Chronic lymphocytic leukemia (CLL) is the most common adult leukemia in Western countries. A critical clinical distinction is between IGHV-mutated CLL (mCLL) and IGHV-unmutated CLL (uCLL), which differ dramatically in prognosis: uCLL is more aggressive and responds differently to therapy.
The epigenome - the set of chemical modifications that control how genes are read without changing the DNA sequence itself - is broadly disrupted in CLL. Proteins that remodel chromatin (the DNA-protein complex that packages the genome) are frequently mutated in CLL, pointing to chromatin organization as a central driver of disease.
While DNA methylation in CLL has been studied extensively, the chromatin accessibility landscape - which regions of the genome are physically open and available for gene regulation - had not been mapped at scale in primary CLL patient samples before this study.
The researchers used ATAC-seq (Assay for Transposase-Accessible Chromatin with sequencing) to map regions of open chromatin across 88 CLL samples from 55 patients. ATAC-seq uses a bacterial enzyme (Tn5 transposase) that preferentially inserts sequencing adapters into nucleosome-free, accessible chromatin regions.
A key advantage of ATAC-seq over other chromatin profiling methods is that it works on small numbers of cells from clinical samples, making it compatible with routine diagnostic workflows. All 88 libraries were sequenced to an average depth of 25.4 million fragments, totaling 2.2 billion sequenced fragments across the cohort.
For a subset of 10 samples, the team also generated ChIPmentation profiles for three histone marks (H3K4me1 marking enhancers, H3K27ac marking active regulatory regions, and H3K27me3 marking repressed regions) and RNA-seq transcriptome data, enabling multi-layered integration of chromatin and gene expression information.
Integrating data across all 88 samples identified 112,298 candidate regulatory regions - genomic locations where chromatin is open in at least some CLL samples. This represents the most comprehensive chromatin accessibility reference for CLL to date, and all data are publicly available for interactive browsing.
Of these regions, about 11.6% were constitutively open across essentially all CLL samples, while 59.1% were open in a substantial fraction of samples, and 29.3% were unique to very few samples. This pattern reveals extensive regulatory heterogeneity between individual CLL patients.
The accessible regions were enriched near gene promoters and enhancers in related B cell types, indicating that the CLL chromatin landscape is broadly similar to normal B cells but with significant CLL-specific alterations. The availability of this map as a public resource enables other research groups to explore regulatory hypotheses in CLL.
Unsupervised principal component analysis of the chromatin accessibility data clearly separated samples by IGHV mutation status, identifying it as the dominant source of variation between patients. This was the first demonstration that the mCLL/uCLL distinction is reflected at the level of chromatin accessibility, not just gene expression or DNA methylation.
Many genes with known roles in CLL biology showed significant variation in chromatin accessibility between patients, including BTK, CD79A/B, NOTCH1, CD38, and KRAS. The differential chromatin accessibility at these loci may underlie the differences in gene expression and signaling that distinguish mCLL from uCLL.
Regions that were more variable among uCLL samples (581 regions) showed strong enrichment for cohesin complex binding sites (CTCF, RAD21, SMC3), while mCLL-variable regions were enriched for B-cell-specific transcription factor binding. This suggests that the two subtypes maintain chromatin heterogeneity through fundamentally different molecular mechanisms.
A random forest classifier was trained to predict IGHV mutation status (mCLL versus uCLL) from chromatin accessibility data across all 112,298 regions. Evaluated by leave-one-out cross-validation, the classifier achieved an AUC of 0.96, corresponding to 95.6% sensitivity and 88.2% specificity.
To confirm this was not overfitting, the researchers repeated the analysis 1,000 times with randomly shuffled class labels - all permuted classifiers performed near the chance level of 0.5 AUC. This rigorous negative control demonstrates that the high performance reflects genuine biological signal in the chromatin data.
The top predictive regions from the classifier were extracted, yielding 719 mCLL-specific and 764 uCLL-specific chromatin signature regions. These data-driven signatures provide a molecular fingerprint of each CLL subtype at the chromatin level, offering a potential basis for epigenome-based diagnostic classification.
The mCLL-specific chromatin signature regions were enriched for normal lymphocyte signaling pathways including CTLA4 inhibitory signaling, IgE receptor signaling, and Fc gamma receptor signaling. This aligns with the known behavior of mCLL cells, which more closely resemble post-germinal center mature B cells responding to antigen stimulation.
In contrast, uCLL-specific regions were enriched for cancer-associated pathways including NOTCH signaling and fibroblast growth factor receptor signaling - pathways associated with uncontrolled proliferation and survival. This provides a regulatory explanation for why uCLL behaves more aggressively.
The study also identified a third intermediate subgroup, iCLL (intermediate CLL), which had previously been proposed based on DNA methylation data. The ATAC-seq data confirmed this subtype exists at the chromatin level, with iCLL samples consistently showing chromatin profiles intermediate between mCLL and uCLL across all measurements including RNA-seq and histone marks.
Currently, IGHV mutation status is determined by sequencing the immunoglobulin variable region gene - a specialized test not available in all clinical settings. The demonstration that chromatin accessibility profiles can predict IGHV status with 96% AUC suggests that ATAC-seq could serve as an alternative or complementary diagnostic approach.
Beyond subtype classification, the chromatin accessibility data identify specific regulatory regions linked to prognostic genes including ADAM29, LPL, CD83, and ZBTB20 - all previously validated as clinical predictors in CLL. The chromatin landscape thus provides mechanistic insight into how these prognostic markers are regulated.
The fact that ATAC-seq works on small clinical samples and can be processed rapidly suggests compatibility with routine clinical workflows. As sequencing costs continue to fall, epigenome-based CLL diagnostics could become practical tools for guiding treatment decisions alongside existing molecular tests.
This study establishes the first large-scale chromatin accessibility reference dataset for CLL, providing a publicly available resource that the broader research community can use to investigate regulatory biology in this disease. The data and analysis tools are accessible through a dedicated interactive web portal.
By combining ATAC-seq with ChIPmentation histone profiling and RNA-seq transcriptomics, the study demonstrates that chromatin accessibility is tightly linked to gene expression and histone modification patterns, validating ATAC-seq as a reliable readout of functional gene regulation in CLL.
The methodological framework developed here - using machine learning to extract disease subtype signatures from chromatin accessibility data - is generalizable to other cancers and other clinical questions. This approach of integrating epigenome maps with clinical annotations could be applied wherever patient heterogeneity is a major clinical challenge, establishing a template for chromatin-based cancer medicine.