Downregulation of ERp57 expression is associated with poor prognosis in early-stage cervical cancer.

Abstract Objective: We investigated the clinical significance of ERp57 in the progression of cervical cancer. Methods: mRNA and protein expression of ERp57 in cervical neoplasias were examined. Results: ERp57 mRNA expression was significantly decreased in cervical cancers. Immunohistochemistry revealed that ERp57 expression in 123 cervical cancers was down-regulated compared to cervical intraepithelial neoplasias or normal tissues (p < 0.001). Low ERp57 expression was significantly associated with worse overall survival (HR = 12.19, p = 0.018). Conclusions: Low ERp57 expression independently predicts a poor outcome for patients with cervical cancer, supporting the notion that ERp57 may be a promising novel cancer target.


Introduction
Cervical cancer is one of the most common gynecologic tumors worldwide and is a leading cause of female cancerrelated death in developing countries (Moody et al., 2010). According to official statistics, more than 50% of all cases and deaths from the disease worldwide were reported in the Asia Oceania region, with the highest incidence and mortality rates in South Central and Southeast Asia (Garland et al., 2012). The high mortality is ascribed to disease recurrence after cervix resection and lack of effective treatment for advanced stage disease (Chun, 2009). Although many clinicopathological factors have been described as prognostic tools, a limited number of these parameters have been incorporated and used in the clinic. Furthermore, molecular markers that accurately predict response to therapy are limited because of their low sensitivity and specificity. Therefore, it is necessary to identify reliable molecular markers predictive of the clinical outcome of patients with cervical cancer.
ERp57, the protein encoded by the PDIA3 on chromosome 15, is a glycoprotein-specific thiol-oxidoreductase and a component of the protein disulfide isomerase (PDI) family that is involved in the major histocompatibility complex (MHC) I antigen processing molecules (Lindquist et al., 1998). The role of the MHC class I molecule is to present small peptides to cytotoxic T lymphocytes (Radcliffe et al., 2002), which can potentially trigger a cascade of immune responses. Defective component expression of the MHC class I antigen processing complex is hypothesized to be an important mechanism of MHC class I downregulation and can lead to immune evasion by tumor cells. Defects in MHC class I expression have been found in many tumor types and are associated with advanced stage and poor survival in patients with cancer (Atkins et al., 2004;Mehta et al., 2008;Ogino et al., 2003).
Because ERp57 has an important role in antigen processing, the regulation of ERp57 protein expression may be associated with human diseases including cancer. Tumor cells utilize impaired antigen presentation as one immune escape mechanism that results in the progression of tumorigenesis (Igney et al., 2002). For example, downregulation or loss of MHC class I antigens has been found in many types of tumors and is associated with an adverse prognosis (Atkins et al., 2004;Ogino et al., 2003;Seliger et al., 2000). Furthermore, loss of ERp57 protein expression has been reported to be associated with poor prognosis in gastric cancer and head and neck cancer (Leys et al., 2007;Ogino et al., 2003). Despite evidence supporting ERp57 involvement in tumorigenesis, the relationship between ERp57 expression and clinicopathological factors in various human cancers, including cervical cancer, remains unclear.
In this study, we investigated the mRNA level of ERp57 in cervical cancer cells, cancer tissues and normal cervical epithelial tissues. Additionally, we studied the subcellular localization of ERp57 in CaSki cells by Western blotting combined with subcellular fractionation. To evaluate the potential relationship between ERp57 and various clinicopathological parameters, we analyzed ERp57 protein expression in a large series of cervical cancer, precursor and corresponding normal tissues, and demonstrated that ERp57 expression levels are altered during cervical carcinogenesis and down-regulation of ERp57 is correlated with poor prognosis.

Patients and specimens
After gaining institutional review board approval, 123 paraffin-embedded cervical cancer, 187 cervical intraepithelial neoplasia (CIN) and 310 matched normal cervix tissues that had been resected at Gangnam Severance Hospital, Yonsei University College of Medicine between 1996 and 2010 were retrieved. Some of the paraffin blocks were provided by the Korea Gynecologic Cancer Bank through the Bio & Medical Technology Development Program of the Korea Ministry of Education, Science and Technology. All tumor tissues were histologically revised and only specimens with sufficient tumor cells were included in the tissue microarray (TMA) construction. Tumor staging was performed according to the International Federation of Gynecology and Obstetrics (FIGO) classification system.
Primary treatment for stage I/II consisted of type II or III radical hysterectomy with pelvic lymph node (LN) dissection. In cases of increased risk of relapse including positive resection margins, parametrial invasion, or positive LN metastasis, platinum-based concurrent chemoradiation was added as adjuvant therapy. Patients with inoperable disease were treated with radiotherapy (both external radiotherapy and brachytherapy) or both cisplatin-based chemotherapy and radiotherapy.
Data collected by reviewing the medical records of patients with cervical cancer included age, diagnoses of prior or current malignancies, Hybrid Capture Õ 2 (HC2) results, SCC (squamous cell carcinoma) antigen levels, surgical procedure and survival status. SCC antigen levels were recorded at primary diagnosis up to one week prior to operation or chemoradiation therapy. Pathological data included tumor grade, cell type, tumor size and LN metastases. Fresh cervical tissue samples including normal cervix (n ¼ 2) and cervical cancer (n ¼ 10) were also collected for mRNA detection. In addition, five cervical cancer cell lines were used for mRNA detection.

SYBR green real-time PCR
The HeLa, CaSki, SiHa and ME180 cell lines were purchased from the American Type Culture Collection (ATCC, Manassas, VA) and the SNU17 cell line was purchased from the Korean Cell Line Bank (KCLB, Seoul, Korea). Cells were maintained in DMEM/F12 supplemented with 10% FBS in the presence of 5% CO 2 at 37 C in a humidified incubator. Total RNA was extracted from 2 normal cervical epithelial tissues, 5 cervical cancer cell lines and 10 cervical cancer tissues using the RNeasy Mini kit (Qiagen, Valencia, CA). Next, cDNA was generated from 2 mg of total RNA extracted from each sample using the SuperScriptÔ III first-strand synthesis system (Invitrogen, Carlsbad, CA) according to the manufacturer's protocol. SYBR Green real-time PCR was performed using an ABI 7300 instrument (Applied Biosystems, Foster City, CA). Primers specific for ERp57 were used (forward: 5 0 -GTCGAAGGGCCTTTC TTG-3 0 , reverse: 5 0 -AGCTGCGTGGCAAGGATAAA-3 0 ). Glyceraldehyde-3-phosphate dehydrogenase (GAPDH), a house-keeping gene, was used as an internal control. Amplification was performed under the following conditions: pre-incubation for 2 min at 50 C, then denaturation for 10 min at 95 C followed by 40 cycles of denaturation for 15 sec at 95 C and annealing/extension for 1 min at 60 C. The comparative 2 ÀDDCt method was used for relative quantification of gene expression as described previously (Livak et al., 2001).

Western blotting
CaSki cells were cultured, harvested and fractionated with the NE-PER Nuclear and Cytoplasmic extraction kit according to the manufacturer's protocol (NE-PER Reagents, Thermo Scientific, Rockford, IL). Ten micrograms of the cellular fractionations were separated by 4-12% SDS-PAGE and transferred to a nitrocellulose membrane. After blocking for one hour with 5% nonfat milk in TBST (50 mM Tris, 150 mM NaCl, 0.05% Tween 20, pH 7.5), the membrane was rinsed with TBST and then incubated overnight at 4 C with polyclonal rabbit anti-ERp57 (1:500; Santa Cruz Biotechnology, Santa Cruz, CA) in TBST containing 5% BSA. For validation of the cellular fractionation process, the membrane was also blotted with antibodies against calnexin (BD Transduction Lab, San Jose, CA) and Lamin B1 (Santa Cruz Biotechnology). Subsequently, the membrane was washed with TBST and incubated with anti-mouse or rabbit conjugated to horseradish peroxidase (1:5000; Millipore, Billerica, MA) for two hours. The signal was detected using a SuperSignal Chemiluminescence kit (Thermo Scientific).

Tissue microarray (TMA) construction and immunohistochemical staining
TMAs were constructed from archival formalin-fixed, paraffin-embedded tissue blocks, as previously described (Noh et al., 2012). Briefly, for each tumor and matched normal epithelial tissue, a representative area was carefully selected from a hematoxylin and eosin (H&E) stained section of the donor block, cored and placed into the recipient array block using a 1.0-mm needle.
For immunohistochemical staining, 5 mm paraffin sections were incubated at 60 C for two hours, deparaffinized in xylene and hydrated in serial alcohol solutions to distilled water for five minutes each. Heat antigen retrieval was performed by submerging slides in pH 6 citrate buffer at 121 C for 15 minutes using a pressure cooker (DAKO, Carpinteria, CA) and blocking in 1% skim milk for 30 minutes. Endogenous enzyme activity was inhibited with 3% hydrogen peroxide containing sodium azide for an additional 30 minutes. After rinsing, the sections were incubated at room temperature for one hour with antibodies against ERp57 (1:200, Santa Cruz Biotech., rabbit polyclonal IgG, Cat.# sc-32761, epitope corresponding to amino acids 108-207 mapping within an internal region of ERp57 of human origin). The remainder of the procedure, including labeling and visualization, was completed using an automated immunostaining system with a NovoLink Polymer Detection System (Leica Biosystems, Newcastle, UK). Stained sections were lightly counterstained with hematoxylin, dehydrated in graded alcohol solutions, cleared in xylene, mounted and cover-slipped. For validation of the immunohistochemical staining on tissue microarray slides, we selected five conventional whole section slides from five cases included in this study, performed immunohistochemical staining for ERp57 and compared the staining pattern of the tissue microarray slides and matched conventional slides from the same case. In addition, stromal cells in whole sections of cervical cancer tissue were used as an internal positive control. The primary antibody was omitted from the negative control.

Statistical analysis
The Mann-Whitney test or the Kruskal-Wallis test for continuous variables was used when variance homogeneity was not assumed. Otherwise one-way analysis of variance (ANOVA) was used when appropriate. Survival curves were generated by the Kaplan-Meier method, and statistical significance was calculated by the log-rank test. Multivariate analysis was performed using the Cox proportional hazards model to determine the independent significance of relevant clinical covariates. Statistical analyses were performed using SPSS version 18.0 (SPSS Inc., Chicago, IL). Statistical significance was set at p50.05 and all p values were determined from two-sided tests.

Clinicopathological characteristics of cases
Patients ranged in age from 21 to 79 years (mean, 42.6 years). The mean ages were 39.3 AE 10.3 years for low-grade CIN, 38.4 AE 11.0 years for high-grade CIN and 48.8 AE 11.4 years for cervical cancer. There was a significant age difference between low-or high-grade CIN and cancer patients (p50.001, respectively). However, ERp57 expression did not correlate with patient age, but was similar in all age groups (Spearman's rho ¼ À0.068, p ¼ 0.233

mRNA expression and subcellular localization of ERp57
The ERp57 expression levels were significantly lower in cancer cell lines (mean 2 ÀDDCt ¼ 0.36, p ¼ 0.014) and tumorbearing cancer tissues (mean 2 ÀDDCt ¼ 0.89, p ¼ 0.013) compared with normal cervical epithelial tissues (mean 2 ÀDDCt ¼ 8.89), reflecting low ERp57 mRNA expression in cancer cell lines (24.7-fold) and tissues (9.9-fold) (Figure 1a). To examine the specificity and capability of anti-ERp57 antibodies, we subsequently investigated the level of ERp57 protein in CaSki cells by Western blotting. Western blot analysis of the cultured cervical cancer cultured cells revealed a clear single band around 60 kDa corresponding to ERp57 (Figure 1b). Although ERp57 was detected in both cytosolic and nuclear fractions, high expression was restricted to cytosolic fractions (Figure 1b). The purities of the cytosolic and nuclear fractions were respectively confirmed with Calnexin and Lamin B1.

Association between ERp57 expression and clinicopathological characteristics
As shown in Figure 2, positive staining for ERp57 protein was located primarily in the cytoplasm of tumor and normal epithelial cells. Expression was higher in normal cervical epithelial tissue and low-grade CIN specimens than in highgrade CIN and cancer tissues (p50.001) ( Table 1). This trend of progressively lower ERp57 protein expression corresponding to the phases of cervical cancer progression was significant by Spearman's rank correlation (p value of À0.589, p50.001). In addition, ERp57 immunoreactivity significantly correlated with tumor grade (p50.001) and cell type (p ¼ 0.004) (Figure 3).

Discussion
ERp57 has been found in many different subcellular locations and it is involved in a remarkable variety of processes. One of the widely studied functions of ERp57 is its role in the assembly of MHC class I antigen processing (Lindquist et al., 2001). The glycosylated heavy chain of MHC class I binds to both calnexin and ERp57, and this complex oxidizes the heavy chain through disulfide bond formation. The oxidized heavy chain allows assembly of soluble subunit b 2 -microglobulin and subsequently contributes to the formation of the MHC class I peptide-loading complex (Garbi et al., 2006). This complex is composed of heavy chain class I protein, b 2 -microglobulin, calreticulin, ERp57, tapasin and the transporters associated with antigen presentation (TAPs). It has recently been reported that the level of ERp57 gene promoter methylation was significantly higher in CIN and cervical cancer cases in comparison to normal tissues. In addition, the high level of ERp57 methylation was correlated with human papillomavirus 16 positivity in human cervical cancer cases (Hasim et al., 2012). Therefore, ERp57 may be linked to cancer progression by an unknown mechanism.
To gain some insight into the timing of ERp57 downregulation during tumor development, we studied ERp57 expression in low-and high-grade CIN, both of which are considered premalignant lesions, and in invasive cervical cancer. IHC analysis revealed a drastic reduction in ERp57 protein expression in high-grade CIN (mean histoscore ¼ 6.81) and invasive cervical carcinoma (mean ¼ 5.75) specimens compared to normal tissues (mean ¼ 10.79). This finding suggests that ERp57 may play an important role in cervical carcinogenesis. As an MHC class I peptide-loading complex component, ERp57 is important for T-cell-mediated anti-tumor or antiviral immunity. Few studies have investigated the expression and distribution of ERp57 in tumors in addition to its clinical significance. Mehta et al. previously reported a 28-60% loss of ERp57 expression depending on histological type. They suggested that downregulation of ERp57 was significantly associated with HLA class I downregulation, a potential contributor to tumor progression (Mehta et al., 2008). Recently, Hasim and colleagues analyzed the association of cervical carcinogenesis with the aberrant regulation of HLA class I and expression of antigen processing machinery components including ERp57 (Hasim et al., 2012). They immunohistochemically examined the expression  of ERp57 in cervical lesions (64 CINs and 63 SCCs) and detected a loss of ERp57 expression in both CIN (42%) and SCC (56%). They also reported that ERp57 protein expression was significantly lower in lower stage tumors than in higher stages and it was inversely correlated with tumor differentiation. Consistent with previous studies, we found that ERp57 expression decreased according to the progression of cervical tumorigenesis (p50.001) and was significantly stronger in well and moderately differentiated tumors in comparison with poorly differentiated ones (p50.001). However, in contrast to their study, there was no significant difference in ERp57 expression between FIGO stages I and II. This discrepancy may be explained by a lack of standardized immunohistochemical methods, different standards of interpretation, small number of examined cases and/or a different disease spectrum in the cases studied.
With regard to different cancer cell types, the fact that ERp57 expression was significantly lower in squamous cell carcinomas (mean histoscore ¼ 5.31) compared with non-SCC types (mainly adenocarcinoma/adenosquamous carcinomas; 81.5%, 22 out of 27) (mean histoscore ¼ 7.30) was a finding of interest, as it suggests that the two aforementioned carcinomas can be distinguished based on genetic alterations. Furthermore, this finding demonstrates that each histologic type of cervical cancer develops via a distinct molecular mechanism. Consistent with our findings, Mehta et al. reported a higher frequency of ERp57 expression loss in squamous cell carcinomas (60%, 45 out of 76) compared with adenocarcinomas (28%, 7 out of 25) in their IHC study HLA class I loss in cervical carcinomas (Mehta et al., 2008). These findings suggest that each histologic type of cervical cancer could develop independently, and further studies are needed to identify the molecular genetic changes and percentage of ERp57 downregulation according to cell type. We also included HPV infection status in our analysis and observed no statistical difference between the expression of ERp57 in HPV-positive and -negative patients. Whether ERp57 expression is directly regulated by high-risk HPV remains unclear.
Our most notable finding is that low expression of ERp57 in cervical cancer predicts a shorter disease-free and overall survival. To our knowledge, the negative influence of ERp57 expression on the prognosis of cervical carcinoma patients has not been described previously. Multivariate Cox hazard analysis indicated that low ERp57 expression, large tumor size and LN metastasis are significantly associated with poorer disease-free survival in cervical cancer patients. Moreover, low ERp57 expression and non-SCC cell type were independent prognostic factors for overall survival on . Kaplan-Meier plots for disease-free survival (a) and overall survival (b) for patients categorized by ERp57 expression. ERp57-, low ERp57 expression (histoscore, 0-6); ERp57þ, ERp57 high expression (histoscore, 7-12).  (Leys et al., 2007). Thus, given the role of ERp57 in the assembly of the MHC class I antigen processing complex, these findings suggest that downregulation of ERp57 may contribute to more aggressive tumor behavior. Furthermore, this study supports the notion that low ERp57 expression could serve as a novel and independent predictor of recurrence and death in early-stage cervical cancer patients. Because downregulation of ERp57 was an independent predictor of poor survival in early-stage disease, one can therefore infer that evaluation of ERp57 in patients with early-stage disease might be of significant value in identifying individuals who would benefit from more aggressive post-operative adjuvant therapies.
In conclusion, this is the first study to find that ERp57 expression is associated with the progression of cervical carcinogenesis, with a gradual reduction in the order of progression from normal epithelium to CIN to cancer. We also discovered important correlations between the downregulation of ERp57 and clinicopathological parameters that might be potentially useful clinically. Moreover, we suggest that downregulation of ERp57 expression is an independent predictor of poor prognosis in early-stage cervical cancer.

Declaration of interest
No potential conflicts of interest were disclosed.