Background
Prostate cancer is currently the second leading cause of cancer death in males[
1]. Despite this, in the era of prostate specific antigen (PSA) screening, researchers have now estimated that clinically insignificant prostate cancer is actually overdiagnosed at a rate of 29% for whites and 44% for blacks, the PSA screen resulting in the detection of cancers that otherwise would only have been detected during autopsy in up to 15% and 37% of tumors in whites and blacks, respectively[
2].
There is currently a limited amount of information in the literature on biomarkers with the potential to discern which cases of prostate cancer have the greatest potential to metastasize versus remain latent[
3]. Evaluating the expression of tumor suppressor proteins that have been previously examined in other cancers may indicate novel biomarkers for prostate cancer that have the potential to assess individual patient prognosis and guide therapy selection.
One such biomarker is ezrin-radixin-moesin-binding phosphoprotein 50 (EBP50), which is also known as Na+/H+ exchanger regulatory factor 1, or NHERF1. A 50 kDa, 358 amino acid adaptor protein whose gene is located at 17q25.1, it consists of two PSD-95/Discs Large/ZO-1 (PDZ) domains and a carboxyl-terminal region that is capable of binding members of the ezrin-radixin-moesin (ERM) protein family[
4‐
6]. With its multiple domains, it has been described as a participant in at least 30 unique cellular interactions, including those involving ion transport, secondarily coupled signaling receptors, and tyrosine kinase receptors[
5]. Among the oncologically relevant functions for the protein that have been demonstrated are its ability to recruit the tumor suppressor PTEN for inactivation of the phosphatidylinositol-3-OH kinase (PI3K)/Akt signaling pathway in glioblastoma multiforme[
5,
7], as well as an ability to provide cortical stabilization of β-catenin at cellular junctions in murine embryonic fibroblast models[
8], both indicative of a tumor suppressor function.
Further supporting this notion, additional work has shown that an allele for EBP50 was deleted in 28 of 48 examined breast cancer cell lines[
9]. Knocking-out existing EBP50 expression in T47D and MCF7 breast cancer cell lines has also been shown to lead to increased cell proliferation[
10]. Zheng et al. have additionally noted that restoring EBP50 expression to a MDA-MB-231 breast cancer line, originally deficient in EBP50, inhibited cell growth and increased apoptosis[
11]. Subsequent to this, the same group prepared a stably transfected HeLa-EBP50 clone, which also demonstrated decreased cell growth, suggesting a tumor suppressor role for EBP50 in cervical cancer as well.
In spite of this, a universal tumor suppressor function for EBP50 has not been observed. EBP50 has been shown to be overexpressed in hepatocellular carcinoma[
12]. Cytoplasmic over expression has also been linked to the progression of colorectal carcinoma[
13]. Furthermore, in contrast to the previously described work, Song et al. have reported that EBP50 immunoreactivity in breast cancer was positively associated with tumor stage and lymph node involvement[
14], prompting others to suggest that its role in oncogenesis or tumor suppression may vary with cellular location, with a membranous or apical distribution supporting a tumor suppressor function and a cytoplasmic distribution conferring oncogenic properties [
5].
To our knowledge, the expression of EBP50 has never been studied in prostate cancer. Here, we compare the immunohistochemical profiles in a series of 11 cases of normal donor prostate (NDP), 37 cases of normal tissue adjacent to prostatic adenocarcinoma (NAC), 15 cases of benign prostatic hyperplasia (BPH), 35 cases of high-grade prostatic intraepithelial neoplasia (HGPIN), 103 cases of primary prostatic adenocarcinoma (PCa), and 36 cases of metastatic prostatic adenocarcinoma (Mets) in order to examine if either a tumor suppressor or oncogenic function for EBP50 can be suggested in prostate cancer, providing further information about its potential as a diagnostic and/or prognostic biomarker.
Discussion
In the EBP50 stained specimens, the average staining intensities were highest in the specimens of HGPIN, and lowest in the specimens of PCa and Mets. HGPIN had significantly higher staining than Mets, NAC, and PCa. Despite the fact that HGPIN was significantly different than the BPH and NDP groups but not the NAC group, the means were relatively similar between BPH, NDP, and NAC, suggesting that the staining intensity does not vary greatly between these classifications.
Previous work has shown that radixin, an ERM protein and binding partner of EBP50 that is responsible for linking F-actin to plasma membrane proteins[
5,
18], also demonstrated higher absolute staining in specimens of HGPIN than in other prostatic tissue types, including PCa and NAC [
19]. This finding may indicate that the higher expression of both proteins in HGPIN may reflect a unique feature of the pre-cancerous tissue physiology, and that both may be down regulated in specimens of prostatic adenocarcinoma.
Mets tissue had significantly lower staining than all other tissue types, including PCa, indicating that loss of EPB50 expression may play a role in select cases of prostate cancer metastasis. This is further echoed by the fact that only 1/36 cases of metastatic tissue showed high intensity range staining for EBP50. Given that EBP50 has been previously shown in murine fibroblast models to promote adherens junction stabilization through mediating the interaction of β-catenin with E-cadherin, its loss of expression is plausible with tumor dissemination[
8].
Despite this, however, it is important to note that these findings may also contain a correlative component that is not prognostic in nature. While all of the tumors in this study were primary tumors at the time of specimen retrieval, definitive follow-up information on these patients was not available. In this sense, from this study it is not possible to rule out that the decreased EBP50 expression, at least in part, may be due to the metastatic location itself. The current results, however, indicate and warrant later phase biomarker studies[
20] that will longitudinally correlate EBP50 expression directly with patient outcomes to further evaluate its potential to predict metastatic risk in prostate cancer.
A significant increase in EBP50 staining between Stage 2 and 3 PCa specimens was also observed. It is possible that this finding represents a change in tumor physiology between the two stages, however, given that this trend was not noted in Stage 4 PCa specimens, it is also possible that this represents a spurious finding, and should be further evaluated before definitive conclusions are reached. No differences were seen between the Gleason score classifications.
In general, EBP50's expression is increased in polarized epithelial cells, such as the liver, kidney, pancreas, small intestine, and the prostate[
5]. Considering the diverse intracellular roles that have been proposed for EBP50, it may be difficult to elucidate a clear, singular mechanism whereby it may promote oncogenesis or tumor suppression within individual tissue types.
Differing hypotheses to explain to the behavior of EBP50 in cancer have been proposed, especially given the multiple studies that seem to support two opposing functions for EBP50 in cancer [
9‐
11,
21]. Zheng, et al have suggested that in many cases of breast cancer where EBP50 is expressed, it may not be expressed in sufficient quantities to halt tumor progression[
11]. Others have demonstrated that the hypoxia associated with tumor necrosis can increase EBP50 (NHERF1), which increases Na+/H+ activity, in turn decreasing local pH and promoting tumor dissemination[
4,
21]. And while it has been shown that EBP50 can cluster with EGFR, PDGFR, and the tumor suppressor NF2 to halt cell signaling and hence cancer progression[
11,
22‐
24], others have posited that EBP50's PDZ domains may actually allow for new tumor-specific interactions[
4].
These different findings, in part, have been reconciled with the hypothesis that EBP50 may have different functions by cellular location, with a tumor suppressor function associated with a membranous/apical distribution, and an oncogenic function promoted by a cytoplasmic location[
5]. In support of this hypothesis, previous work has demonstrated a progression from luminal to cytoplasmic EBP50 expression occurs across normal to ductal carcinoma in-situ to invasive and metastatic breast cancer tissues[
25]. While membranous EBP50 has been shown to stabilize β-catenin at cell membranes[
8], non-stabilized β-catenin is also capable of forming growth-promoting transcription complexes in the nucleus and has been prominently associated with hepatocellular carcinoma[
26]. Hence, it is interesting that an overexpression of EBP50 with a focal nuclear localization has been documented in hepatocellular carcinoma[
12]. A similar phenomenon has been noted in colorectal cancer, where membranous EBP50 loss and increased cytoplasmic expression has been noted in the colorectal adenoma-to-carcinoma transition, with subsequent increases in cellular invasion and epithelial-to-mesenchymal transition, processes that demonstrated reversibility when EBP50 was reexpressed at the apical membrane of intestinal epithelium[
13]. Additionally, in normal astrocytes, EBP50 has demonstrated a membranous distribution, while it has demonstrated a cytoplasmic distribution in many cases of glioblastoma multiforme[
7]. This corresponds with the absence of the EBP50-binding tumor suppressor PTEN and the activation of the growth-promoting Akt pathway, which is traditionally silenced by PTEN through recruitment to the plasma membrane by EBP50[
7].
Relating these examples specifically to prostate cancer, it has been demonstrated that β-catenin can interact with androgen receptor and increase its transcriptional activity, hence contributing to prostate cancer progression[
27,
28]. Moreover, the tumor suppressor PTEN is frequently found mutated in prostate cancer, with subsequent PI3K/Akt signaling shown to promote cell survival[
27,
29]. Interestingly, the PI3K/Akt pathway also increases the stability of β-catenin in prostate cancer[
30]. Based on these findings, it is possible to suggest that EBP50, under the appropriate circumstances, may possess a tumor suppressor function in prostate cancer similar to those described above in other cancer types.
While a cytoplasmic staining pattern for EBP50 was noted across most specimens examined in this study, a membranous/apical staining pattern was clearly more prominent than cytoplasmic staining in many cores, most commonly in the benign and pre-neoplastic specimens (Figure
2). While cores with clearly more prominent membranous staining were found in 73.3% of BPH cases studied (Figure
2B), this finding was only noted in 9.7% of cases of PCa and not in any specimens of Mets. Although an overlap between the expression patterns still existed between many of the benign and cancerous specimens, this trend concurs with the above hypothesis regarding strong membranous expression and tumor suppression and warrants further study to determine its potential to assess metastatic risk.
As a final note of interest, in one model of EBP50 function, its PDZ-2 domain has been shown to bind to the C-terminal ERM-binding domain, inhibiting the binding of other proteins to the PDZ domains, such as PTEN and B-catenin[
31]. In this same model, when ezrin, an ERM protein, binds the C-terminal domain, the PDZ domains are freed up for additional binding partners. In the intestinal epithelium of ezrin knock-out mice, EBP50 has been shown to be displaced to the cytoplasm[
5,
32]. In light of this, it is an interesting finding that ezrin expression has been inversely correlated with tumor differentiation in prostate cancer[
33], and that moesin, another ERM protein, showed higher incidences of lymph node metastases when it was associated with a cytoplasmic distribution as opposed to a membranous one in oral squamous cell carcinoma[
34]. Hence, EBP50's location and function may also be directly linked to the location and presence of the ERM proteins that ultimately enable PDZ domain interactions.
Conclusions
These results provide a basis for the characterization of the staining patterns and intensities of EBP50 in PCa and Mets in comparison to benign prostate tissue. The immunostaining was highest in specimens of HGPIN. Its expression was lowest in Mets, with the expression in Mets significantly lower than that of all other tissue groups, including PCa (p = 0.006). As a significant decrease between Stage 2 and 3 cancer was observed (p = 0.021), it is also possible that EBP50 expression is altered with carcinoma stage, although significant differences were not seen when Stages 2 and 3 were compared to Stage 4. No differences were seen when comparing the Gleason scores of the PCa specimens.
EBP50 staining was a combination of membranous/apical and cytoplasmic in the prostatic tissues examined. Although a clear distinction would not be made in all cases, a predominant membranous staining pattern was more commonly observed in benign specimens than in malignant ones. As an example, 11/15 cases of BPH featured cores with a readily apparent predominant membranous staining pattern compared to the adjacent cytoplasm, while this was only observed in 10/103 cases of PCa and was not seen in any cases of Mets.
It is also important to note, however, as this membranous/cytoplasmic distinction was not noted in all comparisons between benign and cancerous specimens, that an alternate explanation may account for these findings. As a general decrease in overall staining intensity was observed between the benign specimens and the cancerous/metastatic specimens, this lack of membranous staining may simply reflect an overall staining decrease, making it difficult to appreciate the true ratio of membranous to cytoplasmic staining from a visual examination. This is especially true in the cases of metastatic cancer, where the overall staining is very faint across all specimens in the first place.
Given that a significant decrease in staining was noted between specimens of PCa and Mets, further studies of EBP50 are justified to assess its potential for clinical usage in prognosis assessment of patients with prostate cancer.
Competing interests
The authors declare that they have no competing interests.
Authors' contributions
TB assisted in scoring tissue microarrays under the direct supervision of an attending pathologist, performed the statistical calculations, and drafted the manuscript. UC assisted with statistical calculations and reviewed the manuscript. AP conceived of the study, developed and approved the study protocol, approved all tissue microarray scoring, and revised the manuscript. MB also conceived of the study, developed and approved the protocol, and revised the manuscript. All authors have read and approved the final manuscript.