Overexpression of MET in patient derived papillary renal cancer cell lines
UOK345, UOK337, UOK342, andUOK332 were generated in the Urologic Oncology Branch (UOB) from tumors derived from patients with metastatic PRCC, as described previously [21]. Briefly, extensive genomic characterization of these cell lines using DNA sequencing, spectral karyotyping and fluorescence in situ hybridization revealed the presence of either wild type (UOK342, UOK332) or activating mutations in the TK domain of MET (UOK345, H1112R; UOK337, H1106Q) [21]. Additionally, gain of chromosome 7 was demonstrated in all four cell lines [21] (Suppl. Table. S2). We confirmed increased levels of both total and phosphorylated MET (Y1234/1235) in all four PRCC cell lines compared to normal Renal Proximal Tubular Epithelial cells (RPTEC) using western blot analysis (Fig. 1A, Suppl. Fig. S1).
MET tyrosine kinase inhibitor had minimal effects on growth and invasion of PRCC cells
Since MET has been implicated as a driver in at least a subset of PRCC tumors and is considered a clinically relevant target in these tumors, we evaluated the activity of the MET TKI EMD1214063 (Tepotinib) on PRCC cell lines in vitro. EMD1214063 is a potent and selective inhibitor of the MET TK receptor with > 1000-fold selectivity for MET compared to other kinases and is in phase 2 clinical trial for patients with confirmed MET exon 14 skipping mutation with advanced or metastatic non–small-cell lung cancer [28, 29]. We found that EMD1214063 had minimal effects on the proliferation of UOK345, UOK337, UOK342 and UOK332 cells, with growth inhibition comparable to that seen in RPTEC cells (Fig. 1B). Similar effects were observed when these cell lines were treated with other MET selective tyrosine kinase inhibitors (capmatinib and AMG208) (data not shown). Furthermore, PRCC cells treated with EMD1214063 in the presence of HGF (100 ng/ml) demonstrated no change in invasion when compared to vehicle alone-treated cells (Fig. 1C).
Effect of EMD1214063 on MET kinase and downstream signaling pathways
Since auto-phosphorylation of MET on tyrosine residues Y1234/1235 in its C-terminal domain provides a docking site for recruitment of signal transducers that are necessary for activation of downstream signaling pathways including PI3K/AKT and MAPK/ERK [30], we examined the effect of EMD1214063 on phosphorylation of MET (Y1234/1235) and known downstream mediators of MET activity including AKT and ERK1/2 in PRCC cells. EMD1214063 (50 nM) markedly inhibited the auto-phosphorylation of MET as determined by western blot analysis (Fig. 1D). However, there was minimal effect on growth or invasion of these cells at this concentration and up to tenfold higher concentrations of EMD1214063 (Fig. 1B and C). Furthermore, there was no significant modulation of downstream mediators of MET activity, including phospho-AKT (Ser473) and phospho-ERK (Thr202/Tyr204) (Fig. 1D). Together, these results suggest that EMD1214063 has little effect on the growth of PRCC cell lines in vitro, probably attributable to the activation of the MAPK/ERK and PI3K/AKT pathways by alternative mechanisms.
High Throughput drug screening identifies alternative therapeutic targets in PRCC cells
In order to uncover alternative therapeutic strategies for patients with papillary renal cell carcinoma, we tested the dose–response behaviors in two PRCC cell lines- one with a MET TK domain mutation (UOK345, H1112R) and a second with copy number gain of wild type MET (UOK342)- employing a MIPE library which was comprised of a panel of 1912 mechanistically annotated anti-cancer agents that are FDA approved or are currently being evaluated in clinical trials. As demonstrated in our growth inhibition assays (Fig. 1B), most MET inhibitors were characterized by weak antitumor activity in the high throughput screen (Fig. 2A). However, we identified several other classes of agents that displayed notable anti-tumor activity against both papillary cell lines (Fig. 2A). The ranking of these compounds was based on average potency obtained using Z-transformed, area-under-the-curve (Z-AUC) values for each compound (Fig. 2A). PI3KCA, mTOR, TUB, EGFR, HSP90, HDAC and PSMD1 inhibitors were among the most active group of agents in the HTS. Amongst these classes, we selected HSP90 inhibition as the target of this study with SNX2112 chosen as the representative HSP90 inhibitor, due to it having the greatest effect in the screen (Suppl. Fig. S2). Several factors informed our selection of HSP90 inhibition in addition to their effectiveness in the screening data. Firstly, HSP90 is involved in the folding and activation of specific client proteins including several oncologically relevant molecules downstream from MET, PI3K, mTOR and EGFR pathways, thus HSP90 inhibition may simultaneously target many of the pathways identified in the screen [17,18,19]. This is likely to provide a more effective response than specific, individual targeting of each of the other identified pathways. Secondly, addition of HSP90 inhibitors have been shown to overcome resistance to MET TKIs in MET driven tumor models and SNX2112 has been shown to be effective as a single agent in a cell line with acquired MET TKI resistance [19, 20]. SNX2112, via the prodrug SNX5422, is orally bioavailable and acts by competitively binding to the N-terminal adenosine triphosphate (ATP) binding site of HSP90, making it a useful reagent for in vivo studies [25].
HSP90 inhibition results in potent growth inhibition of PRCC cells
We assessed the biologic effects of HSP90 inhibition in a panel of PRCC cell lines including those with MET mutations (UOK345, UOK337) and those with copy number gain of wild type MET (UOK342, UOK332). Treatment with SNX2112 exhibited potent and dose dependent growth inhibition in all PRCC cell lines evaluated, in vitro (IC50 range- 52.5 nmol/l to 105 nmol/l) compared to vehicle treated control (p < 0.05); however, 50% inhibition was not achieved even at higher, micromolar concentrations of the agent in normal renal tubular epithelium derived RPTEC cells (IC50-1129 nM). The PRCC cell line UOK275, was used as an additional wild type MET control [22] and was significantly less sensitive to SNX2112 treatment, with an IC50 (515 nM) that was 5 to tenfold higher than that seen with MET altered PRCC lines (Fig. 2B). Further, 2D colony foci formation was significantly suppressed (p < 0.05) when PRCC cells were seeded at low density (1 × 103 – 1.5 × 103 cells/well) and treated with SNX2112. (Fig. 2C). Similarly, 3D soft-agar colony formation under anchorage-independent conditions was potently inhibited with SNX2112 treatment at nanomolar concentrations (p < 0.05) (Fig. 2D, E).
SNX2112 induces the degradation of HSP90 client proteins
Since HSP90 stabilizes numerous oncogenic client proteins [31], we sought to assess the effects of SNX2112 on the expression of relevant proteins such as MET, pMET (Y1234/1235) and downstream mediators of MET activity, including phosphorylated forms of AKT and ERK1/2 in PRCC cells. We observed that SNX2112 treatment led to degradation of total MET and inhibited pMET (Y1234/1235) in a dose-dependent manner (Fig. 3A). However, unlike TKIs targeting MET, SNX2112 also induced the degradation of other HSP90 client including the native form of AKT while also inhibiting phosphorylated forms of AKT (Ser473) and ERK (Thr202/Tyr 204) (Fig. 3A). In order to better understand if SNX2112 induces degradation and/or decreases protein stability of these proteins, cells pre-treated with the protein synthesis inhibitor cycloheximide (10ug/ml) alone or in combination with the proteasome inhibitor MG132 (10uM) were then exposed to SNX2112 (100 nM) or vehicle for an additional 8 h. In the absence of MG132, SNX2112 resulted in markedly reduced half-life compared to untreated cells of the HSP90 client proteins tMET, pMET, tAKT, pAKT, and pERK1/2. However, in the presence of MG132, we observed protection of these HSP90 clients from the destabilizing effect of SNX2112, suggesting that SNX2112 induces proteasome-dependent degradation of these proteins (Suppl. Fig. S3). These data indicate that the potent activity of SNX2112 against PRCC may be the result of more complete abrogation of both MET and key downstream mediators of MET activity that are not adequately inhibited by MET TKIs.
Biologic effects of SNX2112 are partially dependent on PI3K/AKT and MEK/ERK1/2 signaling
To elucidate whether the anti-tumor effects of SNX2112 in PRCC cells are dependent on the PI3K/AKT and MEK/ERK1/2 signaling axis, cells with MET mutation (UOK345) as well as those with copy number gain of wild type MET (UOK342) were transfected with small interfering RNA (siRNA) for AKT1/2, ERK1/2 or non-targeting control. Reduced expression of AKT1/2 and ERK1/2 proteins was confirmed in the siAKT1/2 and siERK1/2 transfected cells respectively compared to non-targeting control by western blot (Fig. 3B). The percentage decrease in expression for AKT1/2 and ERK1/2 proteins was between 60–80% in UOK345 and UOK342 cell lines as indicated in Fig. 3B. The reduction of either endogenous AKT1/2 or ERK1/2 led to a significant decrease in PRCC cell viability at 48 h and 72 h (p < 0.05) (Fig. 3C). A marked decrease in 2-D colony formation was also seen in both PRCC cell lines when expression of AKT1/2 or ERK1/2 were knocked down (Fig. 3D). To further determine the role played by these pathways in SNX2112-mediated growth inhibition, we evaluated the efficacy of SNX2112 in PRCC cells (UOK345 and UOK342) overexpressing AKT1/2 or ERK1/2. PRCC cell lines overexpressing AKT1/2 or ERK1/2 were marginally less sensitive to SNX2112 treatment than parent cells (not transfected with AKT1/2 or ERK1/2 expression plasmids) (Suppl. Fig. S4). These data suggest that AKT1/2 or ERK1/2 overexpression only partially abrogates SNX2112 biologic effects. These results are concordant with our HTS data which identified up- and downstream regulators of AKT1/2 (PI3KCA and mTOR) and the up-stream components of the ERK1/2 pathway (SRC, EGFR, MAP2K1) as targets of interest. These data further suggest that, while PI3K/AKT and MEK/ERK1/2 signaling are crucial for growth of these PRCC cells, the anti-tumor activity of SNX2112 likely involves important additional signaling pathways.
SNX2112 treatment induces apoptosis and G2/M cell cycle arrest in PRCC cells
HSP90 client proteins such as PI3K/AKT and MEK/ERK1/2 are known to regulate cell death and apoptotic pathways as well progression through the cell cycle [32]. We sought to determine the extent to which growth inhibition of PRCC cells was mediated by apoptosis and/or cell cycle arrest. PRCC cell lines were treated with SNX2112 for 48 h and changes in the proportion of apoptotic cells and markers of apoptosis (caspases 3/7 and PARP) were analyzed.
SNX 2112 treatment resulted in a significant (p < 0.05) increase in the number of cells (expressed in %) undergoing apoptosis [control vs treated: 3.9 vs 10.48% (UOK345); 1.7 vs 10.73% (UOK342); 4.19 vs 10.43% (UOK337); 4.2 vs 10.8% (UOK332) (Fig. 4A and B). SNX2112 treatment also significantly (p < 0.05) increased the activity of caspases 3 and 7 at 48 h (Fig. 4C). Since cleavage and inactivation of PARP by caspases is a hallmark of apoptosis [33], we assessed the effect of SNX2112 on PARP cleavage following 48 h of treatment. Western blot analysis revealed increased levels of cleaved PARP, which further confirmed the induction of apoptosis in SNX2112 treated cells (Fig. 4D).
To evaluate the effects of SNX112 on the cell cycle, cells with MET mutation (UOK345) as well as those with copy number gain of wild type MET (UOK342) were treated with 50 nM of SNX2112, stained with propidium iodide and analyzed by flow cytometry. As shown in Fig. 4E and F, SNX2112 led to a significant (p < 0.01) G2/M arrest in both PRCC cell lines. Collectively, these results suggest that SNX2112 treatment induced apoptosis as well as G2/M arrest in PRCC cells.
Identification of differentially expressed genes by RNA sequencing and additional pathways affected by SNX2112 using gene set enrichment analysis (GSEA)
To study the effects of SNX2112 on global gene expression in PRCC, cells with MET mutation (UOK345) and/or copy number gain of wild type MET (UOK342) were treated with SNX2112 (50 nM) for 48 h and mRNA-sequencing was performed to identify differentially expressed genes (DEGs). The expression of approximately 6,038 genes was significantly altered (false discovery rate [FDR] q < 0.05; fold change > 1.5) in both cell lines treated with SNX2112 compared to vehicle controls (Fig. 5A, Suppl. Table. S3). Of the 6,038 differentially expressed genes, 2,902 genes were downregulated while 3,136 genes were upregulated (Fig. 5A). The heatmap (Fig. 5B) represents the top 100 genes (50 up- and down) modulated by SNX2112 treatment in both cell lines. These genes were ranked by their false discovery rate and fold change between SNX2112 treated versus vehicle treated cells.
Next, GSEA were performed using the 50 Hallmark gene set collections in MSigDB for the identification of specifically enriched biological pathways following SNX2112 treatment. GSEA analysis of the common differentially expressed genes in both MET mutated and WT MET cell lines treated with SNX2112 revealed strong negative enrichment for the gene sets involved in cell cycle progression and proliferation including hallmark_E2F_targets (FDR = < 0.001; NES ≥ 3), hallmark_G2M_checkpoint (FDR = < 0.001; NES ≥ 2) and hallmark_MYC_targets_V1 (FDR = < 0.001; NES ≥ 2) (Fig. 5C and D). Strong positive enrichment of gene sets involved in hallmark_protein_secretion (FDR = < 0.001; NES ≥ 2) was also observed (Fig. 5E). The upregulated and downregulated gene sets are presented in Fig. 5C and tabulated with FDR q value in Suppl. Table. S4.
SNX2112 treatment downregulates unique PRCC-associated prognostic genes
In order to determine the biologic relevance of the genes in PRCC that were modulated by SNX2112, likely due to the downstream consequences of degradation of HSP90 clients, we explored the TCGA dataset from the Kidney Renal Papillary Cell Carcinoma (KIRP) cohort to ascertain whether any of the top 200 (100 up and down) differentially expressed genes identified by RNA seq analysis (Suppl. Table. S3) were aberrantly expressed and/or associated with outcome. Several genes downregulated following SNX112 treatment, including CDC20, TPX2, CEP55, FOXM1, KIFC1, NUF2, BUB1B, BIRC5 and PLK1 were overexpressed in Papillary Renal Cell Carcinoma TCGA cohort (Suppl. Fig. S5). Consistent with observations from the TCGA dataset RNA seq gene expression demonstrated overexpression of these genes in the PRCC cell line with relatively low expression observed in RPTEC cells (Suppl. Table. S5). Kaplan–Meier survival analysis were performed for each of these genes according to its relative expression level, dichotomized as high or low relative to its median expression among the entire TCGA KIRP cohort. Interestingly, high expression of each of these genes was associated with poor overall survival (p < 0.001) (Fig. 5F). Additionally, RCAN1, which figured in the list of the top 100 upregulated genes, demonstrated decreased expression in PRCC (Suppl. Fig. S5), with lower expression predicting poor overall survival in the KIRP TCGA cohort (p < 0.01) (Fig. 5G). Notably, RCAN1 has been reported as a tumor suppressor in many cancers, including thyroid and breast cancers [34, 35].
Of the genes that were downregulated in response to SNX2112 and also appear to be overexpressed and associated with poor prognosis in PRCC, five genes (CDC20, FOXM1, BUB1B, BIRC5 and PLK1) that were components of the most enriched pathways identified by GSEA (E2F, G2/M checkpoint and MYC) were selected for further study. All five genes have previously been shown to be involved in progression of a variety of cancers including prostate, bladder, and gastric cancer [36,37,38,39,40,41]. mRNA levels of all five genes in UOK345 or UOK342 cell lines treated with SNX2112, and vehicle treated controls were analyzed using quantitative Real Time PCR. Expression of all five genes was significantly downregulated in PRCC cell lines treated with SNX2112 compared to vehicle controls (Fig. 5H), a finding consistent with that seen in RNA-seq analysis.
HSP90 inhibition suppresses growth of PRCC xenografts in vivo
To evaluate the in vivo anti-tumor efficacy of HSP90 inhibitors in a mouse model using PRCC xenografts, NSG mice were injected subcutaneously either with UOK345 (MET mutation) or with UOK342 (copy number gain of wild type MET) and treated with SNX5422 (30 mg/kg) 3 times a week until five weeks once tumors were established (100 mm3) as depicted in Fig. 6A. SNX2112 is the active inhibitor but SNX 5422 is a prodrug that is orally available and used in animal studies [19, 25, 26].
SNX5422 treatment significantly inhibited the growth of UOK345 and UOK342 xenografts over the five weeks of treatment compared to the vehicle treated control group (p < 0.01) (Fig. 6B and C). Furthermore, mice treated with SNX5422 demonstrated a marked improvement in overall survival (median OS 91 days for UOK342 and median OS 106 days for UOK345) compared to the vehicle treated control mice (median OS 76.5 days for UOK342 and median OS 78 days for UOK345) (p < 0.001) (Fig. 6D).
Next, the xenograft tumors were examined by immunohistofluorescence to detect the levels of Ki-67, a proliferation marker. As shown in Fig. 6E, SNX5422 treatment resulted in significant decrease in Ki-67 levels. Further, SNX5422 also induced the degradation of MET and inhibits key downstream mediators of MET activity including activated forms of MET(Y1234/1235), AKT (Ser473) and ERK (Thr202/Tyr 204) in tumor-bearing xenografts (n = 3, each group) as demonstrated by western blotting (Fig. 6F). Interestingly, the expression of all five genes associated with PRCC survival including CDC20, FOXM1, BUB1B, BIRC5 and PLK1 was significantly reduced in SNX5422 treated (n = 3) compared to control mice (n = 3) (p < 0.001) (Fig. 6G).