Fig 1.
Heat map of relative TK mRNA expression levels among MIBC patient tumor samples.
TCGA data for MIBC patient tumor samples (n = 408) reported previously [22]. TK gene symbols are listed vertically at right. Samples are grouped on the horizontal axis by mRNA subtype classifications as defined by Robertson et al. [22] indicated at the top (from left to right): neuronal (N, green); basal squamous (BS, red); luminal (L, gray); luminal infiltrating (LI, gold); luminal papillary (LP, purple). TK genes are clustered hierarchically; in the cladogram at left cabozantinib targets are designated with violet squares.
Fig 2.
Kaplan-Meier survival analyses in patients harboring significant mRNA overexpression and/or amplification of TK genes.
Kaplan-Meier analyses of OS (left panels) and DFS (right panels) in patients harboring significant mRNA overexpression (z-score >2) and/or amplification of (A) group 1 TKs (red) vs. those without alteration (blue; median OS 28.22 vs. 61.40 mos., log-rank p = 0.0308; median DFS 27.99 vs. 82.42 mos., log-rank p = 0.0145); (B) group 2 TKs (red) vs. those without alteration (blue; 46.65 vs. 22.14 mos., log-rank p = 3.197e-4; median DFS 51.41 vs. 19.05 mos., log-rank p = 6.673e-4); (C) group 3 TKs (red) vs. those without alteration (blue; 30.91 vs. 59.26 mos., log-rank p = 0.134; median DFS 27.99 vs. 72.34 mos., log-rank p = 0.0391); (D) 14 ctRTKs (AXL, CSF1R, FLT3, KDR, MET, MERTK, NTRK1/2/3, RET, ROS1, TEK, DDR2 and TYRO3; red) vs. those without alteration (blue; 25.56 vs. 54.86 mos., log-rank p = 0.0289; median DFS 25.23 vs. 55.16 mos., log-rank p = 0.0109).
Fig 3.
TK expression in MIBC patient tumor samples and BCa cell lines.
(A) Heat maps of relative mRNA expression levels in 408 tumor samples from patients with MIBC [22] (RNA Seq V2, left) and absolute mRNA copy number in 15 BCa cell lines (right) of 31 TKs (gene symbols listed vertically, middle). MIBC patient tumor samples are grouped on the horizontal axis by mRNA subtype classifications as previously defined [22] indicated at the top (from left to right): neuronal (N, green); basal squamous (BS, red); luminal (L, gray); luminal infiltrating (LI, gold); luminal papillary (LP, purple). TK genes are clustered hierarchically; colored vertical bars at indicate group 1 TKs (black) above group 2 TKs (orange). Highest expression is shown in yellow, moderate expression in blue and lowest expression in black; ND for cell line data indicates transcript was undetectable. (B) Cladogram of 34 receptor TKs (RTKs) most closely related to Met based on TK domain amino acid sequence identity, constructed using the EMBL-EBI ClustalW2 Phylogeny tool (http://www.ebi.ac.uk/Tools/phylogeny/clustalw2_phylogeny). RTK symbol colors identify established subfamilies. Values for cabozantinib doses calculated to provide 50% TK inhibition (IC50, nM) in a cell-free assay, listed at right, were derived from a minimum of 3 x 10-dose binding curves with regression (R) values > 0.98 unless indicated by NC (not calculated), where values were >3 micromolar. Values considered clinically relevant (< 150 nM) are shown in red. TKs for which IC50 values could not be determined are denoted by a single dash. (C) Oncoprint showing alterations for 19 genes encoding ctRTKs (identified previously and/or by cell-free kinase inhibition assay) among 408 MIBC samples [22]. Gene symbols and percentage of samples altered (%) for each are listed at left (Oncoprint generated using visualization tools at the cBioPortal for Cancer Genomics, Memorial Sloan Kettering Cancer Center, https://www.cbioportal.org/ [37]). (D) Linear regression analysis of AXL (left) and MET (right) mRNA abundance (y-axis) vs. protein content (x-axis) for 15 BCa-derived cell lines shows significant direct correlation (AXL: Spearman r = 0.638, p = 0.040; MET: Spearman r = 0.7902, p = 0.003).
Table 1.
Classification of BCa-derived cell lines based on TK-specific molecular phenotypes adapted from those defined using TCGA data analyzed by Robertson et al. [22] (all samples with RNASeq V2 data, n = 408).
Fig 4.
Potential autocrine RTK signaling in MIBC patient samples and BCa cell lines.
(A) Heat map showing relative mRNA expression levels of ctRTK-cognate ligand pairs (RNA Seq V2) among 408 tumor samples from patients with MIBC [22] (left), and among 15 BCa-derived cell lines (absolute mRNA copy number, right). (B) Immunoblot detection of ctRTKs and their cognate ligands, CSF1/CSF1R, GAS6/AXL, MST1/MST1R, and evidence of DDR1, DDR2 and PTK7 proteins in 9 BCa-derived cell lines. GAPDH (bottom panel) serves as a loading control. (C) Left: heat map showing relative mRNA expression levels of 9 EMT-associated transcriptional activators (listed at right) among 408 tumor samples from patients with MIBC [22]. Samples are grouped on the horizontal axis by mRNA subtype classifications defined previously [22] indicated at the top (from left to right): neuronal (N, green); basal squamous (BS, red); luminal (L, gray); luminal infiltrating (LI, gold); luminal papillary (LP, purple), and clustered hierarchically on the vertical axis. Right: heat map showing absolute mRNA copy levels of 9 EMT-associated transcriptional activators among 15 BCa-derived cell lines (identified at bottom).
Table 2.
Significant co-occurrence of potentially oncogenic alterations (RNASeq V2 >2-fold and/or gene amplification) for RTKs and cognate ligands in BCa TCGA dataset analyzed by Robertson et al. (n = 408) [22] and differences in median OS and PFS.
Table 3.
Median months OS and DFS for combined significantly concurrent potentially oncogenic alterations (RNASeq V2 >2-fold and/or gene amplification and/or mutation as noted) for RTKs and cognate ligands in BCa TCGA dataset (n = 408) [22].
Table 4.
Significant co-occurrence of potentially oncogenic alteration (overexpression per RNASeq V2 >2-fold, and/or gene amplification) among EMT TAs in TCGA dataset analyzed by Robertson et al. (n = 408) [22] and associated differences in median OS and PFS.
Table 5.
(A) Significant co-occurrence of potentially oncogenic alteration (overexpression >2-fold, and/or gene amplification) for Group 1 TKs and EMT TAs in TCGA dataset (n = 408) [22] and associated median OS and PFS.
(B) Significant co-occurrence of potentially oncogenic alteration (overexpression >2-fold, and/or gene amplification) for Group 2 TKs and EMT TAs in TCGA dataset (n = 408) [22] and associated median OS and PFS.
Fig 5.
Autocrine RTK signaling in BCa cell lines.
(A) Left: Axl kinase activation (phospho-Axl/total-Axl protein, pAxl/tAxl), as determined by 2-site immunoassay, in serum-deprived J82 cells treated with Gas6 (5.7 nM, 1 h at 37°C) in the presence or absence of cabozantinib (cabo; 300 nM), relative to untreated cells. Maximum possible Axl kinase activation was measured by adding ATP during the immunoassay (20 micromolar, 20 min at 25°C) of lysates prepared from untreated J82 cells. Right top: Axl kinase activation (pAxl/tAxl) in serum-deprived J82 cells that had been transfected with GAS6 siRNA or scrambled (control) siRNA. In both panels, values are the mean +/- SD from triplicate samples; asterisks indicate significant difference from untreated control (p < 0.05). Right bottom: ethidium bromide visualization of PCR products specific for GAS6 mRNA (upper panel) vs 18S rRNA (lower panel) resolved by agarose gel electrophoresis. (B) Left: Akt kinase activation (pAkt/tAkt), as determined by 2-site immunoassay, in serum-deprived J82 cells treated with Gas6 (5.7 nM, 1 h at 37°C) in the presence or absence of cabozantinib (cabo; 300 nM), relative to untreated cells. Right: Akt kinase activation (pAkt/tAkt) in serum-deprived J82 cells that had been transfected with Gas6 siRNA or scrambled (control) siRNA. In both panels, values are the mean +/- SD from triplicate samples; asterisks indicate significant difference from untreated control (p < 0.05). (C) Left: Erk kinase activation (pErk/tErk), as determined by 2-site immunoassay, in serum-deprived J82 cells treated with Gas6 (5.7 nM, 1 h at 37°C) in the presence or absence of cabozantinib (cabo; 300 nM), relative to untreated cells. Right: Erk kinase activation (pErk/tErk) in serum-deprived J82 cells that had been transfected with GAS6 siRNA or scrambled (control) siRNA. In both panels, values are the mean +/- SD from triplicate samples; asterisks indicate significant difference from untreated control (p < 0.05). (D) Upper panels: Immunoblot (IB) of phospho-MST1R (pMST1R) and MST1R (left) or pCSF1R and CSF1R (right) in lysates from serum-deprived FL3 cells after transfection with siRNA directed against MST1 (siMST1, left) or CSF1 (siCSF1, right) or control siRNA (siCont). Lower panels: Ethidium bromide visualization of PCR products (PCR) specific for MST1 (left) or CSF1 (right) mRNA or 18S RNA resolved by agarose gel electrophoresis. (E) Left: Akt kinase activation (pAkt/tAkt) in serum-deprived FL3 cells treated with MST1 (5 nM) or CSF1 (10 nM) in the presence or absence of cabozantinib (cabo; 300 nM), relative to untreated cells. Right: Akt kinase activation (pAkt/tAkt) in FL3 cells that had been transfected with siRNA directed against CSF1 or MST1, or with control siRNA. In both panels, values are the mean +/- SD from triplicate samples; asterisks indicate significant difference from control (p < 0.05). (F) Left: Erk kinase activation (pErk/tErk) in serum-deprived J82 cells treated with MST1 or CSF1 in the presence or absence of cabozantinib (cabo), relative to untreated cells. Right: Erk kinase activation (pErk/tErk) in FL3 cells that had been transfected with siRNA directed against CSF1 or MST1, or with control siRNA. In both panels, values are the mean +/- SD from triplicate samples; asterisks indicate significant difference from untreated control (p < 0.05). All results are representative of at least 3 experiments.
Fig 6.
Autocrine RTK driven migration and proliferation in BCa cell lines.
(A) Migration (16 h, 37°C) of serum deprived J82 cells previously transfected with siRNA directed against AXL (siAXL) or control siRNA (siControl), treated with Gas6 (5.7 nM) in the presence or absence of cabozantinib (cabo; 300 nM), relative to untreated cells. Values are the mean +/- SD from triplicate samples; asterisks indicate significant difference from control (p < 0.05; ns, not significant). All results are representative of at least 3 experiments. Inset: ethidium bromide visualization of PCR products specific for AXL mRNA (upper panel) or 18S rRNA (lower panel) resolved by agarose gel electrophoresis. (B) Migration (16 h, 37°C) of serum deprived FL3 cells previously transfected with siRNA directed against MST1R (siMST1R), CSF1R (siCSF1R) or control siRNA (siControl), treated with MST1 (5 nM) or CSF1 (10 nM) in the presence or absence of cabozantinib (cabo; 300 nM), relative to untreated cells. Values are the mean +/- SD from triplicate samples; asterisks indicate significant difference from untreated control (p < 0.05). All results are representative of at least 3 experiments. Inset: ethidium bromide visualization of PCR products specific for MST1R, (upper panel), CSF1R (middle panel), or 18S RNA (lower panel) resolved by agarose gel electrophoresis. (C) Proliferation (cell no. x 10−5, 6 d, 37°C) of 5637 (left) or J82 (right) cells in 1% FBS left untreated (control; circles) or treated with Gas6 alone (5.7 nM; squares), cabozantinib alone (300 nM, triangles) or Gas6 and cabozantinib (inverted triangles). Media with additives was replaced on days 1, 2, and 4; cells were counted on days 0, 3 and 6. Values are the mean +/- SD from triplicate samples; asterisks indicate significant difference from control (p < 0.05). (D) Proliferation (cell no. x 10−5, 6 d, 37°C) of FL3 cells in 1% FBS left untreated (control; circles) or treated with MST1 alone (5 nM; squares), cabozantinib alone (300 nM, triangles) or MST1 and cabozantinib (inverted triangles). Media replacement and cell counting as per panel C. Values are the mean +/- SD from triplicate samples; asterisks indicate significant difference from control (p < 0.05).
Fig 7.
Autocrine RTK driven proliferation and colony formation in BCa cell lines.
(A) PCR products showing siRNA suppression (upper and middle panels in each group) vs. 18S rRNA loading controls (lower panel in each group) resolved by agarose gel electrophoresis and visualized using ethidium bromide staining, from cell lines used for results shown in panels B–F. Upper left: J82 cells transfected with siRNA directed against GAS6 or AXL used for results shown in panel B. Upper middle: FL3 cells transfected with siRNA directed against MST1 or MST1R used for results shown in panel C. Upper right: FL3 cells transfected with siRNA directed against CSF1 or CSF1R used for results shown in panel D. Lower left: J82 cells transfected with siRNA directed against AXL used for results shown in panel E. Lower right: FL3 cells transfected with siRNA directed against MST1R or CSF1R used for results shown in panel F. (B) Proliferation (abs at 570 nm, 3 d, 37°C) of J82 cells in 1% FBS previously transfected with siRNA directed against GAS6 (siGAS6), AXL (siAXL) or control siRNA (siControl) that were left untreated or treated with Gas6 (5.7 nM) in the absence or presence of cabozantinib (cabo, 300 nM). (C) Proliferation of FL3 cells in 1% FBS previously transfected with siRNA directed against MST1 (siMST1), MST1R (siMST1R) or control siRNA (siControl) that were left untreated or treated with MST1 (5 nM) in the absence or presence of cabozantinib (cabo, 300 nM). (D) Proliferation of FL3 cells in 1% FBS previously transfected with siRNA directed against CSF1 (siCSF1), CSF1R (siCSFR) or control siRNA (siControl) that were left untreated or treated with CSF1 (10 nM) in the absence or presence of cabozantinib (cabo, 300 nM). (E) Anchorage independent growth (abs at 570 nm) of J82 cells in 1% FBS previously transfected with siRNA directed against AXL (siAXL) or control siRNA (siControl) that were left untreated or treated with Gas6 (5.7 nM) in the absence or presence of cabozantinib (cabo, 300 nM). (F) Anchorage independent growth of FL3 cells in 1% FBS previously transfected with siRNA directed against MST1R (siMST1R), CSF1R (siCSFR) or control siRNA (siControl) that were left untreated or treated with MST1 (5 nM) or CSF1 (10 nM) in the absence or presence of cabozantinib (cabo, 300 nM). In panels C—F, values are the mean +/- SD from triplicate samples; asterisks indicate significant difference from untreated control (p < 0.05, ns, not significant). In all panels, data are representative of 3 or more independent experiments.
Fig 8.
Serial passage of J82 cell xenografts in mice.
(A) Tumor volume (mm3) of individual J82 (left) or MDXC1 (middle) cell xenografts in mice, or mean (+/- SD) tumor growth of both cell line xenografts (right), vs. days post implantation. (B) mRNA expression levels of AXL, TYRO3, MERTK, CSF1R, RET, MET, KDR, GAS6, KIT and MST1R in J82 cells and 8 J82 tumor-derived cell lines (MDXC1, MDXC2, MDXC31R, MDXC31L, MDXC33R, MDXC33L, MDXC34R and MDXC34L) as determined by qRT-PCR. (C) Immunoblot of lysates from J82 and MDXC1 lysates for phospho-MERTK (pMERTK, upper panels), MERTK (middle panels) or GAPDH (lower panels); intact cells had been treated prior to lysis with Gas6 (5.7 nM) or TP0903 (300 nM) as indicated. Results are representative of 3 or more independent experiments. (D) Light micrographs (10x objective magnification) of stained J82 (upper panels) of MDXC1 (lower panels) 24 h cell migration in the absence (control, left) or presence of added Gas6 (Gas6, 5.7 nM, right). (E) Heat map showing relative mRNA expression levels of 321 genes of the 770 gene Nanostring PanCancer Progression Panel (clustered hierarchically) that were significantly different among untreated J82 (left) and MDXC1 (right) cells (two-group comparison p = 0.021, q = 0.05; S5 Table in S1 File).