thurj vol. 2 issue 2

The Harvard Undergraduate Research JournalVol. 2 Issue 2 Fall 2009

THURJ

A Renaissance Man of the Genomics Era p. 13

Do you know your own mind? p. 16

Victory by Association: Measuring Coattail Effects p. 26

Variations in HIV-1 Epitope Production p.45

DNA Glycosylase: Still Images of a Motion Picture p. 55

LETTERSVolume 2 Issue 2 | Fall 2009

1www.thurj.org

HARVARD COLLEGE

EVELYNN M. HAMMONDS UNIVERSITY HALL, FIRST FLOOR DEAN OF HARVARD COLLEGE CAMBRIDGE, MASSACHUSETTS 02138 BARBARA GUTMANN ROSENKRANTZ TELEPHONE (617) 495-1560 FAX (617) 496-8268 PROFESSOR OF THE HISTORY OF SCIENCE [email protected] AND OF AFRICAN AND AFRICAN AMERICAN STUDIES

November 11, 2009

Dear Harvard Community,

With this issue of the Harvard Undergraduate Research Journal we continue to see

the impressive scientific accomplishments of Harvard College students.

Undergraduate research is important because it helps students navigate the

journey from being science students to becoming full-fledged scientists. The great

inventor and scientist Edwin Land noted that by participating in undergraduate research

students bring to fruition their own special, unique way of thinking and seeing in

environments that support and encourage their intellectual individuality and creativity.

Providing Harvard College students with more opportunities to engage with faculty in

cutting edge research is one of the highest priorities of the College.

The articles in this issue of the Harvard Undergraduate Research Journal show

that our undergraduates are already making impressive contributions to new scientific

knowledge across a wide variety of disciplines. They are also building and sustaining

their own scientific communities while engaged in this work. I look forward to many

more issues of this outstanding journal.

Sincerely,

Evelynn M. Hammonds, PhD

Barbara Gutmann Rosenkrantz Professor of the History of Science and of African and

African American Studies &

Dean of Harvard College

November, 2009


With this issue of the Harvard Undergraduate Research Journal we continue to see the impressive scientific accomplishments of Harvard College students.

Undergraduate research is important because it helps students navigate the journey from being science students to becoming full-fledged scientists. The great inventor and scientist Edwin Land noted that by participating in undergraduate research students bring to fruition their own special, unique way of thinking and seeing in environments that support and encourage their intellectual individuality and creativity. Providing Harvard College students with more opportunities to engage with faculty in cutting edge research is one of the highest priorities of the College.

The articles in this issue of the Harvard Undergraduate Research Journal show that our undergraduates are already making impressive contributions to new scientificknowledge across a wide variety of disciplines. They are also building and sustaining their own scientific communities while engaged in this work. I look forward to many more issues of this outstanding journal.

Sincerely,

Evelynn M. Hammonds, PhDBarbara Gutmann Rosenkrantz Professor of the History of Science and of African and African American Studies &Dean of Harvard College

LETTERS Volume 2 Issue 2 | Fall 2009

2 The Harvard Undergraduate Research Journal

Sincerely,

John MeiCo-Editor-in-Chief

The Harvard Undergraduate Research JournalNovember, 2009


It is with great pleasure that we present to you the fourth issue of The Harvard Undergraduate Research Journal (THURJ), which showcases undergraduate work in the natural and social sciences. This issue represents our continued commitment to original intellectual pursuit as part of the undergraduate experience at Harvard. We are proud to spotlight research that spans topics across the academic spectra, from economics to public health to molecular biology. This year, our prize-winning research article from Oluwatobi Ogbechie investigates peptide degradation in certain immune cell types in response to HIV-1 infection, which has implications for the way our immune system targets and clears infected cells.

In addition to highlighting undergraduate research, our staffers also bring science endeavors in

our community to the forefront. In this issue our feature writers report on the Concord Field Station’s incredible work on the mysteries of navigational motion in everything from mammals to insects. We look into the way George Church’s projects in genomics will affect our lives, the way our own Program for Research in Science and Engineering (PRISE) affects the lives of undergraduate researchers, and the way a web application reveals our the demographic biases of our society. We hope these articles

As always, this issue has been made possible by the dedicated work of our staff, and they cannot go

unmentioned. We thank our peer reviewers, Harvard faculty, graduate students, and associates who reviewed submissions to ensure this publication’s scientific and written caliber. Moreover, our issue could not have been possible without our staffers on the Content, Design, Business, and Social and Public Relations Boards. A special thanks goes out to Harvard College Dean Evelynn Hammonds, HMS Dean Jeffrey Flier, Professor Steven Freedman, Provost Steven Hyman, FAS Dean Michael Smith, Harvard, FAS Dean for the Physical Sciences Jeremy Bloxham, Harvard University, and Harvard College for their unending support . And thanks to you, the reader, whom we hope to delight and fascinate as you flip through the following pages. We hope you enjoy the read!

Lisa RotensteinCo-Editor-in-Chief

ContentsFeatures6 How do animals gallop, jump, and fly with such speed and skill? Learn how one team of scientists are trying to unravel the secrets behind motion. Sarah Zhang ‘11

10 Students spend a summer “getting their science on” as PRISE research fellows. Alissa D’Gama ‘11

13 Putting your DNA on the web: How one man envisions the future of genomics. Isha Jain ‘12

16 Are we more open minded than our parents?Uncovering hidden biases in the mind. Jen Gong ‘12

Research18 Measuring the immune system’s response to HIV with peptidases. Tobi Ogbechie ‘09

28 Victory by ‘coattails’: Hidden influences in Congress on critical election years. S. Travis May ‘09

34 Salt of survival: Startling information on iodine shortages in India. Shubha Bhat ‘09

39 New technology allows scientists to ‘see’ odors as they are processed in the brain. David Blauvelt ‘09

47 Two proteins that regulate neuron growth offer clues on brain development. Stephanie Mok ‘09

55 Snapping stills of DNA repair. Kimberly Oo ‘09

Image credit:

Noor Beckwith

The Harvard Undergraduate Research Journal

Alán Aspuru-Guzik, Ph.D

Assistant Professor of Chemistry and Chemical Biology

Paul Bamberg, Ph.DSenior Lecturer on Mathematics

Michael Brenner, Ph.DGlover Professor of Applied Mathematics and Applied Physics

Myron Essex, D.V.M., Ph.DMary Woodard Lasker Professor of Health Sciences in the Faculty of Public Health

Brian Farrell, Ph.DProfessor of Biology

Jeffrey Flier, M.D.Dean, Harvard Medical School, and George C. Reisman Professor of Medicine

Nicole Francis, Ph.DAssociate Professor of Molecular and Cellular Biology

Steven Freedman, M.D., Ph.DAssociate Professor of Medicine

Robin Greenwood, Ph.DAssociate Professor of Business Administration

Guido Guidotti, Ph.DHiggins Professor of Biochemistry

David Haig, Ph.DGeorge Putnam Professor of Organismic and Evolutionary Biology

Marc Hauser, Ph.DProfessor of Psychology

Dudley Herschbach, Ph.DFrank B. Baird Jr. Professor of Science

John Hutchinson, Ph.DAbbott and James Lawrence Professor of Engineering and Gordon McKay Professor of Applied Mechanics

David Jeruzalmi, Ph.DAssociate Professor of Molecular and Cellular Biology

Efthimios Kaxiras, Ph.DGordon McKay Professor of Applied Physics and Professor of Physics

George Lauder, Ph.DProfessor of Biology and Alexander Agassiz Professor of Zoology

Richard Losick, Ph.DMaria Moors Cabot Professor of Biology

L. Mahadevan, Ph.DLola England Professor of Applied Mathematics

David Mooney, Ph.DGordon McKay Professor of Bioengineering

Hongkun Park, Ph.DProfessor of Chemistry and of Physics

Steven Pinker, Ph.DJohnstone Family Professor of Psychology

Tobias Ritter, Ph.DAssistant Professor of Chemistry and Chemical Biology

Eugene Shakhnovich, Ph.DProfessor of Chemistry and Chemical Biology

Irwin Shapiro, Ph.DTimken University Professor

Zhigang Suo, Ph.DAllen E. and Marilyn M. Puckett Professor of Mechanics and Materials

David Weitz, Ph.DMallinckrodt Professor of Physics and of Applied Physics

Charles Stiles, Ph.DProfessor of Microbiology and Molecular Genetics

Farhan Ali, BSPh.D Candidate in Organismic and Evolutionary Biology

Sandeep Robert Datta, M.D., Ph.D,Postdoctoral Fellow - Axel Lab

Frank Wensheng Fan, Ph.DProgram Coordinator - Harvard School of Public Health China Initiative

Rachelle Gaudet, Ph.DAssociate Professor of Molecular and Cellular Biology

Amitinder Kaur, M.D.Assistant Professor of Medicine

Daniel Kavanagh, Ph.DInstructor in Medicine

R. Paul Johnson, M.D.Associate Professor of Medicine

Filipe Campante, Ph.DAssistant Professor of Public Policy

David Fisher, M.D., Ph.D,Margaret M. Dyson Professor of Pediatrics

Anthony Letai, M.D., Ph.DAssistant Professor of Medicine

Paul Moorcroft, Ph.DProfessor of Organismic and Evolutionary Biology

Karim Kassam, MScPh.D Candidate in Psychology

Nancy Lou Conklin-Brittain, Ph.DLecturer on Human Evolutionary Biology

Guy Crosby, Ph.DAssociate Professor of Nutrition

Boards

BusinessEric Chen ‘12 - Associate ManagerVarun Bansal ‘13Anne Polyakov ‘12Jeanine Sinanan-Singh ‘13Roy Zhang ‘13

ContentAlissa D’Gama ‘11 - Associate ManagerSophie Wharton ‘11 - Associate ManagerJen Jian Gong ‘12Isha Jain ‘12Sarah Zhang ‘11

Peer Review and SubmissionsMeng Xiao He ‘11 - Associate ManagerMonica Liu ‘12 - Associate ManagerCharlotte Seid ‘10 - Associate ManagerJessica Zeng ‘12 - Associate ManagerHelen Yang ‘11 - Head Copy EditorLisa Chen ‘12 - Copy EditorDarius Li ‘12 - Copy EditorJacob Cedarbaum ‘12Andrew Chen ‘12Eric Chen ‘12Sway Chen ‘12Francis Deng ‘12Ben Dobkin ‘12Eva Gillis-Buck ‘12Jen Gong ‘12Johnny Hu ‘11Edward Kogan ‘12Shravani Mikkilineni ‘12Briana Prager ‘12Nicholas Tan ‘12Jacob Weatherly ‘12Ke Xu ‘11Vanisha Yarbrough ‘10Chi Zhang ‘12

DesignRitchell van Dams ‘11 - Associate ChairRyan Neff ‘13 - Associate ChairAmanda Lu ‘13Esther Moon ‘12Allen Shih ‘13Shelun Tsai ‘13

Social and Public RelationsJanet Song ‘13 - Associate ManagerPreya Shah ‘13 - Associate ManagerCaroline Huang ‘13Shannon Purcell ‘12Jeanine Sinanan-Singh ‘13

Managing Editor of ContentFernando Racimo ‘11

Managing Editor of Peer Review and SubmissionsJohn Liu ‘11

Executive BoardCo-Editors-in-ChiefLisa Rotenstein ‘11 and John Mei ‘12

Business ManagerTengbo Li ‘12

Design ChairFrancis Deng ‘12

Manager for Social and Public RelationsHyunje Grace Cho ‘12

Faculty Advisory Board

Faculty Reviewers

CONTENTS Volume 2 Issue 2 | Fall 2009

www.thurj.org

About UsThe Harvard Undergraduate Research Journal (THURJ) showcases

peer-reviewed undergraduate student research from all science and quantitative social science disciplines. As a biannual publication, THURJ familiarizes students with the process of manuscript submission and evaluation. Moreover, it provides a comprehensive forum for scientific discourse on the cutting-edge research that impacts our world today.

At its core, THURJ allows students to gain insight into the peer review process, which is central to modern scientific inquiry. All THURJ manuscripts are rigorously reviewed by the Peer Review Board (consisting of Harvard undergraduates), and the top manuscripts that they select are further reviewed by Harvard graduate students, post-doctoral fellows, and professors. This process not only stimulates faculty-student collaboration and provides students with valuable feedback on their research, but also promotes collaboration between the College and Harvard’s many graduate and professional schools. In addition to publishing original student research papers, THURJ is also an important medium for keeping the Harvard community updated on science research-related news and developments.

Contact

General: [email protected]

Advertising: [email protected]

Subscribing: [email protected]

Submissions: [email protected]

Website: http://www.thurj.org

Copyright 2009 The Harvard Undergraduate Research Journal.

No material appearing in this publication may be reproduced without written permission of the publisher, with the exception of the rights of photographs which may only be granted by the photographer. The opinions expressed in this magazine are those of the contributors and are not necessarily shared by the editors. All editorial rights are reserved.

CONTENTSVolume 2 Issue 2 | Fall 2009

FEATURE Volume 2 Issue 2 | Fall 2009


Move:Move:

Harvard may not boast the reputation of “Ani-mal House”, but just seventeen miles away from campus, the Concord Field Station

could quite literally take the name. The research facil-ity in Bedford is home to animals ranging from guinea fowls to frogs to African pygmy goats. A retired emu—the last from a former study on emu locomotion—still struts its stuff behind the field station’s main office.

Affiliated with Harvard’s Museum of Comparative Zoology (MCZ) and the Department of Organismic and Evolutionary Biology, the Concord Field Station accommodates the labs of Professors Andrew Biewener and Stacey Combes and serves as storage space for the

MCZ’s oversized collections. The field station’s ample size—65 acres adjoining Harvard’s 650 acre Estabrook Woods—gives the scientists enough lab space to con-struct wind tunnels and outdoor insect habitats. Re-search at the field station focuses on the biomechanics of animal movement, which explains the presence of the rather exotic emu in Bedford, Massachusetts.

From Missile Silo to Zoological Lab The facilities of the Concord Field Station origi-

nally belonged to a missile base built during the Cold War. When Harvard University acquired the land and surrounding woods in 1966, the field station was born. “Until recently there was a whole question of whether missiles were ever actually at this base,” says Biewener, the Lyman Professor of Biology and Director of the Concord Field Station, as he pulled up an aerial photo-graph to prove that there were indeed missiles here.

The base’s buildings have been repurposed for science: the barracks became the field station’s main building, housing offices, an animal surgery room, ani-mal care facilities, and, closer to the building’s original purpose, a small apartment for visiting researchers. Re-minders of the field station’s past are most apparent in the underground bunkers now used as storage. Inside the bunkers, equipment originally used to hoist mis-siles above ground is still visible amidst the dozens of whale, dolphin, and porpoise skeletons that make up the MCZ’s oversized cetacean collection. It is obvious why the whale skeletons, from whales that had beached themselves, have to be stored off-site: the jawbone of each whale easily surpasses the height of a grown man.

More interesting than the skeletons are the live an-imals that inhabit the field station. The late C. Richard Taylor, Biewener’s predecessor and founder of the Con-cord Field Station, was a pioneer in the field of animal locomotion. It was under his direction that the field

By Sarah Zhang ‘11 Photography by Noor Beckwith ‘11

On the

Move:Biomechanics at the

Concord Field Station

Move:Move:

FEATUREVolume 2 Issue 2 | Fall 2009

7www.thurj.org

station once housed kangaroos, wal-labies, antelope, llamas, wolves, coy-otes, ponies, and chimpanzees for the purpose of studying locomotion. The original treadmill from Taylor’s days in the 1970s—the first treadmill to have force plates embedded in it and the first to be used by a kan-garoo—is still used in locomotion studies on larger animals, including goats and humans.

Lights, Camera, Flight

Today, Biewener’s research interests are primarily focused on smaller animals, particularly on avi-an flight. When he came to Harvard from the University of Chicago in 1999, a new wind tunnel, essentially an aerial treadmill, was constructed to study birds in flying. The birds, usually cockatiels, are released and photographed in the wind tunnel, which can generate winds up to 65 miles per hour.

Although the wind tunnels and treadmills have factored extensively

into research at the field station, Biewener stresses the importance of studying animal movements in more natural environments. The repetitive walking motion on a treadmill, after all, is quite different from navigating a rocky, winding path, and the same is true of the constant speeds in a wind tunnel versus natural flight. One of the ongoing projects in Biewener’s lab studies the take-off and landing of birds, in which case wind tunnels are not very useful. Instead, doves are set up indoors to fly back and forth, and the displacement of air is tracked

by laser imaging. In the same way that sunlight reveals dust in the air, laser lights up a thin sheet of air, and when a mist is sprayed, particles in the laser light can be tracked by a camera to calculate the flow field around the bird. Transducers, tiny devices that measure the flexion and extension of muscles, are also implanted in the birds to study how they use muscles in different maneuvers.

The same laser imag-ing set-up is used to study 90 degree and 180 degree turns in flight, the mechanism of which is still poorly under-stood. 180 degree turns are especially interesting because they involve the bird almost

stopping in midair at the instant they are changing direction. Flying at slower speeds actually requires the bird to generate more force, which is why hovering is so difficult for most birds. Biewener hopes that laser im-aging can shed some more light on the mechanisms behind these tricky turns.

However, avian flight is only one area of research in Biewener’s lab. Other active research proj-ects include guinea fowl walking, frog swimming, and locomotion in goats.

Chasing Goats

“The immediate plan is to chase a goat around!” laughs doctoral stu-dent Carlos Moreno when asked about his work on African pygmy goats, another project supervised by Biewener. “It’s pretty nonscientific. You just holler and clap and make a lot of noise. And they run away down the corridor,” Moreno says about his goat-chasing techniques. The goal is to make the goat perform evasive maneuvers in order to study the bio-mechanics of turning and dodging movements. Chasing the goat is only the easy part, though probably the most physically strenuous.

Prior studies on goats at the field station have looked at bone bending in turns. Strain gauges are glued to the bone to detect bending,



and the goat is shooed around an en-closed space. Results showed that of two bones in the leg, the radius has less variability in bending than the metacarpal. The radius is a curved bone, so bending mostly happens in one direction, in contrast to the straight metacarpal that can bend every which way.

The remnants of a “cinderblock mountain” once implanted with force plates, stand next to the goat’s runway as a reminder of past climb-ing studies. Current studies con-tinue to focus on the biomechanics of 90-degree turns. A goat is chased down a plywood runway embedded with force plates that detect the force exerted by the goat on the ground. Additionally, a camera captures the movement of the goat’s joints, spe-cially labeled with joint markers.

The act of chasing a goat down a plywood runway may sound silly in itself, but it simulates evolution-arily relevant behaviors by mimick-

ing the act of a gazelle running from a cheetah. A tiny slippage can be the difference between life and death for the gazelle. In contrast to a treadmill, these more variable environments in which goats run allow for more in-sight into how animals turn and ac-celerate, as a gazelle running from a cheetah would.

Goats are chosen for the stud-ies because they are representa-tive quadrupeds, and also because they’re quite harmless. They don’t bite or kick, and are easily made to run. What about their horns? “Han-dles growing out of their heads,” says Moreno. Chasing goats also seems to be a favored activity of Duchess, a pet dog who sometimes makes in appearance at the field station. She stands in contrast to the dozens of other nameless animals that popu-late the field station, since usually, the scientists here are careful to keep a certain distance from their lab animals, labeling them by numbers

instead of names. The animals are well cared for, especially by Pedro Ramirez, the animal care technician who has worked at the field station for [twenty] years – “He’s the reason that science can happen around here” says Moreno – but it’s inevitable that most animals will eventually reach their terminal surgery.

Technologies developed for these studies have also spilled into experiments on human movement and evolution. Professor Dan Lie-berman, from the department of Human Evolutionary Biology, has adapted the goats’ force plates to an-alyze the dynamics of barefoot hu-man running.

Bees Can’t Fly? Unraveling Insect Flight

Just outside the Station’s main lab space is a large tent-like struc-ture with a running pond and a few plants. In early May, this greenhouse is rather empty, but as summer rolls around, the plan is to bring in local dragonflies to study how they chase prey. “Dragonflies do amazing aerial chases,” says Professor Stacey Comb-es, Assistant Professor of Organis-mic and Evolutionary Biology, who heads the second lab at the field sta-tion studying the biomechanics and behavioral ecology of insect flight.

Combes came to Harvard and the field station in 2008 from the Miller Institute at Berkeley, where she studied the flight of wild orchid bees in Panama. A wind tunnel, smaller than the one for birds, is also planned to be built for her study of insect flight. However, Combes too is interested in behaviors beyond smooth flight in a wind tunnel. “In all the years of studying flight no-body has ever looked at how tur-bulence affects flight,” says Combes highlighting her research interests. In Panama, Combes studied bees

What about their horns? “Handles growing out of their heads,” says

Moreno.


9www.thurj.org

with a fan blowing out into the open air, which causes air to curl in and out creating turbulence. There she found that instead of retracting their large legs to reduce drag at high speeds, the bees actually stick out their legs to reduce rolling and pitching, much like a figure skater sticking out his arms to slow a spin.

Combes’ research builds on foundational work done in the past few decades. Until twenty years ago, the aerodynamics of insect flight was a mystery. Insect flight works differently from the more familiar aerodynamics of a bird or airplane, in which air flows faster over the top than the bottom of the wing to cre-ate lift, and the myth once circulated that “insects can’t fly.” The mystery was solved by studying robotic wings in mineral oil, which found that in-

sects beat their wings so fast as to create a negative vortex around their wings, thus generating lift. Combes is interested in taking the study of insect flight outside of controlled lab spaces like a vat of mineral oil and marrying biomechanics with ecol-ogy, thus the dragonfly habitat out-side. Turbulence, of course, is found in any natural environment, and an insect’s ability to fly in turbulence is likely to have evolutionary impor-tance. Whether or not an insect can fly higher in the tree canopy, where there is turbulence, may affect its suc-cess at mating or evading capture.

Additionally, Combes stud-ies the shape and material proper-ties of insect wings, which bend and flop remarkably in flight, though the aerodynamic effects of this flexibil-ity are unclear. Some of this work is

done in collaboration with Professor Rob Wood’s Microrobotics Lab at the School for Engineering and Applied Sciences, which builds miniature ro-botic insects.

The use of robotics is a theme that runs through other research projects too. Until recently, Biewen-er’s lab group worked in collabora-tion with Boston Dynamics, the biorobotics company famous for Big Dog, an amazingly agile robot dog. Data from the goat studies as well as other dog studies at the Concord Field Station went into figuring out the mechanics of Big Dog’s move-ments.

Researchers at the Concord Field Station continue to undertake a wide range of studies on the biol-ogy and biomechanics of movement.The insights they obtain have appli-cations in diverse fields, like robotics and human locomotion. Throughout its forty year history, the Station has harbored countless animals and sci-entists, and today it continues to be a center for novel research. Harvard’s ‘Animal House’ parties on.

Whether or not an insect can fly higher in the tree canopy, where there is

turbulence, may affect its success at mating or evading capture.



Harvard students have a lot of answers to the ques-tion “What did you do

over the summer?” Some may say “intensive language study in Japan”, “an internship with Goldman Sachs in NYC” or “planning events in the White House,” but there are about one hundred who will always en-thusiastically reply “PRISE!”

Put to TaskWhat is PRISE? The simple

answer to Harvard’s love of acro-nyms is the Program for Research in Science and Engineering. The actual answer begins back in Janu-ary 2005, when then Harvard Uni-versity President Lawrence Sum-mers made some widely criticized remarks about gender and science at the National Bureau of Econom-ics Meeting. He shortly thereafter appointed a faculty task force on Women in Science and Engineer-ing, which decided one of its pri-orities was to develop a “summer scientific research community” for undergraduates.

In September 2005, Dean of the College Evelyn Hammonds, then the Senior Vice Provost for Faculty Development and Diver-sity, and Provost Steven Hyman,

provided three years of pilot fund-ing for such a summer research program. With no time to lose if they were to get the program up and running by the next summer, the steering committee—newly ap-pointed director Gregory Llacer, Dean of Administration Georgene Herschbach, and Fiona Chen from the Provost’s Office—set out to de-sign what would become PRISE.

Now transitioned from pilot program status and having already completed a successful four years, PRISE is a staple for the undergrad-uate research community at Har-vard College, providing not only housing and food over the summer,

but also friendships that last well beyond the closing dinner.

Summer In BostonOver the summer, PRISE

Fellows live together in Leverett House, one of Harvard’s upper-classman residential buildings. They are hosted by House Masters Howard Georgi, well known as the professor of Physics 16 and for his Thursday Physics nights, and his wife Ann Georgi, the Life Sciences Undergraduate Research Adviser, who has helped many PRISE Fel-lows find their labs.

Unlike during the school year, when the budding researchers must

Get Your Science On!with the Harvard Program for Research in Science and Engineering (PRISE)By Alissa D’Gama ‘11, THURJ Staff

Credit: Kate Xie THURJ ‘09


11www.thurj.org

also complete problem sets, take orgo midterms, go to lecture, and participate in a slew of other extra-curricular activities, PRISE “creates an environment for a concentrated group of scientists to share their ex-periences and support each other on a daily basis,” said Llacer.

Besides living in Leverett, the Fellows are provided with a stipend for lunch and also eat dinner togeth-er, where the chatter often revolves around the important questions

of the day—including how to get an overlapping three-piece PCR to work and the good movies coming out next week.

This may seem unusual, but Jer-emy Hsu, a PRISE Fellow in Summer

’09, jokes that you can spot a PRISE Fellow in a crowd because “They’re the ones that simultaneously crowd around both the cheesecake and the science!”

Adds Carol Suh, PRISE Fellow in Summer ’08 and Administra-tive Fellow in Summer ’09, “Some people have told me that one of the special things about a PRISE fellow is that you can have a discussion on the most random science topics for hours at a time, even if it’s about

compost and fertilizer.”This kind of conversation is

just what Director Gregory Llacer hopes for every summer; indeed, his favorite part of PRISE is not just the summer, but the entire year, as

the PRISE email list continues to be filled with questions and the Fellows often end up as lab partners or din-ner dates, continuing to engage with each other and with science.

“PRISE provides early interdis-ciplinary opportunities to network with peers to experience science more broadly and to make connec-tions outside more narrow or tra-ditional definitions of scientific re-search,” adds Llacer.

Outside of their labs, the Fel-lows are able to propose and receive funding for a number of science and social activities—a unique aspect of the program that allows the Fellows to take the initiative.

As Hsu recalls, “There were countless fun things PRISE did – whale-watching, getting hit in the eye at point blank range by a water balloon that didn’t explode, try-ing to figure out what exactly a 3-D shadow of a 4-D object is, and ex-ploring the creepy tunnels at Boston Harbor Islands.”

This past summer, for example, Hsu was part of the team that orga-nized a trip to Tanglewood, the sum-mer home of the Boston Symphony

“One of the special things about a PRISE fellow is that you can have a dis-cussion on the most random science topics for hours at a time, even if it’s about compost and fertilizer.”




Orchestra, to see them perform with Yo-Yo Ma, and part of the team that planned a canoeing trip down the Charles river (cleverly named ‘PRISE takes over the Charles’).

The Fellows aren’t limited to just getting to know each other—there are weekly lectures by distin-guished faculty, optional “faculty chats” where fellows can bring in their principal investigator for an informal meal, and an end of the summer dinner for Fellows, their post-docs and graduate students, and PIs.

And of course, there are the presentations given by PRISE Fel-lows at the end of the summer, de-scribing their research and often attended by their lab mentors. For many Fellows, this event, although

a little nerve-wracking, is the first time they get to learn in detail what all their newfound friends had been-working on over the past ten weeks: “It was a remarkable experience to

hear the cumulation of so many summers of research from all your friends and peers, and to be able to hear from such a wide variety of subjects,” said Hsu.

Get Me In!For Harvard students interest-

ed in PRISE, the application, which is due in the spring semester, con-sists of several parts: essays about your proposed project and contri-butions to the community, letters of recommendation, your transcript, and your hopefully abundant enthu-siasm for PRISE. Due to the sheer

volume of applicants, interviews are not feasible, so the committee relies on the written materials the appli-cants present.

The ultimate question the

committee looks to answer is “Will this person effectively contribute to and benefit from being in the PRISE community?” said Llacer.

The chosen Fellows for next summer certainly have a lot to look forward to. While Llacer notes that it is hard to pick his favorite mem-ory from PRISE, he admits that “If I had to single out a specific event it would be sitting in the dunk tank at the PRISE “carnival” of 2006. It was a beautiful, warm day, and getting dumped in the water over and over actually was pretty fun.”

“There were countless fun things PRISE did – whale-watching, getting hit in the eye at point blank range by a water balloon that didn’t explode, trying to figure out what exactly a 3-D shadow of a 4-D object is.”



13www.thurj.org

Harvard Medical School Professor George Church is what many would call ‘a renaissance man’ in the biological sciences. He has published papers in

fields ranging from synthetic biology to sequencing tech-nologies to bioethics, and he helped found the modern science of genomics. But aside from his groundbreaking research, he has also been advisor to 27 different com-panies. His latest endeavor is the Personal Genome Project, which involves sequencing the genome of 100,000 volunteers. Dr. Church sat down with THURJ and told us about his life, his research and his views on science and technology in the genomics era.

Q: You have a strong back-ground in both computer sci-ence and biology. How did these interests merge and de-velop?

A: When I was very young, I got interested in computers. It was re-ally hard, because I wasn’t exactly in an intellectual hotbed. I would play with electrical parts that I could scrounge from construc-tion sites. By 9th grade I managed to get into a better edu-cational system. We had some sort of a toy network that connected Dartmouth to our high school and I learned a lot. I was constantly looking for a way that I could con-nect my interest in math and computers with biology and medicine. My freshman year of college I found the only

guy who was working in biology in the computer science department. Sophomore year I found the only guy in the biology department doing work in computer science. I worked on solving the crystal structure of tRNAs. One of the programs I wrote was still in use thirty years later and is only now fading away.

Q: How did this lead to your later work with sequencing and the Human Genome Project?

A: At one point we wanted to know if the structure we had de-termined for our tRNA was ap-plicable to the other ones that had been sequenced. I typed in all the nucleotide sequences of the known tRNAs and asked if they could fold up to fit our deter-mined structure. I thought it was so cool that we could just type in the sequence of a tRNA and have it fold up. Then I thought, how cool would it be if we could type in the sequence of a human being and have people “fold up”?

Q: So you actually started thinking about the Human Genome Project at that point?

A: Yeah, it was a very vague and immature way of think-ing about it. Looking back it was kind of visionary, but it was really just stupid at the time. So I started becom-

George Church:

A ‘Renaissance Man’ of the

Genomics EraBy Isha Jain ‘12, THURJ Staff

Image Credit - Edge Research

Then I thought, how cool would it be if

we could type in the sequence of a human

being and have

people ‘fold up?’

“



ing obsessed with how we would actually sequence all this DNA. Sequencing struck me as very unsolved, but very solvable. At that time no one used computers in biology, except for maybe a little bit in crystallography or neurosci-ence. There was actually a bit of a sidetrack at this point that some people find amusing. I finished college two years early and then flunked out of graduate school. I was work-ing so hard on the crystal structure of tRNAs. I figured I had already proven that I could take hard courses and do well in them; did I have to do it again? So, I flunked out of Duke and for some reason Harvard let me into graduate school.

Q: What kind of interdisciplinary projects are you working on now?

A: Probably the most interdisciplinary project we have is the Personal Genome Project. We have actually had to in-novate in these fields. We got to the point that we needed a lot of holistic human data. Many previous studies had been very “tunnel-vision”: looking at this liver enzyme or this liver disease. By specifically recruitinh people that are okay with public disclosure, We made a new paradigm in medical research.

Q: What criteria were involved in selecting the PGP10 (the first 10 experimental subjects of the Personal Ge-nome Project)?

A: They were board members of genomic companies or a chief scientific officer or CEO of a sequencing instru-mentation facility. The IRB (Institutional Review Board) wanted it to be very likely that they knew what they were getting themselves into by having published about genom-ics or made public statements. We wanted them to be di-verse within that definition, so that if I was missing some-thing, they could provide another set of advisers. There are certain things in science that you just can’t do yourself. One of the problems with the old way of thinking was that you would put your work in a vault and then only let the people who think the same way you do, see it. Maybe the person with the answers is a school teacher in England - not American, and with no scientific background. The most likely person to think outside of the box is the least likely person to have the credentials.

Q: Perhaps I could ask you about some of your other projects? Could you tell us something about your ‘ag-ing project’?

As its website explains, the PGP aims at “recruiting volunteers who are willing to share their genome se-quence and many types of personal information with the research community and the general public, so that together we will be better able to advance our understanding of genetic and environmental con-tributions to human traits and to improve our ability

to diagnose, treat, and pre-vent illness.” Ten notable names in science, like Steven Pinker, Esther Dyson and George Church himself, have al-ready volunteered to share their entire genomic se-quences with the world at large. They are the mem-bers of PGP-10, the first phase of the project, which may ultimate-ly enroll 100,000 more

participants in the near future.

For more infor-mation, go to: h t t p : / / w w w.p e r s o n a l g e -nomes.org/

Making Your DNA PublicThe Personal Genome Project

The PGP-10 From left to right, George Church, John Halamka, Esther Dyson, Misha Angrist, Keith Bachelder, Steven Pinker , Kirk Maxley, Rosalynn Gill, Stanley Lapidus , and James Sherley. (Credit: Personal Genomics, Inc.)

The3D Model

of DNA

Image: Wikipedia


15www.thurj.org

Image: “The Future of Medicine “ by Ryan Neff; George Church Picture: Personal Genomics, Inc.

genebook Home Profile Microarray cDNA 19 George Church Settings Logout Find Cures

is making massive amounts of genomic data available to doctors and researchers worldwide! twenty minutes ago clear

Wall Info Alleles Cures +Research!

View My Genes (8527)

Analyze My DNA for Diseases

Edit My Cures

All Posts Posts by George Posts by Others SettingsToday

What genes are you expressing right now? Post

Personalized RNA Sequence for Diabetes 7:30 AM

1 atgaatccaa accaaaagat aataaccatt ggttcggtct gtatgacaat tggaatggct 61 aacttaatat tacaaattgg aaacataatc tcaatatgga ttagccactc aattcaactt 121 gggaatcaaa atcagattga aacatgcaat caaagcgtca ttacttatga aaacaacact 181 tgggtaaatc agacatatgt taacatcagc aacaccaact ttgctgctgg acagtcagtg 241 gtttccgtga aattagcggg caattcctct ctctgccctg ttagtggatg ggctatatac 301 agtaaagaca acagtataag aatcggttcc aagggggatg tgtttgtcat aagggaacca 361 ttcatatcat gctccccctt agaatgcaga accttcttct tgactcaagg ggccttgcta 421 aatgacaaac attccaatgg aaccattaaa gacaggagcc catatcgaac cctaatgagc 481 tgtcctattg gtgaagttcc ctctccatac aactcaagat ttgagtcagt cgcttggtca 541 gcaagtgctt gtcatgatgg catcaattgg ctaacaattg gaatttctgg cccagacaat 601 ggggcagtgg ctgtgttaaa gtacaacggc ataataacag acactatcaa ........

“Maybe the cure for cancer is written... in my genes!”

George Church

A: We have been working on the aging project for a few years. We have one of the best comparative zoology da-tabases. What we ideally want are two animals that are re-ally close to each other in sequence but have very different longevities. Two species that have very similar sequences are the mouse and the naked mole rat. The mouse is an ideal human surrogate. It turns out that the typical mouse lives just over two years and the naked mole rat lives about twenty years. So what we are doing is taking big chunks of DNA from the mole rat and putting them into the mouse, replacing the mouse gene with the mole rat gene to see if we can increase its longevity.

Q: How direct is your involvement with businesses and companies, and what is your motivation for de-veloping biotech products?

A: YYou can choose to just publish [your research] and hope that someone pays attention, which is increasingly not the case. In the early days of molecular biology, com-panies would make the enzymes for you [the academic researcher]. And then that wasn’t enough. So then the

companies would make a kit for you. And then even that wasn’t enough. And eventually they would build a device that would implement the kit. This has made it easier for scientists to use a wide variety of tools, but you can’t al-ways hide in the ivory tower. You need to make sure that you have enough control over whatever method you are developing, and work with a company to ensure they don’t botch it. You don’t want to micromanage your company just like you don’t want to micromanage your lab, but you do want to manage it.

Q: You have a very eclectic range of topics. How do you keep track of them?

A: Well, to me they all fit into one thing. There are a rela-tively small number of applications that make sense to me. For example, I say if you have to pick a disease, what is the thing that everyone dies of? When you are between the age of 20 and 45, not much happens to your body. It is only once you start aging that almost every organ system starts fading. Aging is really the fundamental problem and all the other things are just symptomatic.



age and religion to expose a myriad of implicit biases These tests (located online at https://implicit.harvard.edu/implicit/demo/) are an eye-opening experience for many subjects. Since its launch in 1998, over 4.5 million online visitors have completed the web-based version of IAT, generating an immense data set on the prevalence of prejudice. The data yield a clear conclu-sion: implicit biases not only exist, they are pervasive across large sets of the general population.

Currently, Banaji’s research focuses on how implicit attitudes and preferences first came to be. “We know that many of them are learned by us after we are born…but our minds seem infinitely ready to learn them, for some reason.” This curious “readiness” has led Banaji and her colleagues to conduct studies utilizing young children, as well as adults, in order to look for pat-terns in the appearance and development of implicit attitudes. They studied children’s explicit attitudes to-wards certain groups, but also measured their uncon-scious biases using an IAT. They found that younger children spoke more openly about their preferences, while older children and adults tended to mask their biases with more socially acceptable language. How-ever, “on the IAT, if white American adults are show-ing a certain level of preference for whites over blacks, then young children in that group are showing the exact same preference.” These IAT data seem to in-validate the assumption “that young kids should not really show some of those biases, because they are still evolving.” In fact, implicit bias is present in what looks to be exactly the same form in children (as young as age 6) as it is in adults. Disturbingly enough, implicit biases may begin to develop at a far younger age than we previously thought.

In addition to trying to understand biases from a developmental perspective, Banaji and her colleagues

UncoveringImplicit Biases

By Jen Jian Gong ‘12

Many of us believe we are more impartial than our historic predecessors, whether it be in matters of race, gender or ethnic equality.

Landmark achievements like the Universal Declara-tion of Human Rights, the end of apartheid or the rise of a black president to the highest office in America might seem like enough reason to believe that human-ity’s biases are a thing of the past, at least in the de-veloped world. Recent scientific discoveries, however, suggest that they are not.

Harvard psychology professor Mahzarin Banaji is a leader in the field of implicit social cognition, which investigates what she has termed “implicit biases”- un-conscious prejudices that persist even as our explicit attitudes evolve. She is the current Richard Clarke Cabot Professor of Social Ethics and is one of the prin-cipal investigators of “Project Implicit.” She is joined in this project by Brian Nosek, from the University of Virginia.

According to Nosek, the goal of Project Implicit is “to understand thoughts and feelings that exist out-side of awareness and control.” The project functions as a virtual laboratory, where online visitors can assess their own implicit attitudes using the so-called Implic-it Association Test (IAT). The IAT requires subjects to rapidly categorize pairs of stimuli, using differences in response time as an indicator of implicit bias. If, for example, a subject holds an unconscious bias against a certain group, then one would expect the subject to have more difficulty associating positive stimuli with images of that group’s members than with images of members belonging to a preferred group. This diffi-culty manifests itself in slower response times, making the IAT a reliable indicator of bias.

This technique has been used to explore many forms of prejudice, using group classifications like race, sex, Ill

ustra

tion

by A

man

da L

u

Do you know your own mind?


17www.thurj.org

have begun using neuroimaging techniques to under-stand how our brain activity varies when we distinguish between different groups of people. Collaborating with her is Jason Mitchell, an Assistant Professor in the De-partment of Psychology. As Banaji explains, “Most of us feel that we treat people from different groups largely the same, that our minds focus on them in the same way. Our job is to look at how early in the sequence our brains start doing something different when we look at somebody.”

In one of their recent studies, published in the journal Neuron, subjects heard descriptions of people of varying political views and were asked questions about them. The study showed that “many liberals literally use a different set of neurons when they’re thinking about Joe, the liberal, versus Jeff, the conservative.” Most of the questions the researchers asked about the characters did not touch on political beliefs, for exam-ple: ‘Do you think he does his laun-dry every week?’ The neuroimaging data from these studies showed that there was more activity in the ven-tral medial prefrontal cortex (mPFC) when subjects were asked to make judgments about those with similar political leanings. When they were asked questions about those with dissimilar views, there was more activity in the dorsal mPFC.

This research is clearly relevant to our daily lives, with far-reaching educational and judicial implications. Banaji receives many phone calls about her work – and not just from other psychologists. Individuals in busi-ness and law want to know more about these uncon-scious prejudices because they have the potential to play an integral role in their own jobs. “People don’t know that there is this stuff in their mind that ends up affect-ing how they treat people,” she explains. “Because their intention is to do the right thing, they become really disturbed when they learn that maybe they’re not acting in the interest of their patient.”

Banaji sees these implicit biases as “mental viruses”, analogous to any other illness or physical virus we have discovered. Since everyone is susceptible to implicit bias, Banaji says, “when your role is to be public de-fender or a social worker, you especially don’t think of yourself as harming anyone. But imagine that you too carry those viruses. Are you really able to do your job as well as you might be able to?”

Scholars are also beginning to recognize the relevance of this research to their own fields. Professor Jerry Kang, from the University of California, Los Angeles Law School, has collaborated with Banaji to explore the ways that such scientific findings should shape the study and practice of law. For example, one collaborative paper addresses affirmative action policy and findings from implicit social cognition that might affect its implemen-tation. “Right now, civil rights and racial justice talk seems to be stuck in the mud,” Kang says. Rather than utilizing “new philosophical arguments about justice,” perhaps the findings from research into the brain and mind will eventually provide “insurmountable evidence

that we are not in fact ‘colorblind’.”While it is true that a growing body

of psychology literature is revealing our unconscious tendencies to think about groups of people differently, the im-portant question seems to be whether or not these biases can be changed. Is there a cure for the mental virus?

For Banaji, the answer is a resounding yes: we are adaptable beings, and these implicit biases are not set in stone. She believes there is a “very hopeful mes-

sage that’s coming from a lot of work: that our systems are highly adaptable, that we are very malleable.”

She suggests that a diverse campus like Harvard’s is also a good laboratory in which to test the malleability of our minds. Experiences with different groups of peo-ple can challenge our deep-seated biases and, perhaps, begin to change them. “If we’re indeed adaptable, then these simple notions of who’s conservative and who’s liberal will pose for us opportunities to see if our minds can take that kind of leap.”

Thanks to the IAT and similar research endeavors, we are more aware of implicit biases than past generations. So, Banaji asks, “now that we know they exist…will we do with this information what we do with new medi-cal information?” We now “know that there are things happening between us that we can’t see…and that it’s costing our society.” Because we can no longer claim ig-norance, she argues, “the standard for us is different.”

The first step towards combating these viruses, Bana-ji firmly states, is “awareness, awareness, awareness.” Since they operate outside conscious awareness, it is not always easy to acknowledge our biases. But, they are consequential nonetheless, affecting who we marry, be-friend, hire or convict.

For Banaji, the an-swer is a resound-

ing yes: we are adaptable beings, and these implicit biases are not set

in stone.

RESEARCH Volume 2 Issue 2 | Fall 2009


Imm

unol

ogy Differential peptidase activity in

CD4+ T-cells and monocytes results in variations in HIV-1 optimal epitope

production and degradationOluwatobi Ogbechie

Harvard College ‘09, [email protected]

Human Immunodeficiency Virus 1 (HIV-1) infects several cell types, including CD4+ T-cells and monocytes. These cells are important virus targets because they are substantially depleted during infection and can also store dormant virus, which fuels the disease at later stages. A crucial part of the recognition and clearance of infected cells involves their presentation of pieces of viral protein, called epitopes, at their surface for recognition by immune cells. These pieces originate from the degradation of viral peptides by intracellular peptidases. It is unknown if cell types that are infected by HIV present similar viral peptides to evoke analogous immune responses. Here, I examined whether CD4+ T-cells and monocytes would exhibit differences in the production of viral peptides and if that could affect the HIV-specific immune responses. I measured the activity and kinetics of intracellular peptidases involved in viral peptide degradation in CD4+ T-cells and monocytes using an in vitro fluorescence-based hydrolytic activity assay. Peptidases’ activities were higher and faster in monocytes. Speculating that these differences in activity might affect viral peptide production, I measured the production of a 9-amino acid long HIV epitope (RK9) from a longer peptide using an intracellular degradation assay where the longer peptide is incubated with cytosolic extract from CD4+ T-cells and monocytes. I used RP-HPLC to identify degradation products. The longer peptide’s degradation and RK9 production were faster in monocytes. In addition, monocyte degradation products elicited greater epitope-specific immune responses. I sought to determine the overall effect of these cell-type-specific differences in peptidase activity on the intracellular degradation of randomly selected HIV epitopes using a similar intracellular degradation assay. The epitopes’ half-lives were significantly shorter in monocytes than in CD4+ T-cells. These differences in antigen processing could affect presentation of HIV-1 peptides to immune cells and might influence clearance of infected CD4+ T-cells and monocytes.

Introduction

HIV-1 infectionHuman Immunodeficiency Virus 1 (HIV-1) is a retrovirus that

preferentially infects human nucleated cells with tropism towards CD4+ T or monocyte-derived cell lines (Callaway et al., 1999; Sup. Fig. A). After infection, the virus integrates its retro-transcribed DNA into the host cell’s genome and can either reproduce numerous active progeny or remain quiescent (Stevenson et al., 1990; Hauber et al., 1987). Cells with dormant virus, known as viral reservoirs, can linger in the host indefinitely (Chun and Fauci, 1999).

Of all the Peripheral Blood Mononuclear Cell (PBMC) subsets, researchers implicated CD4+ T-cells and monocytes as key reservoirs of virus during infection (Smith et al., 2003; Brenchley et al., 2004). Severe depletion of CD4+ T-cells occurs during the acute stage of HIV infection with around 40-50% lymphocyte loss (Hel et al., 2006; Sup. Fig. B). However, monocytes do not drastically decline in numbers as CD4+ T-lymphocytes do (Douek, 2007). During chronic infec-tion, viral levels in the blood rise, the immune system weakens, and consequently AIDS symptoms, like Kaposi’s sarcoma, occur (Douek, 2007). Progression to AIDS is faster among HIV-infected patients with drawn out symptoms during acute infection due to a weaker immune response (Pedersen et al., 1989).

Immune response to HIV-1 infection Effective regulation of the acute and chronic phase by the host’s

immune system gives the patient a much better prognosis for the dis-ease (Vanhems and Beaulieu, 1997). Patients with little loss of CD4+ T-cells have slow disease progression, hence validating the impor-tance of an efficient immune response (Sheppard et al., 1993). For this study, I was interested in aspects of the cell-mediated immune response because cell types involved in this response that undertake anti-HIV specific functions could increase the host’s capability to control HIV-infection (Schmitz et al., 2001).

To initiate cell-mediated responses, circulating dendritic cells (DCs) in the bloodstream can phagocytose the foreign pathogens, process them, and present the viral peptides to immature CD4+ and CD8+ T-cells in order to begin their maturation process (Randolph et al., 2008). Mature CD8+ T-cells become cytotoxic T-lymphocytes (CTLs) that target and kill infected cells or produce other antiviral responses, and mature CD4+ T helper cells either assist in the recruit-ment of CTLs, aid in the production of neutralizing antibodies, or produce cytokines that enhance the immune response (Betts et al., 2001; Hel et al., 2006). Thus, infection of CD4+ T-cells greatly weak-ens the immune system. Also, it gives a rationale for the success of long-term non-progressors who have highly efficacious T-cell-medi-ated responses against HIV (Paroli et al., 2001).

RESEARCHVolume 2 Issue 2 | Fall 2009

19www.thurj.org

Imm

unology

Antigen processing One of the most important events in a cell-mediated immune re-

sponse is antigen processing, which encompasses the degradation, transport, and presentation of both self and foreign proteins to CTLs. There are two major antigen processing pathways: endogenous and exogenous. The endogenous pathway

Defective, newly synthesized, or intracellular proteins can be tagged with multiple copies of ubiquitin, a protein that targets them for proteolysis in the proteasome (Schubert et al., 2000). Three of the proteasome’s beta subunits bear the protease’s active sites: caspase-like, chymotrypsin-like, and trypsin-like, which cleave proteins after acidic, hydrophobic, and basic amino acids, respectively (Kopp et al., 1997).

After proteasomal processing, the endogenous pathway utilizes peptidases in the cytosol, such as tripeptidyl peptidase II (TPPII), thimet oligoendopeptidase (TOP), and aminopeptidases, to further degrade the peptide products (Groothuis et al., 2005). Recently, an in vitro intracellular experiment demonstrated that TPPII was neces-sary for the generation of an HIV-1 epitope presented through the endogenous pathway in DCs (Seifert et al., 2003). Like TPPII, TOP can generate C-termini that are different from peptides produced by the proteasome, but TOP prefers to cleave shorter peptides (Oliveira et al., 2001). Saric and colleagues determined that TOP was a criti-cal intracellular endopeptidase that degraded proteasomal products of 9-17 amino acids (Saric et al., 2004). The large family of cytosolic aminopeptidases further processes these peptide products (Hattori and Tsujimoto, 2004). However, an experiment on the capability of mice lacking leucine aminopeptidase to present viral peptides found that that this peptidase was not necessary to generate peptides for the MHC-I pathway (Towne et al., 2005). This evidence does not contra-dict previous findings that implicate aminopeptidases as contribu-tors to antigen processing. Rather, it suggests that many peptidases contribute to the processing of viral peptides.

After cytosolic processing, peptides around 9 amino acids long with a preferred C-terminal residue can be transported to the en-doplasmic reticulum (ER) by transporter associated with antigen processing (TAP) (Larsen et al., 2005). In the ER, degradation by ERAP-1 may continue before the peptide is loaded onto an MHC class I molecule. Stronger binding affinity of the peptide in the MHC binding groove displaces the complex from Tapasin, a protein that anchors MHC-I in the ER. Released, MHC-I/peptide travels through the trans-golgi pathway to be presented at the surface of the cell

(Groothuis et al., 2005). CTLs primed by DCs recognize the epitope presented as well as the MHC-I molecule (Frahm et al., 2007; Fig. 1). If the CTL can recognize a non-self peptide, like an HIV-1 epitope, it kills the infected cell or produces other antiviral responses. The exogenous pathway

Professional antigen presenting cells (APCs), like DCs, use this pathway during the initial activation of the immune response to ex-tracellular pathogens. Here, MHC class II molecules on APCs present HIV-1 epitopes to CD4+ T-cells after cathepsins peptidase degrada-tion (Pajot et al., 2007). Antigen processing in PBMC subsets

It is generally assumed that antigen processing across cell types is similar, though one can deduce that differences in presented peptides can lead to variations in the immune response. For example, cellular production of more antigenic peptides should lead to stronger im-mune responses. Despite the multi-layered comprehension of antigen processing, the process is not well understood in PBMC subsets in-fected by HIV. Since HIV-1 targets certain PBMC cell subsets, such as CD4+ T-lymphocytes and monocytes, it is necessary to understand both the type of immune response that these subsets elicit and the processes by which they are generated.

Although direct comparisons of antigen processing in CD4+ T-cells and monocytes have not been previously reported, some work has been published about differences in antigen processing between other cell types. First, mouse fibroblast and dendritic cell lines in-fected with lymphocytic choriomeningtitis virus (LCMV) presented different viral peptides to CTLs, resulting in non-identical cell type-specific immune responses (Butz and Bevan, 1998). The presented peptides varied in their antigenicity; thus immune responses to DCs and fibroblasts varied, showing the possibility of functional dif-ferences in antigen processing among cell types. Another study in mice found that dendritic and non-dendritic cells presented different epitopes to CTLs during primary and secondary infection of influ-enza (Crowe et al., 2003). This differential presentation affected the immune response through the generation of distinct memory CD8+ T-cells for each cell type. Both of the aforementioned suggest cell-type-specific variations in antigen processing that affect the immune response, making this of high interest to HIV research since HIV-1 has tropism for select cell types.

When this is applied to immune responses to HIV-1, where in-fection is subset-specific and depletion of lymphocytes is skewed to-wards CD4+ T-cells, one realizes the need to identify and describe such differences. Furthermore, this question about the differences in antigen processing between the leukocyte subsets is applicable to vaccine research because of the increasing promise of T-cell vaccines (Johnston and Fauci, 2007). These vaccines attempt to elicit memory T-cell responses, especially CD8+ T-cell responses to kill infected cells as soon as a recipient of the virus is infected with the virus. Also, T-cell vaccines that focus on improving the host’s immune response during acute and/or chronic infection would advance the epidemio-logical control of the HIV pandemic. A necessary prelude to creating such a vaccine is ensuring that infected cells, CD4+ T-cells and mono-cytes, will present peptides that CTLs can recognize.

A thorough understanding of HIV pathogenesis will require knowledge about the immune response that HIV-infectible cells, such as CD4+ T-cells and monocytes, can elicit. First, the two tropisms of HIV-1 strains are for CD4+ T-cell lines and for monocyte lineage cells (Weiss, 2008). For instance, CD16+ monocytes, which are precursors to DCs and macrophages, have increased permissivity to HIV-1 in-fection with HIV-seropositive individuals having elevated numbers of this cell population (Ellery et al., 2007). Besides infectibility, both CD4+ T-cells and monocyte cell lines can serve as latent reservoirs of

Figure 1. Endogenous pathway of antigen processing. 1) HIV in-tegrates its genome into host 2) Transcription of HIV genes and translation of protein occur 3) Ubiquitination and degradation of HIV protein in the proteasome 4) Peptide products are further degraded in aminopeptidase, TOP, and TPPII 5) Optimal epitope is transported into the ER through TAP 6) and is subjected into further degradation in ER by ERAP, but can be loaded onto MHC-I heavy chain/Tapasin complex 7) Tapasin disso-ciates from completely assembled MHC-I/peptide complex 8) Complex moves through Golgi to 9) cell surface. 10) T-cell receptor (TCR) of CTL and co-receptor, CD8, recognize MHC-1/peptide complex.



Imm

unol

ogy

HIV. The creation of reservoirs begins during primary infection in resting CD4+ T-cells, and there is evidence that monocyte-derived macrophages in the gastrointestinal tract can serve as reservoirs for the virus during chronic infection (Chun et al., 1998; Smith et al., 2003).

As important as CD4+ T-cells and monocytes are during HIV in-fection, there is little known about the differences in their antigen processing that can potentially affect the immune response. It is known that gene expression between the PBMCs and T-lymphocyte subsets varies in HIV-seropositive individuals (McLaren et al., 2004). This introduces the possibility of variations in the expression of an-tigen processing genes and in the activity of their protein products, leading to functional differences. Such variations between subsets targeted by the virus may affect the kinetics, the amount, or the iden-tity of peptides presented to CTLs and may alter the capacity of CTLs to kill infected CD4+ T-cells or monocytes.

Materials and Methods

Blood processing Blood was drawn from healthy and HIV-infected donors (Massa-

chusetts General Hospital, Boston, MA). PBMCs were isolated from blood using Ficoll-Hypaque (Sigma, St. Louis, MO) density gradient separation after a 30-minute spin at 1500rpm at room temperature. Cells were washed three times in Hank’s Balanced Salt Solution with 10mM HEPES and 1% PSG. After each wash, cells were spun at 1500rpm for 10 minutes at room temperature. Isolation of PBMC subsets

CD4+ T-lymphocytes and monocytes subsets were isolated from PBMC and magnetically immunosorted using the EasySep® Human CD4+ T-cell Enrichment Kit and EasySep® Monocyte Enrichment Kit according to the manufacturer’s instructions (StemCell, Vancouver, Canada). Enriched cells were washed three times in Dulbecco’s Phos-phate Buffered Saline (Sigma) at 1500rpm for 10 minutes at room temperature. Dry cell pellets were kept at –80°C until cytosol extrac-tion.Cytosolic extracts

Cell pellets were resuspended in a Digitonin buffer (50mM HEP-ES, 50mM KAc, 5mM MgCl2, 1mM DTT, 10% glycerol, 1mM ATP, 0.5mM EDTA, 0.0125% Digitonin) for 5 minutes to allow sufficient detergent permeabilization and then spun at 20,000rcf for 15 min-utes. Protein concentration in both methods was measured with the DC Protein Assay Kit from Bio Rad Laboratories (Hercules, CA). Whole cell extracts were kept at –80˚C until use. Actin normalization

To standardize the amount of extract used in experiments, actin in each sample was compared to a sample known to contain 3µg cy-tosol. The normalized amount of each sample was the volume of cell extract with an actin content that matched 3µg of actin in the known sample after Western Blot analysis.Western Blot Analysis

All samples were subjected to Western Blot analysis. First, whole cell extracts were mixed with laemmli buffer and heated at 85°C for 10 minutes. Lysates were run through a 12% SDS-PAGE gel at 100V through stacking gel and 125V through running gel. The gel was transferred to PVDF transfer membrane (GE healthcare) at 25mA overnight. The membrane was blocked at room temperature for at least 30 minutes in 5% milk and 1% NP40. Primary anti-beta actin antibody (Abcam, Cambridge, MA) was diluted to 1/20,000 in 5% milk and 0.1% NP40, added to the membrane, and incubated for 1 hour at room temperature. Following the primary, the membrane was washed 5 times with 0.1% NP40 every 10 minutes. Secondary

anti-mouse and anti-rabbit antibody coupled to horseradish per-oxidase (GE Healthcare) was diluted to 1/6000 in 5% milk and 0.1% NP40 and incubated for 40-45 minutes in room temperature with the membrane. Membrane was washed again as stated above. For imag-ing, ECL Plus Western Blotting Detection Reagents and Amersham Hyperfilm ECL were used following manufacturer’s instructions (GE Healthcare). Peptidase inhibitors and substrates

Caspase-like, chymotrypsin-like, and trypsin-like active sites of the proteasome were inhibited by the proteasome inhibitor MG132 using a 50mM stock solution in DMSO (Sigma-Aldrich, St. Louis, MO). Aminopeptidase activity was inhibited by Bestatin hydro-chloride using a 12mM stock solution in DMSO (Sigma-Aldrich, St. Louis, MO). Cpp-AAF-pAb, the TOP inhibitor, was diluted in DMSO for 1mM stock solution (Bachem, Torrance, CA). The TPPII inhibitor Butabindide oxalate was diluted with DMSO for a 10mM stock solu-tion (TOCRIS Bioscience, Northpoint, UK). TPPII, chymotrypsin-like, and trypsin-like activities were measured with H-Ala-Ala-Phe-Amc, Suc-LLVY-Amc, and Boc-LRR-Amc [where Amc represents 7-amido-4-methyl-coumarin] respectively and were resuspended in DMSO to produce stock solutions of 10mM, 50mM, and 100mM (Biochem Bioscience, Torrance, CA). The aminopeptidase substrate Leu-AMC and the proteasome caspase substrate ZLLE-Amc were obtained from Calbiochem (San Diego, CA) to be resuspended in DMSO for respective stock solutions of 50mM and 50mM. The TOP substrate Mcc-PLGPK-Dnp was resuspended in DMSO for a stock so-lution of 10mM. All inhibitors and substrates were stored at –20°C. Antigen Processing Peptidase Activity Assays

For proteasome peptidase activities, 3µg of whole cell extract was incubated in a buffer containing 20mM HEPES, 50mM KAc, 5mM MgCl2, 1mM DTT, and 1mM ATP at room temperature for 30 minutes. To ensure the specificity of the assay, 1μM MG132 was incubated with the extract. After the incubation, substrates for the different proteasome active sites were added and fluorescence was immediately measured every 5 minutes over 1 hour at 37°C with a Victor-3 plate reader (Perkin Elmer, Boston, MA). 75μM ZLLE-Amc was used to measure caspase-like activity, 100μM Suc-LLVY-Amc was used for chymotrypsin-like activity, and 25μM Boc-LRR-Amc was used for trypsin-like active site activity. Aminopeptidase, TPPII, and TOP fluorescence assays were conducted in similar conditions, excluding DTT and ATP in the buffer. Specificity was determined by incubation with 120μM Bestatin, 10μM Cpp-AAF-pAb, or 1μM Butabindide for aminopeptidase, TOP, and TPPII activities respec-tively. Following the 30 minute incubation, respective substrates were added: 50μM Leu-Amc, 20μM Mcc-PLGPK-Dnp, and 100μM H-Ala-Ala-Phe-Amc for aminopeptidase, TOP, and TPPII. Fluores-cence was read as stated above. Excitation and emission wavelengths for trypsin-like, aminopeptidase, and TOP fluorescent products were 345nm and 405nm, respectively. Those wavelengths for caspase-like, chymotrypsin-like, and TPPII fluorescent products were 380nm and 460nm, respectively. Synthetic peptides and RP-HPLC peptide analysis

Peptides were made by the AAPPPTEC Apex 396 multiple peptide synthesizer at the MGH peptide core (Massachusetts General Hospi-tal, Boston MA). Peptides were then purified by reverse-phase high pressure liquid chromatography (RP-HPLC) and their sequences were verified by mass spectrometry, which showed greater than 95% purity (Partners Proteomics, Cambridge, MA). Eluted peptides pro-duced defined peaks on the RP-HPLC where the area under the peak was proportional to the amount of peptide in the analyzed sample. Defined peptide peaks were calibrated using a 4.6x50mm 3mm C18 column (Waters, Milford, MA).

21www.thurj.org

Imm

unology

Intracellular peptide degradation assaysPBMC, CD4+ T, and monocyte whole cell extracts from the same

healthy donor were resuspended in buffer containing 20mM HEPES, 137mM KAc, 1mM MgCl2, and 1mM ATP in nano-pure H2O. Subse-quently, peptide was added to the solution, at which point the incuba-tion at 37°C began. At 0, 10, and 30 minutes, aliquots of the reaction containing 30μg whole cell extracts and 6nmol peptide were stopped with 0.3% Trifluoroacetic acid (TFA) (Sigma Aldrich, St. Louis, MO). 6nmol of pure peptide in same buffer and TFA conditions was used to identify the elution time and amount of the undigested peptide through RP-HPLC. All peptides were eluted using a gradient solution of two buffers. The first was 0.05% TFA in nanopure H2O, while the second buffer varied depending on the peptide. For peptides 5RK3, A3-RK9, B57-YY9, Cw3-RL9, A33-ER8, B7-HA9, A3-K11, the second buffer either contained 50% acetonitrile (AcN) and 0.01% TFA or 100% methanol and 0.05% TFA. The second elution buffer for B7-FL9, B7-TL10, and B7-FR10 contained either 100% AcN and 0.03% TFA or 100% methanol and 0.05% TFA. Integrated peaks from digested peptides were compared to the undigested peptide. Identities of some digested peaks were verified by mass spectrometry. Antigenicity of produced peptides

Peptide products from the 17-mer (gag) intracellular degradation assay were purified with 10% trichloroacetic acid precipitation and diluted in RPMI-1640 without serum. Subsequently, pH was adjusted to 7.4. HLA-A3+ B cells derived from HeLa cells served as target cells and were labeled with 51Cr and pulsed with 0.5ug/ml of the peptide products for 30 minutes at 37°C without serum. CTL clones specific for RK9 were incubated for 4 hours with the B cells at a 4:1 effector-target ratio. Cell lysis was defined as [(51Cr release from B cells incu-bated with digested products - spontaneous release) / (total release – spontaneous release)]. Cell lysis values were also compared to that of HLA-A3+ B cells that were incubated with undigested 17-mer peptide at a 0.5µg/ml concentration. Statistical analysis

Maximum slope in the fluorescence assays was calculated with Workout 2.0 using a liner kinetic fit of the fluorescence at different time intervals (Perkin Elmer, Boston, MA). The differences in maxi-mum slopes and fluorescence levels between CD4+ T-lymphocytes and monocytes were analyzed using the non-parametric Mann-Whitney test with GraphPad Prism 5 (La Jolla, CA). This test was used since it does not assume a normal distribution for the data, yet it assumes that both samples are independent. The half-lives of pep-tides used were determined using a non-linear regression of an one phase exponential decay with the same software.

Results

Fluorogenic assays of antigen processing associated peptidases As protein degradation of HIV-1 proteins is crucial to the even-

tual presentation of viral peptides to CTLs, I sought to characterize the activity of peptidases associated with antigen processing—pro-teasome-caspase-like, -chymotrypsin-like, and -trypsin-like; amino-peptidase; TOP; and TPPII—in CD4+ T-cells and monocytes. To do this, I utilized an in vitro fluorescence-based hydrolytic assay already developed by others in the group. Here, I incubated cytosolic extracts from peripheral blood mononuclear cells (PBMC), CD4+ T-cells, and monocytes with substrates that fluoresce upon cleavage by specific peptidases (Figure 2). The measured fluorescence of the sample was directly proportional to the activity of the peptidase over time. Af-ter a 30-minute pre-incubation with peptidase-specific inhibitors, I found that the average specificity of substrates to their respective peptidases in PBMC, CD4+ T, and monocyte subsets was between 80% and 95% (Sup. Table 1). I must note that cleavage of Leu-Amc in monocytes was about 50% specific to aminopeptidases, which indi-cated that another peptidase along with aminopeptidase cleaved Leu-Amc. However, fluorescence from cleavage in monocytes was much greater than in other cell extracts that we considered the discrepan-cies to be negligible (data not shown). CD4+ T-cells and monocytes have different peptidase activity levels

After ensuring the specificity of the fluorescence assay, I sought to characterize the activity of the proteasome, aminopeptidase, TOP, and TPPII in two PBMC subsets targeted by HIV: CD4+ T-cells and monocytes. Prior research indicated that activities of the antigen processing machinery varied according to cell type (Butz and Bevan, 1998). Therefore, we hypothesized that the proteasome, aminopep-tidase, TOP, and TPPII would show differences in activity between CD4+ T-cells and monocytes. Using blood from 8 healthy donors, I purified the PBMC and immunosorted CD4+ T-cells and monocytes. With equal amounts of cell extracts normalized through western blots against actin and GAPDH, I compared peptidase activities of whole PBMC, CD4+ T-cells, and monocytes by using the fluorescence emitted by the sample at 1 hour after the 37°C incubation. All assays

Figure 3. Different activity levels of peptidases associated with antigen processing between CD4+ T-cells and monocytes. Pro-teasome caspase-like and trypsin-like, aminopeptidase, TOP, and TPPII activities were measured in whole cell extracts from PBMC (•), immuno-sorted CD4+ T-cells (■) and monocytes (▶) after a 1 hour incubation with peptidase-specific substrates at 37°C. Fluorescence was compared using a non-parametric Mann-Whitney statistical test. Significant differences between CD4+ T-cells and monocytes were found. *p<0.05, **p<0.01, ***p<0.001. n=8.

Figure 2. Intracellular fluorescence-based hydrolytic activity assay. Cell extracts were incubated with specific fluorogenic substrates at 37°C for 1 hour. Cleavage of those substrates produced fluorescence. Pep-tidase activity is proportional to fluorescence. Sample proteasome-trypsin assay shows fluorescence level over time. Slope and 1-hour fluorescence are indicated on graph.

Imm

unol

ogy

were repeated three times, and the arithmetic mean of all trials was used as the final value for comparison. A representative sample shows the typical fluorescence pattern with monocytes having a higher fluorescence at the 1-hour time point than CD4+ T-cells (Figure 2). Overall, I found that proteasome-caspase-like, proteasome-trypsin-like, aminopeptidase, TOP, and TPPII had significantly higher ac-tivity in monocytes than in CD4+ T-cells (Figure 3). In proteasome-chymotrypsin-like activities, I did not find a significant difference, though other members of my lab performed a similar assay with 14 samples and found significant differences between CD4+ T-cells and monocytes (data not shown). Faster kinetics in peptidases associated with antigen processing in monocytes

In addition to the overall level of peptidase activity, the speed at which the peptidase cleaves proteins is crucial during viral infection since the virus depends on the slow recognition of an infected cell for its replication. We investigated whether observed variations in pep-tidase activity levels between CD4+ T-cells and monocytes occurred with kinetic differences. Therefore, I measured the kinetics of pepti-dase degradation for the proteasome’s caspase-like, chymotrypsin-like, and trypsin-like active sites and for aminopeptidase, TOP, and TPPII. Like in the previous experiment, whole cell extracts of PBMC and immunosorted CD4+ T-cell and monocytes from 8 healthy do-nors were incubated with fluorogenic substrates for 1 hour. A repre-

sentative sample shows the observed differences in slope over time (Figure 2). I used the average of the maximum slope from 3 experi-ments as my comparison values. Peptidases in monocytes have sig-nificantly faster degradation capacities than those in CD4+ T-cells (Figure 4). Again, the proteasome chymotrypsin-like subunit failed to show a statistically significant difference between the two subsets, but other members of the group observed differences when using a larger sample size (data not shown). These differences in antigen processing activities and kinetics imply that monocytes are capable of processing intracellular peptides to a greater extent than CD4+ T-cells, which may have downstream effects on the recognition of the infected cell.

Figure 4. Faster peptidase kinetics in monocytes. The degradation of fluorogenic substrates specific for either the proteasome: caspase-like or trypsin-like active sites, aminopeptidase, TOP, or TPPII were measured in whole cell extracts of PBMC (•), CD4+ T-cells (■), and monocytes (▶) over 1 hour at 37°C. The maximum slopes of degradation in CD4+ T-cells and monocytes were compared using non-parametric Mann-Whitney statisti-cal analysis. Monocytes showed significantly higher kinetics than CD4+ T-cells. *p<0.05, **p<0.01, ***p<0.001. n=8.

Figure 5. Intracellular epitope degradation and 51Cr release as-say. A) Intracellular peptidases in whole cell extracts of PBMC, CD4+ T-cells, and monocytes degrade 5RK3 to smaller products when incubated in 37°C over 2 hours. Peptides were eluted in RP-HPLC and produced unique peaks where the integral under the peak is proportional to the amount of that peptide. B) HLA-A3+ B cells were pulsed with peptide degradation products and used as targets in the 51Cr release assay. CTLs specific for the HLA-A3+ -restricted epitope RK9 (orange) were used as effectors.

Figure 6. Intracellular degradation of 5RK3 to produce RK9.Whole cell extracts of PBMC (+), CD4+ T-cell (■), and monocytes (◊) were incubated with 5RK3, an HIV-1 Gag p-17 fragment, at 37°C. At 2, 10, 30, 60, 120 minutes, samples were collected and eluted through RP-HPLC for analysis and quantification. A) Monocytes showed faster degradation of 5RK3 than CD4+ T-cells. B) Monocytes produced the HLA-A3 restricted optimal epitope, RK9, faster and at sustained levels for over 2 hours. Iden-tities of 5R3 and RK9 were confirmed by mass spectrometry.

Figure 7. Peptide products from intracellular 5RK3 digestion.Whole cell extracts of PBMC, CD4+ T-cell, and monocytes were used to digest an HIV-1 Gag p17 fragment (RWEKIRLRPGGKKKYKL aa 15-31), called 5RK3 (uppermost rectangle). It contains an HLA-A3 restricted epitope, RK9 (black text); N-terminal flanking regions are to the left; C-terminal flanking regions are to the right. Degradation products at 2 and 60 minutes that were found through mass spectrometry are listed. Prod-ucts with dotted outlines are unique to the indicated cell type. Shaded products are previously described antigenic epitopes. Monocytes showed shorter and more antigenic degradation products than CD4+ T-cells.


23www.thurj.org

Imm

unology

Intracellular degradation of longer peptides to produce optimal HIV-1 epitopes

After ascertaining significant differences in activity and kinetics of antigen processing peptidases between CD4+ T-cells and mono-cytes, we wanted to know if those differences affect HIV-1 epitope processing. To do this, I utilized an in vitro intracellular degradation system where I incubated whole cell extracts of PBMC, immunosort-ed CD4+ T-cells, or monocytes with a 17-amino-acid sequence from HIV-1 Gag p17 (RWEKIRLRPGGKKKYKL, aa 15-31), called 5RK3 in this experiment, for 0 to 120 minutes (Figure 5a). When intracellu-lar peptidases degrade 5RK3, one can visualize the peptide products through RP-HPLC where one peak indicates a distinct peptide, and the integral below the peak is proportional to the amount of peptide. At different time points during the incubation, one can examine the quantity of 5RK3 and its degradation products. Members of the lab previously showed this system to be a good approximation of the en-dogenous processing of intracellular proteins (Le Gall et al., 2007). Faster degradation of 5RK3 fragment and sustained levels of RK9, a peptide product, in monocytes

Using the in vitro intracellular degradation assay described in the earlier section, I measured the degradation of 5RK3 and the genera-tion of its optimal epitope RK9 (RLRPGGKKK), which requires a multistep cleavage of the 5 N-terminal residues (RWEKI) and 3 C-terminal residues (YKL) of 5RK3.

I studied the production of RK9 since it is associated with delayed progression to AIDS through HLA-A3 restricted CTL responses in some HIV-infected patients (Altfeld et al., 2006). Furthermore, I could infer that the observed differences in peptidase activity and kinetics affected 5RK3 degradation since this process requires the proteasome, aminopeptidases, and TPPII (Le Gall et al., 2007). I found that the degradation of 5RK3 was faster in monocytes than in CD4+ T-cells, and monocytes degraded all the 5RK3 by the second minute of the incubation (Figure 6a). As for the production of RK9, monocytes not only produced it with faster kinetics, but they also maintained higher levels of RK9 over time compared to CD4+ T-cells (Figure 6b). Antigenic degradation products from monocytes

Besides the degradation of the original peptide, and the produc-tion of RK9, we wanted to understand the differences among peptide products of the 2 PBMC subsets. No study had shown the potential differences in peptide products from the antigen processing machin-ery. Such a difference could eventually affect CTL recognition of the infected cell subset. The design of the intracellular degradation

assay allowed me to attribute any observed differences in the prod-ucts between CD4+ T-cells and monocytes to the antigen processing machinery. We sent the products for mass spectrometry analysis to identify the peptide sequences. I found that peptides produced from 5RK3 in monocytes were shorter than those from CD4+ T-cells, pre-sumably allowing for a better fit on MHC-I (Figure 7). In addition, monocyte degradation products contained already identified optimal HIV-1 epitopes: HLA-B27 restricted IK9 (IRLPPGGKK) and HLA-A3 restricted KK9 (KIRLPPGGK), RK9 (RLRPGGKKK), and RY10 (RLRPGGKKKY) (Korber et al., 2008; Fig. 7). This alluded to differ-ences in the CTL recognition and antigenicity of peptide products between the two subsets. My supervisor performed a 51Cr release as-say on the peptide products from the 5RK3 intracellular degradation assay to determine their antigenicity. She purified the small peptides from the incubated extracts and pulsed identical HLA-matched B cells with them (Figure 5b). Those B cells served as the targets in the release assay, and CTLs specific for HLA-A3 RK9 were the effectors. We found that products from monocytes were more antigenic than those of CD4+ T-cells (Figure 8). In fact, we achieved a maximum specific lysis of 51% from monocyte degradation products whereas those from CD4+ T-cells could not generate greater than 3% lysis. Other members of the lab observed similar results after replicating the assay twice (data not shown). Intracellular stability of HIV-1 epitopes

The previous experiments showed how differences in peptidase activity between CD4+ T-cells and monocytes could affect the pro-duction of HIV-1 epitopes. However, antigen processing is a balance between both the production and degradation of epitopes since intra-cellular peptidases are constitutively active. Differences in the deg-radation of optimal epitopes between CD4+ T-cells and monocytes could affect epitope presentation and CTL recognition of optimal epitopes. Little is known about the intracellular stability of short peptides, though aminopeptidases have been implicated in regularly cleaving short and intermediate peptides in the cytosol (Smyth and O’Cuinn, 1994). Thus, I examined whether the degradation of HIV-epitopes, which are short peptides, was the same in both CD4+ T-cells and monocytes. Other members of my lab had characterized the half-lives of many of the known HIV epitopes in PBMC and found that although some epitopes were highly unstable, others resisted degradation over time (data not shown). We speculated that HIV epitopes that were stable in PBMC could have varying levels of stabil-ity between CD4+ T-cells and monocytes since we already saw cell

Figure 8. Antigenicity of peptide products from intracellular 5RK3 digestion. Extracts from 2, 10, 30, 60, 120 minutes in the 5RK3 degradation in PBMC, CD4+ T-cells, and monocytes were purified. HLA-A3+ B cells were pulsed with these products to be used as targets in 51Cr release assay. CTLs specific for HLA-A3+ restricted epitope RK9 were used as effectors. For each extract from all time points during the degradation, percent specific lysis was measured. Monocyte degradation products (▶) elicited greater CTL killing, and thus, were more antigenic than that of CD4+ T-cells (■).

Figure 9. Intracellular degradation of optimal HIV-1 epitopes. A) 8 optimal HIV-1 epitopes (B57-YY9, Cw3-RL9, A33-ER8, B7-HA9, A3-KK11, B7-FL9, B7-TL10, and B7-FR10) were incubated in whole cell extract of PBMC and immunosorted CD4+ T-cells and monocytes at 37°C for 30 minutes. B) At 2, 10, and 30 minutes, portions of the reaction were stopped and eluted through RP-HPLC. Peptides produce unique peaks where the area under the peaks represents the amount of peptide in the eluate.

Imm

unol

ogy

type differences in epitope production. To measure the intracellular half-life of optimal HIV epitopes, I incubated them with whole cell extracts of PBMC, immunosorted CD4+ T-cells, or monocytes over 30 minutes (Figure 9). At 2, 10, and 30 minutes, I eluted the extract-peptide mixture through RP-HPLC. Like the prior RP-HPLC assay, each peptide produced a unique peak where its integral reflects the amount of peptide present in the eluate allowing me to quantify the

percent left of the original epitope. By examining the proportion of peptide remaining at each time-point, I could map the epitope degra-dation to a non-linear regression of exponential decay that provided me with parameters to calculate the half-life of each epitope. From the existing data in the group, I chose epitopes that were relatively stable overtime in PBMC, in order to measure any notable deviations in stability in CD4+ T-cells and monocytes. Using 8 epitopes (B57-YY9, Cw3-RL9, A33-ER8, B7-HA9, A3-KK11, B7-FL9, B7-TL10, and B7-FR10), I found that the half-life of those optimal HIV-1 epitopes was significantly shorter in monocytes than in CD4+ T-cells (Figure 10). To ensure that my results were replicable, I repeated each degra-dation assay 3 times and averaged all values. Different intracellular degradation of HIV-1 optimal epitopes in CD4+ T-cells and monocytes

The calculated differences in half-life of HIV optimal epitopes were not the only interesting findings revealed in our comparison of intracellular epitope degradation between CD4+ T-cells and mono-cytes. Apart from understanding the overall pattern of optimal epitope degradation in PBMC, we wanted to know if the kinetics of degradation was similar across the PBMC subsets that HIV-1 infects. In the same assay as above, I qualitatively assessed the kinetics of deg-radation for each optimal epitope (B57-YY9, Cw3-RL9, A33-ER8, B7-HA9, A3-KK11, B7-FL9, B7-TL10, and B7-FR10) between the subsets. I found that there was no consistent pattern of degradation among epitopes in both cell subsets, and these differences existed regardless of the epitopes protein’s origin (Figure 11). This suggested functional variations in subset-specific degradation of HIV epitopes.

Discussion

HIV-1 infection depends on CD4+ T-cells and monocytes as vessels for proliferation and latency (Pierson et al., 2000). CD4+ T-cells undergo a more severe depletion than mono-cytes during infection (Pedersen et al., 1989). To delay the onset of immunodeficiency following HIV infection, the host must mount an HIV-specific immune response. One study showed that the presence of CTLs that recognized the HIV-1 envelope glycoprotein allowed the patient to control viral levels during primary infection since CTL recognition of infected cells can stimulate antiviral responses (Borrow et al., 1994). This recognition event requires the intracellu-lar degradation of viral proteins and their assembly on an MHC-I for cell surface presentation.

Although CD4+ T-cells and monocytes are critical in HIV infection, antigen processing and presentation of HIV-1 epitopes have not been studied in these two cell types. More-over, the type and efficacy of the immune response that those infected cell subsets can induce is unknown. Therefore, this study aimed to understand the antigen processing machin-ery in CD4+ T-cells and monocytes in the context of HIV-1 epitope production, degradation, and antigenicity. I focused

Figure 10. Shorter half-life of HIV-1 epitopes in monocytes. 8 optimal HIV-1 epitopes (B57-YY9, Cw3-RL9, A33-ER8, B7-HA9, A3-KK11, B7-FL9, B7-TL10, and B7-FR10) were subjected to intracellular degradation in whole cell extracts of PBMC (circle) and immunosorted CD4+ T-cells (square) and monocytes (triangle). Peptide products at 2, 10 and 30 minutes were analyzed and quantified using RP-HPLC. One-phase exponential decay was used to estimate the half-lives of the epitopes. Non-parametric Mann-Whitney test was used for comparison of the half-lives. Optimal epitopes showed significantly shorter half-lives in monocytes than in CD4+ T-cells. **p<0.01.

Figure 11. Various degradation kinetics of HIV-1 op-timal epitopes in CD4+ T-cells and monocytes. Whole cell extracts of PBMC (+), immunosorted CD4+ T-cells (■), and monocytes (◊) digested various HIV-1 optimal epitopes (B57-YY9, Cw3-RL9, A33-ER8, B7-HA9, A3-KK11, B7-TL10, and B7-FR10) over 30 minutes at 37°C. RP-HPLC was used to analyze the extract-peptide mixture to determine the amount of peptide left. Extensive variation in degradation kinetics was seen among epitopes from the same protein as well as across all epitopes.


25www.thurj.org

Imm

unology

on identifying the levels and kinetics of peptidases associated with the production and degradation of these peptide fragments and found that there were indeed differences in activity level and kinetics, which resulted in variations in the processing of HIV-1 epitopes. Observed heterogeneity in the identity and production time course of peptide products might have important consequences for CTL recognition of infected PBMC subsets and the ensuing immune response.

Using the intracellular fluorescence-based hydrolytic assay for peptidase activity, I showed how proteasome-caspase-like and trypsin-like, aminopeptidase, TOP, and TPPII activity levels and speed in CD4+ T-cells and monocytes differ for both healthy and HIV infected donors. We do not know whether these differences stem from variations in transcription or translation. Moreover, we do not understand what differences in intracellular peptidase quantity lead to observable differences in activity. Therefore, I cannot know if variations in activity come from increased expression of the proteins or from increased activity of each individual peptidase in mono-cytes. Another potential reason for the increase in peptidase activity in monocytes might be a contamination in the cytosol of cathepsins, endosomal peptidases that function in the exogenous pathway of an-tigen presentation. I corrected for this by stabilizing the pH of the whole cell extracts at 7.4, which considerably inhibits the function of cathepsins (Turk et al., 1999). Despite the uncertainty in the source of these functional differences between CD4+ T-cells and monocytes, the effects of these variations are of primary interest because we must fully understand the consequences for the immune response.

One limitation of this study was that it did not measure differ-ences in non-cytosolic peptidases, such as ERAP-1 and ERAP-2, be-tween the two PBMC subsets. Much of the literature about antigen processing of epitopes implicates ERAPs and not cytosolic peptidases (York et al., 2002). York and his group showed that purified ERAP-1 could trim all peptides that were longer than 10 residues and half of the 9-residue peptides. Though their experiment was not performed under the typical intracellular environment of the cell, it suggested that ERAP-1 was important in determining the loaded peptides for MHC-I presentation. My data on intracellular epitope processing does not conflict with the existing research, but it adds to the litera-ture by showing that cytosolic peptidases can produce optimal HIV-1 epitopes from longer peptides. However, it is still important to study the potential differences in ERAP activity between the PBMC subsets as those peptidases can make the final peptide trimmings in the ER that may affect its loading on the MHC-I and CTL recognition.

Even with the exclusion of non-cytosolic peptidases, my study revealed differences among intracellular peptidases. For example, aminopeptidase showed the greatest fold differences in average activ-ity values between CD4+ T-cells and monocytes. Compared to the proteasome, TOP, and TPPII, whose fold differences in the average activities between the subsets fell between 1.5 and 2.1, aminopepti-dase activity in monocytes was about 3.5 times greater than that of CD4+ T-cells. This suggests that each peptidase can have unique dif-ferences in activity in different subsets. Also supporting this theory is the inconclusive result on the subset-specific differences in the pro-teasome-chymotrypsin active site. Although average activity levels of this active site were greater in monocytes, there was no statistically significant difference in values between the two cell types (data not shown). Other active sites within the proteasome showed significant subset-specific differences with 8 samples in the cohort, indicating non-uniform subset-specific variations in the proteasome active sites and perhaps, in most intracellular peptidases.

In addition to studying the differences in intracellular peptidase activity between the CD4+ T-cells and monocytes, it is important to study the entire antigen processing pathway because the presentation

of the epitope is dependent on a multi-step process where each step can affect the identity of the final presented product. This process incorporates the kinetics of peptide production, degradation, trans-port, and MHC-I binding, which must be completed in a timely man-ner before the proliferating virions can escape. Here, I studied one of the earlier steps, intracellular epitope processing, which did not include epitope processing in the ER, peptide transport, or MHC-I binding kinetics. Products from intracellular processing need trans-port into the ER by TAP, which displays binding preferences for cer-tain peptide sequences to transport into the ER (Procko and Gaudet, 2009). Thus, it is possible to infer that TAP might also influence the final product presented by MHC-I. Therefore, further experiments to discern the differences in the entire antigen processing pathway between CD4+ T-cells and monocytes must be completed.

Moreover, my research only focused on 2 subsets that HIV infects: CD4+ T-cells and monocytes, and did not include the entire repertoire of HIV-infectible PBMC subsets, such as macrophages, natural killer (NK) cells, and dendritic cells (DCs), which all contribute to infec-tion (Weiss, 2008; Valentin et al., 2002). Macrophages are monocyte-derived cells that can store dormant virus in the gastrointestinal tract during chronic infection (Smith et al., 2003). Reduced levels of NK cells in infected patients are associated with worse prognoses (Ullum et al., 1995). DCs activate CTLs in cell-mediated immunity, making them especially important in HIV pathogenesis (Zarling et al., 1999). Although most of the infected cells in HIV-individuals are CD4+ T-cells, it is still important to understand the antigen processing capa-bilities of all PBMC subsets that contribute to this disease in order to understand the entire scope of presented HIV-1 peptides.

Apart from exposing peptidase activity differences between the subsets, my results introduce the concept of a balance between pro-duction and degradation of intracellular peptides within the infected cell. The observations that monocytes both produce and degrade optimal HIV-1 epitopes faster than CD4+ T-cells can be seemingly contradictory if one quickly attributes faster degradation of epitopes as a negative factor in antigen processing. However, one must remain aware of the balance between production, degradation, and transport of epitopes in the cytosol compartment. If monocytes have a more efficient transport system into the ER, then the increased degrada-tion of optimal epitopes in monocytes might be negligible. Similarly, slower degradation of HIV-1 epitopes in CD4+ T-cells might allot more time for TAP binding and transport into the ER. With my re-sults, I have reiterated the importance of kinetics in the entire antigen processing pathway and established the value of a balance between intracellular production and degradation of viral peptides destined to be presented to CTLs.

Additionally, I have shown that these observations of differential epitope processing might also be applicable to HIV-infected individ-uals. Though I did not perform the intracellular degradation of an HIV epitope on whole cell extracts from HIV-infected donors, the significant difference in peptidase activities between CD4+ T-cells and monocytes of HIV-infected samples suggests similarities in epitope production (Sup. Fig. C and D). However, the estimated frequency of infected CD4+ T-cells and monocytes waivers between 0.01% and 1% in an HIV-positive individual; therefore, the differences in peptidase activity between the subsets that I detected might stem from the un-infected cells (Poznansky et al., 1991; Smith et al., 2003). Although HIV epitope processing in infected cells remains unclear, a larger study that compares intracellular peptidase activity and epitope pro-cessing in HIV-infected PBMC will elucidate unknown patterns.

Furthermore, since HIV-1 has two main tropisms for PBMC cell subsets—T-cells and monocyte-derived cells—one can speculate that the untimely recognition of infected CD4+ T-cells may skew the viral



Imm

unol

ogy

population in an infected patient to the T-cell tropic strains. Recently, the prevalence of HIV-1 viral strains that preferentially infect CD4+ T-cells was found to be significantly greater in PBMC (Verhofstede et al., 2009). With increased viral levels of T-cell-tropic strains, greater numbers of CD4+ T-cells might be infected, thus contributing to the drastic reduction of CD4+ T-cells during HIV infection. Already, the period following infection before the CTL response has been identi-fied as crucial for the virus to establish reservoirs in T-cells (Daven-port et al., 2004). This study complements my findings to suggest that kinetics of antigen processing and presentation and CTL recognition are important for disease progression. Though it might be imprecise to generalize laboratory findings to the disease progression, my find-ings provide a different way of examining the dissimilar fates of CD4+ T-cells and monocytes during HIV infection.

This research is applicable to other viral infections, especially ones with tropism towards specific cell subsets. For the cell-mediated im-mune response to be effective, the infected cell must present a viral peptide that DC-primed CTLs can recognize. Since uninfected DCs can undergo cross-presentation, where extracellular peptides enter the endogenous pathway, it is important to ensure that the peptides presented by the infected cells are the same ones that the DCs can process for CTL priming. For example, one study on the DC-tropic human cytomegalovirus (HCMV) strain found that whole PBMC in-fected with HCMV could elicit more responses than infected DCs (Gerna et al., 2005). Here, differences in viral peptide presented might be the cause for differences in the elicited immune response. Further research on cell subsets infected by viruses with particular tropisms must be completed to understand the extent of cell-type-specific dif-ferences in antigen processing.

My results also pertain to the current literature on immunodomi-nance of MHC-I restricted epitopes. For this purpose, I will define MHC-I immunodominance as the phenomenon where CTLs against a certain viral epitope are more prevalent in an infected individu-al. Mathematical models based on patient observations show that a dominant CTL response usually arises when a pathogen is homoge-neous; however, this does not explain the molecular reasons for over-representation of that particular viral peptide (Nowak et al., 1995). A recent study on intracellular processing found that flanking sequenc-es near the epitope contributed to the processing of those epitopes and strongly determined the intracellular peptide products (Le Gall et al., 2007). If cell subsets exhibit different kinetics in the production and identity of presented HIV-1 epitopes, then the immunodominant epitope might not be presented by one of the subsets. This would cre-ate dire consequences for the PBMC subset that does not present this dominant epitope, and clearance of infected cells in that subset might be compromised.

Besides understanding the immune response to infect CD4+ T-cells or monocytes, my findings also add to the existing literature on HIV vaccines. Unfortunately, the search for an HIV vaccine has not been very successful due to the complexity of HIV and its interac-tions with the immune system (Johnston and Fauci, 2007). Vaccines function by delivering antigen to a host; the antigen activates the im-mune system and induces the creation of antibody responses and an-tigen-specific memory T-cells, which persist in the host. Upon infec-tion with that same pathogen, infected cells present foreign peptides that optimally trigger proliferation of those memory immune cells to perform their effector functions to curb the disease. An efficacious vaccine necessitates memory T-cells recognizing the viral peptides presented by infected cells. Since HIV infects various cell types that potentially have differences in their antigen processing machinery, a great candidate for a vaccine should contain a viral peptide that all cell types can process efficiently and present to CTLs. Thus, the

consequences of subset-specific differences in antigen processing and presentation are broad reaching. The identity of the peptide present-ed might affect memory T-cell activation and CTL recognition of the infected cell.

Even without observations on other components of the antigen processing pathway or on other cells infected by HIV, my results dem-onstrate that differences in intracellular processing might affect the production, degradation, identity, and antigenicity of HIV-1 epitopes in two PBMC subsets. Monocytes showed increased activity in anti-gen processing peptidases, which might lead to a different repertoire of presented HIV-1 epitopes than CD4+ T-cells. The rapid production and larger amounts of more antigenic peptides in monocytes may increase the probability of transport into the ER for those peptides. If infected monocytes present more antigenic epitopes, then these cells might elicit a better immune response than infected CD4+ T-cells. Accordingly, it is important to further research on subset-specific differences in antigen processing not only to better understand the host’s immune response to infection but also to develop highly effec-tive clinical interventions, such as vaccines.

ReferencesAltfeld, M., Kalife, E., Qi, Y., Streeck, H., Lichterfeld, M., Johnston, M. 1. et al. (2006). HLA Alleles Associated with Delayed Progression to AIDS Contribute Strongly to the Initial CD8+ T-cell Response against HIV-1. PLoS Medicine, 1851-1864.Betts, M., Ambrozak, D., Douek, D., Bonhoeffer, S., Brenchley, J., Casazza, 2. J. et al. (2001). Analysis of Total Human Immunodeficiency Virus (HIV)-Specific CD4+ and CD8+ T-Cell Responses: Relationship to Viral Load in Untreated HIV Infection. Journal of Virology, 11983-11991.Borrow, P., Lewicki, H., Hahn, B., Shaw, G., and Oldstone, M. (1994). Vi-3. rus-specific CD8+ cytotoxic T-lymphocyte activity associated with control of viremia in primary human immunodeficiency virus type 1 infection. Journal of Virology, 6103-10.Butz, E., and Bevan, M. (1998). Differential Presentation of the Same 4. MHC Class I Epitopes by Fibroblasts and Dendritic Cells. The Journal of Immunology, 2139-2144.Callaway, D., Ribeiro, R., and Nowak, M. (1999). Virus phenotype switch-5. ing and disease progression in HIV-1 infection. Proceedings of the Royal Society of London, 266, 2523-2530.Chun, T. W., and Fauci, A. S. (1999). Latent reservoirs of HIV: obstacles 6. to the eradication of virus. Proceedings of the National Academy of Sci-ences, 10958-61.Chun, T. W., Engel, D., Berrey, M. M., Shea, T., Corey, L., and Fauci, A. 7. S. (1998). Early establishment of a pool of latently infected, resting CD4+ T-cells during primary HIV-1 infection. The Proceedings of the National Academy of Sciences, 8869-73.Crowe, S., Turner, S., Miller, S., Roberts, A., Rappolo, A., Doherty, P. et al. 8. (2003). Differential antigen presentation regulates the changing patterns of CD8+ T-cell immunodominance in primary and secondary influenza virus infections. The Journal of Experimental Medicine, 399-410.Davenport, M., Riberio, R., and Perelson, A. (2004). Kinetics of virus-9. specific CD8+ T-cells and the control of human immunodeficiency virus infection. Journal of Virology, 10096-103.Douek, D. (2007). HIV disease progression: immune activation, microbes, 10. and a leaky gut. Topics in HIV Medicine, 114-7.Ellery, P., Tippett, E., Chiu, Y.-L., Paukovics, G., Cameron, P., Solomon, 11. A. et al. (2007). The CD16+ monocyte subset is more permissive to infec-tion and preferentially harbors HIV-1 in vivo. The Journal of Immunol-ogy, 6581-6589.Frahm, N., Yusim, K., Suscovich, T. J., Adams, S., Sidney, J., Hraber, P. 12. et al. (2007). Extensive HLA class I allele promiscuity among viral CTL epitopes. The European Journal of Immunology, 2419-33.Gerna, G., Percivalle, E., Lilleri, D., Lozza, L., Chiara, F., Hahn, G. et al. 13. (2005). Dendritic-cell infection by human cytomegalovirus is restricted to strains carrying functional UL131-128 genes and mediates efficient viral antigen presentation to CD8+ T-cells. Journal of General Virology, 275-284.Groothuis, T., Griekspoor, A., Neijssen, J., Herberts, C., and Neefijes, J. 14. (2005). MHC class I alleles and their exploration of the antigen-process-

ing machinery. Immunological Reviews, 60-76.Hattori, A., and Tsujimoto, M. (2004). Processing of antigenic peptides by 15. aminopeptidase . Biological and Pharmaceutical Bulletin, 777-80.Hauber, J., Perkins, A., Heimer, E. P., and Cullen, B. R. (1987). Trans-acti-16. vation of human immunodeficiency virus gene expression is mediated by nuclear events. Proceedings of National Academy of Sciences, 6364-8.Hel, Z., McGhee, J., and Mestecky, J. (2006). HIV infection: first battle 17. decides the war. Trends in Immunology, 274-281.Johnston, M., and Fauci, A. (2007). An HIV Vaccine – Evolving Concepts. 18. The New England Journal of Medicine, 2073-81.Kopp, F., Hendil, K., Dahlmann, B., Kristensen, P., Sobek, A., and 19. Uerkvitz, W. (1997). Subunit arrangement in the human 20S proteasome. Proceedings of the National Academy of Sciences, 2939-2944.Korber, B., Brander, C., Haynes, B., Koup, R., Moore, J., Walker, B. et al. 20. (2008). HIV Molecular Immunology 2006/2007. Los Alamos, New Mexi-co: Los Alamos National Laboratory, Theoretical Biology and Biophysics.Larsen, M. V., Lundegaard, C., Lamberth, K., Buus, S., Brunak, S., Lund, 21. O. et al. (2005). An integrative approach to CTL eptitope prediction: a combined algorithm integrating MHC class I binding, TAP transport ef-ficiency, and proteasomal cleavage predictions. The European Journal of Immunology, 2295-303.Lazaro, E., Godfrey, S. B., Stamegna, P., Ogbechie, T., Kerigan, C., Zhang, 22. M. et al. (2009). Differential HIV Epitope Processing in Monocytes and CD4 T Cells Affects Cytotoxic T Lymphocyte Recognition. Journal of In-fectious Diseases, 236-43.Le Gall, S., Stamegna, P., and Walker, B. (2007). Portable flanking se-23. quences modulate CTL eptiope processing. Journal of Clinical Investiga-tion, 3563-3575.McLaren, P., Mayne, M., Rosser, S., Moffatt, T., Becker, K., Plummer, F. et 24. al. (2004). Antigen-Specific Gene Expression Profiles of Peripheral Blood Mononuclear Cells Do Not Reflect Those of T-Lymphocyte Subsets. Clini-cal and Diagnostic Laboratory Immunology, 977-982.Nowak, M., May, R., Phillips, R., Rowland-Jones, S., Lalloo, D., McAdam, 25. S. et al. (1995). Antigenic oscillations and shifting immunodominance in HIV-1 infections. Nature, 606-11.Oliveira, V., Campos, M., Melo, R., Ferro, E., Camargo, A., Juliano, M. et 26. al. (2001). Substrate specificity characterization of recombinant metallo oligo-peptidases thimet oligopeptidase and neurolysin. Biochemistry, 4417-4425.Pajot, A., Schnuriger, A., Moris, A., Rodallec, A., Ojcius, D., Autran, B. et 27. al. (2007). The Th1 immune response against HIV-1 Gag p24-derived pep-tides in mice expressing HLA-A02.01 and HLA-DR1. European Journal of Immunology, 2635-2644.Paroli, M., Propato, A., Accapezzato, D., Francavilla, V., Schiaffella, E., 28. and Barnaba, V. (2001). The immunology of HIV-infected long-term non-progressors---a current view. Immunology Letters, 127-129.Pedersen, C., Lindhart, B., Jensen, B., Lauritzen, E., Gerstoft, J., Dickmeiss, 29. E. et al. (1989). Clinical couse of primary HIV infection: consequences for subsequent course of infection. British Medical Journal, 154-7.Pierson, T., McArthur, J., and Siliciano, R. (2000). Reservoirs for HIV-1: 30. Mechanisms for Viral Persistence in the Presence of Antiviral Immune Responses and Antiretroviral Therapy. Annual Review of Immunology, 665-708.Poznansky, M., Walker, B., Haseltine, W., Sodroski, J., and Langhoff, E. 31. (1991). A rapid method for quantitating the frequency of peripheral blood cells containing HIV-1 DNA. Journal of Acquired Immune Deficiency Syndrome, 368-73.Procko, E., and Gaudet, R. (2009). Antigen processing and presentation: 32. TAPping into ABC transporters. Current Opinion in Immunology, E-publishing.Randolph, G., Jakubzick, C., and Qu, C. (2008). Antigen presentation by 33. monocytes and monocyte-derived cells. Current Opinion in Immunology, 52-60.Saric, T., Graef, C., and Goldberg, A. (2004). Pathway for Degradation 34. of Peptides Generated by Proteasomes. Journal of Biological Chemistry, 46723-46732.

35. Schmitz, J., Veazey, R., Kuroda, M., Levy, D., Seth, A., Mansfield, K. et al. (2001). Simian immunodeficiency virus (SIV)-specific cytotoxic T-lym-phocytes in gastrointestinal tissues of chronically SIV-infected rhesus monkeys. Blood Journal, 3757-3761.Schubert, U., Anton, L. C., Gibbs, J., Norbury, C. C., Yewdell, J. W., and 36. Bennink, J. R. (2000). Rapid degradation of a large fraction of newly syn-thesized proteins by proteasomes. Nature, 770-4.Seifert, U., Maranon, C., Shmueli, A., Desoutter, J.-F., Wesoloski, L., 37.


27www.thurj.org

Imm

unology

ing machinery. Immunological Reviews, 60-76.Hattori, A., and Tsujimoto, M. (2004). Processing of antigenic peptides by 15. aminopeptidase . Biological and Pharmaceutical Bulletin, 777-80.Hauber, J., Perkins, A., Heimer, E. P., and Cullen, B. R. (1987). Trans-acti-16. vation of human immunodeficiency virus gene expression is mediated by nuclear events. Proceedings of National Academy of Sciences, 6364-8.Hel, Z., McGhee, J., and Mestecky, J. (2006). HIV infection: first battle 17. decides the war. Trends in Immunology, 274-281.Johnston, M., and Fauci, A. (2007). An HIV Vaccine – Evolving Concepts. 18. The New England Journal of Medicine, 2073-81.Kopp, F., Hendil, K., Dahlmann, B., Kristensen, P., Sobek, A., and 19. Uerkvitz, W. (1997). Subunit arrangement in the human 20S proteasome. Proceedings of the National Academy of Sciences, 2939-2944.Korber, B., Brander, C., Haynes, B., Koup, R., Moore, J., Walker, B. et al. 20. (2008). HIV Molecular Immunology 2006/2007. Los Alamos, New Mexi-co: Los Alamos National Laboratory, Theoretical Biology and Biophysics.Larsen, M. V., Lundegaard, C., Lamberth, K., Buus, S., Brunak, S., Lund, 21. O. et al. (2005). An integrative approach to CTL eptitope prediction: a combined algorithm integrating MHC class I binding, TAP transport ef-ficiency, and proteasomal cleavage predictions. The European Journal of Immunology, 2295-303.Lazaro, E., Godfrey, S. B., Stamegna, P., Ogbechie, T., Kerigan, C., Zhang, 22. M. et al. (2009). Differential HIV Epitope Processing in Monocytes and CD4 T Cells Affects Cytotoxic T Lymphocyte Recognition. Journal of In-fectious Diseases, 236-43.Le Gall, S., Stamegna, P., and Walker, B. (2007). Portable flanking se-23. quences modulate CTL eptiope processing. Journal of Clinical Investiga-tion, 3563-3575.McLaren, P., Mayne, M., Rosser, S., Moffatt, T., Becker, K., Plummer, F. et 24. al. (2004). Antigen-Specific Gene Expression Profiles of Peripheral Blood Mononuclear Cells Do Not Reflect Those of T-Lymphocyte Subsets. Clini-cal and Diagnostic Laboratory Immunology, 977-982.Nowak, M., May, R., Phillips, R., Rowland-Jones, S., Lalloo, D., McAdam, 25. S. et al. (1995). Antigenic oscillations and shifting immunodominance in HIV-1 infections. Nature, 606-11.Oliveira, V., Campos, M., Melo, R., Ferro, E., Camargo, A., Juliano, M. et 26. al. (2001). Substrate specificity characterization of recombinant metallo oligo-peptidases thimet oligopeptidase and neurolysin. Biochemistry, 4417-4425.Pajot, A., Schnuriger, A., Moris, A., Rodallec, A., Ojcius, D., Autran, B. et 27. al. (2007). The Th1 immune response against HIV-1 Gag p24-derived pep-tides in mice expressing HLA-A02.01 and HLA-DR1. European Journal of Immunology, 2635-2644.Paroli, M., Propato, A., Accapezzato, D., Francavilla, V., Schiaffella, E., 28. and Barnaba, V. (2001). The immunology of HIV-infected long-term non-progressors---a current view. Immunology Letters, 127-129.Pedersen, C., Lindhart, B., Jensen, B., Lauritzen, E., Gerstoft, J., Dickmeiss, 29. E. et al. (1989). Clinical couse of primary HIV infection: consequences for subsequent course of infection. British Medical Journal, 154-7.Pierson, T., McArthur, J., and Siliciano, R. (2000). Reservoirs for HIV-1: 30. Mechanisms for Viral Persistence in the Presence of Antiviral Immune Responses and Antiretroviral Therapy. Annual Review of Immunology, 665-708.Poznansky, M., Walker, B., Haseltine, W., Sodroski, J., and Langhoff, E. 31. (1991). A rapid method for quantitating the frequency of peripheral blood cells containing HIV-1 DNA. Journal of Acquired Immune Deficiency Syndrome, 368-73.Procko, E., and Gaudet, R. (2009). Antigen processing and presentation: 32. TAPping into ABC transporters. Current Opinion in Immunology, E-publishing.Randolph, G., Jakubzick, C., and Qu, C. (2008). Antigen presentation by 33. monocytes and monocyte-derived cells. Current Opinion in Immunology, 52-60.Saric, T., Graef, C., and Goldberg, A. (2004). Pathway for Degradation 34. of Peptides Generated by Proteasomes. Journal of Biological Chemistry, 46723-46732.

35. Schmitz, J., Veazey, R., Kuroda, M., Levy, D., Seth, A., Mansfield, K. et al. (2001). Simian immunodeficiency virus (SIV)-specific cytotoxic T-lym-phocytes in gastrointestinal tissues of chronically SIV-infected rhesus monkeys. Blood Journal, 3757-3761.Schubert, U., Anton, L. C., Gibbs, J., Norbury, C. C., Yewdell, J. W., and 36. Bennink, J. R. (2000). Rapid degradation of a large fraction of newly syn-thesized proteins by proteasomes. Nature, 770-4.Seifert, U., Maranon, C., Shmueli, A., Desoutter, J.-F., Wesoloski, L., 37.

Janek, K. et al. (2003). An essential role for tripeptidyl peptidase in the generation of an MHC class I epitope. Nature Immunology, 375-379.Sheppard, H., Lang, W., Ascher, M., Vittinghoff, E., and Winkelstein, W. 38. (1993). The characterization of non-progressors: long-term HIV-1 infec-tion with stable CD4+ T-cell levels. AIDS, 1159-1166.Smith, P., Meng, G., Salazar-Gonzalez, J., and Shaw, G. (2003). Mac-39. rophage HIV-1 infection and the gastrointestinal tract reservoir. Journal of Leukocyte Biology, 642-649.Smyth, M., and O’Cuinn, G. (1994). Alanine aminopeptidase of guinea-40. pig brain: a broad specificity cytoplasmic enzyme capable of hydrolysing short and intermediate length peptides. International Journal of Bio-chemistry, 1287-97.Stevenson, M., Stanwick, T. L., Dempsey, M. P., and Lamonica, C. A. 41. (1990). HIV-1 replication is controlled at the level of T-cell activation and proviral integration. The EMBO Journal, 1551-60.Towne, C., York, I., Neijssen, J., Karow, M., Murphy, A., Valenzuela, D. et 42. al. (2005). Leucine aminopeptidase is not essential for trimming peptides in the cytosol or generating epitopes for MHC class I antigen presentation. Journal of Immunology, 6605-14.Turk, B., Dolenc, I., Lenarcic, B., Krizaj, I., Turk, V., Bieth, J. et al. (1999). 43. Acidic pH as a physiological regulator of human cathepsin L activity. Eu-ropean Journal of Biochemistry, 926-32.Ullum, H., Gotzsche, P., Victor, J., Dickmeiss, E., Skinhoj, P., and Peder-44. son, B. K. (1995). Defective Natural Immunity: An Early Manifestation of Human Immunodeficiency Virus Infection. Journal of Experimental Medicine, 789-799.Valentin, A., Rosati, M., Patenaude, D., Hatzakis, A., Kostrikis, L., Laza-45. nas, M. et al. (2002). Persistent HIV-1 infection of natural killer cells in patients receiving highly active antiretroviral therapy. Proceedings of the National Academy of Sciences, 7015-7020.Vanhems, P., and Beaulieu, R. (1997). Primary infection by type 1 hu-46. man immunodeficiency virus: diagnosis and prognosis. Postgrad Medical Journal, 403-408.Verhofstede, C., Vanderkerckhove, L., Eygen, V., Demecheleer, E., Van-47. denbroucke, I., Winters, B. et al. (2009). CXCR4-using HIV type I variants are more commonly found in peripheral blood mononuclear cell DNA than in plasma RNA. Journal of Acquired Immune Deficiency Syndrome, 126-36.Weiss, R. A. (2008). Special Anniversary Review: Twenty-five years of hu-48. man immunodeficiency virus research: success and challenges. Clinical and Experimental Immunology, 201-210.York, I., Chang, S.-C., Saric, T., Keys, J., Favreau, J., Goldberg, A. et al. 49. (2002). The ER aminopeptidase ERAP I enhances or limits antigen presen-tation by trimming epitopes to 8-9 residues. Nature Immunology, 1177-1184.Zarling, A., Johnson, J., Hoffman, R., and Lee, D. (1999). Induction of pri-50. mary human CD8+ T lymphocyte responses in vitro using dendritic cells. Journal of Immunology, 5197-204.

AcknowledgementsI would like to thank Sylvie Le Gall for giving me the opportunity to conduct

research and for being a great mentor to me. Her dedication to the scientific pro-cess is a quality that I will always cherish. Furthermore, my experiments would have been impossible without the amazing supervision by Estibaliz (Esti) Lazaro. She helped me grow as both a researcher and as a young adult. Furthermore, I truly appreciate working in a world-class laboratory whose director, Bruce Walk-er, gave me encouragement and support during my time there. This project was supported in part by Howard Hughes Medical Institute (HHMI), the National Institutes of Health (NIH), and Bill and Melinda Gates Foundation.

I would also like to individually thank all members of the Le Gall lab, past and present: Sasha Blue Godfrey, Jeremy Ho, Christopher Kerrigan, Mei Zhang, Ser-gio Martinez, Paul Bourgine, Shao Chong Zhang, and Jonathan Chow for sharing their projects with me, assisting me with portions of my research, and staying positive in times of stress. Professor Losick’s HHMI program also supported me immensely in my research.

To end, my family and friends have encouraged me and kept me sane through-out this project. I cannot imagine completing this process without them. They always believed and still believe in my commitment to understanding and com-bating this destructive virus.



Econ

omic

s

Victory by association:Using electronic prediction markets to

measure coattail effectsStephen Travis May


In this paper, we study the magnitude of coattail effects in the 2008 election, or the impact of the presidential elec-tion on congressional elections. We utilize data from electronic prediction markets to measure these effects. In order to eliminate endogeneity biases in our analysis, we measure coattails by using the occurrence of a major event as an instrument. We select days in which major events occurred (such as presidential debates) that affected the presidential election without any direct effect on local elections. Since the shifts that occur on these days in congressional markets are due exclusively to coattails, we then measure the impact the exogenous events had on the congressional elections. We find strong coattail effects in House elections and insignificant effects in the overall Senate race. We then apply our methodology to the Minnesota Senate election, where we find strikingly strong coattail effects. We discuss these results in the context of our simple model of voting preferences.

IntroductionAs the cornerstone of democracy, elections play a critical role in

the United States political system. Every four years, star candidates rise and fall, scandals emerge and disappear, and long campaigns are fiercely fought before settling on a winner. Pundits devise dozens of stories that “explain” what determined the winner, and television and radio talk shows will debate these theories for months. But to the academic empiricist, understanding an election is problematic. To understand the determinants of a specific election, the sample size is only one event—so how can we know whether John McCain would have fared better with a different vice-presidential candidate? To un-derstand election determinants, political economists like Ray Fair have pooled the data of many elections and constructed models of very broad trends; yet because of the small sample size and the com-plexity of the variables impacting elections, these models yield only a few answers about what shapes elections in general, and much less about what shapes specific elections.

In this paper, we propose the use of a new tool for understand-ing elections: electronic prediction markets. These markets have been studied at length, with a broad theoretical and empirical consensus that they yield reasonably unbiased, efficient estimates of the prob-ability that an event occurs. Instead of focusing on their efficiency, as most papers about electronic prediction markets have, we will in-stead assume that they are efficient. In this paper, we will begin to an-swer the following question: given that electronic prediction markets are efficient, what can they tell us about political institutions? We will provide a case study of the potential utility of this methodology in the example of coattail effects. While this paper yields interesting results, our focus is primarily methodological and aims to understand how the existence of a continuous metric of election probabilities can be useful for causal analysis. Background on electronic prediction markets

Electronic prediction markets are futures exchanges in which the value of an asset is tied to the outcome of a particular event. In a vote share market, traders bid in a continuous double auction for the

optimal price that they will pay for a contract that pays $1 for every percentage point that a particular party receives. For example, in the 2000 election, George Bush received 47.9% of the vote, and thus the Republican contract paid off $47.90 at the end of the election. In a winner-take-all (WTA) election, traders bid on contracts for which candidate will win the election and receive $100 per share of the win-ning candidate held.

In a system with no transaction costs and risk neutrality, Wolfers and Zitzwitz (2004) show that the market price of the contract should equal the median market participant’s expected election outcome weighted by trading volume. Furthermore, Berg et al. (2001) show that the market price is a remarkably accurate short-term predictor of the real election outcome, consistently outperforming polls as a last-day prediction tool. In a follow-up study, Berg et al. (2003) find that electronic markets have exceptional long-term predictive abili-ties, and greatly outperform both polls and forecasting models.

The intuition behind these results is clear: traders are forced to “put their money where their mouth is,” and they are thus incentiv-ized to be accurate. Since projection models and poll results are gen-erally publically available, that information can be incorporated into the market price. Thus the market prediction should be at least as accurate as any type of publicly-known prediction mechanism, if not more so. Interestingly, this assumption does not depend on traders being a random sample of the population, as the market can incorpo-rate information without all voters participating. As Berg et al. (2001) point out, market participants are far from a random sample; they tend to be wealthy, young, and highly-educated. Furthermore, the av-erage trader in the market can be biased and trade based on his own bias; empirically, there is a small set of “marginal traders” who are the ultimate price-makers and hunt for arbitrage opportunities—and their probabilistic assumptions are quite accurate.

With this consensus of research demonstrating that prediction markets are accurate estimators, we intend to utilize the market re-sults in this paper as a continuous set of election results —changing over the course of the election only as the expected outcome changes. Thus, we have a much more complete data set, with a continuous set

29www.thurj.org

Economics

of data points that can be evaluated in order to measure the impor-tance of contextual events. Background on coattail effects

In the paper, we apply electronic prediction markets to understand the interplay between national and local elections. While resting par-tially on local issues and candidate personalities, local elections are often a referendum on national political figures. Local elections are heavily correlated with national results, and many elections—such as the 2008 election—have the same party sweep to victory in the House, the Senate, and the presidency.

Some of this correlation is due to contextual shifts, such as macro-economic changes or wars abroad. Other sources of correlation may be due to a demographic transition, leading to partisan realignment. A third part of the correlation comes from the presence of coattail effects, or the effect that a top-level candidate has on a local candi-date. A popular presidential candidate can register new voters, drive partisan turnout and party-line voting, and change voter preferences on issues, leading to coattail effects that change the outcome of con-gressional elections.

Despite the intuitive nature of coattail effects, attempts at effective-ly quantifying these effects have generally floundered. Since elections are correlated for a wide variety of reasons, measuring basic correla-tion between elections is insufficient to establish coattail effects. The methodological obstacles to measuring coattail effects were pointed out by Miller (1955), who noted the simultaneity of decisions taking place in elections. Mondak (1990) asserted that “a growing consensus holds that the presidential vote does exert significant influence on congressional elections,” though he added that these analyses were set back by the “long–recognized difficulties associated with measur-ing political coattails.”

Two primary obstacles exist in measuring coattail effects. First, there is a highly limited sample size: there is only one presidential election every four years from which to draw data, and there are substantial contextual changes that occur over that time frame. To overcome the limited datasets, we use electronic prediction markets to assess the status of elections on a continual basis. A second ob-stacle is one of endogeneity: much of the correlation between candi-dates is due to agreement on issues and shifts in economic context, rather than coattail effects? In this paper, we measure coattail effects within the 2008 election by using exogenous changes in the presi-dential election (that do not directly affect congressional elections) as an instrument. We find substantial coattail effects in the House and insignificant effects in the Senate.

The paper is organized as follows: first, we describe a theoretical model for the existence of coattail effects; then, we propose an em-pirical methodology based on the electronic prediction market data-set; next, we analyze the results of this methodology; and finally, we provide concluding remarks and suggest areas for future research.

Modeling Coattail Effects

Despite the difficulties in measurement, there are intuitive rea-sons why coattail effects are likely to exist. First, voter registration and turnout are often driven by grassroots campaigns and excite-ment for a top-level candidate. If a presidential candidate is able to register millions of first-time voters, those voters are likely to vote in lower-level elections for the same party. Similarly, excitement about a top-level candidate can help drive party-line voting in some states, as voters that are ambivalent about lower-level elections choose out of simplicity and expediency to vote for a party generally rather than evaluating individual candidates. While these effects can help shift votes towards a popular presidential candidate’s party, a more subtle

source of coattail effects shapes the median voter’s preferences direct-ly in lower-level elections. The median voter, typically a swing voter who was planning to vote and does not make a straight party-line vote, may change her lower-level preferences based on a change in her perception of a top-level candidate; her preference for a particu-lar presidential candidate informs her decision of which lower-level candidate to support.

To gain a better understanding of the impact of coattail effects on the election, we will consider a congressional election that is simulta-neous with a presidential election. Define candidate C as a congres-sional candidate in the same party as presidential candidate J. The decision of whether to vote for candidate C can be modeled as:

XiC = 1 if Udif_C (Equation 1)XiC = 0 if –Ū < Udif_C < Ū XiC = –1 if Udif_C < –Ū Where Udif_C = E[UiC]–E[UiC2]

XiC: vote cast by voter i for candidate C. Not voting counts as 0 votes, and voting against counts as a negative vote for the candidate.

UiC: utility voter i receives in the state of the world where candidate C wins

UiC2: utility voter i receives in the state of the world where candidate C’s opponent wins

Ū: absolute utility difference threshold that causes voter i to cast a vote.

Additionally, we can further expand these variables as:

E(UiC) = E(F(ii, viC)) (Equation 2)ii = i(c, ti)E(viC) = v(c, riC, pC, siC)

UiC: utility voter i receives in the state of the world where candidate C wins

viC: vector of expected agreement between voter i and candidate C’s decisions

ii: vector of weights of importance that a voter gives to particular is-sues

c: context of electionti: voter i’s tastespC: personal characteristics of candidate CriC: historical correlation between voter i and candidate C’s known

beliefssiC: signals of future agreement between voter i and candidate C

Under this model, a voter’s tastes (ti) and the election’s context (c) are the same in both the presidential and congressional elections, and are not affected by either candidate. Similarly, the personal charac-teristics of a candidate (pC) and the agreement between a voter and a candidate’s revealed beliefs (riC) are specific to a particular candidate, and we assume that they are not affected across elections. Instead, we model coattails as an informational phenomenon, with signaled beliefs (siC) as the source of coattail effects between the elections.

Intuitively, the model suggests that coattail effects occur as voters use presidential candidates as partial proxies for local, less-known candidates. Beyond any information known about the candidate per-sonally, the candidate’s party affiliation is also visible to a voter —a strong signal of the candidate’s future decisions if elected. Mondak (1990) finds that voters without much knowledge of a political race may use their views of the presidential candidate in the same party as a factor in their votes. Given the time cost of information collection



Econ

omic

s

and the voter’s scarce budget of time, a typical voter spends only part of her time considering the election, and uses the presidential candi-date as a proxy for the views of the local candidate. We formalize this view of election signaling as a simplified model:

siC = s(vij, wic) (Equation 3)siC: signals of future agreement between voter i and candidate C vij: vector of expected agreement between voter i and presidential can-

didate j’s decisionswiC: relative strength of presidential views as a proxy for congressio-

nal candidate’s views

The extent that the voter expects to agree with the presidential candidate’s political decisions (vij) affects the voter’s expectation of how much she expects to agree with the lower-level candidate (vic). The importance of the presidential candidate as a proxy is given a weight (wic) that shows the extent to which the voter relies on the presidential candidate as a proxy (which in turn is affected by the perceived correlation between the presidential candidate and the lo-cal candidate as well as the relative amount known about the views of each). Assuming a candidate’s views are considered to be positively correlated with the presidential candidate of the same party, then ∂sic /∂vij > 0, implying that ∂Uic/ ∂vij > 0. Thus, the probability that the voter votes for candidate c increases with the voter’s expected de-gree of agreement with the presidential candidate (vij). Furthermore, the magnitude of impact of coattails on voter choice increases with weight, so ∂2Uic/ ∂vij∂wic > 0.

This model outlines a mechanism for why, ceteris paribus, a candi-date’s popularity might be affected by a change in political popular-ity of the presidential candidate from the same party. Since a voter often knows more about a presidential candidate than about a local candidate, she uses the presidential candidate as a proxy for the local one – resulting in coattail effects. The problem with traditional measurement approaches

For an observer trying to estimate the impact of coattail effects, many of the variables from Equations 1 and 3 are not directly observ-able; instead, what is observable is the median voter’s decision on the presidential election, Xij and their decision in the lower-level election Xic. Previous attempts at measurement, such as the C-Correlation ap-proach, have effectively regressed lower-level election results directly on presidential election results, or Xic on Xij. Equation 3, however, demonstrates why this is problematic. The unobserved effect of con-text is a factor in both Xic and Xij, causing correlation without coat-

tails.To give a concrete example of this problem, in the 2008 election,

negative events in the economy tended to shift voters’ views in ways that were favorable to Democrats in general. There was substantial correlation between local elections and the presidential election, but much of this relationship was due to the favorable context for Demo-cratic policies that were shared among candidates. The correlation in such a case could not be attributed to coattails, since it was eco-nomic changes—rather than changes in the presidential election—that shaped the association between the two elections.

A New Empirical Methodology

Despite the intuitive theoretical justifications for coattail effects, attempts to measure the effects empirically have been hindered by both causality questions and a limited dataset. To overcome these obstacles, we propose an instrumental variable approach that uses data from electronic prediction markets. There are three steps to our methodology: first, we will broaden the dataset by turning to elec-tronic prediction markets; second, we will select specific events where the presidential elections were exogenously affected without any di-rect effect on congressional elections; and finally, we will analyze the impact of these events on the congressional elections.Step 1: Broadening the dataset

In order to broaden the dataset, we use electronic prediction market data for the 2008 election from the Iowa Electronic Markets. We track daily closing prices for Democrats in the winner-take-all (WTA) market for the presidential election and the seat gain WTA market for the House and Senate elections from August 26, 2008 (the date the congressional markets opened) until November 4, 2008 (Election Day). In these markets, House and Senate prices are ag-gregated total probabilities of the Democrats winning seats in each respective chamber rather than looking specifically at particular dis-tricts. Figures 1 and 2 show the association between weekly changes in presidential markets and changes in the House and Senate races, respectively. As expected, the figures demonstrate a clear correlation, from which we intend to isolate the coattail effects.

Due to light trading and large bid-ask spreads, we use weekly price changes for the congressional prices instead of daily changes. However, if we were to compare weekly prices for each, we would un-intentionally include correlation that does not result from coattails by expanding our time horizon too greatly (for example, economic changes over that time period may affect both if the time horizon is

Figure 2. Senate and presidential election correlation. Figure 2 shows weekly changes in prediction market prices for Democrats winning the Senate (on the y-axis) against weekly changes in price for Obama win-ning the election (on the x-axis). There is a general positive correlation between these two probabilities.

Figure 1. House and presidential election correlation. Figure 1 shows weekly changes in prediction market prices for Democrats winning the House (on the y-axis) against weekly changes in price for Obama win-ning the election (on the x-axis). There is a general positive correlation between these two probabilities.


31www.thurj.org

Economics

too long). Thus, throughout this paper, we are generally comparing weekly congressional shifts with daily presidential shifts, assuming that any congressional price change that occurred in the preceding six days is in a random direction. We also remove from the dataset all periods in which there was no trading volume. Finally, we restrict the dataset to periods on which presidential candidate Barack Obama gained vote share as a proxy for an election event being positive. Sum-mary statistics are shown in Table 1.

Step 2: Choosing eventsIn order to construct an instrument, we establish a binary variable

dependent on whether or not there is a major shift that affects the presidential election without directly impacting lower-level elections. While some factors in the presidential election (such as economic context) are also major direct factors in lower-level elections, certain events only affect the presidential election directly. For example, a presidential debate—designed to influence public opinion about only the presidential candidates—would impact congressional outcomes exclusively through coattail effects. After analyzing major events in the election,1 we select seven such events to set our binary variable equal to one:

8/23/08 – Obama selects Biden for VP• 8/28/08 – Obama Acceptance Speech• 8/29/08 – McCain selects Palin for VP• 9/27/08 – Day after 1• st debate10/3/08 – Day after VP debate• 10/8/08 – Day after 2• nd debate10/16/08 – Day after 3• rd debate

To confirm that these events had a substantial effect on the presi-dential election, we compared prediction market price changes on major event days in which Obama’s market price for winning the presidency improved. As Figure 3 shows, the average gain in the presidential markets nearly doubles on the dates with major events. Step 3: Measuring impact

Now that we have measured the impact of these events on the presidential elections, we turn to the congressional elections to gauge the impact that these major, exogenous events from the presidential election have on the congressional markets. Figure 3 shows a sub-

1 While a number of resources were used to select the largest events of the elec-tion, the Pew Research Center’s “Top 25 Events of 2008 Election” was particularly useful. In choosing events, we selected events that were clearly exogenous to the election and were confined to a discrete time period. While there were several other significant events in the Presidential election (such as Obama travelling to Europe), we specifically selected for events with measurable short-term effects.

stantial effect of major presidential events on the congressional mar-kets that increases average gains by a factor of eight in the Senate and a factor of four in the House.

To quantify this effect, we run a two-stage least squares regression using the presence of a major event as an instrument. As our first step, we regress changes in the probability of Obama winning on the presence of a major event on days where Obama gained popularity in the election:

Change_DemPres_WTA = α0 + α1 × MajorEvent + u (Equation 4)Change_DemPres_WTA: daily price change for prediction market

outcome of Obama winning presidencyMajor_Event: binary variable equal to one on days with major presi-

dential eventsu: error term

We can now plug this result in to measure coattail effects. Equa-tion 4 shows that the probability that a particular local candidate wins the median voter’s support is a function of candidate-specific effects (including the candidate likability and the agreement between the voter and the candidate’s revealed beliefs), time-specific contex-tual effects, and changes in the popularity of the party’s presiden-tial candidate. In a naïve model regressing change in congressional probability on change in presidential probability directly, the er-ror term would include both the candidate-specific effects and the time-specific contextual effects and would thus be correlated with the regressor. To overcome this problem with the naïve model, we instead use the predicted effect from Equation 4 as a regressor. Since we constructed MajorEvents to be uncorrelated with macroeconomic contextual changes and based exclusively on exogenous events in the campaign, its expected correlation with the error term is 0. Aggre-gating local elections into a metric of all ongoing congressional elec-tions, this instrument enhances our model for a change in probability of the Democrats winning the election:

Change_DemCongress_WTA = (Equation 5) β0 + β1 × Predicted_Change_DemPres_WTA + ε Where Predicted_Change_DemPres_WTA = α0 + α1 × MajorEvent

(with α0 and α1 from Equation 4)Change_DemCongress_WTA: weekly price change for prediction

market outcome of Democrats winning seats in Congress

Variable Obs Mean Std. Dev.

Min. Max.

Presidential WTA Price 46 0.693 0.106 0.553 0.903

Pres. Daily Change 45 0.003 0.023 -0.09 0.054

Senate WTA Price 30 0.923 0.045 0.806 0.993

Senate Daily Change 29 0.603 0.044 -0.149 0.059

House WTA Price 30 0.881 0.065 0.764 0.974

House Daily Change 29 0.002 0.054 -0.133 0.069Table 1. Summary statistics. Table 1 provides summary statistics for our dataset. The Winner-Take-All (WTA) Price is the market probability that the Democrats win a particular election.

Figure 3. Average gains in election markets. Figure 3 displays the average daily gains for the Obama WTA contract and the average weekly gains in House and Senate markets on days where the closing price in-creased. Days with major events showed substantially larger average gains in all markets.



Econ

omic

s

Change_DemPres_WTA: daily price change for prediction market outcome of Obama winning presidency

ε: error term

Through the two-stage least squares outlined in Equation 4 and Equation 5, we observe the effect of an exogenous shift in presidential preferences on Congressional prediction markets, yielding an unbi-ased estimate of coattail effects as β1.

Results and Discussion

General results and discussionOur results from the two-stage regression in Equations 4 and 5

are reflected in Table 2. The results display strong, significant coattail effects in the House election but weaker, insignificant results in the Senate. A 1% increase in Obama’s likelihood of winning is associated with a 2.1% increase in the Democrats winning the House, but only a 0.76% change in probability in the Senate. The coattail effect in the House election is quite strong, as an event occurring in the presiden-tial election has a larger effect on candidates in the House than on the presidential candidates themselves.

The general model of coattail effects in Equation 3 helps explain the difference in magnitude of the effects in the House and Senate. One of the model’s core results is that a voter’s perception of the presi-dential candidate matters in congressional elections only to the ex-tent that the presidential candidate’s views are a useful proxy for the congressional candidate’s views, or the magnitude of the variable wic. In an election where the median voter knows less about the candi-date, coattail effects are expected to be stronger, since the party affili-ation of the candidate carries more meaning. Typically, Senate candi-dates are better known than their House counterparts; since senators serve longer incumbencies, wield more personal power, spend more on their campaigns, and have broader constituencies, they generally capture a larger share of news coverage and political buzz. Thus, the median voter often knows more about the candidates in the Senate election than the candidates in the House election, and presidential views are more necessary as a proxy in the House election.Amplification

One of the more surprising results of this analysis is the amplifi-cation of effects across elections. A one percent change in the prob-ability of Obama winning is transformed into a greater than one per-cent change in the probability of Democrats winning in the House election. We propose two hypotheses for why amplification exists: geographic influences and the self-fulfillment of predicted success. In these markets, changes in voting trends in different states are not equally important, as a vote shift in a non-swing state is not as mean-

ingful as a vote in a swing state. The presidential swing states may vary from congressional swing states. Thus, a presidential announce-ment that “rallies the base” may have an amplified effect in states with close congressional elections but that are heavily leaning towards a particular party already in the presidential election.

An alternative hypothesis for amplification is that success is self-fulfilling, as the perception of being likely to win an election can in-crease the actual likelihood of winning. Campaign contributors, key politician endorsers, and even news editorial boards are often eager to support the winning side of an election and reap potential benefits of early support. A company’s executives may choose to donate to a campaign they expect will win in order to gain potential political goodwill, and they must base their decisions on probability estimates made weeks or months before an election. Since these acts of sup-port may directly shift the outcome of an election (and likely have larger effects in elections where candidates are lesser known), success may have inertia that amplifies a small probability change into strong gains. Case study: Minnesota

Turning from the general application of our model across all con-gressional and Senate elections in 2008, we now apply our model to one specific election: the 2008 Minnesota Senate election. Because of the perceived closeness of Minnesota’s election, the Iowa Electronic Markets introduced a market with reasonable trading volume that measured the vote share in state’s senatorial election. Table 3 shows the results of applying the instrumental variable regression from Equation 5 on Minnesota’s results. A 1% change in likelihood of Obama’s victory is associated with a 5.865% increase in the expected vote share for Franken (the Democratic candidate in this election).

The magnitude of the coattail effect is strikingly large, especially for a Senate race. The difference in effect between this election and other Senate elections is likely due to the extreme closeness of this race – a race that was ultimately decided by just a few hundred heav-ily-contested ballots. Thus the impact of any perceived political shift would be heavily amplified in this ultra-close election. Furthermore, Minnesota was not a heavy swing state in the presidential election, and ultimately voted by a substantial margin for Obama. Thus, events in the presidential election that resounded with liberal voters may have had a stronger effect on the Senate election outcome than the general election probabilities. Limitations and further research

There are several limitations to these conclusions, and several questions are raised for further research. First, while this methodolo-gy could be generalized, we only used data from the 2008 elections in this analysis due to the limited history of electronic prediction mar-kets for Congressional elections. As more elections occur and the use

Variable Coefficient

Predicted_Change_Pres_WTA_Price 5.865 ***[1.349]

_const -0.057[0.035]

Observations 39Table 3. 2-stage least squares results in Minnesota. This table shows the results of Equation 5, demonstrating the predicted effect of ex-ogenous changes in presidential market probabilities on the Minnesota Senate election. The election demonstrates strong, significant coattail ef-fects, while the effects in the Senate are minimal. Heteroskedasticity-ad-justed standard errors are displayed in brackets. *** significant at 1%

Variable House Senate

Predicted_Change_Pres_WTA_Price 2.102 **[0.901]

0.764[0.587]

_const -0.020[0.021]

-0.005[0.015]

Observations 29 29Table 2. 2-stage least squares results. This table shows the results of Equation 5, demonstrating the predicted effect of exogenous changes in presidential market probabilities on House and Senate probabilities. The House demonstrates strong, significant coattail effects, while the effects in the Senate are minimal. Heteroskedasticity-adjusted standard errors are displayed in brackets. ** significant at 5%


33www.thurj.org

Economics

of electronic prediction markets expands, a broader analysis could be performed on the impact of coattail effects on elections and the determinants of the magnitude of the effects. In particular, the 2008 election was unique in recent elections because of the relative cer-tainty of the eventual outcome for Senate and House control, as the probabilities that Democrats would control each chamber were over 90% for much of the period evaluated. Thus, an analysis performed in a year when both the presidential and congressional elections are expected to be close may yield different results.

As prediction markets continue to mature, another potential ex-pansion would be to observe coattail effects on a state-by-state level. While state-by-state election details from electronic prediction mar-kets were limited in the 2008 election, further expansion in market availability and trading volume would enable a study on state-specific coattail effects and determinants of the magnitude of coattails.

Finally, several assumptions were made in choosing to use elec-tronic prediction market prices as proxies for probabilities of actual outcomes. First, we assumed that they are unbiased, reasonably accu-rate estimators, claims that are generally supported by both empiri-cal and theoretical literature on prediction markets. A more subtle assumption is that changes in prediction market prices are due to events impacting the election directly, rather than merely changing perceptions of the candidates themselves by market participants. For example, an eloquent speech by a candidate would not only boost his chances in the election directly, it might also signal to predic-tion market participants that he will be similarly eloquent in the future—and traders will factor these future events into the market price. This effect would likely be particularly magnified early in an election, since market participants know less about the strategic and tactical competence of particular candidates. In order to mitigate this effect, we selected events for this study that occurred relatively late in the election and would likely have a primary effect of directly alter-ing the election, such as the selection of a vice-presidential candidate. Furthermore, even if the market changes its perception of a candidate and incorporates information about future expected events because of one of the events we selected, as long as those future events also generate coattails, they will also be incorporated into congressional prediction markets, leaving our estimate unbiased.

Concluding Remarks

Our results provide an intriguing insight into one of the major determinants of United States elections outcomes. Using electronic prediction market trading data, we find that coattail effects have a major impact on House elections, but a more limited impact on Senate elections. By using major exogenous shifts in the presidential election as an instrument, we isolate the coattail effects in the elec-tion and overcome causality concerns intrinsic to most past studies of empirical effects.

While this result generally confirms our intuitive expectations about coattail effects, the methodology provides an intriguing new way to study major factors shaping elections. Similar to coattail ef-fects, many factors that intuitively have large effects on elections are difficult to show empirically, since elections occur only once every few years and substantial differences in context occur from election to election that complicate any attempts to hold multiple factors con-stant. By using market prices of a particular outcome as a proxy for the status of an election, electronic prediction markets promise to

widely expand the data set available for analyzing election phenom-ena. Nonetheless, most literature on electronic prediction markets to date has been either theoretical in nature or an empirical measure-ment of the market’s predictive ability. By assuming the markets are in fact reasonably accurate (which most analysis supports), we use prediction markets as robust, real-time indicators of the status of an election. Our results indicate that – beyond their predictive value – prediction markets can also be a useful tool in isolating empirical effects in presidential elections and uncovering the determinants of political success.

ReferencesBerg, J. et al., “Results from a Dozen Years of Election Futures Markets 1. Research.” Working Paper (2001).Berg, J. et al., “Accuracy and Forecast Standard Error of Prediction Mar-2. kets.” Working Paper (July 2003).Erikson, Robert et al. “Was the 2000 Election Predictable?” Political Sci-3. ence and Politics, Vol. 34, No. 4 (Dec., 2001), pp. 815-819Fair, Ray. The Effect of Economic Events on Votes for President. The Re-4. view of Economics and Statistics, Vol. 60, No. 2 (May 1978), pp. 159-173.Fair, Ray. “Econometrics and Presidential Elections.” Journal of Econom-5. ic Perspectives, Vol. 10, (Summer 1996), pp. 89-102. Fair, Ray. Predicting Presidential Elections and Other Things. Stanford 6. University Press (2002).Ferejohn, John and Randall Calvert. “Presidential Coattails in Historical 7. Perspective.” American Journal of Political Science, Vol. 28, No. 1 (Feb., 1984), pp. 127-146. Frey, B. and Schneider, F. “Economic and Personality Determinants of 8. Presidential Popularity.” Empirical Economics, Vol. 3, No. 2 (June, 1978), pp. 79-89Kaplowitz, Stan. “Using Aggregate Data to Measure Presidential Coat-9. Tails Effects.” Working Paper (1970). Levitt, Steven. “Using Repeat Challengers to Estimate the Effect of Cam-10. paign Spending on Election Outcomes in the U.S. House.” The Journal of Political Economy. Vol. 102, No. 4 (Aug., 1994), pp. 777-798.Lewis-Beck, M. “Economic voting: an introduction.” Electoral Studies, 11. Volume 19, Issue 2-3 (2000), pp. 113-121 Miller, Warren. “Presidential Coattails: A Study in Political Myth and 12. Methodology.” The Public Opinion Quarterly, Vol. 19, No. 4 (Winter 1955), pp. 353-368 Mondak, Jeffery. “Determinants of Coattail Voting.” Political Behavior, 13. Vol. 12, No. 3 (Sep., 1990), pp. 265-288. Mondak, Jeffery and Carl McCurley. “Cognitive Efficiency and the Con-14. gressional Vote: The Psychology of Coattail Voting.” Political Research Quarterly, Vol. 47, No. 1 (1994), pp. 151-175.New York Times. President Map. http://elections.nytimes.com/2008/re-15. sults/president/map.html. Accessed February 2009.Pew Research Center. “Top Events of Campaign 2008.” November 6, 2008. 16. Available at http://pewresearch.org/pubs/1025/election-news-interest Shaw, Daron. “The Effect of TV Ads and Candidate Appearances on State-17. wide Presidential Votes, 1988-96.” The American Political Science Review, Vol. 93, No. 2 (Jun., 1999), pp. 345-361Snowberg, Erik et al. “Partisan Impacts on the Economy: Evidence from 18. Prediction Markets and Close Elections.” The Quarterly Journal of Eco-nomics, Vol. 122, No. 2 (2007), pp. 807-829. Washington Post. Campaign Tracker. http://projects.washingtonpost.19. com/2008-presidential-candidates/tracker/candidates/barack-obama/states/oh/. Accessed in February 2009.Wisconsin Ad Project. “Pres. TV advertising spending continues to 20. grow; Over $28 million spent from September 28-October 4.” Univer-sity of Wisconsin. Available at http://wiscadproject.wisc.edu/wiscads_release_100808.pdf. Wolfers, Justin and Eric Zitzewitz. “Interpreting Prediction Market Prices 21. as Probabilities.” CEPR Discussion Paper No 5676 (May 2006).Wolfers, Justin. “Best Bet for Next President: Prediction Markets.” Wall 22. Street Journal. December 31, 2007.

Publ

ic H

ealth

an

d M

edic

ine

Element of survival: Isolating the causal effect of access to iodized salt on child

health in IndiaShubha Bhat

Harvard University ‘09, [email protected]

Despite India’s longstanding efforts to combat Iodine Deficiency Disorders through a Universal Salt Iodization pro-gram, only 51% of households were using iodized salt in 2005. In order to justify efforts to actively expand the pro-gram, it is crucial to establish a causal link between access to iodized salt and child health. This study examined household salt iodine concentration, anthropometric outcomes, and birth histories of over 18,000 children from the 1998 India National Family Health Survey. To isolate causality, two-stage least squares (2SLS) regressions were used with state-fixed-effects. Innovative district-level instruments were constructed using Geographic Information Sys-tems to predict salt iodine concentration in households and targeting efforts of the government. The 2SLS estimate revealed that increasing the iodine level in salt led to a 1.168 standard deviation increase in height-for-age (p<0.05), a 15.9% decreased likelihood of having below-average birth weight (p<0.05), a 18.6% increased likelihood of having above-average birth weight (p<0.05), and a 2.8% increase in child survival (p<0.10).

Introduction Iodine deficiency disorders (IDD) are one of the most common

causes of preventable mental retardation (X. Y. Cao et al., 1994). A meta-analysis of 18 studies of 2214 subjects comparing the perfor-mance of iodine-deficient children with that of iodine-sufficient peers on a standardized intelligence tests, concluded that iodine deficiency lowered the mean intelligence quotient by 13.5 points, which indi-cates a staggering public health problem (Bleichrodt and Born 1994). This shortcoming affects a child’s ability to learn, and later in life, to earn. In this way, the negative effects iodine deficiency on both men-tal and physical health can significantly impede worker productivity and the economy at large. Clearly, eliminating iodine deficiency has a significant impact on the world’s poor.

To combat IDD, the WHO and UNICEF recommended in 1994 the use of iodized salt as a safe, cost-effective and sustainable strategy to ensure sufficient intake of iodine by all individuals. However, de-spite pushing a Universal Salt Iodization program, progress has been limited. In India, which is a major salt-producing country, only 51% of households were using iodized salt in 2005. This is a particular problem since out of 38 million newborns in developing countries every year that remain unprotected from the lifelong consequences of IDD-related brain damage, a large percentage live in South Asia, as shown in Figure 1 (UNICEF 2008).

However, to justify efforts that the Indian government is making to improve the system, it is important to understand what effect the lack of iodized salt coverage has had on public health. Specifically, it is necessary to determine whether IDD affects the physical as well as mental capacities of children in India. Thus, identifying the causal link between iodine deficiency and child survival and growth will more effectively direct the appropriate resources toward its correc-tion. In doing so, this study also aims to contribute to the method-ological literature on measuring intervention impacts.

Methodology

DatasetsThe primary dataset used in this study was from India’s second

National Family and Health Survey (NFHS-2), which was conducted from 1998-1999 by the International Institute for Population Sciences (IIPS) in Mumbai (IIPS and Macro 2000).1 This dataset was ideal for two reasons. First, it captured key socioeconomic, cultural, and health information about over 91,880 households, 90,303 women and 33,132 children in India. Second, it provided district-level identifiers for each household, which was not available in the most recent 2005 NFHS-3 because of privacy issues. This was crucial because it allowed for the use of within-state variation by district in order to estimate causality, which is particularly helpful for assisting policymakers in planning and implementing strategies for improving population health and nutrition programs.

Six secondary datasets were superimposed on a map of India and linked to the NFHS-2 through district-level household identifiers to create a series of district-level control and instrumental variables. Measurements were made using the ArcGIS9 (ArcMap Version 9.3) 1 At the national level, the overall sample weight for each household or woman is the product of the design weight for each state (after adjustment for non-response) and the state weight. I use these sample weights in my main results.

Figure 1. Children unprotected from IDD (2000-2006).


35www.thurj.org

Public Health

and Medicine

software. First, MLInfoMap: 1991 Census provided map data on district borders, area and population. Second, the Global Precipita-tion and Climatology Center’s Full Data Reanalysis Project provided information on average annual rainfall for each district. Third, the Shuttle Radar Topography Mission provided 90-meter by 90-meter raster images of elevation. Fourth, the Global Self-consistent, Hier-archical, High-resolution Shoreline (GSHHS) Database Version 1.3 provided information on distances to the closest coast. Fifth, Collins Bartholomew World Premium 2007 provided information about the main and secondary railway lines, as well as the road networks in India. Sixth, the Government of India Salt Department’s 2005-2005 Annual Report provided information on how much salt was trans-ported to each state by railroad, road, or sea.Variables

The explanatory variable of interest in this paper was the iodine level of the salt used in a household. In NFHS-2, interviewers mea-sured the iodine content of cooking salt in each interviewed house-hold using a rapid-test kit and recorded the iodine level as 7, 15, or 30 parts per million (ppm) (IIPS and Macro 2000).

Overall, half of households used cooking salt that was iodized at the recommended level of 15 ppm or more, one-quarter of house-holds used salt that was not iodized at all, and 21 percent used salt that was inadequately iodized (less than 15 ppm). The use of iodized salt varied dramatically from one state to another (as shown in Figure 2). The variations could be due to a number of factors, including the scale of salt production, transportation requirements, enforcement efforts, pricing structure, and storage patterns. In particular, salt io-dization was likely to be more common in states where salt was trans-ported exclusively by railways, partly because the Salt Department monitored the iodine content of salt shipped by railways.

The five child health outcomes that this paper examined were height-for-age, weight-for-height, likelihood of being born small, likelihood of being born large, and overall child survival. These out-comes were based on NFHS-2 anthropometric data of 18,521 chil-dren ages 0-5 years, and NFHS-2 birth history of 21,388 children of all ages.

The first two outcomes, children’s height-for-age and weight-for-height were evaluated relative to the median height-for-age and weight-for-height of the 1997 U.S. National Center for Health Sta-tistics (NCHS) reference population, recommended by the World Health Organization. Measures greater than two standard deviations below the reference median considered stunted (height-for-age) or wasted (weight-for-height). Stunting is a sign of chronic, long-term under-nutrition, whereas wasting is a sign of acute, short-term un-der-nutrition. It was therefore hypothesized that access to household iodine would lead to improved height-for-weight outcomes and have no effect on weight-for-height outcomes. Such a result would support the fetal origins of disease hypothesis.

On average, children in India were borderline stunted at 1.91 standard deviations below the reference population. Those children living in households with salt iodine content of 0, 7, 15 and 30ppm were 2.06, 2.07, 1.98 and 1.67 standard deviations below the median, respectively. Thus, there is a clear corresponding trend between io-dine concentration and height-for-age outcome. On the other hand, though children were on average 0.78 standard deviations below the reference median, they were not low enough to be considered wasted. However, unlike the prediction, there seemed to be a similar rising trend of weight-for-height (0.96, 0.89, 0.75 and 0.6 standard devia-tions below the reference), which corresponded with rising iodine concentrations (0, 7, 15, and 30ppm, respectively). Such a pattern pos-sibly indicates that there are other factors such as standard of living that may be driving these similar trends.

The third and fourth outcomes, likelihood of being born small and large, were chosen because small newborns generally face sub-stantially higher risks of dying than do newborns of normal or large size. The average birth weight of the small, average and large children was 2.2 kg, 2.8 kg and 3.4 kg, respectively. According to mothers’ esti-mates, 25 percent were small, 60 percent were average, and 14 percent were large. As the salt iodine concentration went up, there was also a corresponding decrease in fraction of children born small and in-crease in fraction of children born large.

The fifth and final health outcome examined was childhood sur-vival, which was calculated as the fraction of children still living over the total number of children ever born to each woman. In total, the child survival rate was 91 percent. In this case, when broken down by iodine concentration, there seemed to be no difference in child sur-vival rates within households that had 0, 7 or 15ppm. However hav-ing 30ppm iodized salt corresponded with a 93 percent survival rate.Estimation Strategy

To test whether trends in health outcomes are driven by iodine concentration and not other factors ordinary least squares (OLS) and two-stage least squares (2SLS) regressions were conducted. Below, the empirical specification for the OLS regressions run in this analy-sis is presented:

Y = α+β1(I)+β2(M)+β3(H)+β4(D)+β5(S)+ε (Equation 1)

In equation 1, Y is the child health outcome of interest. This in-cludes height-for-age standard deviations from the international mean, weight-for-height standard deviations from the international mean, child’s size at birth, and child survival. The α is the constant, β1 is the coefficient for I, the explanatory variable. Iodine levels of 1, 2, 3 and 4 correspond with iodine concentrations of 0, 7, 15 and 30 parts per million, respectively. β2 is the vector coefficient for M, the mother-level controls. These include characteristics such as age, education level, possession of a health card, body-mass-index, and health status (smoking habits, alcohol consumption, tobacco use, TB, jaundice, asthma, and malaria). β3 is the vector coefficient for H, the Figure 2. Iodine concentration.

Publ

ic H

ealth

an

d M

edic

ine

Equation 2 corresponds with the first stage of the 2SLS estimate. Î is the predicted value of household salt iodine. Unlike equation 1, β1 in equation 2 is the vector coefficient for Z, the instruments. These in-clude both institutional and geographic instruments. The institution-al instruments include distance to any railroad (Figure 3), distance to the nearest main railroad, fraction of salt transported to the state by railroad, and total road length in the district (Figure 4), which pre-dict access to iodized salt. The geographic instruments include aver-age district precipitation (Figure 5), elevation (Figure 6), and distance to the nearest coastline (Figure 7), which predict baseline endemic iodine deficiency.

Equation 3 corresponds with the second stage of the 2SLS esti-mate. Here, Y is the child health outcome of interest. β1 is the coef-ficient for Î, the iodine level predicted by the instruments. As before, the remaining coefficients correspond with mother-, household-, dis-trict- and state-level controls.

The 2SLS regressions were also broken down to capture heteroge-neous treatment effects on male versus female children and on urban versus rural households to determine whether the effect of iodized salt had a greater impact on certain groups. Finally, three robustness checks were carried out to assess how sensitive the results were to the econometric specification.

Results

When district-, household-, and mother-level observable charac-teristics were controlled for in the OLS regression (Column 1, Table 5), it was found that increasing the iodine concentration by one level (i.e. 0 to 7ppm, 7 to 15ppm, or 15 to 30ppm) correlated with a 0.0375 standard deviation increase in height for age (p<0.01). The 2SLS esti-mate (Column 2, Table 5) revealed that increasing iodine concentra-tion by one level led to a 1.168 standard deviation increase in height for age. Looked at another way, increasing the salt iodization con-centration by two levels, from 0 to 15ppm (the concentration recom-mended by the WHO) led to a 2.336 standard deviation increase in height for age. Essentially, such an effect would put children in India (who on average are at 1.91 standard deviations below the median) at

household-level controls. These include characteristics such as the standard of living index, urban environment, and reli-gion of household head. β4 is the vector coefficient for D, the district-level controls. These include district population density and area. β5 is the vec-tor coefficient for S, the state-fixed effects. Finally, ε is the error term.

The disadvantage with OLS is that there are several unob-servable characteristics that cannot be controlled for which may affect both the outcome and explanatory variable. Al-though OLS helps identify pat-terns, it is difficult to use an OLS estimation to determine

causality. Therefore, a two-stage least squares (2SLS) methodology is used. The 2SLS technique employs instrumental variables, which are supposed to be highly correlated with the outcome variable only through a correlation with the explanatory variable. Below, the em-pirical specification for the 2SLS regressions run in this analysis is presented:

Î = α+β1(Z)+β2(M)+β3(H)+β4(D)+β5(S)+ε (Equation 2)

Y = α+β1(Î)+β2(M)+β3(H)+β4(D)+β5(S)+ε (Equation 3)

Figure 3. Railroad. Figure 4. Road length. Figure 5. Precipitation.Figure 6. Elevation.Figure 7. Coast.

Figure 4

Figure 5 Figure 6

Figure 7

Figure 3

37www.thurj.org

Public Health

and Medicine

just over normal, on the scale of international growth standards. Contrary to the prediction, in the OLS estimate (Column 3, Table

5), an increased iodine level correlated with a 0.0265 standard de-viation increase in weight-for-height outcomes for children (p<0,05). However, as expected, the significance of this estimate became nil in the 2SLS (Column 4, Table 5).

Increasing the iodine concentration by one level led to a statisti-cally significant (p<0.05) but small (0.008 percent) decrease in the likelihood of being born small in the OLS estimate (Column 5, Table 5). This value became much larger and more significant in the 2SLS. As shown in Column 6, Table 5, increasing iodine concentration level led to a 16% drop in the likelihood of being born small (p<0.01). The correlation between iodine concentration level and likelihood of be-ing born large was insignificant in the OLS regression (Column 7, Table 5). However, iodine concentration seemed to have a large and significant effect in the 2SLS (Column 8, Table 5). An increase in io-dine level led to an 18.6% increase in likelihood of being born large (p<0.01).

The OLS and 2SLS estimates (Columns 9 and 10, Table 5) showed that an increase in salt iodine concentration level led to a 0.22 per-cent and a 2.8 percent increase in child survival rates, respectively (p<0.1).

When the sample was broken down by gender and place of resi-dence, there were significantly greater effects of iodized salt on fe-male children and rural households.

Discussion

The goal of this study was to establish a causal link between access to iodized salt and child health outcomes. Showing the direct relation between iodine deficiency and child survival and growth would help motivate researchers and policy-makers to identify the barriers that prevent India from reaching more widespread use of iodized salt. In addition, illuminating the differential effects of access to iodized salt on subpopulations would help direct resources more appropriately.

This economic analysis is unique because it utilized Geographic Information Systems to measure precise distance and geographic information and connected this information with the district codes of households in this sample. Used together as instruments, the GIS data served to identify causal effects where a controlled experiment was not possible. In addition, because the analysis used the 1998 NFHS, it covered a representative sample of the Indian population as a whole. The timing of the survey captured the effect of five years of universal salt iodization program in India and might have captured the ban on non-iodized salt that had been instated in 1998.

The results of this study suggest that access to iodized salt posi-tively impacted children’s height-for-age outcomes, especially for

children living in rural settings. The magnitude of the effect is sig-nificant since increasing iodine concentration from 0 to 15 ppm seemed to drive the average height-for-age standard deviation from -1.91 (borderline stunted) to nearly +0.43 standard deviations above the NCHS standard. Such an effect can have a significant effect on the future productivity of the nation’s children. However, iodized salt might not be the answer to the problem of childhood wasting. This is not a huge concern since the average weight-for-height was not close to 2 standard deviations below the NCSH standard, which defines someone as wasted. In addition, the results suggest that use of iodized salt by mothers might reduce the likelihood of being born small and increase the likelihood of being born large, especially for girls and children born in rural areas. Finally, there seemed to be a positive effect of iodine on child survival, especially for males.

It is important to note that only two of the five outcomes—height-for-age and likelihood of being born large—passed the greatest num-ber of robustness checks. This indicates that we can only begin to extend the fetal origins of disease hypothesis to these two outcomes. However, since even these two outcomes failed to pass the most strin-gent robustness check, conclusions can only be made with some de-gree of caution.

There are several measurement issues that this analysis faces. In this sample of the NFHS-2, the reference population used in calcu-lating anthropometric outcomes was the 1997 U.S. National Center for Health Statistics (NCHS) standard. More recently however, a new international reference population was released by the WHO in April 2006 and accepted by the Government of India, which may better as-sess children regardless of ethnicity, socioeconomic status and type of feeding. In addition, the set of child health outcome variables fo-cusing on size of the newborn are especially prone to selection and reporting biases, since mothers tend to self-report their children as being bigger than they actually are, since size is considered a measure of health. Finally, the way that I calculated child survival was simply the number of children living over the total number of children born. It is unclear whether such a measurement can really capture the in-tricacies of child mortality early in life.

For the explanatory variables, an assumption was made that if a household had iodized salt at the time of the survey, it probably has had access to that level of iodized salt throughout the universal salt iodization program, which became more active starting in 1992. However, this assumption may not necessarily hold. The practices of households in 1998 do not necessarily have to reflect the practices of the household five years before.

The instruments used in this analysis showed high relevance in the first-stage regression, yielding high F-stats and significant cor-relations with the explanatory variable. However, the exogeneity re-quirement was less convincing. Though precipitation and elevation

Table 5. Effect of iodine concentration on child health outcomes. Controls = State fixed effects, population density, district area, standard of living, urban, hindu, muslim, christian, mother’s characteristics (age, education, has health card, bmi, smoking habits, alcohol and tobacco use, jaundice, malaria, TB, asthma). Instruments = distance to railroad, distance to main railroad, %salt transported to state by railroad, total road length, annual precipitation, elevatin, distance to coast (F = 153.22). All regressions are run with sample weights. * p < 0.10, ** p < 0.05, *** P < 0.001; Robust standard errors in parentheses.

Iodine

n

Height-for-Age1 2

OLS 2SLS

Weight-for-Height3 4

OLS 2SLS

Small Child5 6

OLS 2SLS

Large Child7 8

OLS 2SLS

Child Survival9 10

OLS 2SLS

0.0375*** 1.168***(0.0131) (0.311)20328 18521

0.0256** -0.0341(0.0107) (0.190)

20328 18521

-0.00798** -0.159***(0.00314) (0.0556)

23441 21388

0.00222 0.186***(0.00250) (0.0517)

23441 21388

0.00167* 0.0281*(0.000999) (0.0163)

23450 21396



Publ

ic H

ealth

an

d M

edic

ine

seem to be less of a problem, distance to the coast and distance to rail-roads were arguably endogenous. Presumably the closer a household is to a railroad or ocean, the more resources the household can access and therefore the higher standard of living it may experience.

The robustness check limiting the set of instruments to only dis-tance to railroad revealed insignificant results, showing that the main findings might be sensitive to the instruments chosen. Moreover this dataset was limited to district level data. Therefore, it is possible that the instrument used in the most stringent specification did not have enough variation to predict access to iodized salt accurately. Having exact coordinates of households would enable a more robust analy-sis.

Conclusion

This paper revealed that an increase in salt iodine levels led to pos-itive and statistically significant child health outcomes—namely, a 1.168 standard deviation increase in height-for-age, a 15% decreased likelihood of being small at birth, a 19% increased likelihood of be-ing large at birth, and a 2% increase in child survival. The paper also provided evidence that access to iodized salt had stronger effects on female children and on children living in rural settings. Therefore, targeting certain populations could potentially have a more power-ful impact on child health. This study is innovative in that it utilized unique instruments constructed through GIS data to predict access to iodized salt and isolate its causal effects on childhood health. Us-ing this estimation strategy, the paper shows that the Indian govern-ment would do well to continue promoting access to iodized salt in order to secure the health and well-being of its children.

ReferencesAllen, L. and S. Gillespie (2001). What Works? A review of efficacy and effec-

tiveness of nutrition interventions, UN (ACC/SN).Bleichrodt, N. and M. Born (1994). A Meta analysis of research on iodine and

its relationship to cognitive development. The damaged brain of iodine deficiency. J. Stanbury. New York, Cognizant Communication: 195-200

Cobra, C., Muhilal, et al. (1997). “Infant survival is improved by oral iodine supplementation.” J Nutr 127(4): 574-8.

Dunn, J. T. and F. Delange (2001). “Damaged reproduction: the most impor-tant consequence of iodine deficiency.” J Clin Endocrinol Metab 86(6): 2360-3.

Field, E., O. Robles, et al. (2007). The Cognitive Link Between Geography and Development: Iodine Deficiency and Schooling Attainment in Tanzania. NBER Working Paper Series. Cambridge, MA, National Bureau of Eco-nomic Research: 64.

Garrow, J. S., A. Ralph, et al. (2000). Human Nutrition and Dietetics. New York, Elsevier Health Sciences.

Hetzel, B. S. (1989). The Story of Iodine Deficiency; An International Chal-lenge in Nutrition. Oxford, Oxford University Press.

IIPS and O. Macro (2000). National Family Health Survey (NFHS-2), 1998-1999: India. Mumbai, IIPS.

Kapil, U., R. S. Raghuvanshi, et al. (1999). “Utility of spot testing kit in the assessment of iodine content of salt--A mutlicentric study.” Indian Pedi-atrics 37: 182-186.

Lamberg, B. A. (1991). “Endemic goitre--iodine deficiency disorders.” Ann

Med 23(4): 367-72.Pelletier, D. L. (1994). “The potentiating effects of malnutrition on child mor-

tality: epidemiologic evidence and policy implications.” Nutr Rev 52(12): 409-15.

Pharoah, P. and K. Connolly (1994). Iodine Deficiency in Papua, New Guinea. The damaged brain of iodine deficiency. S. Stanbury. New York, Cogni-zant Communication: 299-305

Sarkar, S., B. Mohanty, et al. (2007). “Iodine deficiency in school going chil-dren of Pondicherry.” Indian J Pediatr 74(8): 731-4.

Thilly, C., R. Lagasse, et al. (1980). Impaired fetal and postnatal development and high perinatal death-rate in a severe iodine deficient area. Thyroid research VIII. J. Stockigt, S. Nagataki, E. Meldrum, J. Barlow and P. Hard-ing. Canberra, Austrailian Academy of Science: 20-23.

UNICEF (2008). “Sustainable Elimination of Iodine Deficiency: Progress since the 1990 World Summit for Children.” 52.

Whitney, E. N., C. B. Cataldo, et al. (2002). Understanding Normal and Clini-cal Nutrition. Australia, Canada Wadsworth Group, Thomson Learning.

WHO, UNICEF, et al. (1999). Progress towards the elimination of iodine de-ficinecy disorders (IDD). Geneva, WHO.

Acknowledgements

I would like to thank the following people for their tremendous support throughout this research process. Without their help and encouragement, this original work would not have been possible. Erica Field, Assistant Professor of Economics, Harvard University, for being a wonderful thesis advisor and contin-ually guiding me in the right direction. Winnie Fung, PhD candidate in Econom-ics, Harvard University, for being my thesis tutorial leader and helping me clarify my questions and methodology. Konstantin Styrin, PhD candidate in Economics and STATA Teaching Fellow, Harvard University, for answering my countless STATA coding questions. Sebastian Linnemyer, PhD candidate in Economics and Teaching Fellow, Harvard University, for his advice in the Ec980 Junior Tu-torial, “Development, Education, and Health.” Scott Walker, Digital Cartogra-phy Specialist, Harvard Map Collection, for spending so much time teaching me about GIS, providing me with data and helping me construct map images. Bon-nie Burns, GIS Specialist, Harvard Map Collection, for helping me identify the 1991 India Census Data. Fred Arnold and Bridgette James, DHS/NFHS Archives, for providing me with district coding for the NFHS data. Joan Curhan, Debbie Whitney, Suzanne Scudder, for supporting me in the Certificate in Health Policy Program. The Cordeiro Family, for providing me with a generous grant to travel to India to continue on-the-ground research in July 2008. Darpana Academy of Performing Arts, for giving me the opportunity during my Fall 2007 semester in India, to travel to Valsad (pictured above), a rural district in Gujarat, and be-come involved in a UNICEF development project that inspired this thesis. Dr. Chandrakant S. Pandav and Dr. Arijith Chakrabarty, International Council for Control of Iodine Deficiency Disorders (ICCIDD), All India Institute of Medical Sciences, New Delhi, for spending time sharing their insight with me on the most recent accomplishments and challenges of the Universal Salt Iodization Program in India. Dr.Rajan Sankar, Regional Manager for the South Asia branch of Global Alliance for Improved Nutrition, New Delhi, for providing me with extensive literature on the history of iodine deficiency. Mr. S. Sunderesan, Salt Commis-sioner, and Mr. M.A. Ansari, Deputy Salt Commissioner at the Ministry of Salt in Jaipur, Rajasthan, for communicating with me via email regarding questions about the transportation and production of salt throughout the country. Tracy Li, Peter Ganong, Chethan Bachireddy, Prithvi Shankar, and all my friends and classmates for editing, giving me STATA and formatting tips, and helping make this process enjoyable. My parents, for their unconditional love.


39www.thurj.org

Neuroscience

Olfactory bulb glomerular responses of mice in different behavioral states

One of the first steps in processing a smell occurs at the glomerular layer, which is where axons of olfactory receptor neurons make synapses on the dendrites of mitral cells. While olfactory receptor neurons’ pre-synaptic activity have been extensively researched in anesthetized mice, little is known about odor-evoked synaptic activity in awake, freely moving animals. This project attempts to address this problem by adapting a fiber optic bundle imaging technique to the olfactory bulb. The results show that while behavior causes no consistent changes in olfactory receptor neuron activity across all subjects and odors, it does seem to be linked to differences in the nature of odor responses in indi-vidual mice. Importantly, this study is the first demonstration that a fiber optic imaging bundle can be used to image the olfactory bulb of an awake, behaving animal.

Introduction

One of the fundamental concerns of neuroscience is to under-stand how the brain perceives and processes the outside world. From the first stage of detection by sensory nerves to the forging of a mem-ory of an event, our brains are dynamically interacting with our sur-roundings. Olfaction is one of the ways we sense the world, yet the study of the olfactory system has many unanswered questions. How do we process different odors? How do we forge and retrieve odor memories? How do our different mental and physical states affect how we respond to odor stimulation?

Sensing an odor begins with the olfactory receptor neurons (ORN) located on the nasal epithelium (Wilson and Mainen, 2006). A single mouse possesses approximately 44 million ORNs (Albeanu, 2008, Mombaerts, 2006, Wachowiak, 2004), with each ORN contain-ing G-protein coupled receptors known as the olfactory receptors (OR). In order to remain functionally distinct, a single ORN only expresses one or perhaps a small number of different types of ORs (Albeanu, 2008, Mombaerts et al., 1996). Every OR can detect mul-tiple odors, and many types of ORs detect each odor. However, each odor is sensed by a unique combination of ORs that allows it to be distinguished from other odors (Malnic et al., 1999). The axons of the stimulated ORN transmit the sensory signal to approximately two locations in a single olfactory bulb known as glomeruli, which are simply the synapses joining the ORN axons with the dendrites of mitral/tufted (M/T) cells, which output the signal to the olfactory cortex (Albeanu, 2008, Mombearts et al. 1996). Since the glomeruli are spatially defined functional clusters, the glomerular layer can be viewed as an odor map for which each odor is encoded by a unique combination of glomerular activation (Mombaerts et al., 1996). Once the signal reaches the glomerular layer, it is passed to the mitral cell layer, which contains the M/T cells (Wilson and Mainen, 2006).

The structure of the olfactory system, however, is not a simple re-lay. Rather, the system is subject to lateral modulation from regulato-ry cells. Each glomerulus is surrounded by different types of regula-tory juxtaglomerular (JG) cells, including periglomerular (PG) cells and short axon (SA) cells (Albeanu, 2008, Wachowiak and Shipley, 2006). Periglomerular cells, the more populous of the interneurons in the glomerular layer, are inhibitory regulators (Albeanu, 2008,

Wilson and Mainen, 2006). PG cells receive stimulatory input from ORNs, M/T cells, and even other PG cells. They then transmit inhibi-tory signals reciprocally to the ORN axons and the M/T dendrites of the exciting glomerulus. The other type of juxtaglomerular cell is the short-axon cell, which indirectly inhibits M/T cells by exciting PG cells. Contrary to what the name suggests, a short axon cell is a long-range regulatory cell and may act on PG cells several glomeruli away (Albeanu, 2008, Aungst et al., 2003, Wilson and Mainen, 2006).

Mitral/tufted cells also communicate with each other indirectly and directly. M/T cells form dendrodendritic synapses with granule cells in the external plexiform layer, and can inhibit other M/T cells through granule cells (Wilson and Mainen, 2006). M/T cells may also directly influence their neighbors through a process known as spillover. When the M/T cells release glutamate through their den-drites, some of the glutamate excites dendrites of neighboring M/T cells (Isaacson, 1999, Wilson and Mainen, 2006).

Finally, the olfactory system is influenced by centrifugal inputs from the cortex and other areas of the brain (Isaacson, 1999). These inputs may be affected by the behavior of the mouse, genetic predis-positions, and other factors. Centrifugal influences can be separated into three major categories. First, odor intake is affected by areas of the brain outside of the olfactory bulb. Passive sniffing rate is deter-mined by central pattern generators in the medulla (Ramirez and Richter, 1996, Wilson and Mainen, 2006) while active respiration is controlled by the forebrain (Wilson and Mainen, 2006). Second, the bulb receives reciprocal feedback from output regions such as the olfactory cortex. Finally, changing levels of neuromodulators such as norepinephrine, acetylcholine, and serotonin can affect odor re-sponses (Wilson and Mainen, 2006).

The complexity of the olfactory system, designed to accomplish what may seem like a simple task, is astounding. This intricacy makes the system susceptible to outside influences such as alterations in be-havior. Most in vivo imaging research on the olfactory system exam-ines the bulb while an animal is in an anesthetized state. However, anesthesia greatly limits the scope of the field because it makes it im-possible to study the effect of behavior on olfactory response.

One way in which an animal’s behavioral state may affect its re-sponse to odors is by influencing the internal bulb activity. Recently, a team headed by Kenasku Mori studied the dendro-dendritic activ-

David Blauvelt




Neu

rosc

ienc

e

ity in the external plexiform layer (Tsuno et al., 2008). They found that mitral-granule activity was strongest in anesthetized animals and was progressively weaker in lightly sleeping, awake and immo-bile, and awake and mobile animals.

In addition to direct effects upon the olfactory bulb, behavior may affect centrifugal inputs. For instance, while the sniffing rate of an anesthetized animal is passively controlled by the brainstem, a be-having animal has the ability to actively control this rate. Further-more, the levels of norepinephrine, acetylcholine, and serotonin can be altered by different behavioral states (Wilson and Mainen, 2006). As an example, Shea et al. (2008) found that norepinephrine release, when coupled with odor presentation, acts in the olfactory bulb to cause suppression of paired odor responses.

To more fully understand the effects of different behaviors on odor responses, the olfactory bulb needs to be monitored while the animal is awake and behaving. While electrical activity in behaving animals has been extensively studied, they are limited to single cell readouts as opposed to imaging studies, which allow for large-scale population readouts. Sub-glomerular resolution imaging studies of the olfactory bulb in awake animals have been performed on head-fixed animals (Carey et al., 2004), but are limited. Furthermore, to date, sub-glomerular resolution imaging of the olfactory bulb in freely behaving animals has not yet been accomplished. The present study attempts to address this issue by offering a way to image the bulb in a freely behaving mouse.

Two important advances have made it possible to explore odor responses in freely behaving mice. The first was the creation of trans-genic mice expressing synaptopHluorin in the ORNs, which allowed the visualization of pre-synaptic glomerular activity in the olfactory bulb. pHluorins is a mutant form of GFP that was sensitive to pH and would fluoresce in a neutral environment but not in an acidic envi-ronment. SynaptopHluorin (spH) was then made by fusing pHluorin to synaptobrevin (Miesenböck et al., 2000). Synaptobrevin is a vesicle protein required for the release of neurotransmitters into the syn-apse. The idea is to take advantage of the acidic environment of the neurotransmitter vesicles (pH ~5.7). When inside the vesicles, spH would not fluoresce, but once a vesicle binds with the cellular mem-brane to release neurotransmitters into the neutral pH of synaptic space (~7.4), spH would fluoresce (Miesenböck et al., 1998). This laid the groundwork for the use of synaptopHluorin as a genetically en-coded molecular probe that would allow detection of neural activity using simple fluorescence microscopy.

Though there are many other ways to image ORN synaptic activ-ity such as nasal calcium dye injections, bulk calcium dye loading, and intrinsic signaling (Toga and Mazziotta, 2002, Albeanu, 2008), synaptopHluorin was used because of three significant advantages. First, it is genetically encoded unlike methods involving calcium dyes, which need to be exogenously loaded. Second, the signal is mostly specific to pre-synaptic activity. Finally, because it is localized to the axon termini, spH makes it possible to more easily distinguish individual glomeruli.

The second necessary development for this study was the intro-duction of an imaging technique using a flexible fiber optic imag-ing bundle attached to the skull. This method was chosen over other methods such as the head-fixed method (Carey et al., 2004) specifi-cally because it allows the mouse to freely move around while keep-ing the fluorescence image in the focal plane of the microscope.

The first fundamental question that this study attempts to answer is whether the fiber technique is a viable way to image freely behav-ing animals. This technology is still in its very early stages, and has never been used to image the olfactory bulb. However, after adapting and optimizing the procedure, this technique will presumably offer a

good way to image freely behaving animals.The second question of this study is whether behavior has an effect

on the pre-synaptic activity of glomeruli. As discussed earlier, there are several ways that behavior could influence ORN output. Thus, it seems likely that behavior will cause some change in glomerular activity, although attributing observed changes to specific individual behaviors as well as parsing out the causes is beyond the scope of this study. In order to test this hypothesis, the aforementioned fiber optic technique was used to image transgenic mice expressing syn-aptopHluorin in the glomeruli. In addition, a customized odor rack was employed to expose the mice to different odors while monitoring the fluorescence response in real-time. The same mice were imaged in both an awake and an anesthetized state, and the responses were compared.

Fundamentally, the goal of this study is simply to open the doors to a burgeoning field of new imaging techniques while answering some questions about how behavior can affect olfactory sensory ac-tivity. First, this project attempts to determine whether a fiber op-tic imaging bundle provides a way to image the olfactory bulb of an awake, freely-moving animal. Second, this study uses the bundle to track olfactory receptor neuron pre-synaptic activity in anesthetized and freely-behaving mice to find similarities and differences based on behavioral state.

Methods

Subjects and surgeryThe subjects used were transgenic adult (postnatal days 60-100)

synaptopHluorin mice, including both heterozygous and homozy-gous as well as male and female mice (Bozza et al., 2004). Mice were anesthetized for surgery with a cocktail of ketamine and xylazine (ketamine - 100 mg/kg, IP, Fort Dodge Animal Health #440761; 10 mg/kg xylazine, IP, Phoenix Pharmaceutical, Inc. #4111505). The mice were mounted on a stereotaxic frame and the skin was cut to expose the skull. A small hole, approximately 1.5-2 mm in diameter, was opened over the right olfactory bulb by thinning the bone in a circle and removing the piece. A flexible fiber bundle (Schott Inc., #1137189) with a 1.45 mm outer diameter was lowered onto the brain. The other end of the bundle was placed in the focal plane of a wide-field fluorescence microscope fitted with a 10x objective and a high speed imaging camera (SensiCam, Cooke Corp.). The bundle was lowered using an XYZ translator until glomeruli became clear. The fiber bundle provided a continuous two-way light path from the mi-croscope to the glomeruli and back to the microscope, allowing the mouse to move around away from the microscope while still keep-ing an image of the glomeruli within the focal plane of the micro-scope. Once the bundle was in the proper place on the brain, it was cemented onto the skull with RelyX Luting Plus (3M ESPE, #3525). The mouse was allowed to recover for several hours before the analge-sic, Buprenorphine HCl (0.5 mg/kg, BD, IP, Bedford Labs, #1208141), was given. Anesthetized imaging occurred immediately after surgery while the mouse was still anesthetized. Behavioral imaging was per-formed after recovery.Imaging fiber optic bundle

There are two issues regarding the properties of an imaging fiber bundle that need to be addressed. The first is the pixelation of the image. The bundle was composed of thousands of microscopic fibers, each 8 microns in diameter. This led to pixelation that was visible un-der 10x magnification. However, since the pixels were much smaller than the glomeruli (approximately 90 microns), the pixelation did not significantly interfere with the overall quality of the image. The second issue is the lack of focusing optics in the bundle itself. Ide-


41www.thurj.org

Neuroscience

ally, one would use microscopic lenses to focus the light and prevent cross-contamination of signals from nearby glomeruli. Fortunately this was not a problem, since the distance between the bundle and glomerular layer was sufficiently small. As demonstrated by the re-sults, the images obtained were clear enough to distinguish individ-ual glomeruli. Odor delivery apparatus

In order to systematically present odors, a computer-controlled olfactometer (Figure 1) was designed and constructed. The apparatus contained tygon tubing (McMaster-Carr 5046K11) and two main sets of solenoid valves (ASCO Scientific AL4124). The first set consisted of ten valves, each associated with a single odor. When a valve was opened, air would pass through the valve, going into a test tube con-taining the odor and out via another path, carrying the odor with it. The odor would then travel to a second set of valves, which included an exhaust valve, an odor valve, and an air valve. The odor could ei-ther go through the odor valve or the exhaust valve. If the odor valve was open, the odor would be presented to the animal. Alternatively, if the exhaust valve was open, the odor would leave the system through a ventilation system. The air valve was connected to an air line, allow-ing clean air to be presented to the animal. Aside from the two main sets of valves, there was a cleaning valve.

Each trial involved four different phases that were repeated: clean-ing, air presentation 1, odor presentation, and air presentation 2.

During the cleaning phase, the cleaning valve was opened to allow fresh air to flow through the system. In addition, the air valve was open, providing fresh air to the animal. During air presentation 1, the air valve remained open and images were taken to obtain an average baseline fluorescence image for comparison to odor images. During the odor phase, the exhaust and air valves were closed, and the odor valve was opened. The odor flowed through the second set of valves to the animal. Finally, during air presentation 2, the odor valve was closed, and the air valve was re-opened, allowing the fluorescence to return to baseline while images were taken to track the return.Software and experimental design

The odor delivery control and image acquisition was performed by a software program written in LabView (National Instruments) adapted by Tomokazu Sato from a similar program by Edward Soucy. Each experiment could be altered by changing the settings of the pro-gram. During the course of the experiment, the mouse was placed in a small chamber to minimize movement during awake imaging and to maximize and homogenize odor exposure. The air in the cham-ber was constantly evacuated throughout the experiment. However, some delay in clearing an odor from the chamber was expected. While the concentration was not measured, it was assumed that the odor was expunged from the chamber quickly, likely on the order of a few seconds or less.Analysis

Analysis was done primarily using ImageJ (National Institutes of Health) and Microsoft Excel. Time course image stacks were col-lected and used to track the responses in real time. Ratio images were used to quickly identify the presence or absence of responses as well as the strength of response. These images are displayed as functions

Figure 1. Odor rack schematic. During the cleaning phase, air passes though valve 6, cleaning the tubing. Air and contaminants are flushed from the system though valve 3, the exhaust valve. During the air1 phase, either valve 1 or 2 is opened. Odor fills the system but goes out through valve 3. In addition, valve 5 is opened, allowing air to flow to the animal. During the odor phase, valve 1 or 2 remains open, but valves 3 and 5 are closed, replaced by the opening of valve 4, which allows odor to flow to the animal. During air2, valve 4 is closed and valves 3 and 5 are reopened. In addition valve 6 is opened, starting the cleaning process. This process is repeated for each odor.

Figure 2. Odor responses can be visualized with the fiber bun-dle. A) Image of the olfactory bulb as seen through the fiber bundle. Ar-rows point to sample glomeruli. B) Ratio image indicating ΔF/F values. Image is inverted with dark spots indicating an increase in fluorescence. Responding glomeruli can be seen as dark circles. Odor used was isopro-pyl tiglate; subject was the mouse from A. C) Enlargement of the image enclosed by the square in A. The contrast has been increased to more easily visualize fluorescence differences. The circle encloses a glomerulus at rest-ing fluorescence. D) Similar enlargement showing the same glomerulus after odor stimulation. Contrast was increased by the same amount as in C. Comparing the two images reveals a greater amount of fluorescence after odor stimulation. This glomerulus is also the dark spot in the middle of B. E) Graph tracking fluorescence over time. Gray arrow indicates the start of odor stimulation (10 seconds) and the black arrow indicates the stop (30 seconds). This time course was not obtained from the same glom-erulus as in C and D.



Neu

rosc

ienc

e

of the inverse of ΔFluorescence/Base Fluorescence (ΔF/F). Hence, dark spots indicate an increase in ΔF/F. Ideally, one would like to subtract out all fluorescence that is not attributable to glomeruli. However, since ratio images compared images taken within minutes of each other, it was assumed that background fluorescence changed minimally.


Efficacy of fiber bundle imagingUsing a fiber optic bundle to image the brain of an awake, behav-

ing animal is an underdeveloped technique. Hence, there are several potential complications that may arise in its application to this study. First, the fiber method has never been used to image glomeruli. It is hard to predict whether individual glomeruli will be distinguishable because of their proximity to each other and because of variable opti-cal properties of the imaging fiber. Secondly, due to less constraining structural support, the olfactory bulb may be more subject to move-ment when the mouse moves its head. If the bulb were constantly moving relative to the fiber, it would be difficult to maintain clear images throughout a chronic experiment. Furthermore, movement of the brain during the course of imaging may introduce motion ar-tifacts, interfering with results. A third potential problem with the fiber technique is that the stressful nature of the surgery may cause the mice to experience stress, pain, or fear upon recovery. This may cause them to alter their behavior in response, which could affect not only odor intake, but also neural response. While these are interest-ing behaviors that can be studied, too high a level of stress, pain, or fear may limit other behaviors such as natural exploration.

The first goal in the development of the technique was to repro-duce detailed images comparable to simple wide-field images. The concerns about pixelation and lack of optics mentioned earlier would affect the quality of the images. However, the fiber technique was suc-cessful in this respect, and consistently produced clear images, often showing clear glomeruli (Figure 2A).

The next important step was to be able to consistently see glom-erular odor responses. (For the rest of this paper, mentions of “glom-erular responses” means a change in ORN pre-synaptic activity as a result of odor stimulation.) This was also successful (Figure 2B-E), but not as much as simply achieving clear images. While several ani-mals showed clear responses, others did not. Often individual ani-mals or even litters do not respond very well to odor stimulation for a multitude of possible reasons such as sickness, obstruction of the na-sal passages, or other variables. Furthermore, the surgery to implant the fiber bundle is very invasive, and often damage to the dura matter of the brain could cause a failure to respond. This could be due to ac-tual damaging of the glomeruli or, more likely, due to decreased im-age quality caused by dura matter damage. Finally, the bundle does

not cover the entire bulb, so even strong odors may only stimulate glomeruli outside of the field of view. In order to correct for this, a test trial was run before cementing the bundle. If the testing revealed limited responsiveness, the fiber was moved and retested until either responses were seen or it was determined that the mouse would not respond, in which case it was euthanized.

The final step was to determine whether the fiber technique was suitable for imaging awake, freely behaving animals. Strong responses similar to those seen in anesthetized animals were observed. Howev-er, there were several shortcomings, as predicted. One major problem is the small size and high curvature of the olfactory bulb, which cre-ates problems with maintaining focus. The olfactory bulb has space to move around, and during the awake trials, images often contain motion artifacts due to brain movement relative to the fiber (Figure 3A,B). There were two types of motion artifacts observed. The most common was caused by horizontal movement of the brain. When the brain shifted horizontally, so too did the glomeruli in the field of view. The movement was manifested in ratio images as a pattern of very dark and bright spots (Figure 3A). Another type of movement artifact was caused by vertical movement of the brain. Vertical move-ment caused glomeruli to come in and out of focus. If a glomerulus came into focus, the ratio image would show a dark spot at the new position (Figure 3B).

Another focusing problem that was caused by the size and shape of the bulb was simply maintaining the focus over a long time period. Given the mobility of the bulb, it was hard to keep glomeruli in focus even for a day. By the time a mouse fully recovers, the clarity of an image may be lost (Figure 3C,D). While several glomeruli may still be visible, others will be lost. This is one reason to keep the surgery and recovery as short as possible. Usually, an entire experiment can be finished within one day, preventing problems such as this.Glomerular pre-synaptic activity in anesthetized and behaving mice

As discussed earlier, one of the fundamental concerns of this study is to determine whether behavior has an effect on ORN output. To answer this question, images from three mice were analyzed. The images from these three mice were chosen because they were the only ones that matched a set of three criteria necessary to enhance the probability of getting reliable results. First, the images remained clear in both the behaving and anesthetized trials. Second, clear glomeru-lar responses were observed. Finally, the animals were behaving rela-tively normally after recovery and did not display signs of excessive

Figure 3. Olfactory bulb’s size and shape make it difficult to get reliable results. A) Movement artifact caused by horizontal movement of the brain. As the glomeruli shift, their original position becomes darker because of the loss of base fluorescence. By contrast, their new position is brighter. This is seen as a characteristic dark and bright pattern in the ratio image. B) Movement artifact caused by vertical movement of the brain. In this case, the glomeruli came more into focus, causing fluorescence in-crease. Vertical movement is problematic, because it often manifests as an odor response. Note the halo effect. C) Image of the olfactory bulb taken immediately after surgery. Blood vessels and glomeruli are in focus. D) Image of the same olfactory bulb taken one day after surgery. While many glomeruli are still visible, others (arrows) may disappear out of focus.

43www.thurj.org

Neuroscience

stress or pain.The first characteristic of glomerular response studied was the pat-

tern of the responses. Glomerular activation patterns for individual odors were compared in the same animal to determine if the patterns were similar in anesthetized and behaving animals. It was predicted that the patterns would be the same. While behavior may affect the feedback loops that can modulate the signal, it seemed unlikely that behavior would alter odor-ORN binding affinities or the wiring of the ORN. In all three mice it was found that the patterns were con-served in behaving animals (Figure 4).

The next property studied was whether or not the signal strength was significantly different when the animal was awake as compared to when the animal was anesthetized. It was predicted that there would be a difference, although whether it would be stronger or weaker was unknown. ΔF/F values were obtained for several responding glom-eruli in each of the three animals and averaged over 10 trials. Then, the mean values from the anesthetized and behaving conditions were compared. However, since ORN output can vary significantly for different odors and even different glomeruli, comparison was only done between the same glomeruli in each animal for identical odors. Also, in order to correct for motion artifacts, data from trials with large motion artifacts were removed. Somewhat contrary to what was predicted, the results (Figure 5) indicated that only one glomerulus

responded to one odor, ethyl valerate, significantly more strongly in the anesthetized condition (p<0.05 2-tailed paired t-test). The other 11 glomerulus-odor combinations examined had statistically insig-nificant differences.

While the pattern and intensity of the responses appeared to be unaffected by behavior, it was predicted that the nature of the re-sponse over time might be affected. In order to evaluate this, time course data was examined for patterns across glomeruli and animals. Unfortunately motion artifacts greatly limited the quantity of usable data, making it difficult to draw global conclusions

One major difference noted in the time courses was that the slope of the fluorescence change during odor presentation varied. In ani-mal 2, stimulation with ethyl valerate in the anesthetized condition caused a fluorescence change with a steeper slope (Figure 6). By con-trast, in animal 3, stimulation with isopropyl tiglate in the anesthe-tized condition caused a change with a shallower slope (Figure 7). For both animals, this change was seen across all glomeruli but only during the first repeat. Subsequent trials were too variable to draw any conclusions. In addition, data would ideally come from the same odor. Hence, this difference cannot be attributed to animal differenc-es or odor differences. Nevertheless, the differences between the two conditions in both animals are strikingly large. Interestingly, it seems that the slopes all fall around 0.05 for both animals in the anesthe-tized conditions. By contrast, it seems that the variance comes mostly from the behaving condition, which has a slope of only 0.025 in ani-mal 2 but approximately 0.1 in animal 3. There are several possible explanations for this. One possibility is that animal 3 may have had a higher sniffing rate than animal 2. It is possible that animal 3 found isopropyl tiglate more pleasing than animal 2 found ethyl valerate. Animal 3 could have also been more exploratory or aware than ani-mal 2, which may have not only increased its sniffing rate but also its mental awareness. A further study could test this theory by monitor-ing sniffing rate and comparing it to the slope of the response.

The final characteristic examined was rate of recovery from odor stimulation. Only animal 2 was analyzed because the data from ani-mals 1 and 3 had too much noise or large motion artifacts during re-

Figure 4. Glomerular response patterns are similar in both anesthetized and behaving animals. Shown are the Z-stack projec-tion averages derived from ΔF/F images for each of the three mice over 10 trials. Each row represents images from a different mouse. The left column displays images taken from an anesthetized mouse and the right column displays images from the same mouse while behaving. Some of the image slices were removed from the projections because of large motion artifacts. While some motion artifacts remain, the response map appears to be unaf-fected by behavior.

Figure 5. ORN output is similar in the anesthetized and behav-ing conditions. Shown is a bar graph pairing the mean ΔF/F values for the anesthetized and behaving conditions for various animals (A), odors (O), and glomeruli (G). These 11 glomeruli were chosen because they showed some response in either the awake or the anesthetized condition. Only A1-O2-G3 showed a statistically significant difference between the anesthetized and behaving conditions (p<0.05 2-tailed paired t-test). Odor 1 (O1): Isopropyl Tiglate; Odor 2 (O2): Ethyl Valerate. Error bars denote standard error.



Neu

rosc

ienc

e

covery. Also, other than a few exceptions that will be discussed later, only the first trial was analyzed for the same reasons. To quantify the rate of recovery, the downward slope was calculated. It is important to note that the data included in this analysis is somewhat arbitrary because there is no time point defined as the peak and each glomeru-lus peaks at a different time. For both odors and across glomeruli, the odor recovery was slower during the behavior trial (Figure 8). In fact, for isopropyl tiglate, the odor responses did not peak over the course of the entire trial. The small differences in slope and the low R-squared values are less than ideal for drawing solid conclusions. However, when viewed in combination with the time course graphs, there does seem to be a trend for faster recovery under anesthesia. Furthermore, the lowest R-squared values were seen in the behav-ing condition, and as discussed earlier, behavior seems to cause large fluctuations in the signal as the animal recovers from odor stimula-tion. Furthermore, since the slope is often close to horizontal, these vertical fluctuations can significantly affect the correlation coefficient because the mean variance in the y-direction is determined almost completely by these fluctuations. By contrast, as the slope of the trend-line gets steeper, the total variance in the y-direction increases. This means that the proportion of the total variance that is determined by the variance of these fluctuations decreases, thus increasing the R-squared value. Finally, the R-squared value was calculated based on a total sum of squares that only took into account variance of the data

points used to create the trendline. It can be argued, however, that total sum of squares ought to be based on all data points, since that is the true variability of fluorescence. In this case, the R-squared values would be higher. All of this would suggest that even low R-squared values do not necessarily signify an erroneous trend, especially in the cases where the slope is very close to zero. The difference in recov-ery time could be the result of several factors. One possibility is that active exploration after the stimulus presentation may play a role in odor intake or neural response. The expunging of the odor from the stimulus chamber does not happen instantly. Rather, after the odor presentation, a weak vacuum evacuates residual odorants out of the chamber. It is unclear at what concentration or time point the mouse is no longer able to sense the odor, so it is possible that even as the concentration dwindles, the mouse increases its sniffing rate, caus-ing the effect observed. However, as mentioned earlier, while odorant concentration was not measured, it is assumed that it was expunged relatively quickly. Given this, another explanation is that the mouse may actively suppress inhibition in an attempt to try to maintain sen-sation of the smell even after the odor is cleared.

Figure 6. Animal 2 has a steeper odor response slope in the anesthetized condition. A) Time courses for the two responding glomeruli in animal 2. Gray arrow indicates stimulus start; black arrow indicates stimulus end. B) Table showing the slopes and R-squared values. The interval for which each slope was calculated varied based on the start and end of the response. The slopes for the anesthetized condition were roughly twice those of the awake condition. Slope is expressed as change in fluorescence over time.

Glomerulus 1 Glomerulus 2

Slope R2 Slope R2

Anesthetized 0.046 0.91 0.053 0.97

Behaving 0.025 0.71 0.025 0.92

Glomerulus 1 Glomerulus 2 Glomerulus 3 Glomerulus 4

Slope R2 Slope R2 Slope R2 Slope R2

Anesthetized 0.038 0.97 0.061 0.99 0.043 0.96 0.031 0.96

Behaving 0.074 0.96 0.091 0.98 0.119 0.96 0.076 0.96

Figure 7. Animal 3 has a steeper odor response slope in the be-having condition. A) Representative time courses from glomerulus 1 of animal 3. Gray arrow indicates stimulus start; black arrow indicates stimu-lus end. B) Table showing the slopes and R-squared values. The interval for which each slope was calculated varied based on the start and end of the response. In this case, the slopes for the behaving condition were about twice those of the anesthetized condition, which is the opposite of the re-sult found in animal 2.

B.

B.


45www.thurj.org

Neuroscience

Conclusions and Future Directions

From the beginning, this study has had two goals. The first goal was to explore the use of a fiber optic imaging bundle as a viable way to image the olfactory bulb in awake, freely-moving mice. The second was to investigate any differences in glomerular response based on whether the mouse was anesthetized or awake and freely-behaving.

This project is the first demonstration that a fiber optic bundle can be used to image the olfactory bulb of awake, freely-behaving mice. The results illustrate that not only can clear images showing distinct glomeruli be attained, but also changes in fluorescence can be ob-served, captured, and quantified. Furthermore, the unique advantage that the fiber bundle offers over other awake imaging methods such as the head-fixed method is that mice are able to freely explore their environments. Given results that indicate that exploratory behavior may have a significant effect on odor response, the fiber technique offers a way to further investigate this possibility.

However, since this method is still in its early stages, some prob-lems remain. One issue not indicated by the presented results is that the surgery required to install the bundle on the mouse is incredibly difficult. Obtaining and maintaining clear images is a sensitive pro-cess requiring a clean and rapid surgery. Also, as shown in the results, motion artifacts could interfere with results. In order to make the fiber technique practical, it is necessary to optimize the technique to address issues such as motion artifacts. However, the success of this technique in its first application to the olfactory bulb is a testament to its potential as an invaluable tool for imaging behaving animals.

The second major finding of this study is that generalized free be-havior does not seem to cause any consistent differences in glomeru-lar responses across all animals, odors, and glomeruli. Both the pat-tern of the response as well as the overall strength of response were unchanged by the behavioral state. As discussed earlier, one of the main reasons for looking at ORN synaptic output was because ORNs are more upstream than the rest of the olfactory system. Knowing that odor responses do not appear to be globally modulated at the level of the ORN, it is now possible to study the mitral cells to see if a global change occurs in that layer. If a difference is observed, one can more reliably attribute the change to mitral cell modulation as opposed to a modification in ORN activity. One way to explore this possibility would be to use transgenic mice expressing the calcium-

sensitive dye, GCaMP, in mitral cells.The final major finding was that behavior causes localized changes

individual to different mice, glomeruli, or odors. In two animals, a wider variability was observed after the average response began to stabilize. Differences in the rate of odor response were also seen in two mice. One mouse responded more quickly when it was behaving and the other responded more slowly. Finally, a difference in odor recovery rate was observed. For both odors presented, one of the mice recovered from odor stimulation more slowly when it was behav-ing. Interestingly, for one of the odors, differences were found even amongst trials for the same glomeruli. In two of the trials, relative to when it was anesthetized, the mouse recovered more slowly when behaving, but in the third, it recovered more quickly. Furthermore, the recovery rate in the anesthetized condition did not vary much. Rather, the discrepancy was caused by a large variation in recov-ery rate between free-behavior trials. This trend of large variations in odor response in the freely-behaving condition was also seen in the fluctuations of fluorescence, as mentioned earlier. In addition it was seen in the rate of odor response. While the slopes for the anes-thetized condition were similar for both animals, the slopes for the freely-behaving condition were dramatically different. Thus it seems that while differences are individualized and seemingly unpredict-able on a global scale, the general trend seems to be that behavior causes variation in the nature of the odor response.

The field of imaging awake animals holds an exciting future. Little is known about how behavior and brain activity are linked and count-less questions remain unanswered. However, as mentioned earlier, before these questions can be addressed, the fiber technique needs to be optimized. One way the method can be improved is by enhancing the image quality. There are several potential ways to accomplish this. Adding focusing optics such as a gradient-index (GRIN) lens would not only increase the image quality by reducing cross-contamination of light, but it would also allow the fiber to be placed above the bulb as opposed to directly on it. In addition, adding a mechanism by which the fiber can be focused after a surgery is performed would decrease the number of experiments that fail because of bulb movement.

The preliminary results of this study inspire several future stud-ies that can further explore some of the findings presented. Since the results indicate that exploratory behavior, possibly sniffing rate, may have an effect on odor response, a study that concurrently monitors

Glomerulus 1 Glomerulus 2

Slope R2 Slope R2

Anesthetized -0.006 0.60 -0.017 0.95

Behaving -0.002 0.07 -0.011 0.83

Figure 8. Animal 2 recovered from odor stimulus faster in the anesthetized state. A) Representative time courses taken from the first trial of glomerulus 1 and odor 1, isopropyl tiglate. Gray arrow indi-cates stimulus start; black arrow indicates stimulus end. B) Chart showing the slope and R-squared values for each glomerulus stimulated by odor 1. Note the positive slopes in the behaving condition. C) Chart showing the slope and R-squared values for each glomerulus stimulated by odor 2, ethyl valerate.

Glomerulus 1 Glomerulus 2 Glomerulus 3 Glomerulus 4

Slope R2 Slope R2 Slope R2 Slope R2

Anesthetized -0.006 0.66 -0.003 0.33 -0.008 0.62 -0.003 0.25

Behaving 0.003 0.28 0.003 0.46 0.003 0.27 0.002 0.15

B.

C.



Neu

rosc

ienc

e

sniffing rate and glomerular response could shed some light on this variation. One potential outcome is that sniffing rate may match closely with the level of fluorescence observed, which would explain why different mice had different odor responses while freely behav-ing. To further explore this, one could use calcium dyes to more clearly pair sniffing with ORN output. While synaptopHluorin is a good indicator of overall ORN activity, it cannot be used to image individual ORN action potentials. By contrast, calcium dyes react quickly to ORN activity, and one can visualize events on a much shorter time scale, including individual action potentials.

As discussed earlier, the biggest difference between the head-fixed technique and the fiber bundle technique is that with a fiber bundle, mice can freely move and behave. Thus, it would be interesting to ex-amine whether the two techniques produce differences in glomerular response. Hypothetically, any significant differences can be attribut-ed to exploratory behavior, and this may help to resolve whether the differences seen in this study are attributable simply to being awake or rather to being able to move around and freely explore the envi-ronment.

Finally, one could explore the effect of training on odor responses. A basic study could compare mice that are previously exposed to the experimentation setup to naïve mice. It is conceivable that mice that trained in this fashion may reduce their exploratory behavior because the environment they are experiencing is not a novel one. Another study could compare mice trained to a certain odor to mice that are just trained to general odor stimulation or even naïve mice. This might also have an effect on how a mouse approaches and ex-plores the different odors presented.

This project is the first step towards understanding the effect of behavior on the olfactory bulb’s response to odor stimulation. While the field is new and there is a large amount of unexplored territory, hopefully this study will lay the groundwork for similar future stud-ies. Given the promising results of this project, the fiber technique has the potential to offer a way to understand the olfactory bulb in the larger context of the whole brain and even the entire body.

ReferencesAdrian, E.D. (1950). The Electrical Activity of the Mammalian Olfactory

Bulb. Electroencephalography and Clinical Neurophysiology 2, 377-388.Albeanu, D.F. (2008) Precision and diversity in an odor map on the olfactory

bulb [Thesis]. Harvard University. 96pp.Aungst, J.L., Heyward, P.M., Puche, A.C., Karnup, S.V., Hayar, A., Szabo, G.,

and Shipley, M.T. (2003). Center–surround inhibition among olfactory bulb glomeruli. Nature 426, 623-629.

Belluscio, L., and Cummings, D.M. (2008) Charting Plasticity in the Regen-erating Maps of the Mammalian Olfactory Bulb. Neuroscientist 14, 251-263.

Bouret, S., and Sarah, S.J. (2004). Reward expectation, orientation of atten-tion and locus coeruleus-medial frontal cortex interplay during learning. European Journal of Neuroscience 20, 791-802.

Bozza, T., McGann, J.P., Mombaerts, P., and Wachowiak, M. (2004). In Vivo Imaging of Neuronal Activity by Targeted Expression of a Genetically En-coded Probe in the Mouse. Neuron 42, 9-21.

Carey, R.M., Verhagen, J.V., Wesson, D.W., Pirez, N., and Wachowiak, M. (2009). Temporal structure of receptor neuron input to the olfactory bulb imaged in behaving rats. Jorunal of Neurophysiology 101, 1073-1088.

Ferezou, I., Bolea, S., and Peterson, C.C.H. (2006). Visualizing the Cortical Representation of Whisker Touch: Voltage-Sensitive Dye Imaging in Free-ly Moving Mice. Neuron 50, 617–629.

Flusberg, B.A., Nimmerjahn, A., Cocker, E., Mukamel, E.A., Baretto, R.P.J., Ko, T.H., Burns, L.D., Jung, J.C., and Schnitzer, M.J. (2008). High-speed, miniaturized fluorescence microscopy in freely moving mice. Nature Methods 5, 935-938.

Guerin, D., Peace, S.T., Didier, A,., Linster C., and Cleland, T.A. (2008). No-radrenergic neuromodulation in the olfactory bulb modulates odor ha-bituation and spontaneous discrimination. Behavioral Neuroscience 122, 816-826.

Issacson, J. (1999) Glutamate Spillover Mediates Excitatory Transmission in the Rat Olfactory Bulb. Neuron 23, 377-384.

Isaacson, J., and Strowbridge, B.W. (1998). Olfactory Reciprocal Synapses: Dendritic Signaling in the CNS. Neuron 20, 749-761.

Malnic, B., Junzo, H., Sato, T., and Buck, L. (1999). Combinatorial Receptor Codes for Odors. Cell 96, 713-723.

Miesenböck, G., DeAngelis, D.A., and Rothman, J.E. (1998). Visualizing se-cretion and synaptic transmission with pH-sensitive green fluorescent proteins. Nature 394, 192-195.

Mombaerts, P. (2006). Axonal Wiring in the Mouse Olfactory System. Annual Review of Cell and Developmental Biology 22, 713-737.

Mombaerts, P., Wang, F., Dulac, C., Chao, S., Nemes, A., Mondelsohn, M., Edmondson, J., and Axel, R. (1996). Visualizing an Olfactory Sensory Map. Cell 87, 675-686.

Ramirez, J., and Richter, D.W. (1996). The neuronal mechanisms of respira-tory rhythm generation. Current Opinion in Neurobiology 6, 817-825.

Rinberg, D., Koulakov, A., and Gelperin, A. (2006). Sparse Odor Coding in Awake Behaving Mice. Journal of Neuroscience 26, 8857-8865.

Shea S.D., Katz L.C., and Mooney R. (2008). Noradrenergic induction of odor-specific neural habituation and olfactory memories. Journal of Neu-roscience 28, 10711-10719.

Toga, A.W., and Mazziotta, J.C. (2002). Brain Mapping: The Methods. San Diego: Academic Press.

Tsuno, Y., Kashiwadani, H., and Mori, K. (2008) Behavioral State Regulation of Dendrodendritic Synaptic Inhibition in the Olfactory Bulb. Journal of Neuroscience 28, 9227-9238.

Wachowiak, M., and Shipley, M.T. (2006). Coding and synaptic processing of sensory information in the glomerular layer of the olfactory bulb. Semi-nars in Cell & Developmental Biology 17, 411-423.

Wachowiak, M., Denk, W., and Friedrich, R.W. (2004). Functional organiza-tion of sensory input to the olfactory bulb glomerulus analyzed by two-photon calcium imaging. PNAS 101, 9097-9102.

Wilson, R. and Mainen, Z., (2006). Early Events in Olfactory Processing. An-nual Review of Neuroscience 29, 163-201.

47www.thurj.org

Neuroscience

In neural development, Bhlhb5 and Prdm8 may have a cooperative role in the same regulatory processes underly-ing neuronal differentiation and maturation. Three potential models of interaction are proposed. The three models are: 1) Bhlhb5 and Prdm8 regulate the same target genes, but do not bind together in complex, 2) Bhlhb5 and Prdm8 interact in complex to regulate the same gene targets, and 3) Prdm8 and Bhlhb5 each regulate the transcription of unique proteins that are critical mediators of a common pathway. To address the hypothesis that Bhlhb5 and Prdm8 are crucial in the same neurons in development, the protein expression patterns of Bhlhb5 and Prdm8 were examined in the mouse brain and spinal cord via immunohistochemistry. Results show that Bhlhb5 and Prdm8 do not bind directly together in complex. Moreover, Prdm8-expressing neurons are significantly lost in the spinal cord of Bhlhb5 mutants, Bhlhb5 and Prdm8 are highly co-expressed in the brain and spinal cord, and Prdm8 mutants exhibit loss of the corticospinal tract. These results suggest that Bhlhb5 and Prdm8 are crucial regulators of a common pathway that underlies neuronal development in the mammalian central nervous system.

IntroductionThe bHLH family of transcriptional regulators

One group of transcriptional regulators found to be critical in the early development of the central nervous system is a family of proteins known for their highly conserved Basic-Helix-Loop-Helix or BHLH structural domains. Many members of the BHLH fam-ily have been found to perform crucial regulatory roles in nervous system development in a broad array of cellular types (Massari and Murre, 2000). A particular BHLH gene of interest is Bhlhb5, which has been found to be transiently expressed in subpopulations of post-mitotic neurons throughout the central nervous system (Ross et al., unpublished data). These findings suggest that the Bhlhb5 transcrip-tion factor may have a regulatory role in cellular differentiation and maturation during neuronal development of key pathways necessary for proper movement and sensory mechanisms.Bhlhb5 mutants have phenotypes indicating deficits in the ner-vous system

Investigators in the Greenberg Lab examined the potential role of Bhlhb5 in neural development through the generation of a Bhlhb5 knockout mouse (Bhlhb5 -/-). They found that Bhlhb5 mutants ex-hibited a series of physical and behavioral deficits (Figure 1: Supple-mentary Figures). Behavioral phenotypes observed include self injury due to excessive spot-grooming and the subsequent development of open sores, or skin lesions (Figure 1A). Moreover, in behavioral tests that evaluated sensitivity of Bhlhb5 mutants to chemical stimulation through exposure to capsaicin, a natural irritant found in chili pep-pers, Bhlhb5 mutants were found to demonstrate decreased response (measured by paw licking) to the harmful stimulant. Similarly, Bhl-hb5 mutants also exhibited decreased sensitivity to thermal stimu-lation as indicated by the decreased rate of paw withdrawal when placed on a hot plate (Ross et al., unpublished data).

An unusual behavioral phenotype observed in the Bhlhb5 mu-tants was a “handstand” behavior in which the mice walked forward on their forelimbs, while retracting their hindlimbs to display a tran-sient handstand posture (Figure 1C). Bhlhb5 mutants also displayed lack of motor coordination when compared to wildtype controls.

This was indicated by a significant decrease in balancing skills when Bhlhb5 mutants were placed on a rotating rod and fell off at much higher rates than wildtype animals (Figure 1B).

While the nervous systems of Bhlhb5 mutants appeared normal at a gross level, closer inspection revealed several defects at the cellular level. For instance, a statistically significant loss of Bhlhb5-expressing neurons in the most superficial laminae of the dorsal horn of the spi-nal cord was observed in Bhlhb5 mutants when compared to that of wildtype controls (Figure 2). In addition, the corticospinal tract was missing in Bhlhb5 mutants (Ross et al., unpublished data) (Figure 3). The corticospinal tract (CST) consists of the bundle of neuronal fibers extending from the motor cortex of the brain into the spinal cord, and is therefore a key factor in regulating motor coordination in animals. The distinct absence of the CST in mutants thus indicates a potential role of Bhlhb5 in the formation of motor circuits in de-velopment. To identify the genes that are mis-expressed in Bhlhb5 mutants, Affymetrix array analysis examining RNA levels in Bhlhb5 knockout versus wildtype was performed. Prdm8 was the most dra-matically misregulated gene (Sup. Figure 1) (Ross et al., unpublished data). In fact, Prdm8 was found to be up-regulated significantly in Bhlhb5 mutants (p<0.05) (Ross et al., unpublished data), thus sug-gesting that Bhlhb5 may directly repress the transcription of Prdm8 (Figure 4: Supplementary Figures). Similarity of phenotypes in Bhlhb5 and Prdm8 mutants

The Affymetrix RNA array results from the Bhlhb5 mutants sug-gested that Prdm8 may be an important transcriptional regulator in the same neuronal pathways as that of Bhlhb5. A putative transcrip-tion factor that is a member of the PR Domain family known for its histone methyl-transferase activity, Prdm8 has not been researched broadly and little is known about its role in the central nervous sys-tem.

However, Prdm8 analysis in the retina was being explored by in-vestigators at the McInnes Lab; intriguingly, Prdm8 knockout mice exhibited similar behavioral phenotypes and deficits when compared to that of Bhlhb5 mutants (Sup. Figure 2). Like Bhlhb5 knockout mice, the Prdm8 mutants also displayed the unique “handstand” posture during movement. Furthermore, Prdm8 mutants also devel-

Analysis of the role of Bhlhb5 and Prdm8 in neural development

Stephanie I. Mok



Neu

rosc

ienc

e

oped extensive skin lesions like the Bhlhb5 mutants. The combina-tion of these observations in conjunction with the RNA Affymetrix array results indicating that Prdm8 is the most mis-regulated gene in Bhlhb5 mutants suggested that Prdm8 may be involved in a com-mon pathway with Bhlhb5 in the development of the central nervous system in mice.

What is the mechanism through which Bhlhb5 and Prdm8 regu-late transcriptional pathways such that removal of either factor in-duces the same phenotypic deficits in knockout mice? I propose three models of interaction to answer this question (Figure 4).

Three chief objectives were addressed: 1) Which model of interac-tion between Bhlhb5 and Prdm8 is best supported by the observed data? In Figure 8, several models of interaction that may characterize the true collaborative function of Bhlhb5 and Prdm8 in neuronal de-velopment are presented. Is there evidence to support the hypothesis that Bhlhb5 and Prdm8 regulate the same target genes, yet do not bind together directly in complex (Figure 4A)? Do they interact di-rectly with each other to regulate the same gene targets (Figure 4B)? Or, do Prdm8 and Bhlhb5 each regulate distinct groups of genes that are critical mediators of a common pathway (Figure 4C)?

2) Is there evidence to suggest a mechanism of interaction between Bhlhb5 and Prdm8 in the regulation of processes important neuronal development? Do protein expression patterns of Bhlhb5 and Prdm8 in various regions of the brain and spinal cord indicate that these two factors may be crucial in the same neurons during differentiation and development?

3) To design a Prdm8 fusion protein construct for the develop-ment of an antibody that accurately recognizes Prdm8 protein in immunhistochemistry and immunoprecipitation experiments and allows for visualization of endogenous Prdm8 protein in fixed tissue and neural lysates.

Materials and Methods1

Immunoprecipitation of Bhlhb5 or Prdm8 in transfected 293T cells

Supernatant of 293T cell lysate was incubated with primary Bhl-hb5 (rat) antibody (or anti-Myc epitope (mouse)) antibody. Blocked Protein-A agarose beads were added to the cell lysate supernatant, pelleted from the cell lysate, and washed. Bhlhb5 protein (or Prdm8 protein) was then precipitated and bound to the beads through pull-down. Design and generation of Prdm8-directed antibody

Multiple Prdm8 GST-fusion protein constructs were devised that consisted of all 6 combinations of 3 highly conserved domains in the full length Prdm8 sequence (labeled as Domain 1, Domain 2, and Domain 3). Primer-targeted PCR amplification was conducted, trans-forming constructs into competent cells, which were maxiprepped and sequenced. Constructs were then digested with restriction en-zymes, ligated into pGEX vectors, and transformed the vectors into Top10 Competent Cells. The researcher then cultured in high volumes the correctly transformed GST-PRDM8 constructs, maxiprepped, and then transformed them into BL21 (E. coli) cells.

1 (see Supplementary Materials for full Materials and Methods)

Figure 3. Bhlhb5 mutants are missing the corticospinal tract. A) PKC-gamma immunostaining in BHLHB5 wt mice marks the CST (indi-cated by white arrow). B) PKC-gamma immunostaining in BHLHB5 mut mice exhibit no expression of PKC-gamma, thus demonstrating loss of the CST(idicated by white arrow). (Ross et al., unpublished data)

Figure 2. BHLHB5-expressing neurons are missng from a sub-population of cells in the superficial layers of the spinal cord dorsal horn. A) Control BHLHB5 cre/+ mice exhibit BHLHB5-express-ing neurons marked by a reporter gene (eYFP) as shown. B) Mutant BHL-HBS cre/- mice exhibit fewer BHLHB5-expressing neurons as marked by a reporter gene (eYFP) as shown. The most superficial layers of the spinal cord are marked by layers A and B (outlined in white). (Ross et al., unpub-lished data).

Figure 4. Propsed models of Prdm8 and Bhlhb5 regulation of target genes. A) Bhlhb5 and Prdm8 bind to distinct regions of DNA, yet repress the same target genes. B) Bhlhb5 and Prdm8 interact directly with each other in complex to co-repress the same target genes. C) Bhlhb5 and PRdm8 each repress transcription of distinct target genes (X and Y, respectively), which are key proteins involved in a common pathway. In this model, both Bhlhb5 and Prdm8 act as indirect regulators of the same downstream processes, yet Bhlhb5 and Prdm8 do not interact with each other or target directly the same genes.


49www.thurj.org

Neuroscience

Induction and purification of Prdm8 GST-fusion protein con-structs

Colonies of transformed BL21 cells were diluted, cultured in high volumes, and induced with IPTG. Cells were spun down, lysed, and incubated with Protein-G sepharose beads. Protein-coated beads were spun down, washed, and the protein eluted from the beads. A Bradford protein assay was conducted to quantify the eluted protein from the beads, and the purified protein was sent for injection into rabbits. To observe induction of each of the 6 PRDM8 GST-Fusion protein constructs in BL21 cells, cells were lysed in Western Lysis Buffer and protein sizes were examined via SDS-PAGE and coomass-ie staining (using a GelCode Blue Stain Reagent). Immunoprecipitation of endogenous Bhlhb5 and Prdm8 protein in neurons

The supernatants of the neural lysate of Prdm8 mutant and wild-type mice were incubated with primary Bhlhb5 (rat) (or Prdm8 (rab-bit)) antibody. Blocked Protein-A agarose beads were added to the cell lysate supernatant, and the beads were pelleted from the neural lysate and washed. Bhlhb5 protein (or Prdm8 protein) was precipi-tated bound to the beads through pull-down. Western blot analysis of immunoprecipitation results

Lysates from the immunoprecipitation experiments were analyzed via SDS-PAGE and western blot using Prdm8 and Bhlhb5-specific antibodies. Prdm8 rabbit antibody was used to probe the upper half of the blot (Prdm8 protein band is approximately 72 kDa in size in neurons). For immunoprecipitation experiments conducted on 293T cell lysates, an anti-Myc (mouse) antibody to probe for the C-termi-nal epitope tag attached to Prmd8 transfected in cultured 293T cells was used. Bhlhb5 rabbit antibody was used to probe the lower half of the blot (Bhlhb5 protein is approximately 35 kDa in size). Immunohistochemistry analysis

To analyze the cell-specific expression patterns of Prdm8 and Bhl-hb5 in cortical and spinal cord tissue sections, immunohistochem-istry analysis was performed on fixed tissue. Whole specimens were frozen in blocks of optimum cutting temperature (OCT) formula, and sagittal and coronal sections 20 microns thick were collected utilizing a cryostat, and the sections were transferred onto slides. Im-munostaining analysis was conducted to visualize proteins of inter-est.

Results

No co-immunoprecipitation of Bhlhb5 or Prdm8 in 293T cellsTo test the hypothesis of whether or not any evidence of protein-

protein interaction between Bhlhb5 and Prdm8 could be observed, immunoprecipitation experiments were conducted. In this experi-ment, the researcher performed antibody-specific pull-down of a particular protein (Bhlhb5) via adhesion to beads and precipitation from the whole cell lysate. The precipitate was probed with antibod-ies specific for its partner protein (Prdm8) to observe for any co-im-munoprecipitation. Through western blot analysis (Figure 5), no evi-dence was found of a direct interaction between Bhlhb5 and Prdm8. Antibody-specific immunoprecipitation of the Bhlhb5 protein was observed, and any associated Prdm8 pull down via probe identifica-tion of the Myc-epitope tagged to the C-terminal of the PRDM8 pro-tein was analyzed. As seen in Figure 9, there is no observable protein band at approximately 98 kDa in the lane containing the precipitate pulled down via Bhlhb5 interaction. Conversely, immunoprecipi-tation of Prdm8 via its Myc-epitope tag exhibited no pull-down of Bhlhb5 in the precipitate (which would have been indicated by a band ~37 kDa in size).Design and development of Prdm8 GST-fusion protein constructs

In order to conduct immunoprecipitation experiments on endog-enous Bhlhb5 and Prdm8 protein from neurons, use of a Prdm8 an-tibody was required. Due to the lack of available Prdm8 antibody, however, a series of Prdm8 GST-Fusion protein constructs were de-signed and developed. Single constructs to be purified were selected and injected into rabbits for Prdm8 antibody production.

Sup. Figure 3 exhibits the design of the 6 different protein domains cloned into vectors with GST to create 6 individual GST-fusion pro-tein constructs. Sup. Figure 4 displays the induction results for each of the Prdm8 GST-Fusion protein constructs in which the darkest band in each lane corresponds to the appropriate sized protein band for each of the Prdm8 fusion proteins. The middle region of the pro-tein was found to be the only construct isolated in the soluble fraction during the protein purification process. This region (termed Domain 2) was selected as the Prdm8 antigen to be injected into rabbits for Prdm8 antibody production. Designed antibody recognizes Prdm8 protein in neurons

To test for specificity and strength of the developed Prdm8 anti-body to recognize endogenous Prdm8 in neural lysate, an experimen-tal western blot on a wildtype whole mouse cortex and one mutant for Prdm8 was conducted (Sup. Figure 5). Prdm8 antibody derived from the serum of a rabbit injected with purified antigen (derived from the Prdm8 Domain 2 GST-Fusion protein construct) was used to probe for the presence of Prdm8 protein in the neural lysate. The Prdm8 antibody accurately identified the Prdm8 protein in the neurons as predicted. This finding was verified in the western blot through the observation of a ~72 kDa protein band, which corresponds to the ap-proximate size of endogenous Prdm8 in neurons. No co-immunoprecipitation of Bhlhb5 or Prdm8 in neurons

Unlike the previous experiment performed in which 293T cells transfected with Bhlhb5 and Myc-Prdm8, this immunoprecipitaton experiment was performed on proteins expressed from endogenous genes in neural lysate drawn from the whole cortex of Prdm8 wild-type and mutant mice. The advantage of this protocol is that any as-sociated co-factors necessary to facilitate proper Bhlhb5 and Prdm8

Figure 5. Western blot of immunoprecipitation conducted in 293T cells. 293T cells were transfected with both BhlHb5 and myc-tagged Prdm8. First lane displays IP for Bhlhb5, and no co-IP of Prdm8 is observed. Second lane displays IP for PRdm8 via its myc epitope tag, and no co-IP of Bhlhb5 is observed. Bottom half of blot(divded by dotted line) was blotted with antibody for Bhlhb5, upper half of blot probed with antibody for Myc-epitope tag of Prdm8.



Neu

rosc

ienc

e

protein interaction would also be available in the neural lysate to maintain natural protein-protein interactions (such would not be possible in 293T cells).

In this experiment, antibody-specific pull-down of a particular protein (i.e. Bhlhb5) was conducted via adhesion to beads and pre-cipitation from the whole cell lysate. The precipitate with the anti-body specific for the partner protein (i.e. Prdm8) was then probed to look for any co-immunoprecipitation. As indicated in Figure 6, there is no protein band at approximately 72 kDa (MW of Prdm8 protein) in the lane that indicates the precipitate pulled down via Bhlhb5 interaction. Since Prdm8 is evident in the wildtype whole cell lysate when compared to the Prdm8 mutant control, endogenous Prdm8 is clearly present prior to immunoprecipitation. No Prdm8 was co-immunoprecipitated with Bhlhb5 in neurons. Similarly, im-munoprecipiation was performed for Prdm8 and no associated pull-down of Bhlhb5 in the precipitate was found (Figure 7).Prdm8-expressing cells lost in the dorsal horn of spinal cord of Bhlhb5 mutants

In previous research, Bhlhb5 mutants exhibited distinct pheno-types that included skin lesion development, handstand behavior, decrease in motor coordination, and decrease in sensitivity to ther-mal or chemical stimulation (Ross et al., unpublished data). When examined at the cellular level, Bhlhb5 mutants also exhibited a sig-nificant loss of Bhlhb5-expressing neurons in the superficial layers of the dorsal horn of the spinal cord (Ross et al., unpublished data). Recent findings that Prdm8 mutants indicate striking similarities comparable to those of Bhlhb5 mutants (i.e. skin lesion development and handstand behavior) (McInnes, unpublished data) suggested the theory that Bhlhb5 and Prdm8 may function using similar regula-tory pathways of the central nervous system. As a result, it is possible that these two transcription factors are expressed in the same neu-rons during development. This reasoning led to the research hypoth-esis that: 1) Bhlhb5 mutants also exhibit significant loss of Prdm8-expressing neurons in the superficial layers of the dorsal horn of the spinal cord, 2) Prdm8 mutants exhibit the loss of Prdm8 and Bhlhb5

expressing neurons in the dorsal horn of the spinal cord.To test the first hypothesis, immunohistochemistry was per-

formed to compare Prdm8 expression in neurons in the dorsal horn of the spinal cord between wildtype controls and Bhlhb5 knockout mice at P0. A significant loss of Prdm8-expressing neurons in the dor-sal horn of the spinal cords of Bhlhb5 mutants was detected(Figure 8A). The mean number of Prdm8-expressing neurons located in the dorsal horn of the spinal cord in Bhlhb5 mutants and controls were quantified to determine if a statistically significant difference ex-isted. A two-tailed T-test with a confidence value of 95% (p-value = 0.0012) (Table 1) was utilized to perform the statistical analysis. A significant decrease in Prdm8-expressing neurons in Bhlhb5 mutant mice as compared to those of Bhlhb5 wildtype controls was observed. Moreover, the Prdm8-expressing neurons were lost to the outer-most laminae (superficial region) of the spinal cord dorsal horn. Co-expression of Bhlhb5 and Prdm8 in spinal cord dorsal horn

To test the second hypothesis, immunostaining was conducted for Bhlhb5 and Prdm8 in the spinal cord of Prdm8 wildtype controls and mutants at P0. Wildtype mice exhibited high levels of Bhlhb5 and Prdm8 co-expression in the dorsal horn region of the spinal cord (indicated by white dotted lines in Sup Figure 6A).

Co-expression of both transcription factors (indicated in the “merged” panels of Sup. Figures 6A and 6B, where neurons express-ing both Bhlhb5 and Prdm8 appear purple or blue-green in color, respectively) was detected. Interestingly, in the analysis of immunos-taining for Bhlhb5 and a GFP marker for Prdm8-expressing neurons in Prdm8 mutants, a qualitative loss was observed of Bhlhb5 and Prdm8 co-expressing neurons in the dorsal horn on the spinal cord (Sup. Figure 6B).Co-expression of Bhlhb5 and Prdm8 in the frontal cortex

Due to the similarity in phenotypes observed in both Bhlhb5 and Prdm8 mutants (both behaviorally and at the cellular level in the spi-nal cord), there is another key hypothesis. The third hypothesis is that high levels of co-expression of both Bhlhb5 and Prdm8 would also be observed in neurons located in the cortex where important

Figure 6. Western blot of BHLHB5 immunoprecipitation con-ducted in neural lysate. Neural lysate from whole cortex of both PRDm8 wt and Prdm8 mut P0 mice were utilized to IP for Bhlhb5 (both BHLhb5 wt). First and second lane display neural lysate (1% NP-40 lysis buffer) of Prdm8 mut and Prdm8 wt mice, respectively. Second lane clearly indicates the presence of PRDM8 protein in the neural lysate prior to the IP. Third and fourth lanes indicate IP for BHlhb5 in both Prdm8 mut and Prdm8 wt mice, respectively. No evidence of co-IP of Prdm8 is observed in the Prdm8 wt neurons.

Figure 7. Western blot of PRDM8 immunoprecipitation conduct-ed in neural lysate. Neural lysate from whole cortex of both Prdm8 wt and Prdm8 mut P0 mice were utilized to IP for PRDM8. First and second lanes display neural lysate (SDS lysis buffer) of Prdm8 wt and Prdm8 mut mice (both Bhlhb5 wt), respetively. Presence of BHLHB5 protein in these lysates is observed by the dark band at the 35kDa mark in each lane. Third and fourth lanes indicate IP for PRDM8 in both Prdm8 wt and Prdm8 mut mice, respectively. No evidence of co-IP of BHLHB5 is observed in the Prdm8 wt neurons.

51www.thurj.org

Neuroscience

neuronal components of sensory and motor circuits are derived and send projections throughout the body. Immunostaining was con-ducted for Prdm8 and Bhlhb5 in sagittal sections of wildtype mice frontal cortex, which indicated high levels of co- expression for both proteins in populations of neurons at P0 (Figure 9). Figure 9A ex-hibits immunostaining for Bhlhb5 and Prdm8 in a wildtype control brain and Figure 9B exhibits immunostaining for Bhlhb5 and Prdm8 in a Prdm8 mutant cortex. Enlarged boxes of each panel in Figure 9 demonstrate the co-expression of both proteins in the same neurons in wildtype and Prdm8 mutants. In the analysis of the immunostain-ing, no qualitative difference between Bhlhb5 and Prdm8 expression patterns in the cortex was observed. Nevertheless, high levels of co-expression was localized near the outermost layers of the frontal cor-tex of both wildtype and Prdm8 mutants. Prdm8 mutants and double heterozygotes are missing the CST

In previous research, the observation that both Bhlhb5 and Prdm8 mutants demonstrate similar behavioral phenotypes (i.e. develop-ment of skin lesions, handstand behavior) (Ross et al., unpublished data) led to the overarching theory that Bhlhb5 and Prdm8 may be two transcriptional regulators involved in common regulatory pro-cesses in neuronal development. This conceptual idea thus led to the reasoning that the phenotypes observed in Prdm8 mutants mirror-ing that of Bhlhb5 mutants would also extend to the cellular level throughout the central nervous system. Therefore, based on the find-ing that Bhlhb5 mutants exhibit distinct loss of the corticospinal tract (CST) (Figure 3), Prdm8 mutants might also exhibit loss of the CST.

To test this hypothesis, immunohistochemistry analysis of spinal cord sections of wildtype and Prdm8 mutant mice was performed at P0. In the immunostaining analysis, the presence of the CST was assessed through analyzing sectioned spinal cord tissue with a probe that recognizes a protein marker, Protein Kinase C-gamma (PKC-gamma), for the CST. PKC-gamma is highly expressed in the corti-

cospinal tract (Moreno-Flores et al., 2006), which is composed of a bundle of fibers derived from neurons originating in the motor cor-tex of the brain (Purves, 2001). In the analysis of the immunostaining experiment, the CST was found to be lacking in Prdm8 mutants (Sup. Figure 7C).

The observation regarding the absence of the CST in Prdm8 mu-tants is identical to evidence demonstrating that Bhlhb5 mutants are also missing the corticospinal tract, established through research indicating that corticospinal axons may mis-project or stop growth prematurely and do not reach targets in the dorsal spinal cord (Ross et al., unpublished data). Therefore, the combination of past research exhibiting the absence of the corticospinal tract in Bhlhb5 mutants with this project’s findings that Prdm8 mutants may also lack the corticospinal tract further support the overarching hypothesis that Prdm8 and Bhlhb5 are both crucial factors in the development of cir-cuit formation in the central nervous system. Intriguingly, it was also noted that double heterozygous mice for Prdm8 and Bhlhb5 (Prdm8-/+ Bhlhb5 -/+) also indicate loss of the CST through the absence of PKC-gamma protein expression in spinal cord sections at P0 (Sup. Figure 7B). However, this observation regarding the loss of the CST in heterozygous animals is not observed in animals heterozygous for only one gene.

Figure 8. BHLHB5 mutants exhibit significant loss of PRDM8-ex-pressing meurons in dorsal horn of spinal cord. A) Immunostain-ing analysis exhibiting loss of PRDM8 (marked in red) in subpopulations of neurons located in the dorsal horn of the spinal cord (outlined in white). B) Graphical representation of the significant loss of PRDM8-expressing neurons in the dorsal horn of the spinal cord when comparing BHLHB5 mutants vs. wildtype mice.

Figure 9. PRDM8 and BHLHB5 immunostaining in cortex. A) Immunostaining analysis of PRDM8 expression (red) and BHLHB5 ex-pression (blue) in the frontal cortex of wildtype mice (P0). High levels of PRDM8 and BHLHB5 co-expression expression exhibited in the outer layers of the cortex (shown as purple neurons in enlarged box). B) Im-munostaining analysis of PRDM8-expressing neurons via GFP knockin at Prdm8 locus (green) and BHLHB5 expression (blue) in the frontal cor-tex of PRDM8 mutant (-/-) mice (P0). Large population of neurons co-expressing PRDM8 and BHLHB5 also exhibited in the outer layers of the cortex (shown as blue-green neurons in enlarged box).

BHLHB5 WT Sample Size

BHLHB5 MUTSample Size

Mean # PRDM8-expressing cells in

PRDM8 WT(St. Dev)

Mean # PRDM8-expressing cells in

PRDB8 Mut (ST. Dev)

2-Tailed T Test P-Value*

N=20 N=2056 Neurons(19.31608)

28 Neurons(7.483314) 0.0011548

Table 1. PRDM8 mutants exhib-it signifiacnt loss of PRDM8-ex-pressing neurons in dorsal horn of spinal cord. 2-Tailed statistical test (*2 sample unequal variance het-eroskedastic) comparing te number of PRDM8-expressing neurons be-tween Bhlhb5 mut and wt mice in-dicate a statistically significant differ-ence (p-value<0.05, confidence value of 95%).



Neu

rosc

ienc

e

Discussion

No evidence of Bhlhb5 or Prdm8 binding in 293T cellsResults from this experiment (Figure 9) did not present any evi-

dence of co-IP. Conversely, when immunoprecipitation for Prdm8 was conducted, no presence of Bhlhb5 was observed in the precipi-tate. These results suggest that Bhlhb5 and Prdm8 do not interact, and therefore provide evidence in support of models in the Figures 4A and 4C. However, this experiment was conducted in transfected 293T cells, where many endogenous factors necessary for regulato-ry processes in neurons are absent. It is possible that other proteins present in neurons but absent in 293T cells are necessary co-factors to facilitate the complex formation of Bhlhb5 and Prdm8.

The immunoprecipitation experiment was replicated in neurons where pull-down of endogenous Prdm8 and Bhlhb5 in their natural substrate environments could be observed. To conduct this experi-ment in neurons, a direct Prdm8 antibody was required. The previous immunoprecipitation experiment conducted in 293T cells utilized a Myc-epitope tagged construct of Prdm8. Thus, immunoprecipitation and probing for Prdm8 in 293T cells could be performed via the use of an anti-Myc antibody (rather than a direct antibody for Prdm8). A chief obstacle in this study that arose was the lack of available Prdm8 antibody to use in immunoprecipitation and immunohistochemistry experiments on neurons and fixepd tissue. Design and development of the Prdm8 antibody

To resolve the issue of the absence of a Prdm8 antibody, a series of Prdm8 GST-Fusion protein constructs was designed and developed (Supp. Figures 3 and 4), from which a purified Prdm8 antigen could be selected and injected into rabbits for production of an antibody. No evidence of endogenous Bhlhb5 and Prdm8 binding in neu-rons

Immunoprecipitation for Bhlhb5 in Prdm8 wildtype and mutant neurons from the whole cortex of mice demonstrated clear pull-down of the Bhlhb5 protein, yet no evidence of any Prdm8 was found in the precipitate (Figure 6). Similarly, no co-IP of Bhlhb5 was ob-served when pull-down for Prdm8 was conducted (Figure 7). Again, these results suggested that Bhlhb5 and Prdm8 do not interact, and therefore provided evidence in support of models in the Figures 4A and 4C.

Nevertheless, it is also possible that conditions of the immunopre-cipitation experiment itself may have contributed to the dissociation of proteins in complex from one another. Considering the multiple steps involved in immunoprecipitation (i.e. washing of the beads, incubation of the antibody, and incubation with beads), it is very plausible that the nature of the protein-protein interaction between Bhlhb5 and Prdm8 may be very sensitive to the concentration of the lysis buffer ingredients, and that such may have affected the natural affinity between Bhlhb5 and Prdm8. Future experiments to clarify the interactive relationship between Bhlhb5 and Prdm8 should there-fore call for a series of tests examining the effects of different lysis buffer ingredients, concentrations, and wash conditions on the im-munoprecipitation and co-immunoprecipitation of proteins. Significant loss of Prdm8-expressing neurons in Bhlhb5 mutant spinal cord

To further explore the relationship between Bhlhb5 and Prdm8 in neural development, the expression patterns of each protein in both Bhlh5 and Prdm8 mutants were characterized. Immunohistochemis-try analysis was conducted on Bhlhb5 mutant mice, and a significant loss of Prdm8 was observed in the dorsal horn of the spinal cord at P0. The researcher confirmed this observation through immunostaining analysis (Figure 8A) and statistical quantification (Table 1, Figure 8B). These results, in conjunction with past research that Bhlhb5 is

significantly lost from populations of neurons in the superficial layers of the dorsal horn of the spinal cord (Ross et al., unpublished data), combine to support the primary hypothesis: that Prdm8 and Bhlhb5 may be important regulatory factors in the same pathways in central nervous system development. With evidence from past research that Bhlhb5-expressing neurons are lost in the dorsal horn of the spinal cord, this finding that Prdm8-expressing neurons are also lost in the same region of the spinal cord suggested that Prdm8 and Bhlhb5 may be highly expressed in the same neurons. High levels of co-expression of Prdm8 and Bhlhb5 in dorsal horn of spinal cord

Due to the finding that a significant loss of Prdm8-expressing neurons was observed in the same regions of the dorsal spinal cord where a significant loss of Bhlhb5-expressing neurons are absent in Bhlhb5 mutants, it was asked whether or not Bhlhb5 and Prdm8 co-expression would also be observed in the dorsal horn of the spinal cord. Sup. Figure 6A depicts immunostaining results from a wildtype mouse spinal cord. A high degree of co-localization of Prdm8 and Bhlhb5 protein expression in the dorsal horn of the spinal cord was observed. Furthermore, when the wildtype immunostaining results were compared to those from Prdm8 mutant spinal cord, a qualita-tive loss was determined of neurons co-expressing both Prdm8 and Bhlhb5 in the dorsal horn of the spinal cord (Sup. Figure 6B). The finding that Prdm8 and Bhlhb5 are both highly expressed in the same neurons thus provides further evidence in support of the hypothesis that Bhlhb5 and Prdm8 are involved in the same pathways in nervous system development.

Due to the lack of available Prdm8 mutant mice, there was not a sufficient sample size to conduct a quantitative statistical analysis to determine the significance of this loss of neurons co-expressing Prdm8 and Bhlhb5 in the dorsal horn of the spinal cord. Future ex-periments, however, should call for a quantification of the number of co-expressing neurons in wildtype versus Prdm8 and Bhlhb5 mutant mice, and statistical calculation to test for significance in difference between the two groups. High levels of co-expression of Prdm8 and Bhlhb5 in frontal cortex

Due to the high levels of observed Bhlhb5 and Prdm8 co-expres-sion in the spinal cord, in addition to the underlying theory that these two genes are crucial in the development of the central nervous system, Bhlhb5 and Prdm8 expression patterns in the mouse brain were analyzed. Immunostaining analysis was conducted for Bhlhb5 and Prdm8 on sagittal sections of the mouse cortex. Expression was observed for each of these proteins in cortical neurons, and for any distinguishing features between Bhlhb5 and Prdm8 expression pat-terns in wildtype versus Prdm8 mutant P0 mice. From the results (Figure 9), high levels of Prdm8 and Bhlhb5 co-expression in the cor-tex was found. The qualitative analysis further determined that the areas of greatest co-expression were the outermost layers of the fron-tal cortex. Qualitative examination of Bhlhb5 and Prdm8 expression in the frontal cortex between wildtype and Prdm8 mutant mice did not reveal any outstanding disparities in expression patterns. Nev-ertheless, further analysis following the availability of more Prdm8 mutant mice would be necessary to examine critically the differences in Prdm8 and Bhlhb5 expression patterns in the cortex of mutant mice.Prdm8 mutants and Prdm8-Bhlhb5 heterozygotes exhibit loss of CST

From the finding that Bhlhb5 mutants exhibit distinct loss of the corticospinal tract (CST) (Figure 3), it was hypothesized that Prdm8 mutants would also exhibit loss of the CST. Immunostaining analy-sis was conducted utilizing a CST marker (PKC-g) (Sup. Figure 7C),


53www.thurj.org

Neuroscience

which indicates the distinct loss of the CST in Prdm8 mutants. This result supportes the hypothesis that Prdm8 and Bhlhb5 mutants are both missing the CST and further provides evidence in support of the theory that Prdm8 and Bhlhb5 are crucial in the development of neu-rons involved in the same pathways of the central nervous system.

More intriguingly, however, the loss of the CST in double heterozy-gous mice (Bhlhb5 -/+ Prdm8 -/+) (Sup. Figure 7B) was noted. This finding is particularly striking, since no loss of the CST was observed in mice heterozygous for only one gene (i.e. Bhlhb5 -/+ Prdm8 +/+ or Bhlhb5 +/+ Prdm8-/+). Therefore, the finding that a heterozygous copy of both Prdm8 and Bhlhb5 could result in the absence of the CST, a phenotype found only in Bhlhb5 or Prdm8 full knockouts, suggestes that Bhlhb5 and Prdm8 may have an additive impact where the absence of a single copy of both Prdm8 and Bhlhb5 in mice may be equivalent in phenotype to missing both copies of either gene. The discovery of the absence of the CST in double heterozygous mice is exceptionally intriguing and prompts the hypothesis that such mice will also exhibit other identical phenotypes at the cellular (i.e. sig-nificant loss of Bhlhb5 or Prdm8-expressing neurons) and behavioral (i.e. skin lesion, handstand) levels. This hypothesis thus calls for fu-ture experiments to explore the phenotypes of double heterozygous mice through a series of behavioral assays (i.e. rotarod balance exper-iments, observation of self-injurious behavior, thermal and chemical stimulation assays) and examination of cellular expression patterns (i.e. analysis of Bhlhb5 and Prdm8-expressing neurons in the spinal cord and brain) and comparing such results to those of Prdm8 or Bhlhb5 single gene knockout mice.

Future Directions

Although results gathered in this study does not suggest direct interaction between Bhlhb5 and Prdm8, it is still not clear what the precise nature of their regulatory relationship may be. Three poten-tial models of interaction were proposed, of which one (Figure 4B) was not supported by the immunoprecipitation results. However, two other models (Figures 4A and 4C) are possible. Therefore, a future di-rection of this study would be to distinguish between the two mecha-nisms or identify other alternative models that may best describe the relationship between Bhlhb5 and Prdm8.

Conclusions

The major finding in this study was that no indication of direct interaction between Bhlhb5 and Prdm8 proteins was observed. The models in Figures 8A and 8C are potential mechanisms of interaction between Bhlhb5 and Prdm8 in transcriptional regulation. In these models, Bhlhb5 and Prdm8 do not bind to each other in complex, targeting either the same genes (Figure 4A) or distinct genes in the same pathway (Figure 4C). Nevertheless, due to the nature of the experimental conditions and the possible sensitivity of the proteins’ affinities for each other, it is necessary that further experiments be conducted before the model in which Prdm8 and Bhlhb5 interact in complex (Figure 4B) may be rejected.

Despite the lack of evidence identifying the precise model of inter-action between Bhlhb5 and Prdm8, immunostaining results provide strong evidence in support of the overarching theory that Bhlhb5 and Prdm8 function as important regulatory factors in the same pathway of neuronal development in the central nervous system. This con-clusion was drawn from four critical findings: 1) Prdm8-expressing neurons are significantly lost in the dorsal horn of the spinal cord in Bhlhb5 mutants, 2) high levels of co-expression of Prdm8 and Bhlhb5 are observed in the dorsal horn of the spinal cord, 3) dramatic loss of

neurons co-expressing Prdm8 and Bhlhb5 is observed in the dorsal horn of the spinal cord, and 4) loss of the CST is observed in Prdm8 mutants and double heterozygous mice (Bhlhb5 het, Prdm8 het). The combination of these four findings demonstrate that Prdm8 mutants, like Bhlhb5 mutants, express similar Prdm8 and Bhlhb5 expression patterns in the brain and spinal cord. This study additionally sug-gests that the missing subpopulation of neurons in the dorsal horn of the spinal cord co-express both Bhlhb5 and Prdm8, thus providing evidence in support of the hypothesis that Bhlhb5 and Prdm8 are crucial factors in a common pathway underlying neural differentia-tion and development.

Technical difficulties encountered included the lack of available mice of desired Prdm8 and Bhlhb5 genotypes. With a greater number of Prdm8 mutants, double heterozygotes, and double mutants, more immunostaining experiments could be conducted to discern for any nuances in phenotype that distinguish Bhlhb5 mutants from Prdm8 mutants. Furthermore, the availability of more mice of desired mu-tant genotypes would allow for statistical analysis to be performed so that the significance of Bhlhb5 and Prdm8 co-expression may be calculated.

In sum, the overarching motivation that propelled this research was the pursuit of a better understanding of the processes underly-ing central nervous system development in mammals. The striking phenotypes exhibited by both Bhlhb5 and Prdm8 mutant mice was the first signal that a cooperative relationship may exist between the two proteins, and that their function was critical in neural develop-ment. These results provide support for the view that a novel relation-ship between Bhlhb5 and Prdm8 exist in the same neural pathways; they also help illustrate a process within the formation of the central nervous system that have significance for the future understanding of not only how the mammalian nervous system develops, but how it matures and ages with time.

ReferencesAlberts, B.J., Alexander; Lewis, Julian; Raff, Martin; Roberts, Keith; Walter,

Peter, ed. (2002). Molecular Biology of the Cell (New York).Ancelin, K., Lange, U.C., Hajkova, P., Schneider, R., Bannister, A.J., Kouza-

rides, T., and Surani, M.A. (2006). Blimp1 associates with Prmt5 and di-rects histone arginine methylation in mouse germ cells. Nature Cell Biol-ogy 8, 623-630.

Andersson, E., Jensen, J.B., Parmar, M., Guillemot, F., and Bjorklund, A. (2006). Development of the mesencephalic dopaminergic neuron system is compromised in the absence of neurogenin 2. Development 133, 507-516.

Bertrand, N., Castro, D.S., and Guillemot, F.o. (2002). Proneural genes and the specification of neural cell types. Nature Reviews Neuroscience 3, 517-530.

Bramblett, D.E., Pennesi, M.E., Wu, S.M., and Tsai, M.-J. (2004). The Tran-scription Factor Bhlhb4 Is Required for Rod Bipolar Cell Maturation. Neuron 43, 779-793.

Davis, C.A., Haberland, M., Arnold, M.A., Sutherland, L.B., McDonald, O.G., Richardson, J.A., Childs, G., Harris, S., Owens, G.K., and Olson, E.N. (2006). PRISM/PRDM6, a transcriptional repressor that promotes the proliferative gene program in smooth muscle cells. Mol Cell Biol 26, 2626 - 2636.

Derunes, C., Briknarová, K., Geng, L., Li, S., Gessner, C.R., Hewitt, K., Wu, S., Huang, S., Woods Jr, V.I., and Ely, K.R. (2005). Characterization of the PR domain of RIZ1 histone methyltransferase. Biochemical and Biophysi-cal Research Communications 333, 925-934.

Dimou, L., Simon, C., Kirchhoff, F., Takebayashi, H., and Gotz, M. (2008). Progeny of Olig2-Expressing Progenitors in the Gray and White Matter of the Adult Mouse Cerebral Cortex. J. Neurosci. 28, 10434-10442.

Djaldetti, R., Shifrin, A., Rogowski, Z., Sprecher, E., Melamed, E., and Yar-nitsky, D. (2004). Quantitative measurement of pain sensation in patients with Parkinson disease. Neurology 62, 2171-2175.

Elworthy, S., Hargrave, M., Knight, R., Mebus, K., and Ingham, P.W. (2008). Expression of multiple slow myosin heavy chain genes reveals a diversity



Neu

rosc

ienc

e

of zebrafish slow twitch muscle fibres with differing requirements for Hedgehog and Prdm1 activity. Development 135, 2115-2126.

Feng, L., Xie, X., Joshi, P.S., Yang, Z., Shibasaki, K., Chow, R.L., and Gan, L. (2006). Requirement for Bhlhb5 in the specification of amacrine and cone bipolar subtypes in mouse retina. Development 133, 4815-4825.

Fumasoni, I., Meani, N., Rambaldi, D., Scafetta, G., Alcalay, M., and Cicca-relli, F. (2007). Family expansion and gene rearrangements contributed to the functional specialization of PRDM genes in vertebrates. BMC Evolu-tionary Biology 7, 187.

Hamik, A., and Jain, M.K. (2008). Variety is the splice of life. Journal of Mo-lecular and Cellular Cardiology 44, 44-46.

Hand, R., Bortone, D., Mattar, P., Nguyen, L., Heng, J.I.-T., Guerrier, S., Boutt, E., Peters, E., Barnes, A.P., Parras, C., et al. (2005). Phosphoryla-tion of Neurogenin2 Specifies the Migration Properties and the Dendritic Morphology of Pyramidal Neurons in the Neocortex. Neuron 48, 45-62.

Hughes, S.M. (2004). Muscle Differentiation: A Gene for Slow Muscle? Cur-rent Biology 14, R156-R157.

Jiang, X., Tian, F., Du, Y., Copeland, N.G., Jenkins, N.A., Tessarollo, L., Wu, X., Pan, H., Hu, X.-Z., Xu, K., et al. (2008). BHLHB2 Controls Bdnf Pro-moter 4 Activity and Neuronal Excitability. J. Neurosci. 28, 1118-1130.

Kajimura, S., Seale, P., and Tomaru, T. (2008). Regulation of the brown and white fat gene programs through a PRDM16/CtBP transcriptional com-plex. Genes and Development 22, 1397-1409.

Kouzarides, T. (2002). Histone methylation in transcriptional control. Cur-rent Opinion in Genetics & Development 12, 198-209.

Lu, Q.R., Sun, T., Zhu, Z., Ma, N., Garcia, M., Stiles, C.D., and Rowitch, D.H. (2002). Common Developmental Requirement for Olig Function Indicates a Motor Neuron/Oligodendrocyte Connection. Cell 109, 75-86.

Ma, Q., Kintner, C., and Anderson, D.J. (1996). Identification of neurogenin, a Vertebrate Neuronal Determination Gene. Cell 87, 43-52.

Ma, Y.-C., Song, M.-R., Park, J.P., Henry Ho, H.-Y., Hu, L., Kurtev, M.V., Zieg, J., Ma, Q., Pfaff, S.L., and Greenberg, M.E. (2008). Regulation of Motor Neuron Specification by Phosphorylation of Neurogenin 2. Neuron 58, 65-77.

Marsden, C.D. (1994). Parkinson’s disease. J Neurol Neurosurg Psychiatry 57, 672-681.

Massari, M.E., and Murre, C. (2000). Helix-Loop-Helix Proteins: Regulators of Transcription in Eucaryotic Organisms. Mol. Cell. Biol. 20, 429-440.

Moreno-Flores, M.T., Bradbury, E.J., Martin-Bermejo, M.J., Agudo, M., Lim, F., Pastrana, E., Avila, J., Diaz-Nido, J., McMahon, S.B., and Wandosell, F. (2006). A Clonal Cell Line from Immortalized Olfactory Ensheathing Glia Promotes Functional Recovery in the Injured Spinal Cord. Mol Ther 13, 598-608.

Nieto, M., Schuurmans, C., Britz, O., and Guillemot, F. (2001). Neural bHLH Genes Control the Neuronal versus Glial Fate Decision in Cortical Pro-genitors. Neuron 29, 401-413.

Ohinata, Y., Payer, B., O’Carroll, D.n., Ancelin, K., Ono, Y., Sano, M., Barton, S.C., Obukhanych, T., Nussenzweig, M., Tarakhovsky, A., et al. (2005). Blimp1 is a critical determinant of the germ cell lineage in mice. Nature 436, 207-213.

Purves, D., Augustine, J., Fitzpatrick, D., Katz, L. C., LaMantia, A. S., Mc-Namara, J. O., and Williams, S. M. (2001). Neuroscience, Vol 2nd Edition (Sunderland, MA: Sinauer Associates).

Ren, B., Chee, K., and Kim, T.H. (1999). PRDI-BF1/Blimp-1 repression is mediated by corepressors of the Groucho family of proteins. Genes and Development 13, 125-137.

Ross, S.E., Greenberg, M.E., and Stiles, C.D. (2003). Basic Helix-Loop-Helix Factors in Cortical Development. Neuron 39, 13-25.

Vincent, S.D., Dunn, N.R., Sciammas, R., Shapiro-Shalef, M., Davis, M.M., Calame, K., Bikoff, E.K., and Robertson, E.J. (2005). The zinc finger tran-scriptional repressor Blimp1/Prdm1 is dispensable for early axis forma-tion but is required for specification of primordial germ cells in the mouse. Development 132, 1315-1325.

Völkel, P., and Angrand, P.-O. (2007). The control of histone lysine methyla-tion in epigenetic regulation. Biochimie 89, 1-20.


55www.thurj.org

Chemical

Biology

Still images of a motion picture: Using static crystal structures to understand the

behavior of a DNA glycosylaseKimberly (Wei-Wei) Oo


The cell has developed a number of defenses against DNA damage; glycosylases, for example, remove damaged bases. Human 8-oxoguanine glycosylase 1 (hOGG1) is responsible for the excision of the damaged base oxoguanine (oxoG). In this study, we present the first reported structure of hOGG1 containing guanine in a fully wild type active site. In addition, we also present the structure of a novel catalytic intermediate during the excision of oxoG.

IntroductionMaintaining the integrity of the genome is crucial for the sur-

vival of an organism. DNA is vulnerable to damage from a variety of sources, both extrinsic and intrinsic to the cell (Lindahl 1993; Friedberg 2003). Unless the cell can repair this damage, it will cause mutations in the genome. Thus, DNA repair mechanisms in the cell are extraordinarily important (Lindahl 1993). One common kind of DNA damage is the oxidation of guanine to 8-oxoguanine (oxoG) by reactive oxygen species (Neeley 2006; Henle 1997), which are com-monly produced by ionizing radiation, exposure to transition metals, or even as byproducts of aerobic respiration (Henle 1997; Bjelland 2003). Although the Watson-Crick face of oxoG is identical to that of guanine and can still base pair with cytosine, it can also rotate around its glycosidic bond and base pair with adenine through a Hoogsteen interaction (Oda 1991) (Figure 1). During replication, DNA poly-merase preferentially adds adenine across from oxoG, which after two rounds of replication, causes a G:C to T:A transversion, thereby introducing a mutation into the genome (Hsu 2004; Shibutani 2001).

Human 8-oxoguanine glycosylase 1 (hOGG1) locates and excises oxoG from the genome as part of the base excision repair pathway

(Crenshaw 2009; David 1998). hOGG1 is a bifunctional enzyme, act-ing as a glycosylase, which breaks the N-glycosidic bond to the base, and a β-lyase, which nicks the DNA on the 3' side of the sugar (Boi-teux 2001). The enzyme diffuses along the DNA until it locates oxoG. It induces a bend in the DNA and extrudes oxoG out of the DNA helix and into its active site (Bruner 2000). The structural intermedi-ates involved in flipping out the base from the DNA helix make up the base extrusion pathway. Then, hOGG1 excises oxoG, creating an abasic site, and nicks the DNA (Dodson 1994). Other members of the base excision repair pathway then remove the sugar and insert a new guanine nucleotide, repairing the DNA (Vidal 2001).

One goal of the Verdine lab is to answer the “search problem,” namely, how hOGG1 distinguishes oxoG from other bases. Although hOGG1 can distinguish between damaged and normal bases while they are within its active site through specific contacts, it is unclear whether it is necessary to individually flip out each base into the ac-tive site to find rare oxoG substrates hidden in the genome (Bruner 2000). Such a process seems unlikely since hOGG1 slides along DNA at a velocity approaching the upper limit for one-dimensional diffu-sion (Blainey 2006). However, when oxoG is intrahelical, there are no obvious structural distortions to the DNA duplex that could be used to distinguish between oxoG and guanine (Bowman 2008; Lipscomb 1995; Oda 1991).

To determine if hOGG1 can distinguish between guanine and oxoG while the bases are intrahelical, it would be ideal to capture “encounter complexes” of hOGG1 on an intrahelical oxoG and gua-nine. With these structures, one could compare the interactions be-tween the protein and DNA to determine if the interactions differed between oxoG and guanine. Thus, our initial goal was to solve the structure of hOGG1 with an intrahelical guanine.

Another goal of our lab is to elucidate the catalytic mechanism hOGG1 uses to cleave out oxoG. In general, a bifunctional glycosy-lase uses a nucleophilic residue to displace the undesired base from the sugar, opening the ring and forming a Schiff base. β-elimination of the 3' phosphate through deprotonation of the C2' hydrogen cleaves the DNA strand and forms an α, β unsaturated Schiff base (David 1998), which is then hydrolyzed to release the glycosylase from the DNA, completing the mechanism. The structures of several interme-diates have been solved to bring insight into the specific mechanism hOGG1 uses (Fromme 2003; Radom 2006; Chung 2004).

Here, we present two crystal structures of hOGG1-DNA com-Figure 1. The cause of the mutagenic potential of oxoG. Repro-duced from Crenshaw 2009.



Chem

ical

Bi

olog

y

plexes. The first is a structure of hOGG1 containing guanine within its active site, the “G complex.” This structure was solved while we attempted to solve a structure of hOGG1 with an intrahelical gua-nine. The G complex was completely unexpected because previous studies indicated that guanine would be excluded from the active site (Banerjee 2006). In spite of being in the active site of a catalytically active hOGG1, the guanine was not excised out of the DNA. Thus, this study presents the first structure of hOGG1 with an undamaged base in the wild-type active site. The second structure presented is a structure of hOGG1 bound to a trans-α,β-unsaturated aldehyde, a previously unreported intermediate along the catalytic pathway.

Methods

Because hOGG1 has no specific preference for an undamaged base, to capture a structure with guanine it is necessary to restrict hOGG1 to that guanine so that the protein-DNA complex would be homo-geneous enough for crystallographic studies. We formed a disulfide bond between a mutated cystine on hOGG1 and a modified thiol-containing base in the DNA, creating a disulfide crosslink that cova-lently tethers the two together, restricting hOGG1 to the desired base (Banerjee 2005) (Figure 2). This disulfide crosslink was necessary to determine the structure of many of the intermediates captured thus far (Banerjee 2005; Banerjee 2005; Radom 2007). Our lab has also employed this strategy to determine the structure of other systems suffering from similar problems (Huang 2000; Huang 1998).

A number of intermediates along the base extrusion and catalytic pathways have already been captured. In the base extrusion pathway, crystal structures have been solved containing: guanine extruded out of the helical stack, but folded back into the major groove of the DNA (intermediate 1) (Banerjee 2006); oxoG (int. 2) (Radom 2007) and guanine (int. 3) (Banerjee 2005) in an “exo site,” with the base extrahelical, but on a part of hOGG1 adjacent to the active site; oxoG entering the active site (int. 4) (Radom 2007); and oxoG fully in the active site (int. 5-8) (Bruner 2000; Banerjee 2005; Radom 2007; Nor-man 2003). An overview of the previously captured intermediates in the base extrusion pathway is summarized in Figure 3.

In the earliest intermediate so far crystallized in the pathway, (int. 1), guanine was extruded out of the helical stack into the major groove (Banerjee 2006). The guanine was placed adjacent to oxoG, but a disulfide crosslink formed between N149C and the cytosine opposite from the guanine was short enough to restrict hOGG1 to

interrogate the guanine. The crosslink forced the guanine out of the DNA helix because the crosslink physically occupied the same place the intrahelical guanine would have occupied. The extruded gua-nine formed a noncannonical base pair with the neighboring oxoG, stabilizing guanine in the major groove. Likewise, in the structure containing guanine in the exo site, (int. 3), the same disulfide cross-linking site was used, which biased the guanine to be extrahelical (Banerjee 2005).

These two results seemed to imply that when forced out of the he-lix, guanine should have no strong preference for any particular site extrahelical to the DNA, as both the exo site and the noncannonical base pair do not seem offer a particularly large amount of stabiliza-tion (Banerjee 2005; Banerjee 2006). Thus, it is likely that it is more favorable for guanine to be within the DNA helix, where there are specific interactions stabilizing it. As a result, moving the crosslink to a different site that would not sterically displace guanine from the DNA helix might allow the capture of a structure containing an in-trahelical guanine (Radom 2006).

In this study, the crosslinking site was between S292C and an adenine four bases down from the guanine. This site had previously been employed to solve the late-stage intermediate structure (int. 4) with oxoG partially in the active site of hOGG1 (Radom 2007). In addition, the structure of a catalytically inactive mutant of hOGG1 crosslinked to oxoG containing DNA at the S292C site (int. 7) is almost completely identical to the structure of the mutant without the crosslinking site (int. 5) and the structure of the mutant with the N149C crosslinking site (Bruner 2000; Radom 2007) . Thus, the S292C crosslinking site seemed like it might not force the guanine to be extrahelical. As a result, we hoped that crystallization of hOGG1 crosslinked through S292C to guanine-containing DNA would allow the structural determination of a complex containing an intrahelical guanine. In reality, the structure captured was that of hOGG1 with guanine in its active site. An overview of the disulfide crosslinking strategy is presented in Figure 4.

Several intermediates along the catalytic pathway have also been captured. An overview of the theorized catalytic mechanism sum-marizing the previously captured intermediates is presented in Fig-ure 5. Two intermediates in the catalytic mechanism for hOGG1 have been captured using the technique of borohydride trapping (Fromme 2003; Radom 2006). Intermediates containing iminium ions can be reduced by sodium borohydride to the amine, halting catalysis at a particular step and allowing for the characterization of a normally

Figure 2. Functionalization of adenine and crosslinking with hOGG1 at the S292C site

Figure 3. An overview of the previously captured intermediates in the base extrusion pathway. hOGG1 is shown in purple, DNA in blue, and the interrogated base in red. The intermediates captured thus are categorized by: the location of the interrogated base; the crosslinking site, if there is one; and other mutations/information. PDB accession codes: 1. 2I5W29, 2. 2NOF31, 3. 1YQK30, 4. 2NOZ31, 5. 1EBM15, 6. 1YQR30, 7. 2NOL31, 8. 1N3C32.

C4Cysteamine

4 2

hOgg 1 S292C

4

Crosslinking A

Cysteine 292


57www.thurj.org

fleeting intermediate (Zharkov 2002; Gilboa 2002). In hOGG1, the ε-amino group of Lys 249 is a nucleophile and displaces the oxoG base to create the initial Schiff base (II). This intermediate has been captured by treatment with sodium borohydride and the structure determined by X-ray crystallography (III) (Fromme 2003). In ad-dition, the intermediate after β-elimination – the α, β unsaturated Schiff base (IV) – was captured by treatment with sodium cyano-

borohydride (V) (Radom 2006). Although borohydride trapping has been useful in determining

the structures of intermediates, one limitation of borohydride and other chemical trapping methods is that they can only capture inter-mediates with certain functional groups. A more general technique is needed to discover all the intermediates in the catalytic cycle. To address this, Drs. Chris Radom and Seongmin Lee in the Verdine lab developed a “time resolved” crystallography technique (Radom 2006). Instead of chemically trapping an intermediate, the interme-diate would be freeze-trapped by cryoprotection of crystals in liquid nitrogen. In brief, DNA containing a bulky photocleavable adduct attached to oxoG was crosslinked to hOGG1 by a disulfide tether be-tween N149C and the cytosine opposite from oxoG and crystallized. The structure of a complex containing this bulky adduct has been solved, indicating that the bulky adduct prevents oxoG from entering the active site and instead places it at the exo site. Flashing the crystal with ultraviolet light cleaves the adduct off and allows oxoG to enter the active site of hOGG1. Therefore, all the hOGG1 synchronously begins excising the oxoG. After a certain length of time, the crystal is frozen with liquid nitrogen, and the low temperature (77K) prevents further catalysis from occurring. As crystals are usually cryoprotect-ed to protect against radiation damage and increase the diffraction quality and resolution, this method of freeze-trapping the crystals is very convenient (Rodgers 1994; Henderson 1990). Ideally, this allows one to capture the various intermediates along the catalytic pathway by varying the time between photocleaving the adduct and freezing the crystal (Radom 2006).

This technique has been used to capture a late stage intermediate in the base extrusion pathway (Lee 2008). A crystal was cryoprotect-ed immediately after photocleavage of the adduct, and the structure of that complex was solved. The oxoG was almost fully inserted into

Figure 4. An overview of the crosslinking strategy. A) Crystal structures of: (int. 1). K249Q hOGG1 (green) bound DNA (blue) with oxoG (red); (int. 9). S292C hOGG1 (purple) crosslinked to DNA with guanine (green); (int. 3). N149C hOGG1 (yellow) crosslinked to DNA with guanine. B) and C) A comparison of the S292C and N149C crosslink sites.

Figure 5. An overview of the catalytic pathway of hOGG1. Red indicates that the structures have been crystallized. The geometry of double bond in IV and V can be either cis or trans. PDB accession codes: I. Same as (int. 5)-(int. 8), III. IHUO, V. Not submitted, and VII. 1M3H.

SN1-like

I. Before Catalysis

SN2-like

II. Initial Schiff Base

NaBH4

E1cb-likeE2-like

III. Reduced Schiff Base

Hydrolysis

IV. α, β Unsaturated Schiff BaseVI. α, β Unsaturated Aldehyde(This work)

V. Reduced α, β Unsaturated Schiff BaseVII. Ring Closed Sugar

Ring Closing NaBH3CN

Chemical

Biology

Chem

ical

Bi

olog

y

the active site of hOGG1, but the active site had not yet made any of the contacts with oxoG that have been observed in other structures containing oxoG in the active site. Thus, this structure validates this technique as a way to capture otherwise fleeting intermediates.

This study used the time resolved crystallography technique to solve the crystal structure of a catalytic intermediate obtained by freezing the crystal 30 minutes after photocleavage. It is hereafter re-ferred to as the “30 minute structure.”


Crystallization of the G complexThe crystallization of the G complex was surprisingly difficult.

The previous structures of base extrusion intermediates had all been crystallized in similar conditions, approximately 150 mM calcium chloride, 17% polyethylene glycol 8000, and 100 mM sodium cacody-late pH 6.0 within the 24 well hanging drop vapor diffusion method (Bruner 2000; Banerjee 2005; Banerjee 2006). However, when these conditions were used, the crystals were either too thin to be useful in X-ray crystallography or severely branched. In addition, they were fragile and broke apart while being transferred to cryoprotectant.

The screens were broadened to look at the effects of different salts (magnesium acetate and magnesium chloride), temperatures (4 and 20 °C), drop volumes, and protein complex to well solution ratios, without much change in crystal behavior. In addition, other crystal-lization techniques were tried, such as sitting drop vapor diffusion, the use of oils to slow vapor diffusion, macroseeding, streakseeding,

and addition of additives, without much improvement. To find a different crystal form that might be more amenable to

crystallographic studies, ten 96-well Nextal screening suites (Qiagen) were used to screen the hOGG1 complex over a variety of different conditions. However, the crystals that resulted from this large-scale screen were almost all either needles or dendrites and were also dif-ficult to reproduce in the 24-well format. As a result, this effort did not yield any useful results.

Eventually, a crystal yielding diffraction data of sufficient qual-ity was produced. The complex was crystallized with the addition of additives in a hanging drop vapor diffusion format with the 24 well OptiSalts Screen (Hamptom). The resolution of the structure from this crystal was 3.05 Å. Unfortunately, reproduction of this condition did not produce higher resolution data, so the 3.05 Å structure was used. Structure of the G complex

Charisse Crenshaw collected X-ray diffraction data for the G com-plex and scaled it to 3.05 Å (Figure 6). She solved the structure using the phases calculated from a previously solved hOGG1 structure, the K249Q, Q315A mutant containing oxoG in the active site (PDB ac-cession code: 2NOH) (Radom 2007). Surprisingly, this structure was almost identical to structures containing oxoG in the active site, ex-cept that guanine was in the place of oxoG (heavy atom RMSD=0.401 Å with (int. 5), the structure of catalytically inactive hOGG1 (K249Q) bound to oxoG containing DNA) (Bruner 2000).Conformational changes

When hOGG1 binds oxoG in its active site, it undergoes confor-mational changes that allow its residues to make contact with oxoG (Crenshaw 2009; Radom 2007). In a structure without oxoG in the active site, such as in the structures along the base extrusion pathway, the α-O helix is held away from the active site in an “open” conforma-tion (Banerjee 2005; Radom 2007). In this conformation, His 270 on the α-O helix forms a salt bridge with Asp 322 and an aryl-π inter-

Figure 6. Data collection and refinement statistics of the G com-plex. Reproduced from Crenshaw 2009.

Figure 7. A comparison of the open and closed active sites. The G complex is purple. A) open active site (int. 3) is orange. B) closed active site (int. 5) is green.

hOGG1 dDXL G-complex (Space group: P6522, Cell parameters: a = b = 90.91 Å, c = 211.63 Å, α = β = 90.0°, and γ = 120°)Data collectiona

Source APS 24ID

Wavelength (Å) 0.97921

Resolution (Å) 50-3.05 (3.16-3.05)

Rsym (%)a 13.4 (69.7)

Total no. of obs. 130,934

No. of unique obs. 10,527 (1,009)

Completeness (%) 99.7 (100.0)

 / σ 45 (2.1)

Refinement

Resolution (Å) 50-3.05

No. reflections 10,500

No. of non-hydrogen atoms 3,067

Rwork (%)b 22.09

Rfree (%)c 28.72

Mean B value (Å2)d 22.8038

Rmsd bond (Å)d 0.0063

Rmsd angle (°)d 1.244

Ramachandran plot analysise

(% most favored, additional allowed, generously allowed, disallowed)

82.4, 17.3, 0.4, 0.0

aRsym = Σ|I–|/ ΣI where I is the integrated intensity of a given reflection.bRwork = Σ|F(obs)–F(calc)|/ΣF(obs).cRfree = Σ|F(obs)–F(calc)|/ΣF(obs), calculated using 7.5% of the datadFrom CNS.eFrom PROCHECK.


59www.thurj.org

Chemical

Biology

action with Phe 319 (Figure 7a). However, in a structure containing oxoG in the active site, such as the structures captured with catalyti-cally inactive mutants or the structures of the catalytic intermediates, the α-O helix moves towards the active site into a “closed” conforma-tion, and His 270 instead forms a salt bridge with a phosphate of the DNA backbone, freeing Phe 319 to form a π-stacking interaction with oxoG. Furthermore, Gln 315 hydrogen bonds with the Watson-Crick face of oxoG (Figure 7b) (Bruner 2000; Radom 2007).

In the structure of the active site G complex, the aforementioned residues occupy the same positions as the residues in the “closed” conformation, with the exception of His 270, which is rotated 90°. However, this rotation does not affect its ability to form a salt bridge with the DNA phosphate, and it functions in the same manner as the histidine in the “closed” conformation.Discrimination between oxoG and guanine

The previous interactions are characteristic of the “closed” con-formation and stabilize both oxoG and guanine. However, based on quantum mechanical calculations using (int. 3) and (int. 5), there are two sources of discrimination between oxoG and guanine in the ac-tive site of hOGG1 that might destabilize guanine binding in the ac-tive site (Banerjee 2005). First, there is a hydrogen bond between the Gly 42 carbonyl and the oxoG N7 that is lost with guanine. This is calculated to stabilize oxoG by about 3.5 kcal/mol. Second, the salt bridge between Lys 249 and Cys 253 causes a dipole that interacts favorably with the dipole of oxoG. However, since guanine has the opposite dipole, this would be a repulsive interaction, favoring oxoG by about 3.3 kcal/mol.

In the active site G complex, the positions of Gly 42, Lys 249 and Cys 253 stay roughly the same, along with the rest of the residues in the active site, when compared to (int. 5) (Figure 8). The loop contain-ing Gly 42 was previously found to be rigid, which may explain why it does not move to relieve the posited repulsion. However, the distance between the Cα carbonyl of Gly 42 and N7 increases slightly (2.7 Å in (int. 5) to 3.2 Å in the active site G complex) by the base shifting down, which may relieve some repulsion. In addition, it is possible that the N7 of guanine is protonated, which would recreate a favor-able interaction. This is unlikely, however, because N7 is not basic

Figure 8. The effect of the S292C crosslink. A) and B) (int. 1) pink, (int. 3) yellow, (int. 8) green, (int. 5) blue, and (int. 6) purple. The length of the crosslink correlates with the position of the base along the base extru-sion pathway. C) and D) The lowest energy conformations of the crosslink used at the S292C site. Figure reproduced from Crenshaw 2009.

and is unlikely to be perturbed enough to be protonated. The other structures containing oxoG in the active site have distances similar to that of (int. 5): 2.9 Å (int. 6) (Banerjee 2005), 2.8 Å (int. 7) (Radom 2007) and 2.7 Å (int. 8) (Norman 2003), indicating that the increased distance to 3.2 Å may be significant. The last structure (int. 7) indi-cates that the increase is not directly due to the crosslinker, though it does not rule out a more subtle effect.

The salt bridge also appears to be intact in the G complex. The distance between Lys 249 and Cys 253 increases slightly from 2.6 Å in (int. 5) (Bruner 2000) to 2.8 Å in the G complex, which is still consis-tent with a salt bridge (Petsko 2003). Thus, it is unlikely that these re-pulsive interactions were completely removed, though the repulsion may be lessened slightly. Therefore, it is unclear as to why guanine would be inserted in the active site, as is seen in this structure. Crosslinking bias

The authors presenting (int. 2) and (int. 4), which contain oxoG in the exo site or partially in the active site, respectively, posed a similar question (Radom 2007). When the crosslinking site was moved from N149C to S292C, the oxoG moved further along the extrusion path-way. It was proposed that the crystal packing of the particular crystal form of (int. 4) bends the DNA to favor an extrahelical base (Radom 2007; Radom 2006). However, this leaves unresolved why in (int. 4), oxoG moved further along the base extrusion pathway instead of re-maining in the exo site, as presumably the crystal forms and packing interactions were the same (since the crystallization conditions were similar and the space groups identical).

Here, we propose that DNA bending is indeed the key difference but that the DNA is bent because of the crosslinker, not because of crystal packing. The distance between S292 and the adenine, which is the distance a crosslinker would need span if it were placed at the S292C site varies depending on the stage of the base extrusion path-way (Figure 9). The structures of the early intermediates (1 and 3) require a larger span (10.1 Å and 9.6 Å), while the structures of later intermediates (5, 6, and 8) require a smaller span (8.1 Å and 8.3 Å). However, in the nine lowest energy conformations of the four-carbon crosslinker used at the S292C site, the largest span is 8.46 Å, which is too short to span the distances of the early intermediates (Figure 8).

Chem

ical

Bi

olog

y

Thus, it is likely that the early intermediates, such as the exo site or intrahelical intermediates, are disfavored by the S292C four-carbon crosslinker, forcing the base into the active site.

Furthermore, in the previously solved structures containing the S292C crosslinker, (int. 4) and (int. 7), the distances spanned are 8.32 Å and 6.98 Å respectively. In the latter case, it is well under the maxi-mum distance, explaining why there was no observable perturbation of that structure when compared to (int. 5), the structure without a crosslink.

Based on this evidence, it is probably not possible to capture an in-trahelical base with a four-carbon crosslink, and a longer one is nec-essary. This also raises the possibility that different crosslink lengths can be used to capture different stages along the base extrusion path-way, though care must be taken to ensure that the intermediates cap-tured are biologically relevant.Synthesis of a C8 crosslink

An 8 carbon (C8) crosslink was synthesized to test this possibility. However, solubility problems prevented it from being able to func-tionalize the DNA. C8 was activated with Aldrathiol to reduce the size and make it less nonpolar. However, this was difficult to purify and was not ultimately used, as a shorter crosslink, such as a 5 or 6 carbon crosslink, would probably be a simpler and equally useful tool. (See Supplementary Information for more detail.)Structure of the 30 minute structure

The X-ray diffraction data for this structure was collected, scaled to 2.44 Å, and solved (Figure 9). Overall, this structure resembles pre-viously solved catalytic intermediates. Several characteristics of this structure should be noted. First, oxoG is not present in the active site.

Second, C3' and the 3' phosphate are not covalently bound. Third, C1' and Lys 249 are not covalently bound. Fourth, the C2'-C3'double bond is trans. Fifth, the structure is in the “open” conformation as-sociated with base extrusion intermediates, instead of the “closed” conformation associated with catalysis. The first four characteristics point to this structure being a late stage intermediate in catalysis, af-ter the hydrolysis of the Schiff base, but before the isomerization to the end-product structure (Figure 10). However, the fifth characteris-tic seems to contradict this conclusion. A late stage intermediate

Most of the characteristics of this structure favor it as a late stage intermediate. During refinement, when oxoG was modeled into the active site, a strong negative density resulted in the σA-weighted fo-fc electron density maps, indicating that there is not enough electron density to indicate that oxoG is in the active site (data not shown). Instead, three water molecules were placed in the active site. The exis-tence of water molecules within the active site in the absence of oxoG is consistent with other structures, including catalytic intermediates (III) (Radom 2006), (V) (Radom 2006), and (VII) (Chung 2004) and base extrusion intermediates (int. 2) (Radom 2007) and (int. 3) (Ba-nerjee 2005).

In addition, there is no electron density between the C3' and the 3' phosphate. Furthermore, the distance between C3' and the oxygen of the 3' phosphate is 5.7 Å, much larger than a normal C-O bond. Thus, in this structure, C3' and the oxygen of the 3' phosphate are no longer covalently attached and β-elimination of the phosphate has already taken place (Figure 10).

Similarly, there is no electron density between C1' and Lys 249, and the distance between the terminal amine and C1' is 4.4 Å, in con-trast with the normal distance of a C=N bond. Thus, lysine 249 is no longer involved in a Schiff base with C1' and has been hydrolyzed off the sugar (Figure 10).

Taken together, these three details of the structure indicate that this intermediate is after (IV) in the catalytic pathway. Thus, this structure is the latest intermediate currently captured along the cat-alytic pathway. Furthermore, this structure provides evidence that hOGG1 is able to catalyze the excision of oxoG even when it is con-strained by crystal packing forces.The geometry of the C2'-C3' bond

In (III), the structure of the reduced initial Schiff base, the oxoG is already excised but remains bound within the active site of hOGG1, and its N9 is positioned within 4 Å of C2' – a plausible distance for

Figure 9. Data collection and refinement statistics of the 30 minute structure. Figure reproduced from Crenshaw 2009.

Figure 10. The active site of the 30 minute intermediate. In pur-ple is the α,β-unsaturated aldehyde that formed from the sugar of oxoG.

hOGG1 dDXL G-complex (Space group: P6522, Cell parameters: a = b = 91.99 5Å, c = 211.293 Å, α = β = 90.0°, and γ = 120°)Data collectiona

Source APS 24ID

Wavelength (Å) 0.97921

Resolution (Å) 50-2.44 (2.53-2.44)

Rsym (%)a 9.0 (100)

Total no. of obs. 220,303

No. of unique obs. 20,494 (1981)

Completeness (%) 99.3 (99.4)

 / σ 40.6 (2.1)

Refinement

Resolution (Å) 50-2.44

No. reflections 20,492

No. of non-hydrogen atoms 2,975

Rfree (%)c 26.27

Mean B value (Å2)d 51.51

Rmsd bond (Å)d Main chain–1.406, Side chain–1.988

Rmsd angle (°)d Main chain–2.393, Side chain–3.167

Ramachandran plot analysise

(% most favored, additional allowed, generously allowed, disallowed)

86.4, 12.8, 0.7, 0.0

aRsym = Σ|I–|/ ΣI where I is the integrated intensity of a given reflection.bRwork = Σ|F(obs)–F(calc)|/ΣF(obs).cRfree = Σ|F(obs)–F(calc)|/ΣF(obs), calculated using 7.5% of the datadFrom CNS.eFrom PROCHECK.


61www.thurj.org

Chemical

Biology

it to deprotonate C2’ (Fromme 2003). Furthermore, addition of oxoG analogues 8-bromoguanine (8-bromoG) and 8-aminoguanine (8-aminoG) can accelerate strand scission of an abasic site. X-ray crystal structures of the initial borohydride trapped Schiff base with the free oxoG purified out and 8-bromoG or 8-aminoG soaked into the hOGG1-DNA complex show that 8-bromoG and 8-aminoG oc-cupy the same position that oxoG occupies in (III), indicating that oxoG may behave similarly to 8-bromoG and 8-aminoG (Fromme 2003). Taken together, this indicates that oxoG likely deprotonates C2' and causes the elimination of the 3' phosphate.

Molecular modeling calculations have indicated that the depro-tonation of the C2' proR hydrogen by oxoG is favored (80% proR to 20% proS) (Fromme 2003). Nevertheless, it is unclear if elimina-tion of the hydrogen would lead to a cis- or trans-α,β-unsaturated Schiff base. Because the phosphate is antiperiplanar to the proR hy-drogen, it is possible that the elimination is concerted and more E2-like, and thus the elimination of the proR hydrogen would lead to a trans-α,-β-unsaturated Schiff base. However, it is also possible that the elimination instead forms the enamine first and then eliminates the phosphate (thus is more E1cb-like), which may lead to either the trans or cis product. In addition, molecular modeling found that the trans product is more stable than the cis product, which is consistent with elimination reactions in general, but although this supports the trans product, it is possible that this reaction is under kinetic con-trol (Fromme 2003). Furthermore, although the end-product struc-ture (VII) contains the sugar is in the ring-closed form, and thus the double bond between C2' and C3' must be cis eventually, this does not preclude the formation of a trans double bond in the initial elimina-tion, since this bond can later isomerize to the cis form (Chung 2004). Unfortunately, the crystal structure containing the α,β-unsaturated Schiff base caught by borohydride trapping (V) was similarly incon-clusive. Models containing both the cis, trans, or saturated (from potential overreduction by sodium cyanoborohydride) bonds fit into the electron density of that structure, with little difference in the Rfree, so it was not possible to distinguish between the possibilities (Radom 2006). Complicating the matter, it is also possible that hOGG1 may produce a mixture of both the trans or cis products, which may be the reason for the ambiguity of (V).

The structure presented in this study resolves this issue. In this structure, the bond between C2' and C3' was determined to be trans. The sugar of oxoG could be in one of three conformations: cis, trans, or ring closed. To determine which of the sugar forms was likely, we modeled in the three different forms, and after performing one round of energy minimization and individual B factor refinement in CNS

(Brunger 1998; Brunger 2007), found that the trans-α,β-unsaturated aldehyde best fit the electron density (Figure 11). In addition, the Rfree values after the round of refinement were 0.2602 for the trans, 0.2613 for the cis, and 0.2630 for the ring closed, indicating that of these three, the trans best described the structure. In addition, even though the cis fit the second best, it is likely that the cis intermediate may be too short lived to capture, because it can very favorably turn into the ring closed form since it is already in the conformation to do so. Thus, the comparison is between the trans and ring closed forms, and the trans intermediate fits the best.

Thus, the bond between C2' and C3' appears to be trans, implying that the β-elimination of the 3' phosphate occurs to give the trans product. Furthermore, this structure also indicates that isomeriza-tion of the trans to the cis product occurs on the aldehyde and not the Schiff base, since in this structure the Schiff base has already been hydrolyzed.

It is likely that this isomerization occurs through a conjugate ad-dition of a nucleophile, and then the subsequent rotation and conju-gate elimination of that nucleophile to form the cis-α,β-unsaturated aldehyde (Figure 12B). Although there has been no direct evidence of this isomerization, the existence of a religated product (VIII) formed from the addition of 8-aminoG to a crystal containing the end prod-uct structure (VII) implies that the phosphate is able to undergo a conjugate addition, albeit assisted by 8-aminoG (Figure 12A) (Chung 2004). Thus, it is possible that phosphate or a different nucleophile such as water, may be able to undergo conjugate addition if driven by an ultimate thermodynamic preference for the ring-closed struc-ture.The “open” conformation

Surprisingly, this structure is in the “open” conformation. The α-O helix is held away from the active site and His 270 is interacting with Asp 322 and has an edge-face interaction with the π-system of Phe 319, which moves it away from the position it occupies in the “closed” conformation. In addition, Gln 315 is also away from its position in the “closed” conformation. In contrast, the previous two catalytic in-termediates were in the “closed” conformation (Figure 13).

At this time, it is unclear whether the “open” conformation of the 30 minute structure is the natural conformation or if it has been ar-tificially introduced by the system. Biochemical experiments done by Charisse Crenshaw have indicated that the disulfide crosslink, which is one of the differences between this structure and the previous cata-lytic intermediates, does not seem to have a significant affect on the conformation of the DNA, in contrast to the role it played in the G complex. However, this does not preclude bias introduced from other

Figure 11. Electron density maps of the 30 minute structure. A) The trans product is purple. B) The cis product is green. C) The ring closed product is yellow. Maps are σA-weighted 2Fo-Fc map contoured at 0.6 σ.



Chem

ical

Bi

olog

y

aspects of the freeze trapping system or from another unidentified bias from the crosslinker.

The previous two structures obtained using this freeze trapping technique were in the “open” conformation; however, it is unclear if this is because the structures were within the base extrusion pathway or if this system biases hOGG1 to the “open” conformation. In addi-tion, it is possible that the “closed” conformation is only necessary for the initial part of catalysis and the conformation may move to the “open” form later during catalysis. Because this structure does not contain oxoG in the active site, the π-stacking interaction be-tween Phe 319 and oxoG, which stabilizes the “closed” structure, is lost. Thus, it is possible that by this stage in the reaction, hOGG1 has moved back into the “open” form.

Moreover, although the end-product structure, which would be after this intermediate, does contain hOGG1 in the “closed” form (Chung 2004), this may be due to the mutation used to crystallize that structure, D268E, which had previously been known to affect the nearby residues 269-271 (Norman 2003). Because the interaction between His 271 and the DNA is a key part of the “open” conforma-tion, the mutation may be biasing the structure of the molecule to the “closed” form. Again, however, it is not clear whether the “open” conformation is an accurate description of this intermediate or if it is an artifact brought on by the system used.

ConclusionIn the active site G complex, guanine is not cleaved out, even

though the protein is catalytically active, and the active site of the guanine is in the “closed” conformation and presumably ready for catalysis. This is surprising because excision of oxoG by hOGG1 occurs within the 30 minute structure. The G complex and the 30 minute structure are in the same crystal forms and are thus presum-ably affected by the same crystal packing forces. Furthermore, the S292C crosslink – at least by itself – does not prevent catalysis, as hOGG1 with the S292C crosslink has been shown to excise oxoG in biochemical studies (Crenshaw 2009). However, guanine does not seem to have been excised, as the N glycosidic bond and the 3' and 5' phosphodiester bonds are all intact. In addition, biochemical experi-ments have shown that hOGG1 with the S292C crosslink does not excise guanine.

Thus, the G complex raises the possibility that there is a “catalytic checkpoint,” or a way that hOGG1 can prevent excision of guanine even in the likely rare case that guanine is placed in the active site. Such a catalytic checkpoint is not a ubiquitous feature of glycosylases: Uracil DNA glycosylase, (which sterically excludes thymine from en-tering the active site) when modified to allow thymine to enter the active site, will excise thymine (Bennet 2006). However, human thy-mine DNA glycosylase does use a catalytic checkpoint. It allows both thymine from T:G mismatches and cytosine from C:G matches into the active site but only excises thymine because thymine is a better leaving group than cytosine (Otwinowski 1997). OxoG is intrinsical-ly a better leaving group than guanine, and it is possible that hOGG1 discriminates between oxoG and guanine at the level of catalysis, and not only during the search process.

The 30 minute structure is a previously uncharacterized late stage intermediate in the catalytic mechanism of oxoG excision. In addi-tion, this intermediate confirms that elimination of the phosphate creates the trans α, β unsaturated Schiff base. It also lends support to the validity of the freeze trapping technique. However, in order to confirm whether the freeze trapping technique is valid, the struc-tures of the other crystals for which data has already been collected should be solved. It is important to confirm that the stage in catalysis the intermediates are captured at correlates with the interval between photocleavage of the adduct and cryoprotection; if not, the progres-sion of catalysis is too stochastic, and thus the crystals are likely to be too heterogeneous to be characterized accurately. Additionally, it would be useful to ensure that early intermediates captured using this methodology are in the “closed” form, like the other early inter-mediates; otherwise, it would indicate that the freeze trapping system artificially warps the structures.

Figure 13. The 30 minute structure is in the open conformation. The 30 minute structure is in blue, with the sugar from oxoG in purple. III, a closed structure, is in green, with its oxoG and sugar in orange. Here π interactions are shown with dashes.

Figure 12. A possible isomerization mechanism. A) This is a proposed mechanism for the formation of VIII (PDB accession code 1M3Q39) from VII. B) This is the proposed mechanism for the formation of VII from VI.

VII. Ring Closed Sugar VIII. Religated Product

8-aminoG accelerated ring opening Conjugate Addition Tautomerization

Conjugate Addition Conjugate Elimination Ring ClosingRotation

VII. Ring Closed SugarVI. α, β Unsaturated Aldehyde


63www.thurj.org

Chemical

Biology

Materials and Methods

For full methods, see the supplementary information online. In brief, hOGG1 was expressed in E. coli and crosslinked to a synthe-sized DNA strand, yielding a hOGG1-DNA complex. This complex was then crystallized and the diffraction data analyzed, yielding the crystal structure of the complex.

ReferencesLindahl, T. Instability and decay of the primary structure of DNA. Nature 1. 362, 709-715 (1993).Friedberg, E. C. DNA damage and repair. Nature 421, 436–440 (2003).2. Lindahl, T. & Wood, R.D. Quality control by DNA repair. Science 286, 3. 1897-1905 (1993)Neeley, W. L. & Essigmann, J. M. Mechanisms of formation, genotoxic-4. ity, and mutation of guanine oxidation products. Chem. Res. Toxicol. 19, 491–505 (2006).Mutsuo Sekiguchi and Teruhisa Tsuzuki, Oxidative nucleotide damage: 5. consequences and prevention. Oncogene (2002) 21, 8895-8904Henle, E. S., and S. Linn. 1997. Formation, prevention, and repair of DNA 6. damage by iron/hydrogen peroxide. J. Biol. Chem. 272:19095-19098.Bjelland S, Seeberg E. Mutagenicity, toxicity and repair of DNA base dam-7. age induced by oxidation. Mutat Res 2003;29:37-80.Oda Y et. al. NMR studies of a DNA containing 8-hydroxydeoxyguanos-8. ine. Nucleic Acids Res. 1991 Apr 11;19(7):1407–1412.Hsu, G. W., Ober, M., Carell, T. & Beese, L. S. Error-prone replication 9. of oxidatively damaged DNA by a high-fidelity DNA polymerase. Nature 431, 217–221 (2004).Shibutani, S., Takeshita, M. & Grollman, A. P. Insertion of specific bases 10. during DNA synthesis past the oxidation-damaged base 8-oxodG. Nature 349, 431–434 Crenshaw, C. M., Structures and Dynamics of DNA Repair Enzymes: 11. Mechanistic Insights into Catalysis and Damage Targeting [Thesis], Cam-bridge MA, Harvard University, 2009, 142 pp.S. David, V. O'Shea, and S. Kundu, Nature 447, 941-950 (2007)12. David, S. S. & Williams, S. D. Chemistry of glycosylases and endonucleas-13. es involved in base-excision repair. Chem. Rev. 98, 1221–1261 (1998).Boiteux S and Radicella JP (2001) Mechanism of stimulation of the DNA 14. glycosylase activity of hOGG1 by the major human AP endonuclease: by-pass of the AP lyase activity step. Nucleic Acids Research, 2001, Vol. 29, No. 6 1285-1292Bruner, S. D.; Norman, D. P. G.; Verdine, G. L. “Structural Basis for Rec-15. ognition and Repair of the Endogenous Mutagen 8-oxoguanine in DNA,” Nature 2000, 403, 859-866.Dodson, M.L., Michaels, M.L. & Lloyd, R.S. Unified catalytic mechanism 16. for DNA glycosylases. J. Biol. Chem. 269, 32709-32712 (1994).Vidal A.E., Hickson,I.D., Boiteux,S. and Radicella,J.P. Mechanism of 17. stimulation of the DNA glycosylase activity of hOGG1 by the major hu-man AP endonuclease: bypass of the AP lyase activity step. Nucleic Acids Research, 2001, Vol. 29, No. 6 1285-1292Sheehan AM, McGregor DK, Patel A, Shidham V, Fan CY, Chang CC. Ex-18. pression of human 8-oxoguanine DNA glycosylase (hOGG1) in follicular lymphoma. Modern Pathology (2005) 18, 1512–1518.Fan CY et. al. Frequent Allelic Imbalance and Loss of Protein Expression 19. of the DNA Repair Gene hOGG1 in Head and Neck Squamous Cell Carci-noma. Lab Invest 2001, 81:1429–1438O. Minowa, et. al., Mmh/Ogg1 gene inactivation results in accumulation 20. of 8-hydroxyguanine in mice, Proc. Natl. Acad. Sci. U.S.A. 97 (2000), pp. 4156–4161A. Klungland, et. al., Accumulation of premutagenic DNA lesions in 21. mice defective in removal of oxidative base damage, Proc. Natl. Acad. Sci. U.S.A. 96 (1999), pp. 13300–13305.K. Sakumi, et. al. Ogg1 knockout-associated lung tumorigenesis and its 22. suppression by Mth1 gene disruption, Cancer Res. 63 (2003), pp. 902–905.Blainey, P. C.; van Oijen, A. M.; Banerjee, A.; Verdine, G. L.; Xie, X. S. “A 23. base-excision DNA-repair protein finds intrahelical lesion bases by fast sliding in contact with DNA,” Proc. Natl. Acad. Sci. USA 2006, 103, 5752-5757.Bowman, B. R.; Lee, S.; Wang, S.; Verdine, G. L. “Structure of the E. coli 24. DNA Glycosylase AlkA Bound to the Ends of Duplex DNA: A System for

the Facile Structure Determination of Lesion-Containing DNA,” Struc-ture 2008, 16, 1166-1174.Lipscomb LA, et. al., X-ray structure of a DNA decamer containing 7,8-di-25. hydro-8-oxoguanine. Proc Natl Acad Sci U S A. 1995 Jan 31;92(3):719-23.Oda, Y., et. al. 1991. NMR studies of a DNA containing 8-hydroxydeox-26. yguanosine. Nucleic Acids Res. 19:1407-1412.Banerjee, A.; Santos, W. L.; Verdine, G. L. “Structure of a DNA Glycosy-27. lase Searching for Lesions,” Science 2006, 311, 1155-1157. Fromme, J. C.; Verdine, G. L. “Structural insights into lesion recognition 28. and repair by the bacterial 8-oxoguanine DNA glycosylase MutM,” Nat. Struct. Biol. 2002, 9, 544-552.Banerjee, A.; Verdine, G. L. “A nucleobase lesion remodels the interaction 29. of its normal neighbor in a DNA glycosylase complex,” Proc. Natl. Acad. Sci. USA 2006, 103, 15020-15025. Banerjee, A.; Yang, W.; Karplus, M.; Verdine, G. L. “Structure of a repair 30. enzyme interrogating undamaged DNA elucidates recognition of dam-aged DNA,” Nature 2005, 434, 612-618.Radom, C.; Banerjee, A.; Verdine, G.L. “Structural characterization of 31. human 8-oxoguanine DNA glycosylase variants bearing active site muta-tions,” J. Biol. Chem. 2007, 282, 9182-9194. Norman, D. P. G.; Chung, S. J.; Verdine, G. L. “Structural and biochemical 32. exploration of a critical amino acid in human 8-oxoguanine glycosylase,” Biochemistry 2003, 42, 1564-1572.Huang, H., Harrison, S. C. & Verdine, G. L. Trapping of a catalytic HIV 33. reverse transcriptase template:primer complex through a disulfide bond. Chem. Biol. 7, 355−364 (2000) Huang, H., Chopra, R., Verdine, G. L. & Harrison, S. C. Structure of a 34. covalently trapped catalytic complex of HIV-1 reverse transcriptase: im-plications for drug resistance. Science 282, 1669−1675 (1998)Fromme, J. C.; Bruner, S. D.; Yang, W.; Karplus, M.; Verdine, G. L. “Prod-35. uct-assisted catalysis in base-excision DNA repair,” Nat. Struct. Biol. 2003, 10, 204-211.Radom, C.T., Structural and Biochemical Exploration of Human 8-Ox-36. oguanine Glycosylase I, in Chemistry and Chemical Biology. 2006, Har-vard University: Cambridge, MA. p. 172.Zharkov, D.O. et al. Structural analysis of an Escherichia coli endonu-37. clease VIII covalent reaction intermediate. EMBO J. 21, 789-800 (2002).Gilboa, R. et al. Structure of formamidopyrimidine-DNA glycosylase co-38. valently complexed to DNA. J. Biol. Chem. 277, 19811-19816 (2002).Chung, S. J.; Verdine, G. L. “Structures of End Products Resulting from 39. Lesion Processing by a DNA Glycoylase/lyase,” Chem. & Biol. 2004, 11, 1643-1649.Lee, S.; Radom, C. T.; Verdine, G. L. “Trapping and Structural Elucida-40. tion of a Very Advanced Intermediate in the Lesion-Extrusion Pathway of hOGG1,” J. Am. Chem. Soc. 2008, 130, 7784-7785.Rodgers, D.W. Cryocrystallography. Structure 2, 1135-1140 (1994).41. Henderson, R. Cryo-protection of protein crystals against radiation dam-42. age in electron and x-ray diffraction. Proc. R. Soc. Lond. [Biol] 241, 6-8 (1990).Petsko, G. & Ringe, D. Protein Structure and Function (Sinauer Associ-43. ates, Sunderland, Massachusetts, USA, 2003).Kavli, B., et. al., Excision of. Cytosine and Thymine from DNA by Mutants 44. of Human. Uracil-DNA Glycosylase, EMBO J., 1996, vol. 15, no. 13, pp. 3442–3447.Bennett MT, et. al. Specificity of human thymine DNA glycosylase de-45. pends on N-glycosidic bond stability. J. Am. Chem. Soc. (2006) 128:12510–12519.Z. Otwinowski and W. Minor, “Processing of X-ray Diffraction Data Col-46. lected in Oscillation Mode”, Methods in Enzymology, Volume 276: Mac-romolecular Crystallography, part A, p.307-326, 1997, C.W. Carter, Jr. & R. M. Sweet, Eds., Academic Press (New York). A.T. Brunger, et. al.,Crystallography & NMR System (CNS), A new soft-47. ware suite for macromolecular structure determination, Acta Cryst. D 54, 905-921(1998).A.T. Brunger, Version 1.2 of the Crystallography and NMR System, Na-48. ture Protocols 2, 2728-2733 (2007). Zbyszek Otwinowski, W.M., Processing of X-ray diffraction data collected 49. in oscillation mode. Methods in Enzymology, 1997. 276: p. 307-326.Read, R.J., Improved Fourier coefficients for maps using phases from par-50. tial structures with errors. Acta Crystallogr. Sect. A, 1986. 42: p. 104. A. W. Schuettelkopf and D. M. F. van Aalten (2004). PRODRG - a tool for 51. high-throughput crystallography of protein-ligand complexes. Acta Crys-tallogr D60, 1355--1363.

w ONLINE ABSTRACTS Volume 2 Issue 2 | Fall 2009


More research papers are available online at www.thurj.org

Y-Box binding protein 1 is a novel substrate of Granzyme A

David Kopelman ‘09, [email protected]

Granzymes are the cell death effector serine proteases in the granules of natural killer (NK) cells and cytotoxic T lymphocytes (CTL) that target cells during an immune response. Among the five human gran-zymes, Granzyme A (GzmA) is the most abundant. It initiates caspase-independent cell death that is mor-phologically indistinguishable from apoptosis. Once delivered by killer cells into infected cells or tumor cells that have been targeted for elimination, GzmA cleaves a number of proteins inside the cell, including multiple components of the SET complex, to induce apoptosis. The Lieberman Laboratory performed yeast two-hybrid screens that identified Y-box binding protein 1 (YB-1) as a protein that specifically interacts with two SET complex proteins, SET and pp32. YB-1 is a multifunctional nucleic acid-binding protein whose overexpression has been implicated in multiple cancers. We investigated the possibility that YB-1 might also be a substrate of GzmA. Treatment of lysate from cells over-expressing YB-1 with GzmA and treatment of whole cells with GzmA and perforin (PFN) demonstrated dose-dependent proteolysis of YB-1. Purified recombinant YB-1 protein was cleaved by GzmA in the arginine-rich region between R234 and R253.

Consolidation and quality:An examination of the effect of hospital

consolidation on the quality of care over timeTariq Nazir Ali ‘09, [email protected]

The study of hospital consolidation and its effect on quality of patient care has been of great interest in both the economic and legal communities as consolidation activity surged in the mid-1990s. Previous studies have reached conclusions that the impact on quality of hospital mergers and acquisitions is either inconclusive or is detrimental. Hospital care before and after hospital consolidation is examined from 1993 to 1998 from patient data across 14 states. Using inpatient mortality and length of stay for CHF patients as quality of indica-tors, the study incorporates time lag variables to test for any time variance in the effect of hospital consolida-tion on quality of patient care. Initially, in the first year post-merger, ospital consolidation results in an initial increase in inpatient mortality and has a negligible effect on length of stay. In subsequent years post-merger, there is a significant decrease in inpatient mortality and length of stay—both indicating an improvement in quality of care. These results seem to counter the conclusions of existing literature and thus, invite further study.

thurj vol. 2 issue 2

Documents

harvard faculty

undergraduate work

edge research

undergraduate experience

issue ofencourage

great students

prizewinning research

graduate students