Schwanhausser 2011 Global Quantification

8/12/2019 Schwanhausser 2011 Global Quantification

1/6

ARTICLE doi:10.1038/nature10098

Global quantification of mammalian gene

expression controlBjorn Schwanhausser1, Dorothea Busse1, Na Li1, Gunnar Dittmar1, Johannes Schuchhardt2, Jana Wolf1, Wei Chen1

& Matthias Selbach1

Gene expression is a multistep process that involves the transcription, translation and turnover of messenger RNAs andproteins. Although it is one of the most fundamental processes of life, the entire cascade has never been quantified on agenome-wide scale. Here we simultaneously measured absolute mRNA and protein abundance and turnover by parallelmetabolic pulse labelling for more than 5,000 genes in mammalian cells. Whereas mRNA and protein levels correlatedbetter than previously thought, corresponding half-lives showed no correlation. Using a quantitative model we haveobtained the first genome-scale prediction of synthesis rates of mRNAs andproteins.We findthat the cellular abundance

of proteins is predominantly controlled at the level of translation. Genes with similar combinations of mRNA and proteinstability shared functional properties, indicating that half-lives evolved under energetic and dynamic constraints.Quantitative information about all stages of gene expression provides a rich resource and helps to provide a greaterunderstanding of the underlying design principles.

The four fundamental cellular processes involved in gene expressionare transcription, mRNA degradation, translation and protein degra-dation. It is now clear that each step of this cascade is controlled bygene-regulatory events1,2. Although each individual process has beenintensively studied, littleis known about howthe combined effect of allregulatory eventsshapesgene expression. Thefundamentalquestionofhow genomic information is processed at different levels to obtain aspecific cellular proteome has therefore remained unanswered.

With regard to a quantitative description of gene expression,numerous previous studies comparing mRNA and protein levels con-cluded that the correlation is poor3,4. However, the available datasuffer from several limitations. Most studies are limited to a fewhundred genes, mainly due to the technical challenges involved inlarge-scale protein identification and quantification. Also, proteinlevels measured in one experiment are typically compared tomRNA levels determined in a different experiment performed at adifferent time in a different laboratory, making it difficult to interpretwhy the correlation is low. Finally, mRNA and protein levels resultfrom coupled processes of synthesis and degradation. Therefore, ana-lysis of mRNA and protein levels alone cannot provide sufficientinformation to understand gene expression comprehensively.mRNA and protein turnover can be measured with drugs to inhibit

transcription or translation5,6

, but this has severe side effects. Studiesbased on artificial fusion proteins areproblematic because taggingcanaffect protein stability7.

To overcome theselimitations we sought to quantifycellularmRNAand protein expression levels and turnover in parallel in a populationof unperturbed mammalian cells. Pulse labelling with radioactivenucleosides or amino acids is regarded as the gold standard methodto determine mRNA and protein half-lives. Recently, variants of thisapproach based on non-radioactive tracers have been established810.In stable isotope labelling by amino acids in cell culture (SILAC), cellsare cultivatedin a mediumcontaining heavy stable-isotope versions ofessential amino acids11. When non-labelled (that is, light) cells aretransferred to heavySILAC growthmedium,newlysynthesizedproteinsincorporate the heavy label while pre-existing proteins remain in the

light form. This strategy can be used to measure protein turnover1214 orrelative changes in protein translation15,16. Similarly, newly synthesizedRNA can be labelled with the nucleoside analogue 4-thiouridine (4sU).4sU-containing mRNA can be purified and compared with the pre-existing fraction to compute mRNA half-lives10.

Pulse labelling of proteins and mRNAs

We used parallel metabolic pulse labelling with amino acids and 4sU

to measure simultaneously protein and mRNA turnover in a popu-lation of exponentially growing non-synchronized NIH3T3 mousefibroblasts (Fig. 1a). Protein samples were collected at three timepoints, measured by liquid chromatography and online tandem massspectrometry (LC-MS/MS) and analysed with the MaxQuant soft-ware package17. We identified 84,676 peptide sequences and assignedthem to 6,445 uniqueproteins (falsediscovery rate,1% at thepeptideand protein level). A total of 5,279 of these proteins was quantified byat least three heavy to light (H/L) peptide ratios (Fig. 1b). Tissue-specific amino acid precursor pools and recycling rates, a pervasiveproblem forin vivopulse labelling experiments9,18,19, did not appre-ciably affect our results (Supplementary Fig. 1). For constant incorp-oration rates the logarithm of H/L ratios should increase linearly withtime (Fig. 1c). Ninety-three per cent of proteins showed excellent

linear correlation indicated by a variability of the linear regressionslope smaller than 1% (Fig. 1d). Protein abundance did not influenceH/Lratio measurements (Supplementary Fig. 2). In total, we obtaineda confident set of 5,028 protein half-lives calculated from the slope ofthe regression line. Cycloheximide chase experiments for selectedproteins spanning a representative range of half-lives agreed well withhalf-lives determined by pulsed labelling and mass spectrometry(Supplementary Fig. 3). In parallel, we pulse labelled newly synthe-sized RNA for 2 h with 4sU. RNA samples were fractionated into thenewly synthesized and pre-existing fractions. Both fractions and thetotal RNAsample were analysed by mRNA sequencing andquantifiedby mapping reads to their exonic region20. We calculated mRNA half-lives based on the ratios of newly synthesized RNA/total RNA ratioand the pre-existing RNA/total RNA10.

1Max Delbruck Center for Molecular Medicine, Robert-Rossle-Str. 10, D-13092 Berlin, Germany. 2MicroDiscovery GmbH, Marienburger Str. 1, D-10405 Berlin, Germany.

1 9 M A Y 2 0 1 1 | V O L 4 7 3 | N A T U R E | 3 3 7

Macmillan Publishers Limited. All rights reserved2011
http://www.nature.com/doifinder/10.1038/nature10098http://www.nature.com/doifinder/10.1038/nature10098


2/6

Proteins were, on average, five times more stable(median half-life of46 h) than mRNAs (9 h) andspanned a bigger dynamic range (Fig. 2a).Because very long (.200 h) and very short (,30 min) protein half-lives cannot be accurately quantified from our three time points, thetrue dynamic range of protein stabilities may be even higher. Notably,we found no correlation between protein and mRNA half-lives (Fig.2c,R25 0.02, loglog scale).

Absolute mRNA and protein copy numbersWe calculated absolute cellular mRNA copy numbers based on thenumber of sequencingreads in the unfractionatedsample in conjunction

with information on cellular mRNA content20. Absolute protein copynumbers can be inferred from mass spectrometry data21,22. To this end,we used the sum of peak intensities of all peptides matching to a specificprotein. When divided by the number of theoretically observable pep-tides, this value provides an accurate proxy for protein levels (intensity-based absolute quantification or iBAQ, see Supplementary Methods).

Levels of detected proteins spanned approximately five orders ofmagnitude (Fig. 2b). Relatively few proteins had less than 100 copiesper cell, indicating that some proteins of low abundance escapeddetection. Indeed, we observed a moderate detection bias (Sup-plementary Fig. 4) and therefore restricted our analysis to genes thatwere identified at both the mRNA and protein level. In this subset,proteins were, on average, ,900 times more abundant than corres-ponding transcripts. Despite a huge spread, mRNA and protein levelswere clearly correlated (Fig. 2d, R25 0.41, loglog scale). This cor-relation is considerably higher than in any previous study in mam-mals3,4,23. An attempt to improve this correlation further by nonlineartransformation resulted only in a marginal increase (R25 0.44,Supplementary Fig. 5). It seems that for our data set, this is aboutthe maximum correlation between mRNA and protein that can beachieved without additional information.

ReproducibilityTo investigate the experimental noise we performed a second inde-pendent large-scale experiment and measured mRNA and proteinlevels and half-lives again. The overall correlation of half-lives andlevels between both replicates was good (Supplementary Fig. 6 andSupplementary Table 1). Removing less-consistent data points didnot increase correlation between mRNA and protein levels or half-lives (Supplementary Fig. 7). Thus, noise has little impact on theobserved correlationbetween mRNAand protein levelsand half-lives.We also validated absolute mRNA and protein copy numbers usingindependent methods. For mRNA copy numbers we used theNanoString technology, which captures and counts individual tran-scripts without enzymatic reactions24. Correlation between sequen-cing and NanoString data was high (r5 0.79, see also Supplementary

Fig 8a). Absolute protein quantification was validated by spike-in

NewlysynthesizedproteinsH/L ratio

H

L

m/z

Pre-existingproteins

Proteins

SILAC heavy(t1,t2,t3)

SILAC light

400M 4sU (2 h)

mRNAs

Separation

RNA isolation and

biotinylation

Pre-existingRNA

Newly synthesizedRNA

Solexa sequencing

Withoutseparation

TotalRNA

a

Hist1h1c

(SEAAPAAPAAAPPAEK)

H/L ratio = 0.05

L L L

H

H

H

746 748 7500

20

40

60

80

100

746 748 750 7520

20

40

60

80

100

746 748 750 7520

20

40

60

80

100Hist1h1c

(SEAAPAAPAAAPPAEK)

H/L ratio = 0.19

Hist1h1c

(SEAAPAAPAAAPPAEK)H/L ratio = 0.63

Relativeintensity

b

c d

770 772 774 776 770 772 774 776 770 772 774 776

20

40

60

80

100

0

20

40

60

80

100

0

20

40

60

80

100

0

L

L

L

H

H H

Relativeintensity

z/mz/mz/m

t1

(1.5 h)

t1 (1.5 h)

t2

(4.5 h)

t2 (4.5 h)

t3

(13.5 h)

t3 (13.5 h)

Rrm2

(APTNPSVEDEPLLR)

H/L ratio = 0.24

Rrm2

(APTNPSVEDEPLLR)

H/L ratio = 1.26

Rrm2

(APTNPSVEDEPLLR)

H/L ratio = 12.8

z/mz/mz/m

ln(ratio+1)

0

0.5

1

1.5

2

2.5

t1

t2

t3

Harvesting time point

Hist1h1c

Rrm2

t1/2

= 4.5 h

R2= 0.99

t1/2

= 62.1 h

R2 = 0.99

0

200

400

600

10.01106 100104

Variability of linear regression slope (%)

Counts

93%

Intensity

Figure 1| Parallel quantificationof mRNA and protein turnover and levels.a, Mouse fibroblasts were pulse labelled with heavy amino acids (SILAC, left)and the nucleoside 4-thiouridine (4sU,right). Protein and mRNA turnoverwasquantified by mass spectrometry and next-generation sequencing, respectively.b, Mass spectra of peptides from a high- and low-turnover protein reveal

increasing heavy to light (H/L) ratios over time. c, Protein half-lives werecalculated from log H/L ratios at all three time points using linear regression.d, Variability of linear regression slopes assessed by leave-one-out cross-validation was small.

Counts

Average cellular half-life (h)

mRNAmedian: 9 h Protein

median: 46 h

1 10 100 1,000

1 10 100 1,000

1 10

10

100

100

104

104

105

105

106

106

107

107

1,000

1,000

0

200

400

600

800

1,000

0

200

400

600

800

1,000

Counts

mRNAmedian: 17

Proteinmedian: 16,000

Average copies per cell

mRNA copies per cell

Proteincopiespercell

1

10

100

1,000

100.5 101 101.5

mRNA half-life (h)

Proteinhalf-life(h)

a b

c d

R2 = 0.02 R2 = 0.41

Figure 2| mRNA and protein levels and half-lives. a,b, Histograms ofmRNA (blue) and protein (red) half-lives (a) and levels (b). Proteins were onaverage 5 times more stable and 900 times more abundant than mRNAs andspanned a higher dynamic range.c,d, Although mRNA and protein levelscorrelated significantly, correlation of half-lives was virtually absent.

RESEARCH ARTICLE

3 3 8 | N A T U R E | V O L 4 7 3 | 1 9 M A Y 2 0 1 1



3/6

experiments using a mixture of 48 proteins with known concentra-tions (Supplementary Fig. 8b). iBAQ values correlated well withknown absolute protein amounts over at least four orders of mag-nitude and had a higher precision and accuracy than alternative mea-sures of absolute protein abundance (data not shown)21,22. We alsoassessed degradation and synthesis rates for mRNAs and proteins byactinomycin D and cycloheximide treatment, respectively. For highturnover proteins and mRNAs we obtained results consistent with

pulse labelling data (Supplementary Fig. 8cf).

A quantitative model of gene expression

Our data allow us to calculate average synthesis rates of mRNAs andproteins for thousands of genes using a mathematical model (Fig. 3aand Supplementary Methods). The experimental data are based on apopulation of non-synchronized cells. Therefore, our estimated ratesprovide an average over the population and time.

Average cellular transcription rates predicted by the model spannedtwo orders of magnitude with a median of about twomRNAmoleculesperhour(Fig. 3b). An extreme example wasMdm2 with more than 500mRNAs per hour. A microscopic study on the cytomegalovirus (CMV)promoter reportedtranscription termination rates of 5.8 to 8.7 mRNAsper hour25. These values are above the median of our predictions, as

perhaps expected for a strong promoter system. Next, we calculatedtranslation rate constants; that is, how many proteins are made fromeach mRNA template per hour (Fig. 3c). We find a median translationrateconstant of about 40 proteinsper mRNA per hour. Several proteinsinvolved in translational regulationsuch as the translation initiationfactor eIF4G1, fragile X syndrome related protein Fxr2 and tuberinhad extremely low rate constants and were translationally repressed.Plotting translation rate constants against protein levels revealed thatabundant proteins are translated about 100 times more efficiently thanthose of low abundance(Fig.3d).Hence, differenttranslation efficienciescontribute to thehigherdynamic rangeof proteins compared to mRNAs(Fig. 2b). Intriguingly, translation rate constants saturated at around 180protein copies per mRNA per hour. To our knowledge, the maximal

translation rate constant in mammals is notknown.On thebasisof ref. 1,the estimated maximal translation rate constant in seaurchin embryosis 140 copies per mRNA per hour, which is surprisingly close to theprediction of our model.

Control of gene expression

A long-standing question is how much protein abundance is con-trolled at the transcriptional, post-transcriptional, translational and

post-translational levels. Until now, this has mainly been addressedindirectly by analysing mRNA and protein sequence features. Featuresrelated to translation initiation (for example, ShineDalgarno, Kozakand 39 untranslated region (UTR) sequences), elongation (for example,codon bias) and protein stability (for example, degrons) have been ana-lysed and reported to correlate partially with protein/mRNA ratios inbacteria, yeast and mammals23,26,27. We also observed sequence featurescharacteristic of mRNA and protein stability and found that mRNAswith long 39UTRs are, on average, less stable (Supplementary Fig. 9). Inaddition, the density of AU-rich elements and binding motifs of a spe-cific RNA-binding protein (pumilio 2) correlated negatively withmRNA stability (Supplementary Fig. 10). Highly structured proteinswere more stable than unstructured ones (Supplementary Fig. 11a).We also identified amino acids over-represented in unstable proteins

(Supplementary Fig. 11b).Sequence features are at best indirect proxies for mechanisms con-trolling protein abundance. How much efficiencies of differentstepsinthe gene expression cascade contribute to variance of cellular proteincopy numbers can only be revealed by direct parallel genome-scalemeasurements of mRNA and protein levels and half-lives which werenot available previously. In our data the coefficient of determination(R2) between mRNA and protein copy numbers is 0.41 (Fig. 2d).Assuming the absence of technical and biological noise, this meansthat ,40% of the variance in protein levels is explained by mRNAlevelsconsiderably more than previously thought (Fig. 4a). Most ofthis40% is dueto differenttranscription rates, whereas mRNA stabilityhas a smaller role. Considering translation rate constants markedlyboostsR2 to 0.95. Thus, translation rate constants have the dominant

role for control of protein levels. Unexpectedly, the impact of proteindegradation is rather small.In the above analysis the same experimental data were used to

calculate synthesisrates andto estimate their impact on proteinlevels.To avoid this over-fit and to assess reliability of the model predictionswe performedthe same analysis with data from the biological replicateexperiment. In the replicate the coefficient of determination betweenmRNA and protein levels was 0.37 (Fig. 4b). We then used the modelincluding the estimated parameters from the first experiment to pre-dict protein levels from mRNA levels in the replicate data. Predictedprotein levels agreed very well with measured protein levels(R25 0.85, Fig. 4c). Therefore, the model explains ,85% of the vari-ability in protein copy numbers in an independent experiment. Thecorrelation is very similar to the direct comparison of protein levels inboth experiments (R25 0.84, Supplementary Fig. 6d). We concludethat technical and biological noise in our data are low, and that themodel faithfully predicts protein levels from mRNA levels in mousefibroblasts. It also indicates that the estimated impact of transcription,mRNA stability, translation and protein stability on protein abund-ance is reproducible. We finally assessed how much of the efficienciesof the various steps in gene expression are retained in a different celltype and organism. To this end, we quantified mRNA and proteinabundance in the human breast cancer cell line MCF7 by RNA-seqand mass spectrometry, respectively. A total of 2,030 human genesfrom the MCF7 data set had orthologues in the mouse fibroblast data.We then used rates from the mouse fibroblast model to predict proteinlevels from mRNA levels in human breast cancer cells. In MCF7 cells,the model predicted ,60% of the variability in protein levels (Fig. 4a).Although the fraction explained by the model is smaller than in mouse

fibroblasts, this indicates that translation and degradation rates are to

a

vsr (mRNAs per hour)

Counts

0.1 1 10 100

ksp (proteins per mRNA per hour)

Counts

Protein copies per cell

b

c d

mRNA

ksp x [mRNA]

vsr

0.1 1 10 1,000100 104 100 1,000 104 106105 107

ksp

(proteinsperm

RNAperhour)

kdr x [mRNA]

kdp x [protein]protein

350

300

250

200150

100

50

0

250

200

150

100

50

0

0.1

1

10

104

100

103

Figure 3| Quantitative model of gene expression in growing cells.a, mRNAs are synthesized with the rate vsrand degraded with a rate constantkdr. Proteins are translated and degraded with rate constantskspand kdp,respectively.b, Calculated mRNA transcription rates show a uniformdistribution.c, Calculated translation rate constants are not uniform.d, Translation rate constants of abundant proteins saturate betweenapproximately 120 and 240 proteins per mRNA per hour. Red line shows thelocally weighted fit (Lowess). Dashed linesindicate 95% confidence intervals of

the Lowess maximum value calculated by bootstrapping.

ARTICLE RESEARCH

1 9 M A Y 2 0 1 1 | V O L 4 7 3 | N A T U R E | 3 3 9



4/6

someextent independent of thecelltype and conservedbetweenmouseand human. It is noticeable, however, that the drop in prediction ismainly due to the fact that the translation part of the model performsless well.

Half-lives and gene function

Degradation of proteins is critically involved in many cellular processesincluding cell-cycle progression, signal transduction and apoptosis2830.Similarly, mRNA stability is important for the temporal order of geneinduction10,31. Genes mayhave evolvedspecific combinations of mRNAand protein half-lives under functional constraints10,31,32. We thereforeasked if genes withspecificcombinations of mRNA andproteinstabilityhave distinct biological functions. We grouped genes according to theirhalf-lives and used gene ontology to find enriched biological processes(Fig. 5; see Supplementary Table 2 for a complete list).

Genes with stable mRNAs and stable proteins were enriched inconstitutive cellular processes like translation (that is, ribosomalproteins), respiration and central metabolism (glycolysis, citric acidcycle). Hence, many housekeeping genes tend to have stable mRNAsand proteins. In yeast energy costs keep transcription and translationrates under selective pressure33. We reasoned that energy constraintsmay explain why housekeeping genes tend to have stable mRNAs andproteins. On the basis of the model, we calculated the theoreticalenergy required to maintain cellular mRNA and protein levels byrecycling from their building blocks (nucleotide monophosphatesand amino acids, respectively) in terms of high energy phosphates.This is a conservative estimate as splicing, folding and transport arenot included. Protein synthesis consumes more than 90% of theenergy whereas less than 10% is needed for transcription. A total of20% of the proteins consumed 80% of the energy for translation(Pareto principle or 80/20 rule). Consistent with optimization underenergy constraints, abundant proteins were significantly more stable

than less abundant ones (Supplementary Fig. 12a, P, 10215,

Wilcoxon test). This is not necessarily expected because the overallcontribution of protein stability to protein levels is very small(Fig. 4a). In addition, abundant proteins were significantly shorter(Supplementary Fig. 12b). Shuffling protein half-lives and lengthsmarkedly increased theoretical energy consumption (SupplementaryFig.12c).Collectively,theseobservations indicatethat mammaliangeneexpression evolved under energy constraints.

The subset of genes with unstable mRNAs and proteins was stronglyenriched in transcriptionfactors, signalling genes,chromatin modifyingenzymes and genes with cell-cycle-specific functions (Fig. 5). Because

mRNAs and proteins are information carriers, their degradation can be

1

10

100

1,0

00

100.5 101 101.5

mRNA half-life (h)

Proteinhalf-life(h

)

ProteolysisPhosphorylationChemical homeostasisCellular cation homeostasisCell adhesionIntegrin-mediated signalling pathwayCellular iron ion homeostasisGlycogen metabolic processDefence responseRegulation of cell proliferationmRNA processingDephosphorylationtRNA processingRNA splicingncRNA processingRegulation of cytokine productionRibosome biogenesisRegulation of transcriptionTranscription

Cell cycleMitosisCell divisionChromatin modicationChromatin organizationTranslationGluconeogenesisSecondary metabolic processGlycolysisTricarboxylic acid cycleCellular respirationMonosaccharide metabolic processPurine nucleotide metabolic processOxidation reductionGeneration of precursor metabolites/energy

1.5 0StablemRNAs/

stableproteins

UnstablemRNAs/

stableproteins

StablemRNAs/

unstableproteins

UnstablemRNAs/

unstableproteins

StablemRNAs/

stableproteins

Unstable mRNAs/unstable proteins

Unstable mRNAs/stable proteins

Stable mRNAs/unstable proteins

z-transformed

log10P-value

1.5

Figure 5| Functional characteristics of genes with different mRNA and

protein half-lives. Genes were grouped according to their combination ofmRNA andprotein half-lives andanalysedfor enriched gene ontology terms.Aheat map of enrichment P-values reveals functional similarities of genes withsimilar combinations of half-lives.

a

Modeldata

NIH3T3

replicate M

CF7

Predictivepower(%)

mRNA transcription (vsr)

mRNA degradation (kdr)

mRNA levels

Protein translation (ksp)

Protein degradation (kdp)

Noise/variability

c

Proteincopiespercellreplicate

predictedfrommRNAlevelsreplicate

Protein copies per cell, replicate

R2= 0.85

102 103 104 105 106 1071 10 100 1,000

mRNA copies per cell, replicate

Proteincopiespercell,replicate

b

R2= 0.37

100

80

60

40

20

0

102

103

104

105

106

107

102

103

104

105

106

107

Figure 4| Impact of different rates and rate constants on proteinabundance. a, Protein levels are bestexplainedby translation rates, followedbytranscription rates. mRNA and protein stability is lessimportant (leftbar). b, Inthe replicate experiment mRNA levels explained 37% of protein levels inNIH3T3 cells (middle bar ina).c, The model explains 85% of variance inprotein levels from measured mRNA levels (middle bar ina). The mousefibroblast model has some predictive power for human orthologous genes inMCF7 cells (rightbar in a). Errorbars show 95% confidence intervalsestimatedby bootstrapping.

RESEARCH ARTICLE

3 4 0 | N A T U R E | V O L 4 7 3 | 1 9 M A Y 2 0 1 1



5/6


6/6

10. Friedel, C. C., Dolken, L., Ruzsics, Z., Koszinowski, U. H. & Zimmer, R. Conservedprinciples of mammalian transcriptional regulation revealed by RNA half-life.Nucleic Acids Res.37,e115 (2009).

11. Mann, M. Functional and quantitative proteomics using SILAC.Nature Rev. Mol.Cell Biol.7,952958 (2006).

12. Doherty,M. K.,Hammond,D. E.,Clague,M. J.,Gaskell,S. J.& Beynon, R.J. Turnoverof the human proteome: determination of protein intracellular stability bydynamic SILAC.J. Proteome Res.8, 104112 (2009).

13. Milner, E., Barnea, E., Beer, I. & Admon, A. The turnover kinetics of majorhistocompatibility complex peptides of human cancer cells.Mol. Cell. Proteomics5,357365 (2006).

14. Lam, Y. W., Lamond, A. I., Mann, M. & Andersen, J. S. Analysis of nucleolarproteindynamics reveals the nuclear degradation of ribosomal proteins.Curr. Biol.17,749760 (2007).

15. Schwanhausser,B., Gossen,M., Dittmar,G. & Selbach,M. Global analysisof cellularprotein translation by pulsed SILAC.Proteomics9,205209 (2009).

16. Selbach,M. etal. Widespread changesin proteinsynthesisinduced by microRNAs.Nature455,5863 (2008).

17. Cox, J. & Mann, M. MaxQuant enables high peptide identification rates,individualized p.p.b.-range mass accuracies and proteome-wide proteinquantification. Nature Biotechnol. 26,13671372 (2008).

18. Price, J. C., Guan, S.,Burlingame, A.,Prusiner, S. B. & Ghaemmaghami,S. Analysisof proteome dynamics in the mouse brain.Proc. Natl Acad. Sci. USA107,1450814513 (2010).

19. Wu, C. C., MacCoss, M. J., Howell, K. E., Matthews, D. E. & Yates, J. R. III. Metaboliclabelingof mammalianorganisms withstableisotopes forquantitative proteomicanalysis.Anal. Chem.76,49514959 (2004).

20. Mortazavi, A., Williams, B. A., McCue, K., Schaeffer, L. & Wold, B. Mapping andquantifying mammaliantranscriptomesby RNA-Seq.NatureMethods 5, 621628

(2008).21. Lu, P., Vogel, C., Wang, R., Yao, X. & Marcotte, E. M. Absolute protein expression

profiling estimates the relative contributions of transcriptional and translationalregulation. Nature Biotechnol.25,117124 (2007).

22. Malmstrom,J. et al.Proteome-wide cellular protein concentrations of the humanpathogenLeptospira interrogans.Nature460,762765 (2009).

23. Vogel, C. et al. Sequence signatures and mRNA concentration can explain two-thirds of protein abundance variation in a human cell line.Mol. Syst. Biol. 6,400(2010).

24. Geiss, G.K. et al.Direct multiplexed measurement of gene expression with color-coded probe pairs.Nature Biotechnol.26,317325 (2008).

25. Darzacq,X. etal.Invivo dynamicsof RNA polymerase II transcription.NatureStruct.Mol. Biol.14,796806 (2007).

26. Arava, Y.,Boas, F. E., Brown, P. O. & Herschlag, D. Dissecting eukaryotic translationand its control by ribosome density mapping.Nucleic Acids Res.33,24212432(2005).

27. Wu,G., Nie,L. & Zhang,W. Integrative analyses of posttranscriptional regulation intheyeastSaccharomyces cerevisiae using transcriptomicand proteomicdata. Curr.

Microbiol.57,1822 (2008).28. Kirkpatrick,D. S.,Denison, C. & Gygi, S. P. Weighing in on ubiquitin: theexpanding

role of mass-spectrometry-based proteomics. NatureCell Biol.7, 750757 (2005).29. Hershko, A. & Ciechanover, A. The ubiquitin system.Annu. Rev. Biochem.67,

425479 (1998).30. King, R. W., Deshaies, R. J., Peters, J. M. & Kirschner, M. W. How proteolysis drives

the cell cycle.Science274,16521659 (1996).31. Hao, S. & Baltimore,D. The stabilityof mRNA influencesthe temporal orderof the

induction of genes encoding inflammatory molecules.Nature Immunol.10,281288 (2009).

32. Legewie,S.,Herzel, H.,Westerhoff,H. V.& Bluthgen,N. Recurrentdesign patterns inthe feedback regulation of the mammalian signalling network. Mol. Syst. Biol. 4,190 (2008).

33. Wagner,A. Energy constraints on the evolution of geneexpression. Mol. Biol. Evol.22,13651374 (2005).

34. Pedraza, J. M. & Paulsson,J. Effects of molecular memory and bursting onfluctuations in gene expression.Science319,339343 (2008).

35. Mittal,N., Roy,N., Babu, M. M. & Janga, S.C. Dissectingthe expression dynamics ofRNA-binding proteins in posttranscriptional regulatory networks.Proc. Natl Acad.Sci. USA106,2030020305 (2009).

36. Hogan, D. J.,Riordan, D.P., Gerber,A. P., Herschlag, D. & Brown, P.O. DiverseRNA-

binding proteins interact with functionally related sets of RNAs, suggesting anextensive regulatory system. PLoS Biol.6,e255 (2008).

37. Hentze, M. W., Muckenthaler, M. U. & Andrews, N. C. Balancing acts: molecularcontrol of mammalian iron metabolism. Cell 117, 285297 (2004).

38. Ingolia,N. T., Ghaemmaghami,S., Newman, J. R. & Weissman, J. S. Genome-wideanalysisin vivo of translation with nucleotide resolution using ribosome profiling.Science324,218223 (2009).

39. Churchman, L. S. & Weissman, J. S. Nascent transcript sequencing visualizestranscription at nucleotide resolution.Nature469,368373 (2011).

40. Gebauer,F. & Hentze,M. W. Molecular mechanisms of translational control. NatureRev. Mol. Cell Biol.5,827835 (2004).

41. Sonenberg, N. & Hinnebusch, A. G. Regulation of translation initiation ineukaryotes: mechanisms and biological targets. Cell 136, 731745 (2009).

42. Lu, R. et al.Systems-level dynamic analyses of fate change in murine embryonicstem cells.Nature462, 358362 (2009).

43. Rosenfeld, N., Elowitz, M. B. & Alon, U. Negative autoregulation speeds theresponse times of transcription networks.J. Mol. Biol.323,785793 (2002).

44. Rabani, M. etal. Metaboliclabeling of RNAuncovers principles of RNA productionand degradation dynamics in mammalian cells.Nature Biotechnol. doi:10.1038/

nbt.1861 (24 April 2011).

Supplementary Informationis linked to the online version of the paper atwww.nature.com/nature.

AcknowledgementsWethankN. Rajewskyand L. Dolken for fruitful discussions andC. Sommer for technical assistance. M.S. and W.C. are supported by the HelmholtzAssociation,the GermanMinistryof Education andResearch (BMBF) andthe SenateofBerlin by funds aimed at establishing the Berlin Institute of Medical Systems Biology(BIMSB)(grant number 315362A). J.W.is supported by theForSys-programme of theGerman Ministry of Education and Research (grant number315289); D.B. by theHelmholtz Alliance on Systems Biology/MSBN; and N.L. by the China ScholarshipCouncil CSC.

Author ContributionsM.S.conceived, designed and supervised the experiments. B.S.performed wet-labexperiments, massspectrometry andproteomic dataanalysis. D.B.and J.W. developed and employed the mathematical model. N.L. performed RNA-seqexperiments.W.C. designedand supervised RNA-seq experiments. B.S., D.B., J.S., W.C.andM.S.analysed genome-widedata.G.D. helpedin cycloheximide chase experiments

anddata analysis.B.S.,D.B., J.S., J.W., W.C.and M.S.interpretedthe data.M.S. wrote themanuscript.

Author InformationSequences have been deposited in the Sequence Read Archiveunder accession codeSRA030871. Reprints and permissions information is availableat www.nature.com/reprints. The authors declare no competing financial interests.Readers are welcome to comment on the online version of this article atwww.nature.com/nature. Correspondence and requests for materials should beaddressed to J.W. ([email protected] , for mathematical modelling), W.C.([email protected], for transcriptomics) or M.S.([email protected] , for proteomics).

RESEARCH ARTICLE

3 4 2 | N A T U R E | V O L 4 7 3 | 1 9 M A Y 2 0 1 1
http://www.nature.com/naturehttp://www.nature.com/reprintshttp://www.nature.com/naturemailto:[email protected]:[email protected]:[email protected]:[email protected]:[email protected]:[email protected]://www.nature.com/naturehttp://www.nature.com/reprintshttp://www.nature.com/nature

Documents

Schwanhausser 2011 Global Quantification