Research ArticleInflammation

Integrated in vivo multiomics analysis identifies p21-activated kinase signaling as a driver of colitis

See allHide authors and affiliations

Science Signaling  27 Feb 2018:
Vol. 11, Issue 519, eaan3580
DOI: 10.1126/scisignal.aan3580
  • Fig. 1 Multiomics analyses of murine colitis.

    (A) Unsupervised clustering of microarray, mass spectrometry (MS), and phosphoproteomic mass spectrometry (pMS) data. Noninflamed (NI) samples are marked in blue in the dendrograms above the heat maps, whereas inflamed (Inf) samples are marked in red. (B) Probability distributions of Spearman correlations for each pairwise comparison between each data set. (C) Average fold changes in Lima1 (Eplin) RNA, protein, and phosphoprotein abundance between inflamed and noninflamed mouse colons. (D) Scatter plot of Eplin-Ser360 and total Eplin protein counts in individual samples. (E) Venn diagrams summarizing the unique and overlapping differential expression events between the RNA, MS, and pMS data sets. Species refers to RNAs, proteins, or phosphopeptides that were detected in each data set from the comparison. Expression events refer to the direct comparison between the data sets for a given species. (F) Cellular localization of representative phospho-signals that were increased in abundance in colitis. Left: The abundances in individual samples, as detected by MS, for Trim28 Ser473 (top) and Map3k3 Ser337 (bottom). **P < 0.01, ***P < 0.001 by unpaired t test. Middle and right: Immunohistochemical analysis of pTrim28 and pMap3k3 in noninflamed and inflamed colons. In all panels, data are from five control (noninflamed) samples and three inflamed samples.

  • Fig. 2 Differential RNA expression, differential protein abundance, and pathway analysis.

    (A) Heat maps of RNA and total protein measurements most strongly contributing to pathway enrichment scores (ES) of gene set enrichment algorithm (GSEA). (B) Venn diagrams summarizing the unique and overlapping positively and negatively correlated pathway enrichments in the RNA and MS data sets. (C) GSEA plots from the RNA and total protein MS data set.

  • Fig. 3 Differential regulation of RNA and protein.

    (A) Scatter plot of the fold change (inflamed versus noninflamed) in RNA expression plotted against the fold change in protein abundance for species that were present in both data sets. Colored dots represent extracellular matrix proteins (purple), acute-phase proteins (red), and neutrophil proteins (orange), with the arrowheads indicating genes that were further investigated. (B) Collagen expression in the RNA data and abundance in the protein data. Differential abundance of matrix metalloproteinases (MMPs) and tissue inhibitor of matrix metalloproteinases (TIMPs) in the MS data are indicated in the heat map inset with a key. (C) Tissue expression patterns of acute-phase (red) and neutrophil (orange) transcripts. Each gene was normalized to a maximum of 1.0, and all of the genes from each category were averaged to generate bars. (D) Induction of acute-phase RNA in the liver during inflammation. Bars represent the ratio of liver expression to colon expression for Orm1 and Fga. Data are from two inflamed and two noninflamed animals. Assays were performed in duplicate. *P < 0.05 by unpaired t test. P = 0.085 for Orm1. (E) Loss of gene expression in colonic neutrophils. The abundances of the RNAs for Camp and Elane from bone marrow neutrophils relative to those in neutrophils isolated from colon were determined. The plot represents log-transformed data of two inflamed animals, normalized to the smallest expression value for each gene. For both genes, P < 0.05 by paired t test. norm., normalized. (F) Model depicting RNA expression (R) and protein abundance (P) for acute-phase proteins (APP) and neutrophil proteins (represented by a polymorphonuclear cell symbol) in the colon and distant organ sites.

  • Fig. 4 Coexpression and co-abundance network landscapes of RNA, MS, and pMS measurements.

    (A) Correlation networks for RNA, MS, and pMS data sets. Nodes indicate genes or proteins and edges connect two genes or proteins if the Spearman correlation between the expression of two genes or abundance of the two proteins is greater than 0.9 or less than −0.9. (B) Two-step generalized topological overlap matrices (GTOM2) of the RNA, MS, and pMS data sets clustered by unsupervised hierarchical clustering. Correlations greater than 0.9 or less than −0.9 were set to 1 (yellow), and all others were set to 0 (blue). Square regions indicate highly connected clusters of genes or proteins. (C) Plot of the gap statistic versus the number of clusters in each data set. Clustering cutoff points are marked with a star for each data set based on the gap statistic and GTOM2 topology. (D) Network visualization of module overlap. Nodes indicate particular modules; edges are present if there was statistically significant overlap in genes or proteins between the two modules (Fisher exact test, P < 0.05; FDR, q < 0.25). (E) YourCrosstalker network module for pMS module 3. Nodes are colored by differential phosphorylation status in inflamed relative to uninflamed colons (red, hyperphosphorylated; blue, dephosphorylated; Wilcoxon-Mann-Whitney, P < 0.05; FDR, q < 0.25), and edges are colored by pathway membership of the interaction. Gray edges indicate protein-protein interactions that exist but are not part of an enriched pathway in the subnetwork. Pathways with higher statistical significance determine the interaction pathway association for interactions in multiple pathways. Striped nodes were recruited by the algorithm during the random walk procedure as significantly traversed “crosstalker” nodes. (F) YourCrosstalker network for MS module 5. Nodes are colored by total protein differential abundance status in inflamed relative to uninflamed mice (red, increased abundance; blue, decreased abundance; Wilcoxon-Mann-Whitney, P < 0.05; FDR, q < 0.25). Edges and striped nodes are defined as in (E).

  • Fig. 5 Inferring kinase activity from pMS measurements.

    (A) Volcano plot of normalized enrichment score (NES) versus false discovery rate (FDR). Kinases with positive or negative enrichment and an FDR <0.25 are specified. (B) Heat maps of phosphopeptides corresponding to known kinase substrates from noninflamed (NI) and inflamed (Inf) animals. The kinase is indicated to the left of each set of substrates together with the log2 differential abundance (inflamed versus noninflamed) for RNA (left box), protein (right box), and phosphorylation (circles). NES is specified for each kinase. All of the kinases shown had FDR <0.25 and are predicted to be either increased (positive NES) or decreased (negative NES) in abundance in colitis. (C) Validation of Mapk14 and Gsk3α/β phosphorylation in colon samples from inflamed and noninflamed animals. Fluorescence intensity (FI) was measured using Luminex assays specific to each phosphorylation site. Data are from samples from 12 individual noninflamed animals and 25 individual inflamed animals. *P < 0.05 and **P < 0.01 by unpaired t test.

  • Fig. 6 Overlap between mouse model omics data and human inflammatory bowel disease biopsy transcripts.

    (A) Venn diagrams representing the differential expression analysis of human inflammatory bowel disease (IBD) colonic biopsies in inflamed and uninflamed phenotypes (Wilcoxon-Mann-Whitney, P < 0.05; FDR, q < 0.25) compared to differentially expressed RNA, protein abundance, and phosphopeptide abundance between inflamed and uninflamed mouse colons. (B) Differentially expressed genes in the PAK signaling network neighborhood in human IBD. Genes are colored by differential expression direction (red, increased; blue, decreased) in inflamed relative to uninflamed human colonic biopsies. (C) Assessment of the overlap between the genes regulated by kinases that were statistically significantly associated with colitis in the mouse and human genes differentially expressed in the kinase-regulated network (hypergeometric test, P < 0.05).

  • Fig. 7 Validation of Pak as a therapeutic target in colitis.

    (A) Validation of Pak activation in the colons of animals with induced colitis (inflamed). Phosphorylated Pak1 and Pak2 were detected by Western blotting. Each lane represents a sample from a different animal. (B) Inhibition of Pak activity by FRAX597. Merlin phosphorylation on Ser518, a Pak substrate, was assessed by Western blotting analysis of colon samples from inflamed animals that had been treated for 24 hours with FRAX597 (100 mg/kg single dose) or polyethylene glycol and polyvinylpyrrolidone (vehicle). Gapdh, glyceraldehyde phosphate dehydrogenase. (C) Colonoscopic monitoring of colitis. Colonoscopy images of a representative mouse with adoptive transfer–induced colitis. The mouse was imaged before and 7 days after treatment with FRAX597 (100 mg/kg per day). (D) Histological effects of FRAX597 on the colons of mice with induced colitis. (E) Immunological effects of FRAX597 on the colons of mice with induced colitis. The percentages of macrophages and neutrophils were quantified by flow cytometry in the colons of inflamed animals after 7 days of treatment with FRAX597 or vehicle. *P < 0.05 by one-tailed Mann-Whitney test. Data are from three vehicle-treated animals and four FRAX597-treated animals.

Supplementary Materials

  • www.sciencesignaling.org/cgi/content/full/11/519/eaan3580/DC1

    Fig. S1. Histological analysis of colon samples.

    Fig. S2. Schematic of the experimental design.

    Fig. S3. Functional annotation of phosphorylation sites for mammalian proteins.

    Fig. S4. Human PAK1 and PAK2 signaling network neighborhood.

    Fig. S5. Pak activation during acute colitis.

    Table S1. Affymetrix microarray quantification of gene expression in individual samples.

    Table S2. MS-based quantification of proteins in individual samples.

    Table S3. MS-based quantification of phosphopeptides in individual samples.

    Table S4. Differential expression analysis for each data set.

    Table S5. Pathway analysis for each data set.

    Table S6. Modules from trans-omics coexpression network analysis.

    Table S7. Statistics for trans-omics coexpression network analysis.

    Table S8. Phosphosite lists for each kinase used in GSEA.

    Table S9. Statistics for GSEA-based kinase enrichment.

  • Supplementary Materials for:

    Integrated in vivo multiomics analysis identifies p21-activated kinase signaling as a driver of colitis

    Jesse Lyons, Douglas K. Brubaker, Phaedra C. Ghazi, Katherine R. Baldwin, Amanda Edwards, Myriam Boukhali, Samantha Dale Strasser, Lucia Suarez-Lopez, Yi-Jang Lin, Vijay Yajnik, Joseph L. Kissil, Wilhelm Haas, Douglas A. Lauffenburger, Kevin M. Haigis*

    *Corresponding author. Email: khaigis{at}bidmc.harvard.edu

    This PDF file includes:

    • Fig. S1. Histological analysis of colon samples.
    • Fig. S2. Schematic of the experimental design.
    • Fig. S3. Functional annotation of phosphorylation sites for mammalian proteins.
    • Fig. S4. Human PAK1 and PAK2 signaling network neighborhood.
    • Fig. S5. Pak activation during acute colitis.
    • Legends for tables S1 to S9

    [Download PDF]

    Other Supplementary Material for this manuscript includes the following:

    • Table S1 (Microsoft Excel format). Affymetrix microarray quantification of gene expression in individual samples.
    • Table S2 (Microsoft Excel format). MS-based quantification of proteins in individual samples.
    • Table S3 (Microsoft Excel format). MS-based quantification of phosphopeptides in individual samples.
    • Table S4 (Microsoft Excel format). Differential expression analysis for each data set.
    • Table S5 (Microsoft Excel format). Pathway analysis for each data set.
    • Table S6 (Microsoft Excel format). Modules from trans-omics coexpression network analysis.
    • Table S7 (Microsoft Excel format). Statistics for trans-omics coexpression network analysis.
    • Table S8 (Microsoft Excel format). Phosphosite lists for each kinase used in GSEA.
    • Table S9 (Microsoft Excel format). Statistics for GSEA-based kinase enrichment.

    © 2018 American Association for the Advancement of Science

Stay Connected to Science Signaling

Navigate This Article