A gain-of-function mutation in DHT synthesis in castration-resistant prostate cancer

Abstract

Growth of prostate cancer cells is dependent upon androgen stimulation of the androgen receptor (AR). Dihydrotestosterone (DHT), the most potent androgen, is usually synthesized in the prostate from testosterone secreted by the testis. Following chemical or surgical castration, prostate cancers usually shrink owing to testosterone deprivation. However, tumors often recur, forming castration-resistant prostate cancer (CRPC). Here, we show that CRPC sometimes expresses a gain-of-stability mutation that leads to a gain-of-function in 3β-hydroxysteroid dehydrogenase type 1 (3βHSD1), which catalyzes the initial rate-limiting step in conversion of the adrenal-derived steroid dehydroepiandrosterone to DHT. The mutation (N367T) does not affect catalytic function, but it renders the enzyme resistant to ubiquitination and degradation, leading to profound accumulation. Whereas dehydroepiandrosterone conversion to DHT is usually very limited, expression of 367T accelerates this conversion and provides the DHT necessary to activate the AR. We suggest that 3βHSD1 is a valid target for the treatment of CRPC.

Fig. 1

The 3βHSD1(367T) protein encoded by mutant HSD3B1(1245C) increases flux from DHEA to AD, which is otherwise rate-limiting, en route to DHT and expression of AR-responsive genes. (A) Metabolic flux from [³H]-DHEA (100 nM) to AD and downstream to 5α-dione and DHT is robust in LNCaP but limited in LAPC4. The metabolic pathway and steroid structures are shown, indicating sites of modification by 3βHSD1 in converting DHEA to AD. Steroids were quantitated at the indicated time points by HPLC. (B) DHEA induces PSA and TMPRSS2 expression in a concentration-dependent manner in LNCaP but not LAPC4. Expression was assessed by qPCR and normalized to RPLP0 and vehicle control. (C) A substitution converting A → C at position 1245 in HSD3B1 occurs in LNCaP and VCaP encoding a change from N → T at amino acid 367 in 3βHSD1. (D) Wild-type 3βHSD1(367N) and 3βHSD1(367T) have comparable kinetic properties. Michaelis-Menten plot of DHEA metabolism with 3βHSD1(367N) (circle) and 3βHSD1(367T) (square) enzyme. The K_m for 3βHSD1(367N) and 3βHSD1(367T) protein are 32 and 77 μM, respectively. (E) Endogenous expression of 3βHSD1(367T) is associated with increased protein quantity. Error bars in A, B and D represent the SD from experiments performed in triplicate. See also Figure S1.

Fig. 2

Somatic selection for HSD3B1(1245C) encoding 3βHSD1(367T) occurs with resistance to androgen deprivation. (A) Conversion from A → C in HSD3B1 occurs in 3 CRPC tumors from patients with homozygous wild-type inheritance. Sequence of cDNA clones from a fresh-frozen tumor (UTSW7) confirms expression of HSD3B1(1245C) transcript. (B) Three CRPC tumors from patients with heterozygous inheritance exhibit LOH of the wild-type HSD3B1(1245A) allele. Sequencing informative (heterozygous) adjacent 5’ (rs6203) and 3’ (rs34814922 and rs113096733) SNPs confirms LOH. (C) 3βHSD1 protein is abundant in tumors with LOH of the HSD3B1(1245A) allele but not tumors with heterozygous expression or homozygous HSD3B1(1245A) expression. Both tumors with LOH tested also express AR and PSA (20 μg protein loaded per lane for each tumor). (D) Somatic mutation converting A → C in HSD3B1 occurs in two LAPC4 xenograft tumors treated with abiraterone acetate (Abi) after orchiectomy and expression of HSD3B1(1245C) transcript encoding 3βHSD1(367T) is evidenced by sequencing cDNA clones from these tumors. Genomic sequence from two representative control tumors (CTRL#1 and CTRL#2) treated with orchiectomy alone is shown for comparison. All 37 cDNA clones from CTRL#1 and CTRL#2 have HSD3B1(1245A) transcript encoding 3βHSD1(367N). See also Figure S2 and Table S1.

Fig. 3

Genetic silencing of 3βHSD1(367T) impedes conversion of DHEA to DHT, induction of PSA and TMPRSS2 expression, and CRPC growth. (A) Stable lentiviral expression of two independent shRNA constructs against HSD3B1 (shHSD3B1 #1 and shHSD3B1 #2) silences 3βHSD1 protein expression in LNCaP. The 3βHSD1 protein was quantitated and normalized to cells expressing nonsilencing lentiviral vector (shCTRL) and β-actin. (B) Silencing 3βHSD1(367T) blocks flux from [³H]-DHEA (100 nM) to AD as well as further downstream conversion to 5α-dione and DHT. Cells were treated with [³H]-DHEA in triplicate and steroids were quantitated with HPLC at the designated time points. (C) Inhibition of AR-regulated genes. Cells were treated with the indicated concentration of DHEA for 24 hours, and gene expression was assessed by qPCR and normalized to shCTRL-infected cells treated with vehicle and the RPLP0 housekeeping gene. (D) Silencing 3βHSD1(367T) inhibits in vitro growth. Cells were grown in the presence of 20 nM DHEA or vehicle and growth for each cell line is normalized to vehicle for each designated day. (E) 3βHSD1(367T) depletion blocks CRPC growth in LNCaP xenografts. Mice underwent surgical orchiectomy and DHEA pellet implantation concomitantly when xenograft tumors reached a threshold volume of 100 mm³. Fifteen mice were initiated in each cohort, 7, 8 and 10 mice in shCTRL, shHSD3B1 #1, and shHSD3B1 #2 groups, respectively, achieved a tumor volume of 100 mm³ in eugonadal mice, underwent orchiectomy and were included in the CRPC analysis. The number of days from orchiectomy to tumor volume ≥ 600 mm³ is shown. In the comparisons of shCTRL vs shHSD3B1 #1 and shHSD3B1 #2, P = 0.002 and 0.003, respectively, using a log rank test. (F) 3βHSD1(367T) protein is regained in CRPC tumors that grow from LNCaP expressing shHSD3B1 #1 and shHSD3B1 #2. Immunoblot for 3βHSD1 and β-actin were performed on protein from the indicated LNCaP CRPC tumors. Error bars in B, C and D represent the SD for experiments performed in triplicate.

Fig. 4

Resistance to ubiquitination and proteosome-mediated degradation occurs with 3βHSD1(367T) which results in prolonged protein half-life. (A) 3βHSD1(367T) persists after inhibition of protein translation. LAPC4 cells were transiently transfected with constructs encoding for wild-type (N-HA) and (T-HA) protein and treated with cycloheximide (CHX) for the designated incubation times. Western blot with anti-HA antibody was performed, and signal was quantitated and normalized to time zero and β-actin. (B) Treatment with MG132 (10 μM; 8 hours) reverses 3βHSD1(367N) protein loss in LAPC4 and results in no 3βHSD1(367T) protein increase in LNCaP. (C) Proteosome inhibition with MG132 (10 μM; 8 hours) results in an increase in polyubiquitinated 3βHSD1(367N) protein in LAPC4 as evidenced by immunoprecipitation with an anti-ubiquitin antibody. (D) Loss of 3βHSD1(367T) vulnerability to proteosome-mediated degradation is explained by diminished susceptibility to ubiquitination. His-ubiquitin (His-ubi) was expressed with wild-type (N-HA) or (T-HA) protein in 293 cells, followed by pull down with Ni-agarose beads and anti-HA immunoblot.

Fig. 5

The ER-associated degradation (ERAD) pathway and AFMR regulate 3βHSD1 ubiquitination and degradation. (A, B) K70 and K352 ubiquitination on 3βHSD1(367N) is detectable by mass spectrometry. (C) K70, 352R mutant 3βHSD1(367N) is resistant to ubiquitination. K70R and K352R single and double mutant forms of N-HA were expressed with His-ubi in 293 cells, followed by pull down with Ni-agarose beads and anti-HA immunoblot. (D) Treatment with the ERAD inhibitor, Eeyarestatin I (EerI, 10μM), increases endogenous 3βHSD1 protein in LAPC4. (E) AMFR preferentially physically associates with wild-type protein (N-HA). Proteins were expressed in 293 cells, immunoprecipitated with anti-HA antibody, followed by immunoblot for AMFR. (F) Silencing the ubiquitin E3-ligase AMFR increases 3βHSD1 protein detected in LAPC4 cells. In contrast, genetically silencing the ubiquitin E3-ligase SKP2 has no detectable effect on 3βHSD1.

Fig. 6

3βHSD1(367T) increases metabolic flux from DHEA to DHT and elicits CRPC. (A) Transient expression of 3βHSD1(367T) (T, blue bars) leads to increased conversion from DHEA to AD and downstream steroids compared with 3βHSD1(367N) (N, red bars). LAPC4 cells were transfected with the indicated plasmid, treated with CHX, and cultured with [³H]-DHEA (100 nM); steroids were extracted and measured by HPLC at the designated time points (p-value = 0.023 for the difference in DHT synthesis by the N and T forms using by Student's t-test). (B) Transient transfection results in equivalent expression of both transcripts by qPCR. (C) Stable expression demonstrates increased activity of 3βHSD1(367T). Lentiviral constructs expressing luciferase (L), wild-type (N), or (T), were stably expressed (without CHX treatment) and flux from [³H]-DHEA to DHT was assessed, as described previously (p-value = 0.015 for the difference in DHT synthesis by the N and T forms using Student's t-test). (D) Expression of both enzyme transcripts by qPCR is comparable. (E) Increased flux from DHEA to DHT with stable expression of 3βHSD1(367T) leads to amplified expression of PSA in LAPC4. Cells stably expressing the designated constructs were treated with the indicated steroids for 48 hours. PSA expression induced by the DHT positive control is equivalent among the three cell populations. For B, D and E, expression is normalized to RPLP0 and vector, luciferase, or vehicle controls. Error bars represent the SD for experiments performed in triplicate. (F) Development of CRPC occurs more rapidly in LAPC4 xenografts stably expressing 3βHSD1(367T) as compared with 3βHSD1(367N). Time from subcutaneous injection of cells in each flank to tumor size = 50 mm³ is shown for each tumor that developed in a mouse flank (n = 40 mouse flanks in each group). P=0.017 for the comparison using a log rank test. (G) PSA expression is higher in CRPC tumors expressing 3βHSD1(367T) compared with 3βHSD1(367N) (p-value = 0.015 by Student's t-test). Expression is normalized to RPLP0. Bars represent the upper and lower quartiles of individual tumor values. See also Figure S3.