Skip to main page content
Access keys NCBI Homepage MyNCBI Homepage Main Content Main Navigation
, 13 (5), e1005535

Post-transcriptional Regulation Across Human Tissues


Post-transcriptional Regulation Across Human Tissues

Alexander Franks et al. PLoS Comput Biol.


Transcriptional and post-transcriptional regulation shape tissue-type-specific proteomes, but their relative contributions remain contested. Estimates of the factors determining protein levels in human tissues do not distinguish between (i) the factors determining the variability between the abundances of different proteins, i.e., mean-level-variability and, (ii) the factors determining the physiological variability of the same protein across different tissue types, i.e., across-tissues variability. We sought to estimate the contribution of transcript levels to these two orthogonal sources of variability, and found that scaled mRNA levels can account for most of the mean-level-variability but not necessarily for across-tissues variability. The reliable quantification of the latter estimate is limited by substantial measurement noise. However, protein-to-mRNA ratios exhibit substantial across-tissues variability that is functionally concerted and reproducible across different datasets, suggesting extensive post-transcriptional regulation. These results caution against estimating protein fold-changes from mRNA fold-changes between different cell-types, and highlight the contribution of post-transcriptional regulation to shaping tissue-type-specific proteomes.

Conflict of interest statement

The authors have declared that no competing interests exist.


Fig 1
Fig 1. The fraction of total protein variance explained by scaled mRNA levels is not informative about the across-tissues variance explained by scaled mRNA levels.
(a) mRNA levels correlate with measured protein levels (RT = 0.33 over all measured mRNAs and proteins across 12 different tissues). (b) Protein levels versus mRNA levels scaled by the median protein-to-mRNA ratio (PTR); the only change from panel (a) is the scaling of mRNAs, which considerably improves the correlation. (c) A subset of 100 genes are used to illustrate an example Simpson’s paradox: regression lines reflect within-gene and across-tissues variability. Despite the fact that the overall correlation between scaled mRNA and measured protein levels is large and positive RT = 0.89, for any single gene in this set, mRNA levels scaled by the median PTR ratio are not correlated to the corresponding measured protein levels (RP ≈ 0). (d) Cumulative distributions of across-tissues scaled mRNA-protein correlations (RP) for 3 datasets [–22]. The smooth curves correspond to all quantified proteins by shotgun proteomics while the dashed curves correspond to a subset of proteins quantified in a small targeted dataset [22]. The vertical lines show the corresponding overall (conflated) correlation between scaled mRNA levels and protein levels, RT. See Methods and S1 Fig.
Fig 2
Fig 2. Data reliability crucially influences estimates of transcriptional and post-transcriptional regulation across-tissues.
(a) The within-study reliability—defined as the fraction of the measured variance due to the signal—of relative mRNA levels is estimated as the correlation between the mRNA levels measured in the twelve different tissues. Estimates for the levels of each transcript measured in different subjects were correlated (averaging across the 12 tissue-types) and the results for all analyzed transcripts displayed as a distribution for each RNA dataset [29, 30]. (b) The within-study reliability of relative protein levels is estimated as the correlation between the protein levels measured in 12 different tissues [20, 21]. Within each dataset, separate estimates for each protein were derived from non-overlapping sets of peptides and were correlated (averaging across the 12 tissue-types) and the results for all analyzed proteins displayed as a distribution; see Methods. (c) The across-study reliability of mRNA was estimated by correlating estimates as in (a) but these estimates came from different studies [29] and [30]. (d) The across-study reliability of proteins was estimated by correlating estimates as in (b) but these estimates came from different studies [20] and [21]. (e) The fraction of across-tissues protein variance that can be explained by mRNA levels is plotted as a function of the reliability of the estimates of mRNA and protein levels, given an empirical mRNA/protein correlation of 0.29. The red Xs correspond to two estimates of reliability of the mRNA and protein measurements computed from both independent mRNA and protein datasets.
Fig 3
Fig 3. Concerted variability in the relative protein-to-RNA (rPTR) ratio of functional gene-sets across tissue-types.
(a) mRNAs coding for the ribosomal proteins, NADH dehydrogenase and respiratory proteins have higher protein-to-mRNA ratios in kidney as compared to the median across the other 11 tissues (FDR < 1%). In contrast mRNAs genes functioning in Rac GTPase binding have lower protein-to-mRNA ratios (FDR < 1%). (b) The stomach also shows significant rPTR variation, with low rPTR for the ribosomal proteins and high rPTR for tRNA-aminoacylation (FDR < 1%). (c) Summary of rPTR variability, as depicted in panel (a-b), across all tissues and many gene ontology (GO) terms. Metabolic pathways and functional gene-sets that show statistically significant (FDR < 1%) variability in the relative protein-to-mRNA ratios across the 12 tissue types. All data are displayed on a log10 scale, and functionally related gene-sets are marked with the same color. (d) The reproducibility of rPTR estimates across estimates from different studies is estimated as the correlation between the median rPTRs for GO terms showing significant enrichment as shown in panels (a-c). See Methods, S2 and S3 Figs.
Fig 4
Fig 4. Deriving a consensus protein dataset for improved quantification of human tissue proteomes.
We compiled a consensus protein dataset by merging data from [20] and [21] as described in Methods. The relative protein levels estimated from [20, 21], and the consensus dataset were correlated to mRNA levels from [30] (a) or to mRNA levels from [29] (b). The correlations are shown as a function of the median correlation between protein estimates from [20] and [21]. The consensus dataset exhibits the highest correlations, suggesting that it has averaged out some of the noise in each dataset and provides a more reliable quantification of of human tissue proteomes. (c) The datasets from [20], from [21], and the consensus dataset were evaluated by comparison to a targeted MS validation dataset quantifying 33 proteins over 5 tissues [22]. The similarity for each dataset was quantified by the mean squared error (MSE) relative to the targeted MS validation data using 124 protein/tissue measurements that were observed in all datasets. The MSEs are reported for each of the five tissues and for all 5 tissues combined; they indicate that the consensus data have the best agreement with the validation dataset.

Similar articles

See all similar articles

Cited by 30 PubMed Central articles

See all "Cited by" articles


    1. Sørlie T, Perou CM, Tibshirani R, Aas T, Geisler S, Johnsen H, et al. Gene expression patterns of breast carcinomas distinguish tumor subclasses with clinical implications. Proceedings of the National Academy of Sciences. 2001;98(19):10869–10874. 10.1073/pnas.191367098 - DOI - PMC - PubMed
    1. Slavov N, Dawson KA. Correlation signature of the macroscopic states of the gene regulatory network in cancer. Proceedings of the National Academy of Sciences. 2009;106(11):4079–4084. 10.1073/pnas.0810803106 - DOI - PMC - PubMed
    1. Spellman PT, Sherlock G, Zhang MQ, Iyer VR, Anders K, Eisen MB, et al. Comprehensive identification of cell cycle–regulated genes of the yeast Saccharomyces cerevisiae by microarray hybridization. Molecular biology of the cell. 1998;9(12):3273–3297. 10.1091/mbc.9.12.3273 - DOI - PMC - PubMed
    1. Slavov N, Macinskas J, Caudy A, Botstein D. Metabolic cycling without cell division cycling in respiring yeast. Proceedings of the National Academy of Sciences of the United Statesof America. 2011;108(47):19090–19095. 10.1073/pnas.1116998108 - DOI - PMC - PubMed
    1. Slavov N, Airoldi EM, van Oudenaarden A, Botstein D. A conserved cell growth cycle can account for the environmental stress responses of divergent eukaryotes. Molecular Biology of the Cell. 2012;23(10):1986–1997. 10.1091/mbc.E11-11-0961 - DOI - PMC - PubMed

Publication types