NPM1 is overexpressed in PCa tissues and cells
Because the role of NPM1 in cancer cells, especially PCa cells, is poorly understood, we explored the expression level of NPM1 in PCa cells. We first tested the mRNA expression of NPM1 in the most commonly used PCa cell lines, including the androgen-dependent cell line LNCaP and androgen-indenpendent cell lines C4-2, PC-3, DU145 and 22Rv1, as well as the benign prostate cell lines RWPE1 and BPH1. Our data showed that NPM1 expression in PCa cells was higher than that in benign prostate cells (Fig. 1A). We also checked the expression of NPM1 in PCa tissues and prostate noncancer tissues by utilizing GEPIA web tool. Consistently, we observed that NPM1 expression in PCa tissues was higher than that in prostate noncancer tissues (Fig. 1B).
To verify these findings, we measured the protein expression of NPM1 in the PCa surgery specimens by immunohistochemical (IHC) staining of prostate nontumor tissues (n = 21) and PCa ones (n = 35). The IHC results were evaluated by assessment of the staining intensity, and representative images are shown in Fig. 1C. The quantified data are shown in Fig. 1D, which demonstrated that NPM1 expression was increased in PCa tissues compared with paracancerous nontumor tissues. Furthermore, the protein level of NPM1 in cancer tissues and paired noncancerous tissues of 12 individual patients was tested, and the results revealed that NPM1 expression significantly increased in cancer tissues (Fig. 1E, F). We then investigated the expression of NPM1 in silico, and the results demonstrated that NPM1 had higher expression in the PCa cohorts (GSE21034, GSE35988) than in normal prostate tissues. Accordingly, we observed a similar trend in a PCa cohort based on a Chinese population (Chinese Prostate Cancer Genome and Epigenome Atlas, CPGEA) (Fig. 1G-I). Collectively, these data suggested that NPM1 expression is increased in PCa cells and tissues.
NPM1 overexpression leads to the proliferation and invasion of PCa and CRPC cells
Since NPM1 expression is increased in PCa cells and tissues, we explored the biological function of NPM1 expression in PCa. First, we knocked down NPM1 expression by a pair of short hairpin RNAs (shRNAs) in LNCaP (Fig. 2A, B) and 22Rv1 cells (Fig. 2C, D) and found by utilizing MTS assays that cell proliferation was dramatically impaired when NPM1 was knocked down (Fig. 2E, F). Conversely, when NPM1 was overexpressed in 22Rv1 and LNCaP cells by ectopically transfecting HA-tagged NPM1 plasmids (Fig. 2G), we detected accelerated growth of PCa cells (Fig. 2H, I). We further tested the cancer cell invasion ability by Transwell assays, and the results revealed that NPM1 upregulation was positively correlated with the invasion ability of PCa cells. Knockdown of NPM1 resulted in decreased invasion ability in both the LNCaP and 22Rv1 cell lines (Fig. 2J, K), while overexpression of NPM1 increased the cell invasion ability (Fig. 2L, M). Taken together, these data revealed that NPM1 is involved in PCa tumorigenesis by regulating cell proliferation and invasion.
C-Myc is a downstream target of NPM1 in PCa and CRPC cells
The above data showed that NPM1 not only is highly expressed in PCa tissues but also facilitates PCa progression by modulating cancer cell proliferation and invasion. However, the underlying molecular mechanism remains unclear. Interestingly, we found that when NPM1 was knocked down in LNCaP and 22Rv1 cells, c-Myc expression was decreased (Fig. 3A-C). In line with this finding, the expression levels of c-Myc increased when PCa cells ectopically expressed NPM1 (Fig. 3D-F). To elucidate the correlation between these two genes in PCa, we performed IHC staining in 35 patient specimens, and the images of NPM1 and c-Myc staining are shown (Fig. 3G). We observed a positive correlation between NPM1 and c-Myc expression in patient samples (Fig. 3H, I). We further checked the correlation between the two genes in silico, and the result demonstrated a significant positive correlation with a p value < 0.001 (Fig. 3J). Collectively, these data indicated that c-Myc is regulated by NPM1 and is a downstream target of NPM1.
The oncogenic function of NPM1 in PCa and CRPC is performed through a c-Myc-mediated signaling pathway
Since we found that c-Myc is regulated by NPM1 in PCa cells, and that c-Myc is a well-known oncogene that drives multiple cancers, including PCa. We hypothesize that the oncogenic function of NPM1 in PCa is performed through a c-Myc-mediated pathway. To verify this hypothesis, we performed in vitro assays by knocking down NPM1 and c-Myc in LNCaP and 22Rv1 cells (Fig. 4A). MTS assays showed that single knockdown of NPM1 drastically reduced the growth rate of LNCaP and 22Rv1 cells, while single knockdown of c-Myc had a more prominent impact on cancer cell proliferation. Interestingly, combined with knockdown of NPM1 and c-Myc, the cell proliferation rates did not decrease further (Fig. 4B, C). Similar trends were observed in colony formation assays (Fig. 4D-F). Furthermore, the results in the mouse xenograft models confirmed the observation that knockdown of NPM1 or c-Myc alone reduced tumor growth, but double knockdown of NPM1 and c-Myc did not decrease the tumor growth rate more than single knockdown of c-Myc (Fig. 4G-I). The invasion ability of PCa cells was also evaluated, and the results demonstrated that single knockdown of NPM1 or c-Myc reduced the cell invasion ability, as expected, whereas compared with single c-Myc knockdown, double knockdown of NPM1 and c-Myc did not further reduce the invasion rate (Fig. 4J-M). Collectively, these results suggested that the oncogenic function of NPM1 in PCa and CRPC is mediated mainly by c-Myc.
BRD4 is involved in the NPM1–c-Myc oncogenic pathway
To explore the potential therapeutic significance of NPM1 in PCa, especially CRPC, we tested and compared the sensitivity of several small molecule drugs on shControl and shNPM1 CRPC cells (22Rv1 cells). As shown in Fig. 5A, we discovered that CRPC cells were more sensitive to treatment with some small molecule drugs (JQ1, MK2206, docetaxcel, RAD001 and AZD6244) when NPM1was knocked down, while their sensitive to some drugs (LY3214996 and GSK126) changed little, and their resistance to some drugs (SB239063 and PD0332991) was even increased. Among these drugs, JQ1, a bromodomain and extraterminal domain (BET) inhibitor, showed the most remarked reduction in the IC50 from 315.52 nM to 129.67 nM (Fig. 5A), which prompted us to hypothesize that BET family members may be involved in this process.
NPM1 overexpression reduced the sensitivity of CRPC cells (22Rv1 cells) to JQ1 treatment, whereas NPM1 silencing increased the sensitivity of CRPC cells (22Rv1 cells) to JQ1 (the IC50 of JQ1 decreased), as expected (Fig. 5B, C). As bromodomain containing 4 (BRD4) is a well-known and effective BET protein and is well acknowledged in cancers for its oncogenic roles, we tested the relationship of NPM1 and BRD4. The coimmunoprecipitation assay showed that there was a strong endogenous interaction between NPM1 and BRD4 in both PCa and CRPC cells. BRD4 acts as a transcriptional and epigenetic regulator by serving as a reader of acetylated lysine residues on histones. As shown in our previously published data [27-29], BRD4 can promote gene transcription by taking advantage of histone acetyltransferase 1 (HAT1) [29]. It has been reported that BRD4 binds to the P-TEFb complex [30], which is crucial for transcription elongation. Thus, we proposed that NPM1 may influence and cooperate with BRD4 to facilitate downstream target transcription. We already showed that c-Myc may serve as a potential downstream gene of NPM1 (Fig. 3A-F). Hence, we performed a ChIP-qPCR assay with NPM1 knockdown or overexpression. The data demonstrated that the binding of BRD4 to c-Myc was decreased when NPM1 knockdown, whereas the binding increased when NPM1 overexpression (Fig. 5D-G).
Moreover, we assessed the ChIP-seq data and observed a binding peak in the promoter region of Myc (Fig. 5H). These data indicated that both NPM1 and BRD4 are involved in c-Myc regulation. To further address this observation, we generated stable cell lines with single knockdown of NPM1 or BRD4 as well as double knockdown of NPM1 and BRD4. We showed that silencing NPM1 or BRD4 alone reduced the expression of c-Myc, while dual silencing did not further attenuate c-Myc expression (Fig. 5I, J). A similar tendency was observed in both the presence and absence of JQ1 treatment (Fig. 5K, L). In contrast, overexpression of NPM1 increased the expression of c-Myc, but this trend was moderated in the context of BRD4 blockade (Fig. 5M-P). Collectively, these data illustrated that NPM1 could epigenetically regulate c-Myc expression, and that this regulatory mechanism occurs in a BRD4-mediated manner.
BET inhibitor blocks the NPM1–c-Myc oncogenic pathway in vitro and in vivo
Given that BRD4 is involved in NPM1-mediated oncogenic function, we invesgigated whether the BET inhibitor JQ1 can block the detrimental effects of NPM1 in cells and mouse models. We transfected HA-tagged empty vector (EV) and NPM1 plasmids into 22Rv1 cells and treated them with or without JQ1 (Fig. 6A). MTS and colony formation assays demonstrated that JQ1 markedly reduced cell growth and colony numbers in both the HA-EV and HA-NPM1 groups (Fig. 6B-D). Interestingly, we did not observed difference between the HA-EV and HA-NPM1 groups when treated with JQ1, which is consistent with the previous result that NPM1 regulated cancer cell growth through BRD4 (Fig. 5K, L and Fig. 5O, P). Furthermore, we performed in vivo assays by using NOD/SCID mice. JQ1 was injected intraperitoneally at a dosage of 50 mg/kg for 4 weeks (Fig. 6E). As demonstrated in Fig. 6F-I, JQ1 treatment effectively curbed tumor growth in vivo and abolished the growth-promoting effects of NPM1. The tumor samples were excised and embedded in paraffin and were then subjected to IHC staining for c-Myc, Ki67 and cleaved caspase 3 (Fig. 6J). Quantitative data are shown in Fig. 6K, L. Taken together, these data showed that treatment with a BET inhibitor can block the NPM1–c-Myc oncogenic pathway in vitro and in vivo.