CircRNA circ_0075796 is downregulated in breast cancer and suppresses cell proliferation, migration and invasion

Emerging evidence shows that circular RNAs (circRNAs) play crucial parts in tumorigenesis and progression. In this work, the expression, clinical signicance, function and potential mechanism of circ_0075796 in breast cancer were explored. The expression of circ_0075796 in 189 pairs of breast cancer tissues and adjacent normal tissues was detected by quantitative real-time PCR (qRT-PCR). Cell Counting Kit-8 (CCK-8) assay, methyl thiazolyl tetrazolium (MTT) assay and colony formation assay were conducted for cell proliferation. Transwell assay and wound healing assay were used for cell migration and invasion. Flow cytometry analysis was adopted for cell cycle and cell apoptosis. The cellular localization of circ_0075796 was determined by uorescence in situ hybridization (FISH). The circ_0075796/miR-452-3p/SAMD5 axis was screened out by bioinformatics analysis and veried by qRT-PCR. Methylated RNA Immunoprecipitation (MeRIP) was used to detect the N6-methyladenosine (m6A) modication levels of circ_0075796. QRT-PCR was used to detect the expression of RNA binding protein Quaking (QKI) in breast cancer tissues and adjacent normal tissues.

adopted for cell cycle and cell apoptosis. The cellular localization of circ_0075796 was determined by uorescence in situ hybridization (FISH). The circ_0075796/miR-452-3p/SAMD5 axis was screened out by bioinformatics analysis and veri ed by qRT-PCR. Methylated RNA Immunoprecipitation (MeRIP) was used to detect the N6-methyladenosine (m6A) modi cation levels of circ_0075796. QRT-PCR was used to detect the expression of RNA binding protein Quaking (QKI) in breast cancer tissues and adjacent normal tissues.
Results circ_0075796 was downregulated in breast cancer tissues compared with adjacent normal tissues. In addition, circ_0075796 showed satisfactory diagnostic value to discriminate breast cancer and normal controls. Downregulated circ_0075796 expression was correlated with lymph node metastasis, HER2 expression, larger tumor size, high Ki-67 expression, advanced histological grade, aggressive molecular subtypes and advanced clinical stages. Overexpression of circ_0075796 inhibited cell proliferation, migration and invasion in vitro. FISH showed that circ_0075796 was localized in the cytoplasm and nucleus of breast cancer cells. Bioinformatics analysis and qRT-PCR revealed the potential circ_0075796/miR-452-3p/SAMD5 axis. Moreover, circ_0075796 showed lower m6A modi cation levels in breast cancer tissues compared to adjacent normal tissues. QKI was predicted to contain binding sites of circ_0075796 and was downregulated in breast cancer tissues compared to adjacent normal controls.
Conclusions circ_0075796 was downregulated in breast cancer compared to normal controls, and showed potential diagnostic value for breast cancer. Downregulation of circ_0075796 was correlated with aggressive clinical features of breast cancer and overexpression of circ_0075796 inhibited the progression of breast cancer in vitro, indicating that circ_0075796 may be related to tumorigenesis and development of breast cancer.

Background
Breast cancer is a serious threat to the health of women, which has become one of the most prevalent and invasive malignancies for female worldwide [1,2]. According to public data, it is the second leading cause of cancer-related deaths in women [3]. Although there are various treatments for breast cancer (e.g. surgery, radiotherapy and chemotherapy), the incidence and prognosis remain pessimistic [4]. In order to improve the curative effect and prognosis of breast cancer, it is essential to discover new biomarkers and therapeutic targets. Therefore, detailed researches into the mechanisms of breast cancer are urgently needed.
Circular RNAs (circRNAs) is a novel class of endogenous non-coding RNAs (ncRNAs). In contrast to linear RNA, circRNAs are formed by covalently closed loop and have neither 5' to 3' end polarity nor polyadenylation tail [5,6]. Studies have con rmed that circRNAs can play key roles in the progression of various diseases. For example, circular RNA 406961 can regulate the in ammatory response of human bronchial epithelial cells by activating the STAT3/JNK pathway [7]. Circular RNA-ZNF532 can regulate retinal pericyte degeneration and vascular dysfunction caused by diabetes [8]. Moreover, an increasing number of studies have shown that circRNAs are involved in tumorigenesis and progression, including breast cancer [9][10][11]. CircRNAs participate in a number of different biological processes in tumor cells, including cell growth, metastasis, cell cycle control, nuclear and cytoplasmic transport, cell differentiation, RNA decay, transcription, translation and so on [12]. Concretely, microRNA(miRNA) sponges are the most common roles of circRNAs in many different types of cancers [10,13]. Many RNA transcripts share binding sites with miRNAs and compete with each other to act as competitive endogenous RNAs (ceRNAs) to further regulate the occurrence and progress of tumor [14]. Numerous studies have shown that circRNAs can play the role of ceRNAs by acting as miRNA sponges in various tumors [15]. For instance, Tang et al. revealed that circGFRA1 and GFRA1 played the role of ceRNAs in triple negative breast cancer by regulating miR-34a [16].
N6-methyladenosine (m6A) is one of the most abundant modi cation of eukaryotic mRNAs and ncRNAs, which can control many aspects of post-transcriptional regulation of mRNA, including splicing, export, stability, and translation [17,18]. In recent years, m6A modi cation of circRNAs has been brought into the spotlight [19]. For example, Timoteo et al. have showed that the decrease in methylation could affect the conversion of the pre-mRNA into the circRNA and further in uence circRNA levels [20]. Zhao et al.
revealed that m6A modi cation was essential for the protein-coding ability of circE7 [21].
CircRNA circ_0075796 is located on chromosome chr6:18236683-18258636-, and its predicted length is 902nt. At present, there is no research report on circ_0075796 in human diseases. In this study, the expression, clinical signi cance, function in vitro and potential mechanism of circ_0075796 were explored.

Materials And Methods
Tissue specimens

Plasmid construction and cell transfection
For the overexpression of circ_0075796, 902bp cDNA fragment was cloned into pLC5-ciR vector, which was then veri ed by PCR product sequencing analysis ( Supplementary Fig. S1A). MDA-MB-231 cells and BT474 were seeded in 6-well culture plates and cultured to 70-90% con uence before transfection, respectively. According to the manufacturer's instructions, circ_0075796 overexpression plasmid and negative control (Geneseed Biotech Co., Ltd., Guangzhou, China) were transiently transfected using

Colony formation assay
The transfected MDA-MB-231 and BT474 cells (700 cells/well) were plated into 6-well plates. The plate was placed in a 37°C incubator containing 5% CO 2 to induce colony growth. After cultivation for 10 days, cell surface was rinsed using cooled PBS. Cell colonies were xed with 4% paraformaldehyde, stained with crystal violet and nally observed under a microscope.
Transwell assay Fluorescence in situ hybridization (FISH) circ_0075796 speci c probe was used for in situ hybridization. Nuclei were counter by staining with 4,6diamidino-2-phenylindole (DAPI). All the procedures were conducted according to the manufactory's instruction (Geneseed, Guangzhou, China).
The nal specimen was analyzed on a laser confocal microscope, TCS SP2 AOBS (Leica, Wetzlar, Germany). The probe was labeled with digoxin and the sequence is 5'-ggatacattccttgcaaatc-3'.

Bioinformatics analysis
Starbase [30] was applied to predict the potential binding sites between QKI and circ_0075796. Miranda [31] and TargetScan [32] databases were used to predict the relationship between 836 disease-related miRNAs (from the Human microRNA Disease Database, version 2.0) and circ_0075796. The target genes of the candidate miRNAs and the potential binding sites between them were further predicted based on these two databases.

Methylated RNA immunoprecipitation (MeRIP)
In brief, total RNA was extracted from three pairs of breast cancer tissues and adjacent normal tissues using TRIZOL reagent (Invitrogen, Carlsbad, CA, USA) and then fragmented using 2µL fragmentation buffer later. The fragmented RNA was then incubated with anti-m6A onoclonal antibody and A/G magnetic beads in IP buffer at 4°C for 2 hours for immunoprecipitation. Next, the bound RNA was eluted from the beads in IP buffer. SuperScript TM III Reverse Transcriptase (Invitrogen, Carlsbad, CA, USA) was applicated for reverse transcription of the eluted RNA after puri cation. QRT-PCR was conducted using qPCR SYBR Green Master Mix (CloudSeq, Shanghai, China) in QuantStudio 5 real-time PCR System (Thermo Fisher, Waltham, MA, USA). The Ct difference between input and the immunoprecipitated RNA was identi ed, and the relative enrichment was calculated using the 2 −△△Ct method.

Statistical analysis
All statistical analyses were performed using SPSS 20.0 (SPSS, Chicago, IL, USA) and GraphPad Prism 5.0 (SPSS, Chicago, IL, USA). Data differences between two groups were analyzed using the Student's t test. One way analysis of variance (ANOVA) was used to analyze data differences between three or four groups. Receiver operating characteristic (ROC) curves were generated to evaluate the diagnostic value of circ_0075796 for breast cancer. The correlations between circ_0075796 expression level and the clinicopathological parameters of breast cancer patients were analyzed using the chi-squared test. P < 0.05 was considered to be statistically signi cant.

Results
circ_0075796 was signi cantly downregulated in breast cancer tissues compared with adjacent normal tissues To examine the expression levels of circ_0075796, qRT-PCR was performed in 189 pairs of breast cancer tissues and adjacent normal tissues. Results showed that circ_0075796 was prominently downregulated in breast cancer tissues compared to adjacent normal tissues (Fig. 1A, P = 0.0028). Additionally, the area under the receiver operating characteristic curve (AUC) of circ_0075796 was 0.8015 ( Figure 1B, P < 0.0001), showed that circ_0075796 could well distinguish breast cancer from adjacent normal tissues.

Association of circ_0075796 expression with clinicopathological features
The relationship between the expression level of circ_0075796 and clinicopathological features of breast cancer patients was analyzed. The data implied that circ_0075796 was downregulated in patients with lymph node metastasis than those without lymph node metastasis ( Fig. 1C, P = 0.0174). In HER2 positive breast cancer, circ_0075796 showed lower expression level than in HER2 negative breast cancer (Fig. 1D, P = 0.0182). The expression level of circ_0075796 were negatively correlated with HER2 expression (Table 1, P = 0.028). As for tumor size, results showed that circ_0075796 was downregulated in patients with tumor size 2cm than patients with tumor size ≤ 2cm (Fig. 1E, P = 0.0251). The expression level of circ_0075796 were negatively correlated with tumor size (Fig. 1F, Spearman correlation r = -0.2253, P = 0.0018; Table 1, P = 0.029). In addition, patients with higher expression of circ_0075796 had lower Ki-67 expression (Fig. 1G, P = 0.0320). And the expression level of circ_0075796 were negatively correlated with Ki-67 expression (Fig. 1H, Spearman correlation r = -0.1405, P = 0.0279). Meanwhile, the expression level of circ_0075796 showed progressive decrease with the increase of pathological N (pN) stages (Fig. 1I, P = 0.0368). In particular, patients with stage pN0 had higher expression level of circ_0075796 compared to patients with stage pN1 and patients with stage pN2-3, respectively (Fig. 1I, pN0 vs pN1, P = 0.0338; pN0 vs pN2-3, P = 0.0489). As for histological grade, decrease of circ_0075796 expression was found from Grade , Grade to Grade (Fig. 1J, P = 0.0118). Speci cally, circ_0075796 showed higher expression in patients with Grade than patients with Grade II and Grade III, separately (Fig. 1J, Grade I vs Grade II, P = 0.0355; Grade I vs Grade III, P = 0.0038). The negative correlation between circ_0075796 expression and histological grade was also discovered ( Table 1, P = 0.006). Moreover, circ_0075796 represented differential expression levels in different molecular subtypes of breast cancer. circ_0075796 showed higher expression level in Luminal A breast cancer compared with HER2 enriched breast cancer (Fig. 1K, Luminal A breast cancer vs HER2 enriched breast cancer, P = 0.0252). As for clinical stages of breast cancer, circ_0075796 was downregulated in patients with advanced clinical stages than patients with early clinical stages (Fig. 1L, P = 0.0480). The expression level of circ_0075796 was not correlated with estrogen receptor (ER) and progesterone receptor (PR) ( Table 1).

circ_0075796 promoted the progression of breast cancer in vitro
To further investigate the role of circ_0075796 in the tumorigenesis and development of breast cancer, cell function experiments were performed. Since the expression of circ_0075796 was down-regulated in breast cancer tissues, the overexpression vector of circ_0075796 was constructed to transfect the breast cancer cell line MDA-MB-231 and BT474. After transfection, the expression of circ_0075796 was signi cantly up-regulated compared to the controls ( Fig. 2A; Supplementary Fig. S1B). In cellular experiments, MTT (Fig. 2B), CCK-8 ( Supplementary Fig. S1C) and colony formation assays ( Supplementary Fig. S1D) indicated that overexpression of circ_0075796 could signi cantly promote cell proliferation. Transwell (Fig. 2C-D) and wound healing assays ( Supplementary Fig. S1E-F) showed that the migration and invasion abilities of breast cancer cell were facilitated by circ_0075796. Flow cytometry analysis was conducted to detect the in uence of circ_0075796 on cell cycle and apoptosis. According to the consequence, there was no signi cant difference in cell cycle distribution (Fig. 2E-F; Supplementary  Fig. S1G) and apoptosis rates (Fig. 2G-H; Supplementary Fig. S1H) between circ_0075796 overexpression group and the control group. These results revealed that circ_0075796 could promote the progression of breast cancer in vitro.
Bioinformatics analysis and preliminary veri cation of circ_0075796/miR-452-3p/SAMD5 axis Numerous studies have reported that circRNAs can serve as miRNA sponges in cancers and further regulate gene expression [33]. In this work, FISH assays con rmed that circ_0075796 was localized both in the cytoplasm and nucleus of breast cancer cells (Fig. 3A), suggesting its potential of miRNA sponge in breast cancer. To further study the miRNAs related to circ_0075796, Miranda and Targetscan databases were used to predict candidate miRNAs. There were 90 miRNAs with potential binding sites for circ_0075796 (Supplementary Table S2). According to the combination of the number of binding sites, the context score from TargetScan and the thermodynamic properties of the binding site from miRanda, miR-452-3p was predicted to have the highest degree of association with circ_0075796. In addition, the possible target genes of candidate miRNAs were predicted on the basis of the two databases. Among the predicted circ_0075796/miRNA/mRNA interactions, circ_0075796/miR-452-3p/SAMD5 axis is of great interest since both miR-452-3p and SAMD5 have been reported to be related to cancer development and progression [34,35].
circ_0075796 showed different m6A modi cation level in breast cancer tissues and normal controls M6A modi cation of circ_0075796 was predicted by SRAMP website (http://www.cuilab.cn/sramp/). Results showed that there were multiple m6A modi cation sites in circ_0075796 (Fig. 4A). Furthermore, methylated RNA immunoprecipitation (MeRIP) showed that m6A modi cation of circ_0075796 existed in both breast cancer tissues and adjacent normal tissues (Fig. 4B). Compared with normal controls, breast cancer tissues showed lower m6A modi cation levels (Fig. 4B). The mRNA levels of m6A-related enzymes in breast cancer tissues and adjacent normal tissues were also monitored. As expected, YTHDF2 and IGF2BP2, known as m6A readers [36], were downregulated in breast cancer tissues contrast to normal controls ( Fig. 4C-D, P < 0.05).

QKI was downregulated in breast cancer tissues contrast to normal controls
Given that QKI can be involved in the formation of circRNAs and tumor progression, Starbase was applied to further explore the interaction of QKI and circ_0075796. The predictions showed that there were several potential binding sites between QKI and circ_0075796 (Fig. 5A-B). The expression levels of QKI were detected by qRT-PCR in 11 pairs of breast cancer tissues and adjacent normal tissues. Results indicated that QKI was signi cantly downregulated in breast cancer tissues (Fig. 5C, P = 0.0437), consistent with the expression trend of circ_0075796.

Discussion
Increasing studies have revealed the potential of circRNAs in biomarkers and molecular targets for varieties of cancers. For example, Melika et al. revealed the potential of hsa_circ_0005046 and hsa_circ_0001791 as diagnostic biomarkers for breast cancer [37]. Fu et al. provided a circRNA pro le as potential biomarkers and therapeutic targets for breast cancer brain metastasis [38]. In this study, qRT-PCR was used to analyze circ_0075796 expression in 189 pairs of breast cancer and adjacent normal tissues. Results showed that circ_0075796 was signi cantly downregulated in breast cancer tissues, which suggests the potential tumor suppressor role of circ_0075796. In addition, the ROC curve of circ_0075796 demonstrates that circ_0075796 has a certain signi cance in the diagnosis of breast cancer. Furthermore, the analysis of the clinicopathological features indicated that circ_0075796 showed down-expression in patients with larger tumor size, positive HER2 expression, higher Ki-67 expression and advanced histological grade, emphasizing the involvement of circ_0075796 in the tumorigenesis and development of breast cancer. circ_0075796 expression level was also relevant to lymph node involvement and different molecular subtypes, which implies that circ_0075796 may be associated with metastasis and prognosis of breast cancer.
In addition, circRNAs are gaining increasing attention in special regulatory functions in various biological processes of multiple cancers including breast cancer. For example, circ_0069718 was demonstrated to promote the progression of breast cancer by up-regulating NFIB through sequestering miR-590-5p [39]. Shi et al. revealed that hsa_circ_0006220 remarkably inhibited the proliferation, migration, and invasion of TNBC cells, playing an inhibitory role in TNBC progression [40]. Therefore, in-depth researches of the functions and regulatory mechanisms of circRNAs could provide novel insights into the molecular mechanisms of breast cancer progression. In this study, we recognized that circ_0075796 overexpression statistically inhibited cell proliferation, migration, and invasion, which was consist with the expression of circ_0075796 in breast cancer tissues. These results convincingly suggested the tumor suppressor role of circ_0075796 in breast cancer.
Researches have indicated that circRNAs can serve as ceRNAs to accommodate the expression of miRNAs and further in uence the expression of mRNAs to affect tumor progression [10,41,42].To further explore the mechanism of circ_0075796 as ceRNA in breast cancer, we obtained the potential circ_0075796/miR-452-3p/SAMD5 axis by bioinformatics analysis. QRT-PCR analysis preliminarily veri ed that circ_0075796 may serve as ceRNA to eliminate the inhibitory effect of miR-452-3p on SAMD5. Tang et al. have disclosed that miR-452-3p can promote the proliferation and migration of liver cancer cells by directly targeting the CPEB3/EGFR axis [34]. And Li et al. indicated that miR-452-mediated miR-452-GSK3β-LEF1/TCF4 loop could induce colorectal cancer proliferation and migration [43]. In this study, the proposal of potential circ_0075796/miR-452-3p/SAMD5 axis and the veri cation of the upregulation of miR-452-3p in breast cancer tissues revealed the role of miR-452-3p in breast cancer.
SAMD5 is a protein containing SAM domain, which is distributed on about 70 residues and has different effects on cellular processes through polymerization [44]. It has been reported that knockdown of SAMD5 can inhibit cell proliferation in small cell lung cancer [45]. Inversely, Tomoki Yagai et al. demonstrated that knockdown and overexpression of SAMD5 resulted in promotion and inhibition of cell proliferation in cholangiocarcinoma respectively [35]. In this work, the potential circ_0075796/miR-452-3p/SAMD5 axis and the correlation of the expression levels of circ_0075796, miR-452-3p and SAMD5 indicated the role of circ_0075796 in breast cancer.
According to the prediction of m6A modi cation sites of circ_0075796 based on the SRAMP website, we Park et al. found the role of m6A modi cation in circRNA degradation [47]. Additionally, extensive translation of circRNAs driven by m6A modi cation also has been reported [48]. These ndings revealed the close correlation between m6A modi cation and biological behaviors of circRNAs. In the present study, the decreased m6A modi cation degree of circ_0075796 may be concerned with the differential expression of circ_0075796 in breast cancer tissues and adjacent normal tissues.
Over the years, more studies have documented that QKI can regulate the formation of circRNAs and thus affects the progression of diseases [49]. For example, circNRIP1 was upregulated on account of the promotion of QKI in gastric cancer tissues [50]. In prostate cancer, QKI can increase circZEB1 levels [51].
In non-small cell lung cancer, QKI promoted the information of circ-SLC7A6 and facilitated the suppression of tumor progression [52]. In addition, Cao et al. found that QKI could suppress breast cancer via RASA1/MAPK signaling pathway [53]. Our study predicted the potential binding sites of QKI and circ_0075796 and revealed the downregulation of QKI in breast cancer tissue. Decreased expression levels of QKI may be involved in the downregulation of circ_0075796 and then affect the progression of breast cancer, which requires continually further researches.

Conclusions
In summary, this study explored the expression, roles, functions and potential mechanisms of circ_0075796 in breast cancer. The results disclosed the downregulation of circ_0075796 in breast cancer tissues and its diagnostic value for breast cancer. Besides, the close association of circ_0075796 expression and clinicopathological features revealed the possible role of circ_0075796 in the occurrence, progression, metastasis and prognosis of breast cancer. The cell experiments uncovered the potential functions of circ_0075796 in the progression of breast cancer. In addition, the potential circ_0075796/ miR-452-3p/SAMD5 axis was revealed. The speci c mechanisms of the ceRNA axis and the roles of circ_0075796 in the occurrence and development of breast cancer remain to be further studied.   circ_0075796 inhibited cell migration and invasion abilities by Transwell assays. (E-F) Flow cytometry analysis showed that circ_0075796 had no effect on cell cycle. (G-H) Flow cytometry analysis showed that circ_0075796 had no effect on cell apoptosis rates. *P < 0.05, ***P < 0.001. normal tissues (n = 11) demonstrated that miR-452-3p was upregulated in breast cancer tissues, while SAMD5 was downregulated (All P < 0.0001). (F-G) circ_0075796 was negatively correlated with miR-452-3p (Spearman correlation r = -0.4297, P = 0.0459) and positively correlated with SAMD5 (Spearman correlation r = 0.3698, P = 0.0451). (H) miR-452-3p was negatively correlated with SAMD5 (Spearman correlation r = -0.7414, P < 0.0001).

Supplementary Files
This is a list of supplementary les associated with this preprint. Click to download.