NXP800 Activates the Unfolded Protein Response, Altering AR and E2F Function to Impact Castration-Resistant Prostate Cancer Growth

Abstract

Purpose: Advanced prostate cancer is invariably fatal, with the androgen receptor (AR) being a major therapeutic target. AR signaling inhibitors have improved overall survival for men with advanced prostate cancer, but treatment resistance is inevitable and includes reactivation of AR signaling. Novel therapeutic approaches targeting these mechanisms to block tumor growth is an urgent unmet clinical need. One attractive strategy is to target heat shock proteins (HSP) critical to AR functional activity.

Experimental design: We first did transcriptome analysis on multiple castration-resistant prostate cancer (CRPC) cohorts to correlate the association between the Gene Ontology cellular response to heat gene expression signature and overall survival. Next, we analyzed the impact of targeting the heat shock factor 1 (HSF1) pathway, with an inhibitor in clinical development, namely, NXP800 (formerly CCT361814), in models of treatment-resistant prostate cancer. Finally, we confirmed our mechanistic and phenotypic findings using an NXP800-resistant model and an in vivo model of CRPC.

Results: We report that in multiple CRPC transcriptome cohorts, the Gene Ontology cellular response to heat gene expression signature associates with AR signaling and worse clinical outcome. We demonstrate the effects of targeting the HSF1 pathway, central to cellular stress, with an inhibitor in clinical development, namely, NXP800, in prostate cancer. Targeting the HSF1 pathway with the inhibitor NXP800 decreases HSP72 expression, activates the unfolded protein response, and inhibits AR- and E2F-mediated activity, inhibiting the growth of treatment-resistant prostate cancer models.

Conclusions: Overall, NXP800 has antitumor activity against treatment-resistant prostate cancer models, including molecular subtypes with limited treatment options, supporting its consideration for prostate cancer-specific clinical development.

Figure 1.

GO cellular response to heat gene expression signature associates with AR signaling and poorer prognosis in men suffering from CRPC. A and G, Two independent (PCF-SU2C and ICR-RMH) transcriptome cohorts of patients with CRPC. Quantification of GO cellular response to heat gene expression signature in each transcriptome cohort of patients with CRPC in the PCF-SU2C (A) and ICR-RMH (G) CRPC cohorts. Biopsies (red dots) with GO cellular response to heat gene expression signature >80th percentile (dotted line) are shown. B and H, Kaplan–Meier curves for OS from CRPC biopsy by >80th percentile (red) or ≤80th percentile (gray) GO cellular response to heat gene expression signature in PCF-SU2C (B) and ICR-RMH (H) transcriptome cohorts. Median OS is shown. HR with 95% CI and P values for univariate Cox survival model are shown. C–F and I–L, Association between GO cellular response to heat gene expression signature and Hallmark Androgen Response, AR signature, Nelson Response to Androgen Up, and AR-V7 signature in transcriptome cohorts of patients with PCF-SU2C (C–F) and ICR-RMH (I–L). Pearson r and P values are shown.

Figure 2.

NXP800 inhibits AR transactivation and AR signaling to inhibit the growth of AR aberrant prostate cancer models. A–C, VCaP (A), LNCaP95 (B), and 22Rv1 (C) prostate cancer cells were treated with vehicle (DMSO 0.1%) or various concentrations (5, 10, 50, 100, and 250 nmol/L) of NXP800 (active, red line) or CCT365248 (inactive, blue line), and growth was determined after 5 days by CellTiter-Glo Luminescent Cell Viability Assay. Mean growth (compared with vehicle; defined as 1) with SD from a single experiment with six replicates is shown. P values were calculated for active compound compared with inactive compound for each concentration using the unpaired Student t test. P values ≤ 0.05 are shown (*). VCaP (A), LNCaP95 (B), and 22Rv1 (C) prostate cancer cells were treated with vehicle (DMSO 0.1%) or various concentrations (50, 100, and 250 nmol/L) of NXP800 (active) or CCT365248 (inactive) for 48 hours, and AR-FL, AR-V7, PSA, and GAPDH protein expression was determined from one experiment performed in triplicate. Single Western blot representative of three is shown. VCaP (A), LNCaP95 (B), and 22Rv1 (C) prostate cancer cells were treated with vehicle (DMSO 0.1%) or various concentrations (50, 100, and 250 nmol/L) of NXP800 (active, red line), or CCT365248 (inactive, blue line) for 48 hours, and KLK2, KLK3, TMPRSS2, and FKBP5 RNA expression was determined. Mean RNA expression (normalized to average of GAPDH/B2M/HRPT1/RPLP0 and vehicle; defined as 1), with SD from a single experiment with three replicates is shown. P values were calculated for active compared with inactive compound for each concentration using the unpaired Student t test. P values ≤ 0.05 are shown (*). D, PC3 cells were transfected with CONTROL-FLAG, AR-FL-FLAG, or AR-V7-FLAG and PSA-ARE3-luciferase, prior to treatment with vehicle (DMSO 0.1%), 250 nmol/L CCT365248 (inactive), 250 nmol/L NXP800 (active), or 3 μmol/L enzalutamide for 1 hour prior to stimulation with or without 10 nmol/L DHT for 16 hours. Mean luciferase activity (compared with CONTROL-FLAG/vehicle/without DHT) with SD from two experiments with five replicates is shown. P values were calculated for each plasmid (with and without DHT stimulation) with vehicle compared with other treatments using the unpaired Student t test. P values ≤ 0.05 are shown (*). E, ChIP with an AR antibody was carried out in 22Rv1 cells that were plated in starved (charcoal-striped serum/phenol red–free) media for 72 hours prior to treatment with 250 nmol/L CCT365248 (inactive) or 250 nmol/L NXP800 (active) for 1 hour prior to stimulation with or without 10 nmol/L DHT for 5 hours. AR recruitment to AR-responsive genes (KLK2, KLK3, FKBP5, TMPRSS2, CHRNA2, and ANKRD30B) was determined. Mean binding as percentage of input with SD from two experiments with three replicates is shown. P values were calculated for 250 nmol/L CCT365248 (inactive) compared with 250 nmol/L NXP800 (active) with and without DHT stimulation using the unpaired Student t test. P values ≤ 0.05 are shown (*).

Figure 3.

NXP800 inhibits the growth of AR-dependent and AR-independent prostate cancer models with activation of the UPR and inhibition of key signaling pathways. A and B, PDX-O [CP50, CP89, CP129, and CP142 (A)], AR-positive (VCaP, LNCaP, LNCaP95, and 22Rv1), and AR-negative (PC3 and DU145) prostate cancer cell lines (B) were treated with vehicle (DMSO 0.1%) or various concentrations (5, 10, 50, 100, and 250 nmol/L) of NXP800 (active, red line), CCT365248 (inactive, blue line), and in the case of organoids, various concentrations (1 and 10 μmol/L) of enzalutamide (gray scale), and growth was determined after 5 days for cell lines and 7 days for organoids by CellTiter-Glo Cell Viability Assay as defined in “Materials and Methods.” Mean growth (compared with vehicle; defined as 1) with SD from a single experiment with three to six replicates is shown. *Abiraterone given in the castration-sensitive setting. ^CP89 and CP129 derived from two temporally separated mCRPC biopsies from the same patient. P values were calculated for NXP800 compared with CCT365248 for each concentration and for enzalutamide, compared with vehicle using the unpaired Student t test. P values ≤ 0.05 are shown (*). C–E, VCaP, LNCaP95, and 22Rv1 prostate cancer cells were treated with NXP800 (active, 100 or 250 nmol/L) or CCT365248 (inactive, 250 nmol/L) for 48 hours. RNA-seq was performed on each single experiment in triplicate (duplicate for 250 nmol/L NXP800 in LNCaP95). Analysis of RNA-seq with gene set enrichment analysis shows the enrichment and de-enrichment of Hallmark pathways in response to 100 and 250 nmol/L NXP800 (compared with 250 nmol/L CCT365248) in VCaP (C), LNCaP95 (D), and 22Rv1 (E) prostate cancer cells. NES and FDR are shown as volcano plots. Colored dots denote significantly (FDR 0.05) enriched (red dots) and de-enriched (blue dots) pathways with NXP800 (active compound) treatment. Table shows the NES associated with pathways wherein the FDR was ≤0.05 for 100 and/or 250 nmol/L (asterisk indicates those in which FDR was >0.05). F, VCaP (A), LNCaP95 (B), and 22Rv1 (C) prostate cancer cells were treated with NXP800 (active, 100 or 250 nmol/L) or CCT365248 (inactive, 250 nmol/L) for 48 hours. RNA-seq was performed on each single experiment in triplicate (duplicate for 250 nmol/L NXP800 in LNCaP95). Log₂ fold expression level changes of “activating” E2F (E2F1–3) family members treated with NXP800 (active, 100 or 250 nmol/L) were compared with CCT365248 (inactive, 250 nmol/L). P values were calculated by DESeq2 using the Wald test. P values ≤ 0.05 are shown (*). G, VCaP, LNCaP95, and 22Rv1 prostate cancer cells were treated with 250 nmol/L CCT365248 (inactive) or 250 nmol/L NXP800 (active) for 24 hours. PERK, phospho-eIF2α, and ATF4 (PERK arm); ATF6 (ATF6 arm); IRE1 (IRE1 arm); E2F1 (E2F); and GAPDH (housekeeping) protein expression was determined by Western blot from one experiment performed in triplicate. H, Association between GO cellular response to heat gene expression signature and Hallmark E2F Targets in transcriptome cohorts of patients with PCF-SU2C and ICR-RMH. Pearson r and P values are shown.

Figure 4.

NXP800 activates the UPR and inhibits key signaling pathways identifying a novel mechanism of action in prostate cancer models. A, VCaP, LNCaP95, and 22Rv1 prostate cancer cells were treated with 250 nmol/L CCT365248 (inactive) or 250 nmol/L NXP800 (active) for 24 hours prior to the addition of puromycin for 30 minutes. PERK, phospho-eIF2α, and ATF4 (PERK arm); ATF6 (ATF6 arm); IRE1 (IRE1 arm) protein; E2F1; puromycin (incorporation a surrogate for protein synthesis); and GAPDH (housekeeping) protein expression was determined by Western blot from one experiment performed in triplicate. B, VCaP, LNCaP95, and 22Rv1 prostate cancer cells were treated with 250 nmol/L NXP800 (active) or 250 nmol/L CCT365248 (inactive) with various concentrations of the small-molecule ISRIB (0.1, 0.5, and 1 μmol/L), and growth was determined after 5 days by CellTiter-Glo Luminescent Cell Viability Assay. Mean growth from a single experiment with four replicates was determined. Growth (fold change) between 250 nmol/L NXP800 and 250 nmol/L CCT365248 with various concentrations of ISRIB (0.1, 0.5, and 1 μmol/L) is shown. C, VCaP, LNCaP95, and 22Rv1 prostate cancer cells were treated with 250 nmol/L CCT365248 (inactive) or 250 nmol/L NXP800 (active) with various concentrations of ISRIB (0.1, 0.5, 1, 2, and 5 μmol/L) for 24 hours prior to the addition of puromycin for 30 minutes. PERK, phospho-eIF2α, and ATF4 (PERK arm); ATF6 (ATF6 arm); IRE1 (IRE1 arm); E2F1; puromycin (incorporation a surrogate for protein synthesis); AR-FL and AR-V7; and GAPDH (housekeeping) protein expression was determined by Western blot from one experiment. D, VCaP, LNCaP95, and 22Rv1 prostate cancer cells were treated with 250 nmol/L CCT365248 (inactive) or 250 nmol/L NXP800 (active) with various concentrations of ISRIB (0.1, 0.5, and 1 μmol/L) for 48 hours, and KLK2, KLK3, TMPRSS2, and FKBP5 RNA expression was determined. Mean RNA expression (normalized to average of GAPDH/B2M/HRPT1/RPLP0) from a single experiment with three replicates is shown. RNA expression (fold change) between 250 nmol/L NXP800 and 250 nmol/L CCT365248 with various concentrations of ISRIB (0.1, 0.5, and 1 μmol/L) is shown.

Figure 5.

NXP800-resistant 22Rv1 prostate cancer cell sublines demonstrate the reversal of the NXP800-mediated phenotype. A, Long-term treatment of 22Rv1 prostate cancer cells with increasing concentrations (up to 2.5 μmol/L) of DMSO (vehicle-C, white), CCT365248 (inactive-C, blue), and NXP800 (NXP800-R, red) led to the generation of 22Rv1 prostate cancer cell–derived sublines. Mean growth was determined after 5 days by CellTiter-Glo Luminescent Cell Viability Assay compared with day 0 and vehicle-C cells with SD for each subline developed is shown. P values were calculated for subline compared with vehicle-C using the unpaired Student t test. P values ≤ 0.05 are shown (*). B, The impact of NXP800 and CCT365248 on the growth of vehicle-C, inactive-C, and NXP800-R sublines was determined. Mean growth in response to various concentrations of NXP800 (active, red line) and CCT365248 (inactive, blue line) was determined after 5 days by CellTiter-Glo Luminescent Cell Viability Assay compared with day 0 and vehicle-C cells with SD for each subline developed is shown. P values were calculated for active compared with inactive for each subline using the unpaired Student t test. P values ≤ 0.05 are shown (*). C, Inactive-C and NXP800-R sublines were treated with 250 nmol/L CCT365248 (inactive) or 250 nmol/L NXP800 (active) for 24 hours prior to the addition of puromycin for 30 minutes. PERK, phospho-eIF2α, and ATF4 (PERK arm); ATF6 (ATF6 arm); IRE1 (IRE1 arm); E2F1; puromycin (incorporation a surrogate for protein synthesis); and GAPDH (housekeeping) protein expression was determined by Western blot from one experiment performed in triplicate. D–F, Inactive-C and NXP800-R sublines were treated with either 250 nmol/L CCT365248 (inactive) or 250 nmol/L NXP800 (active) for 48 hours. RNA-seq was performed on each single experiment in triplicate. Analysis of RNA-seq with gene set enrichment analysis shows the enrichment and de-enrichment of Hallmark pathways comparing 250 nmol/L CCT365248 (inactive) vs. 250 nmol/L NXP800 (active) treated inactive-C sublines (D), 250 nmol/L CCT365248 (inactive) vs. 250 nmol/L NXP800 (active) treated NXP800-R sublines (E), and 250 nmol/L CCT365248 (inactive) treated inactive-C sublines vs. 250 nmol/L CCT365248 (inactive) treated NXP800-R sublines (F). NES and FDR are shown as volcano plots. Colored dots denote significantly (FDR 0.05) enriched (red dots) and de-enriched (blue dots) pathways. Log₂ fold expression level changes of “activating” E2F (E2F1–3) family members comparing 250 nmol/L CCT365248 (inactive) vs. 250 nmol/L NXP800 (active) treated inactive-C sublines (D), 250 nmol/L CCT365248 (inactive) vs. 250 nmol/L NXP800 (active) treated NXP800-R sublines (E), and 250 nmol/L CCT365248 (inactive) treated inactive-C sublines vs. 250 nmol/L CCT365248 (inactive) treated NXP800-R sublines (F). P values were calculated by DESeq2 using the Wald test. P values ≤ 0.05 are shown (*).

Figure 6.

NXP800 activates the UPR and inhibits E2F-mediated transcription to drive antitumor activity against the castration-resistant VCaP prostate cancer cell line–derived mouse xenograft. A and B, Castration-resistant emergent VCaP prostate cancer cell line–derived mouse xenografts were treated with vehicle (n = 7, gray) or 35 mg/kg NXP800 (active, n = 8, red) after tumors had established castration-resistant growth with a defined dosing schedule for 38 days. Mean tumor volume (normalized to start; defined as 100%) with SEM is shown. P values were calculated for NXP800 compared with vehicle using the unpaired Student t test at 38 days. P values ≤ 0.05 are shown (*; A). Time to reach 200% starting tumor volume was used as a surrogate endpoint for survival. HR with 95% CI and P value for univariate Cox survival model are shown (B). C–G, Castration-resistant VCaP prostate cancer cell line–derived mouse xenografts were treated with vehicle (n = 4) or 35 mg/kg NXP800 (active, n = 4) daily after tumors had established castration-resistant growth for 5 days with tumor collection 6 hours after final dose for pharmacodynamic studies. RNA-seq was performed on tumor samples. Analysis of RNA-seq with gene set enrichment analysis shows the enrichment and de-enrichment of Hallmark pathways comparing NXP800 (active) and vehicle. NES and FDR are shown as volcano plots. Colored dots denote significantly (FDR 0.05) enriched (red dots) and de-enriched (blue dots) pathways (C). Log₂ fold expression level changes of “activating” E2F (E2F1–3) family members treated with NXP800 (active) were compared with vehicle. P values were calculated by DESeq2 using the Wald test. P values ≤ 0.05 are shown (*; D). PERK, phospho-eIF2α, total-eIF2α, and ATF4 (PERK arm); ATF6 (ATF6 arm); IRE1 (IRE1 arm); and GAPDH (housekeeping) protein expression was determined by Western blot (E). Cleaved caspase 3 (F) and Ki-67 (G) protein expression was determined by IHC on formalin-fixed, paraffin-embedded tumors. Representative micrographs are shown. Scale bar, 50 μm. Mean percentage positive cells with SD are shown. P values were calculated for NXP800 (red) compared with vehicle (white) using the unpaired Student t test. P values ≤ 0.05 are shown (*).