Patient-derived Models of Abiraterone- and Enzalutamide-resistant Prostate Cancer Reveal Sensitivity to Ribosome-directed Therapy

Abstract

Background: The intractability of castration-resistant prostate cancer (CRPC) is exacerbated by tumour heterogeneity, including diverse alterations to the androgen receptor (AR) axis and AR-independent phenotypes. The availability of additional models encompassing this heterogeneity would facilitate the identification of more effective therapies for CRPC.

Objective: To discover therapeutic strategies by exploiting patient-derived models that exemplify the heterogeneity of CRPC.

Design, setting, and participants: Four new patient-derived xenografts (PDXs) were established from independent metastases of two patients and characterised using integrative genomics. A panel of rationally selected drugs was tested using an innovative ex vivo PDX culture system.

Intervention: The following drugs were evaluated: AR signalling inhibitors (enzalutamide and galeterone), a PARP inhibitor (talazoparib), a chemotherapeutic (cisplatin), a CDK4/6 inhibitor (ribociclib), bromodomain and extraterminal (BET) protein inhibitors (iBET151 and JQ1), and inhibitors of ribosome biogenesis/function (RNA polymerase I inhibitor CX-5461 and pan-PIM kinase inhibitor CX-6258).

Outcome measurements and statistical analysis: Drug efficacy in ex vivo cultures of PDX tissues was evaluated using immunohistochemistry for Ki67 and cleaved caspase-3 levels. Candidate drugs were also tested for antitumour efficacy in vivo, with tumour volume being the primary endpoint. Two-tailed t tests were used to compare drug and control treatments.

Results and limitations: Integrative genomics revealed that the new PDXs exhibited heterogeneous mechanisms of resistance, including known and novel AR mutations, genomic structural rearrangements of the AR gene, and a neuroendocrine-like AR-null phenotype. Despite their heterogeneity, all models were sensitive to the combination of ribosome-targeting agents CX-5461 and CX-6258.

Conclusions: This study demonstrates that ribosome-targeting drugs may be effective against diverse CRPC subtypes including AR-null disease, and highlights the potential of contemporary patient-derived models to prioritise treatment strategies for clinical translation.

Patient summary: Diverse types of therapy-resistant prostate cancers are sensitive to a new combination of drugs that inhibit protein synthesis pathways in cancer cells.

Keywords: Abiraterone; Androgen receptor; Castration-resistant prostate cancer; Enzalutamide; Explant; Neuroendocrine prostate cancer; Organoid; Patient-derived xenograft; Prostate cancer; Ribosome.

Fig. 1 –

Establishing PDXs of CRPC from rapid autopsy specimens. (A) Sites that tumour material was collected from for two patients (patients 27 and 201) who consented to the CASCADE rapid autopsy programme. The metastases where there was sufficient fresh tissue for xenografting are marked. A subset of tumours formed successful PDXs. The four tumours that produced serially transplantable PDXs (27.1, 27.2, 201.1, and 201.2) are shown in colour, while other sites that were initially grafted, but did not establish long-term PDXs that are shown in white. (B) Serum PSA measurements from diagnosis to death for patient 27 and patient 201 and the systemic treatments they received. (C) Time course of the first 400 d of PDX growth for the four PDXs from patients 27 and 201. Each step represents a new generation (27.1—blue; 27.2—green; 201.1—red; 201.2—yellow). AR = androgen receptor; CRPC = castration-resistant prostate cancer; LN = lymph node; PDX = patient-derived xenograft; PSA = prostate-specific antigen.

Fig. 2 –

PDXs maintain the genomic features of patient tumours. (A) Heatmap showing the Spearman’s correlation between the copy number profiles of all pairs of samples based on low-coverage whole genome sequencing. (B) Summary of key somatic alterations in each sample based on low-coverage whole-genome sequencing, targeted DNA sequencing, and FISH. Crosses indicate that the TMPRSS2:ERG fusion was present in 201.2 according to FISH, but not detected by RNA-Seq. FISH = fluorescence in situ hybridisation; PDX = patient-derived xenograft; PDX E = early-generation PDX; PDX L = late-generation PDX.

Fig. 3 –

PDXs exhibit diverse mechanisms of resistance to AR-directed therapies. (A) Average weight of grafts from each PDX grown in control mice bearing testosterone implants (+T, red) or established in mice bearing testosterone implants and then castrated (CX, blue) for 3–28 d. Fast growing PDXs from patient 201 were harvested early due to animal ethics guidelines (unpaired T tests for +T versus CX at each time point, n = 4–8 grafts/group, ±SEM). (B) Schematic of the AR gene showing the locations of two genomic inversions in PDXs 27.1 and 27.2. (C) Plots showing sequencing coverage across coding and noncoding regions of the AR gene in PDXs 27.1 and 27.2. (D) Representative images of immunohistochemistry for the AR N terminal (AR-N), AR C terminal (AR-C), ARv567es, and AR-V7 in each PDX. Scale bars equal 50 μm. (E) Transcriptional activation assays with an AR-responsive probasin luciferase construct comparing wild-type AR (AR-wt) with single and double mutant forms. Transfected PC3 cells were treated for 24 h with vehicle or DHT alone or in combination with dexamethasone (Dex), enzalutamide (Enz), bicalutamide (Bic), flutamide (Flut), hydroxyflutamide (OH-Flut), or nilutamide (Nil). Data represent the average luciferase signal relative to Renilla (±SEM) for a representative experiment with n = 6 wells per treatment. (F) Heatmap showing the relative expression of genes associated with AR-null and neuroendocrine prostate cancer. Data represent log₂ fold changes in transcript abundance relative to the median for each gene. (G) GSEA of up-and downregulated genes in the Beltran signature in tumour 201.2 versus tumour 201.1. AR = androgen receptor; AR-FL = AR full length; AR-GSR = AR genomic structural rearrangement; CRPC = castration-resistant prostate cancer; FDR = false discovery rate; GSEA = gene-set enrichment analysis; NE = neuroendocrine; NES = normalised enrichment score; PDX = patient-derived xenograft; PDX E = early-generation PDX; PDX L = late-generation PDX; SEM = standard error of the mean. * p < 0.05. ** p < 0.01.

Fig. 4 –

Integrated models for rapid preclinical testing. (A) Tumour 201.1 as a PDX, an explant, organoids, and regrafted organoids (Organoid-PDX) from the same generation (scale bars: upper panel—2, 2, 0.1, and 2 mm from left; lower panel—50 μm). Average percentage of (B) Ki67 and (C) cleaved caspase 3 (Casp3)-positive cells in explants after 48 h. Each point represents average data for three explants per PDX generation (n = 3–6 generations ± SEM). (D) The average percentage of individual cells from each PDX that form organoids (% organoid-forming efficiency, OFE). Each point represents a different PDX generation. (E) Average diameter (μm) of organoids for each PDX over time. PDX = patient-derived xenograft; SEM = standard error of the mean.

Fig. 5 –

Diverse tumours are sensitive to dual inhibition of ribosome biogenesis and function. The results of explants treated for 48 h with (A) AR-directed inhibitors (Gal—10 μM galeterone; Enz—10 μM enzalutamide), (B) cell cycle or DNA repair inhibitors (PARPi—1 μM talazoparib; CDDP—1 μM cisplatin; CDKi—1 μM ribociclib), (C) BET inhibitors (I-BET—1 μM I-BET 151; 1 μM JQ1), or (D) ribosome-targeted inhibitors (1 μM CX-5461; 1 μM CX-6258; 1 μM CX-5461 + 1 μM CX-6258). Graphs summarise the combined data from multiple experiments and represent average changes in the percentage of Casp3- (red) and Ki67-positive (blue) cells versus the vehicle control (n = 3, ±SEM, * p < 0.05, **p < 0.01, T test of absolute Ki67 and Casp3 values in each treated group compared to its matching vehicle control from the same sample of PDX tissue). (E) Western blots of 201.1 and 201.2 organoids treated for 4 h with 1 μM CX-5461 and 5 μM CX-6258 alone or in combination. In vivo treatment of (F) 201.2 and (G) 305 PDXs for 2 wk with CX-5461 and CX-6258 (TX) compared to the vehicle control. Data represent average tumour volume per mouse (n = 4 grafts/mouse **p < 0.01, paired T test). Representative images of grafts are shown (scale = 250 μM). AR = androgen receptor; Av = average; Casp3 = caspase 3; PDX = patient-derived xenograft; SEM = standard error of the mean.