Proteogenomic Features Define Novel Subtypes of Mantle Cell Lymphoma

Abstract

Introductions Recent advancements in mass spectrometry-based proteomics have unveiled novel translational and post-translational aspects of tumor biology. The joint characterization of tumor proteomics with genomics and transcriptomics enables proteogenomic analysis, improving our understanding of the molecular mechanisms, identifying new proteome-specific markers associated with clinical outcomes, and discovering novel treatment approaches. However, proteogenomic features remain largely uncharted due to a scarcity of information on MCL proteome. Methods We conducted a proteogenomic profiling study with 4 normal and 27 MCL B cells derived from healthy donors and MCL patients, integrating whole-exon sequencing (WES), transcriptome sequencing and mass spectrometry-based proteomics with data-independent acquisition (DIA) mode. We assessed the relationship among genetic lesions, RNA and protein expression, and further classified MCL based on omics information using k-means clustering. Results To this end, we first identified 1107 differentially expressed proteins between MCL and normal B cells (FDR<0.05) (329 upregulated and 778 downregulated in MCLs). The up-regulated proteins are highly enriched for RNA splicing and DNA repair pathway while the down-regulated proteins are primarily involved in cytoskeleton and endosomal transport. When integrated with 33 recurrent genetic lesions we previously reported, we found that mutations in CCND1 and NOTCH1 had significant impact on protein expression. Key genes including ATM and BCL10 were differentially expressed between samples with and without NOTCH1 mutation. Next, we computed Pearson correlation coefficients between RNA and protein levels for each gene to define the relationships between them. RB1 (0.88), PML (0.81) and CDK2 (0.81) showed high correlation efficiency, however, the overall correlation between RNA and protein abundance was low (0.26), suggesting that proteome abundance profiles may provide unseen information not previously discovered from WES and transcriptome data. Last, we evaluated protein expression level and their associations with progress free survival and overall survival to identify proteins potentially influencing clinical outcomes. Our comprehensive analysis led us to identify 40 proteins that could play a crucial role in disease progression. To assess the prognostic value of the omics data, we performed multi-omics clustering by incorporating genomics, transcriptomics and proteomics. This enabled us to identify four distinct clusters, each characterized by key driver genetic lesions as well as RNA and protein signatures. Notably, patients in these four clusters exhibited significantly different median overall survival (C1 to C4: not reach, 28.9, 15.8 and 7.9 months; log-rank test, P<0.001). Importantly, the protein signature remained significant even when considering the MIPI stage as a covariate (P<0.001), suggesting its potential as an independent prognostic factor. We then conducted an in-depth protein feature analysis and found that cluster 1 exhibited a significant enrichment of genes involved in ubiquitin-mediated proteolysis pathway. Cluster 2 showed a remarkable upregulation in DNA replication and cell cycle pathways. C3 displayed a significant upregulation of genes associated with BCR pathway and oxidative phosphorylation. Cluster 4 had an upregulation of RNA polymerase, pyrimidine metabolism, and RNA splicing, along with a notable downregulation in the BCR pathway. Our cluster analysis not only stratified MCL patients but also uncovered critical cellular pathways that could serve as potential therapeutic targets. Conclusions In this study, we investigated proteome changes and identified protein signatures associated with overall survival in MCL. Our integrative analysis unveiled four clusters, each characterized with unique genetic features, distinct gene and protein signatures, and different clinical outcomes. This study emphasizes the significance of protein abundance data as a nonredundant layer of information in MCL biology. Moreover, we have provided a protein expression reference map for MCL, offering valuable insights for future research and clinical applications.