A systems approach to sub-cellular localisation of proteins

Kathryn S. Lilley

Cambridge Centre for Proteomics, Department of Biochemistry, University of Cambridge, United Kingdom, CB2 1QR

SPEAF 2012Rouen

Organelles of the cell

http://media.web.britannica.com/eb-media

Eucaryote cells have many different types of sub-cellular compartments (some specific to a cell type)

Many proteins reside in multiple locations

Within these locations many form functional units

Dynamic changes in these locations (and binding partners) reflect biological processes in which a protein functions

Most proteomics protocols start with addition of detergents which destroy delicate sub-cellular structures

Changes in sub-cellular dynamics are as important as changes in abundance, post translational status and interacting partners.

Tagging of a fluorescent

fusion proteins

Immunofluorescence Mass spectrometry based methods

LC-MS/MS derived catalogue

Quantitative proteomics methods to show enrichment of proteins within different subcellular preparations

Prediction from sequence?LimitedStatic

Databases often carry contradictory assignmentsDifferent approaches can lead to conflicting data

....but can be highly complementary

FFP IF Pure SubtractiveMultiple locations ✔ ✔ x ✔

Sensitivity ✔ ✔ x xSpecificity ✔ variable ✔ ✔Hypothesis generating x x ✔ ✔

Hypothesis driven ✔ ✔ x xMultiplexing limited limited x ✔

No perturbation of protein x possibly ✔ ✔

Throughput x x ✔ ✔Cost x x ✔ ✔

high quality reagents ✔ ✔ x xUniversal approach limited x ✔ ✔Isoform specificity x ✔ ✔ ✔

Dynamic ✔ ✔ x ✔

Organelle “enrichment” rather than organelle “purification”

Purification strategies can result in contamination by other organelles and false-positive hits

Dynamic proteome: proteins in transit (cargo proteins) may be just passing through!

Purification of an organelle gives no information about steady state location of proteins with multiple localisations

Sampling the cell as a whole?

de Duve’s Principle (adapted)

Based on the principle that during analytical centrifugation, organelle structures will migrate until they reach their buoyant densities

Proteins from the same organelle will have identical distribution profiles through the gradient

Novel organelle residents can be assigned by matching their profiles to the distribution of known marker proteins

……..

Winner of the 1974 Nobel Prize in physiology or medicine for his discovery of the lysosome and the peroxisome.

PCP and LOPIT

or in fact any method that gives differential enrichment

Equilibrium density centrifugation

Organelle Fractionation Western blot

LOPIT WorkflowDensity gradient centrifugation Differential Centrifugation

Reporter ion intensities mimic the peptide distribution profiles

050000

100000150000200000250000300000350000400000450000500000

1 2 3 4 5 6

Protein X

TMT126 TMT127 TMT128 TMT129 TMT130 TMT131 P

Gatto et al., 2010

Steady State Position

Mixed Locations

LOPIT in a whole organism

Drosophila embryos

Tan et al (2009) J. Prot Res 8(6):2667-2678

Arabidopsis thaliana root derived callus

Nino NikolovskiPaul DupreeDenis Rubstov

Increased coverage by combining experiments

membrane + membrane associated

2205 proteins identified

1826 quantified in all replicates after imputation

163 Golgi proteins

320 ER266 PM

Nikolovski 2012 in press

Saccharomyces cerevisiae

Y. Wang and S. Oliver - unpublished

Comparison with GFP dataset (Huh et al, 2003) revealed good

overlap for some organelles, but not for PM or Golgi

>1500 proteins

Plasma membrane

Endoplasmic reticulumGolgi

Mitochondrion

Lysosome

Chicken DT40 cell line

Hall et al 2009

Tony JacksonStephanie HallMatthew Trotter

Combine with next slide

B C D

E F G

-4

-2

-8 -6 -4 -2 0

PC1

PC

2

ClathrinIgM

B-cell receptor and clathrin show average position away from plasma membrane cluster

IgM

IgM

clathrin

Rab4

B+C

E+F

-6

-4

-2

0

2

4

6

-8 -6 -4 -2 0 2 4 6 8

Hall et al 2009.

Dynamic system – in action

E14TG2a mouse embryonic stem cell line

Dppa5a

Sox2

Oct4Dppa4

Utf1

Mcl-1Tdgf-1 Risc

Fgf4LIF receptor

Alkaline phosphataseE-ras

β-catenin

Erk-2

Andy Christoforou

HEK293T Human cell line

Andy Christoforou

Comparison with Human Protein Atlas

Andy Christoforou

Organelle Fractionation

Western blot

Trypsinization and labelling

Combine

MS/MS

MSnBASE ? ?

LOPIT pipeline

Machine learning methods to allow greater data mining

Dynamics changes in locationPredict multiple locations

Gatto and Lilley , Bioinformatics 2012

Peroxisome

Proteasome

Ribosomal (60S) cluster

Nucleus

Ribosomal (40S) cluster

Cytoplasm

Original Dataset

Protein-organelle prediction with supervised KNN

Identification and assignment of

proteins to organelles with

phenoDisco

Drosophila embryos

PC

2P

C2

PC

2

PC1

PC1

PC1

Tan et al. J .of Proteome Res. (2009) 8(6):2667-78

TGN

ABC transporters

Ribosomal (40S) cluster

Ribosomal (60S) cluster

ER membrane associated

PC1

PC

2P

C2

PC1

PC1

PC

2

LOPIT on Arabidopsis

Original Dataset

Protein-organelle prediction with supervised KNN

Identification and assignment of

proteins to organelles with

phenoDisco

Dunkley et al. PNAS (2006) 103(17):6518-23

Chloroplast envelope

FFP IF Pure Subtractive LOPITMultiple locations ✔ ✔ x limited ✔

Sensitivity ✔ ✔ x x xSpecificity ✔ variable ✔ ✔ ✔Hypothesis generating x x ✔ ✔ ✔

Hypothesis driven ✔ ✔ x x xMultiplexing limited limited x ✔ ✔

No perturbation of protein x possibly ✔ ✔ ✔

Throughput x x ✔ ✔ ✔Cost x x ✔ ✔ ✔

high quality reagents ✔ ✔ x x xUniversal approach limited x ✔ ✔ ✔Isoform specificity x ✔ ✔ ✔ ✔

Dynamic ✔ ✔ x ✔ ✔

Summary

The ‘LOPITEERS’

Andy ChristoforouLaurent GattoArnoud GroenAdam GutteresClaire Mulvey

Dan NightingaleNino Nikolovski

Konstanze SchottPavel Shliaha Lisa Simpson

Matthew TrotterYuchong WangHoujiang Zhou

Cambridge Collaborators

Paul DupreeStephanie HallTony Jackson

Alfonso Martinez Arias

Ludovic Vallier

Dirk WaltherMatthias MannPeter James

Isaac Newton Trust