Elevated mRNA expression of CHAC1 splicing variants is associated with poor outcome for breast and ovarian cancer patients

Background: The role of CHAC1 (cation transport regulator-like protein 1), a recently identified component of the unfolded protein response (UPR) pathway, in gynaecological cancers has not yet been characterised. Now, this work illustrates CHAC1 mRNA expression and associated clinical outcome in breast and ovarian cancer. Methods: The prognostic value of CHAC1 and its two transcript variants was investigated in 116 breast and 133 ovarian tissues using quantitative real-time reverse-transcriptase PCR. Subsequently, we conducted functional studies using short-interfering RNA-mediated knockdown and plasmid-mediated overexpression of CHAC1 in breast and ovarian cancer cells. Results: Poorly differentiated tumours exhibited higher CHAC1 mRNA expression (breast cancer: P=0.004; ovarian cancer: P=0.024). Hormone receptor-negative breast tumours and advanced-staged ovarian cancers demonstrated elevated CHAC1 mRNA expression levels (P<0.001 and P=0.026, respectively). The multivariate survival analysis showed a prognostic value of both transcript variants in breast cancer (transcript variant 1: RRdeath 6.7 (2.4–18.9); P<0.001), RRrelapse 6.7 (2.1–21.3); P=0.001); (transcript variant 2: RRdeath 4.9 (2.0–12.4); P<0.001), RRrelapse 8.0 (2.4–26.8); P<0.001). Ovarian cancer patients aged younger than 62.6 years with high CHAC1 mRNA expression showed poorer relapse-free- and overall-survival (P=0.030 and P=0.012, respectively). In functional studies CHAC1 knockdown suppressed cell migration, whereas ectopic overexpression opposed these effects. Conclusion: High CHAC1 mRNA expression could be an independent indicator for elevated risk of cancer recurrence in breast and ovarian cancer.

genes, was shown to be differentially expressed in pancreatic ductal adenocarcinoma in comparison with normal pancreatic ducts (Buchholz et al, 2005). Two alternatively spliced transcript variants of this molecule have been described until now. Transcript variant 1 (GenBank: NM_024111.3) represents the longer transcript and encodes the longer isoform a. Transcript variant 2 (GenBank: NM_001142776.1) lacks an alternate in-frame segment, compared with variant 1, resulting in a shorter protein (isoform b), compared with isoform a.
In this study we investigated the prognostic role of CHAC1 mRNA expression and its transcript variants in breast and ovarian cancer patients, and we describe the effect of CHAC1 on cell migration in breast and ovarian cancer to better understand its role in tumour biology and potential implication for cancer progression.

General study design, patients and samples
We retrospectively studied CHAC1 mRNA expression by applying qRT -PCR to prospectively collected breast and ovarian tissue samples from patients with primary breast neoplasm treated at our Department (Department of Gynaecology and Obstetrics, Innsbruck Medical University, Austria). Samples have been collected during primary surgery. Univariate and multivariate survival models were used to explore the potential of CHAC1 mRNA expression in predicting relapse-free (RFS) and overallsurvival (OS).
Furthermore, we re-examined the breast and ovarian cancer cases with respect to CHAC1 transcript variants 1 and 2 mRNA expression. Finally, we conducted functional studies using shortinterfering RNA (siRNA)-mediated knockdown and plasmidmediated overexpression of CHAC1 in breast cancer cells (Hs578T,  and ovarian cancer cells (HOC-7).
Clinical, pathological and follow-up data were stored in a database according to our hospital privacy rules. The patients were treated at our Department between February 1989 andMarch 2004. Written informed consent is not available from all patients. But in accordance with the Austrian law, the study was approved by the Ethics Committee of the Innsbruck Medical University (reference number: AM3634) and conducted in accordance with the Declaration of Helsinki Principles. Patient consent was not obtained under a waiver IRB approval. All samples were anonymised to guarantee the protection of privacy before performing the analysis. The study was performed in concordance with the Reporting Recommendations for Tumour Marker Prognostic Studies of the National Cancer Institute (McShane et al, 2005).
Frozen breast-tissue samples from 106 patients with primary breast cancer (aged 35.5-89.7 years; median age at diagnosis, 60.4 years) and 10 patients with benign breast diseases (aged 27.3-66.9 years, median age at diagnosis, 40.1 years), frozen ovarian-tissue samples from 103 patients with ovarian cancer (aged 24.1-87.1 years; median age at diagnosis, 62.6 years) and 30 patients with benign ovarian diseases (aged 38.2-85.8 years; median age at diagnosis, 55.7 years) were analysed. In ovarian cancer patients staging was performed in accordance with the International Federation of Gynaecology and Obstetrics (FIGO) classification system.
All patients were monitored within the outpatient follow-up programme of our Department.
The median observation period of the breast cancer patients was 7.5 years (0.9 -17.0) and 4.13 years (0.09 -20.0) for the ovarian cancer patients. No neoadjuvant chemotherapy was applied to the patients included in the study.
A platinum-based chemotherapy was part of the treatment for all but 12 ovarian cancer patients (8, 2 and 2 who had FIGO stage I, II and III, respectively). Clinicopathological features of all patients are summarised in Table 1. Tumour specimens were obtained immediately after surgery, brought to our pathologist, a part of the tissue was pulverised under cooling with liquid nitrogen and stored at À70 1C. Oestrogen receptor (ER) and progesterone receptor (PR) status was identified by immunohistochemistry.

Cell culture
The human breast cancer cell lines Hs578T and BT-20 were obtained from the American Type Culture Collection (ATCC, Manassas, VA, USA) and cells were cultured as recommended by the ATCC. Human ovarian carcinoma cell line HOC-7 was kindly provided by Dr C Dittrich (University of Vienna, Vienna, Austria) and cultured under standard conditions (Marth et al, 1997). Amplification of 15 STR loci and the gender-specific locus amelogenin was carried out in the Institute of Legal Medicine, Innsbruck Medical University, to authenticate these cell lines as described recently (Parson et al, 2005).
Small-interfering RNA transfections were performed according to the manufacturer's protocol. All cell lines were transfected with 5 nmol l À1 siRNA against CHAC1 or control siRNA. Two days after transfection, the cells were treated further for in-vitro scratch assay or proliferation assay, respectively, and subsequently collected for western blots and real-time quantitative PCR analysis.

CHAC1-overexpressing cells
We used CHAC1 -pcDNA6 plasmid (CHAC1 -V5eGFP construct) provided by the University of California at Los Angeles, USA. The control plasmid was produced by excision of the CHAC1 sequence using Xba1 and Xho1 (Fermentas, Leon-Rot, Germany), from the pcDNA6-V5eGFP backbone. Plasmid DNA was purified using miniprep and/or midiprep kits from Qiagen.

In vitro scratch assay
The assay was performed as recently described (Liang et al, 2007;Berger et al, 2010). Untransfected cells (BT-20, Hs578T and HOC-7), CHAC1-knockdown cells, CHAC1 overexpressing and mock-treated cells were scratched by a pipette tip when cell confluence reached B90% and further cultivated for 24 h in fresh medium with reduced FCS amounts. The same fields were photographed immediately (0 h), and 4, 8 and 16 h later (Hs578T), or 24 and 48 h later (BT-20), or 14 and 48 h later (HOC-7) using a Kappa PS30 camera (Kappa opto-electronics GmbH, Gleichen, Germany). The experiments were repeated at least three times. Gap widths were measured using the free ImageJ software (http:// rsbweb.nih.gov/ij/).

Apoptosis detection
Biotin-labeled POD TUNEL Apoptosis detection kit for adherent cell was used according to the manufacturer's protocol (GenScript, Piscataway, NJ, USA). Cells were counterstained with DAPI and analysed using an Olympus 1 Â 70 inverted microscope (Olympus, Tokyo, Japan) in conjunction with Kappa ImageBase software V2.7.2.

RNA isolation and mRNA expression analysis
Procedures were performed as previously described (Widschwendter et al, 2000;Mueller et al, 2003). Primers and probe for qRT-PCR for CHAC1 were purchased from Applied Biosystems (Foster City, CA, USA, Applied Biosystems Assay ID: Hs00899499_g1). Primers and probes for the TATA box-binding protein (endogenous RNA control) were used according to Bièche et al (1999). All reactions were checked if they are specific for mRNA and do not amplify genomic DNA.

Statistical analysis
Descriptive results are presented as median and interquartile range. For the comparison of CHAC1 expression between groups, a two-sided t-test, or in case of comparisons between more than two groups, ANOVA was applied to the log-transformed values of CHAC1 expression. The correlation between the log-transformed CHAC1 expression values was analysed using the Pearson's correlation coefficient. For survival analysis, the CHAC1 mRNA expression was dichotomised into low and high using the median expression value.
Relapse-free survival was defined as the time from surgery to histopathological confirmation of distant metastases or regional recurrence. Overall survival was defined as the time from surgery to death from any cause or to the last clinical inspection. To estimate hazard ratios with 95% confidence intervals, we first calculated univariate Kaplan -Meier curves for dichotomised age, tumour size, stage, grade, chemotherapy and CHAC1 mRNA expression using the log-rank test to compare the survival distributions between groups. For breast cancer specimens additionally menopausal and lymph node status, hormone receptor (HR) and HER-2/neu status and the application of endocrine and radiation therapies were considered.
A time-independent Cox proportional hazard approach was used for multivariate survival analysis using all variables of the univariate analysis (Table 2). Furthermore, a backward conditional stepwise variable selection procedure was used, with Po0.05 for entering and P40.1 for removing a variable in the model. For both multivariate models dichotomised CHAC1, CHAC1 transcript In sensitivity analysis we further aimed to flexibly investigate the effect of CHAC1 and CHAC1 transcript variants 1 and 2 on the above endpoints, including CHAC1 as a log-transformed continuous variable in univariate and multivariate models, using penalised splines (P-splines) in extended, restricted maximumlikelihood optimal Cox-type additive hazard regression (Strasak et al, 2009). Data points of the 10 -90% interpercentile range of CHAC1 were used for the calculation of Supplementary

CHAC1 mRNA expression and clinical outcomes in breast cancer
We found an increase of CHAC1 mRNA expression from benign neoplastic tissues through to grade III cancer tissues (P ¼ 0.014; Figure 1A). Clinicopathological analysis of cancer tissues provided a similar result (P ¼ 0.004; Table 1). Increased CHAC1 mRNA expression was found in tumours lacking ER (Po0.001) and   Table 1a). Furthermore, we identified a positive association with tumour size (P ¼ 0.011; Table 1a). Univariate survival analysis of all 106 breast cancer patients revealed no significant association of total CHAC1 mRNA expression with the clinical endpoints (Table 2a), whereas high total CHAC1 mRNA expression was strongly associated with a high risk of death and relapse in the multivariate Cox regression model (RR death (1.3 -7.1); P ¼ 0.012, RR relapse 4.8 (1.6 -14.6); P ¼ 0.005) ( Table 3). CHAC1 also remained as strongest independent factor in the backward Cox regression model (RR death 2.9 (1.3 -6.4); P ¼ 0.008, RR relapse 5.2 (1.8 -15.2); P ¼ 0.002).

Statistical validation of survival associations in breast cancer
Validating the association of CHAC1 and its transcript variants with survival within breast-cancer entities, a P-spline regression

CHAC1 mRNA expression and clinical outcomes in ovarian cancer
To analyse whether an aberrant CHAC1 mRNA expression is also associated with poor outcomes in other female malignancies, we analysed 103 ovarian cancer tissues and 30 normal ovarian specimens. The mRNA expression values of total CHAC1 strongly correlated with the CHAC1 transcript variants 1 and 2 also in ovarian cancer (r ¼ 0.88; Po0.001 and r ¼ 0.96; Po0.001).
We observed a significant increasing trend of CHAC1 mRNA expression from non-neoplastic tissues to grade III neoplastic tissues (Po0.001; Figure 1B). Again, poorly differentiated cancers demonstrated higher CHAC1 levels (P ¼ 0.024; Table 1b). A high CHAC1 mRNA expression was also associated with advanced tumour stage (P ¼ 0.026; Table 1b).
Univariate analysis of 103 ovarian cancer patients revealed prognostic significance for CHAC1 mRNA expression only in younger patients (omedian age of 62.3 years) for OS (P ¼ 0.012; Table 2b, Figure 1E) and RFS (P ¼ 0.03; Table 2b, Figure 1F). The same findings were revealed for premenopausal women (data not shown).
In ovarian cancer, the multivariate survival analysis showed no significant prognostic value of CHAC1 (data not shown).

CHAC1 influences cell migration and proliferation
Owing to the identified associations of CHAC1 mRNA expression and clinical outcome data in breast and in ovarian cancer patients, we were interested in functional effects of CHAC1 knockdown and CHAC1 overexpression in cancer cells.
Hence, we measured cell migration by means of an in vitro scratch assay and proliferation in Hs578T and BT-20 breast cancer and HOC-7 ovarian cancer wild-type cells, CHAC1 knockdown cells and cells treated with a scrbl siRNA as negative control.
In Hs578T breast cancer cells a 96% or 35% CHAC1 knockdown was revealed at the mRNA level or the protein-level, respectively, in comparison with scrbl siRNA-treated cells (Figure 2A). We identified a significantly reduced migration and proliferation in CHAC1 knockdown cells ( Figure 2B). Apoptosis measurements by means of FACS analysis or TUNEL staining, respectively, showed no differences between knockdown and control cells (data not shown).
Next, we performed again an in vitro scratch assay to measure cell migration of CHAC1-overexpressing Hs578T cells and cells transfected with a control plasmid as negative control. In comparison with the control cells a 83-fold or 10-fold increase was revealed in CHAC1 mRNA or protein expression, respectively, in Hs578T cells ( Figure 2C). We identified significantly increased migration and proliferation in CHAC1-overexpressing cells in comparison with mock-treated cells, respectively ( Figure 2D), but no effect on apoptosis (data not shown). In BT-20 breast cancer cells, a 56% CHAC1 knockdown was revealed at the protein level (Supplementary Figure S2A). In this cell line only a tendency of a reduced migration and proliferation was observed (Supplementary Figure S2B).
In CHAC1-overexpressing BT-20 cells (1.5-fold increase in protein expression; Supplementary Figure S2C), we identified an increased migration and only a tendency of an increased proliferation (Supplementary Figure S2D).
To elucidate the role of CHAC1 in ovarian cancer we analysed CHAC1 knockdown and overexpression in HOC-7 ovarian cancer cells. In HOC-7 cells, a 21% CHAC1 knockdown was revealed at the protein level (Supplementary Figure S3A). Again, we identified a reduced migration in the knockdown cells in comparison with the control cells without affecting proliferation (Supplementary Figure S3B). In CHAC1overexpressing HOC-7 cells (1.3-fold increase in protein expression; Supplementary Figure S3C) an increased migration was observed without affecting proliferation (Supplementary Figure S3D).

DISCUSSION
This is the first pilot study, which shows an association of CHAC1 mRNA expression in tumour tissues with the survival of breast and ovarian cancer patients. CHAC1 has been identified as a novel proapoptotic component of the UPR pathway, which itself responds to endoplasmic reticulum stress (Gargalovic et al, 2006;Mungrue et al, 2009).
We identified a positive correlation between poor tumour differentiation and higher CHAC1 mRNA expression levels in breast and ovarian cancer. Recently it was described in human parthenogenetic-induced pluripotent stem cells that CHAC1 is negatively regulated by miRNA-370, which is an imprinted miRNA (Stelzer et al, 2011). It was suggested that distinct regulatory imprinted small RNAs such as miRNA-370 and their targets such as CHAC1 have substantial roles in cellular differentiation (Stelzer et al, 2011).   Moreover, we found that a high mRNA expression of CHAC1 or related transcript variants were an independent poor prognostic marker for outcome in breast cancer patients. In the present study, the overall 5-year survival rate of breast cancer patients with low CHAC1 expression was 49% compared with only 35% for patients with high CHAC1 expression. Although this difference appears substantial when inspecting the Kaplan -Meier curves, it only reached statistical significance for the expression of both transcript variants in the univariate survival model for RFS (74% vs 53% and 78% vs 49%, respectively) and only for transcript variant 2 in OS (50 vs 34%). Considering the significant association of CHAC1 expression in breast cancer samples with most of the clinicopathological features (and their role as confounders), an inclusion of the features in the multivariate model showed a consistent strong independent prognostic role of CHAC1 and its splicing variants for RFS and OS, which was confirmed by additional sensitivity analyses. This finding is also supported by the significant correlation of CHAC1 expression and CHAC1 transcript variants. As mentioned above, on the basis of the increased expression of CHAC1 in HR-negative breast tumours, a stratified multivariate subgroup analysis might provide more insight and address the question, if the prognostic value of CHAC1 differs between patients with poorly differentiated or aggressive tumour cells and tumour entities with presence of HR on their surface. As a limitation, our case number did not provide this possibility. Furthermore it should be considered that the case -control design hinders the establishment of a causal relationship between elevated CHAC1 and poor outcome.
In ovarian cancer we observed only in younger patients (age omedian age of 62.6 years as well as premenopausal women) an association between high CHAC1 mRNA expression levels and poor OS and RFS. The prognostic significance of CHAC1 mRNA expression determined specifically for younger women could provide indication that strong CHAC1 mRNA expression may be associated with an earlier development of more aggressive tumours.
Owing to the associations identified between high CHAC1 mRNA expression levels and poor survival, especially with RFS, in breast and partly in ovarian cancer patients, we hypothesised that CHAC1 may have a role in cell migration and proliferation. We found a significantly reduced migration and proliferation in vitro, in CHAC1 knockdown Hs578T breast cancer cells and, conversely, we witnessed increased migration and proliferation in CHAC1-overexpressing cells. But in CHAC1 knockdown BT-20 cells only a tendency of a reduced migration in vitro was observed, whereas in CHAC1 overexpression BT-20 cells an increased migration was identified.
In CHAC1-knockdown HOC-7 ovarian cancer cells we found a significantly reduced migration and conversely a tendency of an increased migration in CHAC1-overexpressing cells.
Recently, CHAC1 was described as a novel proapoptotic component in human aortic endothelial cells and in human embryonic kidney cells (HEK 293;Mungrue et al, 2009). However, in our CHAC1-knockdown and overexpression experiments no association with apoptosis was observed. It is known that the UPR pathway activation in cancer might result in apoptosis and disease resolution, or an anti-apoptotic, pro-angiogenic drive, resulting in disease progression (Ma and Hendershot, 2004;Scriven et al, 2007). Hence, it remains unclear how UPR activation in solid tumours balances cell survival and cell death.
As UPR is suspected to be responsible for the failure of some patients to respond to chemotherapy, it could provide a target for improving existing treatments or the discovery of new anti-cancer targets (Scriven et al, 2009). Additional studies will improve the understanding of the link between CHAC1 and patient resistance to breast cancer therapies.
In summary, our data show that CHAC1 correlated with tumour differentiation and survival in breast and partly in ovarian cancer. Cell migration was revealed to be in part affected by CHAC1 expression. Owing to the limitations of this pilot study, further studies should elucidate the role of CHAC1 and its transcript variants as potential biomarkers for identifying patients with high risk of cancer recurrence.