Epigenetically driven upregulation of immunoproteasome component PSMB8 characterizes diffuse large B-cell lymphoma (DLBCL) with inflammatory microenvironment in African ancestry patients

Abstract Diffuse large B-cell lymphoma (DLBCL) is the most prevalent and biologically heterogenous hematologic malignancy, representing approximately one-third of all non-Hodgkin lymphomas (NHL). Although standard chemoimmunotherapy achieves long-term remission in up to 65% of patients, the remaining 35% experience poor outcomes with limited therapeutic options available for relapsed or refractory disease. In addition to its genetic diversity, DLBCL is profoundly shaped by the tumor microenvironment. Transcriptomic analyses have defined four distinct lymphoma microenvironment (LME) subtypes. Among them, the inflammatory LME (IN-LME), characterized by co-enrichment of inflammatory and immunosuppressive cell populations, has been independently linked to poor clinical outcomes. One potential mechanism contributing to this LME subtype is epigenetic dysregulation. Particularly, DNA methylation changes have been implicated in reshaping immune gene expression and promoting the inflammatory and immunosuppressive features characteristic of the IN-LME subtype in DLBCL. Individuals of African ancestry (AA) with DLBCL show poorer survival (PMID: 22800091) potentially due to increased prevalence of the inflammatory LME (IN-LME), which emerging evidence suggests may be shaped by ancestry-associated tumor biology. Specifically, AA individuals are known to exhibit systemic inflammation in various disease contexts (PMID: 31633808) and display distinctive patterns of somatic mutations in lymphoma-associated genes (PMID: 32469082), features that may predispose tumors toward inflammatory microenvironments such as IN-LME. Given that ancestry influences methylation profiles in both normal and cancerous tissues, it is likely that epigenetic differences linked to AA may drive IN-LME biology in DLBCL. To explore this, we performed genome-wide DNA methylation profiling on DLBCL tumor samples (n=94 – post-QC), stratified by ancestry (self reported and genetic, AA samples (n=22) vs. european/other (n=72, using the Illumina EPIC array. Differential methylation analysis (limma, p-adj ≤ 0.05) identified 25,245 probes, including multiple hypomethylated sites within the main catalytic subunit of the immunoproteasome PSMB8 promoter region in the AA cohort. Matched RNA sequencing confirmed increased PSMB8 expression in the same samples, and this upregulation was further validated at the protein level using COMET multiplexed immunostaining in this cohort and an additional one (n=27). As a critical component of the immunoproteasome, PSMB8 facilitates antigen processing for HLA class I presentation and has been linked to enhanced response to immune checkpoint inhibitors. Given its central role in shaping the peptide repertoire, we next asked whether ancestry-associated HLA genotype variation might help explain the immune remodeling characteristic of AA DLBCL. HLA class I alleles were inferred from genomic data and their frequencies compared between AA and non-AA DLBCL patients using Fisher's exact test with FDR correction. Several alleles were significantly enriched in AA patients, including A*74:01, B*15:03 (FDR < 0.005). These alleles, which are more common in African populations (PMID: 39282263, PMID: 39098805), have not been as extensively characterized as canonical alleles like A*02:01. However, structural and ligandome studies suggest that differences at key anchor residues may influence the diversity and predictability of bound peptides (PMID: 34581761). While further immunopeptidomics analyses are needed to determine their functional impact, the overrepresentation of these alleles raises the possibility of ancestry-specific patterns of antigen presentation that could effective tumor immune clearance. In this context, we are currently evaluating whether the observed upregulation of PSMB8 represents a compensatory mechanism to enhance peptide generation in response to persistent inflammatory signaling within the tumor microenvironment, thereby reinforcing the IN-LME phenotype.