The tumor suppressor miR-138-5p targets PD-L1 in colorectal cancer

microRNAs (miRNAs) play critical roles in cancer development and progression. This study investigated the effects of miR-138-5p in human colorectal cancer (CRC) development. miR-138-5p was frequently downregulated in CRC tissues and was associated with advanced clinical stage, lymph node metastasis and poor overall survival. We found that miR-138-5p decreased expression of programmed cell death ligand 1 (PD-L1) through interaction with its PD-L1 3′ untranslated region. miR-138-5p also dramatically suppressed CRC cell growth in vitro and inhibited tumorigenesis in vivo. PD-L1 and miR-138-5p levels were inversely correlated in human CRC tumors, and miR-138-5p inhibited PD-L1 expression in tumor models. These results suggest that miR-138-5p is a tumor suppressor in CRC, and its effects are exerted at least partially through PD-L1 downregulation. Low miR-138-5p and high PD-L1 levels correlated with shorter overall CRC patient survival, indicating that miR-138-5p and PD-L1 may serve as CRC biomarkers for risk group assignment, optimal therapy selection and clinical outcome prediction. Targeting PD-L1, possibly by administering miR-138-5p mimics, might be a clinically effective anti-CRC therapeutic strategy.


Research Paper
correlates with poor prognosis in several carcinomas [32][33][34][35]. In this study, miR-138-5p was shown to be frequently downregulated in CRC tissues and might act as a tumor suppressor. We found that miR-138-5p markedly suppressed CRC cell growth in vitro and inhibited tumorigenesis in vivo by targeting PD-L1; miR-138-5p expression was inversely correlated with that of PD-L1 in CRC. Low miR-138-5p expression was associated with advanced clinical stage, lymph node metastasis and poor overall patient survival, and high PD-L1 expression correlated with decreased overall patient survival.
PD-L1 small interfering RNA (siRNA) transfected into NCM460 cells reduced PD-L1 protein levels, while the negative control (scramble) had no effect ( Figure 4A). NCM460 and CCD841CoN cells (with high endogenous miR-138-5p expression) transfected with miR-138-5p inhibitors showed increased proliferation (P<0.05), and this was rescued by PD-L1 siRNA transfection ( Figure 4B & 4D). In NCM460 and CCD841CoN cells transfected with miR-138-5p inhibitors, the number of cells in S phase of the cell cycle increased and the number in G1 phase decreased (P<0.05, Figure 4C & 4E); this was rescued by PD-L1 siRNA transfection (P<0.05). Collectively, these results suggest that miR-138-5p expression in CRC cells is important for both proliferation and cell cycle progression, and miR-138-5p acts by downregulating PD-L1.

Ectopic miR-138-5p expression suppresses CRC cell tumorigenicity in vivo
Two stable cell lines derived from SW620, SW620-miR-138-5p and SW620-scramble, were established using LV-miR138-5p and LV-scramble, respectively. SW620-miR-138-5p and SW620-scramble cells were inoculated into mice (four in each group). SW620-miR-138-5p cells exhibited significantly reduced tumor growth compared with SW620-scramble cells ( Figure 5A&5B). On day 12, average tumor size in SW620-miR-138-5p mice was reduced as compared to controls (P<0.05). Immunohistochemical (IHC) ( Figure 5C) and western blot www.impactjournals.com/oncotarget adjacent normal tissues as determined by qRT-PCR analysis A. U6 was the endogenous control. Relative miR-138-5p expression in CRC tissues B. miR-138-5p expression in two normal epithelial cell lines and four colon cancer cell lines as determined by qPCR (**P<0.01) C. U6 was the endogenous control. Data are presented as the mean ± SD from at least three separate experiments. Detection of miR-138-5p by ISH in CRC and corresponding adjacent normal tissues D. Tissue samples were cut into serial paraffin sections. "Scramble" and "NC, without probe" were used as controls. Row b shows magnified images of staining in adjacent normal tissues (red box in row a). Row c shows magnified images of staining in tumor tissues (green box in row a). Magnifications: ×100 (a) and ×200 (b and c). Kaplan-Meier analysis indicated that miR-138-5p downregulation was associated with poorer overall CRC patient survival (P=0.001) E. analyses ( Figure 5D) revealed that PD-L1 expression was lower in tumors derived from SW620-miR-138-5p mice as compared to controls. These studies demonstrated that miR-138-5p downregulated PD-L1 and inhibited CRC cell tumorigenicity in nude mice.

miR-138-5p and PD-L1 levels are negatively correlated in CRC
We assessed PD-L1 and miR-138-5p levels in CRC tissues using ISH, IHC and qRT-PCR. We observed an inverse correlation between PD-L1 and miR-138-5p levels in tumor tissues by all methods (Figure 6A-6C). Patients with high PD-L1 expression had an increased risk of death (P=0.0024, Figure 6D), indicating that PD-L1 expression could be a prognostic factor for CRC.

DISCUSSION
miRNAs, such as oncomiRs or anti-oncomiRs, play critical roles in the initiation and progression of human cancers through post-transcriptional regulation of gene expression [36]. Previous studies indicated that miR-138 might be a tumor suppressor in some cancers, including colorectal cancer, ovarian cancer, head and neck squamous cell carcinoma, nasopharyngeal carcinoma and pancreatic cancer [8,24,[37][38][39][40][41]. However, the functional role and mechanistic action of miR-138-5p in CRC remained largely unclear. Our study showed miR-138-5p was downregulated in CRC tumor samples as compared with corresponding adjacent normal tissues (Figure 1). We examined the effects of miR-138-5p on CRC cells in vitro and in vivo. Our results indicated that miR-138-5p reduced CRC cell proliferation and blocked the G1/S transition ( Figure 3). In addition, we demonstrated that miR-138-5p overexpression suppressed tumor growth in vivo ( Figure 5).
PD-L1 expression has been studied in different cancers including melanoma and cancers of the kidney, lung, pancreas, esophagus, ovary, colorectal, breast, head and neck, and was correlated with clinicopathological tumor features in several studies [42][43][44][45][46][47][48][49]. Recent clinical trials demonstrated that blocking the PD-1/PD-L1 pathway induces durable remission in patients with advanced solid tumors. Colorectal cancer appeared to respond poorly to PD-1 or PD-L1 antibody blockade in a clinical trial [50]. Several other factors may be involved in CRC patient survival time [51,52], and the role of PD-L1 in CRC cells must be further investigated.
miRNAs are known to perform their biological functions by downregulating expression of their target genes. The predicted miR-138-5p binding site was present in PD-L1 3′ UTRs. Whether PD-L1 is the only direct target of miR-138-5p is still unknown. Luciferase activity assay results verified that PD-L1 was a target of miR-138-5p. PD-L1 levels were decreased by ectopic miR-138-5p expression in CRC cells (Figure 2Da). The present study showed that PD-L1 overexpression in CRC cells can rescue them from miR-138-5p-induced cell cycle arrest and reduced proliferation (Figure 3). The tumor suppressive role of miR-138-5p in CRC is thus at least partly realized by downregulating PD-L1. A previous study demonstrated that PD-L1 knockdown by siRNA reduced cell proliferation in vitro similarly to miR-138 restoration [33]. We demonstrated a direct link between miR-138-5p and PD-L1 expression in CRC patients, and observed that PD-L1 and miR-138-5p levels were inversely correlated in human CRC specimens ( Figure  6C). We also showed that low miR-138-5p expression correlated with advanced clinical stage and lymph node      metastasis. Importantly, low miR-138-5p ( Figure 1D) and high PD-L1 levels ( Figure 5D) were correlated with shorter overall CRC patient survival, indicating that miR-138-5p and PD-L1 may serve as CRC biomarkers for risk group assignment, optimal therapy selection and clinical outcome prediction. Targeting PD-L1, possibly by administering miR-138-5p mimics, might be a clinically effective anti-CRC therapeutic strategy.

Cell lines, plasmids and miRNAs
The human CRC cell lines HCT116, SW620, SW480 and HT29 and human normal colonic epithelium cell lines NCM460 and CCD841CoN were obtained from the Chinese Academy Medical Science (Beijing, China). CRC cell lines were cultured in RPMI-1640 medium containing 10% fetal bovine serum (FBS). NCM460 and CCD841CoN cell lines were cultured in DMEM medium containing 15% FBS. Cells were cultured at 37°C with 5% CO 2 . miRNA mimics, inhibitors and negative control sequences were synthesized and purified by Shanghai Gene Pharma Co (Shanghai, China). The PD-L1 expression plasmid was obtained from OriGene (Beijing, China).

Tumor samples
CRC patient tumor samples were collected from the Xiangya First, Second, and Third Hospitals of Central South University. Written informed consent was obtained from all study participants. Tissue sample collection and use protocols were approved by the ethical review committees of the Xiangya Hospital Ethic Committee of Central South University.

Quantitative real-time PCR
Total RNA was isolated from CRC surgical specimens and reverse transcribed as previously described [53]. Real-time quantitative PCR was performed using an IQ5 Multicolor Detection System (Bio-Rad) with a Hairpin-it miRNAs qPCR Quantification Kit (GenePharma, Shanghai, China) according to the manufacturer's instructions. The primers used for miRNA detection by qRT-PCR were designed based on sequences provided by the Sanger Center miRNA Registry. The U6 snRNA was used as an endogenous control. The following program was used for the RT reaction: 16°C for 30 min, 42°C for 30 min and 85°C for 10 min. The following program was used for qPCR: 95°C for 3 min, 40 cycles of 95°C for 12 s and then 62°C for 35 s. The primers used were PD-L1 (forward, 5′-CATCTTATTATGCCTTGGTGTAGCA-3′; reverse, 5′-GGATTACGTCTCCTCCAAATGTG-3′) and β-actin (forward, 5′-GCATCCCCCAAAGTTCACAA-3′; reverse, 5′-AGGACTGGGCCATTCTCCTT-3′). The following program was used for PCR: 94°C for 5 min, 35 cycles of 94°C for 30 s, 58°C for 40 s and 72°C for 30 s, and then 72°C for 10 min. Relative expression changes were calculated using the 2 -ΔΔCT (where CT is threshold cycle) method. Three parallel repeats were performed for each sample in each experiment and results were expressed as the mean of three independent experiments.

Cell proliferation assay
CRC cells were transfected with miR-138-5p mimics or inhibitors for 24 h, and proliferation was examined. Briefly, 2000 cells from each group (untreated, miR-138-5p mimics, miR-138p inhibitors and negative control) were plated in each well of five 96-well plates in 200μL of medium. Cells were treated with 20μL of 5mg/ ml MTT (tetrazoliumsalt 3-[4,5-dimethylthiazole-2-yl]-2,5-diphenyltetrazolium bromide) in complete medium for 4 h, and were then dissolved in dimethylsulfoxide (DMSO). Absorbance at 490 nm was measured using an enzyme-linked immunosorbent assay (ELISA) plate reader (Sunrise remote, Tecan Austria GmbH). One plate was analyzed immediately after cells adhered (approximately 4 h after plating). One plate per day was examined for the next four days. Each experiment was repeated three times independently.

Flow cytometry analysis
Cells were washed twice with phosphate-buffered saline (PBS) and fixed in 70% ethanol overnight. Then, cells were centrifuged at 1,200×g for 5 min, resuspended in 50μg/ml propidium iodide (Sigma-Aldrich) in PBS and immediately analyzed by flow cytometry using a MOFLO XDP flow cytometer (Beckman Coulter, Fullerton, CA). The appropriate forward and side scatter gates were used, and 1×10 5 cells were examined per experiment. Data were analyzed with Modfit software (Verity Software House, Topsham, ME) and values were expressed as the mean and standard deviation of three independent experiments.

In situ hybridization and immunohistochemistry
ISH was used to detect miR-138-5p expression in CRC tissue samples. Oligonucleotide probes complementary to hsa-miR-138-5p (where hsa indicates Homo sapiens) (product 38511-01; Exiqon) and scramble (normal control) were digoxigenin labeled at the 5′ terminal end. Tumor and corresponding adjacent normal tissues were formalin fixed, paraffin embedded and sectioned. Probes were detected using an enhanced sensitive ISH detection kit (Boster Inc., Wuhan, China) according to the manufacturer's instructions. Negative controls did not include probes. MiR-138-5p staining was scored according to intensity and proportion: 0-1 (0-25%), 1-2 (26-50%), 2-3 (51-75%), and 3-4 (76-100%). The sum of intensity and percentage counts was used as the final score. Expression scores greater than or equal to 2 were defined as high, whereas scores lower than 2 were indicative of low expression. The detailed method was applied as previously described [55].
To examine PD-L1 expression in tissues via IHC, sectioned specimens were deparaffinized and dehydrated. Samples that were not exposed to the primary antibody were used as negative controls. The scoring system used in grading PD-L1 expression was described previously [56]. Tumors with strong or moderate immunostaining intensity were classified as having positive (+) expression, whereas tumors with absent or weak immunostaining were classified as having negative (-) expression. miR-138-5p and PD-L1 staining were scored independently by two pathologists blinded to patient clinical characteristics.

Experiments in nude mice
Four-week-old specific-pathogen-free (SPF) BALB/c nude mice were purchased from SLAC Laboratory Animal Co., Ltd. (Shanghai, China). All animal procedures were performed according to approved protocols and in accordance with recommendations for the proper use and care of laboratory animals. A xenograft model was used to examine the effects of miR-138-5p on tumor growth in CRC cells in vivo. SW-620 cells were infected with LV-miR138-5p or LV-scramble, and six million cells in 0.2 ml of growth medium were subcutaneously injected into BALB/c nude mice. The first tumor was detected one week after inoculation. Mice were killed 9 d after the initial tumor detection, and tumors were harvested, weighed and snap frozen.

Statistical analysis
The differences between groups were tested examined using the Student's t test or a one-way analysis of variance (ANOVA) test. Survival analysis was performed using Kaplan-Meier plots and log-rank tests. Correlations between miR-138-5p expression and clinicopathological characteristics were analyzed by Pearson's χ 2 test and Fisher's exact test. Univariate and multivariate Cox proportional hazard regression models were used to assess survival hazard. Statistical analyses were performed using SPSS v16.0. Differences were considered significant when P<0.05.

CONFLICTS OF INTEREST
The authors declare no conflicts of interest.

GRANT SUPPORT
This work was supported by the National Natural Science Youth Foundation of China (81402272), the Hunan Provincial Natural Science Foundation of China www.impactjournals.com/oncotarget (11JJ4071) and the Freedom Explore Program of Central South University (2011QNZT192).