Aberrant promoter hypermethylation of miR-335 and miR-145 is involved in breast cancer PD-L1 overexpression

PD-L1 is one of the most important immune checkpoint molecules in breast cancer that plays an important role in suppressing the immune system when confronted with tumor cells and is regulated by various microRNAs. Among them, microRNA-335-3p and microRNA-145-5p, regulated by DNA methylation, have tumor suppressor activities. We studied the role of miR-335 and -145 on PD-L1 suppression in breast cancer. The expression of miR-355 and miR-145 was significantly downregulated in BC tissues and cell lines compared to their controls, and their downregulation was negatively correlated with PD‐L1 overexpression. In-silico and luciferase reporter systems confirmed that miR-335 and -145 target PD-L1. In BC tissues and cell lines, cancer-specific methylation was found in CpG-rich areas upstream of miR-335 and-145, and up-regulation of PD-L1 expression was connected with hypermethylation (r = 0.4089, P = 0.0147, and r = 0.3373, P = 0.0475, respectively). The higher levels of miR-355 and -145 in BC cells induced apoptosis, arrested the cell cycle, and reduced proliferation significantly. In summary, we found that miR-335 and -145 are novel tumor suppressors inactivated in BC, and these miRs may serve as potential therapeutic targets for breast cancer treatment.


PD-L1 expression is upregulated in BC tissues and cell lines.
First, we analyzed PD-L1 mRNA expression in 50 breast cancer and noncancerous tissues. PD-L1 was upregulated (4.297 ± 0.4107) in breast cancer compared to noncancerous tissue (Fig. 1a). PD-L1 expression was evaluated based on the TCGA database to assess its prognostic significance in breast cancer. The database showed breast cancer tumors have higher expression than normal tissue. (P = 4.930500E−02) (Fig. 1b). PD-L1 expression was unrelated to patient age, estrogen receptor status, tumor size, etc. (P > 0.05). PD-L1 positive expression was higher in patients with lymph node metastases, and Ki-67 expression was 20% (P < 0.05) (Fig. 1c). Next, PD-L1 mRNA in human breast cancer was assessed using the Kaplan-Meier Plotter online database. An online KM plotter database using microarray data showed that PD-L1 mRNA expression was not significantly correlated with PFS (Progression-Free Survival) in 458 breast cancer patients (HR = 0.84, 95% CI = 0.59-1.19, P = 0.32) (Fig. 1d). In 1880 breast cancer patients, PD-L1 mRNA was associated with overall survival (HR = 0.68, 95% CI = 0.52-0.9, P = 0.0062) (Fig. 1e). We determined the expression of PD-L1 protein in BC tissues by immunohistochemistry. The presence of PD-L1 protein was high in breast cancer samples (Fig. 1f). In this study, we used ROC curves (AUC = 0.8, 96% sensitivity, and 44% specificity) to establish the predictive significance of these changed expression levels separately (Fig. 1g). According to Fig. 1h, we showed that there is a significant relationship between PD-L1 expression and TNBC or non-TNBC status of patients participating in this study. As shown in Fig. 1i, MDA-MB231 and BT549 have higher PD-L1 levels than normal breast tissue cells (MCF10a). PD-L1 expression may be diagnostic for BC based on these data.
Prognostic values of miR-335 and -145 in breast cancer and in silico exploration of their target genes. To evaluate the prognostic values of miR-335 and -145 in breast cancers, the expression of these miRs was assessed based on the breast cancer TCGA miRNA database. In the database, significantly lower expression was observed in breast cancer tumors (n = 749) compared to normal tissue (n = 76) (P = 2.098e−09, < 1e−12, respectively) (Fig. 3a). Furthermore, a breast-cancer survival analysis was performed based on the breast-cancer METABRIC miRNA database. The results showed that the low expression of miR-335 and -145 was significantly correlated with overall survival rate (BC, P = 0.00025 and P = 0.047, respectively), which demonstrates its bio-   www.nature.com/scientificreports/ logical importance (Fig. 3b). Based on this analysis, miR-335 and -145 and their target genes might potentially be considered biomarkers for breast cancer prognosis. In this study, miR-145 and 335 were significantly downregulated (0.736 ± 0.087, 0.606 ± 0.09 respectively) in breast cancer tissue compared with noncancerous tissue (Fig. 3c). We also determined the predictive value of these altered miR-335 and -145 expression levels using ROC curves (AUC = 0.9) (Fig. 3d). As compared with noncancerous patients with low PD-L1 levels, BC cases with high PD-L1 levels showed low miR-145 and -335 levels, with Pearson correlation analyses revealing a clear inverse relation between miR-145 and -335 expression and PD-L1 mRNA level (r = − 0.5431; P-value = 0.0013, r = − 0.6207; P-value = 0.0002 respectively) (Fig. 3e). We showed a significant correlation between miR-335 and -145 expression and the TNBC or non-TNBC status of study participants (Fig. 3f). As shown in Fig. 3g, miR-335 and-145 were more downregulated in breast cancer cell lines, with the highest expression level of PD-L1 compared with the normal breast tissue cell line (MCF10a).

Hypermethylation of miR-335 and miR-145 promoters might be responsible for upregulation of PD-L1.
Another question is why down-regulation of miR-335 and -145 occurs in the malignant transition of breast cancer cells. Tumor suppressor miRs, including miR-335 and -145, have been reported to be extinguished by abnormal DNA hypermethylation. Analysis of the miR-335 and -145 promoter showed three CpG islands upstream of the transcriptional start site, indicating that it may regulate the mature miR-335 and -145 and its precursor through DNA methylation. BC tissues showed hypermethylation in the miR-335 and -145 promoters, whereas the methylation levels of miR-335 and -145 were lower in noncancerous tissues (Table 1). We also determined the methylation rate in the miR-335 and -145 promoters in BC cells. MDA-MB231 cells had the highest methylation level, and MCF7 cells had the lowest level of methylation in the miR-335 and -145 promoters (Fig. 4a, b). Statistical analysis also showed that the level of methylation was strongly linked to the expression of miR-335, miR-145, and PD-L1 in both cancerous and healthy tissues (Fig. 4c). The surface of an activated T cell contains PD-1. Both immunological and cancerous cells express the protein PD-L1. By attaching to PD-1, PD-L1 prevents T-cell activation (Fig. 4d).

MiR-335 and miR-145 induce apoptosis in cell lines.
To further illustrate PD-L1 as a direct target of miR-335 and -145, we examine the significance of PD-L1 in miR-335 and -145-mediated cell survival. Using the Annexin V/PI assay, we also investigated the effect of these miRs on apoptosis induction. As shown in Fig. 6a, b, miR335 and -145 could increase the apoptosis rate in MDA-MB231 to 2.96% ± 0.38% and 3.19% ± 0.06%, respectively; in BT549 to 2.5% ± 0.2% and 2.62% ± 0.1%, respectively; in and MCF7 cell lines to 2.93% ± 0.4% and 1.9% ± 0.1%, respectively, when compared to the control groups. These results further suggest PD-L1 as a major target of miR-335 and -145.

Overexpression of miR-335 and -145 inhibits BC cell lines proliferation and cell cycle.
To understand the functional role of miR-335 and -145 on BC cell lines, we evaluated the impact of these miRs on the cellular proliferation of MDA-MB231, BT549, and MCF7 cell lines. As shown in Fig. 6c, 48 h after transfecting miR-335 and -145, the proliferation of these cells in comparison with non-transfected cells was decreased, in MDA-MB231 to 52.3% ± 2.4% and 33% ± 1.1%, respectively; in BT549 to 35.3% ± 0.9% and 23% ± 0.4%, respectively; and in MCF7 to 10.3% ± 0.3% and 14.4% ± 1.7% (mean ± SE), respectively, which was significantly different from non-transfected cells and transfected cells with scrambled miR. The cell cycle found by FCM also showed that overexpression of miR-335 and -145 stopped cells in the G0/G1 phase to stop the growth of cancer cells (Fig. 6d, e). In this study, we investigate miR-335 and -145 promoter hypermethylation in breast cancer. MiR-335 and -145 promoter hypermethylation is higher in breast cancer cell lines and malignant tissues. PD-L1 suppresses apoptosis and promotes breast cancer cell proliferation and the cycle. MiR-335 and -145 regulate PD-L1, which contributes to pro-tumor effects (Fig. 6f).

Discussion
Breast carcinogenesis is a process involving the dysregulation of tumor suppressor genes and oncogenes 33 . The PD-1/PD-L1 pathway is important as an immunosuppressant, and it is deregulated in a wide range of human cancers 34 , including BC. PD-L1 is involved in tumorigenesis and immunosuppression, making it a valuable therapeutic target for a variety of human cancers [35][36][37] . The correlation between decreased T-cell proliferation and increased apoptosis and tumor immune escape, with increased expression of PD-L1 proteins on cancer cells, creates an understanding of controlling PD-L1 expression using cancer cells as a model. In addition, significant advances in cancer therapy have been made in early clinical trials using Food and Drug Administration (FDA)approved antibodies that target PD-1/PDL1, such as pembrolizumab 38 . However, to increase the number of patients benefiting from the blockade of the safety checkpoint, these inhibitors are more commonly prescribed in combination with other therapies, including chemotherapy. Nevertheless, cumulative studies have shown that many common chemotherapy agents, including 5-Fluorouracil and Paclitaxel, lead to rearrangement of PD-L1. This phenomenon, in turn, reduces cell-mediated antitumor T reactions and promotes immune system escape 39 . Therefore, identifying the molecular mechanisms involved in regulating the PD-1/PD-L1 pathway through breast   www.nature.com/scientificreports/ using publicly available data and a variety of bioinformatics methodologies. Based on what we found, PD-L1 expression levels were strongly linked to shorter patient survival times and were much higher in BC tumor tissue than in normal tissue. High expression of PD-L1 was also significantly associated with lymph node metastases, and Ki-67 expression was 20%. Among the 50 cases of BC patients included in this study, immunohistochemically stained results showed that PD-L1 was positive in BC patients, which was brownish.
In the context of epigenetic regulation of the genome generally, promotor methylation is a distinct and reversible mechanism for gene silencing. Accordingly, promoter hypermethylation of tumor suppressor genes is found in many human cancers. MiRNAs are another epigenetic mechanism for controlling definitive gene expression because the translation of transcripts is expressly abrogated via mRNA degradation after miRNA binding. In this study, we found that BC is associated with low expression of miR-335 and -145. MiR-335-3p was found to be significantly downregulated as a tumor suppressor in cancers such as gastric cancer 40 , multiple myeloma 41 , human glioma 42 , etc. Furthermore, miR-145-5p downregulation has been found in a series of human cancers. Previous studies have highlighted its role as a tumor suppressor by modulating multiple oncogenes in cancer cells 43 , including breast cancer cells 44 . MiR-335 and -145 expression differs in benign and malignant tumors in women 44,45 . Furthermore, miR-335 and -145 were found to suppress the oncoproteins PD-L1, inhibiting the growth of breast cancer cells 46,47 . In the present study, using bioinformatics analysis, we identified the 3′-UTR PD-L1 region as a potential target for miR-335 and -145 targeting, suggesting its possible role in the escape of immune tumor cells. Subsequently, the analysis of clinical samples showed that the expression of miR-335 and -145 in tumor tissues was significantly lower than that in adjacent tissues, and its expression level was negatively correlated with PD-L1, suggesting that miR-335 and -145 may be involved in post-transcriptional regulation of PD-L1 expression. With this in mind, we used the luciferase assay, which confirmed that miR-335 and -145 could target specific regions of the PD-L1 3′-UTR and significantly coordinate its expression. In BC, several miRs undergo transcriptional inactivation by hypermethylation of their promoters. This can lead to the overactivity of their oncogenic targets 48 . CpG island hypermethylation-mediated silencing of miR-34b/c, miR-148, and miR-9-3 is correlated with the loss of oncogenic target gene regulation, such as C-MYC, E2F3, CDK6, and TGIF2, and promotes invasion and metastasis 49 . We detected the promoter hyper methylation status of miR-335 and -145 in BC tissues and cell lines through HRM. We discovered that BC tissues and cell lines, particularly metastatic cell lines, had abnormally high levels of methylation compared to normal tissues and cells; miRNA methylation levels were positively associated with PD-L1 expression in BC patients. We analyzed the correlation between miR-335 and -145 methylation status and PD-L1 expression in BC tissues. We showed that up-regulated expression of PD-L1 was significantly correlated with CpG island DNA hyper methylation in the promoter region of miR-335 and -145. In addition, DNA is less prone to degradation than RNA, frequency, stability, and variability among patients, which may indicate clinical usefulness. So, we think that, in addition to miRNA expression, abnormal methylation of the miR-335 and -145 promoters is another great epigenetic tumor marker. Unlike genetic modifications, the reversibility of epigenetic changes makes them a potential therapeutic target. The previous study results indicated that the expression of some silenced miRs in cancer www.nature.com/scientificreports/ cells could be restored by treatment with demethylating agents like 5-aza-2′-deoxycytidine, which inhibits the growth, invasion, and metastasis of cancer cells 50 . The US Food and Drug Administration (FDA) has approved 5-azacytidine (Azacitidine) 46 and 5-aza-2-deoxycytidine (Decitabine) 51 , which modify DNA methylation, for the treatment of patients with MDS. Previous reports of miR-335 and -145's involvement in human cancer have confirmed their role as a regulator of cancer cell growth and invasion 52,53 and the correlation between metastasis and human cancer. Exogenous overexpression of miR-335 and -145 in BC cells led to a big drop in PD-L1 expression at both the mRNA and protein levels. This showed that miR-335 and -145 change how the PD-L1 gene is expressed in BC.
This study has highlighted the efficacy of miR-335 and -145 in restraining tumor proliferation, arresting the cell cycle, and stimulating tumor apoptosis in BC cells. Cancer proliferation and progression are mainly associated with the ability of anti-tumor immunity to evade cancer cells 54 . Ghebeh et al. 55 have shown that the expression of PD-L1 in tumors could be linked to the development of breast cancer.
According to recent studies, PD-L1 plays a specific tumor-intrinsic role in promoting cancer metastasis, development, and treatment resistance 56 .Our study demonstrated that downregulated PD-L1 expression, through overexpression of candidate miRs, in BC cell lines significantly inhibited cell proliferation along with induction of G0/G1 phase arrest and apoptosis induction. The findings in other human tumors were in accordance with these results. In human AML cancer, the PD-L1 expression level was linked to a high number of cancer cells that were dividing 57 , and it has been shown that too much PD-L1 expression makes tumor cells grow 58 . Additionally, suppressing the expression of PD-L1 in gastric cancer cells could significantly increase apoptosis and reduce invasion, migration, and cell proliferation 59 . Moreover, in the present study, miR-335 and -145 were shown to induce apoptosis in BC cells by altering their significant regulators' expression.
These findings suggest that the methylation status and expression levels of miR-335 and -145, and PD-L1 overexpression as a consequence, may serve as markers of BC tumors. Further studies on the epigenetic regulation of miRNA expression are necessary, and the regulation of miRNA expression by epigenetic drugs might be of great promise for BC prevention and therapy.

Materials and methods
Clinical tissue samples and cell lines. 50 breast cancer patient samples were obtained from Khatamal-Anbya Hospital in Tehran, Iran. The hospital records included sex, age, lymph node metastases, and clinical grade (Fig. 1c). Radiotherapy or chemotherapy patients before surgery were excluded. Dual Luciferase assay. MiR-335, miR-145, and target gene 3′-UTR interaction was demonstrated using a dual-luciferase assay. The 3′-UTR of PD-L1 and a special scramble sequence (AAG CTT CAT AGG GCA TAG C) as a negative control were cloned into the psiCHECKTM-2 vector (Promega, C8021) to perform luciferase reporter assays (PCR primers are listed in Table 2 (Table 2). Referencing U6.

DNA Extraction and Bisulfite Modification.
Per the manufacturer's instructions, genomic DNA was isolated from BC tissues and adjacent non-tumor tissues using the Universal Genomic DNA Extraction Kit (Qiagen, Germany). The DNA from these tissues was tested for quality and integrity using electrophoresis on a 1% agarose gel and quantified spectrophotometrically. Genomic DNA (2 µg) was then subjected to bisulfite conversion using an EZ DNA Methylation Kit (Qiagen, Germany Methylation studies by high-resolution melting analysis. Bisulfite-modified DNA was amplified using primers that targeted methylated regions of miR-335 and -145 promoters ( Table 2). Diluting 100% methylated DNA with unmethylated bisulfite-modified DNA produced 25%, 50%, and 75% standard controls (Qiagen, Hilden, Germany). Each experiment had these standards. The reaction mixture contained 10 μl of Meltdoctor master mix, 10 ng of bisulfite-treated DNA, 200 nM of each primer, and 20 μl of PCR-grade water.
Step one PCR and high-resolution melting (HRM) were performed (ABI) 62 .
Step One: The program performed HRM-curve analysis, then compared fluorescence at the melting Cell proliferation analysis. Cell proliferation was assessed using the MTT kit (Sigma, St. Louis, MO, USA). Briefly, MDA-MB231, BT549, and MCF7 (3 × 10 4 cells/well in 96-well microplates) were transfected with pre-miR-335, miR-145, or negative control. Cells in the logarithmic growth phase were harvested and seeded on a 96-well plate. At 24, 48, and 72 h after seeding, 10 μl of MTT was added to each well, and the cells were incubated for 4 h. We supplemented each well with 150 μl of DMSO, and the optical density (OD) was recorded at 540 nm. Viability = 100 × (absorbance of the treated sample) / (absorbance of control). All experiments were repeated three times, and the average results were calculated. The experiments were repeated three times, and average results were calculated.

Cell cycle and apoptosis assay.
To determine the effects of miR-335 and -145 on apoptosis and cell cycle in BC cells, MDA-MB231, BT549, and MCF7 cells were seeded in 24-well plates with a density of 1 × 10 6 cells/mL in growth medium one day before transfection. The cells were transfected with scramble oligonucleotides or pre-miR-335 and pre-miR-145 by HiPerFect (Qiagen, Germany) according to the manufacturer's protocol. A cold 70% ethanol solution was applied to the cells for 24 hours at − 20 °C. The cells were stained with a propidium iodide (PI) solution after a washing step and left to dry for 30 min at room temperature. Flow cytometry was used to analyze the cell cycle (BD Biosciences, San Jose, CA, USA). According to the manufacturer's instructions, cells were stained twice with FITC-labeled Annexin V and PI using the Annexin V-FITC/PI Apoptosis Detection Kit (Solarbio, CA1020). We repeated this test three times, and the data it produced was analyzed by flow cytometry in FL1 and FL3 channels in the Partec Flow cytometer (Supplementary file).

Statistical analysis.
At least three real-time PCR, cell proliferation, cell cycle, and apoptosis tests were run.
The data is mean ± SEM. Each test's data was entered into GraphPad Prism V.8 for statistical analysis using oneway ANOVA. The t-test was used to compare cancerous and non-cancerous breast cells and samples. All realtime PCR results were analyzed using the CT technique with REST 2009 software (Qiagen, Hilden, Germany) and normalized against GAPDH for mRNA and U6 for miRNA. Pearson's correlation coefficient tested the two groups' correlation. P < 0.05 signifies statistical significance. Log-rank was used to compare Kaplan-Meier survival curves.
Ethics approval and consent to participate. The present study was conducted under the instructions accepted by the Ethics Committee of Pasteur Institute of Iran (Ethical code: IR.PII.REC.1398.023), written informed consent to participate, and consent to publish forms was obtained from all participants involved in the present study.
Informed consent. Written informed consent was obtained from all enrolled subjects.

Data availability
All data generated or analyzed during this study are included in this published article.