Quantitative maps of protein phosphorylation sites across 14 different rat organs and tissues

Abstract

Deregulated cellular signalling is a common hallmark of disease, and delineating tissue phosphoproteomes is key to unravelling the underlying mechanisms. Here we present the broadest tissue catalogue of phosphoproteins to date, covering 31,480 phosphorylation sites on 7,280 proteins quantified across 14 rat organs and tissues. We provide the data set as an easily accessible resource via a web-based database, the CPR PTM Resource. A major fraction of the presented phosphorylation sites are tissue-specific and modulate protein interaction networks that are essential for the function of individual organs. For skeletal muscle, we find that phosphotyrosines are over-represented, which is mainly due to proteins involved in glycogenolysis and muscle contraction, a finding we validate in human skeletal muscle biopsies. Tyrosine phosphorylation is involved in both skeletal and cardiac muscle contraction, whereas glycogenolytic enzymes are tyrosine phosphorylated in skeletal muscle but not in the liver. The presented phosphoproteomic method is simple and rapid, making it applicable for screening of diseased tissue samples.

Figure 1. Workflow for phosphoproteome analysis of rat tissue.

A total of 14 different tissues were isolated from four male rats, followed by snap freezing, homogenization and solubilization of the tissues. For each tissue, 10 mg of protein was subjected to tryptic digestion, succeeded by duplicate steps of phosphopeptide enrichments using titanium dioxide beads. Both enrichments were analysed by high-resolution LC-MS/MS yielding a total of 31,480 phosphorylation sites.

Figure 2. Tissue distribution of phosphoproteins.

(a) Hierarchical clustering of phosphoproteins and tissues based on label-free quantification of protein intensities from both enrichment steps (1 and 2). Low-intensity phosphoproteins are depicted in blue and high-intensity phosphoproteins are depicted in yellow. Red boxes highlight clusters of phosphoproteins that are either specific to a certain tissue or present in all tissues. (b) Histograms depicting the total number of phosphoproteins identified from each of the two enrichment steps (shown in blue) as well as the total number of phosphoproteins identified in each tissue obtained from merging the data from the two enrichment steps (shown in grey).

Figure 3. Blood-specific phosphoprotein network.

Protein interaction network based on phosphoproteins specifically identified in blood (green nodes) and expanded to include direct interaction partners (grey nodes). Protein clusters obviously relevant to blood biology, such as coagulation, Kell blood group glycoprotein complex, haemoglobin and haem biosynthesis, inflammatory responses, immune regulation and albumin-mediated transport, are highlighted. The network is available as a Cytoscape session file including accession numbers and protein names for all proteins in the network at http://cpr1.sund.ku.dk/cgi-bin/PTM.pl.

Figure 4. Tissue specificity of phosphoproteins.

(a) Histogram depicting the number of phosphoproteins identified in 1 to 14 tissues. (b) Phosphoproteins identified in one tissue are classified as tissue specific, whereas those identified in 14 tissues are classified as global. GO term analysis were made for tissue specific as well as global phosphoproteins. The P-values for over-representation in either of the two categories were calculated with the Wilcoxin–Mann–Whitney test and a Benjamini–Hochberg false discovery rate test was applied to account for multiple testing. Enriched GO terms for cellular component (CC), molecular function (MF) and biological process (BP) are shown. The numbers next to each bar indicate how many proteins form basis for any given term, and how many of those are found among the enriched fraction.

Figure 5. Tissue distribution and amino-acid sequence features of localized phosphorylation sites.

(a) Histograms depicting the percentages of serine, threonine and tyrosine phosphorylation sites identified in each rat tissue as well as in human skeletal muscle samples. The total number of S, T and Y phosphorylation sites across all tissues and their relative abundance in percent is stated above each histogram. (b) Heat map visualizing relative abundance of sequence motifs matching the indicated kinases from phosphopeptides identified in all tissues investigated. (c) Top: amino-acid sequence patterns for all S and T phosphorylation sites in brain and testis. Bottom: amino-acid sequence patterns for tissue-specific versus non-specific S and T phosphorylation sites for brain and testis. (d) Hierarchical cluster of pathway analysis of phosphotyrosine-containing proteins. (e) Protein–protein interaction network build from tyrosine-phosphorylated proteins identified in human skeletal muscle using InWeb shows that these proteins significantly interact with each other (Adj. P=2e-4, using a permuation test). The resulting network shows that tyrosine-phosphorylated proteins collaborate in muscle contraction, oxygen transport and cell proliferation to carry out physiological processes relevant to the tissue in question. The input proteins from human skeletal muscle are depicted as yellow spheres, whereas interacting proteins reported in the literature are depicted as grey spheres.

Figure 6. Muscle and liver glycogenolysis.

The chemical pathway for glycogenolysis is shown as are the phosphorylated enzymes involved in each of the steps identified in rat or human skeletal muscle (left) or in liver (right). The protein isoforms with the greatest number of identified phosphorylation sites were chosen for visualization. The number of phosphorylation sites on each of the proteins is indicated by small circles that are colour coded according to the amino acid phosphorylated with yellow for tyrosine, purple for threonine and blue for serine. The normalized protein intensity count is provided next to each protein.

Figure 7. Proline-directed phosphorylation.

The fraction of localized phosphorylation sites that harbour a proline residue at the adjacent position +1 for six different cellular compartments (a) and for protein kinases and transcription factors (b) are depicted as histograms. The number of identified phosphorylation sites with a proline at position +1 for each category is indicated above the bars. Students t-test, ***P>0.001. (c) Amino-acid sequence pattern for phosphorylation sites on nuclear proteins with a proline at position +1 versus phosphorylation sites with a proline at position +1 on all non-nuclear proteins. (d) Amino-acid sequence pattern for phosphorylation sites on transcription factors compared with all other phosphorylation sites. ER, endoplasmic reticulum.