Targeting the KIF4A/AR Axis to Reverse Endocrine Therapy Resistance in Castration-resistant Prostate Cancer

Prostate cancer is one of the most common malignancies and ranks second among causes of male cancer-related death worldwide (1). Androgens play an essential role in the growth of prostate cancer cells. Androgen deprivation therapy (ADT) is the main method for the treatment of advanced prostate cancer. However, after an average of 2 years of ADT treatment, the tumor often progresses from castration-sensitive prostate cancer (CSPC) to castration-resistant prostate cancer (CRPC; ref. 2). As the androgen receptor (AR) signaling pathway is an important pathway for prostate cancer survival and progression, targeting AR expression and inhibiting its activity has become an effective way to treat CRPC. Enzalutamide and abiraterone are novel androgen receptor (AR) inhibitors that are used to treat patients with metastatic CRPC after chemotherapy (3). However, prostate cancer can become therapy resistant through upregulation of androgen synthesis, mutation, and abnormal expression of AR, AR splice variants (AR-Vs), and neuroendocrine transitions. Therefore, revealing the complicated molecular mechanism of endocrine therapy resistance of prostate cancer is an urgent global concern.

AR-V7 is the most common AR splice variant. Levels of AR-V7 are low before treatment with enzalutamide or abiraterone but increase significantly after progression on either agent (4). A large-scale clinical trial found that AR-V7 expression in circulating tumor cells of patients with prostate cancer was associated with shorter survival and was also closely related to resistance to endocrine therapy drugs (5). Increased AR-V7 expression may represent one of the mechanisms of resistance to these agents. However, the precise role of AR-V7 in the progression of CRPC is still unclear.

KIF4A, a member of the kinesin superfamily (KIFs), is located on chromosome Xq13.1 and encodes a protein consisting of 1232 amino acids. KIF4A is involved in multiple cellular activities, particularly spindle formation and centrosome assembly in mitosis (6), chromosome concentration and separation (7), and DNA damage repair (8). KIF4A plays a critical role in a variety of tumors, such as lung cancer (9), oral cancer (10), liver cancer (11), and colorectal carcinoma (12). Estrogen regulates the expression of KIF4A through the estrogen receptor (ERα) in breast cancer, whereas ANCCA, a coregulator of ERα, is involved in the regulation of KIF4A (13). KIF4A is abundantly expressed in prostate cancer, and this expression is associated with poor prognosis in patients with prostate cancer (14). These findings suggest that KIF4A may have functions that contribute to prostate cancer progression, but whether KIF4A can regulate the AR signaling pathway remains unclear.

In this study, we demonstrate for the first time that KIF4A is positively correlated with AR levels in prostate cancer. Patients with higher KIF4A expression have worse survival. AR binds to the promoter region of KIF4A and activates its transcription. In addition, KIF4A binds and stabilizes the AR and AR-V7 proteins to modulate the AR pathway in prostate cancer cells. Furthermore, our results indicate that KIF4A promotes human CSPC and CRPC cell growth through AR-dependent signaling. Most importantly, we show that depletion of KIF4A reverses enzalutamide resistance and enhances the sensitivity of CRPC cells to endocrine therapy.

A human prostate cancer tissue microarray (TMA) was purchased from Alenabio (PR1921b). All detailed clinical information including pathology, diagnosis, stage, Gleason scores, and PSA level is freely available on the Web (http://www.alenabio.com/public/details?productId=59058&searchText=). The following antibodies were used: anti-KIF4A (1:50; Proteintech) and anti-AR (1:100; sc-816; Santa Cruz Biotechnology). The TMA was evaluated using by scoring the staining intensity in a range of 0 to 3 (0 = absent, 1 = weak, 2 = intermediate, 3 = strong) and the percentage of positively stained cells in a range of 0 to 4 (0 ≤ 5%, 1 = 5%–25%, 2 = 25%–50%, 3 = 50%–75%, 4 ≥ 75%). The two scores were multiplied to obtain an immunoreactivity score (IRS) ranging from 0 to 12. The IHC data were evaluated by two blinded pathologists.

293T, LNCaP, C4-2, and 22Rv1 cells were purchased from ATCC. 293T cells were maintained in DMEM. LNCaP, C4-2, and 22Rv1 cells were maintained in RPMI1640 medium. All media were supplemented with 10% FBS and 1% penicillin–streptomycin solution. All cells were cultured at 37°C in an incubator containing 5% CO₂. Drug concentrations (unless otherwise indicated) were enzalutamide (20 μmol/L), bicalutamide (10 μmol/L), DHT (10 nmol/L), and cycloheximide (CHX; 10 μmol/L).

The starting treatment concentration of enzalutamide in the parental C4-2 cell culture medium was 0.2 μmol/L. At this concentration, the cells were stably passaged three times. The drug concentration was then increased, and the culture was continued. The concentration gradients of enzalutamide were 0.4, 0.8, 1, 2, 4, 8, 10, 20, and 40 μmol/L. After 4 months of induction, the stable enzalutamide-resistant cell line was obtained and named C4-2-ENZ-R.

The pLent-GFP shRNA targeting KIF4A and pENTER-Flag KIF4A were purchased from Vigenebio. GV298-shRNA targeting CHIP, MYC-CHIP, eGFP-AR, and HA-AR-V7 were purchased from Genechem. pCDNA3.1-3 × Myc-Ub was purchased from miaolingbio. Briefly, 3 × 10⁵ cells were seeded in six-well plates and transfected with 2 μg of plasmid using Lipofectamine 2000 (Invitrogen) according to the manufacturer's procedure.

Lentivirus was packaged by cotransfection of the constructs with the plasmids pMD2.G, pRRE, and pRSV/REV in 293T cells. The supernatants collected at 72 and 96 hours after transfection were filtered by a 0.45-μm filter, and the filtrates were then added to prostate cancer cells. After being selected with puromycin (1 μg/mL), the protein expression level was verified by Western blotting.

IHC was conducted as described previously (15). Tissues were sequentially fixed in formalin, dehydrated, and embedded in paraffin. Then, IHC was conducted by incubating the tissue sections with primary antibodies including KIF4A, PSA, or Ki67 overnight at 4°C. Subsequently, after washing three times with PBS, the sections were incubated with the secondary antibody (1:200; GB23303; Servicebio, Inc.) at room temperature for 2 hours. Immunofluorescence (IF) was performed as described previously (16). Cells were fixed in 4% paraformaldehyde, permeated by 0.3% Triton X-100, and blocked with 3% BSA for 1 hour at 37°C, followed by incubation with primary KIF4A and AR antibodies.

The ChIP assay was performed according to the protocol of the ChIP Assay Kit (CST). 22Rv1 and C4-2 cells were cultured in 10-cm dishes. Then, the chromatin in the cells was cross-linked by adding formaldehyde to a final concentration of 1% at room temperature for 10 minutes, followed by washing twice with 2 mL of ice-cold PBS containing protease inhibitors, lysis in ChIP lysis buffer, and sonication to completely lyse nuclei. The digested, cross-linked chromatin was diluted with ChIP buffer. A 10-μL sample of the diluted chromatin was removed as a 2% input sample. The remaining 500 μL of diluted chromatin was incubated with anti-AR or anti-IgG at 4°C overnight with rotation. After elution of chromatin, reversal of cross-links and DNA purification, qRT-PCR was performed to amplify the potential AR binding sites on the KIF4A promoter region.

A total of 3 × 10³ cells were added to each well of a 96-well plate. The cell proliferation rate was determined using Cell Counting Kit-8 (CCK-8; Dojindo) every 24 hours. Briefly, 10 μL of CCK-8 solution was added to each well. After 4 hours, the absorbance was measured at 450 nm.

The collected cells were lysed in RIPA buffer containing protease inhibitor. The lysate was kept on ice for 30 minutes and centrifuged at 12,000 rpm for 5 minutes. Then, the supernatant was collected in two parts: a small amount of lysate was taken as input, and the remaining lysate was incubated with 2 μg of the corresponding antibody and 30 μL of agarose beads at 4°C overnight. The immune complexes were centrifuged at 3,000 rpm for 2 minutes. The supernatant was carefully discarded, and the agarose beads were washed three to four times with 200 μL of lysis buffer. Finally, 40 μL of RIPA lysate and 10 μL of loading buffer were added to the beads and boiled for 10 minutes. For Western blotting, total proteins were separated by 10% SDS-PAGE and transferred to polyvinylidene fluoride (PVDF) membranes. The PVDF membranes were blocked and then incubated with primary antibodies. The antibodies used in Western blotting were KIF4A (ab124903; Abcam), AR (sc-7305; Santa Cruz Biotechnology), AR-V7 (ab198394; Abcam), Flag (20543-1-AP; ProteinTech), MYC (60003-2-Ig; ProteinTech), HA (66006-2-Ig; ProteinTech), eGFP (66002-1-Ig; ProteinTech), and β-actin (AC026; ABclonal). The proteins were visualized using ChemiDoc-XRs+ (Bio-Rad).

Total RNA was extracted from cells using TRIzol reagent (Invitrogen) and then reverse transcribed by the PrimeScript RT reagent Kit with gDNA Eraser (Takara). Real-time PCR was conducted using Maxima SYBR Green/ROX qPCR Master Mix (Thermo Fermentas) on the ABI-7500 qRT-PCR system. The primers used for qRT-PCR are listed in Supplementary Table S1.

Cells were seeded in a 24-well plate at 70% confluence and transiently transfected with 0.8 μg of expression vector plasmids and 0.4 μg of promoter reporter plasmids. The fluorescence intensity was measured after 48 hours. The luciferase activity of the gene promoter was normalized to Renilla luciferase activity as an internal standard control. The plasmids of KIF4A-luc and mutated KIF4A-luc were designed and synthesized by Genechem.

Three- to 4-week-old castrated male nude mice were purchased from Beijing Vital River Laboratory Animal Technology Co., Ltd. Approximately 5 × 10⁶ C4-2-ENZ-R cells with knockdown of KIF4A or control cells suspended in 100 μL of serum-free medium were injected subcutaneously into the armpit on the right side of the mice. Tumor volumes (V) were measured every 5 days based on measurements of length (L) and width (W) and calculated as V = (L × W²⁾/2. When the tumors reached 100 to 200 mm³, the mice were randomized equally, treated with vehicle or enzalutamide (10 mg/kg) twice per week, and sacrificed approximately 40 days later.

All statistical analyses were performed using Prism 5.0 (GraphPad) and SPSS 22.0 (IBM Corporation). All in vitro experiments were repeated three times. All data are presented as the mean ± SD. Survival information was verified by Kaplan–Meier analysis and compared using the log-rank test. A two-tailed unpaired Student t test was used to determine the P values, which were considered significant at less than 0.05.

AR is a ligand-activated transcription factor that plays critical roles in normal prostate development and prostate tumorigenesis (17). The growth of advanced prostate cancer (both CSPC and CRPC) depends on AR signaling. In this study, by screening four independent prostate cancer gene sets (TCGA, Arredouani, Grasso, and Varambally; refs. 18–20), three genes (KIF4A, BIRC5, ATP8A2) that correlated with AR were selected on the basis of their significant differences in expression in prostate cancer tissues compared with normal tissues (logFC > 1.5, P < 0.05; Fig. 1A; Supplementary Fig. 1A). Furthermore, Kaplan–Meier survival analysis and log-rank tests were conducted to determine whether the disease-free survival (DFS) and overall survival (OS) of patients were associated with KIF4A, BIRC5, and ATP8A2 expression in tumors. As shown in Fig. 1B, only patients with high levels of KIF4A mRNA expression had worse DFS (P = 0.0002) and OS (P = 0.033) than those with low KIF4A expression. Higher BIRC5 or ATP8A2 expression did not appear to affect OS in prostate cancer (Fig. 1B). A comprehensive consideration of these results led us to focus on KIF4A for further study.

Previous studies have shown that increased KIF4A mRNA expression is a potential prognostic factor in prostate cancer (14). KIF4A may play a key role in prostate cancer progression. Next, to further confirm the previous findings and the interaction between KIF4A and AR, we examined KIF4A and AR protein levels with IHC staining using a tissue array containing 160 cases of human prostate cancer and 32 adjacent normal tissue specimens. Similar to a previous result, we observed that KIF4A was positively correlated with AR expression in human normal prostate tissues (r = 0.37, P < 0.05) and human prostate cancer tissues (r = 0.55, P < 0.0001; Fig. 1C). KIF4A protein levels were significantly higher in the prostate cancer tissues than in the normal tissues and were related to tumor stage and PSA levels (Supplementary Figs. S1B–S1D). Taken together, these results suggested a functional interaction between KIF4A and AR during prostate cancer progression.

AR acts as a transcription factor to regulate the expression of its downstream target genes and promote prostate cancer progression (21). Given that KIF4A and AR are positively correlated in prostate cancer, there is a possibility that AR may transcriptionally regulate KIF4A expression. To explore this possibility, LNCaP cells were treated with DHT for 12 hours. As shown in Fig. 2A, KIF4A protein levels were elevated in LNCaP cells upon androgen treatment. To confirm that this effect of androgen occurred through AR, we directly modulated the levels of AR in PC3 and C4-2 cells. As expected, the changes in KIF4A protein levels were in accordance with the change in AR when we overexpressed AR in PC3 cells and C4-2 cells or downregulated AR in C4-2 cells (Fig. 2B–D). These results indicated that KIF4A is elevated by AR activation.

In the nucleus, AR binds to specific DNA sequences termed androgen response elements (AREs) in the promoter regions of target genes, such as prostate-specific antigen (PSA) and transmembrane protease serine 2 (TMPRSS2; refs. 22, 23). Next, we identified a canonical ARE ∼531 base pairs upstream of the KIF4A transcription start site (Fig. 2E). We then cloned luciferase reporter constructs into a pGL3 plasmid containing the putative ARE (ARE-luc) or its mutant (AREmut-luc; Fig. 2E). As expected, ectopic AR expression significantly activated ARE-luc activity with or without DHT treatment. However, no activation of AREmut-luc activity was observed upon overexpression of AR (Fig. 2F). To further verify the binding sites on the promoter region, we analyzed and designed primers covering different sequences of interest in the promoter of KIF4A, which were denoted P1, P2, P3, P4, P5, and P6 (Fig. 2G). ChIP assay revealed a significant increase in AR recruitment to the P5 region compared with the IgG control in C4-2 and 22Rv1 cells (Fig. 2H–I). Furthermore, AR recruitment to the P5 region was significantly increased upon DHT treatment compared with the IgG control in castration-sensitive LNCaP cells (Supplementary Fig. S2A). These results demonstrated that AR directly activates KIF4A transcription in prostate cancer.

To explore the role of KIF4A in prostate cancer, we knocked down KIF4A in C4-2 cells using KIF4A-specific shRNA and observed that the shRNA decreased the expression levels of AR (Fig. 3A). Furthermore, the transcript levels of PSA and TMPRSS2, AR canonical target genes, were also decreased (Fig. 3B). However, the transcript level of AR was not downregulated by KIF4A knockdown (Fig. 3B). AR-Vs lacking the ligand-binding domain, such as AR-V7, have been implicated in the pathogenesis of CRPC and in mediating resistance to new endocrine therapies that target the androgen axis (5). Next, we knocked down KIFA in 22Rv1 cells, another CRPC cell line expressing high levels of AR-V7. As shown in Fig. 3D, the expression levels of AR and AR-V7 were also decreased by KIF4A knockdown in 22Rv1 cells. However, the mRNA levels of AR and AR-V7 were not decreased (Fig. 3E). Because the loss of KIF4A decreased AR and AR-V7 protein levels and inhibited AR signaling without downregulation of AR and AR-V7 transcript levels, we hypothesized that KIF4A regulates the AR and AR-V7 proteins and their functions through a posttranslational modification. To test this possibility, we first performed immunoprecipitation experiments. AR was immunoprecipitated from C4-2 cell lysates with an anti-AR antibody and analyzed for KIF4A binding by Western blot analysis. The results showed that endogenous KIF4A coimmunoprecipitated with AR (Fig. 3C). KIF4A interacted with AR-V7 proteins in 22Rv1 cells (Fig. 3F; Supplementary Fig. S2B). To prove their direct interaction, Flag-KIF4A and eGFP-AR or HA-AR-V7 were coexpressed in 293T cells. Flag-KIF4A was immunoprecipitated from cell lysates with an anti-Flag antibody, and eGFP-AR was pulled down and detected by Western blot analysis (Fig. 3G). When we used an anti-eGFP antibody to precipitate eGFP-AR, Flag-KIF4A was also observed in the immunoprecipitate by using an anti-Flag antibody (Fig. 3H). Furthermore, HA-AR-V7 was pulled down by exogenous KIF4A (Fig. 3I). Moreover, IF assays confirmed that KIF4A and AR had obvious colocalization (Fig. 3J–K).

On the basis of the above results, we expected that KIF4A controls the function of AR and AR-V7 by an additional mechanism that may involve regulation of AR and AR-V7 expression at the posttranslational level. To test this hypothesis, we constructed the enzalutamide-resistant cell line C4-2-ENZ-R (Supplementary Fig. S3). We knocked down KIF4A expression by using shRNA in C4-2-ENZ-R cells and treated cells with CHX to block de novo protein synthesis. We found that downregulation of KIF4A significantly accelerated the degradation of AR (Fig. 4A). In addition, the half-life of AR and AR-V7 protein was shortened upon KIF4A knockdown (Fig. 4C; Supplementary Fig. S4A). Similarly, we overexpressed KIF4A in 22Rv1 cells and treated cells with CHX for different durations. The results showed that overexpression of KIF4A reduced the rate of degradation of AR-V7 and AR and prolonged the half-life of the AR-V7 and AR proteins (Fig. 4B and D; Supplementary Fig. S4B). In addition, KIF4A overexpression decreased the ubiquitination levels of endogenously expressed AR and AR-V7 in C4-2-ENZ-R and 22Rv1 cells (Fig. 4E and F), suggesting that KIF4A contributes to AR and AR-V7 stability by blocking its ubiquitination.

Several studies have reported that AR levels are regulated by the ubiquitin-proteasome degradation pathway (24–27). Several ubiquitin E3 ligases have also been implicated in AR-V7 degradation (28, 29). It appears that only CHIP E3 ligase interacts with both the AR and AR-V7 proteins. This finding led us to expect that CHIP is involved in KIF4A depletion-induced AR and AR-V7 degradation. First, the effects of KIF4A on CHIP-mediated degradation were examined. In 293T cells, the coexpression of KIF4A and CHIP resulted in significant increases in the protein levels of AR and AR-V7 compared with overexpression of CHIP alone (Fig. 4G and H). This result suggested that KIF4A might affect CHIP-mediated ubiquitination of AR and AR-V7. Next, the potential role of KIF4A in AR and AR-V7 ubiquitination was examined. In 293T cells, CHIP-dependent ubiquitination of AR was enhanced by the overexpression of CHIP protein (Fig. 4I), but KIF4A expression rescued CHIP-mediated AR ubiquitination (Fig. 4I). Indeed, the amount of pulled down CHIP protein was significantly decreased by KIF4A expression (Fig. 4I). Similarly, AR-V7 ubiquitination was inhibited by KIF4A expression, and the binding of CHIP protein to AR-V7 was also reduced (Fig. 4J). Meanwhile, CHIP-dependent ubiquitination of AR and AR-V7 was decreased by the deletion of endogenous CHIP protein in 293T cells. And additional KIF4A expression further reduced the ubiquitination level of AR and AR-V7 (Supplementary Figs. S5A and S5B). Taken together, these results suggest that KIF4A inhibits CHIP-mediated ubiquitination of AR and AR-V7 by blocking the interaction between AR or AR-V7 and CHIP.

Given that KIF4A interacts with AR and inhibits its degradation, we assumed that the inhibition of cell growth by silencing KIF4A is dependent on AR. To further investigate this possibility, we transfected AR plasmids in LNCaP and C4-2 cells in which KIF4A was silenced. The results showed that the inhibition of cell proliferation induced by KIF4A knockdown was reversed by transient overexpression of AR in LNCaP-stable cell lines (Fig. 5A). Similarly, overexpression of AR abolished the inhibition of cell proliferation induced by downregulation of KIF4A in C4-2 cells (Fig. 5B). Also, there was no growth inhibitory effect upon KIF4A knockdown in AR-negative cells, PC3 and DU145 (Supplementary Figs. S6A and S6B). These results demonstrated that AR plays a critical role in KIF4A-regulated prostate cancer cell growth.

Bicalutamide, a pharmaceutical drug commonly used as an antiandrogen therapy to treat recurrent prostate cancer, is a competitive inhibitor of AR. Enzalutamide, a novel AR signaling inhibitor, blocks the growth of CRPC in cellular model systems and was shown in a clinical study to increase survival in patients with metastatic CRPC. The rapid development of therapy resistance in patients with prostate cancer receiving bicalutamide and enzalutamide treatment is becoming a major clinical challenge (30, 31). The sustained expression of AR and AR-V7 is a hallmark of endocrine therapy resistance in prostate cancer. To further investigate whether KIF4A contributes to bicalutamide and enzalutamide resistance in prostate cancer cells, we examined the sensitivity of prostate cancer cells to these drugs under different conditions through CCK8 assays. As shown in Fig. 5C, we observed that knockdown of KIF4A enhanced inhibition of growth by bicalutamide in castration-sensitive LNCaP cells. Similarly, downregulation of KIF4A increased inhibition of growth by enzalutamide in castration-resistant C4-2 cells and enzalutamide-resistant C4-2-ENZ-R cells (Fig. 5D–E). Overexpression of KIF4A alleviated the inhibition of growth by enzalutamide in C4-2 cells (Fig. 5F). Furthermore, we have established the stably expressed additional AR cells named PC3-AR. As shown in Supplementary Fig. S6E, we observed that knockdown of KIF4A enhanced inhibition of growth by Enzalutamide in AR-positive PC3-AR cells. But knockdown of KIF4A did not enhance inhibition of growth by Enzalutamide in AR-negative cells (Supplementary Figs. S6C and S6D). Finally, Enzalutamide did not inhibit the growth of C4-2-ENZ-R cells, and overexpression of KIF4A had same effect (Supplementary Fig. S6F). These results indicated that inhibition of KIF4A potentiates the effects of enzalutamide and bicalutamide in prostate cancer cells. KIF4A knockdown reverts endocrine therapy-resistant CRPC progression.

Our previous studies showed that KIF4A knockdown effectively reverts endocrine therapy resistance in CRPC cells. To further confirm this function of KIF4A, we transplanted C4-2-ENZ-R cells with stable knockdown of KIF4A into castrated male SCID mice. The mice were then treated with enzalutamide (10 mg/kg) twice a week. As shown in Fig. 6A–C, there was no effect of enzalutamide treatment alone on the tumor growth and tumor weight of the mice bearing tumors without KIF4A knockdown. By contrast, KIF4A knockdown induced inhibition of tumor growth and tumor weight by enzalutamide treatment in C4-2-ENZ-R tumors, indicating that KIF4A knockdown can restore enzalutamide treatment sensitivity. The knockdown efficiency of KIF4A was detected by qRT-PCR. The mRNA level of KIF4A was significantly downregulated upon stable KIF4A knockdown (Fig. 6D). Similarly, the protein level of KIF4A was verified in tumor xenografts by IHC assays (Fig. 6F). Next, we determined the expression of the AR target genes PSA and TMPRSS2 in tumor tissues. Similar results were observed. As shown in Fig. 6E, there was no effect of enzalutamide treatment alone on PSA and TMPRSS2 expression in mice bearing tumors without KIF4A knockdown. By contrast, KIF4A knockdown induced inhibition of PSA and TMPRSS2 expression by enzalutamide treatment in C4-2-ENZ-R tumors (Fig. 6E). Finally, IHC staining of tissue sections of tumors from the four groups was performed to evaluate the protein expression of KIF4A, PSA, and Ki67. The representative images show that the combination of KIF4A knockdown and enzalutamide usage resulted in weaker staining intensity of PSA and Ki67 compared with the KIF4A knockdown-only group, indicating that deletion of KIF4A enhanced the inhibition of AR activity and proliferation ability by enzalutamide in C4-2-ENZ-R tumors (Fig. 6G; Supplementary Fig. S7A and S7B). Taken together, these results suggested that KIF4A knockdown effectively reversed endocrine therapy resistance in CRPC and that KIF4A is a novel therapeutic target for CRPC (Fig. 6H).

The KIF proteins are involved in many essential cellular biological functions, including mitosis and transport of intracellular vesicles and organelles (32). Increasing evidence indicates that KIF members participate in the genesis and development of human cancers (33–36). KIF4A, a member of the KIF family, has been reported to be abnormally expressed and to play a critical role in the progression of various solid cancers (9–13). However, the expression and function of KIF4A in prostate cancer have not been fully researched.

In this study, we demonstrated that KIF4A was upregulated in prostate cancer tissues compared with paired normal tissues. Moreover, elevated KIF4A expression was significantly correlated with several clinicopathologic parameters, such as tumor stage and PSA levels. In addition, high KIF4A expression was associated with poorer OS and DFS of patients with prostate cancer, which was partially consistent with a previous study in prostate cancer (14). These findings revealed that KIF4A plays an essential role in the progression of prostate cancer and could act as a potential clinical prognostic indicator for patients with prostate cancer.

The AR-mediated androgen signaling pathway plays an important role in the development of prostate cancer (22). Upon androgen binding, AR dissociates from heat shock proteins and translocates to the nucleus. The AR dimer then binds to AREs in the promoter regions of androgen-dependent genes, thereby activating/inhibiting their transcription (37). PSA and TMPRSS2 are two canonical AR target genes. A binding site at 13 kb upstream of the TMPRSS2 transcription start site is necessary for AR regulation of the TMPRSS2 gene (38). A group confirmed that the upstream sequence of the PSA promoter (−539-320 bp) is necessary for androgen regulation. AR regulates the expression of PSA by interacting with the 5′-AGAACAgcaAGTGCT-3′ sequence (39). Here, we identified a half ARE site in the promoter region of the KIF4A gene. AR ChIP analysis revealed that KIF4A might be upregulated directly via AR binding to the KIF4A promoter. KIF4A expression is enhanced by AR upon androgen treatment. These facts strongly indicate that KIF4A is a direct AR downstream target that is upregulated by androgens.

Infinite proliferation of cancer cells is a hallmark of progression of prostate cancer (40). Previous studies have showed that KIF4A is involved in regulating the proliferation of cancer cells (11). Our data revealed that KIF4A depletion inhibits prostate cancer cell proliferation. More importantly, the regulation of prostate cancer cell proliferation by KIF4A is dependent on AR. These results reveal a novel mode of AR-dependent signaling that is involved in regulating cell biological behavior. The importance of the involvement of KIF4A in oncogenic regulation was reinforced by our finding that KIF4A bound with AR/AR-V7 and inhibited their ubiquitination and degradation. The E3 ligase CHIP forms a complex with AR/AR-V7 and participates in protein ubiquitination. KIF4A blocked CHIP and AR/AR-V7 complex formation, leading to AR/AR-V7 protein stabilization. Introducing KIF4A protein restored CHIP-mediated AR/AR-V7 degradation. Our findings on complex formation by KIF4A and CHIP/AR or CHIP/AR-V7 are critical as this mechanism may represent a general ubiquitin–proteasome mechanism for the regulation of AR/AR-V7 protein stability that may be involved in endocrine therapy resistance in prostate cancer progression.

The upregulation of AR protein is a hallmark of CRPC and seems to be an adaptive response to ADT (41). An increase in AR protein levels is observed in most refractory cases (41). Several mechanisms account for increased AR levels because some factors confer stability of AR in CRPC or, importantly, in therapy-resistant prostate cancer. There is evidence that AR proteins are degraded by the ubiquitin–proteasome system (42). A number of E3 ligases have been implicated in AR regulation by protein degradation. RNF6, an E3 ligase, is involved in regulating the AR protein and induces AR ubiquitination to increase AR transcriptional activity (43). Speckle-type POZ protein (SPOP) is a ubiquitin E3 ligase that regulates AR protein stability (26). Mutations in SPOP cause failure of SPOP to interact with AR, leading to stabilization of AR in prostate cancer cells. Other E3 ligases that regulate ubiquitination of AR include Siah2, PIAS1, MDM2, SKP2, and CHIP (24, 44–47). Apart from E3 ligases, some proteins affect AR stability. Deleted in breast cancer 1 (DBC1), for instance, binds and stabilizes AR (48). PC-1 reduces AR stability by enhancing the interaction between AR and CHIP, resulting in degradation of AR by proteasomes (49). BMI1, a polycomb group protein (PcG), stabilizes AR by competitive inhibition of MDM2 and thereby decreases proteasome degradation (50). Our present data suggest that KIF4A stabilizes AR by physical interaction with AR and inhibiting the CHIP binding activity required for degradation. We hypothesize that binding KIF4A could mask the CHIP binding site on AR. However, the mechanism by which KIF4A stabilizes AR needs to be further clarified. In summary, our results further reveal the regulatory networks conferring AR protein stability.

AR-V7, a major splice variant of AR, is constitutively expressed in refractory prostate cancer and can drive prostate cancer progression even under enzalutamide treatment (51). AR-V7 mRNA is generated from an alternative RNA splicing process that is enhanced under castration conditions (52). The AR-V7 protein is then stabilized and perform its functions in prostate cancer cells. However, little is known about the regulatory mechanism of the expression and stability of AR-V7 protein. Protein phosphatase-1 (PP-1) and AKT kinase can govern AR-V7 phosphorylation status. PP-1 can phosphorylate AR-V7 at serine 213, preventing MDM2-mediated AR-V7 protein degradation by the ubiquitin–proteasome pathway (28). BMI1 can also stabilize AR-V7 by physical interaction at the N-terminus of AR and thereby inhibit MDM2-mediated degradation (50). A recent study demonstrated that DBC1 could stabilize AR-V7 by facilitating the DNA-binding activity of AR-V7 and inhibiting CHIP-mediated ubiquitination and degradation of AR-V7 by competing with CHIP for AR-V7 binding (53). Our functional studies show that KIF4A binding to AR-V7 acts as a positive regulator for AR-V7 stability and functions by inhibiting the E3 ligase activity of CHIP. As a consequence, downregulation of KIF4A decreases AR-V7 protein levels. In addition, CHIP can block the interaction of the AR/V7 co-chaperone protein, HSP70, with AR-V7 in enzalutamide- and abiraterone-resistant prostate cancer cells, thus leading to AR-V7 protein degradation (54). This regulatory pathway provides an in-depth understanding of the AR-V7 regulatory network.

In the past few years, much effort has been made to improve ADT. The development of novel anti-androgenic drugs such as abiraterone and enzalutamide has resulted in promising effects (55, 56). However, resistance to these drugs, especially enzalutamide, occurs, and models have been developed to investigate the underlying mechanisms. Herein, we investigated the potential of KIF4A inhibition to improve current prostate cancer treatment strategies. Functional experiments showed that KIF4A knockdown increases the inhibition of the growth of prostate cancer cells by ADT drugs, including bicalutamide and enzalutamide. Most importantly, using our enzalutamide-resistant cell model, the in vivo study revealed that targeting KIF4A significantly inhibits CRPC tumor growth, and the use of enzalutamide in combination with KIF4A knockdown further attenuated tumor growth compared with a single treatment.

In conclusion, the present data provide a strong theoretical basis for clinical KIF4A targeting either alone or in combination with ADT drugs, such as bicalutamide or enzalutamide, to treat CRPC overexpressing AR/AR-V7 and to improve enzalutamide treatment in prostate cancer. Targeting of the KIF4A/AR axis could be used to reverse endocrine therapy resistance in CRPC.

No potential conflicts of interest were disclosed.

Conception and design: Q. Cao, X. Zhang

Development of methodology: Q. Cao, C. Wang, X. Zhang

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): Q. Cao, H. Ruan, C. Wang, X. Yang, K. Wang, G. Cheng, T. Xu, W. Xiao, Z. Xiong, D. Zhou, X. Zhang

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): K. Chen, Q. Cao, Z. Song, H. Ruan, C. Wang, M. Yang

Writing, review, and/or revision of the manuscript: K. Chen, Q. Cao, H. Ruan, H. Yang, X. Zhang

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): Q. Cao, Z. Song, L. Bao, D. Liu, X. Zhang

Study supervision: Q. Cao, H. Yang, X. Zhang

This work was supported by grant from National Natural Sciences Foundation of China (No. 81672524, No. 81672528, No. 31741032, No. 81874090), Hubei Provincial Natural Sciences Foundation of China (2018CFA038), Foundation of Health and Family Planning Commission of Hubei Province in China (WJ2017M124), Independent Innovation Foundation of Huazhong University of Science and Technology (2016YXZD052, No. 118530309) and Clinical Research Physician Program of Tongji Medical College, Huazhong University of Science and Technology (No. 5001530015).

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.