KDM1A Promotes Immunosuppression in Hepatocellular Carcinoma by Regulating PD-L1 through Demethylating MEF2D

Background Immune checkpoint inhibitor therapy targeting antiprogrammed cell death-1 (anti-PD-1) or its ligand (anti-PD-L1) is effective in the treatment of some hepatocellular carcinomas (HCC). Hence, further identification of biological targets related to PD-L1 regulation in HCC is beneficial to improve the clinical efficacy of immunotherapy. Some HCC cells express lysine-specific demethylase 1A (KDM1A), which is implicated in the reduced survival time of patients. Here, we studied whether the level of PD-L1 and the immunosuppression are regulated by KDM1A and its miRNA in HCC cells. Methods In the present study, we studied clinical data from The Cancer Genome Atlas (TCGA) database. We performed qPCR and western blotting assays to measure the expression level of genes of interest. PD-L1 expression was also analyzed by FACS. Clustered regularly interspaced short palindromic repeats (CRISPR)/Cas9 was used to generate gene knockout cells to investigate the relationships of genes of interest. We also developed a reporter gene assay (RGA) to explore the changes in T cell-induced antitumor immunity relative to PD-L1 expression in HCC cells. The binding between proteins and promoters or miRNAs and their target genes was explored by luciferase reporter assays. Results The results showed that PD-L1 and KDM1A were increased in HCC patients and cells, and KDM1A promoted the expression of PD-L1 in HCC cells. Our findings showed that the enhancement of PD-L1 expression was not attributed to mitochondrial dysfunction caused by increases in KDM1A in HCC cells. Furthermore, we observed a lower level of MEF2D methylation in HCC cells than in normal human liver cells. Demethylated MEF2D could bind to the promoter of PD-L1 and activate its expression, while KDM1A interacted with MEF2D and acted as a demethylase to reduce its methylation. Moreover, a new miRNA, miR-329-3p, targeting KDM1A was found to regulate the PD-L1 expression profile in HCC cells. In the xenograft model, the tumors treated with miR-329-3p showed growth inhibition. Conclusions Mechanistically, miR-329-3p inhibits tumor cellular immunosuppression and reinforces the response of tumor cells to T cell-induced cytotoxic effect by targeting KDM1A mRNA and downregulating its expression, which contributed to MEF2D demethylation and activation of PD-L1 expression.


Introduction
Liver cancer, which is a global health problem, is the sixth most frequent malignancy and the fourth leading cause of cancer-related death worldwide [1]. Hepatocellular carcinoma (HCC), accounting for 90% of liver cancer, is the third most common cause of death globally, with a 5-year survival rate of 18% [1,2]. Notably, HCC is a kind of immunogenic liver injury influenced by tumor-infiltrating lymphocytes, and lymphocyte-mediated antitumor immunity is able to prevent HCC malignancy as well as progression [3].
Programmed cell death-ligand 1 (PD-L1), encoded by the CD274 gene, is the ligand of programmed cell death-1 (PD-1) and exists in several cancers, and it has the ability to inhibit T cell activation [4]. Neutralizing antibodies against immune checkpoints, such as PD-L1 or PD-1, show great performance as therapies for many cancers [4]. Therefore, the identification of biological targets related to PD-L1 regulation in HCC is helpful to improve the clinical efficacy of immunotherapy. Thus, it is necessary to further understand the pathways controlling PD-L1 expression in HCC to heighten the efficacy of PD-L1/PD-1 blockade.
Lysine-specific demethylase 1A (KDM1A, also named LSD1) was the first discovered histone-specific demethylase and is well known because of its ability to catalyze lysine demethylation in a flavin adenine dinucleotide-(FAD-) dependent oxidative reaction [5]. KDM1A could demethylate H3K4me1/2 (Lys-4) and H3K9me1/2 (Lys-9), which means it acts as a coactivator or a corepressor depending on the context [6][7][8]. In addition to histones, KDM1A is also able to demethylate lysine residues in several nonhistone proteins, such as p53 [9], Dnmt1 [10], E2F1 [11], and MYPT1 [12]. Because of its ability to control such a wide range of proteins, KDM1A is associated with an expansive spectrum of biological processes, such as cell proliferation [13], hematopoiesis [14], and embryonic development [15]. Evidence has shown that the dysregulation of KDM1A plays an important role in tumorigenesis in several cancers [12], including HCC. However, little is known about the relationship between KDM1A and T cell-mediated antitumor immunity in HCC.
MicroRNAs (miRNAs) are small noncoding RNAs that regulate a number of biological processes posttranscriptionally by binding to the 3 ′ untranslated region (UTR) of target mRNAs and suppressing their translation or accelerating their degradation [16]. Many miRNAs have been proven to play an important role in cancer progression by regulating the epithelial-mesenchymal transition (EMT), energetic metabolism, tumor angiogenesis [17][18][19][20], and other processes.
Here, we revealed that KDM1A could control the PD-L1 expression level in HCC by demethylating myocyte enhancer factor 2D (MEF2D), a transcription factor promoting PD-L1 expression without methylation. We also found that miR-329-3p could repress PD-L1 expression by targeting and downregulating KDM1A mRNA.

Materials and Methods
2.1. Cell Culture. The normal human liver cell line L-O2, human HCC cell lines HepG2 and SMMC7721, and mouse HCC cell line H22 were purchased from Procell Life Science & Technology. The L-O2, SMMC7721, and H22 cell lines were cultured in RPMI-1640 (Gibco) containing 10% fetal bovine serum (FBS) (Gibco), 100 μg/mL streptomycin, and 100 U/mL penicillin (Gibco). HepG2 cells were cultured in MEM (Gibco) containing 100 μg/mL streptomycin, 100 U/mL penicillin, and 10% FBS. All cell lines were cultured at 37°C with 5% CO 2 . All cell lines were routinely tested for mycoplasma contamination and found to be negative.

Clustered Regularly Interspaced Short Palindromic
Repeats (CRISPR)/Cas9-Generated KDM1A, PD-L1, or MEF2D Knockout Cell Lines. The Cas9-GFP protein and sgRNA were obtained from GenScript. All sgRNA sequences are shown in Table S1. The CRISPR/Cas9 system was used as described in a previous study [21]. Briefly, before transfection, SMMC7721 cells (3 × 10 4 cells per well) were seeded in a 6-well plate. Then, Opti-MEM containing the Lipofectamine Cas9 Plus™ Reagent (Invitrogen) was used to transfect a mixture of 50 pmol Cas9-GFP protein and 50 pmol sgRNA into SMMC7721 cells, which were cultured for 2 days at 37°C. Single clones were generated with the limiting dilution method in 96-well plates after sorting for GFP-positive cells by using a FACSAria instrument (BD Biosciences). All of the single clones were identified by sequencing and western blotting analyses. Cells with the successful KDM1A knockout (SMMC7721 KDM1A-/-), PD-L1 knockout (SMMC7721 PD-L1-/-), and MEF2D knockout (SMMC7721 MEF2D-/-) were used for subsequent experiments.
The miR-329-3p inhibitor (inhibitor) and negative inhibitor control (inhibitor NC) were generated by RiboBio. For plasmid transfection, pcDNA3.1 (Invitrogen) was constructed and contained the target genes. Small interfering RNAs (siRNAs) were obtained from Sangon Biotech. Transfection using the Lipofectamine 2000 reagent (Thermo Fisher Scientific) was performed according to the standard protocol described in the manufacturer's instructions.
The sequences of miRNAs and inhibitors are shown in Table S1.

Quantitative Real-Time Polymerase Chain Reaction
(qRT-PCR). The TRIzol reagent (Invitrogen) was used for extracting total RNA, and the PrimeScript RT reagent kit (TaKaRa) was used for synthesizing cDNAs except for those corresponding to the miRNA, which was generated with a TaqMan Advanced miRNA cDNA synthesis kit (Waltham).

2
Journal of Immunology Research qPCR quantification was performed by using the TB Green® Fast qPCR Mix (TaKaRa) according to the manufacturer's manual on the ABI-7500 (Applied Biosystems). For testing miRNA expression, stem-loop qRT-PCR was used referring to others' work [23,24]. Primers are shown in Table S1.
2.6. Western Blotting Analysis. Proteins were extracted from cells (RIPA, Beyotime Biotechnology) and tumors (One Step Animal Tissue Active Protein Extraction Kit, Sangon Biotech), and quantification of protein lysates was performed by a bicinchoninic acid assay kit (Boster). After that, the proteins were boiled for 5 min at 100°C in a loading buffer (20 μg) and separated in an SDS-PAGE gel at 80 V for 30 min and then at 120 V for 1 h. After that, the proteins were electrotransferred into a nitrocellulose membrane (Boster) at 300 A for 1.5 h and blocked for 1 h with Tris-buffered saline containing 0.1% Tween 20 (TBST) and 5% fat-free milk. After that, the membranes were incubated with corresponding primary antibodies overnight at 4°C and secondary antibodies for 2 h at room temperature after washing. The membranes were measured by an image analysis system (Image-Pro Plus 6.0, Media Cybernetics) after incubating with a high-signal electrochemiluminescence kit (Fdbio Science). Cholesterol-conjugated miR-329-3p or mi-NC (50 nmol) was intratumorally injected into the two groups three times per week for two weeks. Tumor growth was detected every 6 days. Tumors were weighed and isolated after mice were sacrificed at 6 weeks after grafting.

Chromatin Immunoprecipitation-(ChIP-) PCR
Analysis. ChIP was performed using ChIP Kit-One Step (Abcam) according to the manufacturer's protocol. 2 μL of antibodies was used in the ChIP process. Anti-MEF2D antibodies were the same in the IP assay. Anti-H3K4me1, anti-3 Journal of Immunology Research H3K4me2, anti-H3K4me3, anti-H3K9me1, anti-H3K9me2, and anti-H3K9me3 antibodies were obtained from Abcam. 2 μL from 100 μL input DNA and eluted DNA was detected by qPCR as described previously. The primers are shown in Table S1.
2.16. LDH-Based T Cell Killing Assay (Cytotoxicity Assay). 1 × 10 4 target cells (SMMC7721) were seeded in triplicate in a 96-well U-bottom plate with an effector (human peripheral blood mononuclear cell (PBMC)) (SailiBio, Inc.) at a ratio of 10 : 1 (effector : target (E : T)) and coincubated in 200 μL of RPMI-1640 containing 4% FBS for 6 h at 37°C with 5% CO 2 . The cytotoxicity was measured by the lactate dehydrogenase (LDH) release assay (Non-Radioactive Cytotoxicity Assay, Promega). The absorbance at 490 nm was recorded by using a microplate reader (BioTek Instruments, Inc., USA). Except for wells with E+T (samples), other wells were designed for controls: wells only with a medium (blank), wells only with PBMC (E only), and wells only with target cells (T only). Each control well was triplicated. The T only wells were completely lysed by adding 30 μL lysis solution (10x). The cytotoxic rate % was calculated according to the following formula: cytotoxic rate% = ðsamples − E only − blankÞ/ðT only − blankÞ * 100.
Cell-free supernatants from T cell and tumor cell coculture were harvested at 6 h for the presence of interleukin-2 (IL-2) and interferon-γ (IFN-γ). These cytokines were quantified by ELISA kits (Boster, China). All samples were tested in duplicate according to the manufacturer's instructions.
2.17. Statistics. All quantitative data were evaluated to determine the normality of the distribution using the Shapiro-Wilk test. Student's t-test was performed using Prism (version 5; GraphPad Software). To compare multiple groups, one-way ANOVA followed by Tukey's multiple comparison test was performed using Prism. The data are presented as the means ± SD. p < 0:05 was considered to indicate a statistically significant difference.

Results
3.1. KDM1A Controls the Level of PD-L1 in HCC. First, we analyzed KDM1A expression in data available from TCGA database with the ENCORI website (http://starbase.sysu.edu .cn) to evaluate the importance of KDM1A in HCC. We observed a 2-fold higher level of KDM1A in HCC patients compared with healthy people in the clinic (p < 0:05; Figure 1(a)). The importance of KDM1A was also emphasized by its performance based on the overall survival curve for HCC, which demonstrated a significant difference in clinical outcomes between HCC patients and healthy persons ( Figure 1(b)). In addition, the clinical coexpression analysis performed using the ENCORI website indicated that a high level of PD-L1 was related to high KDM1A expression, with a p value = 9:46E − 08 in HCC (Figure 1(c)). These findings implied that KDM1A may be involved in PD-L1 regulation and contribute to immunosuppression in HCC.
Based on the results of the clinical data analyses, we hypothesized that KDM1A could regulate the PD-L1 in HCC. Thus, we detected PD-L1 and KDM1A expression in different cell types, including the normal human liver cell line L-O2, the human HCC cell lines HepG2 and SMMC7721, and the mouse HCC cell line H22. KDM1A expression, as determined by qPCR and western blotting, was much higher in HCC cells than in normal human liver cells, which is similar to the tendency of PD-L1 expression observed in those cells (Figures 1(d) and 1(e)). Based on PD-L1 expression, we used SMMC7721 as our cell model in the in vitro assays.
Moreover, we want to know whether the PD-L1 expression variation was attributed to its chromosome openness affected by KDM1A, as it is a well-known histone demethylase. KDM1A can demethylate both "Lys-4" (H3K4me) and "Lys-9" (H3K9me) of histone H3 to control the H3K4me1/2/3 and H3K9me1/2/3 levels, which are associated with the chromosome openness [26]. Thereby, we have performed ChIP-qPCR to investigate this question. We measured the level of the PD-L1 promoter immunoprecipitated with anti-H3K4me1/2/3 or anti-H3K9me1/2/3 antibodies in wild-type SMMC7721 treated with the si-KDM1A or KDM1A overexpression plasmid. The results showed that the level of the PD-L1 promoter did not display a significant change responding to the KDM1A variation, indicating that KDM1A failed to affect the chromosome openness of PD-L1 directly (Figure 1(j)).
In general, these results suggested that KDM1A promotes the expression of PD-L1 independent with H3K4 and H3K9 demethylation in HCC.

The Alteration of PD-L1 Expression Is Irrelevant to
Mitochondrial Dysfunction Caused by KDM1A in HCC. Several reports have implied that PD-L1 expression is connected with mitochondrial function to some extent in cancers, and KDM1A injures mitochondrial function in cancers [27][28][29]. Thus, it was necessary to inquire whether the regulation of the PD-L1 level was attributed to the mitochondrial function change caused by KDM1A. First, we used SMMC7721 KDM1A-/cells to confirm mitochondrial function. The mitochondrial membrane potential (ΔΨm) is a key indicator for evaluating mitochondrial function. It was obvious that the knockout of 4 Journal of Immunology Research KDM1A restored the ΔΨm in SMMC7721, as shown by the increase in the relative ratio of FL 590 /FL 525 , which is a value used to assess the level of ΔΨm (Figure 2(a)). Another key indicator is the efficiency of ATP synthesis, which is driven by ΔΨm and indicated by the relative ratio of ADP/ATP. The ATP synthesis efficiency was more robust in 7 Journal of Immunology Research SMMC7721 KDM1A-/cells with a lower ADP/ATP ratio than SMMC7721 cells (Figure 2(b)). We also detected ROS in cells by staining with MitoSOX. A lower percentage of MitoSOXpositive cells resulted in a decreased ROS level in SMMC7721 KDM1A-/- (Figure 2(c)). These results suggest that abolishing KDM1A rescues mitochondrial function in SMMC7721 cells.
Furthermore, we studied the relationship between PD-L1 expression and mitochondrial dysfunction by using rotenone to inhibit mitochondrial function in SMMC7721 KDM1A-/cells. SMMC7721 KDM1A-/cells treated with 3 mM rotenone exhibited a change in the ΔΨm, decreased ATP synthesis, and increased ROS levels, indicating that mitochondrial function was impaired (Figures 2(a)-2(c)). However, there was no significant difference in PD-L1 levels between the groups with or without rotenone treatment for SMMC7721 KDM1A-/cells, which indicated that mitochondrial dysfunction did not contribute to PD-L1 regulation in HCC (Figures 2(d) and 2(e)).
In addition to PD-L1 expression, we also determined the ability of SMMC7721-based cell lines to inhibit T cellmediated antitumor immunity through a RGA. This was based on the stable expression of a luciferase reporter under the control of the nuclear factor of activated T cell (NFAT) response elements in Jurkat cells (Figure 2(f)). The Jurkat cell line is an acute lymphocytic leukemia cell line derived from T lymphocytes, which can be activated by APCs or tumor cells with the CD3 agonist antibody OKT3 in T cell-like pathways, and this activation can be inhibited by immune checkpoints. In the Jurkat-based RGA, luciferase production only occurred in cells activated by SMMC7721 and OKT3, whereas the process was influenced by variation in the PD-L1 abundance on the SMMC7721 surface. Obviously, rotenone-induced mitochondrial dysfunction failed to decrease luciferase production in SMMC7721 KDM1A-/cells (Figure 2(g)). In contrast, the groups of SMMC7721 KDM1A-/overexpressing KDM1A showed a reduced luciferin signal compared to the SMMC7721 group (Figure 2(g)).
Together, these findings revealed that the changes in the PD-L1 steady state in HCC are not associated with changes in mitochondrial function caused by KDM1A, suggesting that another mechanism is involved in PD-L1 expression regulated by KDM1A.  The data are presented as the means ± SD. n = 3 experiments in (a-g). * p < 0:05, * * p < 0:01, and * * * p < 0:01.

8
Journal of Immunology Research

11
Journal of Immunology Research regulates PD-L1 expression by its acetylation in HCC [25], we thought that MEF2D might play a significant role in the regulation of PD-L1 by KDM1A. We also found that the clinical outcome analysis showed an increased level of MEF2D in HCC patients (Figure 3(a)). Therefore, we adopted an IP assay to determine whether KDM1A could cause the demethylation of MEF2D. The anti-K267me antibody was used to identify the MEF2D methylation regulated by KDM1A [22]. Flag-KDM1A and HA-MEF2D recombinant proteins were cotransfected into 293T cells, and the methylation level in MEF2D was detected by western blotting. The results showed a reduction in MEF2D methylation in the presence of KDM1A, suggesting that KDM1A could induce MEF2D demethylation (Figure 3(b)).
Compared with that in L-O2 cells, MEF2D expression was higher in SMMC7721 cells, which was consistent with the clinical data (Figures 3(c) and 3(d)). However, the methylation of MEF2D was reduced by half in SMMC7721 cells, as analyzed by IP (Figure 3(c)). We next evaluated the MEF2D methylation level in SMMC7721 KDM1A-/cells with or without KDM1A overexpression. The MEF2D methylation level decreased dramatically with KDM1A overexpression together with elevated PD-L1 expression (Figure 3(e)). The level of MEF2D displayed a slight increase, which may compensate for the change in its methylation and PD-L1 expression (Figure 3(e)). These results implied that the upregulation of PD-L1 expression caused by KDM1A occurred via demethylation of MEF2D in HCC.
Xiang et al. found that the binding sites of MEF2D were located in the -535 to -516 (P3) and -246 to -245 (P4) positions of the PD-L1 promoter [25]. According to the results, we used a reporter gene assay to investigate the influence of MEF2D methylation on its ability to bind to the PD-L1 promoter (Figure 3(f)). The reporter vectors consisted of a luciferase gene activated by the wild-type PD-L1 promoter (P0WT) and mutant PD-L1 promoter (P3+4M) that were transfected into different cells. The demethylation of MEF2D caused by KDM1A was inhibited by ORY-1001, a micromolecule inhibiting the demethylase activity of KDM1A. It was obvious that the luciferase activity was decreased in SMMC7721 cells with KDM1A or MEF2D abolishment (Figure 3(g)). In wild-type SMMC7721 cells with the P0WT promoter, the reduction in luciferase activity exhibited a dose-dependent effect with ORY-1001 (Figure 3(g)). SMMC7721 cells without MEF2D were generated by CRISPR/Cas9, similar to the SMMC7721 KDM1A-/cells, with 17 bp and 7 bp deletions within the alleles, respectively (Figure 3(h)). Then, we measured the binding between the MEF2D and the PD-L1 promoter in different conditions by ChIP-qPCR. The results showed a decreased enrichment of PD-L1 promoter tendency along with the KDM1A knockout or activity inhabitation, while its enrichment increased with KDM1A overexpression (Figure 3(i)). We also found that the PD-L1 level kept a stable state in SMMC7721 MEF2D-/cells treated with si-KDM6A or KDM6A overexpression (Figures 3(j) and 3(k)).
Moreover, we investigated the influence of KDM1A activity on immunosuppression in SMMC7721 cells. We performed a FACS assay to monitor the PD-L1 level on the surface of SMMC7721 cells treated with different concentrations of ORY-1001. ORY-1001 reduced PD-L1 abundance at the cellular surface associated with the inhibited MEF2D methyl-ation level (Figures 3(n) and 3(o)). Consistent with the FACS results, the luciferase activity showed an ORY-1001 dosedependent tendency, with the highest signal observed in KDM1A-or MEF2D-lacking cells and the lowest signal observed in wild-type cells (Figure 3(p)). Similarly, the LDH-based T cell killing assay also displayed a progressive cytotoxic rate, as well as the level of IFN-γ and IL-2, with the higher ORY-1001 concentration (Figures 3(q)-3(s)). These results suggested that the suppression of KDM1A activity subdued the immunosuppression of SMMC7721 cells and promoted the T cell response.
Together, these data demonstrated that KDM1A-driven MEF2D demethylation promotes PD-L1 abundance and plays a suppressive role in T cell-mediated antitumor immunity in HCC.
To further confirm the predicted results for miR-329-3p, we performed a dual-luciferase reporter assay that utilized wild-type and mutant KDM1A 3 ′ UTR luciferase plasmids as well as miR-329-3p and mi-NC mimics according to the predicted alignment (Figures 4(a) and 4(b)) [31]. Cells cotransfected with the wild-type KDM1A 3 ′ UTR luciferase and miR-329-3p mimics showed a conspicuous reduction in luciferase activity relative to the cells cotransfected with the wild-type KDM6A 3 ′ UTR and mi-NC, while there was no difference in activity for cells with the mutant KDM1A 3 ′ UTR luciferase and miR-329-3p or mi-NC mimics, indicating that KDM1A mRNA is the target of miR-329-3p.
Then, we constructed SMMC7721 cells stably overexpressing miR-329-3p (SMMC7721 miRNA ) or mi-NC (SMMC7721 mi-NC ) and monitored the variation in KDM1A and PD-L1 expression. At the mRNA level, KDM1A and PD-L1 mRNA levels decreased dramatically, resulting in nearly 9-fold higher miR-329-3p expression (Figure 4(c)) compared with that in the mi-NC groups. In contrast, SMMC7721 miRNA cells exhibited increased KDM1A and PD-L1 mRNA levels when transfected with the miR-329-3p inhibitor, which decreased miRNA expression (Figure 4(d)). Consistent with the mRNA results, miR-329-3p overexpression decreased KDM1A expression and further decreased PD-L1 levels, leading to the upregulation of MEF2D methylation, while the miR-329-3p inhibitor reversed this effect (Figure 4(e)). Furthermore, we detected the effect of miR-329-3p on the anti-immune activity of SMMC7721 cells by the Jurkat-based RGA. Relative to that in the SMMC7721 mi-NC group, the luciferase activity increased greatly in the SMMC7721 miRNA group, but it was suppressed by transfection with the miR-329-3p inhibitor (Figure 4(f)). What is more, an elevated level of the T cell-induced cytotoxic rate as well as IFN-γ and IL-2 release levels was observed in miR-329-3p transfected target cells, whereas all of them were repressed dramatically when treated with the miR-329-3p inhibitor (Figures 4(g)-4(i)).
Our findings suggested that hsa-miR-329-3p reduced the immunosuppression in HCC cells and promotes its response toward T cell-induced cytotoxic effect by inhibiting PD-L1 expression in HCC through targeting KDM1A mRNA and indirectly reducing MEF2D demethylation.
3.5. miR-329-3p Enhances the T Cell Response toward HCC Tumors by Controlling the Expression of PD-L1 in the KDM1A/MEF2D Pathway. To study the impact of miR-329-3p on the immune response of HCC tumors, a wildtype BALB/c mouse xenograft model was developed by implanting H22 cells with high PD-L1 and KDM1A expression. After two weeks of cell implantation, we intratumorally injected cholesterol-conjugated mi-NC or miR-329-3p mimics into mice three times per week. It was easily observed that miR-329-3p prevented tumor growth, which was verified again by the tumor volume measurements (Figures 5(a) and 5(b)). Moreover, the tumor weight in the miR-329-3p group, which was measured at least weekly, was much lower than that in the mi-NC group ( Figure 5(c)). These results suggested that miR-329-3p prevents HCC tumor growth in vivo. Thus, we focused on the mechanism of HCC tumor growth inhibition by miR-329-3p. The mRNAs of KDM1A and PD-L1 as well as miR-329-3p in separated tumors at the end of 6 weeks were detected first. Compared with the tumors injected with mi-NC, the tumors injected with mi-329-3p displayed decreased expression of KDM1A and PD-L1 mRNA ( Figure 5(d)). We also found similar results at the protein level, with lower expression of KDM1A and PD-L1 in miR-329-3p-injected tumors and elevated MEF2D methylation ( Figure 5(e)). In addition, we tested the