Regulatory network analysis identifies changes in ahr, TFAP4, and AP-1 transcription factors as top drivers of symptomatic disease progression in Waldenstrom's macroglobulinemia. Free

We previously described three subtypes of Waldenstrom's Macroglobulinemia (WM) using a multi-omic data set derived from 249 untreated patients with WM who had detectable MYD88 mutations. The B-cell like (BCL) and Plasma cell like (PCL) subtypes have distinct cell of origin but evolve from a shared transcriptional subtype (Early WM) found at early stages of disease development (Hunter et al, Manuscript in Review. Research Square, preprint 2025). We also described the WM evolutionary score (EScore) that corresponds to the progression from smoldering WM to symptomatic disease correlating with time to first therapy and WM bone marrow involvement. The transcription factor (TF) network differences that underlie these findings remain poorly understood. In this study, we examined changes in implied gene regulatory networks between WM and healthy donor memory B-cells (HDMB; CD19+CD27+), with EScore, and between WM subtypes. Findings were further analyzed for expression, alternative splicing, and TF motif enrichment in differentially regulated ATAC peaks.

Sample specific gene regulatory networks were modeled using PANDA (Glass et al. Plos One, 2013) followed by LIONESS (Kuijer et al. iScience, 2019). Together, these calculate an interaction score for each TF/gene pair for per sample. These scores were summed up by TF and by gene to model to net regulatory connectivity known as the target score (TS). LIONESS and TS data were analyzed using limma in R and top networks were derived from the results. Top TF and gene hits were analyzed further for changes in expression and isoform composition based on Gencode augmented by novel isoforms discovered in our IsoSeq long read analysis of HDMB and WM samples (Hunter et al, ASH 2024). TF motif enrichment from differentially regulated ATACSeq peaks were calculated using MEME Suite's SEA.

Network analysis of EScore TF enrichment revealed AHR and TFAP4 as top drivers. AHR is a regulator of xenobiotic metabolism which was already known to be negatively associated with EScore from our previous Hallmark gene set enrichment analysis of the expression data. Expression of AHR was negatively correlated with EScore (r = -0.73; p<0.0001) with the median expression falling by a factor of 12 from the earliest to latest EScore levels. AHR has been reported to inhibit B-cell class switching and plasma cell differentiation while upregulating the pro-inflammatory PTGS2 (cyclooxygenase-2) and anti-apoptotic BCL2L1 (BCL-XL) genes. The correlation with PTGS2 (r=0.641; p<0.0001) and BCL2L1 (r=0.266; p<0.0001) was observed in this study. Notably BCL2 was negatively correlated (r=-0.552; p<0.0001) consistent with a change in BCL2 family dependency during progression to symptomatic disease. Unlike AHR, TFAP4 was both transcriptionally upregulated and showed increased TS with EScore, consistent with its reported role as an oncogene. Most of the AP-1 transcription factor family also appeared in the top EScore network results including JUNB, FOS, FOSL1, FOSL2, ATF3, and MAF. Moreover, JUNB, FOS, FOSL2, and FOSB were all top enriched motifs in differentially closed chromatin between early and late EScore (p<0.0001 for all). Transcriptional levels of FOS and FOSL2 decreased >15x with EScore (p<0.0001 for both) while FOSB and JUNB levels were largely steady despite notable TS enrichment. Isoform analysis further revealed that while FOSB expression in WM is over 2x higher than HDMB expression, HDMB expression is primarily intron retained (ENST00000587358) while WM primarily expressed the full-length transcript (ENST00000353609). Comparing HDMB to WM Subtypes revealed strong differences in TF and gene TS with corresponding isoform dysregulation including PRDM1, PRDM4, IL15, RCAN3, CRIP2, and CD19. Despite increased plasmacytic differentiation, CD19 was expressed at only slightly lower levels than HDMB. However, HDMB primarily expressed a novel isoform skipping exon 2 while WM expressed the canonical transcript (p <0.0001). Given the important role of CD19 in BCR and TLR signaling, our findings indicate a critical dependency for CD19 in WM.

Our extensive regulatory network analysisutilizing a large multi-omic data set identified several novel therapeutic targets including TFAP4 and CD19, as well as the AHR, TFAP4, and AP-1 transcription factors as key drivers of inflammatory signaling, BCL2 family dysregulation and malignant cell growth for symptomatic disease progression in WM.