Relating Protein Abundance & mRNA Expression

Mark B GersteinYale (Comp. Bio. & Bioinformatics)

NIDA site visit at Yale

2007.12.19

[Greenbaum et al. Bioinformatics 2002, 18, 587.]

Why relate amounts of protein & mRNAGene expression - major place for regulation (easy to measure)

vs. Concentration of protein - major determinant of activity

where ks,i and kd,i are the protein synthesis and degradation rate constants

At steady state: Pi =ks;i [mRNAi ]

kd,i

= ks,i [mRNAi ] – kd,i Pi dPidt

Expectations from simple kinetic models:

Outliers from trend interesting

[Graphic: Jeong et al, Nature, 41:411]

Protein interaction networks

[Graphic: http://proton.chem.yale.edu]

Protein complexes

In protein complexes, one expects stoichiometric abundance of component proteins and that mRNA expression levels should be correlated with protein abundance

…Among pathways, this is expected to a lesser degree between interacting proteins

Relationship of Protein Abundance to Complexes and Pathways

mRNA expression levelsMicroarrays

AffymetrixPCR

SAGE

Protein abundance2D Gel Electrophoresis Multiple staining options

Small dynamic range

DIGE Cy3 vs. Cy5 labelingLarge dynamic range

ICAT, iTRAQ MS-basedRelative abundance (ratio of isotopically

labeled species)Large dynamic range

MudPIT LC-MS/MS

SILAC Stable isotope labeling with amino acids in cell culture - for MS analysis

TAP-Tag Weissman and O’Shea (Oct. 2003)

Sources of experimental data

[http://www.biology.ucsc.edu/mcd/images/microarray.gif]

PARE: proteomics.gersteinlab.org

Upload or use pre-loadedmRNA, protein datasets

Analyze all oranalyze MIPS or GO subset

[Yu et al., BMC Bioinfo. '07]

Open-source codeDownloadable

PARE: a web-based tool for correlating mRNA expression and protein abundance

PAREmain page

(1) Select mRNA, protein datasets: -use pre-loaded datasets -upload datasets

(2) Choose categorization method: -correlate all -MIPS complexes -GO biological processes -GO molecular function -GO cellular component

Select MIPS, GO subsets (opt.)

Display results (3) Display -Linear or log-log correlation for selected subset(s) -Tabulate data, correlation values for selected subset(s) -Label (on plot) and tabulate outlying datapoints

[Yu et al., BMC Bioinfo. '07]

Correlated data

Log-log plot of correlation -linear fit -outliers labeled

Calculation of mutual information

PARE output

[Yu

et a

l., B

MC

Bio

info

. '07

]

Correlation of subsets (GO, MIPS)Yeast ref. datasets: “Correlate all”vs.GO cellular component subsets for particular cellular locations

[Yu

et a

l., B

MC

Bio

info

. '07

]

PARE: pre-loaded datasets

[Yu et al., BMC Bioinfo. '07]

Connecting PARE with datasets from NIDA investigators

Protein abundance (iTRAQ datasets)

Mouse (Nairn lab) - samples from 3 brain regions: cortex, striatum, hippocampus

Green monkey (Taylor lab) - several brain regions -caudate, dlPFC, mPFC, NacC, NacS, PFC11, PFC13, PrCO, putameneach treated with saline, PCP, and cocaine

mRNA datasets obtained from expression database

Mouse - Sandberg et al. PNAS 2000, 97, 11038.

0

0.5

1

1.5

2

2.5

3

0 0.5 1 1.5 2 2.5 3mRNA (hippocampus/cortex)

pro

tein

(h

ipp

oca

mp

us/

cort

ex)

mRNA expression ratio (hippocampus/cortex)

pro

tein

ab

un

dan

ce r

atio

(h

ipp

oca

mp

us/

cort

ex)

Mouse brain: correlation of mRNA & protein expression

For 96 genes with differential expression in hippocampus vs. cortex

Plan to correlate abundance for individual pathways and complexes with C. Bruce ("John", talking later)

Acknowledgements

iTRAQ datasets:Angus Nairn, Erika Andrade, Dilja KruegerJane TaylorChris Colangelo, Mark Shifman (YPED)

PARE:Anne Burba, Eric Yu