development and application of control tools for use in optogenetics

Development and Application of Control Tools for #V4,*ESMASSACHUSETTS 1WTUTE

Use in Optogenetics Research OF TECHNOLOGY

by OCT 332014

Patrick Erin Monahan III LIBRARIES

M.S., Cell Biology, Neurobiology and Anatomy, Loyola University Chicago, 2009B.S., Biomedical Sciences, Marquette University, 2005

Submitted to the Program in Media Arts and Sciences, School of Architecture andPlanning, in partial fulfillment of the requirements for the degree of

Master of Science

atthe

Massachusetts Institute of TechnologySeptember 2014

This work is licensed under a Creative Commons Attribution 3.0 Unported License

The author hereby grants MIT permission to reproduce and distribute publiclypaper and electronic copies of this thesis document in whole or in part.

Signature redacted

AuthorPatrick Monahan

Program in Media Arts andSignature redacted Sciences

August 8, 2014

Certifiedby4Prof. Edward S. Boyden

Leader, Synthetic Neurobiology GroupAssociate Professor, MIT Media Lab and McGovern Institute

Departments of Biological Engineering and

Signature redacted Brain and Cognitive Sciences

Accepted byProf. Pattie Maes

Interim Academic HeadProgram in Media arts and Sciences

Development and Application of Control Tools for

Use in Optogenetics Research

by

Patrick Erin Monahan III

Submitted to the Program in Media Arts and Sciences,School of Architecture and Planning,

on August 8, 2014, in partial fulfillment of therequirements for the degree of

Master of Science

Abstract

Optogenetic actuators such as Channelrhodopsin-2 (ChR2) are seven-transmembrane proteins that function as light-gated ion channels. These naturallyoccurring proteins are found in green algae and serve as sensory photoreceptorscontrolling phototaxis. Operationally, they contain the light-isomerizablechromophore all-trans-retinal that, upon absorption of a photon at or around473nm, a conformational change to 13-cis-retinal is induced. This change opens thechannel allowing cations to flow through. In the absence of light, the 13-cis-retinalrelaxes back to the resting all-trans-retinal conformation and the channel closes.When an actuators packaged into a lox-containing Adeno-associated virus is used inconjunction with a mouse that expresses the Cre recombinase enzyme in a specificcell type, cell specific expression of the opsin is achieved. When used with LEDs,lasers, or specifically fabricated light delivery tools, control of very specific neuralnetworks is realized. This thesis provides a review of optogenetics and details thedevelopment and application of a novel wireless device to optically control neuralcircuits and behavior.

Thesis Supervisor:Prof. Edward S. BoydenLeader, Synthetic Neurobiology GroupAssociate Professor, MIT Media Lab and McGovern InstituteDepartments of Biological Engineering andBrain and Cognitive Sciences

3

Acknowledgements

I would like to thank Ed Boyden, Mitch Resnick and Linda Peterson for helping me

complete this thesis.

I would also like to thank the love of my life, my wife Sheena and my son Paddy, and

the rest of my family for putting up with (and sometimes enabling) my shenanigans.

5

Contents

Abstract...............................................................................................................................3

Acknow ledgem ents............................................................................................................5

Contents..............................................................................................................................6

Introduction........................................................................................................................7

Light-Activated Ion Pumps and Channels for Temporally Precise OpticalControl of Activity in Genetically Targeted Neurons................................... 10

A wirelessly powered and controlled device for optical neural control offreely-behaving anim als................................................................................ 44

6

Introduction

This thesis contains two previously published works in the field of

optogenetics. The first paper is a review discussing types of microbial opsins and

their functional characteristics as well as tools for controlling such opsins. The

second paper presents a novel wireless tool for neural circuit control and behavioral

modification using optogenetics.

Optogenetics gives neuroscientists the ability to turn on or off very specific

cell types and neural circuits in the brain in a temporally precise fashion. This

development has enabled the casual manipulation of neural activity to explore the

sufficiency and necessity of different neural patterns as they pertain to normally and

abnormally functioning neural circuits. When incorporated into in vivo systems,

optogenetics allows neuroscientists to dynamically control genetically specific

neurons by delivering light to deep brain nuclei to alter behavior, manipulate

individual components of cortical microcircuits to study local network dynamics

and target malfunctioning pathways to shed light on disease.

Beyond enabling our understanding of brain function, continuing

development of opsins and the related optical toolbox may spawn a new generation

of optical prosthetics for treating currently intractable brain disorders.

The second paper looks at the invention and application of a wireless

transmitter for controlling neural circuits and behavior with optogenetics. This is an

important advancement, as current optogenetic efforts require animals to be

tethered to an optical fiber and/or electrical system in conjunction with a

7

commutator that can restrict natural motion and are impractical for long-term or

high-throughput experimentation.

My specific contributions to the first paper includes building the polyimide

cannula system to deliver light to specific deep structures of the brain, creating

schematics and aiding in the writing of the paper (for example, see Figure 8 and

text).

My specific contributions to the second paper includes performing all

surgical implantations of the wireless devices, monitoring post-surgery mice,

performing behavioral experiments, analyzing results and contributing to the

manuscript (detailed in Materials and Methods).

8

Neuromethods (2012) 67: 305-338DOI 10.1007/7657-2011-100 Springer Science+Business Media, LLC 2011Published online: 13 December 2011

Ught-Acivated Ion Punps and Channelsfor TeMporaly Precise Optical Control of ActivityIn Geneicafly TgPetsd Neurons1

Brian Y. Chow, Xue Han, Jacob G. Bernstein, Patrick E. Monahan,and Edward S. Boyden

AbStrad

The ability to turn on and off specific cell types and neural pathways in the brain, in a temporally precisefashion, has begun to enable the ability to test the sufficiency and necessity of particular neural activitypatterns, and particular neural circuits, in the generation of normal and abnormal neural computations andbehaviors by the brain. Over the last 5 years, a number of naturally occurring light-activated ion pumps andlight-activated ion-channels have been shown, upon genetic expression in specific neuron classes, to enablethe voltage (and internal ionic composition) of those neurons to be controlled by light in a temporallyprecise fashion, without the need for chemical cofactors. In this chapter, we review three major classes ofsuch genetically encoded "optogenetic" microbial opsins-light-gated ion channels such as channelrho-dopsins, light-driven chloride pumps such as halorhodopsins, and light-driven proton pumps such asarchaerhodopsins-that are in widespread use for mediating optical activation and silencing of neurons inspecies from Caeuorhabditu elegans to nonhuman primates. We discuss the properties of these molecules-including their membrane expression, conductances, photocycle properties, ion selectivity, and actionspectra-as well as genetic strategies for delivering these genes to neurons in different species, and hardwarefor performing light delivery in a diversity of settings. In the future, these molecules not only will continueto enable cutting-edge science but may also support a new generation of optical prosthetics for treatingbrain disorders.

Key words: Channelrhodopsin, Optogenetics, Photosensitive proteins, Retinal, Halorhodopsin,Archaerhodopsin, Light-sensitive cation channel, Light-sensitive chloride pump, Light-sensitive pro-ton pump, Photocontrol of behavior

1. IWtdu on

The ability to turn on and off specific cell types and neural pathways inthe brain, in a temporally precise fashion, has begun to enable theability to test the sufficiency and necessity of particular neural activitypatterns, and particular neural circuits, in the generation of normal

'This chapter is an updated version of reference (141).

305

10

306 B.Y. Chow et al.

and abnormal neural computations and behaviors by the brain. Mostelectrophysiological and imaging experiments in neuroscience arecorrelative-comparing a neural signal observed in the brain to abehavior or pathology. In contrast, the power to manipulate specificcells and circuits is opening up the ability to understand their causalroles in brain functions. Over the last 5 years, a number of naturallyoccurring light-activated ion pumps and light-activated ion channelshave been shown, upon genetic expression in specific neuron classes,to enable the voltage (and internal ionic composition) of those neu-rons to be controlled by light in a temporally precise fashion. Thesemolecules are microbial (type I) opsins, seven-transmembrane pro-teins naturally found in archaea, algae, fungi, and other species, andwhich possess light-driven electrogenic activity or contain light-gatedion pores. These molecules, when heterologously expressed in neu-rons or other cells, translocate ions across cell membranes in responseto pulses oflight ofirradiances that are easily achievable with commonlaboratory microscopes, LEDs, and lasers. These molecules havebegun to find widespread use in neuroscience due to their ease ofhandling and use (each is a single gene, under 1 kb long, encoding fora monolithic protein), their lack of need for chemical supplementa-tion in many species (they utilize the naturally occurring chromo-phore all-trans retinal, which appears to occur at sufficient quantitiesin the mammalian nervous system), and their high speed of operation(they can respond within tens of microseconds to milliseconds, upondelivery of light, and shut off rapidly upon cessation of light, asneeded for neuroscience experiments).

Three major classes of such "optogenetic" microbial opsinshave been described to date. The first class, channelrhodopsins, isexemplified by the light-gated inwardly rectifying nonspecific cat-ion channel channelrhodopsin-2 (ChR2) from the green algaeGblamydomonareinhardtii (1), which, when expressed in neurons,can mediate sizeable currents up to 1,000 pA in response to milli-second-timescale pulses of blue light (2-5), thus enabling reliablespike trains to be elicited in ChR2-expressing neurons by blue lightpulse trains (Fig. 1b). Several additional channelrhodopsins usefulto biologists and bioengineers have been discovered or engineered,with faster or slower kinetics, red-shifted activation, and cell-regionspecific targeting, and explored in detail below (6-10). The chan-nelrhodopsins have been used to activate neurons in neural circuitsin animals from worms to monkeys, and have proven powerful andeasy to use. The second class of microbial opsins utilized forbiological control to date, halorhodopsins, is exemplified by thelight-driven inwardly directed chloride pump halorhodopsin, fromthe archacal species Natronomas pbaraonis (Halo/NpHR/pHR;(11)), which, when expressed in neurons, can mediate modest inhibi-tory currents on the order of 40-100 pA in response to yellow lightillumination (12, 13), enabling moderate silencing of neural activity(Fig. 1c). Halorhodopsins have some intrinsic kinetic limitations,

11

Uight-Activated Ion Pumps and Channels for Temporally... 307

al btwo dferen ruromn,

responding to SaM PoHseon train (1 100 Ms)

0 mVL

all C

Ught1.nowotion

Light

I - s

d

7 7.0' I 10 0

Fig. 1. Three dm88.e of microbial opeins that enable optical neural activation and silencing tools. (a) Neuron expressinghChr2-mCherry (at; bar, 20 pmn) and Halo-GFP (aN). (b) Polason tralns of spikes elicited by pulses of blue light (Mwt dhes,in two different neurons. (C) Light-driven spike blockade, demonstrated (Top~for a representative hippocampal neuron, and(Bottom) for a population of neurons (N =7). I-itkiid, neuronal firing induced by pulsed somatic current injection(300 pA, 4 ins). Liht hyperpolarization induced by periods of yellow light (yeI~ow dashws. I-iiectki + Light, yellow lightdrives Halo to block neuron spiking, leaving spikes elicited during periods of darkness intact. Panels a-c adapted fromreferences Boyden at al. (3) and Han and Boyden, (12). (d) Photocurrents of Arch vs. Halo measured as a function of575 +25 nm ight irradlance (or effective light irradlance), in patch-clamped cultured neurons (N = 4-18 neurons foreach point), for low (i) and high (ii) light powers. The line is a single Hill fit to the data. (a) Top, Neural actIvity in a

rersnaieneuron before, during, and after 5 s of yellow light illumination, shown as a spike rester plot (2.*, and as ahistogram of Ins flne urng rate averaged across trials (&uttwn; bln size, 20 me). Bot~v, population average ofinstantaneous filing rate before, during and after yellow light illumination (Nick line, mean; pray liws, mean i SE;S= 13 unIte). Panels d-e adapted from reference Chow at al. (15).

12


with some photocycles taking tens of minutes to complete (e.g.,Fig. 4a, b; (12, 14)), and halorhodopsins also require improvedtrafficking for expression at high levels (15-17). A third class ofmicrobial opsin, the bacteriorhodopsins, is exemplified by the light-driven outward proton pump archacrhodopsin-3 (Arch/aR3), fromthe archacal species Halorubrum sodomense. Arch can mediate stronginhibitory currents of up to 900 pA in vitro (Fig. ld), and in vivo iscapable of mediating near-100% silencing of neurons in the awakebehaving mouse or monkey brain in response to yellow-green light(Fig. Ic, (15, 18)). Protons are extremely effective as a charge carrierfor mediating neural silencing, and proton pumps have greatlyimproved kinetics with respect to halorhodopsins (Fig. 4c, d), aswell as a fast photocycle, and efficient trafficking to membranes.Furthermore, outward proton pumps, perhaps surprisingly, do notalter pH to a greater extent than do other opsins (such as ChR2) orthan does normal neural activity. The broad and ecologically diverseclass of outward proton pumps, which includes blue-green lightdrivable outward proton pumps such as the Leptosphaeria maculansopsin Mac (Fig. 5), enables, alongside the yellow-red drivable Halo,multicolor silencing of independent populations of neurons (15).Also, because the neural silencers Halo and Arch are activated byyellow or yellow-green light, and the neural depolarizer ChR2 isdriven by blue light, expression of both a silencer and a depolarizerin the same cell (either by using two viruses, or by using the 2Alinkerto combine two opsins into a single open reading frame (12, 19))enables bidirectional control of neural activity in a set of cells, usefulfor testing necessity and sufficiency of a given set of neurons in thesame animal, or disruption of neural synchrony and coordinationthrough "informational lesioning" (19).

In the sections below we describe the properties of these threeopsin classes, as well as genetic (e.g., viral, transgenic) and hardware(e.g., lasers, LEDs) infrastructures for using these opsins to parseout the function of neural circuits in a wide variety of animalnervous systems. A theme of this field is that extremely rapidprogress and adoption of these technologies has been driven bytechnology development curves in other fields such as gene therapyand optical imaging. Our hope is to convey a snapshot of thisrapidly moving field as of 2011, summarizing the first half-decadeof its existence, to teach both neuroengineers hoping to innovateby inventing new tools, as well as neuroscientists hoping to utilizethese tools to answer new scientific questions. We will first surveythe general properties of these opsins (Sect. 2), then go into thechannelrhodopsins (Sect. 3), followed by the neural silencingpumps (halorhodopsins and bacteriorhodopsins, Sect. 4), themolecular strategies for delivering these genes to cells for appropri-ate expression in vitro and in vivo (Sect. 5), and the hardware forillumination of these opsin-expressing neurons in vitro and in vivo(Sect. 6).

13

Ught-Activated Ion Pumps and Channels for Temporally... 309

2. Properdasof "Optogenetic"Microbial (Type I)Opsins The three classes of molecule described to date are from organisms

such as unicellular algae, fungi, and archaea, whose native environ-ments and membrane lipid composition are very different fromthose of mammalian neurons. Thus, the performance of thesemolecular tools in neurons can be difficult to predict based solelyupon their properties in other species, and must be assessed empiri-cally for assurance of efficacy and safety. Nevertheless, there areseveral molecular properties that contribute to efficacious, tempo-rally precise optical control of neurons, which can be explored in aunified and logical fashion:

" Initial protein exprezion levels. The efficiency of ribosomaltranslation of a molecule is largely affected by codon optimiza-tion. It is recommended that genes be used that are codon-optimized for the target species.

* Membrane insertion propertie, protein folding, andinteractions wilh local environment. Increased membrane local-ization will result in more functional molecules and thusincreased photocurrents. This property may also be inverselyassociated with the potential property of taicijy since poorlytrafficked or fblded molecules may aggregate in the cytosol andendoplasmic reticulum. On the other hand, if a molecule hasadverse intrinsic side effects, enhanced trafficking may exagger-ate them. Furthermore, any given channel or pump will bestoperate under defined conditions (e.g., chloride conductance,pH, lipid environment, etc.), which may not exist in a giventarget cell type.

* Innate conductance and permeability Channels translocatemore ions per photocycde than pumps, since they open up apore in the membrane. On the other hand, pumps can moveions against concentration gradients, unlike channels. Eachopsin furthermore passes a precise set of ions in a specificcellular context, and not others.

- Photocycle kinetics. Both light-driven channels and pumps aredescribed by a photocycle, the list of states that a molecule goesthrough after light exposure, including ion-translocating or ionpore-forming steps. The faster the photocycle, the more tem-porally precise the molecular function might be, and fur apump, the more ions will be translocated. (For a channel, afaster photocycle may result in the channel entering the ionpore-forming state more often, but may also reduce the timespent in the open state.) If a molecule enters an inactive

14


photocyde state for an enduring period of time, it may beeffcctively nonfunctional.

" Photositivity Molecules may require different amounts oflight to begin moving through their photocyce, based on thechromophore absorption efficiency. Furthermore, from a end-user standpoint, effective photosensitivity will appear to be afunction of the overall photocycle; fur example, a pump thathas a slow photocycle may appear to be light insensitive(because incident photons may have no effect on the moleculeduring the photocyce), whereas a channel that inactivatesextremely slowly may appear to be light sensitive (becauseeach photon will result in large charge transfers).

* Action spectrum. Different molecules are driven by differentcolors of light. Multiple cell types can be orthogonallyaddressed with different colors of light, if they express opsinswhose action spectra minimally overlap.

* Ion seectivity Unlike traditional electrodes, microbial opsinscan generate ion specific currents, since they will pass specificions such as chloride (CL) or calcium (Ca' 2 ). This opens upnovel kinds of experimental capability, such as the ability to testthe sufficiency of a given ion, in a given location, for a givenbiological function.

We will, in the fbllowing sections, frame current knowledgeabout cell-type specific optical control of neurons, in the context ofdecades of research in structure-function relationships of microbial(type I, or archaeal) opsins. In many ways, these molecules aresimilar in tertiary structure to mammalian (type II) rhodopsins(20), the pigments that confer photosensitivity to the rods andcones of the human retina. Both types are composed of seventransmembrane (7-TM) cx-helices, linked by six loop segments,and their photosensitivity is enabled by a retinal bound to a specificlysine residue near the C-terminus, forming a Schiff base thatundergoes a trans-cis or cir-grans isomerization upon illumination,that then induces conformational changes in the protein. However,they are evolutionarily unrelated, and their differences have impor-tant implications for their use in perturbing neuronal activity.Mammalian rhodopsins (21) are very sensitive photon detectors,optimized for sensitivity rather than speed. They utilize 11-cisretinal as the primary chromophore, which isomerizes to all-transretinal upon absorbing a photon. The resultant structural changeactivates an associated G-protein, transducin, which then initiates acascade of secondary messengers. The all-trans retinal dissociatesfrom the opsin, is converted back to its 11-cis form, and thenreassociates with the apoprotein to reconstitute a functional mole-cule-a process that typically takes hundreds of milliseconds, tooslow to enable fast control ofneurons in the central nervous system.

15

Light-Activated Ion Pumps and Channels for Temporally... 311

On the other hand, a microbial opsin utilizes all-trans retinal as itschromophore, which isomerizes to 13-cis retinal upon absorbing aphoton. The chromophore does not undergo a quasi-irreversibledissociation event, but rather thermally relaxes to its active all-transform in the dark (although this process can be facilitated by light).The trans-cis isomerization sets off several coupled structural rear-rangements within the molecule that accommodate the passiveconduction or active pumping of ions (22-24). Ultimately, thismeans that at the expense of light sensitivity, archaeal opsins candeflect the membrane potential of a cell on the scale hundreds ofmicroseconds to a few milliseconds. However, it should be notedthat genetically targetable, optical neural silencing has also beendemonstrated using mammalian G-protein coupled receptors,which can couple to potassium channels (4), and genetically target-able optical neural activation has been demonstrated using mela-nopsins and invertebrate-style rhodopsins, at the price of temporalprecision (25, 26).

As an exemplar of a well-characterized microbial opsin reagent,with crystal structure and photocycle both well-described, Fig. 2shows the crystal structure of the light-driven chloride-pump halor-hodopsin (Fig. 2a; (27)), as well as schematized structural rearran-gements that are hypothesized to occur as the molecule pumps achloride ion across the membrane and into the cytoplasm (Fig. 2b;(24)). While it is convenient to consider the molecular tools dis-cussed here as toggle switches for turning neurons on and off, it iscritical for use of these opsins to realize that the translocation ofions by microbial opsins is not as simple as a two-state toggleswitch. The structural rearrangements constitute an active advance-ment through a complex photocycle with various intermediatestates beyond the initial phototransition (Fig. 2c). There exists avery rich literature on type I microbial opsins from an evolutionaryand protein structure-function perspective. The canonical mole-cules include the proton-pumping bacteriorhodopsin (BR), thechloride-pumping halorhodopsin (HR/sHR/HsHR) from Halo-bacterium salinarum (halobium), and the halorhodopsin from N.pharaonis (Halo/NpHR/pHR). These were some of the firstmembrane proteins crystallized, and a myriad of structure-functionstudies have been performed on these molecules (11, 14, 20,22-24, 27-45). These studies are not reviewed here, but it isimportant to point out their existence because much of what weknow with respect to the photocyde and structure of opsins comesfrom these studies and from sequence homology of novel opsins tothese canonical molecules. As we show later, for example, a deepunderstanding of the literature has enabled some researchers toderive powerful new variants of channelrhodopsin (5, 6, 8, 10),even though no crystal structure exists for this molecule.

16

. .... ..... __ :: ........ _ _ - .... .... ..... - -


b

Uptake and isomenzation

HR640 -20 m.

HRlSa0

HROOO

+ C1." M * o E C ( h V

H R 410

R640 H1 +HpS,0)

HR520*-/

Release Switch

HR520

Fig. 2. Structure and function of halorhodopsin. (a) Crystal structure of halorhodopsin, which is composed of7-transamembrane ot-helices (7-TM) and a retina chromophore that forms a Schiff base to a lysine near the C-terminus(from Kolbe et al. (27)). (b) Schematic of the halorhodopsin structural rearrangements and their relation to pumping activityat various points in the photocycle. Image modified from Essen (24). (C) The halorhodopsin photocycle at high-powercontinuous illumination on the timescale of a typical photocycle (.e., condItions used for neural silencing, >fewmilliseconds). The HR410 intermediate is the origin of the long-lived inactivation in neural silencing (e.g., Bamberget al. (14); Han and Boyden (12)).

. Optical NeuralStmulation:Channel-rhodop-ins

Channelrhodopsins arc the primary photoreceptors in the eyespotof the unicellular algae that are responsible for phototactic andphotophobic responses (46-48). Their name is derived from thefact that despite the sensory function, the 7-TM segment is in itselfa light-activated ion channel. While the channel pore and propertiesremain poorly understood, it has recently been realized that chan-nelrhodopsins are likely proton pumps like many other microbialopsins, but with a leaky step in the photocycle during which theopsin lets positive charge into cells (49). In C. reinhardiii, twoseparate channelrhodopsins were originally identified (48), onewith fast kinetics but poor light sensitivity, channelrhodopsin-I(ChRL) (50), and another with slower kinetics but improved sensi-tivity, channelrhodopsin-2 (ChR2) (1). Two more channelrhodop-sins from Valvax carteri(VChR1, VChR2) have also been identified(7, 51), and as we discuss later, many more ChRs are expected to

17

aa

HR'500 HR580

C

14A 580

R_410

HR6a5 It

Isomertzation

HR565 HR600

11. toewbwaace,

&Wd Coniurt

32. Basic MWewOf kwtfts awWaI&gfh Se"ecUy

Uight-Activated lon Pumps and Channels for Temporally... 313

exist. ChRL-style channelrhodopsins have red-shifted action spec-tra (peak Achj = 500 nm, Avchm = 535 nm) relative to ChR2(peak chR2 = 470 rnm), and thus in principle ChRi-stylc andChR2-style opsins could be used together to drive separate sets ofneurons with two different colors of light, if suitably spectrallyseparated opsin pairs could be found.

Channelrhodopsins (abbreviated as ChRs or chops) are light-activated, inwardly rectifying cation channels that are, at neutralpH, permeable to physiologically relevant cations such as H', Na',K, and Ca2*, with permeabilties (relative to sodium) of 1 x 106, 1,0.5, and 0.1 respectively (1, 6, 46, 50). It is of particular note thatthe proton conductance (GH+) is 106-fold larger than the sodiumconductance (GNa.), and thus near physiological pH, perhaps halfthe photocurrent is carried by protons (46); thus, ChRs may rapidlyequilibrate the intracellular pH with its environment (10). Kineticselectivity analysis has shown that the mechanism ofion selectivity islikely to be due to differential binding affinity of channelrhodopsinchannel residues for different ions, not differential ion transportrates (46).

It was originally believed that ChRi was a selective protonchannel (50); however, it was later discovered that the poor photo-currents at mammalian pH were likely attributable topoor membrane localization (6), and the apparent lack of sodiumcurrents in the original report were due to the low pH used toperform experiments in that study; the sodium conductance ofChR1 lessens at low pH (6, 46), unlike that of ChR2 (6). Thishighlights what is a recurrent theme throughout this chapter, thateffective conductance in a heterologous system is determined notonly by the innate kinetic and transport properties of the moleculebut also by its trafficking and performance in the environment ofthe heterologous system.

The single ion channel conductance of ChR2 has been esti-mated at 50 fS (1), which corresponds to approximately 3 x 10"ions per second, or 300 ions per photocycle event, assuming a10 ms turnover. This is considerably less than a typical voltagedependent sodium channel that may have a conductance on theorder of -10 pS. It has been estimated from electrophysiologicaldata that 10 -106 membrane embedded ChR2 molecules arerequired to cause reliable spiking in cultured rodent hippocampalneurons (52), with saturation blue light densities of several milli-watts per square millimeter both in vitro (10) and in vivo (53).

Figure 3a shows a typical photocurrent trace from a voltage-clamped neuron expressing ChR2 (top), and the spiking patternthat would result in current-clamp mode (bottom). There is a largetransient peak with an opening time constant near 1 ms (1, 6, 10),although photocurrent onset can be measured at <200 ps (1, 3,

18

............ ....... ............... ......... .. ... ...... .. .. ...... ........ .. ........


a" mpu

400PAK

SW

bD470 hv

hv P500

P480 C-H-(EC)

P520 4 AZ+H-(CP)#M+leak

Fig. 3. Channelrhodopsin kinetic and photocycle properties, and impact on neural activity. (a) ChR2 currents elicited in avoltage-clamped hippocampal neuron expressing ChR2 and illuminated by blue light (b$*, and ChR2-drven spikes elicitedin a current-clamped hippocampal neuron (three repetitions of the same blue light pulse in the same neuron) (bottom),under 1 s of blue light illumination. Adapted from Boyden et al. (3). (b) The photocycle of ChR2 determined by acombination of spectroscopy, site-directed mutagenesis, and electrophysiology, adapted from Feldbauer et al. (49).The inner circle summarizes the effective appearance of the photocycle, an approximation to the outer photocycle.(c) The interplay between photocycle, wavelength, and electrophysiological activity. ChR2 is excited with blue light for abrief period, and then a green light is turned on. The photocurrent initially diminishes because the channel is forcedto close, but then increases because the green light also pumps the molecule into its most highly efficient state.Image modified from Bamann et al. (54).

54); this transient peak quickly decays to a stationary component thatis typically <20-50% of the initial peak photocurrent (1, 3, 6, 10).Upon removing the light, ChR2 doses with a time constant of10-20 ms (1, 6, 10). The transient photocurrent peak is highlydependent on the illumination intensity (51, 55) and history (1, 3,10); the history dependence results from a desensitization of thetransient component that takes -5 s to recover from in the dark (3).The stationary component on the other hand, is less photosensitiveand effectively history-independent (55). ChR2 absorbs maximally at460 nm (1, 10), and the action spectra of both temporal componentsare nearly identical in ChR2.

The large and fast-onset peak enables ChR2-expressing neu-rons to spike with exquisite temporal precision on the millisecondtimescale (Fig. 1b), the timescale of an action potential. However,the large inactivation (or alternatively, the small stationary compo-nent) and its slow recovery in the dark, as well as the slow dosingrate of -10-15 ms, ultimately limit the ability to drive reliable spikerates >25 Hz (3, 10) because (1) the stationary photocurrent maybe too small to sufficiently depolarize a neuron to spike threshold,and (2) the channel cannot physically close quickly enough toenable dc-inactivation of sodium channels. It should be noted,though, that many neurons, such as pyramidal cells, seldom fire

19

C

40 .

100 pA200 m


action potentials at this rate on the individual neuron level (vs.population synchrony or rhythmogenesis).

ChRi-style channelrhodopsins (VChRI, ChR1) (7, 50) on theother hand demonstrate dramatically faster kinetics than ChR2-stylcchannelrhodopsins (VChR2, ChR2). The stationary photocurrentsof ChRis are >70% of the peak photocurrents, and the channelsopen and close approximately two- to threefold faster than doesChR2. Therefore, one would expect that given comparable expres-sion, protein folding, membrane localization, and photosensitivity(i.e., factors contributing to effective conductance), ChRis wouldbe capable of driving spike rates with greater fidelity than ChR2s.However, poor membrane expression limits the performance ofnatural ChRL -style channelrhodopsins (7,50). Chimeras composedof the first five helices of ChR1 and last two helices of ChR2 havebeen constructed (6, 10, 56), and these new variants exhibit thesmall inactivation and action spectrum of ChRi, but the overalleffective conductance of ChR2. These structure-function studiesare discussed in detail later in this chapter. A point mutant of thischimera dubbed "ChIEF" (based on its composition as a [Ch]annelrhodopsin chimera with an [1]190 V substitution withdomains swapped between ChRI helix-[E] and ChR2 helix-[F]),developed by Tsien and coworkers (10), appears to be a highlyimproved tool for stimulating neurons. Its large stationary photo-current and very fast channel closing, the latter conferred by the1190 V mutation, contribute to far more reliable spiking (up to100 Hz) than ChR2.

Based on the available characterization of the channelrhodop-sins from V carteri (7, 51), the general characteristics are similar tothose of the analogous molecules in C. reinuardhii. VChR2 andChR2 have nearly identical photocycles and action spectrum (51).VChRI and ChR1 exhibit the similar reduced inactivation, and areboth red-shifted from their respective VChR2/ChR2 counterparts.It has been proposed that VChR1 could be used for multicoloroptical stimulation in conjunction with ChR2, which is blue-shiftedby -70 nm, but further improvements are likely required for reli-able spiking because the VChR1 photocurrents are unfortunatelyapproximately four- to fivefold smaller (7), and also there is signifi-cant spectral overlap between VChR1 and ChR2.

3.1 06taed MWs As previously mentioned, it is critical to realize that the transloca-of ONNO aw tion of ions by opsins is not as simple as the operation of an on/offWamehwego SeWa~fsty switch, but rather these opsins traverse a complex photocycle with

various intermediate states beyond the initial opening of the chan-nel. Figure 3b shows the photocycle of ChRs based on photophy-sical studies performed primarily by laser flash spectroscopy,physiology, and site-directed mutagenesis (8, 49, 51, 54, 57).Importantly, the intermediates of the photocyces themselves canalso undergo photoreactions, and thus they may be optically driven

20


or "short-circuited" (54) between photointermediates at muchfaster rates (Fig. 3c). The ChR2 photocyde initially begins in itsclosed dark-adapted state D470 (where the number in the statename corresponds to the peak light absorption, in nm, of themolecule in that state). The channel opens when D470 absorbs aphoton, after which the molecule will become a green absorbingphotoproduct or P-intermediate, P520, via thermal relaxation fromshorter lived photoproducts. This initial cascade of events takes0.2-1.5 ins, depending on the transmembrane potential. Theopen ChR2 can be closed by either optically pumpingP520 -+ D470 with green light, or by decaying to P480 (via a yetto be determined intermediate), a process that takes -6 s. Theinactivation toward the stationary photocurrent may be due tomolecules making the P520 - P480 transition, rather than theoptically induced P520 -- D470 transition that would allow themolecule to quickly open again. Assuming that ChRI and ChR2photocycles arc topologically similar, e.g., the ChR2 D470 andP480 equate to the ChRI peaks at 464 and 505 nm, this interpre-tation of the transient and stationary photocurrents is consistentwith the finding that, for ChRL, the stationary photocurrent is red-shifted from the transient photocurrent (6).

The various wavelengths of absorption of rhodopsins and theirintermediates throughout the photocycle arises from the differentconformations of the chromophore and its local environment,which influences the chromophore charge distribution and theprotonation of the retinylidene Schiff base. Figure 3c demonstratesthis complex interplay in an experiment by Bamberg and coworkers(54). After blue light-excited ChR2 has reached its steady state, agreen light costimulation is introduced. The photocurrent brieflydiminishes because the open channel is forced to close, but thestationary photocurrent quickly improves because many moleculeshave been pumped back into their highly efficient, peak producingstate. Thus, it is possible that slightly red-shifted or broadbandillumination of ChR2 may strike a balance between optimally excit-ing the dark state (transient component) and reprinting the darkstate (driving the red-shifted intermediate photoproduct). As wediscuss, optimal silencing with N. pharaonis halorhodopsin is anal-ogously achieved by using both yellow light to hyperpolarizc theneuron and blue light to drive the molecule out of its inactive state(12, 14).

3A Mtants As previously mentioned, even though no ChR crystal structureaW Valants exists at the time of this writing, useful structure-function studies

have been performed based largely on sequence homology toH. salinarum bacteriorhodopsin. The E90Q mutation (57) hasincreased sodium selectivity (with respect to GH,) vs. wild-typeChR2, and the H134Rmutant (5) demonstrates increased conduc-tance by approximately twofold. Various mutations to C128 (8)

21


corresponding to bacteriorhodopsin T90, drastically slow down therate of ChR2 closure from the open state, thus effectively creating abistable open P520 state until illuminated with green light. Bylengthening the time that ChR2 spends open on a per-photonbasis, this mutation effectively decreases the amount of lightneeded to activate the channel, at the expense of temporal preci-sion. In contrast, the E123T mutant, combined with the H134Rmutant, speeds channel closure and increases the precision of neuralaction potential firing at the expense of photocurrent and lightsensitivity (58), resulting in a reagent nicknamed ChETA.

Chimeras of ChRI and ChR2 have been constructed by severalresearchers (6, 10, 56), one of which was that composed of ChRLhelices A-E and ChR2 helices F-G (called abcdeFG, ChEF, orChR1/2s/ 2 by various investigators). These chimeras displayedthe small inactivation of ChR1, but the large photocurrents ofChR2 on account of improved membrane localization and lightsensitivity (based on quantitative confocal fluorescence microscopy,(6)). An 1190 V substitution to ChEF led to the molecule,"ChIEF," capable of driving more reliable fast spiking due to themuch larger stationary current and faster channel closing kineticsafter light offset (10). During these studies, it was also discoveredthat a single point mutation to wild-type ChR1, E87Q, eradicatesits pH-dependent spectral shifts, and increases inactivation duringillumination (56).

The fact that the poor effective conductance of ChRi can belargely attributed to membrane localization rather than its photo-physical properties highlights the importance of considering andimproving the trafficking of heterologously expressed molecules. Inparticular, ChRs are not localized to the outer membrane, butrather the eyespot, in C. reinhardtii, and the membrane composi-tion of the organism is less than 20% phosphoglyceride (59), aprimary lipid type in mammalian neurons. As we later discuss inthe context of halorhodopsins, the use of signaling peptides canimprove outer membrane localization and reduce aggregation inthe cytosol, endoplasmic reticulum, and Golgi apparatus.

Along similar lines of using signal peptides to alter trafficking,the myosin binding domain (MBD) peptide promotes subcelularlocalization of opsins to neuronal dendrites (9). This subcellularlocalization strategy may prove to be helpful for enabling driving ofelectrical activity in specific neural compartments, or for high-reso-lution connectomic mapping in vivo. Two-photon excitation is apowerful laser excitation technique that enables submicron resolu-tion in 3-D (60) relatively deep into the brain (-750 pm, or asignificant fraction of the thickness of the mouse cortex), butits ability to induce action potentials in a neuron expressing ChR2is limited by the interplay between molecule density and the extentof optical depolarization with respect to time (61, 62).The probability of inducing an action potential, at low powers

22

318 B.Y. Chow et a].

that are not destructive to tissue, is relatively low using a traditionalraster scan because the fraction of molecules excited at any point intime is small, and most photons that do hit the membrane arewasted (since the open time of ChR2 is long relative to a f&mtosec-ond laser photon delivery rate). Thus, with most conventional two-photon laser scanning methods, the aggregate contributions of theserially excited molecules never sufficiently depolarize the wholeneuron to spike threshold. However, Rickgauer and Tank havedemonstrated that neurons expressing ChR2 can be reliably excitedby two-photon microscopy by optimizing the scan pattern to deliverlight optimally to the cell membrane, in a fashion that reaches themaximum surface area while minimizing wastage of photons onalready-light-driven channelrhodopsin molecules (62).

Unlike microbial rhodopsins from archaea, significantly less isknown about the photoelectrogenic molecules of unicellularalgae, the only organisms known to date to have naturally occurringlight-activated channels. Photoelectric responses have beenmeasured in several green flagellates, as well as phylogeneticallydistant cryptophytes (47, 63-65). Interestingly, the two-compo-nent phototaxis strategy employed by C. reinhardtii, in which theresponse is mediated by a fast (ChRi) and slow (ChR2) rhodopsin,appears to be general (47), which begs the question whether chi-meras of their respective rhodopsins will also result in kineticimprovements and variants with interesting properties. Thus,as more phototaxis-mediating rhodopsins are isolated andsequenced, or as perhaps new depolarizing rhodopsin types arediscovered, new molecular tools for controlling neurons will surely

emerge.

4. Opt|cal NeuralSilencing: Halorho-dopsiks anddRCI~riormo-dopsins

Whereas traditional electrodes can stimulate neurons with temporalprecision (albeit without cell type specificity), they are incapable ofsilencing neurons in order to assess their necessity for given neuralcomputations, behaviors, and pathologies. Therefore, there is a largeneed for spatio-temporally precise methods for optical inhibition ofneurons. Inwardly rectifying chloride pumps and outwardly rectifyingproton pumps, halorhodopsins (HRs, hops) and bacteriorhodopsins(BRs, bops), respectively, are electrogenic pumps that when heterol-ogously expressed are capable ofsufficientlyhyperpolarizing a neurontosilence its activity (Fig. c-e;(12, 15,66)). They are thus far knowntoexistin everykingdom except for animals: archaca (22,23,67-69),bacteria (70-76), fungi (77, 78) and algae (79). In addition to theiropposite electrophysiological effect, HRs and BRs differ primarily

23

a Dvrufy

4.1. Iredpssnw:

ad coitet


from channelrhodopsins in that their physiological functions is chieflydue to their role as pumps as opposed to operating as passive chan-nels, and thus can translocate ions against concentration gradients(but, typically only one ion per photocyce). Much is known aboutthe photocydes and structure-function relationships of HRs and BRsbecause they have been crystallized (24, 27, 28, 80, 81) and heavilycharacterized via spectroscopy, mutagenesis, and physiology for dec-ades.

This section focuses on two molecules in particular: N. phar-aonis halorhodopsin (Halo/NpHR) and H. sodomens bacteriorho-dopsin (Arch/AR-3), also known as an archaerhodopsin (that is, abacteriorhodopsin from the halorubrum genus). Halorhodopsinswere shown in 2007 to be capable of mediating modest optical neuralhyperpolarizations, and since have been improved in trafficking toboost their currents (12, 13, 16); bacteriorhodopsins were shown in2009 to be able to mediate very powerful and kinetically versatilesilencing of multiple neural populations with different colors of light(18). We discuss in the following sections "Conductance, permeabil-ity, and context" and "Kinetics and wavelength selectivity" of halor-hodopsins and bacteriorhodopsins for these two classes separately,followed by a joint discussion of the "Mutants and variants" andgenomic "Diversity" in a unified section.

N. pharaomis halorhodopsin (NpHR, Halo) is a highly selective,inwardly rectifying, chloride pump, which can also conduct largermonovalent anions (82). It has a reversal potential of approximately-400 mV (82), and its chloride-dependence of pumping activity(full- and half-saturating chloride concentrations: [Cl-].,. = 20mM, [Cl~]1, 2 = 2.5 mM) (83) is appropriate for operation in

mammalian cells, in contrast to H. malmarum halorhodopsin, whichis not capable of effective operation in mammalian neurons (15),presumably because of its large chloride dependency: [Cl-]. .,.= 5 M and [Cl~]1 2 = 200 mM (83, 84). In the absence of anysignal peptide sequences to improve trafficking and membrane local-ization, Halo has been reported to generate 40-100 pA ofhyperpo-larizing current (12, 18, 66), with the differences in measuredphotocurrents between studies largely attributable to the power andwavelength of excitation used. This photocurrent is approximately10-25-fold less than typical peak depolarizing currents generated byChR2, highlighting one potential disadvantage inherent to a pumpthat translocates one ion per photocycle (e.g., Halo) vs. a channel thatconducts 300 ions per photocycle (e.g., ChR2). To mediate thesecurrents, there are an estimated ten million membrane-embeddedHalo molecules per neuron (as assessed in hippocampal neuron cul-ture). Because the expression levels are so high, Halo is known toform puncta or intraceilular blebs, aggregating in the endoplasmicreticulum (ER) and Golgi apparatus (16, 17). These issues are some-what addressed by attaching trafficking enhancement sequences to

24

... ....... .......... ... ...... .... ..... - I


42. ffahw lepshsw:bmatIcs aWVWM" SAIWfVIRy

the molecule, e.g., sequences from the KiR2.1 protein (eNpHR,eNpHR3.0), which increases the effective conductance several foldby increasing membrane expression (16, 85).

Recently, we have discovered that the crux-halorhodopsin (HRfrom the haloarcula genus) from Haloarcula marismortui, canoni-cally known as cHR-5 (69, 86), produces similar photocurrents toHalo with more uniform expression; even when highly over-expressed under high copy number transfcction conditions, nopuncta or intracellular blebbing is observed (15). This moleculemay better express than Halo in vivo, but it is unknown at thismoment whether it will ultimately be more efficacious at alteringmammalian behavior, given their statistically insignificant differ-ence in photocurrent. However, the prolactin (Prl) ER-locationsequence in conjunction with a signal sequence from a MHC classI antigen triples the Halo photocurrent (15, 18); we are now tryingout multiple trafficking sequences in combination to see if theyboost current further. However, it is important to note that ifhalorhodopsins have other side effects that are due to the protein'sintrinsic properties-as an example, one paper quantitates the sig-nificantly altered neuronal capacitance that results from expressinghalorhodopsin in neurons in vivo (87)-then boosting expressionmay only make such side effects worse.

N. pbaraonis halorhodopsin is capable of silencing weakly firingneurons on the millisecond timescale with its -100 pA-scale cur-rents, with rapid onset and offset (12), but during long periodsof illumination, all halorhodopsins that we have tested so f&r andthat have current (from N. pharamis, H. sodamense, Haoarculavalismortis, H. marismortui, and Salinibacter ruber) inactivate bapproximately 30% every 15 s of illumination at 1-10 mW/mmyellow (593 nm) light (Fig. 4a, b; (12, 18)). This slow inactivationstands in contrast to ChR2, which responds to light with a largetransient peak that decays within seconds, followed by a stablestationary photocurrent. For all of the halorhodopsins namedabove, recovery in the dark from light-induced inactivation isslow, with a time constant of tens of minutes, as has been describedfor some halorhodopsins earlier (12, 14, 39) (Fig. 4a, b). This long-lasting inactivation property may hinder the use of halorhodopsinsfor silencing for prolonged periods, e.g., during repeated behavioraltrials. Importantly, for all halorhodopsins investigated, the inactivephotoproduct can be driven back into its active pumping state with ashort (e.g., subsecond duration) pulse of blue or UV light (12, 14);thus, optimal use of Halo for neural silencing requires both yellowand blue light to be delivered to the same set of neurons, which ispossible (88) but can complicate optics setups.

The Halo photocyde is shown in schematic form in Fig. 2c.The time constants listed are the limiting ones, with all othertransitions <100 ps. It should be noted that the names for

25

Qight-Activated Ion Pumps and Channels for Temporally... 321

C

Halo

1 20% Tgp

15!so%

I00% '

Arch15s 30s 30sda da* dark

30s 30sdark dark

15S 18 Is 1 is 1 Sight light light ht ht light

1OmV ""P'hUUJLIJL"S,5s 0 I

0.890.7f1 T% 06~ 30 60

1 4 0 3 W 9 1 1

s Tirne trom end of in"ta 15 second iMumination (sec)

LL

Fg. 4. inetic comparisons between halorhodopsins and archasrhodopeins. (al) Time course of Halo-mediatedhyperpolarizations in a representative current-damped hippocampal neuron during 15 s of continuous yelow lightfollowed by four 1-s test pulses of yellow light (one every 30 s, starting 10 s after the end of the first 15-s period ofyelow light). (u1) Time course of Halo-mediated hyperpolarizaton for the same cell exdibited in (a), but when Halofunction is facilitated by a 400-ms pulse of blue ight in between the 15-s period of yellow ight and the first 1-s test pulse.(b) Population data for blue-ight facilitation of Halo recovery (N= 8 neurons). Plotted are the lypMrpolartzOnS elicited bythe four 1-s test pulses of yellow light, normalized to the peak hyperpolarizatlon induced by lt original 15-s yellow lightpulse. Dots represent mean SEM. ack d*s represent experiments when no blue ight pulse was delivered (as inFig. Sal.). (Orm MW dols represent experiments when 400 ms of blue ight was delivered to facilitate recovery (as inFig. Sali.). (c) Raw current trace of a neuron lentivirally infected with Arch, illuminated by a15 s ight pulse (575 25 nm,irradiance 7.8 MW/mm2), followed by 1 s test pulbes delivered starting 15,45, 75,105, and 135 s after the end of the 15 slight pulse. (d) Population data of averaged Arch photocurrents (N = 11 neurons) sampled at the times indicated by thevertcal dttd nes that extend into Fig. 4c.

canonical spectroscopic states of H. salinarum halorhodopsin-theK, L, N, and 0 photointermediates-have not been used herebecause the order of the N- and O-states in N. pharaois halorho-dopsin is still somewhat debated (42, 89, 90). In the dominantphotocycde, Halo absorbs a photon and then within tens of micro-seconds, quickly releases a chloride ion into the cytoplasm duringthe HR520 -* H640 transition, via short-lived intermediates that"switch" the chloride location within the molecule from the extra-cellular loading domain to the cytoplasmic release domain. Themolecule then reisomerizes from the HR640 state and takes up achloride ion from the extracellular side, a process that takes-1.5 ms; it then forms the HR' state, which finally relaxes to theactive ground state with a time constant of -20 ms. However,

26

a

b

. .. .... .... ............. .... ..... ...... .. .....


and CO~grt

halorhodopsins can enter an alternate photocycle (middle trajec-tory within Fig. 2c), most notably under prolonged or brightillumination, as might occur during in vivo neural silencing. The13-cis retinalydene Schiff base becomes deprotonated (releasing aproton into the cytoplasm and thus introducing a small depolariz-ing proton current) and forms a long-lived intermediate HR410(14, 39). In the dark, the halorhodopsin will remain in this inactivestate for a duration on the order of 30 min (39). This formation ofHR410 is the origin of the long inactivation observed in neuronsexpressing halorhodopsin and the ability to recover the active stateusing the short blue light pulse (12, 14). In contrast, as we discussin detail in the next section, archaerhodopsins (bacteriorhodopsinsfrom the Halorubrum genus) spontaneously recover in the darkunder physiological conditions.

Arch, canonically known as archaerhodopsin-3 (AR-3) fromH. sodomenxs, is a yellow-green-light sensitive outwardly rectifyingproton pump with nearly an order of magnitude increase in hyper-polarizing current over any characterized natural halorhodopsin(15, 18), attaining neuronal currents up to 900 pA in vitro inresponse to light powers easily achievable in vitro or in vivo. Theefficacy of these proton pumps is surprising, given that protonsoccur, in mammalian tissue, at a millionfold-lower concentrationthan the ions carried by the optical control molecules describedabove. This high efficacy may be due to the fast photocycle of Arch(see also (91,92)), but it may also be due to the ability of high-pKaresidues in proton pumps to mediate proton uptake (91, 93).

Arch is a highly efficacious tool in vivo, with cortical neurons inthe awake behaving mouse undergoing a median of 97.1% reduc-tions in firing rate for periods of seconds to minutes (Fig. le) (18),and safely expresses for months in both mice and monkeys whenvirally delivered in vivo. Due to the larger currents, Arch enablesvery large (e.g., order of magnitude-scale) increases in addressablevolume of tissue silenceable, over earlier reagents. We thoroughlyinvestigated the safety of Arch function. To date, blebbing issuesthat have affected the usage of halorhodopsins have not beenobserved in vitro or in vivo for Arch, but membrane traffickingsequences may still prove effective at boosting expression beyondthe natural state (the PrI sequence, which greatly magnifies Halocurrent, slightly increases Arch current in neurons). Furthermore,we have not observed changes in cell membrane capacitance orother passive neural properties, as has been reported with halorho-dopsin expression (see above). From a end-user standpoint, illumi-nation of Arch neurons was safe: spike rates measured in vivo werenot significantly different before vs. after periods of optical neuralsilencing. Biophysically, pH changes in neurons expressing Archand undergoing illumination were minimal, plateauing rapidly atalkalinizations of -0.1-0.15 pH units; the fast stabilization of pHi

27

Light-Activated Ion Pumps and Channels for Temporally...

44 BacAIrlwho-

and Wavleangihfti m- fy f

may reflect a self-liniting influence that rapidly limits proton con-centration swings, and may contribute to the safe operation of Archin neurons, as observed in mice and monkeys. Indeed, the changesin pH observed in cells expressing Arch and being illuminated arecomparable in magnitude to those observed during illumination ofChR2-expressing cells (10) (due to the proton currents carried byChR2 (1, 46)) and are also within the magnitudes of changesobserved during normal neural activity (94-97). We have observedthat other archaerhodopsins from other Halorubrum strains arealso particularly powerful molecular reagents (work in progress).In contrast, the canonical H. sainarum bacteriorhodopsin, wellknown to poorly function in Eschericbia coli, successfully producedmodest photocurrents in mammalian neurons, which highlightsthe importance ofnot assuming all molecules will express and trafficthe same in different organisms. (In contrast, E. coli does notsupport any detectable expression of ChR2, which expresses wellin neurons; target species influences in modulation of opsin func-tion should not be underestimated).

Unlike all of the halorhodopsins we have screened to date (includ-ing not only the natural halorhodopsins described above, but alsoproducts of halorhodopsin site-directed mutagenesis aimed atimproving kinetics), which after illumination remained inactivatedfor tens of minutes, Arch spontaneously recovers function in sec-onds in the dark (Fig. 4c, d), more like the light-gated cationchannel channelrhodopsin-2 (ChR2) than like halorhodopsins.This feature is particularly useful for in vivo behavior work becauseit dramatically simplifies the necessary optical hardware; the need touse only one wavelength of light also increases the available band-width for multicolor silencing in multiple cell-types. This sponta-neous recovery has also been observed by us with otherarchaerhodopsins, and thus may be a general feature of archaerho-dopsins as a whole (work in progress).

Arch is maximally excited with green-yellow light (A = 561 nm),a fairly common peak wavelength for proton pumps. Thus, it isbackward compatible with halorhodopsin-driving equipment. Pro-ton pumps naturally exist that are activated by many colors of light,in contrast to chloride pumps, which are primarily driven by yellow-orange light (even with significant mutagenesis of retinal-flankingresidues, (15)). The light-driven proton pump from L. maculans,here abbreviated Mac, has an action spectrum strongly blue-shiftedrelative to that ofthe light-driven chloride pump Halo (Fig. 5a). Wefound that Mac-expressing neurons could undergo 4.1-fold largerhyperpolarizations with blue light than with red light, and Halo-expressing neurons could undergo 3.3-fold larger hyperpolariza-tions with red light than with blue light, when illuminated withappropriate powers and filters (Fig. 5b). Accordingly, we coulddemonstrate selective silencing of spike firing in Mac-expressing

28

.. .......... ........

323


a

Mac "r band Halo exr ban!d

JIM 450 SWo SWo 6 00 66 0 Mwavelength (nm)

b

E

Halo Mac

20-

630 470 630 47053 2.1 5.3 It

Light wavelength (nm)

CLgX Onaenc (mWfrm)

Halo-expressing neuron

Mac-expressing neuron

50 mV

15 in-W mV - -65 m

Fig. 5. Multicolor silencing of two neural populations, enabled by blue- and red-lght drivable ion pumps of different classes.(a) Action spectra of Mac vs. Halo; rectangos indicate filter bandwidths used for multicolor silencing in vitro. Blue lightpower is via a 470 20 nm flter at 5.3 mW/mm 2, and red light power is via a 63W 15 nm fter at 2.1 mW/mm 2. (b)Membrane hyperpolaizations elicited by bluevs. red light, in cells expressing Halo or Mac (N = 5 Mac-expressing neurons,N= 6 Halo-expressing neurons). (c) Action potentials evoked by current injection into patch clamped cultured neuronstransfected with Halo (c) were selectively silenced by the red light but not by the blue light and vice versa in neuronsexpressing Mac (cil). &ay basm in the fint(Cl) indicate periods of patch clamp current injection.

neurons in response to blue light, and selective silencing of spikefiring in Halo-expressing neurons in response to red light (Fig. 5c).Thus, the spectral diversity of proton pumps points the way towardindependent multicolor silencing of separate neural populations.

29

Light-Activated Ion Pumps and Channels for Temporally...

10 ms

0640 BR570

from 5 ns + H+ hvcytoplasm from EC

300 ms hvN560 1 us K590

2 msk M1 412

M2412 L550(cytoplasm) it

IA+ 11*- M,412- * (extracellular)to extracellular

environment (EQ hv

M412 (alternate?)lifetime > 2.5 minutes

Fig. 6. The pholocycle of the H. Slkwzu bactarihodopsan. As in Figs. 2c and 3b, thephotocycle has been simplified to reflect the dominant photocycle at large continuousilluminatlon on the timescale of a typical photocycle (.., conditions used for neuralsilencing, >few milliseconds). The M412 alternate hntermediate is the origin of long-livedinactivation with bacterilorhodopsn. In contrast, Arch spontaneously quscidy recoversfrom this state in the daiL

This result opens up novel kinds of experiment, in which, for exam-pie, two neuron classes, or two sets of neural projections from asingle site, can be independently silenced during a behavioral task.

Figure 6 shows the photocycle of the canonical H. salinarumbacteriorhodopsin, as representative of the photocycle of the classof proton pumps (the photocycles of archaerhodopsins are not aswell characterized, and may well be different) (22, 93). The datathat has led to the synthesis of the modern model of the bacterio-rhodopsin photocycle provide many of the insights that have led tosolid investigations into this field as a whole. As in Figs. 2c and 3b,it has been simplified to represent the dominant photocycdeexpected at large and continuous illumination on the timescale ofa typical photocycle (e.g., many milliseconds, as would be used forneural silencing). Upon absorbing a photon, bacteriorhodopsinforms its L550 intermediate (L = lumi) within microseconds,after quickly transferring a proton from the retinylidene Schiffbase (at the Lys-216 position) to the Asp-85 proton acceptor.This transfer triggers the proton releasing group (PRG), containingGlu-204 and Glu-194, to release its own proton (98, 99), duringthe L -+ M transition. After a "switching" step during which theSchiff base reorients itself on the cytoplasmic side, it is then

30

325

326 BY. Chow et at

44 Hatkwlide-PL40 and Bacbrto-dtaw MdWWUawv Vaanta

4A ohrWy

reprotonated during the M -+ N transition via the Asp-96 donorresidue, which in turn picks up a proton from the cytoplasm in thenext transition (N -+ 0). The chromophore reisomerizes to the all-transstate during this transition as well. Finally, the bacteriorhodop-sin relaxes back to its ground state as the proton release group isreloaded from the Asp-85 residue. Like halorhodopsin, bacterio-rhodopsin can become trapped in a long-lasting light-unresponsivestate (Fig. 6, "M412 alternate") that requires blue light to reenterthe normal photocycde; this could partly explain the lower currentsobserved with BR when compared to Arch.

For decades, researchers have been making mutants of bacteriorho-dopsins and halorhodopsins for structure-function studies reviewedin many earlier publications (e.g., (22, 24, 93)). However, pointmutations that improve these molecules as neural silencing toolshave yet to be reported. Given that the conductance of a pump islimited by the fact that they move only one ion per photocyce, itwould be highly desirable to mutate halorhodopsin into a channel,or channeirhodopsin into an anion channel. Tuning the actionspectrum of both bacteriorhodopsins and halorhodopsins may beachieved by mutating the retinal-flanking residues (75, 76, 100,101). Mutations that alter ion selectivity, such as the most well-known example of this is the bacteriorhodopsin D85T mutationthat converts it into a chlorideyump (102), could allow ion-specificcurrents to be mimicked. A Ca selective pump, for example, wouldhave powerful impact on enabling powerful studies of plasticity,synaptic transmission, and cellular signaling. Given that crystalstructures fur many bacteriorhodopsins and halorhodopsins exist,we anticipate that functionally oriented and applied site-directedmutagenesis will be a highly active field of research in neuroengi-neering. The fact that archaerhodopsins on the whole represent aclass of molecules that express particularly well in mammalian cellspossibly indicates properties of their lipid-interacting residues, orperhaps "signal sequence"-like activity by their loop regions. Theknown crystal structures of these molecules (80, 81) may providekey insights into the trafficking of heterologously expressed mole-cules and their membrane insertion.

light-activated proton and chloride pumps are known to exist in farmore organisms than do light-gated cation channels, and protonpumps are particularly prevalent, as all opsins described to datelikely have at least some proton pumping capability (22, 23,67-79, 103, 104). Even though many proton pumps maximallyabsorb blue-green to green wavelengths (which opens up, as shownin Fig. 5, the possibility, alongside yellow-light driven chloridepumps, for multiple-color silencing of distinct neural populations),light-activated hyperpolarizing currents carried by protons havebeen observed across the whole visible spectrum, from deep blue

31


(via a sensitizer) (105) to far red (106) (>650 nm), although thered-light sensitive current likely originates from a receptor thattriggers an H'-ATPase, as opposed to direct light-mediated iontranslocation.

The discovery or creation of a purely genetically encoded light-activated inhibitory channel would be highly desirable. In additionto seeking natural molecules and site-directed mutants, linking anon-light-gated ion-channel to a type I archaeal rhodopsin may beanother promising approach to creating such a molecular tool(107). In this way, a light-activated shunt could be created, whichwould more closely mimic natural mechanisms of neural inhibitionin the brain.

5. MolecularTargetingof Mcroblal Opsinsto Different CalTypes

The number of papers using optical neural control in species rang-ing from C. degam to mouse to nonhuman primate is increasingexponentially each year, and so we do not attempt to review theliterature comprehensively. In each species, opsins have been usedto test the necessity and sufficiency of neurons, cell types, muscles,neural pathways, brain regions, and other entities in behaviors,pathologies, and neural computations. Opsins have proven valuablein exploring neural dynamics in multiple mammalian brain struc-tures as well. We focus on highlighting principles that govern howto best use these opsins in various settings, from a molecularbiology standpoint (Sect. 5) and from a physical-optical standpoint(Sect. 6). From a molecular biology standpoint, these opsins can bedelivered to neurons in almost any conventional way that genes aredelivered into cells or into organisms. Transgenic mice have beenmade with ChR2, for example (108), mice and monkeys have beeninjected with lentiviruses, adeno-associated viruses (AAV), andother viruses encoding for light-gated proteins (2, 52, 109-111),and rodents and chicks have been electroporated in utero withplasmids encoding for light-gated proteins (4, 53, 112). In eachcase, difffrent parameters of the technique can be selected so as toenable specific cell types, pathways, or regions to be selectivelylabeled or to selectively express the opsin. Transgenic C. clegamhave been made expressing ChR2 using conventional methods (5),as have transgenic Drosophila (113) and zebrafish (114). For theselatter species, supplementation with all-tram-retinal may be bene-ficial, whereas mammalian brains seem to operate microbial opsinswithout need for supplementation. (It is possible that in the future,genetically engineering retinal-lacking organisms, such as inverte-brates, to produce retinal within their nervous systems may be ofuse, e.g., by expressing within them enzymes that can produceretinal from vitamin precursors (115)).

32

................. ....... .... .................. . .


For mammalian nervous systems, a large variety of possiblestrategies exist for conveying opsin genes into specific cell types.For example, transgenic mice can be made through BAC trans-genic, knock-in, or other methods, but such strategies are notcommon for other species yet. For viral delivery, cell type specificpromoters can be inserted upstream of the opsin to target variousexcitatory, inhibitory, and modulatory neurons (e.g., (19,116-119)); the size of the promoter is limited by the virus type(AAV viruses hold typically 4-4.5 kb total, whereas lentiviruseshold typically 8-10 kb maximum). The surface or coat proteinsthat a virus bears can also modulate which cell types will take up thevirus; for example, lentiviruses may favor excitatory neurons of thecortex, whereas certain AAV serotypes may favor inhibitory cells(120). Lentiviruses can be pseudotyped-fabricated with a coatprotein of desired targeting capacity, e.g., with rabies glycoproteinthat leads to lentiviruses that travel retrogradely (121)-whereasAAVs can bc engineered with biotinylation sites that enable, uponstreptavidin conjugation, targeting of potentially arbitrary sub-strates (122). Retroviruses, which preferentially label dividingcells, have been used to deliver ChR2 to newborn neurons of thedentate gyrus of the hippocampus (123). Other viruses such asrabies virus and pseudorabies virus can be used, with unique tracingcapabilities including the ability to go retrogradely across multiplesynapses (124, 125). Viral particles typically have to be injecteddirectly into the brain, often through stereotactic targeting to aspecific brain area, since the potent blood-brain barrier typicallyprecludes systemic delivery of large viral particles (although see(126)). We have recently described a parallel injector array thatcan deliver viruses into complex three-dimensional configurations(127). Finally, in utero electroporation of ChR2-GFP into embry-onic mice at embryonic day -15.5 has been found to selectivelylabel pyramidal cells in layers 2/3 (53, 112).

Of potential interest is the possibility of a new generation ofneural prosthetics, which can accomplish the synthetic neurobiol-ogy mission of repairing the nervous system by enabling opticalinput of information to sculpt neural dynamics and overcomepathology. To deliver these genes in a safe, efficacious, andenduring way, viruses such as AAV may be valuable; AAV hasbeen used in over 600 human patients in gene therapy clinical trialswithout a single serious adverse event (128) and has been success-fully used with opsin delivery. The ability to optically control spe-cific targets within the brain may enable more potent and side-effect free therapies than possible with existing electrical and mag-netic neuromodulation therapies, or with drugs which often arenonspecific and have side effects. Already several groups have pro-totyped blindness therapies that may enable new approaches to a

33

Light-Activated Ion Pumps and Channels for Temporally.. . 2

a LoxP LoxP

LoxP pn

Lox2272 1(2)

LoxP OpW '

Ag. 7. Transgenic mouse expressing Cre within specific calls, coupled to lox-containing MV viruses, enable cell-typespecific opsin expression. (a) A pair of loxP recombination sequences meditate the removal of the transciptrional andiranslational stop cassette containing multiple polyadenylalon signals (STOP), in the presence of Gre (provided in

transgenic mice within specific cell types), to initiate gene expression. (b) Two pairs of loxP-type recombinaion sequences(FLEX) for stable inversion proceedB in two slaps: (1) inversion followed by (2) eacision. loxP and lox2272 are orthogonalrecombination sites. (Adapted from Kuhiman and Huang (111) and Atasoy et al. (110)).

currently intractable set of disorders, those in which photoreceptorsdegenerate within the retina (129, 130). To that end, the recentassessment of brain and immune function in nonhuman primatesafter ChR2 expression and activation, which showed a lack ofharmful efficts in a preliminary study (109), may pave the waytoward new ideas fur neural prosthetics for humans.

One strategy that has become widely popular, is to inject thebrain of a mouse expressing Cre recombinase in a specific cell type,with a virus that either bears an opsin preceded by a lox-flanked stopcassette, or a virus that bears an opsin reversed and flanked by pairsoflox sites in a specific configuration (Fig. 7; (110, 111)). Given thevery large number of Cre transgenic mice in existence, and that arebeing generated, this strategy is likely to be very useful, at least formice. Transgenic mice can be made that express Cre in extremely cellspecific ways (e.g., through 3' UTRknockins at the ends of cell type-specific genes), and then viruses can be rapidly made as new opsinsare created, thus enabling cell type-specific expression withoutrequiring the difficult process of trimming cell-specific promotersto fit within the small viral payload, or the difficult process of makingtransgenic mice for each new opsin tool that is developed (given therapid pace of innovation, as shown in Sects. 1-4).

34

329

330 B.Y. Chow et ai.

6 Hardwarefor Optical NeuralControl

For in vitro use, xenon lamps (e.g., Sutter DG-4, Till PhotonicsPolychrome) equipped with fast-moving mirrors or monochroma-tors can be used for flexible delivery of fast (millisecond-timescale), brigh light pulses to biological samples on microscopes.Fluorescence filters can be used to deliver light of the appropriatewavelength (e.g., GFP excitation filter for ChR2, rhodamine orTexas Red excitation filter for Arch, Texas Red excitation filter forHalo, and GFP or YFP excitation filter for Mac). Recently, manycompanies such as Thorlabs have begun to sell LEDs or LED arrayscompatible with microscope fluorescence illuminators, which sellfor a small fraction ofthe price of a full lamp setup. Or fiber-coupledLEDs can simply be placed nearby to the sample (131). Confocaland two-photon microscopes, or more generally scanning lasermethodologies, can be used to drive opsins, as have been describedin a variety of papers (112, 132, 133). Recently, digital micromirrordisplays (DMDs) have come forth as potentially useful for photo-stimulating in complex patterns, comprising millions of individualpixels that can be toggled either on or off (134, 135).

In vivo, the brain scatters light starting within a few hundredmicrons of an optical source, and absorbs light starting within a fewmillimeters of distance. Thus, most efforts for in vivo neuromodu-lation focus on delivering light to a volume of tissue, ranging from avery small volume containing a few hundred cells to a large volume(say, a few cubic millimeters) containing many thousands of cells.One widespread method is to use a laser coupled to an optical fiber(Fig. 8a shows a versatile setup that couples multiple colors of laserlight into a single fiber; simpler commercially available single-colorlaser-coupled fibers with TTL control are also available), and toinsert the fiber into a cannula implanted in the brain (Fig. 8b showsa hand-built one for mice; commercial versions from companiessuch as Plastics One can also be built) or directly into the brain(Fig. 8c shows a setup for monkey), or to couple the fiber to animplanted fiber via a ferrule. An optical commutator (e.g., fromDoric Lenses) can be placed between the fiber that inserts into thebrain and the fiber that connects to the laser, to allow free rotation(66). Arrays of custom-targetable optical fibers, each coupled to aminiaturized light source (e.g., a raw die LED) and targeted to aunique target, will open up the ability to drive activity in complex 3-D patterns, enabling the perturbation of complexly shaped struc-tures as well as the ability to perturb targets in a patterned fashion(136, 137).

Aside from the key advantage of being able to manipulate aspecific cell type, another key advantage of optical stimulation is thelack of electrical artifact as compared to conventional electrical

35

Ught-Activated [on Pumps and Channels for Temporally...

a n*W(nUbb (adus"W)l.

b C3 5- electrode drive

2 _two guide tubes

- - fiber

1 electrode

-4

FIg. 8. (a) Schematic design (left) and picture (rdg, of an Optics assembly used to couple blue and yellow laser light into asingle optical fiber for In vivo neural modulation. A pictured assembly, lacking a neutral density filter, shows the hardwarelaid out on a standard optics breadboard. (b) Schematic design (left) and picture (tikt) of a system for targeting andsecuring optical fibers within the brain. A polyimide cannula (1, 250 pIm ID), designed to terminate at the locus of opticalmodulation, is epoxied to a stack of hex nuts (2, sized 2-56) which will be secured to the skull with dental cement. Ventedscrews (3, sized 2-6), which have holes in their centers, screw into the nuts while leaving a path open to the cannula. Adummy wire (4,230 lAm stainless steel wire) may be epoxied to the screw to seal the Cranlotomy when the optics are not inuse. An optical fiber (5, 230 pim 00 silca fibe) is allowed free rotation without vertical displacement by a plastic washer (6,homemade) which is epoxied to the fiber and sandwiched in between the vented screw above and the cannula below. (c)Apparatus for optical activation and electrical recording. Photograph, showing optical fiber (200 pm diameter) andelectrode (200 pm shank diameter) in guide tubes. Adapted from Han et al. (19).

stimulation methods. However, despite the lack of electrical arti-fact, light does produce a voltage deflection when the electrode tipwas illuminated (19, 138). For example, Fig. 9 shows tracesrecorded on a tungsten electrode in saline being illuminated by apulsed laser beam, as adapted from (19). This voltage deflectionslowly evolved over many tens of milliseconds, and accordingly wasonly recorded when the electrode voltage filtered at 0.7-170 Hz toexamine local field potentials, (Fig. 9, top traces of each panel). Thisvoltage deflection was not recorded when the electrode voltage wasfiltered at 250-8,000 Hz to detect spike signals; (Fig. 9, bottomtraces of each panel). When light illuminated parts of the electrodeother than the tip, no artifact was recorded; only illumination of the

36

331


a In saline b In braini

field potential field potentialchannel - Chanl(0.7-170 Hz) (0.7-170 Hz) 05 mV0.1 MV 100 m8

1 t0m V 00m

spike channelspike channel 5OpV (250-8000 Hz)(250-8000 Hz) S0 m O5m V

100 Ms 01 Ma

lI iifield potential field potentialchannel channel(0.7-170 Hz) (0.7-170 Hlz) O.5 mV

IOf' 4' 1MV 1o0 Mal OOms

spike channel nspike channel AAA

(250-8000 Hz) oo5OPV spik chane Tyy 'II P 51100 M (250-8000 Hz) V I 5OjN0.5 ns

Fig. 9. Voltage deflections observed on tungsten electrodes immersed in saline (a) or brain (b), upon tip exposure to200 me blue light pulses (bI) or trains of 10 ms blue light pulses delivered at 50 Hz (bi). Ught pulses are indicated by bluedesheas Bectrode data was hardware filtered using two data acquisition channels operating in parallel, yielding a low-frequency component ("field potential channel) and a high-frequency component ("spike channel"). For the "spikechannel" traces taken in brain (b), spikes were grouped into 100 me bins, and then the binned spikes were displayedbeneath corresponding parts of the simultaneously acquired *field potential channel" signal (59 and 53 repeated lightexposures for bi and bi respectively). (Shown are the spikes in eight such bins-the two bins before light onset, the twobins during the light delivery period, and the four bins after light cessation.) For all other signals shown, ten overlaid tracesare plotted. From Han et al. (19).

tip-saline interface resulted in the voltage transient. This phenom-enon is consistent with a classical photoelectrochemical finding, theBecquerel effect, in which illumination of an electrode placed insaline can produce a voltage on the electrode (139, 140). Consis-tent with the generality of the Becquerel effect as a property ofelectrode-electrolyte interfaces, this artifact is observed on variouselectrode materials, such as stainless steel, platinum-iridium, silver/silver chloride, gold, nichrome, or copper. Similar slowly evolvingvoltage deflections were observed when tungsten electrodes wereused to record neural activity in the brain, during optical stimula-tion. Because the optical artifact was slowly evolving over many tensof milliseconds, spike waveforms were detected without corruptionby the artifact. However, local field potentials and field oscillations,which reflect coherent neural dynamics in the range of Hertz totens of Hertz, may be difficult to isolate from this Becquerel artifactusing the electrodes here tested. Notably, we have not seen theartifact with pulled glass micropipettes (such as previously used in(3) and (12), or in the mouse recordings with pulled glass pipettesin (109)). Thus, for recordings of local field potentials and otherslow signals of importance for neuroscience, hollow glass electrodesmay prove useful.

37

Ught-Activated [on Pumps and Channels for Temporally... 333

Rfrmme

1. Nagel G et al (2003) Channclrhodopsin-2, adirectly light-gated cation-selective mem-brane channel. Proc NatI Acad Sci USA100:13940-13945

2. Ishizuka T, Kakuda M, Araki R, Yawo H(2006) Kinetic evaluation of photosensitivityin genetically engineered neurons expressinggreen algae light-gated channels. NeurosciRes 54:85-94

3. Boyden ES, Zhang F, Bamberg E, Nagel G,Deisseroth K (2005) Millisecond-timescale,genetically targeted optical control of neuralactivity. Nat Neurosci 8:1263-1268

4. 1i X et al (2005) Fast noninvasive activationand inhibition of neural and network activityby vertebrate rhodopsin and green algae chan-nelrhodopsin. Proc Natl Acad Sci USA102:17816-17821

5. Nagel G et al (2005) ight activation of chan-nelrhodopsin-2 in excitable cells of Caenor-babdisis elegaxs triggers rapid behavioralresponses. Curr Biol 15:2279-2284

6. Wang H et al (2009) Molecular determinantsdifferentiating photocurrent properties oftwo channelrhodopsins from chlamydomo-nas. J Biol Chem 284:5685-5696

7. Zhang F et al (2008) Red-shifted optogeneticexcitation: a tool for fast neural controlderived from Volvox carteri. Nat Neurosci11:631-633

8. Berndt A, Yizhar 0, Gunaydin IA, Hege-mann P, Deisseroth K (2009) Bi-stable neuralstate switches. Nat Neurosci 12:229-234

9. Lewis TL Jr, Mao T, Svoboda K, Arnold DB(2009) Myosin-dependent targeting of trans-membrane proteins to neuronal dendrites.Nat Neurosci 12:568-576

10. Lin JY, Lin MZ, Steinbach P, Tsien RY (2009)Characterization of engineered channelrho-dopsin variants with improved properties andkinetics. Biophys J 96:1803-1814

11. Lanyi JK, Duschl A, Hatfield GW, May K, Oes-terhelt D (1990) The primary structure of ahalorhodopsinnfrom Nammxobacterim pbao-nis. Structural, functional and evolutionaryimplications for bacterial rhodopsins and halor-hodopsins. J Biol Chem 265:1253-1260

12. Han X, Boyden ES (2007) Multiple-coloroptical activation, silencing, and desynchroni-zation of neural activity, with single-spiketemporal resolution. PLoS One 2:c299

13. Zhang F et al (2007) Multimodal fast opticalinterrogation of neural circuitry. Nature446:633-639

14. Bamberg E, Tittor J, Oesterhelt D (1993)Light-driven proton or chloride pumping byhalorhodopsin. Proc NatI Acad Sci USA90:639-643

15. Chow B et al (2010) High-performancegenetically targetable optical neural silencingby light-driven proton pumps. Nature 463(7277):98-102

16. Gradinaru V, Thompson KR, Deisseroth K(2008) eNpHR a Natronomonas halorho-dopsin enhanced for optogenetic applications.Brain Cell Biol 36:129-139

17. Zhao S et al (2008) Improved expression ofhalorhodopsin for light-induced silencing ofneuronal activity. Brain Cell Biol 36:141-154

18. Chow B, Han X, Qian X, Boyden ES (2009)High-performance halorhodopsin variants forimproved genetically-targetable optical neuralsilencing. Frontiers in Systems Neuroscience.Conference Abstract Computational and sys-tems neuroscience. doi:10.3389/conf.neuro.10.2009.03.347

19. Han X, Qian X, Stem P, Chuong AS, BoydenES (2009) Informational lesions: optical per-turbation of spike timing and neural syn-chrony via microbial opsin gene fusions.Front Mol Neurosci. doi:10.3389/neuro.02.012.2009

20. Henderson R, Schertler GF (1990) The struc-ture of bacteriorhodopsin and its relevance tothe visual opsins and other seven-helix G-pro-tein coupled receptors. Philos Trans R. SocLond B Biol Sci 326:379-389

21. Palczewski K (2006) G protein-coupledreceptorrhodopsin. Annu Rev Biochem75:743-767. doi:10.1146/annurev.biochem.75.103004.142743

22. Lanyi JK (2004) Bacteriorhodopsin. AnnuRev Physiol 66:665-688. doi:10.1146/annurevphysiol.66.032102.150049

23. Lanyi JK(1986) Halorhodopsin: a light-drivenchloride ion pump. Annu Rev Biophys BiophysChem 15:11-28. doi:10.1146/annurev.bb.15.060186.000303

24. Essen LO (2002) Halorhodopsin: light-driven ion pumping made simple? Curr OpinStruct Biol 12:516-522

25. Zemelman BV, Lee GA, Ng M, Miesenbock G(2002) Selective photostimulation of geneti-cally chARGed neurons. Neuron 33:15-22

26. Tin B, Koizumi A, Tanaka N, Panda S, Mas-land RH (2008) Restoration of visual func-tion in retinal degeneration mice by cctopic

38

. ........... _ .- .. ... ...... ..... I ......... .... : .... .1__: .... ... ... ............

334 B.Y. Chow et a].

expression of melanopsin. Proc Natl Acad ScUSA 105:16009-16014

27. Kolbe M, Besir H, Essen LO, Ocsterhelt D(2000) Structure of the light-driven chloridepump halorhodopsin at 1.8 A resolution. Sci-ence 288:1390-1396

28. Luecke H, Schobert B, Richter HT, CartaillerJP, Lanyi JK (1999) Structure of bacteriorho-dopsin at 1.55 A resolution. J Mol Biol291:899-911

29. Braiman MS, Stern LJ, Chao BH, KhoranaHG (1987) Structure-function studies onbacteriorhodopsin. IV. Purification and rena-turation of bacterio-opsin polypeptideexpressed in Ercherichia coli. J Biol Chem262:9271-9276

30. Gilles-Gonzalez MA, Engelman DM, Khor-ana HG (1991) Structure-function studies ofbacteriorhodopsin XV. Effects of deletions inloops B-C and E-F on bacteriorhodopsinchromophore and structure. J Biol Chem266:8545-8550

31. Mogi T, Stern LJ, Chao BH, Khorana HG(1989) Structure-function studies on bacteri-orhodopsin. VIII. Substitutions of the mem-brane-embedded prolines 50, 91, and 186:the effects are determined by the substitutingamino acids. J Biol Chem 264:14192-14196

32. Mogi T, Marti T, Khorana HG (1989) Struc-ture-function studies on bacteriorhodopsin.IX. Substitutions of tryptophan residues affectprotein-retinal interactions in bacteriorho-dopsin. J Biol Chem 264:14197-14201

33. Mogi T, Stern LJ, Hackett NR, Khorana HG(1987) Bacteriorhodopsin mutants contain-ing single tyrosine to phenylalanine substitu-tions are all active in proton translocation.Proc NatI Acad Sci USA 84:5595-5599

34. Marti T et al (1991) Bacteriorhodopsinmutants containing single substitutionsof serine or threonine residues are allactive in proton translocation. J Biol Chem266:6919-6927

35. Marinetti T, Subramaniam S, Mogi T,Marti T, Khorana HG (1989) Replacementof aspartic residues 85, 96, 115, or 212affects the quantum yield and kinetics of pro-ton release and uptake by bacteriorhodopsin.Proc Natd Acad Sci USA 86:529-533

36. Mogi T, Stern LJ, Marti T, Chao BH, Khor-ana HG (1988) Aspartic acid substitutionsaffect proton translocation by bacteriorho-dopsin. Proc Natl Acad Sci USA85:4148-4152

37. Subramaniam S, Greenhalgh DA, KhoranaHG (1992) Aspartic acid 85 in bacteriorho-

dopsin functions both as proton acceptor andnegative counterion to the Schiff base. J BiolChem 267:25730-25733

38. Brown US, Needleman R, Lanyi JK (1996)Interaction of proton and chloride transferpathways in recombinant bacteriorhodopsinwith chloride transport activity: implicationsfor the chloride translocation mechanism.Biochemistry 35:16048-16054

39. Hegemann P, Oesterhelt D, Steiner M (1985)The photocycle of the chloride pump halor-hodopsin. I: Azide-catalyzed deprotonationof the chromophore is a side reaction ofphotocycle intermediates inactivating thepump. EMBO J 4:2347-2350

40. Blanck A, Oesterhelt D (1987) The halo-opsin gene. II. Sequence, primary structureof halorhodopsin and comparison with bacte-riorhodopsin. EMBO J 6:265-273

41. Rudiger M, Oesterhelt D (1997) Specificarginine and threonine residues controlanion binding and transport in the light-driven chloride pump halorhodopsin. EMBOJ 16:3813-3821

42. Varo G et al (1995) Light-driven chloride iontransport by halorhodopsin from Natronobac-terium pharaouis. 1. The photochemicalcycle. Biochemistry 34:14490-14499

43. Tittor J et al (1997) Chloride and protontransport in bacteriorhodopsin mutantD85T: different modes of ion translocationin a retinal protein. J Mol Biol 271:405-416

44. Tittor J, Oesterhelt D, Bamberg E (1995)Bacteriorhodopsin mutants D85N, D85Tand D85,96N as proton pumps. BiophysChem 56:153-157

45. Varo G, Needleman R, Lanyi JK (1995)Light-driven chloride ion transport by halor-hodopsin from Natronobacterium pharaois.2. Chloride release and uptake, protein con-formation change, and thermodynamics. Bio-chemistry 34:14500-14507

46. Berthold P et al (2008) Channerhodopsin-1initiates phototaxis and photophobicresponses in Chlamydomonas by immediatelight-induced depolarization. Plant Cell20:1665-1677. doi:10.1105/tpc.108.057919

47. Sineshchekov OA, Govorunova EG, SpudichJL (2009) Photosensory functions of chan-nelrhodopsins in native algal cells. PhotochemPhotobiol 85:556-563

48. Sineshchekov OA, Jung KH, Spudich JL(2002) Two rhodopsins mediate phototaxisto low- and high-intensity light in Chlamydo-moas reinuardtii. Proc Natl Acad Sci USA99:8689-8694

39

Ught-Activated Ion Pumps and Channels for Temporally...

49. Feldbauer K et al (2009) Channelrhodopsin-2 is a leaky proton pump. Proc NatI Acad SciUSA 106:12317-12322

50. Nagel G et al (2002) Channelrhodopsin-1: alight-gated proton channel in green algae.Science 296:2395-2398

51. Ernst OP et al(2008) Photoactivation ofchan-neirhodopsin. J Biol Chem 283:1637-1643

52. Zhang F, Wang LP, Boyden ES, Deisseroth K(2006) Channelrhodopsin-2 and optical con-trol of excitable cells. Nat Methods3:785-792

53. Huber D et al (2008) Sparse optical microsti-mulation in barrel cortex drives learned beha-viour in freely moving mice. Nature451:61-64

54. Bamann C, Kirsch T, Nagel G, Bamberg E(2008) Spectral characteristics of the photo-cycle of channelrhodopsin-2 and its implica-tion for channel function. J Mol Biol375:686-694

55. Nikolic K, Dcgenaar P, Toumazou C (2006)Modeling and engineering aspects of chan-nelrhodopsin2 system for neural photostimu-lation. Conf Proc IEEE Eng Med Biol Soc1:1626-1629

56. Tsunoda SP, Hegemann P (2009) Glu 87 ofchannelrhodopsin-1 causes pH-dependentcolor tuning and fast photocurrent inactiva-tion. Photochem Photobiol 85:564-569

57. Ritter E, Stelifest K, Berndt A, Hegemann P,Bartl FJ (2008) Monitoring light-inducedstructural changes of channelrhodopsin-2 byTV-visible and Fourier transform infrared spe-

ctroscopy. J Biol Chem 283:35033-35041.doi:10.1074/jbc.M806353200

58. Gunaydin IA et al (2010) Ultrafast optoge-netic control. Nat Neurosci 13(3):387-92

59. Harwood J, Guschina IA (2009) The versa-tility of algae and their lipid metabolism.Biochimie 91:679-684

60. Zipfel WR, Williams RM, Webb WW (2003)Nonlinear magic: multiphoton microscopy inthe biosciences. Nat Biotechnol21:1369-1377

61. Mohanty SK et al (2008) In-depth activationof channeirhodopsin 2-sensitized excitablecells with high spatial resolution using two-photon excitation with a near-infrared lasermicrobeam. Biophys J 95:3916-3926

62. Rickgauer JP, Tank DW (2009) Two-photonexcitation of channelrhodopsin-2 at saturation.Proc NatI Acad Sci USA 106:15025-15030

63. Sincshchekov OA et al (2005) Rhodopsin-mediated photoreception in cryptophyte fla-gellates. Biophys J 89:4310-4319

64. Sineshchekov OA, Litvin FF, Keszthelyi L(1990) Two components of photoreceptorpotential in phototaxis of the flagellatedgreen alga Haematococcu pluvialis. BiophysJ 57:33-39

65. Utvin FF, Sineshchekov OA, SineshchekovVA (1978) Photoreceptor electric potentialin the phototaxis of the alga Hamaicoccuspluviai.w Nature 271:476-478

66. Gradinaru Vet al (2007) Targeting and read-out strategies for fast optical neural controlin vitro and in vivo. J Neurosci27:14231-14238

67. Ihara K et al (1999) Evolution of the archaealrhodopsins: evolution rate changes by geneduplication and functional diffirentiation.J Mol Biol 285:163-174

68. Mukohata Y, Ihara K, Tamura T, Sugiyama Y(1999) Halobacterial rhodopsins. J Biochem125:649-657

69. Klare JP, Chizhov I, Engelhard M (2008)Microbial rhodopsins: scaffolds for ionpumps, channels, and sensors. Results ProblCell Differ 45:73-122

70. Ant6n J et al (2005) Salinibacter ruber: geno-mics and biogeography. In: Gunde-Cimer-man N, Plcmenitas A, Oren A (eds)Adaptation to life in high salt concentrationsin archaea, bacteria and eukarya. Kluwer Aca-demic Publishers, Dordrecht, Netherlands,pp 257-266

71. Balashov SP et al (2005) Xanthorhodopsin: aproton pump with a light-harvesting caroten-oid antenna. Science 309:2061-2064

72. Beja 0, Spudich EN, Spudich JL, Leclerc M,DeLong EF (2001) Proteorhodopsin photo-trophy in the ocean. Nature 411:786-789

73. Beja 0 et al (2000) Bacterial rhodopsin: evi-dence for a new type of phototrophy in thesea. Science 289:1902-1906

74. Friedrich T et al (2002) Proteorhodopsin is alight-driven proton pump with variable vec-toriality. J Mol Biol 321:821-838

75. Kelemen BR, Du M, Jensen RB (2003) Pro-teorhodopsin in living color diversity of spec-tral properties within living bacterial cells.Biochim Biophys Acta 1618:25-32

76. Kim SY, Waschuk SA, Brown LS, Jung KH(2008) Screening and characterizationof protcorhodopsin color-tuning mutations

40

.... ........ .......... . ........ . ...........

335


in Ercherichia coli with endogenousretinal synthesis. Biochim Biophys Acta1777:504-513

77. Brown LS (2004) Fungal rhodopsins andopsin-related proteins: eukaryotic homolo-gucs of bacteriorhodopsin with unknownfunctions. Photochem Photobiol Sci3:555-565

78. Waschuk SA, Bczerra AG, Shi L, Brown LS(2005) Leptosphacria rhodopsin: bacteriorho-dopsin-like proton pump from a eukaryote.Proc Nat Acad Sci USA 102:6879-6883.doi:10. 1073/pnas.0409659102

79. Tsunoda SP et al (2006) H pumping rho-dopsin from the marine alga Acetabulaia.Biophys J 91:1471-1479

80. Yoshimura K, Kouyama T (2008) Structuralrole of bacterioruberin in the trimeric struc-ture of archaerhodopsin-2. J Mol Biol375:1267-1281

81. Enami N ct al (2006) Crystal structures ofarchacrhodopsin- I and -2: common struc-tural motif in archacal light-driven protonpumps. J Mol Biol 358:675-685

82. Seki A et al (2007) Heterologous expressionof Pharaonis halorhodopsin in Xenopus laevisoocytes and electrophysiological characteriza-tion of its light-driven Cl- pump activity. Bio-phys J 92:2559-2569

83. Okuno D, Asaumi M, Muneyuki E (1999)Chloride concentration dependency of theelectrogenic activity of halorhodopsin. Bio-chemistry 38:5422-5429

84. Muncyuki E, Shibazaki C, Wada Y, YakushizinM, Ohtani H (2002) Cl(-) concentrationdependence of photovoltage generation byhalorhodopsin from Halobacterium ali-narum. Biophys J 83:1749-1759

85. Gradinaru Vet al (2010) Molecular and cellu-lar approaches for diversifying and extendingoptogenetics. Cell 141(1):154-65

86. Baliga NS et al (2004) Genome sequence ofHaloarcula marismortusi: a halophilic archaeonfrom the Dead sea. Gcnome Res14:2221-2234

87. Tonnesen J, Sorensen AT, Deisseroth K,Lundberg C, Kokaia M (2009) Optogeneticcontrol of epileptiform activity. Proc NatlAcad Sci USA 106:12162-12167

88. Bernstein JG et al (2008) Prosthetic systemsfor therapeutic optical activation and silencingof genetically-targeted neurons. Proc SocPhoto Opt Instrum Eng 6854:68540H

89. Ludmann K, Ibron G, Lanyi JK, Varo G(2000) Charge motions during the photo-cycle of pharaonis halorhodopsin. Biophys J78:959-966

90. Chizhov 1, Engelhard M (2001) Tempera-ture and halide dependence of the photo-cycle of halorhodopsin fromNatronobacurium pharaonis. Biophys J81:1600-1612

91. Ming M et al (2006) pH dependence of light-driven proton pumping by an archaerhodop-sin from Tibet: comparison with bacteriorho-dopsin. Biophys J 90:3322-3332

92. Lukashev EP et al (1994) pH dependence ofthe absorption spectra and photochemicaltransformations of the archaerhodopsins.Photochem Photobiol 60:69-75

93. Lanyi JK (2006) Proton transfers in the bac-teriorhodopsin photocycle. Biochimica etBiophysica Acta (BBA) - Bioenergetics1757:1012-1018

94. Bevensee MO, Cummins TR, Haddad GG,Boron WF, Boyarsky G (1996) pH regulationin single CAI neurons acutely isolated fromthe hippocampi of immature and mature rats.J Physiol 494(Pt 2):315-328

95. Chesler M (2003) Regulation and modula-tion of pH in the brain. Physiol Rev83:1183-1221

96. Meyer TM, Munsch T, Pape HC (2000)Activity-related changes in intracellular pHin rat thalamic relay neurons. Neuroreport11:33-37

97. Trapp S, Luckermann M, Brooks PA, BallanyiK (1996) Acidosis of rat dorsal vagal neuronsin situ during spontaneous and evoked activ-ity. J Physiol 496(Pt 3):695-710

98. Brown LS et al (1995) Glutamic acid 204 isthe terminal proton release group at the extra-cellular surface of bacteriorhodopsin. J BiolChem 270:27122-27126

99. Phatak P, Ghosh N, Yu H, Cui Q, Elstner M(2008) Amino acids with an intermolecularproton bond as proton storage site in bacteri-orhodopsin. Proc Natil Acad Sci USA105:19672-19677

100. Henderson R et al (1990) Model for thestructure of bacteriorhodopsin based onhigh-resolution electron cryo-microscopy.J Mol Biol 213:899-929

101. Man-Aharonovich D et al (2004) Characteri-zation of RS29, a blue-green proteorhodop-sin variant from the Red Sea. PhotochemPhotobiol Sci 3:459-462

102. Sasaki J et al (1995) Conversion of bacterio-rhodopsin into a chloride ion pump. Science269:73-75

103. Iwamoto M et al (2004) Proton release anduptake of pharaonis phoborhodopsin (sen-sory rhodopsin II) reconstituted into phos-pholipids. Biochemistry 43:3195-3203

41


104. Sudo Y, Iwamoto M, Shimono K, Sumi M,Kamo N (2001) Photo-induced proton trans-port of pharaonis phoborhodopsin (sensoryrhodopsin II) is ceased by association withthe transducer. Biophys J 80:916-922

105. Boichenko VA, Wang JM, Ant6n J, Lanyi JK,Balashov SP (2006) Functions of carotenoidsin xanthorhodopsin and archaerhodopsin,from action spectra of photoinhibition ofcell respiration. Biochimica et BiophysicaActa (BBA) - Biocnergetics1757:1649-1656

106. Serrano EE, Zeiger E, Hagiwara S (1988)Red light stimulates an electrogenic protonpump in Vicia guard cell protoplasts. ProcNatl Acad Sci USA 85:436-440

107. Moreau CJ, Dupuis JP, Revilloud J, Arumu-gam K, Vivaudou M (2008) Coupling ionchannels to receptors for biomolecule sens-ing. Nat Nanotechnol 3:620-625

108. Wang H et al (2007) High-speed mapping ofsynaptic connectivity using photostimulationin channelrhodopsin-2 transgenic mice. ProcNati Acad Sci USA 104:8143-8148

109. Han X et al (2009) Millisecond-timescaleoptical control of neural dynamics in the non-human primate brain. Neuron 62:191-198

110. Atasoy D, Aponte Y, Su HH, Sternson SM(2008) A FLEX switch targets channelrho-dopsin-2 to multiple cell types for imagingand long-range circuit mapping. J Neurosci28:7025-7030

111. Kuhlman SJ, Huang ZJ (2008) High-resolu-tion labeling and functional manipulation ofspecific neuron types in mouse brain by Cre-activated viral gene expression. PLoS One 3:e2005

112. Petreanu L, Huber D, Sobczyk A, Svoboda K(2007) Channelrhodopsin-2-assisted circuitmapping of long-range callosal projections.Nat Neurosci 10:663-668

113. Schroll C et al (2006) Light-induced activa-tion of distinct modulatory neurons triggersappetitive or aversive learning in Drosophilalarvae. Curr Biol 16:1741-1747

114. Douglass AD, Kraves S, Deisseroth K, SchierAF, Engert F (2008) Escape behavior elicitedby single, channelrhodopsin-2-evoked spikesin zebrafish somatosensory neurons. CurrBiol 18:1133-1137

115. Yan Wet al (2001) Cloning and characteriza-tion of a human beta, beta-carotene-15,15'-dioxygenase that is highly expressed in theretinal pigment epithelium. Genomics72:193-202

116. Dittgen T et al (2004) Lentivirus-basedgenetic manipulations of cortical neurons

and their optical and electrophysiologicalmonitoring in vivo. Proc Natl Acad Sci USA101:18206-18211

117. Chhatwal JP, Hammack SE, Jasnow AM,Rainnic DG, Ressler KJ (2007) Identificationof cell-type-specific promoters within thebrain using lentiviral vectors. Gene Ther14:575-583

118. Adamantidis AR, Zhang F, Aravanis AM,Deisseroth K, de Lecea L (2007) Neural sub-strates of awakening probed with optogeneticcontrol of hypocretin neurons. Nature450:420-424

119. Tan W et al (2008) Silencing preBotzingercomplex somatostatin-expressing neuronsinduces persistent apnea in awake rat. NatNeurosci 11:538-540

120. Nathanson JL, Yanagawa Y, Obata K, Call-away EM (2009) Preferential labeling ofinhibitory and excitatory cortical neurons byendogenous tropism of adeno-associatedvirus and lentivirus vectors. Neuroscience161:441-450

121. Wickersham IR, Finke S, Conzelmann KK,Callaway EM (2007) Retrograde neuronaltracing with a deletion-mutant rabies virus.Nat Methods 4:47-49

122. Stachler MD, Chen I, Ting AY, Bartlett JS(2008) Site-specific modification ofAAVvec-tor particles with biophysical probes and tar-geting ligands using biotin ligase. Mol Ther16:1467-1473

123. Toni N et al (2008) Neurons born in the adultdentate gyrus form functional synapses withtarget cells. Nat Neurosci 11:901-907

124. Wickersham IR et al (2007) Monosynapticrestriction of transsynaptic tracing from sin-gle, genetically targeted neurons. Neuron53:639-647

125. Banfield BW, Kaufman JD, Randall JA, Pick-ard GE (2003) Development of pseudorabiesvirus strains expressing red fluorescent pro-teins: new tools for multisynaptic labelingapplications. J Virol 77:10106-10112

126. Foust KD et al (2009) Intravascular AAV9preferentially targets neonatal neurons andadult astrocytes. Nat Biotechnol 27:59-65

127. Chan SC, Bernstein JG, Boyden ES (2010)Scalable fluidic injector arrays for viral target-ing of intact 3-D brain circuits. J Vis Exp35:1489

128. (2007) Retracing events. Nat Biotechnol25:949

129. Bi A et al (2006) Ectopic expression of amicrobial-type rhodopsin restores visualresponses in mice with photoreceptor degen-eration. Neuron 50:23-33

42


130. Lagali PS et al (2008) Light-activated chan-nels targeted to ON bipolar cells restore visualfunction in retinal degeneration. Nat Neu-rosci 11:667-675

131. Campagnola L, Wang H, Zylka MJ (2008)Fiber-coupled light-emitting diode for loca-lized photostimulation of neurons expressingchannclrhodopsin-2. J Neurosci Methods169:27-33

132. Rickgauer JP, Tank DW. In: Neuroscience2008. Society for Neuroscience

133. Petreanu L, Mao T, Sternson SM, Svoboda K(2009) The subcellular organization of neo-cortical excitatory connections. Nature457:1142-1145

134. Farah N, Reutsky I, Shoham S (2007) Pat-terned optical activation of retinal ganglioncells. Conf Proc IEEE Eng Med Biol Soc2007:6369-6371

135. Guo ZV, Hart AC, Ramanathan S (2009)Optical interrogation of neural circuits inCAenorhabditis elegans. Nat Methods6:891-896

136. Bernstein J et al (2008) A scalable toolbox forsystematic, cell-specific optical control of

entire 3-D neural circuits in the intact mam-malian brain. Society fur Neuroscience

137. Bernstein JG et al (2009) Modulation of fearbehavior via optical fiber arrays targeted tobilateral prefrontal cortex. Society for Neuro-science

138. Ayling OG, Harrison TC, Boyd JD, Gorosh-kov A, Murphy TH (2009) Automated light-based mapping of motor cortex by photoacti-vation of channelrhodopsin-2 transgenicmice. Nat Methods 6:219-224

139. Honda K (2004) Dawn of the evolution ofphotoelectrochemistry J Photochem Photo-biol A Chem 166:63-68

140. Gratzel M (2001) Photoelectrochemical cells.Nature 414:338-344

141. Chow YC, Han X, Bernstein JG, MonahanPE, Boyden ES (2011) Light-activated ionpumps and channels for temporally preciseoptical control of activity in genetically tar-geted neurons. In: Chambers JJ, Kramer RH(eds) Photosensitive molecules forcontrolling biological function. Neuro-methods, vol 55. Springer, New York, doi:10.1007/978-1-61779-031-7

43

J. Neural Eng.8 (2011)046021 (10pp) doi: 10. 108811741-2560/8/4/046021

A wirelessly powered and controlleddevice for optical neural control offreely-behaving animalsChristian T Wentz' 23, Jacob G Bernstein23 , Patrick Monahan 2,3,Alexander Guerra2,3, Alex Rodriguez23 and Edward S Boyden2,3,4, 6

' Department of Electrical Engineering and Computer Science, Massachusetts Institute of Technology,77 Massachusetts Ave, Cambridge, MA 02139, USA2 McGovern Institute for Brain Research, Massachusetts Institute of Technology, 77 Massachusetts Ave,Cambridge, MA 02139, USA' Media Lab, Massachusetts Institute of Technology, 77 Massachusetts Ave, Cambridge, MA 02139.USA4 Department of Brain and Cognitive Sciences, Massachusetts Institute of Technology,77 Massachusetts Ave, Cambridge, MA 02139, USA5 Department of Biological Engineering, Massachusetts Institute of Technology, 77 Massachusetts Ave,Cambridge, MA 02139, USAE-mail: [email protected]

Received 28 January 2011Accepted for publication 19 May 2011Published 23 June 2011Online at stacks.iop.orgJNE/8A046021

AbtractOptogenetics, the ability to use light to activate and silence specific neuron types within neuralnetworks in vivo and in vimr, is revolutionizing neuroscientists' capacity to understand howdefined neural circuit elements contribute to normal and pathological brain functions.Typically, awake behaving experiments are conducted by inserting an optical fiber into thebrain, tethered to a remote laser, or by utilizing an implanted light-emitting diode (LED),tethered to a remote power source. A fully wireless system would enable chronic orlongitudinal experiments where long duration tethering is impractical, and would also supporthigh-throughput experimentation. However, the high power requirements of light sources(LEDs, lasers), especially in the context of the extended illumination periods often desired inexperiments, precludes battery-powered approaches from being widely applicable. We havedeveloped a headborne device weighing 2 g capable of wirelessly receiving power using aresonant RF power link and storing the energy in an adaptive supercapacitor circuit, which canalgorithmically control one or more headborne LEDs via a microcontroller. 'e device candeliver approximately 2 W of power to the LEDs in steady state, and 4.3 W in bursts. We alsopresent an optional radio transceiver module (1 g) which, when added to the base headbornedevice, enables real-time updating of light delivery protocols; dozens of devices can becontrolled simultaneously from one computer. We demonstrate use of the technology towirelessly drive cortical control of movement in mice. These devices may serve as prototypesfor clinical ultra-precise neural prosthetics that use light as the modality of biologicalcontrol.

[S] Online supplementary data available from stacks.iop.org/JNE/8A)46021/mmedia

6 Author to whom any correspondence should be addressed.

1741-2560/11/046021+10$33.00 1 0 2011 1oP Publishing Ltd Printed in the UK

44

.... .......

10P PUBUSHING JoURLNAL oF NEURAL ENGiNERNG

J. Neural Ens. 8(2011) 046021 CT Wcntz ci at

1. Introduction

Technologies for using light to activate and silence specific celltypes in the brain of the behaving animal are revolutionizingneuroscientists' capacity to understand how defined neuralcircuit elements contribute to normal and pathological brainfunctions. In such experiments, viral or transgenic approachesare used to deliver genes that encode light-gated ion transportproteins, such as ChR2, Halo/NpHR, Arch, Mac and others,to specific cells, pathways, or regions in the brain [1-5].In order to deliver light to the neural circuit in vivo, typicallyexperimenters insert an optical fiber into the brain, tetheredto a remote laser [6, 7], or implant a light-emitting diode(LED) over the top of the brain, tethered to a remotepower source [8]. However, tethered experimental setupspresent several problems for the experimenter. Animals mustbe handled at the beginning of each experiment, which canalter behavior. Tethering may be impractical for optogeneticexperiments over long periods of time such as desired in manydevelopmental, longitudinal, disease-progression, or otherchronic-perturbation experiments, where fibers and cablesrisk breakage or binding over time. Finally, the need fortethering sets limits on the number of animals that can bemanipulated in a single experiment (e.g. a social experimentwith multiple mice could result in tangling or breakageof the tethers) or in parallel experiments (e.g. in high-throughput screening of large sets of animals). Ideally, forsuch experiments a fully wireless system would be available,enabling chronic and longitudinalexperiments, and supportingexperimentation with large numbers of animals. Wirelesselectrical stimulation and optical modulation approaches havealready been demonstrated utilizing batteries for energystorage [9, 10]. While certainly useful in many applications,battery-based systems suffer from a limited operational timebetween charges and, particularly with the high currentrequirements of LED light sources, may be prohibitively largefor rodent experiments when many LEDs are required, or whenhigh current LEDs are utilized (e.g. for targeting of deep brainstructures). The ability to update modulation protocols in realtime to suit behavior-dependent paradigms or other complexexperimental paradigms would also be of great use.

In order to satisfy the power and control requirements forfreely-behaving optogenetic experiments, we have developeda supercapacitor-based headborne device which can controlmultiple headborne LEDs, receiving power (and optionally,real-time delivered stimulation protocol information) in a fullywireless fashion. The device is small (< l cm3), and weighsapproximately 2 g when operated autonomously with pre-programmed modulation protocols, or 3 g when equippedwith optional wireless telemetry, both implementations beingappropriate for use in small animals suchasmice. Inthis paperwe present the design, which centers around a high-efficiencyresonant wireless power transfer system and headbornesupercapacitor-based energy storage, appropriate to supportthe reliability and high-power operation requirements of theoptogenetic research. The power transmitters are low-profiledevices that can fit under behavioral arenas or cages. Weshow that such systems can sustain multi-wan power delivery

both continuously and in burst mode, and demonstrate controlof behavior in untethered mice expressing ChR2 in motorcortex pyramidal cells. Such systems will not only enable anumber of fundamentally new kinds of experiment, but mayalso serve as prototypes for a new generation of clinical neuralprosthetics that achieve great precision through the use of light-targetable molecules as the transducers of cell-type-specificneural control.

2. Materials and methods

2.1. Design andfabrication

The complete wireless optical neural control system consistsof a headborne device (depicted in figure 1), a wireless powertransmitter (depicted in figure 2(a)) and a USB-connectedwireless base station (shown in figure 2(b)) for communicationwith the headborne device. We describe here the design andfabrication of these elements.

2.1.1. Headborne device. The headborne device comprisesfour distinct modules-the power, radio, motherboard andoptics modules. The optics module, containing the LED lightsources, is surgically affixed to the skull, while the remainderof the device (the motherboard module, the power module, andthe optional radio module) is attached to the optics module viaa low insertion force connector (figure 1(f)).

Unless otherwise noted, construction of modules isas follows: circuit schematics for the power, radio,motherboard, and optics modules (see supplementaryfiles available at stacks.iop.org/JNE8/046021/mmedia) arecreated using Eagle CAD Professional (Cadsoft). Theradio and motherboard modules are designed as four-layer printed circuit boards (PCBs) using Eagle CADProfessional and fabricated by Advanced Circuits. PCBsare populated with parts (see supplementary files available atstacks.iop.org/JNE/8/046021/mmedia) using standard solderpaste and reflow oven techniques.

As identified in the board-level drawings (figure 1(b))and module photographs (figure 1(c)), the radio moduleshown at top contains a surface mount antenna (1; note:numbers here refer to the flagged items in figures 1(a)-(c)), a1 Mbit s- radio chipset operating in the ISM 2.4-2.485 GHzband (2), and a six-pin radio-to-motherboard dockingconnector (3). Calculated matching network componentvalues for the radio module were verified in simulation usingLT Spice IV/SwitcherCAD III (Linear Technology).

The power module (shown second from top in figures 1(b)and (c)) consists of a receiver element for power reception(4), a full-wave rectifier for ac to dc conversion (5),a supercapacitor (6) and a power-to-motherboard dockingconnector (7). Calculated passive component values for thepower module tank circuit were simulated using LT SpiceIV/SwitcherCAD Ill. To minimize device size, the powermodule is assembled without a PCB, instead using the body ofthe supercapacitor as a structural element onto which all theparts are mounted using epoxy. The male power connector isused as a structural element to affix the power receiver antenna.

2

45

C T Wentz et alJ. Neural Eng. 8 (2011) 046021

u

111111

*.

U0 1 ~i1'Ibug

A1.1 I I

~

-A

A .,I I

iI

J

IJItia 1

J. Neural Eng. 8 (2011) 046021

(a)A system for wirelesslypowered & controlledoptogenetics

Freely-movingmouse equipped withheadbome opticalneural control device

Wireless power The wireless base station:transmitter a PC equipped with USB-connected

wireless base station

(c)

Pw To

C T Wentz er al

(b)The wireless base station

USB-connectedinterface board

Computer

DockedMotherboard Module+ Radio Module (partof the slack shownin Fig. 16) herealso doubles aswrelesstransceiver

Dock for wireless transceiver

(d)

Demonstration of wirelessoptical control of mouse movement

Wireless device operation characteristics

E uJ. .W

Wireless Power (time-averaged throughout

zDeviceat rest

Devicepowered and

executingprotocol

LEDs Disabled

DeviceDevice powered andat rest executing

protocolLEDs Enabled

train, A * PW * PR, in Watts)

Figure 2. A simple wireless power and communication interface for operation of headborne optical neural control devices on the awakebehaving mouse. (a) Photograph of an arena equipped with a power transmitter coil (a 120 kHz LC tank circuit), containing a mouseequipped with a headborne optical neural control device; schematic showing operation with a computer, base station attached via a USB port(detailed photograph of the base station in (b)). When powered up, the microcontroller automatically initializes the radio module if attached,wirelessly connecting to the base station to receive instuction from the experimenter; if no radio module is attached, can operate inopen-loop fashion. (b) Photograph of a USB-connected wireless interface board (red) docked to a laptop, and equipped with a copy of aheadborne device (green) to serve as the transceiver for wireless communication (collectively denoted the base station). (c) Guidelines forusage of wireless headborne optical control devices for typical neuroscience experiments involving pulse trains of light delivery. (i) Weintroduce a schema defining various properties of pulse trains (i.e. within-train duty cycle = PW/(l/PR), and within-train average power =A *PW/(1/PR)), for assistance with visualizing typical protocols for device operation. (ii) Plot of the range of typical protocols for deviceoperation, expressed as a function of within-train duty cycle and within-train average power, assuming that the between-train pause (i.e. IT -2 * TD) is at least 3 s. Any point under the curve is easily achievable. Because the power antenna continuously receives 2 W, the device canrun indefinitely with a time-averaged power of 2 W; the device can exceed 2 W using the supercapacitor's excess capacity, but with reducedduty cycle, and never crossing the device maximum peak power of 4.3 W. In addition, there is no hit in device performance withbetween-train pauses of less than 3 s if 2 W or less is consumed; if more than 2 W is consumed, because the supercapacitor's excess capacityis needed, some time (up to 3 s, depending on how much energy is used) will be required to recharge the supercapacitor in between theLED-on periods that drain the supercapacitor (d) Unilateral optogenetic control of motor cortex neurons in a freely behaving Thyl-ChR2mouse, which expresses ChR2 in layer 5 pyramidal cells (right), eliciting reliable drive of mouse rotation compared to the no-light conditionbefore stimulation (n = 9 trials across two subjects, positive value indicates CCW rotation, * indicates p < 0.01, paired t-test). When theLEDs were turned off (left), no rotation was apparent (p > 0.5, n = 5 trials, paired s-test).

A wiring harness connects the supercapacitor assembly to the and a ceramic tuning capacitor with a resonant frequency ofconnector/antenna assembly. The power receiver element of 120 kHz.the RF link consists of a resonant parallel LC tank cicuit, built The motherboard module, shown second from bottom inof a 15 mm long, 2 mm diameter wound copper ferrite core figures 1(b) and(c), contains the microcontroller (9), two-stage

47

ii

J. Neural Eng. 3(2011) 046021 CT Wcntz eta)

LED power conditioning circuitry (11), and connectors fordocking the power (8), radio (10), and optics modules (12); theten pins on connector 12 that are not used by the optics moduleare used for programming of the microcontroller (e.g. as seenin figure 2(b), which shows a motherboard + radio connectedto a laptop through the USB-connected interface board (redboard). Note well that figure 2(b) shows a USB-connectedinterface board with acopy of the motherboard + radio moduledocked to it; when such a copy of the motherboard + radiomodule is docked to the interface board, this docked pair ofmodules serves as a transceiver to communicate with one ormore headborne devices in an experiment (see below). Whenassembled as shown in figure 2(b), the device is referred to asthe wireless base station.

Programs running on the motherboard'smicrocontroller (in supplementary files available atstacks.iop.org/JNE/8/046021/mmedia) were writteninCusingthe IAR Embedded Workbench development environment(IAR Systems). Radio communication protocols were builton top of the SimpliciTi network protocol developed byTexas Instruments. Compiled programs were downloadedto the microcontroller on the motherboard module using theUSB-connected interface board, as described in the previousparagraph. The interface board comprises a microcontrollerdevelopment tool (Texas Instruments EZ430-RF2500) thatwe modified with a custom docking circuit board so as tointerface the EZ430 USB tool to the motherboard module.

The optics module, shown at bottom in figures 1(b) and(c), comprises a connector that docks to the motherboardmodule (14), a copper thermal sink that acts as a structureto mate bare die LEDs (16) up to 1 mm x 1 mmin size each, and also serves as a cathode for LEDs(15), and a thermistor acting as temperature sensor(mounted on bottom of the copper, see supplementaryfiles available at stacks.iop.org/JNE8/046021/mmedia forschematic), as well as the LED multiplexer (attached to theside of the connector; see supplementary files available atstacks.iop.org/JNE/8/046021/mmedia). A customPCB milledfrom gold plated FR4 board stock acts as a common structuralelement to link the copper sink containing LEDs, the optics-to-motherboard mating connector and integrated circuitry. Theoptics module PCB schematic and layout was designed usingEagle CAD Professional, and then exported using a customPython script to be machined by a three-axis tabletop millingmachine (Modella). The copper thermal block was machinedwith a waterjet cutter (Omax) from copper bar stock andthe underside LED mounting surfaces were then milled outusing the tabletop mill. In this instantiation of the device,1 mm x 1 mm bare die LEDs (16) were reflow solderedto milled out pedestals in the copper, and the resultingLED + thermal block assembly was reflow soldered to the topside of the PCB. Finally, anodic electrical connections fromindividual LEDs to PCB traces were made using aluminumwedge wire bonds (0.001" diameter). For robustness, bare diewire-bonded LEDs were coated with a protective layer of UVcured clear optics glue (Thorlabs) and the remainder of theunderside of the device was potted in black epoxy to ensureelectrical isolation from tissue. Following these assembly

steps, a miniature 0201-sized thermistor is epoxied to thecopper thermal block using thermally conductive epoxy fortemperature monitoring via the microcontroller.

2.1.2. Wireless power transmitter. Power is coupledwirelessly from an under-arena power transmitter (FerroSolutions) to the headborne device using resonant energytransfer over distances of several centimeters using a low-strength oscillating magnetic field (peak 300 A m-),as diagrammed in figure 2(a). (For a plot of themagnetic field, M, versus vertical distance above thetransmitter, see supplementary figure 5 available atstacks.iop.org/JNE/8/046021/mnimedia.) The arenas used inour prototyping and experiments were either a circular mousecage or a circular bowl of -8" in diameter. The under-cagepower transmitter consists of a resonant series LC circuittuned to the frequency of the power receiver circuit (in thisinstantiation 120 kHz) and an asynchronous bridge driver todrive the LC circuit at this frequency, as described previouslyin [It].

In this instantiation of our system, the under-cage powertransmitter was powered from an external dual channel powersupply (Agilent E3646A); separate power supplies serve theLC circuit and the bridge driver control circuitry. A controlsignal to the bridge driver was supplied externally via aprogrammablefunctiongenerator(Stanford Research SystemsDS345) connected via 50 Q terminated BNC cable. Thus,the receiver element of the headborne device is orthogonal tothe floor of the cage, seen in figures 1 and 2(a), such that itmaximally couples the transmitter's oscillating magnetic field.

2.1.3. USB-connected base station. The USB-connectedbase station serves two functions: (1) to program theassembled headborne electronics unit before attaching to

the animal (as described above), and (2) to wirelesslycommunicate with headbome devices. The USB-connectedbase station consists of an interface board (figure 2(b), red-Texas Instruments), a custom docking board, a motherboardmodule and a radio module identical to those used in theheadborne devices. When equipped with a copy of theheadborne device to serve as a radio transceiver (comprisingthe base station), the microcontroller provides a seamlesswireless communication interface between headborne deviceson animal subjects and the PC to which the base station isattached.

2.2. Animal preparation and experimental setup

Adult Thy-I-ChR2 transgenic mice (eight-ten weeks old,-30 g, line 18 in [12, 131) were utilized in this experiment. Allanimal procedures and protocols were approved by the MITCommittee on Animal Care.

2.2.1. Surgery and animal preparation. Mice wereanesthetized using 1.5% isoflurane in oxygen. The top ofthe head was shaved and the animal was placed in a stereotax(Kopf). Betadine and 70% ethanol was used to sterilize thesurgical area and ophthalmic ointment was applied to the eyes.

48

- X: - - 1 -11, . ..... ..... ........ .............. ...................

J. Noural Eng. 8 (2011) 046021 C T Wentz et at

3. Neural Ens. 8(2011) 046021 CT Wentz et at

A mid-line scalp incision was made, and the skin retracted. Ata center point of anterior-posterior (AP) 2.0, medio-lateral(ML) 2.0 relative to bregma, the skull thickness was thinnedusing a dental drill to create a 2 m x 2 mm bone window forthe LED implant to sit in. Three surgical screws were insertedinto the skull to secure the implant. Kwik-Sil (World PrecisionInstruments) silicone elastomer was used to fix the implantin place temporarily, and to improve the light transmissionefficiency between the LED and brain tissue. Dental acrylic(Stoelting) was then used to affix the implant to the screws andthe skull. The animals were allowed to recover for five daysprior to behavioral testing.

One LED was targeted unilaterally to MI motor cortex(AP -1.0 mm, ML 0.5 mm). An additional LED was includedfor debugging purposes, targeted to M2 cortex (AP -0.3 mm,ML 0.5 mm), but not used in the experiment.

2.2.2. Motor control experimental setup. Mice wereacclimated with headborne devices for at least 15 min priorto behavioral testing in the experiment arena consisting of aplastic bowl (approximately 8" diameter) with an under-cagepower transmitter operating at 5 A. A CCD camera (Logitech)was suspended above the arena to record behavior. Wedelivered LED illumination to Ml cortex (15 ms pulse width,30 Hz) at 250 mW input power to the LED for 30 s followed by90 s off time, repeated four-five times in a session. After theexperiment, animal rotation was scored in 30 s epochs beforeand during the illumination phase. Rotation was scored as thenumber of net rotations within the 30 s epoch to the nearestquarter rotation (positive rotations being counter-clockwise),and is plotted in figure 2(d). An additional control experimentwas performed, in which identical power levels and controlsignals were transmitted, but with LEDs disabled via the LEDcontroller chipset. Thus, in this control experiment subjectsare exposed to identical electrical and magnetic fields, but arenot driven optically. Statistical analyses were performed inExcel (Microsoft).

2.2.3. Measurement of the emitted magnetic field Toquantify the magnetic field produced by the wireless powertransmitter, we applied a dc current to the transmitter coil(5 A), identical in magnitude to that used in normaloperation, and measured the resulting magnetic field, M,at varying heights above the center of the transmitter coil,where field strength is maximal, using a Lakeshore 450Gauss meter. A plot of the resulting magnetic fieldprofile is shown in supplementary figure 4 available atstacks.iop.org/JNE/8/046021/mmedia.

2.2.4. Bench measurment of optics module temperature.To measure optics module temperature, we utilized aK-type thermocouple (Omega Engineering) affixed to thetissue interface side of a non-implanted optics module atroom temperature and cycled power to the LED via theheadborne electronics unit using parameters identical tothose in our motor control experiments (30 Hz, 15 msduty cycle, and 250 mW input power to the LED). The

resulting temperature of the tissue interface side of the opticsmodule was measured using a calibrated Omega HH506RAdata logger (plotted in supplementary figure 5 available atstacks.iop.org/JNE/8/046021/mmedia).

3. Results

3.1. Device description and rationalefor design

We set out to develop a compact. robust optogenetic neuralcontrol device that is capable of supporting experiments inwhich use of untethered freely-behaving animals is desired.In this section, we describe the design of the system andprinciples of operation. In the following section we describeactual use of the wireless optogenetic tool in an experimentalparadigmn.

To achieve the functionality desired for freely-behavingoptogenetic experiments, we developed a modular headborneelectronic device (figure 1), accompanied by a wireless powertransmitter which resides underneath the behavioral arena(e.g. cage or maze), and a USB-connected base station toenable wireless control of headborne devices using a standardcomputer (figure 2). The headborne device consists of fourdifferent modules, namely the optics, motherboard, power andradio modules, which dock together to provide, respectively,the light delivery, waveform generation, power receiving andcommunication functions necessary to achieve untetheredoptogenetic control. Light delivery to target tissue is achievedvia an optics module (figures 1(b) and (c), shown at thebottom) that contains a number of LEDs mounted in a compactstructure that provides both electrical power delivery to theLEDs and dissipation of heat generated during their operation.In the current embodiment. the optics module can hold upto 16 LEDs, which are individually controllable via an on-module multiplexing circuit that is operated by digital signalsfrom a microcontroller residing on the motherboard module.The optics module is the only part of the headborne device thatis chronically implanted on the mouse skull; the other threemodules dock together into an electronics stack that plugsinto the optics module, either after the surgery is complete orbefore the experiment is about to begin. The microcontrollerthat controls the optics module is located on the motherboardmodule (figures 1(b) and (c), second from bottom), whichserves not only as the core structural element of the electronicsstack, but also helps the device achieve efficient wirelesspower by actively regulating the LED power circuitry andthe transfer of energy, received wirelessly via resonant powertransfer, to a storage supercapacitor. To facilitate real-timeupdating of control waveforms, the microcontroller can alsoreceive radioed instructions via the optional radio module(figures 1(b) and (c), top), which docks to the motherboardmodule, and communicates with a base station equippedremote PC (figure 2(b)). To wirelessly receive and storepower for device operation, the power module containsboth a resonant wireless energy receiver as well as astorage supercapacitor, and docks to the motherboard module(figures 1(b) and (c), second from top).

The modular headborne device described above isdepicted in block diagram fashion in figure 1(a), each

6

49

C T Wentz er atJ. Ncurad Eng. 8 (2011) 046021

J. Neural Eng. 8(2011)046021

module being represented by a gray box, and with the majorsubcomponents enabling the functionality of each of the optics(left), radio (top-center), power (top-left), and motherboard(bottom-right) modules shown in white boxes. Black tracescrossing between each gray box represent the major controland power interconnects between modules. As can be seen infigure 1(b), the power and control interconnects are made usingsmall form factor low insertion force connectors (Samtec),allowing their disassembly, but ensuring the device remainsconnected during a long-term experiment. The completeelectronics stack consisting of radio, power and motherboardmodules seen in figure 1(c) is shown as a unit in figures 1(d)and (e), and affixed to a subject in figure 1(f) (supercapacitorremoved to show connector interfaces).

Each of these modules was designed with a set ofstrategies in mind to achieve three core goals: (i) highpower density to support operation of several LEDs, (ii)flexible and simple device operation in a variety of experimentprotocols and (iii) robust, reliable operation. Throughoutall modules, compactness, robustness and flexibility wereachieved by using off-the-shelf parts, chip-scale packaging andhighly integrated system-on-chipcircuitcomponents whereveravailable. The modularity of the physical device layout, witheasily replaceable subsystems, allows for rapid incrementalimprovement of the design. Specific modules were furtheroptinized to meet the above goals as described below.

The ceptral strategy that enables this system to achievehigh power density is the use of a supercapacitor-basedpower module to store and buffer wirelessly receivedpower. By utilizing a supercapacitor energy storage elementand wireless recharging approach rather than battery-basedimplementation, this device is able to achieve a greater thanlOx reduction in size over other battery-based approaches,while increasing the maximum peak power deliverable toLEDs by approximately 15 x over previous designs [141,and allowing the system to operate at these power levelsindefinitely. The reason for such substantial size reductionwith increased power delivery over previous battery-basedapproaches lies in the physical constraints: although batteriesare capable of high energy density, they achieve powerdensities several orders of magnitude lower than those ofsupercapacitors [15]. This is a critical observation becauseoptics systems utilizing LEDs to drive pulsatile or enduringillumination protocols, as is commonly desired in optogeneticexperiments, have high-power requirements but only modestenergy requirements (that is, the peak-to-average power ratiois high, due to the low duty cycle of most optogeneticexperiments). Most batteries are able to supply a dischargecurrent rating of 1-5C- (where C is the total energy capacityof the battery). Because the peak power requirement of anLED-based system is quite high (approximately 400 mA @3.6 V per LED with 1 mm x 1 mm LEDs), but the time-averaged power of a pulse train may be more modest, the peakpower discharge rate limitation of the battery-based solutionnecessitates use of a battery much larger than needed basedon the energy requirements of the system. Thus, by utilizinga constant wireless energy transfer link operating at the time-averaged power of the system (which may be at a lower level

C T Wenz et at

than the peak power utilized by the LEDs), the supercapacitorprovides an energy buffer to deliver bursts of pulse trainsat multi-watt levels, without needing to store on the animalany of the unused energy capacity required for the battery-based approach. Operating the device at matched resonantfrequencies with a high Q (>100) receiver circuit maximizeswirelessly coupled energy transfer efficiency [15].

The flexibility and simplicity of use is achieved throughthe software reconfigurability of the headborne device. Toenable many different types of experiments to be performed,the motherboard module can be programmed to generatemultiple independent waveforms, either running continuouslyin pre-programmed fashion or triggered and updated remotelyby commands via the wireless communication link (e.g. basedon a behavioral input to the computer such as a nose pokeor other cue, or manually triggered). Multiplexers residingon the optics module and controlled by commands from themotherboard module (or through the wireless communicationlink) allow these waveforms to be addressed to particularLEDs, and therefore particular neural circuits, such thatmultiple neural circuits may be modulated independentlywithin the same animal. One motherboard module can beused with many differently designed optics modules (e.g.with different numbers of LEDs, different neural targets, etc);indeed, a key aspect of our design is the independence ofthe modules and the resulting economies of scale for typicalacademic or industrial neuroscience labs.

Because wireless power by its very nature will vary inefficiency over time (e.g. with the distance betweentransmitterand receiver), we ensure safe, robust operation of the deviceusing both passive and active/algorithmic strategies. Thepower module's resonant LC tank circuit is tuned to generateopen circuit voltages that, when rectified using a full waverectifier topology, are below the maximum safe operatingvoltage of a supercapacitor energy storage element (-5.5 V).Nevertheless, to ensure device integrity, we have implementedfurther safety elements to limit device operation to safevoltage ranges. To maintain a supercapacitor voltage thatis within the safety tolerances of the device (that is, below thebreakdown voltage of the supercapacitor), an adaptive controlloop running on the microcontroller monitors supercapacitorvoltage and regulates a switch placed between the rectifier andsupercapacitor, limiting current delivered to the supercapacitoras appropriate (e.g. when the LEDs are not operated). Theheadborne supercapacitor is further protected using a Zenershunt circuit, which acts to short current from the positiveterminal of the power rectifier output to the analog powerground terminal, when voltage levels approach the capacitor's5.5 V rating. It should be noted explicitly that this analogpower ground is isolated from the animal, such that no currentis dissipated through the animal; in addition, the entire deviceis physically electrically insulated from the animal (as notedin the methods above). Finally, to maintain safe operatingtemperatures near tissue, the temperature of the optics modulecontaining the LEDs is continuously monitored using athermistor mounted on the underside optics module heat sinkand a control loop running on the microcontroller. LEDoperation is disabled once a set temperature is reached, settable

50

........ .... ...... .... ..... ......... .

J. Ncuxal Eag. 5(2011)046021 CT Wentz ~ al

by the user in software, and the device prompts the user viawireless telemetry that a temperature fault has occurred. In ourexperiments, we limited the allowable operation temperatureto within 1 "C of baseline body temperature. We additionallyverified that this control circuit was operating at expectedtemperatures in a bench test setup utilizing a calibratedthermocouple and data logger. Maximum temperature risein steady state was observed to be approximately 0.6 'C (for aplot of this thermal performance, see supplementary figure 5available at stacks.iop.org/JNE/8/046021/mmedia). It shouldbe noted that this setup, in which the optics module is notsurgically affixed to skull but freely exposed to air, representsa conservative worst-case scenario of heat transfer efficiency,as the total heat capacity of the headborne device is effectivelyincreased when affixed to the animal skull [16, 17].

3.2. How to use the system

Once the headborne device has been assembled as describedin section 2 and the optics module is surgically affixed to theskull as described in section 2.2.1 (with coordinates of LEDschosen based on intended brain targets), the animal is allowedto recover from surgery. With the headbome device attachedto the optics module, the animal is able to freely move andbehave (figure 1(f)). Once placed in the behavioral arena, thepower module automatically activates and begins to charge theonboardsupercapacitor, which inturnenables the motherboardand radio modules to self-initialize. The headborne systemimtiates network pairing with a nearby USB-connected basestation if available, and then awaits the instruction of theexperimenter. From the base station connected computer, theexperimenter may use any serial communication program (e.g.a MATLAB program, custom script, etc) that uses the USBdevice as a COM port to update the LED activation patternson each headborne device within range.

Figure 2(c) provides guidelines for usage of wirelessheadborne optical control devices for typical neuroscienceexperiments involving pulse trains of light delivery. Herewe define terminology for various properties of the pulsetrain (i.e. within-train duty cycle = PW/(l/PR), and within-train average power = A * PW * PR, where A = LED poweramplitude, PW = pulse width and PR = pulse rate),for assistance with visualizing typical protocols for deviceoperation. Figure 2(c)(i) graphically shows the range oftypical protocols for device operation, expressed as a functionof within-train duty cycle and within-train average power,assuming that the between-train pause (i.e. 1TI -2* TD, whereITI = inter-train interval, TD = train duration) is at least3 s (a simplifying assumption that would be compatible withmany neuroscience experiments-this assumption allows usto assume that the supercapacitor always has time to rechargebetween trains; by taking into account partial charging, thepotential for other protocols can be estimated). Any pointunder the purple curve is easily achievable. Because thepower antenna continuously receives 2 W, the device canrun indefinitely with a time-averaged power of 2 W; thedevice can exceed 2 W using the supercapacitor's excesscapacity, but with reduced duty cycle, and never crossing

the device maximum peak power of 4.3 W (higher maximumpeak powers are easily achievable with use of higher capacitysupercapacitors; we chose a 400 mF capacitor in thisinstantiation based on expected current draw for motor controlapplications). At 4.3 W, these performance figures translateto four I mm x I mm LEDs being driven at 50% duty cyclefor 3 s with 3 s lTI, or eight of the same LEDs for 1.5 s burstsat 50% duty cycle, or 16 LEDs for 0.75 s bursts at 50% dutycycle. In continuous operation, 2 of these LEDs can be drivenat 100% duty cycle to consume 2 W. There is no hit in deviceperformance with between-train pauses of less than 3 s if 2 Wor less is consumed; if more than 2 W is consumed, becausethe supercapacitor's excess capacity is needed, some time (upto 3 s, depending on how much energy is used) will be requiredto recharge the supercapacitor in between the LED-on periodsthat drain the supercapacitor during the pulse train delivery.

The total energy is stored on the supercapacitor scalesas the square of the voltage on the capacitor. The LEDforward voltage is typically -3.6 V, well within the range ofthe 5.5 V maximum voltage allowed on the supercapacitorused in this design. To make the stored energy on thesupercapacitor available for LED operation during periodsin which supercapacitor voltage transiently drops below this3.6 V (e.g. high current pulsatile trains, changes in headdistance from the transmitter), a two-stage power conversionscheme is implemented to boost circuit output voltage. Inthis manner, nearly the full operating voltage range ofthe supercapacitor (and therefore, nearly the full range ofenergy stored) may be utilized for LED power by utilizinga combination of a first stage boost converter circuit topology,which up converts voltage on the supercapacitor from as littleas 0.3 V to the LED's typical 3.6 V forward voltage, as wellas a second stage of switched capacitor LED drivers whichprovides current regulation.

3.3. Demonstration of remote motor control

We demonstrate the applicability of this technology tountethered freely-behaving animal optogenetics, using awell-validated and easily-quantified behavioral paradigm ofcortically-driven unilateral motor control in the mouse [6].A simple paradigm in which 470 em light was deliveredunilaterally to Ml motor cortex at 30 Hz with 15 ms pulsewidth at 250 mW LED input power in 30 s epochs followedby 90 s periods of rest resulted in increased rotation to thecontralateral side during stimulation as compared to the epochbefore stimulation (figure 2(d), right, p < 0.01, paired t-test,n = 9 trials in two animals). Thus, we can drive behaviorally-relevant protocols at sufficiently high power levels to elicitrobust behavioral changes.

We measured the magnetic field produced by the powertransmitter and found that peak field strength was less than300 A m-1, orders of magnitude less than that shown toaffect neural excitability or drive neural activity [181. Weadditionally performed a direct control experiment, repeatingthe motor control experiment with just the LED outputsdisabled (i.e. by disabling the LED controller (shown infigure 1(a)) via program pins). In this control experiment,

8

51

C T Wentz et alJ. Neural Eng. 8 (2011) 046021

J. Neural Eng.8 (2011) 046021

the animal experiences the electromagnetic field, power andcontrol signals are received by the headbome device, and thepower receiver dissipates heat via shunting of received currentin the supercapacitor control circuit, but no light is emittedfrom the LED. We observed no change in rotational behaviorin this experiment (figure 2(d), left, p > 0.50, n = 5 trialsin one subject). As mentioned above, we further control fortemperature-related effects by limiting the maximum rise inoptics module temperature to less than 1 *C, itself insulatedfrom the brain by high thermal resistivity epoxy and skull.

4. Discusslon

The system described here expands upon existing tetheredbehaving animal optogenetic strategies, by enabling trulywireless freely-behaving optogenetics. The device describedhere is able to operate indefinitely while delivering up to2 W of power to an array of LEDs from a compact under-cageor under-behavioral-arena power transmitter operating at low-strength magnetic fields (300 A m-), and is able to deliver upto 4.3 W of power intermittently by utilizing a supercapacitor-based energy storage system to buffer the wireless power hnThis system provides the ability to control many headbornedevices simultaneously using software polling techniquesthat are amenable to high-throughput in vivo neural circuitscreening. Without the requirement to physically connect anddisconnect each animal as in previous tethered approaches,this untethered system further reduces potential behavioralartifacts associated with animal handling and possible torqueon the animal's head from a tethered fiber. Animals in ourexperiment were observed for at least three months withimplanted optics modules, and exhibited natural grooming,nesting and exploratory behavior throughout this time frame,suggesting that this approach does not interfere with normalanimal livelihood. We are currently evaluating futureiterationsof the design presented here, and applying the untetheredoptical neural control setup to new optogenetic researchparadigms, briefly described in the following sections.

4.1. Futumr directions

The headborne device presented here is sufficiently compactin size to enable use in mouse experiments, at approximately2 g total weight in pre-programmed operation and less than3 g with the optional radio module, which enables real-timeupdating from a PC. The supecapacitor-based storage elementenables up to 4.3 W of intermittent power delivery directableto up to 16 LEDs. In future implementations, it would befeasible to target deep brain structures with this design byattaching optical fibers to the surface of the LEDs. Severalunder-cage power transmitters may also be tiled to cover largearea mazes or other environments, enabling a new class ofoptogenetic research paradigms.

Nearly all of the underlying technologies enablingthis device (system-on-chip telemetry and microcontrollersystems, solid state light sources, wireless power transfer,thin film supercapacitors) are being improved daily, driven bytheir applications to this and other fields such as computing,

C T Wentz e: at

telecommunications, and energy. Thus, we expect theperformance of this systemto be further improved as individualtechnologies mature. The use of solid state miniature lasersin particular would enable a substantial reduction in the netpower requirements of the system (LEDs are an uncollimatedlight source with Lambertian distribution, and thus the lightspread is broad). Altematively or in addition, developmentof light-sensitive ion channels compatible with existing solidstate laser technology, e.g. with activation spectra shiftedto the red wavelengths, would substantially increase thepower efficiency of this system overall. In energy storage,development of higher energy density supercapacitors orhybrid supercapacitor + rechargeable battery systems willimprove the performance of this systemoverlonger timescales,such that wireless charging requirements may be relaxed.The use of far-field wireless power delivery techniques mayfurther simplify the delivery of charging power. By ridingtechnology development curves, the system described heremay eventually be miniaturized to a few mm2 , furtherincreasing its applicability to research and potentially enablingclinical applications not addressable with today's devices.

Clinically, modulation of neural circuits via electricalstimulation (deep brain stimulation, or DBS) has beendemonstrated successfully in the treatment of Parkinson'sdisease and dystonia symptoms, with several other promisingindications in clinical trial. These DBS approaches to treatingneurological disorders rely on an implanted pacemaker-likedevice, implanted either in the chest [191 or cranium [20],itself composed primarily of a large battery, providing powertocurrentsources, which in turn deliver therapeutic stimulationto several electrode contacts by way of a flexible single lead orpairofleads.Whilethisapproachhas provenhighlyefficaciousfor the neurological disorders mentioned above (amongothers), side effects are often reported with this modality andare believed to originate from the unwanted recruitment ofnearby neuronal cell bodies and axons [21]. The single-cell-type specificity enabled by optogenetic implementations maytherefore be a potential avenue to pursue to eliminate the sideeffects observed with existing electrical DBS approaches.

Finally, substantial risk of clinical device failure has beenattributed to breakage of the flexible leads connecting thelarge implantable pulse generator and the implanted electrodes[22]. As the number of target sites of therapeutic interestincreases, particularly as DBS technologies are deployed indiseases where the pathology cannot be traced to a single,statically defined brain region (e.g. epilepsy), the issue ofdelivering therapeutic stimulation to the target site withoutrisk to the patient will become increasingly difficult. Onecan therefore envision a system in which this lead is replacedby a wireless power link, and the electrode is replaced by aminiaturized version of the headbome system described here,such that many brain regions may be independently addressedtherapeutically, each receiving power and control signalswirelessly from a small implantable antenna on the surfaceof the brain. Using such an approach, it may be possibleto develop truly adaptable, brain-wide DBS-like therapiesto target disorders intractable to current pharmacological ordevice-based therapies.

9

52

... .........

J. Neural Eng. 8 (2011) 046021

Acknowledgments

ESB acknowledges funding by the NIH Director'sNew Innovator Award (DP20D002002) as well as NIHGrants 1R01DA029639, 1RC1MH088182, 1RC2DE020919,1R01NS067199 and 1R43NS070453; the NSF CAREERaward, as well as NSF Grants EFRI 0835878, DMS0848804 and DMS 1042134; Benesse Foundation, Jerryand Marge Burnett, Department of Defense CDMRPPTSD Program, Google, Harvard/MIT Joint GrantsProgram in Basic Neuroscience, Human Frontiers ScienceProgram, MIT McGovern Institute and the McGovernInstitute Neurotechnology Award Program, MIT MediaLab, MIT Mind-Machine Project, MIT NeurotechnologyFund, NARSAD, Paul Allen Distinguished Investigator inNeuroscience Program, Alfred P Sloan Foundation, SFNResearch Award for Innovation in Neuroscience and theWallace H Coulter Foundation. CTW designed, fabricated andtested the wireless system and contributed to the manuscript,with technical support for power transmitter implementationfrom Ferro Solutions, which provided the wireless transmitter.JGB designed the core of the optics module and fabricatedthese components with assistance from AG. PM performedanimal surgeries and assisted in behavioral experiments, withassistance from AR. ESB provided many insights into systemand experiment design and invaluable contributions to themanuscript. The authors thank X Han for her assistance withbehavioral experiments.

References

[1] Boyden E S et al 2005 Millisecond-timescale, geneticallytargeted optical control of neural activity Nat. NeAurosci.8 1263-8

[21 Han X and Boyden E S 2007 Multiple-color optical activation,silencing, and desynchronization of neural activity, withsingle-spike temporal resolution PLaS ONE 2 e299

[31 Zhang F el al 2007 Multimodal fast optical interrogation ofneural circuitry Nature 446 633-9

[41 Chow B Y et al 2010 High-performance genetically targetableoptical neural silencing by light-driven proton pumpsNature 463 98-102

[5] Gradinaru V et al 2010 Molecular and cellular approaches fordiversifying and extending optogenetics Cel 141 154-65

[6] Aravanis A M et al 2007 An optical neural interface: in vivocontrol of rodent motor cortex with integrated fiberoptic andoptogenetic technology J. Neural Eng. 4 S143

C T Wentz es al

[7] Bernstein J G et al 2008 Prosthetic systems for therapeuticoptical activation and silencing of genetically-targetedneurons Proc. Soc. Photo-Opt. Instrum. Eng.6354 68540H

[8] Huber D ea! 2008 Sparse optical microstimulation in barrelcortex drives learned behaviour in freely moving miceNatwre 45161-4

[9] Arfin S K et a 2009 Wireless neural stimulation infreely behaving small animals J. Neumphysiol.102 598-605

[101 Iwai Y et a 2011 A simple head-mountable LED device forchronic stimulation of optogenetic molecules in freelymoving mice Neursci. Res. 70 124-7

[11] O'HandleyRC, HuangJ K,Bono DC and SimonJ 2008Improved wireless, transcutaneous power transmission forin vivo applications IEEE Sensors 18 357-62

[12 Arenkiel B Ret a! 2007 In vivo light-induced activation ofneural circuitry in transgenic mice expressingchannelrhodopsin-2 Neuron 54205-18

[131 Wang H et al 2007 High-speed mapping of synapticconnectivity using photostimulation inC8annelrhodopain-2 transgenic mice Pmc. Nat! Acad Sci.USA 104 8143-8

[14] Parallikar K, Cong P, Santa W, Dinsmoor D, Hocken B,Munns G, Giftakis J and Denison T 2010 An implantable5 mW/channel dual-wavelength optogenetic stimulator fortherapeutic neuromodulation research IEEE Solid StateCircuits Conference Digest of Technical Papers (ISSCC)2010 (San Diego, CA) pp 238-9

[15] Sarpeshkar R 2010 Ultm-Low Power Bioelectronics(Cambridge: Cambridge University Press)

[16] Janssen F ,Van Leeuwen G M and Van Steenhoven A A2005 Modelling of temperature and perfusion during scalpcooling Phys. Med Biol. 50 4065-73

[17] Kassakin J G, Schlecht M F and Verghese G C 1991Princiles of Pbwer Electronics (New York:Addison-Wesley)

[18] Radman T etal 2009 Role of cortical cell type andmorphology in subthreshold and suprathreshold uniformelectric field stimulation in vitro Brain Stimulation2 215-28

[19] Caparros-Lefebvre D et a 1993 Chronic thalamic stimulationimproves tremor and levodopa induced dyskinesias inParkinson's disease J. Neurl. Neurosurg. Psychiatry56 268-73

[20] Skarpass T L and Morrell M J 2009 Intracranial stimulationtherapy for epilepsy Neurotherapeutics 6 238-43

[21] Schuurman P R et al 2000 A comparison of continuousthalamic stimulation and thalarnotomy for suppression ofsevere tremor N. Engl. J. Med 342 461-8

[22] Oh M Y et al 2002 Long-term hardware-related complicationsof deep brain stimulation Neurosurgery 50 1268-741274-6 discussion

10

53

development and application of control tools for use in optogenetics

Documents