Background
Prostate cancer (PCa) is the second most common cancer and fifth leading cause of cancer-associated mortality in men, with an estimated number of 1.1 million cases worldwide [
1]. Despite early detection and treatment involving radical prostatectomy, radiotherapy and/or androgen deprivation, PCa continues to be a major cause of cancer-associated morbidity and mortality. Development of resistance to androgen deprivation therapy and acquisition of metastatic potential are the major causes of PCa mortality [
2]. Cellular and animal models of PCa have elucidated many signaling pathways that render PCa refractory to treatment and contribute to metastatic dissemination [
3,
4]. This knowledge can be exploited not only to develop personalized therapies for the recalcitrant disease but also to the development and testing of biomarkers for early detection of the aggressive disease.
The suppressor of cytokine signaling 1 (SOCS1) protein is considered a tumor suppressor because of frequent repression of the
SOCS1 gene promoter by CpG methylation in many types of cancers including hepatocellular carcinoma, leukemia and pancreatic adenocarcinoma [
5‐
9]. SOCS1 expression is also inhibited by microRNAs such as miR-19 and miR-155 in human cancers [
10‐
12].
SOCS1 is one of the genes that are under-expressed in the androgen-independent PCa cell line LNCaP-C81 compared to the androgen-dependent LNCaP-33 cell line [
13]. Even though methylation of the
SOCS1 promoter occurs only in 20% of PCa cases, increased expression of the SOCS1-targeting micro-RNA, miR-30d, has been reported to occur frequently in PCa [
14,
15]. In fact, elevated miR-30d expression in PCa specimens correlates with early biochemical recurrence, supporting a tumor suppressor role for SOCS1 in PCa [
15]. A number of studies have shown that SOCS1 attenuates growth of prostate cancer cells in vitro and in vivo [
16‐
18].
The
SOCS1 gene is induced by diverse cytokines and growth factors, and inhibits their signaling in a negative feedback manner [
8,
19]. SOCS1 has been shown to inhibit IL-6 and hepatocyte growth factor (HGF) signaling, which are implicated in PCa pathogenesis [
16‐
18]. SOCS1 can exert its anti-tumor functions through diverse mechanisms. SOCS1 contains a central SH2 domain and C-terminal SOCS box. The SH2 domain binds to JAK kinases activated by cytokines and growth factor receptor tyrosine kinases (RTK), and thus blocks downstream signaling events [
8,
19]. The SOCS box mediates ubiquitination of SOCS1-bound proteins, thereby promoting their degradation by proteasomes. We have shown that SOCS1 regulates HGF signaling by promoting ubiquitination and proteasomal degradation of the MET RTK [
20,
21]. In cellular systems, SOCS1 co-operates with p53 to enforce oncogene-induced senescence [
22]. However, in a mouse model of hepatocellular carcinoma, SOCS1 deficiency is associated with increased expression of a p53 target gene, the cyclin-dependent kinase inhibitor p21
CIP1/WAF1 (p21) [
23]. Even though p21 generally functions as a tumor suppressor, its cytosolic localization may promote tumor growth [
24,
25]. Indeed, overexpression of p21 occurs in several human cancers and correlates with poor prognosis [
24]. The relative contribution of the different downstream targets of SOCS1 in mediating tumor suppression in diverse cancers has not been studied yet. In PCa, even though p53 mutation is uncommon, MET and p21 are implicated in disease progression. Expression of MET occurs in 40% of localized PCa and correlates with Gleason score and lymph node metastasis, reaching nearly 100% in bone metastases [
26‐
29]. Similarly, increased expression of p21 mRNA and protein in PCa has been associated with progression to androgen-independent cancer and resistance to apoptosis induction by chemotherapeutic agents [
30,
31].
In this study, we investigated SOCS1 protein expression in prostatectomy specimens and its correlation to disease severity, with the goal of testing its utility as a prognostic biomarker. We examined the correlation between SOCS1 expression to those of its putative downstream targets of tumor suppression namely, p53, MET and p21 [
21‐
23]. We observed significant inverse correlation between SOCS1 expression and disease severity. However, SOCS1 expression did not correlate with that of p53, MET or p21 in the whole study cohort. Within the study population, cases with regional lymph node metastasis, albeit small in number, showed decreased SOCS1 and increased MET and p21 expression.
Discussion
Given that the expression of SOCS1 is regulated at post-transcriptional level by microRNAs [
10‐
12,
15], detection of SOCS1 protein in cancer tissues would represent a direct approach to evaluate SOCS1 as a potential cancer biomarker. However, detection of endogenous SOCS1 protein has been particularly difficult for two reasons. Firstly, the basal
SOCS1 gene expression in normal tissues is very low; however, it is induced by myriad of inflammatory cytokines, growth factors, chemokines and other mediators such as prostaglandins and androgens [
17,
40‐
42]. Secondly, most available SOCS1 antibodies are not sensitive enough to detect the endogenous SOCS1 protein. Hence, the literature on SOCS1 expression is mostly restricted to quantification of
SOCS1 mRNA with only a few reports on SOCS1 protein expression [
9,
17,
35,
43,
44]. Given that
SOCS1 gene repression occurs in many cancers by diverse mechanisms, clearly there is a need for developing and testing more sensitive and specific SOCS1 antibodies for diagnostic purpose in surgical pathology.
We observed different staining intensities of SOCS1 in prostatectomy specimens that showed strong inverse correlation with the ISUP grade groups and Ki67 staining. Moreover, tumors with regional lymph node involvement showed significantly lower SOCS1 expression compared to those without. These findings indicate that SOCS1 expression is diminished during PCa progression and support the potential tumor suppressor role of SOCS1 in this cancer. Moreover, the finding that SOCS1 staining was reduced but not abolished in PCa specimens is consistent with the earlier reports that repression by CpG methylation occurs only in a small subset of PCa, whereas the expression of the SOCS1-targeting miR-30d is more frequent [
14,
15]. Hence, evaluating SOCS1 protein would be more informative than quantifying CpG methylation of the
SOCS1 gene,
SOCS1-targeting micro-RNA or
SOCS1 mRNA levels.
In the current study, we examined how reduced expression of SOCS1 protein in PCa specimens impacted on p53, MET and p21, which are regulated by SOCS1 [
21‐
23] and are implicated in PCa progression [
29,
30,
38]. In the PCa grade groups, we observed variable protein staining for p53, MET and p21, of which only p53 correlated with either disease severity, and none of them correlated with SOCS1protein expression. Evaluation of p53 in human cancers has restricted diagnostic/prognostic utility because increased expression generally correlates with mutated or dysfunctional p53 [
45]. Nonetheless, we postulated based on the requirement of SOCS1 to mediate p53-dependent cellular senescence [
22] that SOCS1 deficiency might increase p53 expression, similarly to inactivating p53 mutations. Likewise, we expected increased protein levels of MET and p21 in cases where SOCS1 was limiting, as the latter attenuates MET signalling and p21 expression at least partly via ubiquitination and proteasomal degradation [
21,
23]. Even though there was a tendency of inverse correlation between SOCS1 and p53, MET or p21, they were not statistically significant (Fig.
4f). Clearly, studies on larger cohorts are needed either to confirm this tendency, or to support the alternate possibility that the decrease in SOCS1 expression need not necessarily affect all of its downstream mediators of tumor suppression, as the SOCS1-mediated suppression pathways may vary in individual cancers. Besides, other molecules/pathways might influence the downstream mediators of SOCS1 to a variable extent in individual cancers.
Major prognostic determinants of PCa are grade, local extension, lymphovascular invasion and lymph node metastasis. Still, many intermediate grade carcinomas extend outside the prostate gland or metastasize, and many high-grade carcinomas are organ-confined at time of surgery. In such situations, additional prognosis markers would be important in determining whether more aggressive treatment is warranted. A lot of effort has been made in the last two decades to develop and refine diagnostic and prognostic PCa biomarkers [
46]. The former group (e.g., % of free PSA over total PSA, PCA3/DD3 gene expression) strives to achieve higher specificity than total serum PSA levels in order to overcome false-positive results and avoid unnecessary biopsies. On the other hand, prognostic biomarkers such as early PCa antigen-2 (EPCA-2) and post-operative PSA velocity (PSAV) and doubling time (PSAD) have shown promise in predicting PCa aggressiveness, risk of non-organ confined disease and disease recurrence. SOCS1 immunostaining may complement the latter group in predicting aggressive disease, although further studies in larger cohorts with non-organ confined disease and metastatic PCa as well as development of refined reagents with improved sensitivity and specificity are needed.
Androgen receptor (AR) signalling is a major oncogenic driver in PCa progression via induction of promitotic genes, and the AR gene is often amplified or mutated in PCa that have become resistant to androgen deprivation therapy [
47]. PCa with acquired resistance to therapy display distinct AR-responsive gene expression signature [
48]. In recent studies, the AR splice variant AR-V7 has emerged as a predictive biomarker of PCa responsiveness to next generation androgen targeting therapies [
49]. We observed that AR expression did not vary significantly across the PCa grades or the tumor stages T2 and T3. However, a strong positive correlation was observed between AR expression and Ki67 staining, in agreement with the ability of AR signalling to stimulate cell proliferation [
47]. Even though androgen stimulation induces
SOCS1 gene expression in PC3 cells [
17], we did not find significant correlation between AR and SOCS1 protein expression. Nonetheless, SOCS1 protein level showed significant negative correlation with Ki67 staining, reflecting the loss of SOCS1-dependent control of other growth stimulatory pathways.
Tumors with regional lymph node metastasis, despite being limited in number, showed significantly lower SOCS1 and prostein expression and higher levels of Ki67, MET and p21 than those without lymph node involvement (Fig.
7). SOCS1 might impinge on multiple signaling pathways that promote aggressive growth, migration, invasion and metastasis, resulting in metastatic spread to regional lymph nodes. Increased MET expression had been reported in metastatic PCa [
28]. We have recently shown using PCa cell lines that SOCS1 inhibits HGF-induced MET signaling and attenuated their migration and invasion [
18]. Consistent with this report, MET overexpression correlated with high proliferation and regional lymph node metastasis (Fig.
5 and Fig.
7). However, we did not observe significant correlation between SOCS1 and MET protein expression. Several factors could contribute to this discrepancy between the studies on cell lines and in primary cancers. For instance, in primary cancers retaining SOCS1 expression, MET mutations might allow escape from SOCS1-mediated regulation. Alternatively, signaling pathways other than MET might contribute to aggressive growth of cancer cells retaining SOCS1 expression. The association between p21 and metastasis, but not with tumor grade is also intriguing. Although the number of metastatic cancers included in our study is small, our finding raises the possibility that the proposed function of cytosolic p21 in promoting cell motility [
50] might contribute to PCa metastasis.
Even though the loss of SOCS1 did not correlate with increased MET or p21 expression as we had expected, our findings support the tumor suppressor function of SOCS1, and the oncogenic potential of MET and p21 in PCa. The expression of p21 also showed a strong correlation to AR levels, which is expressed in androgen-independent PCa and contributes to disease progression in a ligand-independent manner, renewing the interest in AR blockade [
51,
52]. MET and AR levels also showed a marked tendency for positive correlation, although this was not statistically significant. Overall, the expression of SOCS1, MET and p21 in PCa specimens may be useful to identify cases that are prone to metastatic spread and thus to put them on more aggressive treatment and/or under close monitoring. While highly specific, clinical grade anti-p21 Abs are available from DAKO, anti-MET Abs suitable for such use are still in development [
53] and efforts must be made to develop clinical grade SOCS1 Abs.
Our study design did not include patient follow-up for responsiveness to therapy or survival. Therefore, we analyzed correlation between
SOCS1 gene expression and patient survival in datasets obtained from the cBioportal (
http://www.cbioportal.org/) [
54] and PrognoScan (
http://www.abren.net/PrognoScan/) [
55] cancer web portals. We did not find any significant correlation between
SOCS1 gene expression and patient survival in both datasets (Additional file
2: Figure S2). The currently available public cancer databases do not contain protein expression data. Therefore, it is necessary to carry out prospective studies on SOCS1 protein expression in PCa and other cancers with a wider scope, including treatment responsiveness, disease-free survival and overall survival. The present study lays foundation for such future investigations.
Acknowledgements
Not applicable.