Development and Characterisation of a New Patient-Derived Xenograft Model of AR-Negative Metastatic Castration-Resistant Prostate Cancer

Abstract

As the treatment landscape for prostate cancer gradually evolves, the frequency of treatment-induced neuroendocrine prostate cancer (NEPC) and double-negative prostate cancer (DNPC) that is deficient for androgen receptor (AR) and neuroendocrine (NE) markers has increased. These prostate cancer subtypes are typically refractory to AR-directed therapies and exhibit poor clinical outcomes. Only a small range of NEPC/DNPC models exist, limiting our molecular understanding of this disease and hindering our ability to perform preclinical trials exploring novel therapies to treat NEPC/DNPC that are urgently needed in the clinic. Here, we report the development of the CU-PC01 PDX model that represents AR-negative mCRPC with PTEN/RB/PSMA loss and CTNN1B/TP53/BRCA2 genetic variants. The CU-PC01 model lacks classic NE markers, with only focal and/or weak expression of chromogranin A, INSM1 and CD56. Collectively, these findings are most consistent with a DNPC phenotype. Ex vivo and in vivo preclinical studies revealed that CU-PC01 PDX tumours are resistant to mCRPC standard-of-care treatments enzalutamide and docetaxel, mirroring the donor patient's treatment response. Furthermore, short-term CU-PC01 tumour explant cultures indicate this model is initially sensitive to PARP inhibition with olaparib. Thus, the CU-PC01 PDX model provides a valuable opportunity to study AR-negative mCRPC biology and to discover new treatment avenues for this hard-to-treat disease.

Keywords: androgen receptor (AR); castration-resistant prostate cancer (CRPC); double-negative prostate cancer (DNPC); neuroendocrine (NE); patient-derived xenograft (PDX).

Figure 1

Generation of the CU-PC01 PDX model. (A) Graph displays the donor patient serum PSA levels over time from the point of diagnosis to death in relation to treatment (blue dashed line = lymph node image guided biopsy collection timepoint). (B) Axial T2 MRI and (C) coronal T1 MRI of the pelvis was performed to guide lymph node biopsy. (D) Plot displays the growth trajectory of the CU-PC01 PDX model from propagation to 419 days, involving six serial transplantations into adult male NSGs and two subsequent serial transplantations into adult athymic nude males (strain switch = red dashed line). The testosterone supplement was removed at the final passage (passage 8). (E) Representative H&E images of donor patient primary prostate adenocarcinoma at diagnosis (prostate carcinoma, left panel) and morphologically heterogeneous high-grade carcinoma with focal NE features in the lymph node metastasis biopsy (middle) and CU-PC01 PDX tumour (passage 1, right panel). Scale bar = 50 µm.

Figure 2

CU-PC01 PDX tumours genocopy the donor patient tumour with high concordance. (A) Alluvial plot generated using R ggalluvial package displaying the frequency of consensus SNV mutations that are present or absent in the patient lymph node biopsy and CU-PC01 PDX tumour specimens collected at P1, P5 and P7. A large majority of mutations are conserved in the CU-PC01 PDX tumours across all passages. (B) Enrichment of selected gene signatures in a consensus set of gene mutations present in all samples. The overlap of consensus genes with those in signatures is shown, as compared to the level expected by chance in 1,000,000 simulations. * indicates significant enrichment, defined as p < 0.05. (C) Heatmap shows genes from signalling pathways commonly deregulated in mCRPC (detailed in side bar), and genes with SNPs/INDELs present in the coding sequence (and UTRs) in the patient lymph node biopsy and/or CU-PC01 PDX tumour specimens collected at P1, P5, and P7. SIFT scores were transformed to 1 − x, thus a highly significant score is represented by a higher number. INDELs, Stop gain, and missense SNVs without a SIFT score are marked ‘NA’. Where multiple SIFT scores are present (e.g., multiple gene mutations detected), the most significant is used. INDELs = grey dot.

Figure 3

Characterisation of prostate epithelial cell populations within the CU-PC01 PDX model. Representative images of IHC staining of the patient lymph node metastatic biopsy and CU-PC01 PDX tumours collected at passage 1 (P1), a rederived cryopreserved tissue fragment at passage 2 (P2-Cryo) and at passage 5 (P5) to detect (A) human mitochondria (Hu. Mito), cytokeratin-8 (CK8) and cytokeratin-5 (CK5) staining and (B) NE markers CD56, INSM1, SYP and CHGA. Low magnification scale bar = 200 µm, high magnification scale bar = 50 µm (n = 3). Boxes indicate the high-magnification region.

Figure 4

CU-PC01 mCRPC PDX tumours are AR-negative. (A) Representative IHC images for AR in the donor patient biopsy and CU-PC01 PDX tumours collected at P1, P2-cryo and P5 (n = 3). Low magnification scale bars = 200 µm, high magnification scale bars = 50 µm. Boxes indicate the high-magnification region. QRT-PCR analysis was performed on CU-PC01 PDX tumour specimens (P2) to detect (B) AR and (C) AR-v7 splice variant mRNA transcripts. 22Rv1 human CRPC cells = AR and AR-v7 positive control. PC-3 human CRPC cells = AR and AR-v7 negative control. Error bars = S.E.M (n = 3, 3 independent repeats). One-Way ANOVA with Tukey correction, *** p < 0.001.

Figure 5

The CU-PC01 PDX model displays activated Wnt signalling and PTEN loss. Representative IHC images for (A) β-catenin and (B) PTEN in the donor lymph node metastatic patient biopsy and CU-PC01 PDX tumours collected at P1, P2-cryo, and P5 (n = 3) show a high level of nuclear β-catenin and the absence of PTEN, respectively. Low magnification scale bars = 200 µm, high magnification scale bars = 50 µm. Boxes indicate the high-magnification region.

Figure 6

CU-PC01 PDX tumours are resistant to enzalutamide and docetaxel. (A) Chart displays tumour growth for subcutaneous CU-PC01 PDX tumours in adult male athymic nude mice treated with either vehicle, enzalutamide (10 mg/kg by gavage 5 days on 2 days off), or docetaxel (10 mg/kg i.p. once a week) for 15 days. Error bars represent S.E.M, Two-way ANOVA = not significant, p = 0.9748 (n = 4 or 6/treatment arm). (B) Representative IHC images and (C,D) quantitation for the apoptosis marker cleaved-caspase-3 (CC3) and proliferation marker PCNA in CU-PC01 PDX tumours treated with either enzalutamide or docetaxel in vivo for 15 days (n = 3/treatment arm). No significant difference was observed between the treatment arms (one-way ANOVA with Tukey correction, p ≥ 0.5683). (E) Representative IHC images and (F,G) quantitation for CC3 and PCNA CU-PC01 PDX tumour explant cultures treated with either enzalutamide or docetaxel for 48 h ex vivo (n = 3/treatment arm). Error bars represent S.E.M. No significant difference was observed between the treatment arms (one-way ANOVA with Tukey correction, p ≥ 0.6390). Scale bars = 100 µm.

Figure 7

CU-PC01 PDX ex vivo explants are sensitive to PARP inhibition. (A) Representative IHC images for CC3 and PCNA, and quantitation of the percentage of (B) CC3 and (C) PCNA positive cells in CU-PC01 PDX tumour explant cultures treated with either vehicle (DMSO) or olaparib (10 µM) for 48 h (n = 3 independent repeats, in triplicate). Olaparib significantly increased the percentage of CC3-positive cells and significantly reduced the percentage of PCNA-positive cells (unpaired two-tailed t test; * p < 0.05, ** p < 0.01). (D) Representative IHC images for CC3 and PCNA, and quantitation of the percentage of (E) CC3 and (F) PCNA positive cells in CU-PC01 PDX tumour explant cultures treated with either vehicle (DMSO) or capivasertib (1 µM) for 48 h (n = 3 independent repeats, in triplicate). No significance was observed (unpaired two-tailed t test, p ≥ 0.0731). Error bars = S.E.M.