hnRNP K is frequently overexpressed in prostate tissues and cell lines
To detect hnRNPK protein levels in prostate cancer, we first performed IHC analysis on 22 primary prostate adenocarcinoma. We found that hnRNPK was mainly located in the nucleus of PrCa cells, and it levels were high in 13 cases (59.1%). Moreover, hnRNPK expression was obviously higher in intermediate/high-risk tissues as compared to low-risk tissues (Fig. 1A). Further, we analyzed hnRNPK expression in lysates from 27 freshly harvested tissue samples of PrCa patients by Immunoblotting compared with matched noncancerous tissues. Among 27 randomly selected PrCa and paired noncancerous prostate tissues, 15 tumors (55.6%) showed an increase in hnRNPK protein (Fig. 1B). Moreover, we detected hnRNPK mRNA expression in 53 paired PrCa tissues and its levels were significantly higher than adjacent noncancerous prostate tissues (Fig. 1C). Additionally, a public data set (Gene Expression Omnibus, GSE70770) containing 35 PrCa tissues and 14 normal prostate tissues also showed that hnRNPK mRNA expression was upregulated in PrCa tissues (Fig. 1D). To further investigate the clinicopathological and prognostic significance of hnRNPK levels in PrCa patients, the mRNA levels of hnRNPK in above cohort of 53 PrCa tissues in which absence of androgen deprivation therapy, chemotherapy, radiotherapy or other anticancer treatment before surgery were classified according to age, Serum PSA, Gleason score, pT stage, Lymph node metastasis, Seminal vesicle invasion and Biochemical recurrence. We found that high hnRNPK mRNA expression was significantly associated with higher PSA levels and biochemical recurrence (P<0.05, Supplement Table 1). This prognostic value was also confirmed using a larger cohort of 203 PrCa patients retrieved from the GSE70770 database, high expression of hnRNPK was associated with higher biochemical recurrence rates (P<0.05, Fig. 1E). As shown in Fig. 1F and G, hnRNPK mRNA and protein expression were remarkably high in C4-2, DU145 and PC-3 cell lines compared with the normal prostate epithelial cells RWPE-1 or other PrCa cell lines. Thus, hnRNPK expression was frequently higher in PrCa tissues and cell lines than in the normal ones, which predicted poor prognosis in PrCa patients.
Knockdown of hnRNPK inhibits cell growth and cell cycle progression of prostate cells
To determine the biological functions for hnRNPK, we firstly performed loss-of-function experiments and knocked down hnRNPK by using short hairpin RNAs (shRNAs) in PC-3 and DU145 cells, respectively. The efficiency of knockdown were confirmed by mRNA and protein levels with RT-qPCR and Immunoblots (Supplement Fig. 1). The results of MTS assay showed that the proliferation capacity of PC-3 and DU145 cells was significantly reduced after silencing of hnRNPK experssion (Fig. 2A and 2B). Colony formation assays further confirmed significantly inhibited of cellular growth in both PrCa cells line following hnRNPK silencing (Fig. 2C and D). Moreover, Flow cytometry results indicated that the proportion of cells in the G0/G1 phase was significantly higher and the proportion of cells in the S phase was significantly lower in hnRNPK-silenced cells than in the control cells (Fig. 2E and F). Consistently, control sh-NC DU145 cells and the corresponding stable hnRNPK-silenced cells were inoculated into BALB/C athymic mice. As shown in Fig. 2G and H, tumors formed by the hnRNPK-silenced cells were retarded in size and weight than those formed from the control cells. Then the relative expression of hnRNPK in xenograft tumor tissue were verified by RT-qPCR (Fig. 2I). Collectively, these results indicate that hnRNPK inhibits PrCa cell proliferation in vitro and vivo via its effects on the cell cycle and may play oncogenic roles in prostate cancer progression.
miR-206 and miR-613 expression are downregulated in prostate tissues and cell lines and directly targets hnRNPK
Among the 53 randomly selected paired tissues from primary PrCa patients (Same cases above-mentioned), miR-206 and miR-613 expression were found to be reduced in tumor tissues when compared with paired noncancerous tissues (Fig. 3A and 3C), which further confirmed by using two public datasets, GSE21036 and GSE60117, respectively (Fig. 3B and 3D). Moreover, we analyzed miR-206 and miR-613 mRNA expression in six PrCa cell lines and found that miR-206 in C4-2, VCap, PC-3 and DU145 cell lines (Fig. 3E) and miR-613 in C4-2, LNCap, PC-3 and DU145 cell lines (Fig. 3F) had lower levels than in the normal prostatic cell line RWPE-1. Further we used TargetScan7.1 algorithms, a bioinformatic tool for microRNA target prediction to identify additional novel targets of miR-206 and miR-613, and found both of them could bind to the 3’-UTR of hnRNPK mRNA(Fig. 3G and 3H). Therefore, it is possible that both of miRNA inhibits hnRNPK expression in PrCa tissues and cell lines by directly binding to its 3’-UTR region. By employed a dual-luciferase reporter system, we subcloned 3’-UTR of hnRNPK mRNA including the predicted miR-206 and miR-613 recognition site (Wt) or the mutated sequences (Mut) into the pGL3 vector, downstream of the luciferase open reading frame. miR-LacZ as a miRNA blank vector control. Obviously, miR-206 and miR-613 inhibited the activity of luciferase with the wild-type but not mutant 3’-UTR of hnRNPK in DU145 cells, respectively (Fig. 3I). We further identified that whether miR-206 or miR-613 negatively regulated the expression of hnRNPK in PrCa. Our data firstly confirmed that the efficiency of overexpression and knockdown for miR-206 and miR-613 in PrCa cells which transfected with their corresponding mimics or inhibitor, respectively. (Supplement Fig. 2). Then, in line with the expression of miR-206 and miR-613, the level of hnRNPK showed a tendency toward an inverse correlation as determined by using RT-qPCR and Immunoblot analysis (Fig. 3J-3L). Taken together, these data support the bioinformatic prediction suggesting the 3’-UTR of hnRNPK is a direct target of miR-206 and miR-613.
Overexpression of miR-206/miR-613 can inhibit PrCa cell proliferation in vitro and vivo
Next, we sought to understand the biological effects of miR-206 and miR-613 in Prostate cancer, we induced overexpression of miR-206/miR-613 by using of miR-206/miR-613 mimics and silencing of miR-206/miR-613 using the of miR-206/miR-613 inhibitor in PrCa cells and studied the effects on cell growth. CCK-8 and colony formation assays showed that PrCa cells overexpressing miR-206 or miR-613 had significantly lower proliferation ability than the control cells (Fig. 5A and 5C). In contrast, silencing of endogenous miR-206 or miR-613 led to a significantly higher proliferation rate than observed in the control cells, with the exception of the PC-3 cell lines (Fig. 5B and 5D). Next, Flow cytometry results indicated overexpressing miR-206 or miR-613 dramatically increased the cell population in the G0/G1 phase, whereas it reduced the cell population in the S and G2/M phases (Fig. 5E). On the contrary, depletion of endogenous miR-206 or miR-613 decreased the cell population in the G0/G1 phase and increased the cell population in the S and G2/M phases in DU145 cells (Fig. 5F). Consistently, induction of miR-206 or miR-613 by shRNA in DU145 cells significantly retarded tumor growth in xenograft mouse models (Fig. 5G and 5H). Then the relative expression of hnRNPK, miR-206 and miR-613 in tumor tissue were verified by RT-qPCR (Fig. 5I-K). Taken together, miR-206 or miR-613 could exert a significant inhibitory effect on tumorigenesis by repressing hnRNPK in vitro and vivo.
In line with the results for hnRNPK-silenced cells mediated by miR-206 or miR-613, the protein abundance of Cycle D1, CDK4, CDK6, which promote the cell cycle, were significantly decreased, whereas the protein abundance of P27, which arrests the cell cycle, was obviously increased than in the control cells. Moreover, depletion of endogenous hnRNPK significantly up-regulated the protein levels of apoptosis, such as that for BAX, NF-Κb/p65, c-JUN, and inhibited the anti-apoptotic protein levels for BCL-2 and oncogenes c-MYC. Interestingly, hnRNPK-silenced cells also exhibited increased protein abundance of the transcription factor Snail and oncogenes c-SRC, decreased the biomarkers of EMT, such as E-cadherin and β catenin(Fig. 5L). On the contrary, hnRNPK-upregulated cells mediated by miR-206 or miR-613 inhibitor showed an inverse tendency for above mentioned proteins abundance (Fig. 5M). The correlation network dataset (GSE 88808) also indicated that hnRNPK shared similar correlation with above mentioned proteins (Supplement Fig. 3). These data strongly suggest that hnRNPK regulates several biological functions crucial for prostate cancer development, including proliferation, apoptosis and EMT.
Cul3SPOP E3 ligase degrades hnRNP K protein
Above findings have revealed a new post-transcriptional mechanism of hnRNPK regulation via miR-206 and miR-613 and thereby represses prostate cancer cell proliferation. Interestingly, with the results for hnRNPK-silenced prostate cancer cells mediated by miR-206 and miR-613, led to higher SPOP expression than in the control cells (Fig. 5A). In addition, the SPOP mRNA expression in the correlation network dataset (GSE 88808) also showed a tendency toward a strongly inverse correlation with hnRNPK in PrCa (Supplement Fig. 3). Two public datasets further confirmed that SPOP mRNA expression was significantly downregulated in human prostate cancer tissues (GSE60329) (P<0.05, Fig. 5B) and low expression of SPOP was associated with higher biochemical recurrence rates (GSE46602) (P<0.05, Fig. 5C).
SPOP, a Cullin 3-based E3 ubiquitin ligase, has been shown to participate in diverse cellular processes and plays tumor suppressive and oncogenic roles in prostate cancer[20, 22], suggested Cullin-Ring family E3 ligase(s) may be the upstream regulator(s) to control hnRNPK protein stability. To uncover the underlying regulatory mechanisms, we firstly screened a panel of Cullin scaffolding proteins to identify the potential E3 complex for HnRNPK. Cullin 3, and to a much lesser extent, Cullin 4A, but not other Cullin family members specifically interacted with HnRNPK in cells (Fig. 5D). In keeping with this notion, deletion of endogenous Cullin 3 in Prostate cancer cell lines including C4-2 and 22Rv1, could dramatically upregulate the protein abundance of endogenous HnRNPK (Fig. 5E and 5F). Previous studies demonstrated that Cullin 3 recruits downstream ubiquitin substrates through interaction with BTB-domain-containing proteins as substrate-specific adaptors, including but not limited to SPOP, KEAP1, KLHL1, PLZF and KLHL12. However, we found that only SPOP, but not other Cullin 3-based BTB-domain-containing adaptor proteins we examined, specifically interacts with HnRNPK(Fig. 5G). Notably, SPOP promoted HnRNPK degradation in a dose-dependent manner in prostate cancer cell (Fig. 5H). More importantly, SPOP-mediated degradation of HnRNPK could be efficiently blocked by MG132 (Fig. 5I), indicating that SPOP regulates HnRNPK abundance through the ubiquitin-proteasome pathway. Consistent with these findings, depletion of endogenous SPOP by shRNAs or CRISPR-mediated knockout in multiple human prostate cancer cell lines or MEFs led to a marked increase in the protein abundance of HnRNPK as well as other identified SPOP substrates, including TRIM24, SRC3, AR and DEK (Fig. 5J-N). Moreover, we found that SPOP, but not KEAP1, another Cullin 3-based adaptor protein we examined, specifically promotes HnRNPK ubiquitination in cells (Fig. 5O). And subsequently the half-life of HnRNPK was significantly extended in SPOP -depleted cells (Fig. 5P and 5Q). Our results collectively suggest that the HnRNPK is potential downstream substrates of the Cullin 3 SPOP E3 ubiquitin ligase.
Patients-associated SPOP mutants are incapable to degrade HnRNPK
SPOP is a member of the MATH-BTB protein family containing an N-terminal MATH domain and a C-terminal BTB domain . The MATH domain is responsible for substrate recognition and interaction, while the BTB domain binds Cullin 3 forming the functional E3 ubiquitin ligase complex. Recent genome-wide sequencing studies have revealed that SPOP, as the major frequently mutated gene in prostate cancer, plays pivotal roles for prostate tumorigenesis and metastasis. Interestingly, Most of the somatic mutations identified in prostate cancers such as Y87C, F102C, W131G and F133V, occurred in MATH domain and play a dominant-negative role in substrate binding and degradation (Fig. 6A). In keeping with previous reports [18, 21], we found that deletion of the MATH domain abolishes SPOP interaction with HnRNPK (Fig. 6B) and both the MATH domain and BTB domain are required for SPOP-mediated HnRNPK ubiquitination and degradation (Fig. 6C). Consistently, we sought to explore whether prostate cancer-associated SPOP mutations affect HnRNPK stability. Conceivably, prostate cancer patients-associated SPOP mutants W131G and F102C failed to interact with (Fig. 6D), and were thereby deficient in promoting the degradation of HnRNPK (Fig. 6E). Moreover, ectopic expression of SPOP-WT, but not the SPOP mutants, significantly shortened the half-life of HnRNPK (Fig. 6F and 6G) and promoted HnRNPK ubiquitination in cells (Fig. 6H). These data indicate that patient-associated SPOP mutants are deficient in promoting ubiquitination and destruction of HnRNPK, therefore partially providing a molecular mechanism to explain the aberrant accumulation of HnRNPK in prostate cancer tissue.