Fig 1.
Sequential enrichment of post translational modification proteomics (SEPTM-proteomics) procedure. (A) Principle of SEPTM peptide enrichment; modification-specific antibodies pull down specific PTMs sequentially. (B) Proteomics data generation procedure and drug-cell line combinations analyzed. (C) Graphical outline of workflow. PTMs from PTM proteomics were subjected to t-SNE-based clustering to create the CCCN; a cluster-filtered network (CFN) was created by filtering PPI interactions to exclude all interactions save those from proteins with co-clustered PTMs. Similarly, relationships between BioPlanet pathways were defined by the extent to which their member proteins have PTMs that co-cluster in the CCCN. Namely, two pathways, A and B, have potential crosstalk if protein(s) from Pathway A and protein(s) from Pathway B have PTMs that co-cluster.
Fig 2.
Pathway crosstalk networks with the EGF/EGFR signaling pathway.
(A) Three pathways linked by strong PTM cluster weight (purple edges). Note that the EGF/EGFR signaling pathway has no genes in common with Glycolysis and gluconeogenesis and Transmembrane transport of small molecules, but the latter two pathways have 11 genes in common (green edges represent pathway Jaccard similarity). (B) Nearest neighbors of pathways in A (★) with additional edges filtered to show only strong associations (PTM cluster weight > 0.065 and/or pathway Jaccard similarity > 0.5).
Fig 3.
EGFR signaling and small molecule transport pathway interactions.
(A) Combined CFN/CCCN showing composite shortest paths from the Bioplanet pathways EGF/EGFR signaling pathway (top row) and Transmembrane transport of small molecules (bottom row). PTM clusters are apparent as cliques connected by yellow correlation edges. (B) same as A but showing CFN edges only. (C) “Mutual friends” (center row) defined as proteins that connect to at least one member of each pathway in the CFN. (D) CFN interactions between RTKs, SFKs and the Transmembrane transport of small molecules pathway. (E) Direct CFN interactions between members of the EGF/EGFR signaling and Transmembrane transport of small molecules pathways. PTMs that were significantly changed (>2.25-fold) in response to erlotinib are shown. Node size and color represents the median log2 fold change in response to erlotinib (A, B, C, E), or dasatinib (D); color scale bar shown in C. Node border and shape and edge colors are defined in S2C Fig.
Fig 4.
EGFR signaling and glycolysis pathway interactions.
(A) Direct CFN connections between proteins in the EGFR signaling and glycolysis pathways and their PTMs that are significantly changed by erlotinib (median fold change in erlotinib-treated vs. control cells at least 2.25). (B) CFN connections between RTKs and SFKs and Glycoylsis and gluconeogenesis pathway proteins. (C, D) Sub-network from EGFR-glycolysis shortest paths CFN (see S10B Fig) showing links between EGFR and top proteins with dasatinib-inhibited PTMs. Node shape and edge color are defined in S2C Fig. Node size and color represents log2 fold change (bar in A) for all EGFR-mutant cell lines treated with erlotinib (A, C) or dasatinib (B, D).
Fig 5.
Heatmaps showing subsets of drug-affected phosphorylation sites in cluster A, grouped by cell signaling pathway.
In the left hand panels, sites in cluster A whose median abundance ratio in drug-treated vs. control cells was at least 2.25-fold lower are blue and those that were at least 2.25-fold higher are yellow (black is below threshold). The right hand panels show the drug-treated to control ratio (median of samples in each group) for each PTM site (see key at right; black represents missing data). (A) Sites known to be phosphorylated by ABL-family kinases. (B) Sites known to be phosphorylated by SRC-family kinases (SFKs). (C) Sites that are phosphorylated directly by EGFR or downstream of EGFR activation. (D) Sites involved in negative regulation of growth factor signaling. Sample groups: h3122.criz: H3122 cells treated with crizotinb; hcc78.criz: HCC78 cells treated with crizotinib; pc9.erl: PC9 cells treated with erlotinib; all.criz: H3122, HCC78, H2228, and STE1 cells treated with crizotinib; all.erl: PC9, HCC4006, and HCC827 cells treated with erlotinib; all.dasat: H2286 and H366 cells treated with dasatinib; all.drug: all samples from all.criz, all.erl, and all.dasat as well as H1781 cells treated with afatinib.
Fig 6.
Heatmaps showing PTM sites in Clusters B and C.
In the left hand panels, sites whose median abundance ratio in drug-treated vs. control cells was at least 2.25-fold lower are blue and those that were at least 2.25-fold higher are yellow (black is below threshold). The right hand panels show the drug-treated to control ratio (median of samples in each group) for each PTM site (see key at bottom; black represents missing data). Sample groups are as in Fig 5. (A) Sites in cluster B changed at least 2.25-fold in at least one sample group that are found in focal adhesions and/or are involved in actin dynamics. (B) Heatmaps of all PTM sites in Cluster C. Many of these sites are involved in regulation of transcription and/or are histone subunits. Sample groups: h3122.criz: H3122 cells treated with crizotinb; pc9.erl: PC9 cells treated with erlotinib.
Fig 7.
Common core CFNs from different lung cancer PTM data sets.
The 43 proteins (excluding the ribosomal protein clique) that were in the highest cores of both CFNs graphed with CFN edges from this study. Node size and color indicates PTM changes in PC9 cells treated with erlotinib. Node border and shape and edge colors are defined in S2C Fig.