Chromatin Remodeling-Related PRDM1 Increases Stomach Cancer Proliferation and Is Counteracted by Bromodomain Inhibitor

Gastrointestinal (GI) cancers are some of the main public health threats to the world. Even though surgery, chemotherapy, and targeted therapy are available for their treatments, these approaches provide limited success in reducing mortality, making the identification of additional therapeutic targets mandatory. Chromatin remodeling in cancer has long been studied and related therapeutics are widely used, although less is known about factors with prognostic and therapeutic potential in such areas as gastrointestinal cancers. Through applying systematic bioinformatic analysis, we determined that out of 31 chromatin remodeling factors in six gastrointestinal cancers, only PR/SET domain 1 (PRDM1) showed both expression alteration and prognosis prediction. Analyses on pathways, therapies, and mediators showed that cell cycle, bromodomain inhibitor IBET151, and BET protein BRD4 were, respectively involved in PRDM1-high stomach cancer, while cell line experiments validated that PRDM1 knockdown in human stomach cancer cell line SNU-1 decreased its proliferation, BRD4 expression, and responsiveness to IBET151; accordingly, these results indicate the contribution by PRDM1 in stomach cancer formation and its association with BRD4 modulation as well as BET inhibitor treatment.


Introduction
Gastrointestinal (GI) cancers are some of the most menacing diseases globally [1,2], containing tumors that originate from the head and neck, stomach, liver, bile duct, pancreas, and colon [1,2].In the US, cancers originating from the colon and pancreas rank 3rd and 4th in both sexes [3]; liver cancer ranks 5th in males and 7th in females; while head and neck cancers rank 7th in males.In Taiwan, cancers originating from the liver, colon, head and neck, pancreas, and stomach rank 2nd, 3rd, 6th, 7th, and 8th in terms of incidence, respectively.Common risk factors for GI cancers include obesity, smoking, and alcohol consumption [1,2] while unique risk factors also exist such as gastroesophageal reflux for stomach adenocarcinoma [4].Additionally, GI cancers share similarities regarding genetic mutation and treatments [5][6][7] where mutations such as KRAS and TP53 are frequently observed among GI cancers [5].Even though surgery [8], chemotherapy [9], radiotherapy [10], and targeted therapeutic [11] treatments are available, patients still confront issues of drug resistance and tumor metastasis [12,13], ensuring that novel therapeutic targets continue to warrant further investigation [6,7].
We focus on chromatin remodeling in GI cancer as it has been recently revisited as an important disease modulator and therapeutic target by us and other researchers [14,15].For the storage of genetic information, chromatin is composed of the nucleosome, which contains DNA and histones [15,16], and for proper regulation of gene expression, nucleosomes are modulated by a chromatin remodeling complex in an ATP-dependent manner [16,17].These chromatin remodeling complexes include switch/sucrose nonfermentable (SWI/SNF), initiator of SWI (ISWI), chromodomain helicase DNA binding protein (CHD), and INO80 [16], which are frequently mutated and participate in tumorigenesis in GI cancers.
Many therapeutics for epigenetic regulation such as inhibitors for bromodomains and extra terminal domain (BET) [18] and enhancers of zeste 2 polycomb-repressive complex 2 subunit (EZH2) [19], counteract chromatin remodeling-related GI cancers.To identify whether there are additional chromatin remodeling factors in GI cancers that might predict disease progression and offer treatment opportunities, we performed systematic bioinformatic analysis on The Cancer Genome Atlas (TCGA) datasets of GI cancers in attempting to determine along with our previous report [14] of chromatin remodeling gene list whether there was any potential target for disease prediction and treatment suggestion utilizing GEPIA for such analyses on target gene expression and prognosis prediction [20], thereby finding only PR/SET domain 1 (PRDM1) was upregulated in stomach cancer while predicting its poor prognosis.
For liver cancer, Llovet et al. reviewed the risk factors including viral infection by hepatitis virus (type B, C, or D), alcohol, nonalcoholic steatohepatitis, age, and gender [37], determining that liver cancer could also be classified into different molecular subclasses including progenitor, macro-trabecular massive, steato-hepatic, and cholestatic modalities.Gene mutation in the progenitor subclass is in axin 1 (AXIN1); for macro-trabecular massive subclass is in TSC complex subunit 1 or 2 (TSC1, TSC2) and the cholestatic subclass is in catenin beta 1 (CTNNB1).Mutations in the promoter of telomerase reverse-transcriptase (TERT) or gene body of TP53 span multiple subclasses [37].Jia et al. reported that with previously identified 870 chromatin regulators, they developed a chromatin regulator-based prognostic risk score model that predicts survival status and associates with immunity as well as drug sensitivity in liver cancer [38].
Li et al. reported that PRDM1 induces cancer immune evasion via ubiquitin-specific peptidase 22 (USP22), Spi-1 proto-oncogene (SPI1), and programmed death ligand 1 (PDL1) while also continuing their previous works on PRDM1 resultantly finding it increased PDL1 expression in liver cancer.With tandem mass tag-based quantitative proteomic analysis, they found SPI1 was a modulator linking PRDM1 and PDL1, and as PRDM1 did not affect SPI1 mRNA expression, the authors explored its effect on protein expression and proteolysis, determining via transfection of ubiquitin-carrying plasmid that PRDM1 decreased SPI1 proteolysis.With the deubiquitinating enzyme (DUB) siRNA library, they found USP22 was the DUB that was affected by PRDM1 and then in turn affected SPI1 out of 98 DUBs.Ultimately, they applied single-cell RNA sequencing on liver cancer specimens and validated the relationship between PRDM1 expression and liver cancer immunity as well as treatment response [39].
For pancreatic cancer, Dr. Klein reviewed the risk factors including tobacco use, diabetes mellitus, obesity, alcohol consumption, pancreatitis, allergies, and familial history of cancer [43] while also determining that typical tumor suppressors for this cancer type include BRCA2 DNA repair-associated (BRCA2), BRCA1 DNA repair-associated (BRCA1), and CDKN2A [43].Hayashi et al. further illustrated that lysine methyltransferase 2C (KMT2C), AT-rich interaction domain 1A (ARID1A), GATA binding protein 6 (GATA6), TP53, SMAD4, KRAS proto-oncogene, and GTPase (KRAS) are all involved in pancreatic cancer formation [44], while Kawakubo et al. revealed that the epigenetic regulation of pancreatic cancer could affect its response to immunotherapy, including the combined applications of inhibitors against DNMT, HDAC, BET, and EZH2 [45].Chiou et al. applied a mouse model for pancreatic cancer and determined that in a highly metastatic subpopulation, PRDM1 is essential to maintain this phenotype.PRDM1 is regulated by hypoxia-inducible factor 1 subunit alpha (HIF1A), and the hypoxia-response elementcontaining region 240 kb upstream of the PRDM1 transcription initiation site is responsible for such regulation [28].
For colon cancer, Keum et al. reviewed the risk factors including obesity, dietary patterns, alcohol use, and tobacco use [46].Li et al. determined that typical oncogenes and tumor suppressors for this cancer type included APC, KRAS, and TP53 [47], while Jung et al. discovered how DNA methylation and histone modifications and their inhibitors contributed to colon cancer formation and treatment [48].For PRDM1 in colon cancer, Kim et al. reported that it promotes chemoresistance [25], Liu et al. reported that it inhibits proliferation [27], and Wan et al. reported that it induces cell cycle arrest [49].
With the above reports and reviews, the importance of GI cancers regarding development and treatment is well-documented, contributing to chromatin remodeling-related events in such circumstances.On the contrary, the role of PRDM1 in GI cancers has been less reported and our systematic bioinformatic analysis showed that its high expression in stomach cancer further predicted poor prognosis.This result inspired us to dissect the therapeutic and pathway modalities for PRDM1 in stomach cancer and validate them via wet-lab analysis.

GEPIA Analysis
To investigate the roles of chromatin remodeling-related factors in GI cancers, we applied such gene lists from our previous report [14] and analyzed their expression alterations and prognostic predictions in GEPIA [20].Default settings for expressions as (1) |log 2 FC| > 1, (2) p-value < 0.01, and (3) log-scale as log 2 (TPM + 1) were applied along with those for prognosis as (1) overall survival and (2) median group cutoff.As PRDM1 in stomach cancer was the only chromatin remodeling-related factor displaying both expression alteration and prognosis prediction, its enriched pathways and therapeutics were further explored.

cBioPortal Analysis
The coexpressed genes for PRDM1-high stomach cancer in TCGA were extracted from cBio-Portal [50], and the top five hundred positively or negatively correlated genes were uploaded to Reactome and L1000CDS 2 for analyses on enriched pathways and therapeutics, respectively.

L1000CDS 2 Analysis
The L1000CDS 2 gene-drug interaction search engine [51] is based on the Connectivity Map and L1000 [52] to accelerate potential therapeutic identification.The above coexpression signature was uploaded and therapeutics having opposite signatures were identified for potential counteraction.As epigenetic inhibitors were enriched, especially those against BET, the most enriched BET inhibitor IBET151 was selected to test its effect on PRDM1-high stomach cancer.

RNA Interference
Cells were seeded in 6-well plates (92006, TPP; Zollstrasse, Switzerland) or 2-well chamber slides (154461, NUNC/Thermo Fisher Scientific), and transfected with 1 µg shRNA plasmid together with 3 µL HyFect (LDG0001RA, LEADGENE; Tainan, Taiwan).Two-day post-transfection cells were analyzed for target gene expression and proliferation.shRNAs were from RNA Technology Platform and Gene Manipulation Core Facility (RNAi core), Academia Sinica (Taipei, Taiwan) and the sequence for luciferase control was GCGGTTGCCAAGAGGTTCCAT, with the sequence for shPRDM1 being CATCTACTTCTACACCATTAA.

Cell Proliferation Assay
Cell proliferation was assayed by trypan blue exclusion as previously described [53].A total of 200,000 SNU-1 cells were seeded in 6 wells and transfected with 1 µg indicated plasmid as well as 3 µL transfection reagent HyFect for two days.Transfection efficiency was confirmed by immunofluorescence and polymerase chain reaction (Supplementary Method).Cells were also subjected to trypan blue exclusion assay for proliferation analysis.

Immunofluorescence
As PRDM1 knockdown significantly decreased the proliferation of SNU-1 and rendered protein harvest unsuccessful even with triplicated cell number, we utilized immunofluo-rescence to analyze knockdown efficiency and BRD4 expression change.Two-day posttransfection cells were washed with PBS and fixed with 10% formalin for 10 min at room temperature, then, primary antibody was added at 1:100 dilution in 0.2% BSA-PBS wash buffer onto cells for overnight incubation at 4 • C.After washing with the buffer, cells were stained with fluorescent secondary antibody (GTX213110-04, GeneTex; Hsinchu, Taiwan) at 1:500 dilution for 1 h and DAPI (D9542, Sigma) at 5 µg/mL for 5 min at room temperature.Following washing with ddH 2 O, slides were mounted and the signals were analyzed using fluorescent microscopy (BX53, Olympus; Tokyo, Japan) and ImageJ (version 1.54f).Antibodies against PRDM1 (GTX132087) and BRD4 (GTX130586) were from GeneTex.

BET Inhibitor Treatment
IBET151 (HY-13235, MedChemExpress, Monmouth Junction, NJ, USA) at the lowest predicted concentration of 1 µM [54] was administrated onto cells and the effect of this inhibitor on PRDM1-high stomach cancer was assayed regarding proliferation.

Statistical Analysis
Statistical analyses were performed with GraphPad (Boston, MA, USA).The statistical difference between the control and experimental groups was analyzed with a t-test.p < 0.05 was considered statistically significant.

PRDM1 Was Increased in Stomach Cancer and Predicted Poor Prognosis
To identify the roles of the chromatin remodeling-related factors in GI cancers in a systematic manner, GEPIA [20] was applied to assay expression alterations and prognosis predictions in TCGA datasets.These factors are mainly from families including DNA methyltransferase (DNMT), histone deacetylase (HDAC), PRDM, and protein arginine methyltransferase (PRMT).DNMTs control the methylation of DNA [55], and HDACs regulate the deacetylation of protein [56].PRDMs manipulate gene transcription [57] while PRMTs control methylation on arginine [58].Both histone and non-histone proteins are targets for HDACs [59] and PRMTs [60].
According to our previous report [14], the following factors are included in this analysis: DNMT1/2/3A/3B/3L, HDAC1/2/3, PRDM1/2/4/5/6/7/8/9/10/11/12/13/14/15/16, and PRMT1/2/3/5/6/7/8/10.As shown in Table 1, some of the chromatin remodelingrelated factors were dysregulated during GI cancer formation.For DNMT, member 1 was upregulated in head and neck squamous cell carcinomas (HNSC) and pancreatic adenocarcinoma (PAAD), and member 3B was increased in HNSC and esophageal carcinomas (ESCA).In the HDAC family, member 1 was upregulated in PAAD, while member 2 showed increment additionally in rectum adenocarcinoma (READ) and stomach adenocarcinoma (STAD).For PRDMs, there were potential oncogenes and tumor suppressors in GI cancers, such as that member 1 was upregulated in PAAD, READ, and STAD while member 8 was increased in PAAD.On the contrary, PRDM8 as well as PRDM6 were downregulated in COAD and READ.READ is a cancer type with frequent PRDM dysregulations, as PRDM1 upregulation and PRDM6/8/11 downregulation were all observed.For PRMTs, members 1, 2, 5, and 7 were increased in PAAD, while member 3 was augmented in COAD and READ.Nevertheless, with all the expression alterations mentioned above, only PRDM1 in STAD predicted prognosis and was in accordance with its increment in this cancer type (Figure 1 and Figures A1-A12), so this result induced us to focus on PRDM1 in STAD as an important pair in chromatin remodeling-related factors in GI cancers.PRDM1 was increased in stomach cancer and predicted a poor prognosis.Chromatin remodeling-related factors identified from our previous report were analyzed with GEPIA for expression alterations and prognosis predictions across GI cancers in TCGA datasets.Among 31 candidates and six cancer types, only PRDM1 in stomach cancer showed both expression alteration (A) and prognosis prediction (B).In (A), those who fulfilled the criteria as |log 2 FC| > 1 and q value < 0.01 are emphasized with green color for downregulation and red color for upregulation.

PRDM1-High Stomach Cancer Was Enriched for Chromatin-Related Pathways and Was Targetable by BET Inhibitor In Silico
To further investigate the pathway and therapeutic strategies for PRDM1-high stomach cancer, we uploaded the above-mentioned co-expression signature to Reactome [61] and L1000CDS 2 , respectively.Reactome is a well-established pathway exploration database in cancer research [61], and L1000CDS 2 [51] is based on a series of the Connectivity Map and L1000 [52] for signature-based therapeutic identification.We started with L1000CDS 2 analysis in order to link back to the potential pathway once a candidate therapeutic was identified.While a few inhibitors possibly counteracting PRDM1-high stomach cancer were discovered, the most frequent ones were those for epigenetics (Figure 2A), with the targets of these epigenetic inhibitors all pointing to BET protein (Figure 2B).These BET inhibitors that hampered the function of BET proteins such as BRD2, BRD3, and BRD4 [62] were initially identified in 2008.These proteins utilize their BET domains to regulate gene expression in development and disease [63], and their expressions are frequently dysregulated in the latter [64].We wished to further investigate whether expressions of BET proteins were associated with that of PRDM1 in stomach cancer in the TCGA dataset and Cancer Cell Line Encyclopedia (CCLE), and such analysis with cBioPortal showed that BRD4 was positively associated with PRDM1 (Figure 2C), while BRD2 and BRD3 were not; additionally, PRDM1-high stomach cancer cell lines tended to express more BRD4 (Figure 2D).These cell lines [65,66] were selected due to their domestic availability so that the above observation could be validated.In the clinical aspect, BRD4 was also expressed at higher levels in the cancer portion rather than in the normal stomach (Figure 2E).Regarding pathway analysis, the presence of BET inhibitors as therapeutic for PRDM1-high stomach cancer allowed confirmation as to whether such pathways were enriched in the above coexpression signature; consequently, we uploaded this signature to Reactome [61] and found pathways including chromatin remodeling as well as epigenetic regulation were enriched in PRDM1-high stomach cancer (Figure 3).The figures were extracted from Reactome under the criteria of "Voronoi visualization" and "flattened view" [61,67].As in Figure 3A and reference [68], WD repeat domain 5 (WDR5) is reported to participate in BRD4-regulated gene expression.The former factor affects histone trimethylation and the latter one is associated with histone acetylation [68].The involvements of the chromatin remodeling-related pathway, BET inhibitor, and BRD4 expression in PRDM1-high stomach cancer galvanized us to test whether these associations could be validated in wet-lab assembly using SNU-1: the human stomach cancer cell line with high PRDM1 expression and level 1 biosafety.
Figure 2. PRDM1-high stomach cancer was targeted by a BET inhibitor and was associated with BRD4 expression in silico.The therapeutic values for PRDM1-high stomach cancer were identified with the above co-expression signature and L1000CDS 2 .Identified therapeutics were divided according to their targets (A).In (A), the epigenetic therapeutics were BET inhibitors I-BET, I-BET151, and PFI-1 (B).The potential mediator for PRDM1 in stomach cancer was identified as BRD4 in cBi-oPortal (C) and CCLE (D), while the expression pattern of BRD4 in stomach cancer was identified with GEPIA (E).*, p < 0.01.Reactome was applied to identify enriched pathways for PRDM1-high stomach cancer, and the results were overlapped with the therapeutic analysis mentioned in Figure 2 and shown as "cell cycle-chromosome maintenance" (A) and "gene expression (transcription)-epigenetic regulation of gene expression" (B).

PRDM1 Knockdown Decreased Cell Proliferation, BRD4 Expression, and IBET151 Sensitivity in Stomach Cancer
Following bioinformatic analysis, we performed cell-line experiments to confirm the effects of PRDM1 on proliferation, BRD4 expression, and response to IBET151 in stomach cancer.Five shRNAs were screened and clone 71 was found to be effective in reducing PRDM1 expression in SNU-1 (Figures 4A and A14).This shRNA was then utilized for further experimentation revealing that it greatly decreased expression of BRD4 in SNU-1 (Figure 4B), which was in accordance with the positive association between PRDM1 and BRD4 in TCGA and CCLE (Figure 3).shPRDM1 also decreased SNU-1 proliferation (Figure 4C), which echoed the association between PRDM1 expression and cell cycle progression in stomach cancer mentioned in Figure 2. We next tried whether RNA extraction would yield enough material to analyze the effect of shPRDM1 on the expressions of PRDM1 and BRD4, and indeed found RNA extraction was successful and shPRDM1 decreased the expression of not only itself but also that of BRD4 (Figure A14).When analyzing the band intensity of the product of polymerase chain reaction with ImageJ (version 1.54f) [69], the signals of PRDM1 and BRD4 were weaker in the group of shPRDM1.Whether BRD4 loss resulted from PRDM1 knockdown rendering stomach cancer insensitive toward IBET151 was of concern, so further testing found shPRDM1-SNU-1 displayed decreased sensitivity toward IBET151 while this inhibitor showed an obvious suppression on shLuc-SNU-1 in terms of proliferation (Figure 4D).PRDM1 expression thus indicated the potential effectiveness of BET inhibitors in stomach cancer treatment, as similar therapeutics have entered clinical trials [18].
PRDM1 expression in SNU-1 (Figures 4A and A14).This shRNA was then utilized for further experimentation revealing that it greatly decreased expression of BRD4 in SNU-1 (Figure 4B), which was in accordance with the positive association between PRDM1 and BRD4 in TCGA and CCLE (Figure 3).shPRDM1 also decreased SNU-1 proliferation (Figure 4C), which echoed the association between PRDM1 expression and cell cycle progression in stomach cancer mentioned in Figure 2. We next tried whether RNA extraction would yield enough material to analyze the effect of shPRDM1 on the expressions of PRDM1 and BRD4, and indeed found RNA extraction was successful and shPRDM1 decreased the expression of not only itself but also that of BRD4 (Figure A14).When analyzing the band intensity of the product of polymerase chain reaction with ImageJ (version 1.54f) [69], the signals of PRDM1 and BRD4 were weaker in the group of shPRDM1.Whether BRD4 loss resulted from PRDM1 knockdown rendering stomach cancer insensitive toward IBET151 was of concern, so further testing found shPRDM1-SNU-1 displayed decreased sensitivity toward IBET151 while this inhibitor showed an obvious suppression on shLuc-SNU-1 in terms of proliferation (Figure 4D).PRDM1 expression thus indicated the potential effectiveness of BET inhibitors in stomach cancer treatment, as similar therapeutics have entered clinical trials [18].

Discussion
In the present study, we identified chromatin remodeling-related PRDM1 as a contributor to stomach cancer formation via modulations on cell proliferation and BRD4 expression; specifically, via systematic bioinformatic analysis on the chromatin remodeling-related gene list suggested by our team [14].Besides, chromatin remodeling-related AT-rich interaction domain 1A (ARID1A) has been widely reported in stomach cancer formation.Lu et al. in 2023 mentioned that multiple ARID1A-related stomach cancer clinical trials were underway and that synthetic lethality combination for ARID1A-deficient stomach cancer might be inhibitors against PARP, PI3K, EZH2, and PD-L1 [70].Yan et al. in 2014 found that ARID1A inhibited stomach cancer invasion by increasing β-catenin membrane translocation and E-cadherin transcription [71].For other chromatin remodeling-related factors, Neil et al. in 2023 reported that SWI/SNF-related, matrix-associated, and actindependent regulator of chromatin, subfamily a, member 4 (SMARCA4) was mutated as truncated, unperceived, or mis-sensed in cancers of esophagus and stomach [72].Liu et al. in 2023 reported that SWI/SNF-related, matrix-associated, and actin-dependent regulator of chromatin subfamily c member 1 (SMARCC1) was a poor prognosis predictor for overall and disease-free survival for stomach cancer and was further associated with invasion, lymph node involvement, and stage [73].Hashimoto et al. in 2020 reported that chromodomain helicase DNA-binding protein 5 (CHD5) in stomach cancer was associated with pathological N status and good prognosis in both overall and recurrence-free survival [74] while also revealing that CHD5 overexpression decreased stomach cancer proliferation and invasion [74].Our focus on PRDM1 in stomach cancer via systematic bioinformatic analysis and in vitro validation could add additional clues to chromatin remodeling factor-regulated stomach cancer appraisal.
For BET inhibitors in the treatment of gastrointestinal cancers, PRDM1 in stomach cancer is one of the potential targets, and Sun et al. in 2022 reviewed the effects of BET inhibitors on gastrointestinal cancers and their advancements in clinical trials [18], while Montenegro et al. in 2016 reported that an isoxazole PNZ5 might be a potential BET inhibitor and it increased stomach cancer apoptosis even in 3D spheroid [75].
To sum up, the present study utilized bioinformatic analysis and in vitro validation to show that PRDM1 increased stomach cancer formation by modulating cell proliferation and BRD4 expression but was counteracted by a BET inhibitor, which provides additional clues to stomach cancer research and treatment.

Conclusions
Summarizing the present study from systematic bioinformatic analysis for chromatin remodeling-related factors in GI cancers, we found PRDM1 in stomach cancer was the only pair showing expression alteration and prognosis prediction out of 31 candidate genes and six GI cancer types.Further therapeutic and pathway analyses revealed that PRDM1-high stomach cancer might be targeted by BET inhibitor and was enriched for chromatin remodeling-related features.Applying databases of cBioPortal and CCLE for proper in vitro validation, we found human stomach cancer cell lines tended to have greater BRD4 expressions once their PRDM1 expressions were higher, with this association also being observed in the TCGA STAD dataset.Cell-line experimentation indeed validated that PRDM1 knockdown in human stomach cancer cell line SNU-1 decreased BRD4 expression, proliferation, and sensitivity to BET inhibitor.
For basic research, we intend to perform promoter assay, chromatin immunoprecipitation, and RNA sequencing for shPRDM1-SNU-1.With promoter assay utilizing BRD4 promoter and shPRDM1-SNU-1, we can identify whether PRDM1 activates BRD4 promoter; with chromatin immunoprecipitation utilizing shPRDM1-SNU-1, we can identify whether PRDM1 binds to BRD4 promoter and whether this binding is lost after PRDM1 knockdown; and with RNA sequencing, we can identify how loss of PRDM1 in stomach cancer changes transcriptome, and with these clues, the source behind epigenetic regulation can be explored.
For clinical research, in the NCBI gene expression omnibus database [76], Cui et al. performed transcriptome analysis on 80 pairs of normal and cancerous stomach tissues with microarray [77], and from this result, we additionally identified that PRDM1 was upregulated in cancerous tissues as shown in Figure A13.This demonstrates that PRDM1 is detectable in stomach cancer in another independent cohort, and with up-to-date RNA sequencing on biopsy [78] or even liquid biopsy [79], the expression of PRDM1 could serve as a factor in the prediction of stomach cancer development and BET inhibitor treatment.
We additionally validated the positive correlation between PRDM1 and CD274 (programmed death ligand 1) in stomach cancer in the TCGA dataset (Figure A15).We also additionally reviewed the importance of chromatin remodeling in immunotherapy for GI cancers to further emphasize its clinical potential.
For head and neck cancer, Brennan et al. followed up their previous work and reported that nuclear receptor binding SET domain protein 1 (NSD1) was an indicator for the immunologically cold and DNA hypomethylated subtype of this cancer [80].The inactivation of this transcriptional regulator led to decreased infiltration of T cells even in a mouse model, as the authors applied NOD-scid IL2Rgamma null (NSG) mouse to establish tumor of control or NSD1 shRNA and human peripheral blood mononuclear cell (PBMC) injection.
For liver cancer, Shen et al. reviewed the potential enhancement by inhibitors against histone deacetylases on the efficiency of immunotherapy.These inhibitors included vorinostat and sodium valproate, which increased the expressions of PD-L1 and MHC class I polypeptide-related sequence B (MICB; for activation of natural killer cell), respectively [81].On the contrary, Tao et al. reviewed how epigenetic regulation affected resistance to immunotherapy and addressed the molecular mechanisms within [82].They emphasized that in liver cancer the inhibitors against DNA methyltransferase increased interferons and activated T cells, while the inhibitors against histone methyltransferase increased UL16 binding protein 1 (ULBP1) and activated natural killer cells.Chen et al. reported that chromatin organization-related gene signature predicted response to immunotherapy [83].Wu et al. analyzed bioinformatically that there was a relation between genes for epigenetics and inflammation [84].Cai et al. reported that the epigenetic regulator SWI/SNF-related, matrix-associated, and actin-dependent regulator of chromatin subfamily c member 1 (SMARCC1) was associated with decreased cytotoxic T cell and increased programmed cell death 1 (PDCD1, PD-1) [85].
For stomach cancer, Lin et al. analyzed bioinformatically the landscape of histone deacetylases and established the histone deacetylase score (HDS) model to predict immunotherapy response [86].Yuan et al. reported that histone modification was involved in the stomach cancer subtype of stroma activation, which had a poor prognosis and immunotherapy resistance [87].This epigenetic-modification-dysregulated (EMD) subtype employed mutations in chromatin regulators of the families of lysine methyltransferase, lysine demethylase, and histone deacetylase.Gu et al. reported that mutation in chromatin regulator AT-rich interaction domain 1A (ARID1A) was enriched for PD-1 signaling and showed a superior response to PD-1 blockade [88].Wang et al. reported that mutation in lysine methyltransferase 2 was associated with PD-L1 positivity, cytotoxic lymphocyte, and efficiency of immune checkpoint inhibitor treatment [89].
For pancreatic cancer, Li et al. analyzed bioinformatically that lysine demethylase 5B (KDM5B) was increased in the tumor part and contributed to immunologically cold tumor microenvironment in respect of CD8 + T cell and interferon γ and validated with a subcutaneous mouse model [90].Zhou et al. reported that histone deacetylase 5 (HDAC5) suppressed PD-L1 expression via deacetylation of lysine 310 residue of p65, with this deacetylation subsequently inhibited NF-κB activation as well as PD-L1 induction [91].HDAC5 inhibition thus sensitized pancreatic cancer to PD-1 blockade [91].

Figure 1 .Figure 1 .
Figure 1.PRDM1 was increased in stomach cancer and predicted a poor prognosis.Chromatin remodeling-related factors identified from our previous report were analyzed with GEPIA forFigure 1.PRDM1 was increased in stomach cancer and predicted a poor prognosis.Chromatin remodeling-related factors identified from our previous report were analyzed with GEPIA for expression alterations and prognosis predictions across GI cancers in TCGA datasets.Among 31 candidates and six cancer types, only PRDM1 in stomach cancer showed both expression alteration (A) and prognosis prediction (B).In (A), those who fulfilled the criteria as |log 2 FC| > 1 and q value < 0.01 are emphasized with green color for downregulation and red color for upregulation.

Figure 2 .
Figure2.PRDM1-high stomach cancer was targeted by a BET inhibitor and was associated with BRD4 expression in silico.The therapeutic values for PRDM1-high stomach cancer were identified with the above co-expression signature and L1000CDS 2 .Identified therapeutics were divided according to their targets (A).In (A), the epigenetic therapeutics were BET inhibitors I-BET, I-BET151, and PFI-1 (B).The potential mediator for PRDM1 in stomach cancer was identified as BRD4 in cBi-oPortal (C) and CCLE (D), while the expression pattern of BRD4 in stomach cancer was identified with GEPIA (E).*, p < 0.01.

Figure 2 .
Figure2.PRDM1-high stomach cancer was targeted by a BET inhibitor and was associated with BRD4 expression in silico.The therapeutic values for PRDM1-high stomach cancer were identified with the above co-expression signature and L1000CDS 2 .Identified therapeutics were divided according to their targets (A).In (A), the epigenetic therapeutics were BET inhibitors I-BET, I-BET151, and PFI-1 (B).The potential mediator for PRDM1 in stomach cancer was identified as BRD4 in cBioPortal (C) and CCLE (D), while the expression pattern of BRD4 in stomach cancer was identified with GEPIA (E).*, p < 0.01.

Figure 3 .
Figure 3. PRDM1-high stomach cancer was enriched for chromatin remodeling-related pathways.Reactome was applied to identify enriched pathways for PRDM1-high stomach cancer, and the Figure 3. PRDM1-high stomach cancer was enriched for chromatin remodeling-related pathways.Reactome was applied to identify enriched pathways for PRDM1-high stomach cancer, and the results were overlapped with the therapeutic analysis mentioned in Figure2and shown as "cell cycle-chromosome maintenance" (A) and "gene expression (transcription)-epigenetic regulation of gene expression" (B).

Figure 4 .
Figure 4. PRDM1 knockdown in stomach cancer decreased its BRD4 expression, proliferation, and response to BET inhibitor.Human stomach cancer cell line SNU-1 was transfected with shPRDM1 or shLuc control (A), and the effects on BRD4 expression (B), proliferation (C), and response to BET inhibitor (D) were analyzed two days later for post-transfection.*, p< 0.05.

Figure 4 .
Figure 4. PRDM1 knockdown in stomach cancer decreased its BRD4 expression, proliferation, and response to BET inhibitor.Human stomach cancer cell line SNU-1 was transfected with shPRDM1 or shLuc control (A), and the effects on BRD4 expression (B), proliferation (C), and response to BET inhibitor (D) were analyzed two days later for post-transfection.*, p< 0.05.

Figure A2 .
Figure A2.DNMT3B was increased in ESCA and HNSC.The expressions of DNMT3B from TCGA datasets were extracted from the GEPIA database.

Figure A2 .
Figure A2.DNMT3B was increased in ESCA and HNSC.The expressions of DNMT3B from TCGA datasets were extracted from the GEPIA database.

Figure A2 .
Figure A2.DNMT3B was increased in ESCA and HNSC.The expressions of DNMT3B from TCGA datasets were extracted from the GEPIA database.

Figure A3 .
Figure A3.HDAC1 was increased in PAAD.The expressions of HDAC1 from TCGA datasets were extracted from the GEPIA database.

Figure A4 .
Figure A4.HDAC2 was increased in PAAD, READ, and STAD.The expressions of HDAC2 from TCGA datasets were extracted from the GEPIA database.

Figure A4 .
HDAC2 was increased in PAAD, READ, and STAD.The expressions of HDAC2 from TCGA datasets were extracted from the GEPIA database.

Figure A5 .
Figure A5.PRDM6 was decreased in COAD and READ.The expressions of PRDM6 from TCGA datasets were extracted from the GEPIA database.

Figure A5 .
Figure A5.PRDM6 was decreased in COAD and READ.The expressions of PRDM6 from TCGA datasets were extracted from the GEPIA database.

Figure A6 .
Figure A6.PRDM8 was altered in COAD, PAAD, and READ.The expressions of PRDM8 from TCGA datasets were extracted from the GEPIA database.

Figure A6 .
Figure A6.PRDM8 was altered in COAD, PAAD, and READ.The expressions of PRDM8 from TCGA datasets were extracted from the GEPIA database.

Figure A6 .
PRDM8 was altered in COAD, PAAD, and READ.The expressions of PRDM8 from TCGA datasets were extracted from the GEPIA database.

Figure A7 .
Figure A7.PRDM11 was decreased in READ.The expressions of PRDM11 from TCGA datasets were extracted from the GEPIA database.

Figure A7 . 22 Figure A8 .
Figure A7.PRDM11 was decreased in READ.The expressions of PRDM11 from TCGA datasets were extracted from the GEPIA database.J. Pers.Med.2024, 14, x FOR PEER REVIEW 16 of 22

Figure A8 .
Figure A8.PRMT1 was increased in PAAD.The expressions of PRMT1 from TCGA datasets were extracted from the GEPIA database.

Figure A8 .
PRMT1 was increased in PAAD.The expressions of PRMT1 from TCGA datasets were extracted from the GEPIA database.

Figure A9 .
Figure A9.PRMT2 was increased in PAAD.The expressions of PRMT2 from TCGA datasets were extracted from the GEPIA database.

Figure A9 .
Figure A9.PRMT2 was increased in PAAD.The expressions of PRMT2 from TCGA datasets were extracted from the GEPIA database.

Figure A10 .
Figure A10.PRMT3 was increased in COAD and READ.The expressions of PRMT3 from TCGA datasets were extracted from the GEPIA database.

Figure A10 .
Figure A10.PRMT3 was increased in COAD and READ.The expressions of PRMT3 from TCGA datasets were extracted from the GEPIA database.

Figure A10 .
PRMT3 was increased in COAD and READ.The expressions of PRMT3 from TCGA datasets were extracted from the GEPIA database.

Figure A11 .
Figure A11.PRMT5 was increased in PAAD.The expressions of PRMT5 from TCGA datasets were extracted from the GEPIA database.

Figure A12 .
Figure A12.PRMT7 was increased in PAAD.The expressions of PRMT7 from TCGA datasets were extracted from the GEPIA database.

Figure A12 .
Figure A12.PRMT7 was increased in PAAD.The expressions of PRMT7 from TCGA datasets were extracted from the GEPIA database.

Figure A13 .
Figure A13.PRDM1 upregulation was observed in microarray GSE27342.For potential clinical application, we searched the NCBI gene expression omnibus database for systematic transcriptomic analysis on stomach cancer clinical specimens and found in GSE27342 Cui et al. analyzed 80 pairs of normal and cancerous stomach tissues, and we additionally found PRDM1 was upregulated in cancerous tissues in this dataset independent of TCGA STAD dataset.

Figure A13 .
Figure A13.PRDM1 upregulation was observed in microarray GSE27342.For potential clinical application, we searched the NCBI gene expression omnibus database for systematic transcriptomic analysis on stomach cancer clinical specimens and found in GSE27342 Cui et al. analyzed 80 pairs of normal and cancerous stomach tissues, and we additionally found PRDM1 was upregulated in cancerous tissues in this dataset independent of TCGA STAD dataset.

Figure A14 .
Figure A14.The effect of shPRDM1 on BRD4 expression at mRNA level.SNU-1 of shLuc or shPRDM1 clones was subjected to RNA extraction, and the effects of PRDM1 knockdown on expressions of PRDM1 and BRD4 were analyzed with polymerase chain reaction and gel electrophoresis.The signal intensities were analyzed and quantified with ImageJ (version 1.54f).

Figure A14 .
Figure A14.The effect of shPRDM1 on BRD4 expression at mRNA level.SNU-1 of shLuc or shPRDM1 clones was subjected to RNA extraction, and the effects of PRDM1 knockdown on expressions of PRDM1 and BRD4 were analyzed with polymerase chain reaction and gel electrophoresis.The signal intensities were analyzed and quantified with ImageJ (version 1.54f).

Figure A14 .
Figure A14.The effect of shPRDM1 on BRD4 expression at mRNA level.SNU-1 of shLuc or shPRDM1 clones was subjected to RNA extraction, and the effects of PRDM1 knockdown on expressions of PRDM1 and BRD4 were analyzed with polymerase chain reaction and gel electrophoresis.The signal intensities were analyzed and quantified with ImageJ (version 1.54f).

Table 1 .
Dysregulated chromatin remodeling-related factors in GI cancers..Pers.Med.2024,14,xFOR PEER REVIEW 6 of 22above, only PRDM1 in STAD predicted prognosis and was in accordance with its increment in this cancer type (Figures1 and A1-A12), so this result induced us to focus on PRDM1 in STAD as an important pair in chromatin remodeling-related factors in GI cancers.

Table 1 .
Dysregulated chromatin remodeling-related factors in GI cancers.