CCDC170 affects breast cancer apoptosis through IRE1 pathway

Genome-wide association studies have revealed that multiple single-nucleotide polymorphisms in the intergenic region between estrogen receptor 1 and coiled-coil domain containing 170 (CCDC170) are associated with breast cancer risk. We performed microarray and bioinformatics analyses to identify genes that were induced upon CCDC170 overexpression, and confirmed our findings by evaluating paraffin-embedded breast cancer tissues and conducting cellular assays. In CCDC170-overexpressing MCF7 breast cancer cells, microarray analyses revealed that inositol-requiring enzyme 1 (IRE1) was the most elevated gene in enriched pathways. In breast cancer tissues, IRE1 expression correlated positively with CCDC170 and X-box binding protein 1 expression at both the mRNA and protein levels. In a survival analysis, patients with higher CCDC170 levels exhibited better disease-free survival. Western blotting indicated that overexpressing CCDC170 in MCF7 cells increased protein levels of IRE1α, estrogen receptor α and X-box binding protein 1, while silencing CCDC170 reduced them. CCDC170 overexpression promoted apoptosis in MCF7 cells, and this effect was more obvious under endoplasmic reticulum stress. MCF7 cells overexpressing CCDC170 were more sensitive to paclitaxel. Our study showed that higher CCDC170 expression is associated with a better prognosis in breast cancer patients and that CCDC170 may promote apoptosis through the IRE1α pathway.


INTRODUCTION
Breast cancer is one of the most common malignant tumors threatening women's health worldwide [1]. Environmental and genetic factors are known to influence breast cancer risk. Hereditary breast cancer accounts for 5-10% of all breast cancer cases, and pathogenic variants in the BRCA1/2 genes have been detected in approximately 90% of hereditary breast cancer cases [2,3]. In addition, many genetic polymorphisms have been reported to be associated with breast cancer risk [4]. Genome-wide association studies have revealed that single-nucleotide polymorphisms residing in the intergenic region between estrogen AGING receptor 1 (ESR1) and coiled-coil domain containing 170 (CCDC170) at 6q25.1 are associated with breast cancer risk. While the function of ESR1 in breast cancer has been well studied, that of CCDC170 remains elusive [5].
CCDC170, also known as C6orf97, is a largely uncharacterized open reading frame located about 69 kilobases upstream of ESR1 on 6q25.1, and spanning around 127 kilobases (http://genome.ucsc.edu/, [GRCh38/hg38]). CCDC170 was found to be coexpressed with ESR1 in breast cancer tissues [6], and an ESR1-CCDC170 rearrangement was discovered in luminal B breast tumors [7]. ERα, the protein encoded by ESR1, is well known to be an independent prognostic factor in breast cancer, and thus is a target of endocrine therapy [8]. We speculated that CCDC170 might have important functions and clinical significance in breast cancer, considering its relationship with ESR1.
Inositol-requiring enzyme 1 alpha (IRE1α), encoded by IRE1, is an essential signal transducer in the most conserved unfolded protein response signaling branch. IRE1α has multiple functions, including unconventionally shearing X-box binding protein 1 (XBP1), triggering apoptosis and inducing autophagy [9][10][11][12]. Upon the activation of the unfolded protein response, IRE1α cleaves 26 nucleotides from the unspliced XBP1 mRNA sequence, yielding a spliced form that encodes the active transcription factor XBP1s. XBP1s subsequently enhances the expression of genes involved in protein folding and degradation [10]. Notably, ESR1 and XBP1 are co-expressed in breast cancer tissues [13,14], and XBP1s can bind to the Xbox in the ESR1 promoter, thus inducing its transcription [15,16].
Apoptosis is one of the principal mechanisms used to induce cell death in cancer therapeutics. Tumor resistance occurs when tumor cells adapt to therapy by activating anti-apoptotic pathways and altering the tumor microenvironment [17]. Apoptotic cell death following the unfolded protein response depends primarily on IRE1α [18]. Thus, the IRE1α-XBP1 pathway has been an attractive cancer therapeutic target [19,20], and has been identified as a vital determinant of drug resistance [21][22][23][24].
In this study, we constructed a gene co-expression network to identify genes that might be associated with CCDC170 function. Our findings suggested the existence of a regulatory loop involving CCDC170, IRE1α, XBP1 and ERα. Thus, we investigated the effects of CCDC170 on apoptosis, and examined whether CCDC170 activity depended on the IRE1α pathway.

The influence of CCDC170 on the gene expression profile of MCF7 cells
To identify genes that could be induced by CCDC170, we transfected MCF7 breast cancer cells with either a pCMV-N-Flag-CCDC170 vector or a pCMV-N-Flag control vector. We extracted mRNA from the cells 24 or 48 hours post-transfection, and used microarrays to assess their gene expression profiles. Then, we performed an enrichment analysis of the differentially expressed genes (DEGs) between the CCDC170overexpressing group and the control group. When we compared the top 20 significantly enriched pathways between the 24-hour group ( Figure 1A) and the 48-hour group ( Figure 1B), we identified eight common pathways ( Figure 1C). We also determined the top 20 genes with consistent expression-change tendencies in the 24-and 48-hour groups ( Figure 1D). Among these genes, only IRE1 and AKT1 were involved in the eight pathways mentioned above. The change in IRE1 expression (24h: log[fold ratio] = 0.69; 48h: log[fold ratio] = 0.33) was the most obvious in the apoptotic pathway ( Figure 1E).
Next, we determined the correlation between CCDC170 and IRE1 expression in breast cancer tissues from The Cancer Genome Atlas (TCGA) and the Gene Expression Omnibus (GEO). There was a significant positive correlation between CCDC170 and IRE1 levels in TCGA (r = 0.273, P = 5.89×10 -14 ; (Figure 1F), as well as in GEO (r = 0.262, P = 9.19×10 -55 ; (Supplementary Figure 1A). When cases in TCGA were stratified by subtype, CCDC170 mRNA levels were higher in the luminal A and luminal B subtypes, but were lower in the human epidermal growth factor receptor 2 (Her2)-positive and Basal-like breast cancer subtypes ( Figure 1G). A similar tendency was found in the GEO database (Supplementary Figure 2F).

The expression of CCDC170, IRE1α and XBP1s in breast cancer tissues
To investigate the prognostic value of CCDC170, IRE1α and XBP1s in breast cancer, we used immunohistochemistry to assess the expression of these proteins in 100 patients with invasive ductal carcinoma. General information about the included breast cancer patients can be found in Supplementary Table 1. IRE1α levels were found to correlate with CCDC170 levels (r = 0.233, P = 0.020; Figure 2B) and XBP1s levels (r = 0.212, P = 0.034; Table 1). Additionally, CCDC170 levels correlated positively with XBP1s levels (r = 0.339, P = 0.001; Table 1). In terms of clinicopathological characteristics, CCDC170 levels AGING correlated positively with ERα levels (r = 0.389, P = 9.90×10 -5 ). IRE1α levels correlated positively with Her2 levels (r = 0.293, P = 0.003) and Ki67 levels (r = 0.208, P = 0.038). XBP1s levels exhibited a notable positive correlation with ERα levels (r = 0.286, P = 0.004). Moreover, CCDC170 levels (r = 0.333, P = 0.006) and IRE1α levels (r = 0.353, P = 0.003) correlated with the molecular subtypes. However, CCDC170, IRE1α and XBP1s levels were not significantly associated with the tumor size, lymph node status, Tumor-Node-Metastasis (TNM) stage or histologic grade (Table 1). Based on these findings, we explored the correlation of CCDC170 levels with XBP1 and ESR1 levels in TCGA and the GEO database. In TCGA, CCDC170 levels were positively associated with XBP1 levels (r = 0.636, P = 2.

Survival analysis based on CCDC170, IRE1α and XBP1s expression
Next, we performed Kaplan-Meier analyses in our 100 patients according to their expression of CCDC170, IRE1α and XBP1s. The CCDC170 high-expression group exhibited better disease-free survival (DFS) than the low-expression group (hazard ratio [HR] [95% confidence interval (CI)] = 0.37 [0.14 -0.98], P = 0.037; Figure 2D). However, there was no significant difference in overall survival (OS) between the CCDC170 low-and high-expression groups (HR [95% CI] = 0.42 [0.11 -1.57], P = 0.183; Figure 2C). We did not find differences in DFS or OS between the high and low-expression groups for IRE1α ( Figure 2E, 2F) or XBP1s ( Figure 2G, 2H). Considering that our patients had received different post-operative treatments, we also analyzed the association between CCDC170 expression and prognosis within cohorts that had received the same treatment. The results indicated that higher CCDC170 expression predicted better DFS in the chemotherapy cohort (Supplementary Figure 5D) and the radiation cohort (Supplementary Figure 5H), and predicted better OS in the radiation cohort (Supplementary Figure 5G).
We also performed Kaplan-Meier analyses using mRNA expression data from TCGA and GEO. In the GEO database, the high-expression groups for CCDC170 (HR   ER, estrogen receptor α; PR, progesterone receptor; Her-2, human epidermal growth factor receptor 2; TNBC, triple negative breast cancer; HER2, Her-2 overexpressed breast cancer; -, negative; +, positive.

CCDC170 promoted apoptosis and enhanced the sensitivity of MCF7 cells to paclitaxel
We then investigated the effects of CCDC170 on breast cancer cell apoptosis and viability. In cells that were transiently transfected with a eukaryotic CCDC170 expression vector for 24 hours, the percentage of apoptotic cells was 1.31 times higher than in the control group (P = 0.044; Figure 4A, 4B). On the other hand, when CCDC170 was transiently silenced using small interfering RNA (siRNA), flow cytometry revealed that the percentage of apoptotic cells decreased by 28.30% (P = 0.047; Figure 4C, 4D). We also found that the number of apoptotic nuclei increased in cells that transiently overexpressed CCDC170 (P = 7.00 ×10 -3 ; Figure 4E, 4F). The expression of Caspase-12 increased when CCDC170 was upregulated transiently (P = 0.005; Figure 4G, 4H) or stably (P = 0.002; Figure 4I, 4J). Furthermore, whether CCDC170 was overexpressed transiently (P = 0.001; Figure 5J) or stably (P = 0.003; Figure 5K) in MCF7 breast cancer cells, the cell viability was significantly lower than that of the control group in a 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT) assay. AGING Next, we used immunohistochemistry to assess the localization of CCDC170 in paraffin-embedded breast cancer tissues, and found that it was expressed in the cytoplasm. Immunofluorescence staining further indicated that CCDC170 partially overlapped with calnexin, a marker of the endoplasmic reticulum ( Figure  5A). Nevertheless, no binding between CCDC170 and IRE1α was observed in a co-immunoprecipitation assay (data not presented).
Endoplasmic reticulum stress is ubiquitous in the tumor microenvironment; thus, we simulated endoplasmic reticulum stress in vitro by treating MCF7 cells with 300 nm thapsigargin for 24 hours. Western blot analyses demonstrated that MCF7 breast cancer cells that transiently overexpressed CCDC170 expressed higher levels of IRE1α (P = 0.018), cleaved poly [ADP-ribose] polymerase (PARP, an apoptotic regulatory protein; P = 0.001) and Caspase-7 (P = 0.001) and lower levels of Bcl-2 (an apoptosis-inhibiting protein; P = 0.013) than control cells under endoplasmic reticulum stress ( Figure 5B, 5C). Likewise, in MCF7 cells that stably overexpressed CCDC170, the expression of IRE1α (P = 0.009), cleaved PARP (P = 9.27×10 -4 ) and Caspase-7 (P = 4.36×10 -4 ) increased while the expression of Bcl-2 decreased (P = 0.025) compared with control cells under endoplasmic reticulum stress for 24 hours (Figure 5D, 5E). Flow cytometry analysis demonstrated that the proportion of apoptotic cells was 1.37 times higher in the CCDC170-overexpressing group than in the control group after 24 hours of endoplasmic reticulum stress (P = 0.001; Figure 5F, 5G). In addition, we found that the percentage of apoptotic cells decreased when IRE1 was knocked down without endoplasmic reticulum stress in MCF7 cells that transiently overexpressed CCDC170 (P = 0.046; Figure 5H, 5I).
To compare the paclitaxel chemosensitivity between breast cancer cells with different CCDC170 levels, we treated CCDC170-stably-overexpressing MCF7 cells and control MCF7 cells with 100 nM paclitaxel. An MTT assay indicated that paclitaxel treatment inhibited growth more effectively in CCDC170-stablyoverexpressing cells than in control cells ( Figure 5L).

DISCUSSION
Several single-nucleotide polymorphisms around human CCDC170 have been identified as important breast cancer risk indicators in Chinese women [25][26][27]. In addition, CCDC170 was found to be tightly coexpressed with ESR1 in breast tumor biopsies and cells [6]. A previous study demonstrated that CCDC170 was fused to ESR1 and employed the constitutively active ESR1 promoter to induce the expression of a truncated form of CCDC170. The fused gene was enriched in luminal B breast tumors and was found to promote a more aggressive phenotype by enhancing cell migration, invasion, anchorage-independent growth and endocrine therapy resistance [7]. However, it has been proposed that CCDC170 can function as either an oncogene or a tumor suppressor [28]. Indeed, higher CCDC170 expression has been associated with a better prognosis in certain breast cancer subtypes, but with a poorer prognosis in others [6,29]. Though these data have highlighted the importance of the CCDC170 gene and its fusion protein in breast cancer, the pathobiology and clinical relevance of CCDC170 have remained unclear.
Microarrays are a powerful technology for assessing expression profiles across entire genomes and discovering DEGs that contribute to the phenotypes, treatment responses and heterogeneity of complex diseases [30][31][32][33][34]. In the present work, we first used a gene chip to identify genes that were differentially expressed upon ectopic CCDC170 expression in MCF7 breast cancer cells. Bioinformatic analyses indicated that IRE1 was the most notably elevated gene in the top 20 apoptotic pathways.
IRE1α is a stress sensor that performs a myriad of cellular functions in a complex signaling network [35][36][37][38][39]. IRE1α is well known to cleave XBP1 in the most conserved arm of the unfolded protein response, which is an important determinant of cell death and survival [40]. IRE1α expression has been reported to be higher in breast cancer tissues than in the surrounding noncancerous tissues [41]. In the present study, the positive correlation between CCDC170 and IRE1 levels in our microarray analysis was validated at both the mRNA and protein levels in breast cancer tissues and cells.
XBP1s is a transcription factor that can induce the expression of ESR1, and these two genes are coexpressed in breast cancer tissues [13][14][15][16]. A recent study revealed that an XBP1 gene signature was expressed at significantly higher levels in ERα+ or nontriple-negative breast cancer samples than in ERαor triple-negative breast cancer samples, respectively [42]. Our immunohistochemical analyses demonstrated that CCDC170 and XBP1s were preferentially expressed in ERα+ breast cancer tissues, consistent with previous observations [6,42]. Western blotting and immunofluorescence analyses indicated that ERα, IRE1α and XBP1s levels correlated with CCDC170 levels in MCF7 breast cancer cells. These findings suggested the existence of a potential regulatory loop among these proteins.
While the physiological function of CCDC170 has not been well understood, previous studies have suggested AGING that CCDC170 expression alters the prognosis of breast cancer patients. A study conducted in Japanese populations demonstrated that higher CCDC170 expression correlated with better relapse-free survival in luminal A subtype patients, but correlated inversely with relapse-free survival in luminal B subtype patients [29]. Dunbier et al. reported that higher CCDC170 expression prolonged the DFS of ERα+ patients [6]. Similar to the latter results, our survival analysis revealed that higher CCDC170 expression was associated with better DFS in the entire patient cohort, but was not associated with OS.
In further analyses, we found that higher CCDC170 expression predicted significantly better DFS in the chemotherapy cohort and the radiation cohort, and tended (though not statistically significantly) to be associated with a better prognosis in the endocrine therapy cohort and the non-endocrine therapy cohort. We further analyzed the relevance of CCDC170, IRE1 and XBP1 expression to breast cancer prognosis in two independent cohorts (TCGA and GEO), and found that higher expression of CCDC170, IRE1 and XBP1 correlated with better OS in GEO. Combined with our initial observation that CCDC170 overexpression mainly induced genes involved in apoptosis, these results suggested that CCDC170 is a tumor inhibitor. Notably, although significant results were obtained in our breast cancer tissues, the sample size was relatively moderate. Though we reached consistent conclusions using data from TCGA and GEO, larger samples size will be needed to validate these findings in subsequent research.
In our functional analyses, we discovered that the overexpression of CCDC170 increased the apoptosis of MCF7 cells, while the silencing of CCDC170 reduced it. Remarkably, we found that Caspase-12 was induced in both transiently-and stably-CCDC170-overexpressing MCF7 cells. We also observed that either stable or transient CCDC170 overexpression clearly inhibited the growth of breast cancer cells. Caspase-12, which is located on the cytoplasmic side of the endoplasmic reticulum, is specifically activated during endoplasmic reticulum stress and is a key initiator of cell death [43,44]. In tumor cells, endoplasmic reticulum stress is common because the endoplasmic reticular environment is disturbed [45]. We simulated endoplasmic reticulum stress in vitro by treating MCF7 cells for 24 hours with thapsigargin, which is well known to induce endoplasmic reticulum stress by inhibiting the Ca 2+ -ATPase pump [46]. CCDC170 more noticeably induced apoptosis (and upregulated IRE1α) in breast cancer cells under endoplasmic reticulum stress. Western blotting revealed that pro-apoptotic molecules such as cleaved PARP and Caspase-7 were upregulated in CCDC170overexpressing cells compared with control cells under endoplasmic reticulum stress, while the anti-apoptotic protein Bcl-2 was downregulated. In addition, the rate of apoptosis was higher in CCDC170-overexpressing cells than in control cells under endoplasmic reticulum stress. However, the knockdown of IRE1α impaired the proapoptotic effects of CCDC170. Interestingly, although we found that CCDC170 was partly located on the endoplasmic reticulum membrane, we detected no direct binding between CCDC170 and IRE1α in our coimmunoprecipitation experiments. Thus, the effects of CCDC170 on breast cancer cells may partly depend on the biochemical activation of the IRE1α pathway A previous study indicated that CCDC170 could bind to perinuclear microtubules, enhance their stability and suppress cell migration [28]. Paclitaxel, one of the taxanes, is an antineoplastic chemotherapeutic agent that stabilizes microtubules, and is frequently used to treat cancers such as breast cancer [47]. We found that CCDC170-overexpressing MCF7 cells were more sensitive than control cells to paclitaxel treatment. We also found CCDC170 overexpression increased the expression level of ERα. Although ERα is an important target of endocrine therapy, many proteins work together to maintain a functional niche environment. CCDC170 is also expected to become a therapeutic target in cancer.
The human cDNA microarray analysis in the present study has provided clues into the functions of CCDC170 in breast cancer. Our DEG analysis in MCF7 breast cancer cells revealed that IRE1 was upregulated in CCDC170-overexpressing cells. Higher CCDC170 expression was associated with a better prognosis in breast cancer patients, and the expression of IRE1α, XBP1s and ERα correlated with the expression of CCDC170. The impact of CCDC170 on cell fate could be partially attributed to the IRE1α-XBP1s pathway. Thus, CCDC170 may have considerable potential as a therapeutic target for breast cancer. The broader involvement and clinical relevance of CCDC170 in the pathogenesis of breast cancer will be the focus of future investigations.

Clinical samples
We obtained 100 formalin-fixed, paraffin-embedded tumor tissues from patients with breast invasive ductal carcinoma who had undergone surgical resection at Tianjin Medical University Cancer Institute and Hospital from December 2008 to May 2009. All the patients were followed up effectively through telephone calls or outpatient electronic medical records until September 2017. Detailed demographic data and clinicopathological information were collected retrospectively. The use of the patients' specimens and information was approved by the AGING ethics committee of Tianjin Medical University Cancer Institute and Hospital.

Breast cancer data from public databases
We collected the gene expression profiles of 742 patients with stage I, II or III invasive ductal carcinoma from TCGA (http://www.cbioportal.org/index.do) and 3409 samples from GSE96058 in the GEO repository (https://www.ncbi.nlm.nih.gov/geo/) All gene expression data were uniformly normalized, and pathological information was downloaded. Details on the patients involved in our study are shown in Supplementary Tables 2, 3.

Cell maintenance
MCF7 breast cancer cells were obtained from the American Type Culture Collection. The cells were maintained in RPMI-1640 medium (Invitrogen, USA) supplemented with 10% fetal bovine serum (Thermo Fisher Scientific, USA) at 37° C in a humidified atmosphere containing 5% CO 2 .

Cell transformation
DH5α E. coli were transformed with plasmid DNA (either the pCMV-N-Flag-CCDC170 vector or the pCMV-N-Flag control vector) using the heat shock method. Specifically, 10 ng of the plasmid was mixed with 5 uL of DH5α E. coli and placed on ice for 30 min. Then, the mixture of chemically competent bacteria and DNA was incubated at 42° C for 90 s (heat shock) and placed back on ice for 10 min. Subsequently, 500 uL of lysogeny broth medium was added, and the mixture was incubated at 37° C for 0.5-1 hour. Then, the mixture was centrifuged for 5 min at 8000 x g, and the supernatant was removed. Freshly prepared agar plates with ampicillin were inoculated with 150 uL of this mixture overnight at 37° C.
Subsequently, 4-6 mL of lysogeny broth medium was inoculated with single colonies picked from the freshly prepared agar plates and incubated on a shaker at 37° C for 12-16 hours. The plasmid was extracted and purified using the EZgene™ EndoFree Plasmid ezFlow Miniprep Kit II (PD1222, BIOMIGA) in strict adherence to the manufacturer's instructions. The plasmid concentration was measured on a NanoDrop 2000c Spectrophotometer (Thermo Fisher Scientific, Waltham, MA, USA).

Transfection
MCF7 cells (3 × 10 5 ) were seeded on six-well plates for 12-16 hours. For the transient CCDC170 overexpression experiments, the cells were transfected with the pCMV-N-Flag-CCDC170 or pCMV-N-Flag control vector using Lipofectamine TM 3000 Reagent (Invitrogen, USA). For the CCDC170 and IRE1 knockdown experiments, the cells were transfected with CCDC170 siRNA, IRE1 siRNA or the corresponding negative control (NC) siRNA (GenePharma, China) using Lipofectamine TM RNAiMAX transfection reagent (Invitrogen, USA) according to the manufacturer's instructions.

Lentiviral infection
CCDC170 was stably overexpressed using the recombinant lentiviral vector LV8(EF-1a/RFP+Puro) (Shanghai GeneChem Co., Ltd., China), and the empty vector was used as a negative control. MCF7 cells were plated on six-well dishes at 30-40% confluence and infected with the retroviruses.
Polybrene was added at a concentration of 5 μg/mL to enhance the infection efficiency. Seventy-two hours after the infection, 2 μg/mL puromycin was applied for 10 days to select the stable pooled cell population.

RNA extraction, amplification and labeling
Total RNA was extracted using an RNeasy Mini Kit (Qiagen, Valencia, CA, USA) in accordance with the AGING manufacturer's instructions. The overall RNA concentration and purity were measured on a NanoDrop 2000 Spectrophotometer. Total RNA was amplified and labeled using a Low Input Quick Amp Labeling Kit (Agilent Technologies, USA).

Microarray construction
The effects of CCDC170 on mRNA expression in MCF7 cells were analyzed using microarrays. After transcription, purification and fragmentation, samples were hybridized onto Agilent Whole Human Genome 4×44K 60 mer oligonucleotide arrays at 60° C for 16 hours. The arrays were then washed and scanned so that gene expression could be quantified. The fluorescence scanning image signals from the chip were converted to digital signals using Feature Extraction software v.10.7 (Agilent Technologies, USA).

DEG identification
Raw microarray data were normalized using the Quantile algorithm after background correction. Each gene expression value was calculated as the weighted average of all the forward or reverse probe sets. The ratio of the signal intensity in the experimental group (with 24-or 48-hour CCDC170 overexpression) to that in the control group was used to analyze the chip data for each gene (fold ratio = experimental group/control group). The DEGs between the CCDC170overexpressing and control MCF7 cells were selected according to the cut-off criteria of log|(fold ratio)| > 0.1761 and P < 0.05.

Pathway enrichment analysis
The Kyoto Encyclopedia of Genes and Genomes database was used for the pathway enrichment analysis of the DEGs, including Annotation, Visualization and Integrated Discovery. The data were analyzed using the "clusterProfiler" package in R, and P < 0.05 was considered statistically significant.

Immunohistochemistry
Formalin-fixed, paraffin-embedded tissue sections were stained at 4° C overnight using the following antibodies: 1:100 anti-CCDC170 (PA5-34723, Thermo Fisher Scientific, USA), 1:400 anti-XBP1 (PA5-27650, Thermo Fisher Scientific, USA) and 1:50 anti-IRE1α (H-190, Santa Cruz Biotechnology, USA). For antigen retrieval, the tissue sections were boiled in a sodium citrate solution (pH 6.0) in a pressure cooker for 2.5 min. For antibody visualization, an Ultra View DAB Detection kit (Dako, 20015510) was applied. Negative control samples were treated in the same manner without the addition of primary antibodies.

Western blotting
Total proteins were electrophoretically separated on sodium dodecyl sulfate polyacrylamide gels (10-12%) according to the molecular size of the target protein, and were subsequently transferred onto polyvinylidene difluoride membranes. After being blocked with 5% skim milk, the membranes were incubated at 4° C overnight with the following primary antibodies . Then, the membranes were washed thoroughly and incubated with secondary antibodies (1:3000 anti-mouse or 1:5000 anti-rabbit) at room temperature for 1 hour. The signals were visualized using the enhanced chemiluminescence method (Immobilon Western Chemiluminescent HRP Substrate, Millipore, USA). The samples were analyzedin duplicate, and the experiment was performed three times.

Immunofluorescence
Cells were grown on a microscope cover glass (PA 15275, Thermo Fisher Scientific, USA) laid on the bottom of the well of a 24-well plate (1 × 10 4 cells per well). The cells were fixed with 4% paraformaldehyde for 15 min, permeabilized with 0.2% Triton X-100 for 20 min and blocked with 10% goat serum for 30 min at room temperature. Then, the slides were incubated with anti-CCDC170 (1:200, PA5-34723, Thermo Fisher Scientific, USA) or anti-calnexin (1:500, MA3-027, Invitrogen, USA) primary antibodies overnight at 4° C.

Flow cytometric apoptosis detection
The percentage of apoptotic cells was ascertained using flow cytometry, based on the binding of Annexin V-PE and 7-ADD. Samples of 2x10 4 to 2x10 5 cells were prepared in 10% bovine serum albumin, according to the manufacturer's instructions. Then, 100 uL of Guava Nexin® Reagent (4500-0450, Millipore, USA) was added to each sample, and the mixture was incubated for 20 min at room temperature in the dark. Finally, the samples were analyzed on a Guava system. The samples were analyzed in duplicate, and the experiment was performed three times.

Terminal deoxynucleotidyl transferase dUTP nick end labeling (TUNEL) assay
Late-stage apoptosis was detected with a TUNEL BrightGreen Apoptosis Detection Kit (Vazyme Biotech, A112-03). Adherent MCF7 cells were cultured on Chamber Slides (PA 15275, Thermo Fisher Scientific, USA) in 24-well plates at a density of 1 × 10 4 cells per well. The cells were then fixed, washed, labeled and detected in accordance with the kit manual. After the TUNEL labeling, the nuclei were labeled with DAPI. Fluorescein isothiocyanate-12-dUTP-labeled DNA was observed directly under a fluorescence microscope. For the positive control slide, cells were permeabilized and treated with DNase I. For the negative control, no terminal deoxynucleotidyl transferase enzyme was added.

MTT assay
The MTT assay involves the conversion of the watersoluble yellow dye MTT to insoluble purple formazan by the action of mitochondrial reductase. MCF7 cells were reseeded in a sterile 96-well plate at a density of 3000 cells per well and grown for three days. Then, 50 µL of diluted MTT was added to each well and incubated with the cells for 4 hours in an incubator. The medium was removed, and 100 µL of dimethyl sulfoxide was added to each well to dissolve the formazan. Finally, the optical density of each well was measured on a spectrophotometer at a wavelength of 595 nm. The samples were analyzed in triplicate, and the experiment was repeated three times.

Statistical analyses
Statistical analyses were performed with SPSS 19.0 (SPSS Inc., Chicago, IL, USA). Categorical variables were compared using the Chi-square test, while continuous variables were analyzed using nonparametric tests (Kruskal-Wallis test and Mann-Whitney test), oneway analysis of variance (ANOVA) and Student's t-test. Spearman correlations were calculated because the data were not normally distributed, even after logtransformation.
The survival time was measured in months from the date of breast cancer diagnosis until recurrence, metastasis or death. OS was evaluated from the date of diagnosis to the date of death or last follow-up. DFS was calculated as the time from diagnosis to the observation of disease progression or death for any reason. Kaplan-Meier analyses and log-rank tests were performed to estimate the probability of OS and DFS. Cox logistic regression models with 95% CIs were used to evaluate the independent prognostic factors.
Data from cell experiments were analyzed using unpaired Student's t-tests, and images based on the statistical analyses were made in GraphPad Prism 5.0 (GraphPad Software Inc, La Jolla, CA, USA). All hypothetical tests were two-sided, and P-values less than 0.05 were considered statistically significant in all tests.

Ethics approval and consent to participate
Permission to use the paraffin-embedded tissues from this study for research purposes was provided by the Department of Breast Pathology, Tianjin Medical University Cancer Institute and Hospital. The ethics committee of Tianjin Medical University approved the study.

Availability of data and materials
The gene expression profiles used in the current study are available on the websites of TCGA (www.cbioportal.org/index.do) and GEO (www.ncbi. nlm.nih.gov/geo; accession number GSE96058).

AUTHOR CONTRIBUTIONS
Lina Zhang and Kexin Chen conceived and designed the study. Lina Zhang, Qiong Wang, Yanrui Zhao and Hongwei Han performed all the experiments and assisted in data interpretation. Qiong Wang and Qinghua Wang performed the bioinformatics data mining and assisted in data interpretation. Wei Wang and Ben Liu assisted in data interpretation. The expression of CCDC170 showed no significant effect on the overall survival and disease-free survival in non-Chemotherapy cohort. (C, D) The CCDC170 high-expression group exhibited better disease-free survival in Chemotherapy cohort, but no significance for overall survival. (E, F) The expression of CCDC170 showed no significant effect on the overall survival and disease-free survival in non-Radiotherapy cohort. (G, H) Higher expression of CCDC170 predicted better overall survival and disease-free survival in Radiotherapy cohort. (I, J) The expression of CCDC170 showed no significant effect on the overall survival and disease-free survival in non-endocrine therapy cohort. (K, L) The expression of CCDC170 showed no significant effect on the overall survival and disease-free survival in endocrine therapy cohort. P < 0.05(two-tailed) was considered as statistically significant.