Cell Line Characteristics Predict Subsequent Resistance to Androgen Receptor-Targeted Agents (ARTA) in Preclinical Models of Prostate Cancer

Abstract

Treatment of prostate cancer (PCa) has changed considerably in the last decade due to the introduction of novel androgen receptor (AR)-targeted agents (ARTAs) for patients progressing on androgen deprivation therapy (ADT). Preclinical research however still relies heavily on AR-negative cell line models. In order to investigate potential differences in castration-resistant PCa (CRPC) growth, we set out to create a comprehensive panel of ARTA-progressive models from 4 androgen-responsive AR wild-type PCa cell lines and analyzed its androgen response as opposed to its ADT-progressive counterparts. Parallel cultures of VCaP, DuCaP, PC346C, and LAPC4 were established in their respective culture media with steroid-stripped fetal calf serum (FCS) [dextran-coated charcoal-stripped FCS (DCC)] without androgen (ADT) or in DCC plus 1 μM of the ARTAs bicalutamide, OH-flutamide, or RD162 (an enzalutamide/apalutamide analog). Cell growth was monitored and compared to those of parental cell lines. Short-term androgen response was measured using cell proliferation 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT) assay. qRT-PCR was performed to assess the mRNA expression of markers for AR signaling, steroidogenesis, glucocorticoid receptor (GR) signaling, epithelial-mesenchymal transition (EMT), and WNT signaling. Out of 35 parallel cultures per cell line, a total of 24, 15, 34, and 16 CRPC sublines emerged for VCaP, DuCaP, PC346C, and LAPC4, respectively. The addition of bicalutamide or OH-flutamide significantly increased CRPC growth compared to ADT or RD162. VCaP, DuCaP, and PC346C CRPC clones retained an AR-responsive phenotype. The expression of AR and subsequent androgen response were completely lost in all LAPC4 CRPC lines. Markers for EMT and WNT signaling were found to be elevated in the resilient PC346C model and CRPC derivatives of VCaP, DuCaP, and LAPC4. Although the resistant phenotype is pluriform between models, it seems consistent within models, regardless of type of ARTA. These data suggest that the progression to and the phenotype of the CRPC state might already be determined early in carcinogenesis.

Keywords: ARTA; CRPC; DuCaP; LAPC4; PC346C; VCaP.

Figure 1

Success rates and cell culture dynamics of CRPC sublines. (A) Survival rate of CRPC sublines per cell line and hormonal condition. Coral sectors represent a fraction of established clones relative to attempts. Overall effect of type of ARTA (ADT vs. BIC vs. FLU vs. RD162), p < 0.0001. Overall primary cell line effect (VCaP vs. DuCaP vs. PC346C vs. LAPC4), p < 0.0001. Drug effect per cell line: VCaP, p = 0.0002; DuCaP, p = 0.0002; PC346C, p = 0.10; LAPC4, p = 0.01 (χ² test). (B) Total passages in culture per clone until the end of the experiment or failure. † = failure to proliferate beyond the noted passage. (C) Smooth spline curves of average passage time at subsequent passages stratified per cell line and ARTA condition. Overall, passage time did not stabilize before the 10th passage. Top right numbers indicate the mean time in days until the 10th passage per ARTA condition for each cell line group. “ns” is “not significant”; “*” is “* p<0.05”; “***” is “*** p<0.001”.

Figure 2

Androgen growth response and antiandrogen potency in CRPC clones. (A) Heatmap of growth response of individually tested clones. First column represents fold growth relative to day 0. Subsequent vehicle columns represent growth induction (red) or inhibition (blue) relative to no R1881/vehicle. Each column with antiandrogen (bicalutamide, OH-flutamide, RD162) represents a normalized response to the corresponding vehicle control at 1 μM ARTA in the presence of different concentrations of R1881. Gray = missing data. (B) Average proliferation for parental and CRPC clones. Average fold growth relative to day 0 in the MTT assay for parental cells (mint bars) and all CRPC cells combined (coral bars). The p value represents parental vs. CRPC in two-way ANOVA. ^†p < 0.05 vs. no R1881/vehicle. ^‡p < 0.05 vs. vehicle control at a given amount of R1881.

Figure 3

Summary of qPCR analysis of parental cell lines. (A) Relative expression of castration resistance-associated genes in parental samples. Summarized qPCR results for parental samples of VCaP (n = 5), DuCaP (n = 4), PC346C (n = 5), and LAPC4 (n = 4). Expression levels are -ΔCt relative to GAPDH and PBGD, meaning that every ΔCt represents a 2-fold difference. * p < 0.05, ** p < 0.01, *** p < 0.001 for Tukey post-hoc test. (B) Heatmap of clusters identified by differential gene expression of castration resistance-associated genes. Colors indicate -ΔCt vs. median, with red showing a higher expression and blue showing a lower expression. Numbers in the top row indicate median Ct for detection per gene. nd is not detected.

Figure 4

Heatmap of clusters identified by differential gene expression of the 15-gene signature between parental cell lines and ARTA-resistant lines. Clockwise from top left: VCaP, DuCaP, LAPC4, and PC346C. Note that in VCaP, DuCaP, and LAPC, parental samples all have a distinct distance from resistant samples, whereas in PC346C, parental samples mix with resistant samples. In all 4 backgrounds, different types of ARTA did not lead to the rise of a specific cluster, except for OH-flutamide-resistant clones in PC346C. Colors indicate -ΔCt vs. median, with red showing a higher expression and blue showing a lower expression. Numbers in the top row indicate median Ct for detection per gene.