Identification of a ternary protein-complex as a therapeutic target for K-Ras-dependent colon cancer

Abstract

A cancer phenotype is driven by several proteins and targeting a cluster of functionally interdependent molecules should be more effective for therapeutic intervention. This is specifically important for Ras-dependent cancer, as mutated (MT) Ras is non-druggable and targeting its interaction with effectors may be essential for therapeutic intervention. Here, we report that a protein-complex activated by the Ras effector p38γ MAPK is a novel therapeutic target for K-Ras-dependent colon cancer. Unbiased proteomic screening and immune-precipitation analyses identified p38γ interaction with heat shock protein 90 (Hsp90) and K-Ras in K-Ras MT, but not wild-type (WT), colon cancer cells, indicating a role of this complex in Ras-dependent growth. Further experiments showed that this complex requires p38γ and Hsp90 activity to maintain MT, but not WT, K-Ras protein expression. Additional studies demonstrated that this complex is activated by p38γ-induced Hsp90 phosphorylation at S595, which is important for MT K-Ras stability and for K-Ras dependent growth. Of most important, pharmacologically inhibition of Hsp90 or p38γ activity disrupts the complex, decreases K-Ras expression, and selectively inhibits the growth of K-Ras MT colon cancer in vitro and in vivo. These results demonstrated that the p38γ-activated ternary complex is a novel therapeutic target for K-Ras-dependent colon cancer.

Figure 1. p38γ overexpression in CRC specimens correlates with shortened patient survival and p38γ specifically forms a ternary complex with Hsp90 and K-Ras in K-Ras MT colon cancer cells

(A) Colon cancer specimens were stained for p38γ expression by IHC [20] and Kaplan-Meier plot of the overall survival according to p38γ levels was shown (left), and associated clinical parameters were given at left. (B, C) Flag immune-precipitates were separated by SDS-PAGE and samples from the numbered region (1-6) were digested, which were then subjected to proteomic analysis as described [22]. MS/MS spectra of heat shock protein 90α peptide HLEINPDHSIIETLR and 90β peptide ADLINNLGTIAK with representative peak matches for b and y ions {detected from the sample 1 but not sample 4 of B}. Additional proteomic data were given in Supplementary Figure S1B/C). (D) Western blotting (WB) analysis of Flag precipitates revealed the p38γ interaction with Hsp90 in HCT116 but not in Hke-3 cells (top left). Endogenous p38γ and Hsp90 proteins were also isolated after treatment with 17-AAG or DMSO for their interactions with K-Ras (top right and bottom right) with a portion of whole cell lysates analyzed by direct WB as an input (bottom, left).

Figure 2. K-Ras is converted to an Hsp90 client protein upon mutation

(A) NCM460 cells were transiently expressed with increasing concentrations (0.5, 1.0, and 1.5μg DNA) of WT and MT (G12V) K-Ras constructs for 24h, followed by incubation with 0.1 μM of 17-AAG or DMSO for additional 24 h, which were then subjected to WB analysis. *, the band intensity was quantitated by the NIH software, which was normalized with GAPDH, and K-Ras protein level in 17-AAG-treated group was expressed as relative to that treated with DMSO. A separate experiment showed similar results. (B) Colon cancer cells with indicated K-Ras phenotype were incubated with indicated concentration of 17-AAG for 24 h, which were then analyzed for protein expression by WB. Similar results were obtained in a separate experiment.

Figure 3. Hsp90 is a therapeutic target for K-Ras MT colon cancer

(A) HCT116 and its MT K-Ras disrupted sub-lines were incubated with 17-AAG for 24 h and assessed for cell viability using trypan blue staining [22]. Results are means of three experiments (+ SD, * vs. either Hke3 or HK2-8 cell line). (B) Indicated cells were cultured in the presence and absence of 17-AAG for about 2 weeks and colony formed was manually counted. Results are expressed as % of solvent treated control (mean ± SD, n = 3, * vs. Caco2 cells). (C, D) Indicated human colon cancer cells (2 × 10⁶ in 0.1 ml of PBS) were s.c. injected into nude mice on both sides of front flank and 17-AAG (80 mg/kg) or solvent (DMSO) solution (in 50 μl) was i.p. administrated daily for 14 days for HCT116 tumors and for 6 days for Ls-174T xenograft. Changes in tumor volume and body weight were monitored every other day [20]. Results in all groups are means of 4-5 tumors (± SEM, n = 4-5 mice, * vs. DMSO). (E, F) The same amounts of lysate proteins from DMSO or 17-AAG treated tumors were immune-precipitated with indicated antibodies and precipitates were then analyzed by WB for their complex formation, with a portion of whole lysates analyzed by direct WB as an input.

Figure 4. p38γ phosphorylates Hsp90, and the p38γ pharmacological inhibitor pirfenidone (PFD) blocks Hsp90 phosphorylation, more effectively inhibits the K-Ras dependent growth in vitro, and selectively synergizes with 17-AAG to inhibit the colony formation of K-Ras MT colon cancer cells

(A) Flag-Hsp90 proteins isolated from 293T cells were incubated with bacterially expressed His-p38γ or His-p38α, and kinase activity assays were performed [24]. Phosphorylated Hsp90 was detected by WB using a specific phospho-MAPK substrate (p-S/TP) antibody (left, top) and the same amounts of His-p38α and His-p38γ proteins were analyzed by WB as input controls (left, bottom). (B) Indicated constructs were transfected into 293T cells for 48 h, with 20 μg/ml of PFD or solvent added for the last 24 h, and cells were then analyzed for protein expression and phosphorylation (right). (C, D) Colon cancer cells were incubated with indicated concentrations of PFD and/or 17-AAG for about 2 weeks. Colonies formed were manually counted and presented as % of solvent treatment. Results are mean of 3-4 experiments (± SD, * P < 0.05 vs. either WT K-Ras line for C; ** P < 0.05 vs. either inhibitor, * P > 0.05 vs. either inhibitor for D).

Figure 5. p38γ depletion selectively depletes K-Ras protein in K-Ras MT colon cancer cells

(A) Colon cancer cells were stably depleted of endogenous p38γ protein by lentiviral mediated shRNA delivery [20] and analyzed for protein expression by WB. Similar results were obtained in a separate experiment. (B) HCT116 cells with and without p38γdepletion were transiently transfected with HA-UB for 48 h with and without incubation with 17-AAG (0.1 μM) for the last 24 h, followed by the last 6-h incubation with and without MG132 (100 μM). Cell lysates were collected for WB analyses. (C) Colon cancer cells were incubated with 100 μg/ml of PFD or solvent DMSO for 24 h, and then analyzed for protein expression by WB. Similar results were obtained in a separate experiment.

Figure 6. PFD inhibits K-Ras-dependent colon cancer growth in mice, disrupts the p38γ/Hsp90/K-Ras complex, and depletes K-Ras protein in tumor tissues

(A, B) Tumor cells were s.c. injected as described above and daily therapy with PFD or DMSO control was i.p. administrated into mice when tumor becomes palpable. Tumor volume was measured every other day and results shown are mean of 4-5 tumors (± SEM). Tumors were also weighted at the end of experiments (inserts, mean ± SD, * vs. PFD). (C, D) PFD disrupts the ternary protein-complex and decreases K-Ras protein expression in tumor tissues. The same amount of lysate proteins from PFD or DMSO treated tumor were subjected to IP with indicated antibodies and precipitates were analyzed for complex formation by WB with a portion of total lysates as an input control. (E) An experimental model elucidates that K-Ras mutation stimulates p38γ expression/phosphorylation leading to Hsp90 phosphorylation resulting in the K-Ras stabilization by a ternary complex. An inhibition of p38γ activity by PFD and/or Hsp90 by 17-AAG will attenuate the complex formation, decrease the K-Ras expression, and suppress the K-Ras dependent growth and progression.