MicroRNA-326 attenuates immune escape and prevents metastasis in lung adenocarcinoma by targeting PD-L1 and B7-H3

Tumor-infiltrating T cells are highly expressive of inhibitory receptor/immune checkpoint molecules that bind to ligand expressed by tumor cells and antigen-presenting cells, and eventually lead to T cell dysfunction. It is a hot topic to restore T cell function by targeting immune checkpoint. In recent years, immunotherapy of blocking immune checkpoint and its receptor, such as PD-L1/PD-1 targeted therapy, has made effective progress, which brings hope for patients with advanced malignant tumor. However, only a few patients benefit from directly targeting these checkpoints or their receptors by small compounds or antibodies. Since the complexity of the regulation of immune checkpoints in tumor cells, further research is needed to identify the novel endogenous regulators of immune checkpoints which can help for developing effective drug target to improve the effect of immunotherapy. Here, we verified that microRNA-326 (miR-326) repressed the gene expression of immune checkpoint molecules PD-L1 and B7-H3 in lung adenocarcinoma (LUAD). We detected that the expression of miR-326 in LUAD tissue was negatively correlated with PD-L1/B7-H3. The repression of PD-L1 and B7-H3 expression through miR-326 overexpression leads to the modification the cytokine profile of CD8+ T cells and decreased migration capability of tumor cells. Meanwhile, the downregulation of miR-326 promoted tumor cell migration. Moreover, blocking PD-L1 and B7-H3 attenuated the tumor-promoting effect induced by miR-326 inhibitor. In tumor-bearing mice, the infiltration of CD8+ T cells was significantly increased and the expression of TNF-α, and IFN-γ was significantly enhanced which contributed to tumor progression after miR-326 overexpression. Collectively, miR-326 restrained tumor progression by downregulating PD-L1 and B7-H3 expression and increasing T cell cytotoxic function in LUAD. Our findings revealed a novel perspective on the complex regulation of immune checkpoint molecules. A new strategy of using miR-326 in tumor immunotherapy is proposed.


INTRODUCTION
Lung cancer is the most frequent malignant tumor in the world [1]. Most patients are in the advanced stages when diagnosed [2]. Metastasis is the primary cause of death. Tumor immunotherapy has achieved rapid development and become one of the important strategies for advanced malignant tumors since the application of the first programmed cell death protein 1 (PD-1) antibody Opdivo [3,4]. Antibodies that block inhibitory immune checkpoints, such as PD-1, programmed cell death ligand 1 (PD-L1), enhance the killing capacity of T cells against cancer cells, have achieved significant therapeutic effects in cancers, especially lung cancer and melanoma [4][5][6]. PD-1/PD-L1 immunotherapy effectively prolongs the survival time of patients with advanced non-small cell lung cancer (NSCLC). But not all patients are sensitive to immunotherapy. More research on the regulatory mechanism of immune checkpoint molecules is required for precise targeted therapy.
In this study, we demonstrated that Hsa-miRNA-326 (miRNA-326) downregulates the expression of PD-L1 and B7-H3 in B7 family members and thus prevents tumor immune escape. PD-L1, also known as B7-H1 and CD274, is continuously expressed in many solid tumors, such as lung cancer, ovarian cancer, and renal cell carcinoma [7,8]. PD-L1 interact with PD-1 on T cells to make them lose their killing ability and tumors escape the surveillance and killing of the immune system, which promote tumor progression [7,9,10]. B7-H3, also called CD276, is overexpressed in many malignant tumors and closely related to the increased metastasis, recurrence, resistance to therapy, and poor prognosis of malignant tumors [11][12][13]. High expression of B7-H3 were proved in 74% of NSCLC [14] and showed positively associated with poor prognosis [11,15]. B7-H3 repressed T cell proliferation and attenuated activity of T cells, thus promotes cancer cell immune escape [16][17][18].
In this study, we identified the direct regulation of PD-L1 and B7-H3 expression by miR-326 and the subsequent suppression of CD8 + T cells in the tumor microenvironment and highlighted a new potential therapeutic target to overcome immune evasion by tumor cells.

RESULTS
miR-326 directly binds to 3′ UTRs of PD-L1 and B7-H3 Aberrantly expressed miR-326 participates in malignant progression of LUAD [31]. Nevertheless, how miR-326 regulates tumor immunity remains unclear. We utilized miRBase, PicTar, and TargetScan software to make prediction of miR-326 targeted genes and focused on the immune checkpoint-related genes. As shown in Fig. 1A, PD-L1, B7-H3, and ICOSLG genes in B7 family were supposed to be targeted by miR-326. We performed luciferase reporter assay to verify the direct binding of miR-326 to the putative targeted genes determined by bioinformatics target prediction analysis. Plasmid with the wild-type (WT) or mutant type (Mut) 3′ UTR of the targeted genes (PD-L1, ICOSLG, and B7-H3), miR-NC mimics, and miR-326 mimics were transfected into 293T cells. Obviously, miR-326 repressed wild-type luciferase reporter activity other than the mutant group ( Fig. 1B-D). This result suggests that miR-326 directly binds to the 3′ UTRs of PD-L1, B7-H3, and ICOSLG, and might repress the expression of these genes. The predicted binding sites and mutant sequence are shown in Fig. 1E.
MiR-326 is downregulated in LUAD and negatively related to PD-L1 and B7-H3 MiR-326 quantitation in LUAD tissues and adjacent normal tissues was accomplished by qRT-PCR. Consistent with previous reports and analysis from public datasets ( Supplementary Fig. 1A), miR-326 had lower expression in LUAD tissues than adjacent normal tissues ( Fig. 2A). Meanwhile, miR-326 expressed lower in LUAD cell lines than normal lung epithelial cell line BEAS-2B (Fig. 2B). The relationship of miR-326 and PD-L1/B7-H3 expression in LUAD tumor tissues were explored by qRT-PCR and Spearman's rankorder correlation analysis. As shown in Fig. 2C, D, miR-326 is negatively expressed compared with PD-L1 and B7-H3 in tumor tissues. What is more, miR-326 and PD-L1/B7-H3 showed a higher negative correlation in lymph node metastasis than in in situ tissues (Fig. 2E, F). This result implies a possible correlation related to tumor metastasis. However, the expression of miR-326 has no substantial relevance to ICOSLG in primary tumor and lymph node . E Schematic diagram of the binding sites between miR-326 and the target genes (PD-L1, B7-H3, and ICOSLG) and their mutant sequence. All data were presented as mean ± SEM. Comparisons between groups for statistical significance were performed with two-way ANOVA with Tukey's post hoc test (B-D). ***P < 0.001 versus 3′ UTR + miR-NC. ### P < 0.001 versus 3′ UTR + miR-326. metastatic tissues ( Supplementary Fig. 1B, C). Overall, these findings prompt that miR-326 probably mediates the immunoreaction in LUAD through regulating the expression of PD-L1 and B7-H3.

MiR-326 attenuates PD-L1 and B7-H3 expression in LUAD cells
We performed lentivirus infection to overexpress miR-326 in LUAD cell lines A549 and H1734 to confirm the direct regulation of miR-326 on PD-L1 and B7-H3. Figure 3A shows the efficiency of overexpression. As testified by qRT-PCR and western blot in Fig.  3B-F, enforced miR-326 decreased the expression of PD-L1 and B7-H3. However, overexpressed miR-326 only leads to considerable recession of ICOSLG at the RNA level but not in the protein level ( Fig. 3B-F). In addition, flow cytometry analysis also revealed that PD-L1 and B7-H3 on A549 and H1734 cell lines was remarkably downregulated (Fig. 3G, H). Conversely, PD-L1 and B7-H3 expression increased when miR-326 was downregulated through miR-326 inhibitor in Calu-3 and H1975 cell lines ( Supplementary Fig. 2). Overall, the results indicate that miR-326 repressed PD-L1 and B7-H3 expression in LUAD.
MiR-326 modifies the cytokine profile of CD8 + T cells and represses tumor cell migration in vitro PD-L1 and B7-H3 were supposed to cause T cell dysfunction and accelerate tumor progression, and miR-326 was suggested to repressed PD-L1 and B7-H3 expression. Thus, we intended to determine whether miR-326 affects T cell function by modulating PD-L1 or B7-H3. We constructed a co-culture system of tumor cell and T cells induced by peripheral blood. Then, we detected changes in T cell function. As shown in Fig. 4A-F, TNFα, IFN-γ, and IL-2 increasingly secreted in the supernatant of miR-326 overexpressed cells. Meanwhile, IL-1β, IL-10, and TGFβ were downregulated. To further verify whether the change of cytokines' profile was derived from T cells, we isolated CD8 + T cells and examine their cytokine expression by flow cytometry. As shown in Fig. 4G, the TNF-α and IFN-γ positive population in CD8 + T cells was significantly increased after miR-326 overexpression, while the IL-10 positive population was not changed and the IL-1β positive population was slightly reduced. Moreover, miR-326 and PD-L1/B7-H3 had a stronger negative correlation in lymph node metastasis than in situ, . All data were presented as mean ± SEM. Comparisons between groups for statistical significance were performed with Student's t test (A, B). Spearman's rank correlation coefficient was used to measure the association between miR-326 and PD-L1 (C, E) or B7-H3 (D, F). **P < 0.01, ***P < 0.001 versus tumor or BEAS-2B.
which implied that miR-326 might be related to metastasis. Therefore, we performed transwell assay for the cells in miR-326 overexpressed cells and found that enforced miR-326 substantially inhibited cell migration ( Fig. 4H-J). Meanwhile, MTS assay analysis showed that miR-326 overexpression did not significantly affect the cell proliferation (Fig. 4K), which demonstrated that miR-326 probably promotes the tumor metastasis. . All data were presented as mean ± SEM. Comparisons between groups for statistical significance were performed with Student's t test. *P < 0.05, **P < 0.01, ***P < 0.001 versus LV-NC. MiR-326 regulating the cytokine profile of CD8 + T cells and tumor cell migration is PD-L1/B7-H3-dependent We speculate that miR-326 inhibits T cell function through regulating PD-L1 and B7-H3 expression. First, we repressed miR-326 expression in H1975 cell lines and then co-cultured the cells with T cells. As expected, the downregulation of miR-326 in H1975 cells leads to the decrease of TNF-α, IFN-γ, and IL-2 secretion and the increase of IL-1β, IL-10, and TGF-β secretion ( Fig. 5A-F). Moreover, administration of PD-L1 and/or B7-H3 antibody considerably reverted the level of cytokine secretion ( Fig. 5A-F). Consistently, the downregulation of TNF-α and IFN-γ positive population and upregulation of and IL-1β positive population in CD8 + T cells by miR-326 inhibitor were also rescued by PD-L1 and B7-H3 antibody incubation (Fig. 5G). These findings suggest that miR-326 affects the function of CD8 + T cell secretion by interfering PD-L1 and B7-H3 function. Furthermore, downregulated miR-326 markedly enhanced the migration ability of tumor cells, and blocking PD-L1 or B7-H3 could reverse the effect caused by repressed miR-326 ( Fig.  5H-J). Overall, miR-326 prevents tumor cell migration through repressing PD-L1 and B7-H3.

DISCUSSION
Immunotherapy has made great progress in recent years. Blocking immune checkpoints is one of the most promising strategies for activating antitumor immunity [32,33]. In recent years, advances in targeted therapy for immune checkpoints, such as PD-L1/PD-1, brings hope for patients with advanced stages [8,34]. However, not all patients benefit from immunotherapy. The heterogeneous level of PD-L1 in patients indicates that PD-L1 expression is affected by many regulatory mechanisms. Even some PD-L1positive patients are refractory for treatment. The immune-related gene expression is under control of many regulators with diverse regulatory mechanisms [35]. Thus, exploring the upstream modulators of immune checkpoint molecules can broaden our understanding to develop novel drug targets.
In recent years, miRNAs have attracted much attention as a class of important gene regulatory factors. These noncoding RNAs repress gene expression and participate in pathological processes [20]. Some miRNAs are also involved in immune regulation [36,37]. MiR-326 is involved in the progress of various tumors, including lung cancer, and acts as an inhibitory factor [21,38,39]. However, the regulatory effect of miR-326 on immune checkpoint has been rarely explored.
PD-L1, ICOSLG, and B7-H3 are all members of B7 family. B7 family molecules are the immunoglobulin superfamily members that can provide stimulation signals to enhance and maintain T cell immune responses and also produce inhibitory signals to limit and weaken T cell immune responses. Dysregulated expression of B7 family will lead to T cell dysfunction and the immune escape of tumor cells, which is of great importance in tumorigenesis, development, and tumor-targeted therapy. PD-L1 is the most concerned molecule in immunotherapy research. PD-L1 has attracted much attention because of its ability to inhibit T cell function and its adjuvant effect on tumor immune escape. B7-H3 is a promising immune checkpoint molecule that is highly expressed in 74% of NSCLC. B7-H3 can reduce the release of interferon by T cells, weaken the cytotoxic activity of natural killer cells, and play a role in negative immune regulation [1]. Studies in NSCLC have revealed that the dual blockade of B7-H3 and PD-L1 enhances the antitumor reaction compared with a single blocking antibody [11,14,40]. That is to say, dual PD-L1 and B7-H3 signaling blockade therapy is a promising treatment strategy for NSCLC.
The analysis of clinical samples showed that the expression of miR-326 in tumor tissues was remarkably lower than that in adjacent tissues, and its expression level was negatively correlated with PD-L1 and B7-H3 (Fig. 2C, D). Furthermore, the negative correlation coefficient between miR-326 and PD-L1/B7-H3 was higher in lymph node metastasis than in situ (Fig. 2E, F), which prompted a possible correlation between miR-326 and metastasis.
B7-H3 and PD-L1 contribute to circumvent CD8 + T cellmediated immune surveillance. Hence, we co-cultured T cells with miR-326 overexpressing tumor cells to detect the levels of the cytokines secreted by T cells and determine the changes in T cell activity. T cells co-cultured with miR-326 overexpressed cells secreted more TNF -α, IFN -γ and IL-2 and less IL-1β, IL-10, and TGF-β. Furthermore, we determine the migration capacity of miR-326 overexpressed cells through transwell assay. The tumor cells that overexpress miR-326 had weakened migration ability (Fig.  4G-I). The co-cultured T cells were dysfunctional and secreted more immunosuppressive factors when miR-326 expression was repressed in H1975 cells (Fig. 5A-F). Moreover, PD-L1 and B7-H3 blocked the negated tumor-promoting efficacy induced by the miR-326 inhibitor; hence, an improved antitumor immunity is mediated by PD-L1/B7-H3.
Finally, we constructed tumor-bearing mice to explore the antitumor effect and immune regulation mechanism of miR-326 in vivo. The IHC staining of tumor tissue showed that enhanced miR-326 expression resulted in increased T cells infiltration. Furthermore, T cells in miR-326 overexpressed tissues secreted more antineoplastic cytokines (TNF-α and IFN-γ) a. CD8 + TILs were supposed to determine the metastatic potential in LUAD models. As expected, enhanced miR-326 restrained pulmonary metastasis in tumor-bearing mice. Additionally, inhibition of miR-326 by antagomir largely attenuated antitumor effect of miR-326.
This study confirmed the regulatory role of miR-326 in the expression of key immune checkpoints molecular PD-L1 and B7-H3, and highlighted a promising target for immunotherapy. Notably, miR-326 overexpression partially repressed T cell function by downregulating PD-L1 and B7-H3. This finding highlights the fact that tumor immunity is a complex process regulated by many factors. In addition, PD-L1 and B7-H3 expression is under complex regulation and miR-326 is one of the many contributing factors. Further research is needed for detailed regulatory mechanism.

Tissues
Tissues of patients (n = 50) were captured during the operation of LUAD. All patients were newly diagnosed in Shenzhen People's Hospital without any prior treatment. The project has been approved by the Ethics Committee of Shenzhen People's Hospital. All patients had informed consent.

Animal experiments
BALB/c mice (male, 6-8 weeks) were obtained from SPF (Beijing, China) Biotechnology Co., Ltd. The mice were bred in an SPF environment and injected subcutaneously with tumors cells (5 × 10 5 cells in 100 μL of PBS per mouse) when their condition was stable. The mice were randomly divided into four groups with 8 mice in each group to prevent unexpected situations. Tumor size was measured every other day 4 days after the injection. MiR-326 antagomir or negative control (10 nmol) was injected into the tumor every 4 days for 28 days.

Western blot
The harvested cells were lysed in a lysis buffer (50 mM Tris-HCl (pH 6.8), 2% sodium dodecyl sulfate, 1.5% DL-Dithiothreitol, 10% glycerol and 0.2% Bromophenol blue), and lysed protein was quantified before use. The same amount of lysed protein was loaded for electrophoresis and electrotransfer. The antibody (PD-L1, Cell Signaling Technology, 13684T; B7-H3, Cell Signaling Technology, 14058T; ICOSLG, Novus biological, NBP2-46011; GAPDH, Abcam, ab9485) was incubated according to the concentration indicated in the instruction manual. The expression of protein was detected by enhanced chemiluminescence (Millipore, Burlington, MA, USA) and analyzed by densitometry using Image J. The background correction was done with the value of 50 (called rolling disc in the software). To calculate the relative expression of specific protein, the glyceraldehyde-3phosphate dehydrogenase (GAPDH) serves as a reference for the sample loading.
Tumor and T cell co-culture system Peripheral blood mononuclear cells (PBMCs) from healthy human donors were isolated using Lymphoprep density gradient centrifugation (Stemcell, 07851). PBMCs were cultured in six-well plates at a density of 3 × 10 6 per well with H1734 or H1975 cell lysate (0.5 mg/ml) and IL-2 (20 ng/ml) for 72 h to promote T cell activation. Stimulated PBMCs were then harvested and purified by Lymphoprep density gradient centrifugation, and cocultured with the H1734 or H1975 cells at a 10:1 ratio for 16 h. The CD8 positive expressing T cells were purified by CD8 magnetic beads (Miltenyi, 130-096-730). The flow cytometry was then performed to analysis the expression of TNF-α, IFN-γ, IL-10, and IL-1β in CD8 + T cells. Co-culture media were assayed for TNF-α, IFN-γ, IL-2, IL-10, IL-1β, and TGF-β using a cytokine ELISA assay.

Lentivirus infection
The lentivirus used for overexpression and interference was purchased from Genechem (Shanghai, China). Cells (5 × 10 5 ) were laid on the six-well plate and incubated for 24 h until the cell density reached about 80% confluence. Lentivirus was added with transfection reagent according to the instruction from the manufacturer. Puromycin (1 μg/mL) was added 72 h after virus addition to screen for stable infection cells. The infection efficiency was verified by qRT-PCR and western blot.

Transwell assay
The cells were resuspended in FBS-free DMEM in the upper transwell chamber at a concentration of 1 × 10 5 cells/100 μL. Medium (500 μL) containing 20% FBS was added into the bottom chamber as a chemoattractant. After incubation for 24 h, the remaining cells in the inner side of the chamber were removed with cotton swabs. The cells that migrated to the outer side of the chamber membrane were fixed with fixative, stained with 0.5% crystal violet, dried at 80°C for 30 min, and photographed, Five fields were randomly selected under an inverted microscope to determine the number of cells.

Luciferase activity assay
293T cells in logarithmic growth phase were made into cell suspension, inoculated in a 24-well culture plate, and cultured until the cell fusion reached about 60%. Transfection was carried out with X-tremeGENE HP (Roche) for 48 h. Luciferase was detected using a Dual-Luciferase Reporter Assay System (Promega, Madison, WI, E1960) following a previously established protocol [41,42].

Enzyme-linked immunosorbent assay (ELISA)
The levels of cytokines were detected using an ELISA kit (Abcam, Cambridge, UK, USA). The experiment was performed following the manufacture's instruction. Then, enzyme activity was measured at 450 nm using a SPARK 10 M microplate reader (Tecan, Switzerland).

Statistical analysis
Statistical analysis was conducted using GraphPad Prism 5.0 evaluate the differences between different groups. All data are shown as mean ± standard deviation. Comparisons between groups for statistical significance were performed with Student's t test or analysis of variance (ANOVA) with post hoc test in multiple groups. P < 0.05 was considered statistically significant.