What Do We Have to Know about PD-L1 Expression in Prostate Cancer? A Systematic Literature Review. Part 5: Epigenetic Regulation of PD-L1

Epigenetic alterations (including DNA methylation or miRNAs) influence oncogene/oncosuppressor gene expression without changing the DNA sequence. Prostate cancer (PC) displays a complex genetic and epigenetic regulation of cell-growth pathways and tumor progression. We performed a systematic literature review (following PRISMA guidelines) focused on the epigenetic regulation of PD-L1 expression in PC. In PC cell lines, CpG island methylation of the CD274 promoter negatively regulated PD-L1 expression. Histone modifiers also influence the PD-L1 transcription rate: the deletion or silencing of the histone modifiers MLL3/MML1 can positively regulate PD-L1 expression. Epigenetic drugs (EDs) may be promising in reprogramming tumor cells, reversing epigenetic modifications, and cancer immune evasion. EDs promoting a chromatin-inactive transcriptional state (such as bromodomain or p300/CBP inhibitors) downregulated PD-L1, while EDs favoring a chromatin-active state (i.e., histone deacetylase inhibitors) increased PD-L1 expression. miRNAs can regulate PD-L1 at a post-transcriptional level. miR-195/miR-16 were negatively associated with PD-L1 expression and positively correlated to longer biochemical recurrence-free survival; they also enhanced the radiotherapy efficacy in PC cell lines. miR-197 and miR-200a-c positively correlated to PD-L1 mRNA levels and inversely correlated to the methylation of PD-L1 promoter in a large series. miR-570, miR-34a and miR-513 may also be involved in epigenetic regulation.


Introduction
As the discovery of novel biomarkers is urgently required to develop tailored therapies for various malignancies [1], increasing attention has been paid to immunotherapy targets such as Programmed death-1 (PD-1) and its ligand (PD-L1). They are type I transmembrane glycoproteins transcribed by PDCD1 (located on chromosome 2) and CD274 genes (located on chromosome 9), respectively [2,3]. PD-1 is expressed by activated T, B, NK cells, and monocytes, while PD-L1 is found on hematopoietic and non-hematopoietic cells: their expression is inducible by microenvironmental conditions [2,3]. Indeed, pembrolizumab monotherapy (anti-PD-1 monoclonal antibody) recently revealed good therapeutic activity, and the 2021 United States National Comprehensive Cancer Network (NCCN) guidelines have considered this drug indicated in selected prostate cancer (PC) patients [4,5]. So, at least in the US, patients with metastatic castration-resistant PCs showing microsatellite instability/mismatch-repair protein system deficiency (MSI-H/dMMR) could be treated with pembrolizumab as a second-line therapy setting or beyond. Unfortunately, the prevalence of MSI-H/dMMR PCs is low, and the administration of immunotherapy in PC patients is still limited in the current clinical practice [4,5].
Epigenetic alterations induce reversible and heritable changes, promoting differences in the expressions of oncogenes and oncosuppressor genes without changing the DNA sequence [6]. DNA methylation, covalent histone modifications, histone variants, microR-NAs (miRNAs) effects, and chromatin-remodeling complexes are well-identified epigenetic mechanisms. However, PC displays a complex genetic and epigenetic regulation, leading to changes in cell growth pathways and overall tumor progression [6].
Epigenetic modifications accumulated by cancer cells influence gene expression and may contribute to tumor immune escape, also by targeting checkpoint inhibitors such as PD-L1. Moreover, epigenetic modulating drugs may be promising in reprogramming tumor cells, reversing epigenetic modifications, and cancer immune evasion [7][8][9][10][11].
Unfortunately, despite the increasing attention on molecular and epigenetic regulators in PCs, there is still limited evidence concerning the complex network of epigenetic factors modulating PD-L1 expression in PC [12]. In our systematic literature review, we have tried to describe the current knowledge on this topic.  Figure 1 presents the "Preferred Reporting Items for Systematic Reviews and Meta-Analyses" (PRISMA) (http://www.prisma-statement.org/, accessed on 8 May 2021) flow chart, summarizing the research method and results of our systematic literature review.   155 articles were considered eligible, as they seemed to report clinic-pathologic studies on human patients or experimental research on pre-clinical models (tumor cell lines, mouse models, etc.) investigating the role of PD-L1 in PC. After reading the full texts of all these papers, 7 articles were excluded for being unfit according to the inclusion criteria or for presenting scant or aggregated data. 148 articles were finally included in our study [5,. Further details are available in Section 4.

Epigenetic Regulation of PD-L1 Expression: Pre-Clinical Models
PD-L1 expression is under epigenetic control, as demonstrated by functional and correlation studies on PC cell lines (Tables 1 and 2) [7,14,26,60,73,115,116,123,127,159]. We identified 263 articles on PubMed (Available online: https://pubmed.ncbi.nlm. nih.gov, accessed on 8 May 2021), 385 articles on Scopus (Available online: https:// www.scopus.com/home.uri, accessed on 8 May 2021), and 399 articles on Web of Science databases (Available online: https://login.webofknowledge.com, accessed on 8 May 2021). After duplicates exclusion, 560 records underwent a screening of titles and abstracts. 155 articles were considered eligible, as they seemed to report clinic-pathologic studies on human patients or experimental research on pre-clinical models (tumor cell lines, mouse models, etc.) investigating the role of PD-L1 in PC. After reading the full texts of all these papers, 7 articles were excluded for being unfit according to the inclusion criteria or for presenting scant or aggregated data. 148 articles were finally included in our study [5,. Further details are available in Section 4.
Histone modifiers also influence the PD-L1 transcription rate in PC cell lines: data suggested that the deletion or silencing of the histone modifiers MLL3 and MML1 may positively regulate PD-L1 expression [14,60]. Xiong et al. [60] found that MLL3 bound to the PD-L1 enhancer, while MLL3 depletion decreased the binding of the methylated histone H3 on Lysine 4 in the PD-L1 enhancer and Pol II Ser5p in the PD-L1 promoter. Moreover, it impaired mouse xenograft growth, decreasing the response to the PD-L1 antibody treatment.
The transcription factor IRF-1 could recruit p300 to the CD274 promoter, inducing the acetylation of histone H3 and increasing CD274 transcription [116]. In PC cell lines, protein levels of the histone deacetylase HDAC1 negatively correlated to PD-L1 levels, while there was no correlation between HDAC2/3 and PD-L1 [116].
Functional studies ( Table 2) revealed that treatments with epigenetic drugs promoting a chromatin-inactive transcriptional state (such as bromodomain or p300/CBP inhibitors) [116,123,127] induced a reduction of PD-L1 expression; conversely, epigenetic drugs inducing a chromatin-active state (i.e., HDAC inhibitors) increased PD-L1 expression [116]. In an experimental study [116], p300 inhibitors (but not anti-PD-L1 antibodies) significantly enhanced the efficacy of HDAC inhibitors on limiting tumor progression by blocking the HDAC inhibition-induced PD-L1 expression.
Class I and II HDAC inhibitors such as SAHA (vorinostat) and LBH589 (panobinostat), as well as IFN-γ, significantly increased CD274 expression in PC cell lines. RNA polymerase II was also enriched at the CD274 promoter after HDAC treatment [116]. A485 may enhance the efficacy of treatments with anti-PD-L1 antibodies, decreasing the PD-L1 expression and reducing the exosomal PD-L1 secreted by PC cell lines; the combined administration of these drugs inhibited the androgen-independent metastatic tumor growth in syngeneic PC models [116].
The histone methylation regulator WDR5 (WD repeat-containing protein 5) was overexpressed in PCs with advanced clinic-pathological features and seemed important for PD-L1 transcription [14]. In PC cell lines, the IFN-γ-induced PD-L1 mRNA and protein levels were significantly abrogated by WDR5 or MLL1 knockdown (not by c-MYC silencing), as well as by OICR-9429 (a highly selected and potent antagonist of WDR5 interactions with MLL1, c-MYC, and other partners) [14]. Some cell cycle, anti-apoptosis, DNA repair, and immune-related genes (such as AURKA, CCNB1, E2F1, PLK1, BIRC5, XRCC2, and CD274) were directly regulated by WDR5 and OICR-9429 in a H3K4me3 (histone H3 Lysine 4 tri-methylation)-and c-MYC-dependent manner; WDR5 knockdown and OICR-9429 could reduce c-MYC recruitment and cell proliferation, increasing apoptosis and chemosensitivity to cisplatin in vitro and in vivo [14].
EZH2 (enhancer of zeste homolog 2) is the methyltransferase catalytic subunit of the polycomb repressive complex 2 (PRC2), which trimethylates Lysine 27 of histone H3 (H3K27me3) to promote transcriptional repression [158]. In PC models [158], EZH2 inhibition activates a double-stranded RNA-STING (stimulator of interferon genes)-ISGs (interferon-stimulated genes) stress response in tumor cells, upregulating genes involved in antigen presentation, Th1 chemokine signaling, and interferon response, including PD-L1 (dependent on STING activation) [158]. Moreover, EZH2 inhibition substantially increased the intratumoral trafficking of activated CD8+ and CD4+ T cells (decreasing the relative number of regulatory T cells, Tregs) and increased M1 tumor-associated macrophages (TAMs) (decreasing the tumor-promoting M2 macrophages). This pathway reversed the resistance to anti-PD-1 therapy in B6-MYC-CaP PCs in vivo [158]. mRNA translation is modulated by the rate of the elongation phase of protein synthesis through the phosphorylation of eukaryotic elongation factor 2 (eEF2) by its Ca 2+ -dependent protein kinase (eEF2K), which promotes PD-L1 expression [159]. miRNAs can regulate PD-L1 at a post-transcriptional level. In cell lines, Chen et al. [26] found that miR-15a negatively regulated PD-L1 expression. Moreover, miR15a overexpression promoted the cytotoxicity of CD8+ T cells against PC cells via directly targeting PD-L1, also decreasing tumor cell viability, migration, and invasion. miR-15a mimics downregulated pathways involved in the epithelial-mesenchymal transition (EMT) and RAS/ERK signaling. In contrast, PD-L1 overexpression in miR-15a mimics-transfected PC cells reversed these phenotypes; PD-L1 may influence the tumor-suppressive activity of miR-15a and stimulate multiple malignant phenotypes by activating the RAS/ERK signaling [26]. The long non-coding RNA gene KCNQ1 overlapping transcript 1 (lncRNA KCNQ1OT1) sponged miR-15a to upregulate the expression of PD-L1, thus inhibiting the cytotoxicity of CD8+ T cells and promoting tumor evasion [26]. Knocking down KCNQ1OT1 lowered PD-L1 expression and inhibited the viability, migration, invasion, and EMT of tumor cells, favoring their apoptosis; moreover, it enhanced the cytotoxicity and proliferation of CD8+ T-cells, reducing their apoptosis.
Functional experiments have demonstrated that miR-195 and miR-16 influenced the PD-L1-associated apoptosis in a co-culture model of human PC cell lines and human T cells [73]. miR-195 and miR-16 enhanced the radiotherapy efficacy in PC cell lines by activating the cytotoxic T cell response (repressing T cell dysfunction), inhibiting myeloid-derived suppressor cells (MDSCs) and Tregs, and increasing the secretion of pro-inflammatory cytokines (such as IFN-γ, TNF-α, and IL-2) in the tumor microenvironment through a PD-L1-dependent pathway [73].
All the above-mentioned observations referred to acinar PCs. Epigenetic methylation could regulate the JAK/STAT pathway-which is involved in PD-L1 expression-also in non-acinar PC histotypes. Indeed, the abstract of Sun et al. reported that PD-L1 expression was increased in small cell neuroendocrine PCs (in human cases, PC cell line, and mice model under IFN-γ stimulation); the block of the JAK1/STAT1 pathway inhibited PD-L1 expression, while demethylation suppressed JAK1 signaling [115].

Epigenetic Regulation of PD-L1 Expression: Studies on Human Patients, Including Data from The Cancer Genome Atlas (TCGA) Database
The epigenetic control of PD-L1 expression has also been confirmed in some studies on human PC tissues, sometimes using data derived from the TCGA database. In these studies, PD-L1 expression was evaluated on tumor tissue by real-time polymerase chain reaction (RT-PCR) analysis [14,26,60,73,90,93] and/or by immunohistochemistry [14,46,83,90,93,116] ( Table 3). Table 3. Epigenetic regulation of PD-L1: human studies on prostatic adenocarcinoma.
In the study of Xiong et al. [60], PD-L1 RNA levels were higher in PC metastases than in primary tumor specimens (n = 35), correlating with MLL3 either in their series or in cases derived from the TCGA database (p < 0.01) [60]. MLL3 and PD-L1 RNA levels positively correlated to PSA levels (not Grade Group, age, or stage) [60].
PD-L1 DNA methylation was associated with the pT stage (p < 0.001) and the Grade Group (p = 0.001).
Another study [93] found that high PD-L1 expression and aberrant CXCL12 methylation (mCXCL12) correlated to significantly shorter BRFS than either PD-L1 low /mCXCL12 normal or PD-L1 high /mCXCL12 normal cases. The aberrant mCXCL12 group included either hypoor hyper-methylated cases, which were combined in the analysis. Concordant results were found between radical prostatectomies and biopsies; unlike the mCXCL12 profile, CXCL12 immunohistochemical expression was not associated with the outcome [93].
The novel lncAMPC transcript produced by the RNF165 (RING finger protein 165) gene apparently promoted metastatic behavior and immunosuppression in PCs via LIF/LIFR stimulation. In mouse PC models, PD-L1 immunohistochemical staining positively correlated to the lncAMPC-activated LIF levels, while LIF inhibition weakened the PD-L1mediated immunosuppression in PC. The TCGA dataset analysis on human patients confirmed that PD-L1 expression was positively associated with lncAMPC-activated LIF levels and RNF165 gene transcripts [108].
The epigenetic control of miRNAs concerning PD-L1 expression has also been investigated in some studies on PC patients. In the series of Tao et al. [73] (n = 40), miR-195 and miR-16 expression inversely correlated to PD-L1, PD-1, CD80, and CTLA-4 levels, showing a potentially positive association with longer BRFS [73]. In silico analysis of GSE21032 dataset (n = 131) confirmed that high miR-195/miR-16 levels were negatively associated with PD-L1 expression: both miRNAs also correlated to longer BRFS [73]. An inverse correlation between miR-15a expression and PD-L1 mRNA has been observed in another cohort of 30 PC tissues [26].
In contrast, miR-197 and miR-200a-c positively correlated to PD-L1 mRNA levels, being inversely associated with the methylation of PD-L1 promoter in a large series [90], suggesting that mPD-L1 is correlated to decreased mRNA expression, destabilizing miR-197 and miR-200a-c. miR-570 was only associated with mPD-L1, while miR-34a inversely correlated to mPD-L1 and mRNA expression [90]. miR-513 was not differentially expressed with regard to methylation and PD-L1 mRNA expression [90].
DNA methylation (mDNA) is involved in cell differentiation and is often aberrantly deregulated in cancers, favoring PC tumorigenesis and progression [5,14,78,90,97]. mDNA analysis is a promising diagnostic tool for specimens with a limited DNA quantity and/or in the case of formalin-fixed paraffin-embedded tissue samples with degraded DNA [90,97]. However, routine tests cannot identify the heterogeneous methylation patterns of the different cell types of a specimen; instead, microdissection techniques may be helpful [90,97].
DNA promoter methylation is frequently correlated to gene silencing [5,14]. In normal tissues, about 80% of CpGs DNA sequences are methylated, while CpG islands in the promoter regions of active genes are hypomethylated. Cancers (such as PC) usually shift the mDNA pattern toward a global hypomethylation, but CpGs in promoter regions of tumor suppressor genes may undergo hypermethylation, resulting in the inhibition of gene expression and gene loss of function. Complex aberrant methylation may also occur. Moreover, genes frequently silenced in the normal human genome (such as "long interspersed nuclear element-1", LINE-1) can be re-expressed in PC cells [6][7][8][9][10][11]90,97,[162][163][164].
In some studies, mPD-L1 suppressed PD-L1 expression in PCs [7,90,93]. In a large series, despite some limits, high mPD-L1 seemed an independent prognostic biomarker for BRFS in PC patients after radical prostatectomy, being also associated with the pT stage and Grade Group [90]. DNMT1 and DNMT3b may cooperatively maintain mDNA and gene silencing in cancer cells, synergizing their biological function, improving methylation efficacy, and suppressing PD-L1 expression more efficiently than DNMT3ac alone; this hypothesis was supported by pre-clinical studies using recombinant constructs (expressing the C-terminal domains of DNMT3a and/or DNMT1 fused with a zinc finger domain specifically binding to the PD-L1 promoter) [7].
The C-X-C chemokine receptor type 4 (CXCR4) and its endogenous ligand CXCL12 are expressed in various tumors [168]. They seemed to be involved in favoring androgendependent proliferation, tumor cell motility, and metastatic growth in PC [169], co-operating with the PD-1/PD-L1 pathway to suppress anti-cancer immunity [170]. CXCR4 expression favors chemotactic cell migration toward compartments releasing high levels of CXCL12, such as bone marrow (a frequent PC-metastatic site) [171]. Constant CXCL12 production causes CXCR4a downregulation and desensitization, resulting in a resting state of tumor cells and antagonizing the metastatic process [171]. CXCL12 promoter hypermethylation downregulates CXCL12 protein expression in PC, disrupting the cellular feedback internalization of membranous CXCR4 and so favoring tumor cell motility and metastatic potential [93]. In the series of Goltz et al. [93], high PD-L1 expression and aberrant mCXCL12 were associated with significantly shorter BRFS than either PD-L1 low /mCXCL12 normal or PD-L1 high /mCXCL12 normal cases.
Epigenetic methylation may regulate the JAK/STAT pathway (involved in PD-L1 expression) also in non-acinar PCs (such as neuroendocrine carcinomas); however, limited data are available [115].
Different cancer types show aberrant expressions of HDACs, representing a promising target for cancer therapy [172]. Androgen receptor (AR) is a driver of PC progression: its downstream signaling events are closely regulated by epigenetic modifications [116]. HDAC inhibition could downregulate AR protein levels and significantly induce PD-L1 expression by increasing the acetylation of the CD274 promoter, resulting in an immuneevasive microenvironment for tumor progression [173].
p300 is a coactivator of AR, regulating its transcriptional program and signaling axis and being involved in PC recurrence and chemoresistance. p300 directly acetylates AR or binds to AR, enhancing its transcriptional activity, inducing the expression of oncogenes, and promoting tumor growth [174]. Moreover, p300 could prevent AR protein degradation [174]. p300 is also involved in PC progression through PD-L1 upregulation, favoring tumor immune escape. The transcription factor IRF-1 could recruit p300 to the CD274 promoter, inducing its transcription via histone acetylation. In an experimental study [116], the p300 inhibitor but not the anti-PD-L1 antibody significantly enhanced the efficacy of HDAC inhibitors on limiting tumor progression by blocking the HDAC inhibition-induced PD-L1 expression. Data on human patients and the TCGA dataset suggested that PD-L1 and p300 expression (unlike HDACs) negatively correlated to the Grade Group and overall survival, favoring an increase of function during cancer progression [116].
A485 (p300/CBP catalytic inhibitor) can decrease the proliferation of hematological tumors and AR-positive PCs [175]. A485 may enhance the efficacy of anti-PD-L1 antibody treatment, reducing the PD-L1 expression and exosomal secretion by PC cell lines; combined treatments inhibited the androgen-independent metastatic tumor growth in syngeneic PC models [116].
HDAC inhibitors have been approved for the treatment of T-cell lymphoma (vorinostat, SAHA; belinostat; and romidepsin) or multiple myeloma (panobinostat and LBH589). They have also been proposed in clinical trials for the treatment of solid malignancies (including PC) despite poor clinical responses in some cases [173,176].
Transcription factors usually bind to distal cis-regulatory regions (enhancers), regulating gene expression in normal conditions and during cancer development or progression [60]. MLL3 and MLL4 are huge molecular weight mono-methyltransferases of the MLL/COMPASS family; their regulation is still largely unknown. MLL3 and MLL4 favor the activity of enhancer regions (such as that of PD-L1) by the methylation of histone H3 on Lysine 4 and through the recruitment of other coactivators (such as p300). PD-L1 and MLL3 seemed positively correlated in PC patients and pre-clinical models [60].
The histone methylation regulator WDR5 is an important component of the SET1/MLL histone-methyltransferase complex and a critical co-activator of oncogenic pathways via the H3K4me3/c-MYC-dependent transcriptional activation of target genes, favoring tumor proliferation, metastases, chemoresistance, and AR-mediated castration-resistance [14]. WDR5 activates cell cycle, DNA repair, anti-apoptosis, and PD-L1 signaling, promoting PC progression; WDR5 seems an independent prognostic factor for progression-free survival and overall survival in PC. In PC cells, the IFN-γ-induced PD-L1 expression is blocked by WDR5 or MLL1 knockdown (not by c-MYC silencing) or by OICR-9429, suppressing proliferation and enhancing apoptosis, sensitivity to cisplatin, and immunotherapy [14].
EZH2 is overexpressed in PCs, contributing to tumor initiation and progression, and negatively regulating interferon-stimulated genes, including Th1-type chemokines, immune checkpoint molecules, and the major histocompatibility complex (MHC) [158]. Increased EZH2 function may favor immunosuppressive tumor microenvironments and immunotherapy resistance [158]. In PC models, EZH2 inhibition upregulated PD-L1 (dependent on STING activation) and other genes involved in antigen presentation, Th1 chemokine signaling, and interferon response. It also increased the intratumoral trafficking of activated CD8+ T cells and M1 TAMs, overall reversing the resistance to anti-PD-1 inhibitors. EZH2 actually regulates CD4+ T and Tregs differentiation [158,177,178]. In Tregs, the loss of EZH2 resulted in the degradation of FOXP3, allowing the reprogramming of Tregs to T-helper cells [179]. It did not change the total number of intratumoral Tregs, but significantly increased the intratumoral CD4+ and CD8+ T cells. EZH2 inhibition in CD8+ T cells may induce PD-1 downregulation and increase cytotoxic activity [179]. MDSCs secrete IL-23 and may activate AR signaling at least in a subset of castrationresistant PCs, representing an important component of the tumor immunosuppressive microenvironment [161]. The EZH2 inhibitor did not dramatically alter intratumoral MDSCs, while significantly reprogrammed TAMs infiltrates, decreasing tumor-promoting M2 TAMs and increasing tumor-inhibiting M1 TAMs [159,161,179]. mRNA translation is regulated by the rate of the elongation phase of protein synthesis through the eEF2K-dependent phosphorylation of eEF2. eEF2K is activated under stress conditions, while it is inhibited by the anabolic mechanistic target of the rapamycin complex 1 (mTORC1) signaling pathway. eEF2K plays a role in cancer cell survival (decreasing protein synthesis and energy/amino acids consumption during nutrient depletion) and migration, angiogenesis, and the synthesis of integrins and other proteins. eEF2K also promotes PD-L1 expression in PC [159].
Bromodomains are protein domains recognizing acetylated Lysine residues on histone tails and other nuclear proteins, promoting gene transcription. The BET protein family comprises four transcriptional coactivators of cell cycle, regulating apoptosis, migration, and invasion (BRD2, BRD3, BRD4 and the testis-specific isoform BRDT). They are frequently overexpressed in various tumors, enhancing the transcription of oncogenic drivers (such as AR and ERG) in PC. BRD4 directly associates with P-TEFb (positive transcription elongation factor b) or interacts with DNA-specific transcription factors (p53, c-MYC, AR, ERG, etc.) [180]. The inhibition of BRD4 reduces the levels of AR-driven target genes in PC, decreasing tumor burden in murine models. BRD4 also regulates immune networks, as it reduces PD-L1 expression by directly binding to the PD-L1 promoter (mediating its transcription). It also increases MHC class I expression and alters the expression of immunerelated genes. Moreover, it increases the number of tumor-infiltrating lymphocytes and susceptibility to the CD8+ T cell-mediated cytotoxicity [123,127,[180][181][182][183].
In some pre-clinical studies, treatment with JQ1 (bromodomain inhibitor) suppressed PD-L1 expression [123,127]. JQ1 stimulates the antigen presentation pathways promoted by IFN-γ. JQ1 upregulates TAF9, a subunit of the Transcription Factor IID (TFIID) required for the initiation of transcription by RNA Polymerase II. TAF9 associates with CIITA (MHC class II transactivator), forming a complex responsible for MHC class I gene upregulation after IFN-γ stimulation. JQ1 differentially modulates the expression of MHC class I alleles, and it may limit the transcription of inhibitors of RelA/NF-kB, thus enhancing their ability to bind selectively to HLA-A and B and increasing mRNA and protein levels. A combined treatment with bromodomain inhibitors and IFN-γ upregulated TRIM36 (E3 ubiquitinprotein ligase) in a dose-dependent manner. Increased TRIM36 expression was associated with the inhibition of PC proliferation and cell-cycle progression through the inhibition of the MAPK/ERK pathway [123,184]. TRIM36 is also involved in antigen processing [32].
Preclinical models documented resistance to BET inhibitors through largely unknown molecular mechanisms. Speckle-type POZ protein (SPOP) is a E3 ubiquitin ligase of the MATH-BTB family, containing MATH and C-terminal BTB domains, which are both required for BRD4 ubiquitination and degradation. In PC, SPOP or BRD4 mutations confer resistance to BET inhibitors; they inhibit the SPOP-mediated BRD4 destruction by disrupting the SPOP-BRD4 interaction, stabilizing BRD4, and leading to its cooperation with AR, ERG, and other oncogenic transcription factors [180].
In a study, miR-195 and miR-16 expression were inversely associated with PD-L1, PD-1, CD80, and CTLA-4 levels, potentially favoring a longer BRFS. They may regulate cytokine secretion in the tumor microenvironment through a PD-L1-dependent pathway and influence the PD-L1-associated apoptosis in PC cell lines [69]. In addition, miR-195 and miR-16 overexpression increases the radiosensitivity of cancer cells; in vitro and in vivo restoration of their expression enhanced radiotherapy via T cell activation in the tumor microenvironment by blocking the PD-L1 immune checkpoint in PC cells, suggesting a synergistic effect of immunotherapy and radiotherapy [73].
LncRNA transcripts may represent useful prognosticators of PC metastasis and proliferation [12]. LncAMPC (a RNF165 transcript) seemed to promote metastasis and immunosuppression in PC by stimulating LIF/LIFR expression: in TCGA datasets, PD-L1 expression positively correlated to the lncAMPC-activated LIF level and RNF165 gene transcripts [104]. Among various targets, miR-15a may bind to the 3 -UTR of PD-L1 and to the lncRNA KCNQ1OT1, located into the same sequence of miR-15a and involved in promoting oncogenic phenotypes and chemoresistance in multiple cancers (colon, lung, breast, liver, etc.). In PC cell lines, KCNQ1OT1 sponged miR-15a, suppressing the miR-15amediated inhibition of PD-L1, thus leading to PD-L1 upregulation, the inhibition of CD8+ T-cells cytotoxicity, and the promotion of tumor evasion [26]. The mechanism responsible for KCNQ1OT1 upregulation is still unknown, while the KCNQ1OT1/miR-15a/PD-L1 axis apparently promotes RAS/ERK signaling activation, inducing tumor immune evasion. ERK signaling may directly activate PD-L1 transcription, stabilizing PD-L1 mRNA as in other contexts [26,[194][195][196].
The miR-200 family inhibits EMT, cancer growth, invasion, and metastasis via the inhibition of ZEB1 and ZEB2 (transcriptional regulators of E-Cadherin). The loss of the miR-200 family through mDNA results in aggressive PC features. PDGF (involved in EMT) and other growth factors regulate miR-200 expression, while miR-200c directly influences the EGFR and TGF-β receptor signaling [202][203][204]. miR-200b seems significantly downregulated in castration-resistant PCs: miR-200 influences Notch1 expression, and together they regulate EMT progression [205]. miR-200c expression is negatively regulated by ERG in PC cells lines [206]. miR-200 overexpression can reduce PC growth and modulate chemosensitivity [207].
In a PC sudy, miR-570 was only associated with mPD-L1, while miR-34a was inversely correlated to mPD-L1 and mRNA expression. miR-513 was not differentially expressed with regard to methylation and PD-L1 mRNA expression [90]. miR-34a, miR-143, miR-148a, and the miR-200 family are involved in chemoresistance by the inhibition of apoptosis and the activation of signaling pathways. miR-34a expression levels were decreased in androgenresistant PC3 and DU145 cell lines (vs. androgen-sensitive LNPCa and normal prostatic tissue). TP53 involvement in miR-34a/miR-34c-mediated apoptosis was found in AR+ PC cells: miR-34a expression appeared completely absent in p53-null PC3 cells [208,209].
In a study, miR-424 was highly expressed in metastatic subclones of DU145 cell lines, while the effects on EMT promotion were controversial [210,211]. The PD-1/PD-L1 pathway regulates T cell activation during inflammatory processes, while CTLA-4 is a protein receptor on T cell surface, inhibiting T cell activity during the priming phase; miR-424 may inhibit the PD1/PD-L1 and CD80/CTLA-4 activity, inducing tumor suppression [46].
In PC patients, miR-424-3p expression by in situ hybridization significantly correlated to CTLA-4 (p < 0.001) and PD-L1 (p = 0.040) immunohistochemical positivity of tumor cells. Low miR-424-3p expression was significantly correlated with reduced clinical failurefree survival, aggressive PC features (high Grade Group, large tumor size, perineural and vascular invasion), and CTLA-4/PD-L1 expression on tumor cells, while no association with T cells subsets was found. CTLA-4 was associated with CD3+ and CD4+ T cells and PD-1 expression by tumor cells and stroma (considered separately or as one compartment) [46,83].
Recently, some authors reported that patients with advanced PC release fully functional circulating extracellular vesicles containing miR-424, which facilitate the acquisition of stem-like traits by low tumorigenic cells, favoring metastatic behavior and cancer progression; circulating miR-424-expressing extracellular vesicles were more frequent in patients with metastatic PCs [212][213][214].

Materials and Methods
Systematic literature reviews (SLRs) and meta-analyses have become increasingly important in health care as: (1) clinicians read SLRs to keep themselves up to date; (2) they are often a starting point for the development of clinical guidelines or further studies/trials; (3) granting agencies may require the results of SLRs to ensure the justification for research financial support. For these reasons, impacted healthcare journals frequently ask contributing authors to conduct their SLRs according to the PRISMA guidelines (http://www.prisma-statement.org/; accessed on 8 May 2021), which include an evidence-based minimum set of items for reporting and are useful for a critical evaluation of the submitted manuscripts. So, we have conducted our SLR according to these guidelines, searching in multiple databases, as previously described in the various topics/contexts in which they are applicable .
Our study aimed to answer the following PICO (population, intervention, comparison, outcomes) questions: Study selection: two independent reviewers selected the studies using a 2-step screening method. In the first step, the screening of titles and abstracts was performed to verify the eligibility/inclusion criteria and exclude irrelevant studies. In the second step, full texts of the selected articles were screened by 2 reviewers to verify the eligibility/inclusion crite-ria and to avoid duplication of the articles. Two other authors screened the reference lists to look for additional relevant publications. Finally, two authors checked the extracted data.
Objects of the systematic review: (1) to update and summarize the literature concerning the role of PD-L1 in PC cells, and (2) to report any information regarding clinicpathological features, treatment strategies, and patients' outcomes.
Data collection process/data items: data collection was study-related (authors and year of study publication) and case-related (tumor stage at presentation, Grade Group, type of specimen, treatment, test methods and results of PD-L1 expression, follow-up and outcomes, and experiment type).
Statistical analysis: the collected data were reported as continuous or categorical variables. Categorical variables were summarized by frequencies and percentages; continuous variables were summarized by ranges, mean, and median values. Time-to-recurrence was the time from the primary treatment to the disease recurrence. The survival status was the time from the primary treatment to the last follow-up.
To better present the results of our SLR and discuss the multiple interesting facets of PD-L1 expression by PC in detail, we have divided the presentation and discussion of our results into different articles, representing independent, self-sufficient chapters/parts of our work. They highlight various subtopics, including PD-L1 immunohistochemical expression in PC with a discussion of pre-analytical and interpretation variables; the clinic-pathological correlations of PD-L1 expression in PC; PD-L1 intracellular signaling pathways in PC and the influence of the tumor microenvironment; the data of pre-clinical studies (cell lines and mouse models) about the effects of experimental treatments on PD-L1 expression by PC cells; an investigation of the correlations of PD-L1 expression with the status of the mismatch repair system, BRCA, PTEN, and other main genes in PC; PD-L1 expression in liquid biopsy samples; the information of clinical trials, etc. [250][251][252][253]. We direct the readers to these papers for further details.

Conclusions
Epigenetic alterations influence oncogene/oncosuppressor gene expression without changing the DNA sequence. PC displays a complex genetic and epigenetic regulation of cell growth pathways and tumor progression.
In PC cell lines, CpG island methylation of the CD274 promoter negatively regulated PD-L1 expression. Histone modifiers are involved in the PD-L1 transcription rate: the deletion or silencing of histone modifiers (such as MLL3 and MML1) may regulate PD-L1 expression.
Epigenetic drugs could be promising in reprogramming tumor cells, reversing epigenetic modifications, and cancer immune evasion. Drugs promoting a chromatin-inactive transcriptional state (such as bromodomain or p300/CBP inhibitors) reduced PD-L1 expression, while those favoring a chromatin-active state (i.e., histone deacetylase inhibitors) increased PD-L1 expression. miRNAs can regulate PD-L1 at a post-transcriptional level. miR-195/miR-16 were negatively associated with PD-L1 expression and positively correlated to longer BRFS. They also enhanced the radiotherapy efficacy in PC cell lines. miR-197 and miR-200a-c were positively correlated to PD-L1 mRNA levels and were inversely associated with the methylation of PD-L1 promoters in a large series. miR-570, miR-34a and miR-513 may also be involved in the epigenetic regulation of PC.