A tissue-specific atlas of mouse protein phosphorylation and expression

Abstract

Although most tissues in an organism are genetically identical, the biochemistry of each is optimized to fulfill its unique physiological roles, with important consequences for human health and disease. Each tissue's unique physiology requires tightly regulated gene and protein expression coordinated by specialized, phosphorylation-dependent intracellular signaling. To better understand the role of phosphorylation in maintenance of physiological differences among tissues, we performed proteomic and phosphoproteomic characterizations of nine mouse tissues. We identified 12,039 proteins, including 6296 phosphoproteins harboring nearly 36,000 phosphorylation sites. Comparing protein abundances and phosphorylation levels revealed specialized, interconnected phosphorylation networks within each tissue while suggesting that many proteins are regulated by phosphorylation independently of their expression. Our data suggest that the "typical" phosphoprotein is widely expressed yet displays variable, often tissue-specific phosphorylation that tunes protein activity to the specific needs of each tissue. We offer this dataset as an online resource for the biological research community.

Figure 1. Overview of Tissue-specific Phosphorylation

A. Numbers of phosphopeptides, sites, and phosphoproteins. B. Numbers of localized and non-localized phosphorylation sites. C. Relative frequencies of Ser, Thr, and Tyr within all phosphoproteins and their relative likelihood of phosphorylation. D. Histogram depicting numbers of sites observed per protein. E. Box plots indicating the relative extents of modification for residues prone to phosphorylation within all observed phosphoproteins. F. Histogram indicating numbers of sites detected in multiple tissues. G. Distribution of tissue-specific sites across organs. H. Hierarchical clustering of sites and tissues based on spectral counts. Only the 11,414 sites containing 5 or more spectral counts in at least one tissue are shown. See also Figures S1, S3, S4 and S7 and Tables S1–S3.

Figure 2. Distribution of Phosphorylation Site Classes across Tissues and Phosphoproteins

Phosphorylation sites were classified as acidic, basic, proline-directed, tyrosine, or other as described (Villen et al., 2007). A. The relative frequencies with which each class is observed overall and for each tissue are plotted as pie charts. P-values reflect the likelihood that the tissue-specific sites were drawn randomly from a population with frequencies matching the entire dataset (χ² Test). B. The heat map presents the numbers of sites of each class observed for 6,296 phosphoproteins. Proteins and site classes have been clustered to highlight similarities. C. Histogram indicating proportions of phosphoproteins containing phosphorylation sites from variable numbers of classes. D. Bar graph indicating relative proportions of tissue-specific, moderate, and global phosphorylation sites in each class. E. Heat maps indicating the tissue distributions and relative abundances of phosphorylation sites along the lengths of proteins Mark1 and Dennd1a. Sites are labeled according to their classes: A = acidic; B = basic; O = other; P = proline-directed; T = tyrosine. See also Figures S3, S4 and S7 and Tables S1–S3.

Figure 3. Cross-Tissue Comparison of Protein and Phosphoprotein Expression

Overlap among phosphoproteins and proteins, overall (A) and in each tissue (B). C. Clustering of proteins based on spectral counts with and without phosphopeptide enrichment. Columns represent 12,000 observed proteins, while rows represent tissues with or without phosphopeptide enrichment. D. Phosphorylated and non-phosphorylated abundance profiles for selected proteins. E. Bar charts reflecting proportions of global, moderate, and tissue-specific proteins identified in non-phosphorylated form. F. Western blotting confirms spectral counting quantification of proteins and phosphorylation sites (Figure S4). G. Heat maps depicting spectral counts observed across tissues for sites along the length of each protein, revealing variable phosphorylation within proteins (Figure S3F). Abundances for each non-phosphorylated protein are also displayed. See https://gygi.med.harvard.edu/phosphomouse for plots of all proteins. See also Figures S2, S4–S6, S7 and Tables S1–S4).

Figure 4. Mapping Protein Expression and Phosphorylation Data onto Protein Interaction Networks

Protein observations with and without phosphopeptide enrichment were mapped onto the mouse STRING database of known protein interactions (Jensen et al., 2009). Only high confidence interactions (score > 0.7) were considered. Proteins that interact with Syk, Vamp, and Bad are depicted in networks above. Labeled diagrams are provided on the left that indicate the identities of all proteins. Each network is displayed twice for each tissue: the network on the left reflects phosphorylation, while the network on the right depicts protein expression. Proteins detected in phosphorylated or nonphosphorylated form in each tissue are represented as colored nodes, with colored edges connecting detectable proteins. Interacting proteins within the STRING database that were not detected in each tissue are shown in gray. See also Tables S1–S4.

Figure 5. Integrating Phosphorylation Data with Known Signaling Pathways

Shown above are all members of the MAPK pathway as recorded in KEGG (Kanehisa et al. 2010). Each protein is represented as a node containing up to nine squares, each indicating presence of phosphorylation in one tissue. Edges represent interactions among proteins in the MAP Kinase pathway and are colored when their proteins were phosporylated in each tissue. See also Tables S1–S4.