Skip to main page content
Access keys NCBI Homepage MyNCBI Homepage Main Content Main Navigation
. 2020 Jan 29;11(1):577.
doi: 10.1038/s41467-019-14081-6.

Chromatin Mapping and Single-Cell Immune Profiling Define the Temporal Dynamics of Ibrutinib Response in CLL

Free PMC article

Chromatin Mapping and Single-Cell Immune Profiling Define the Temporal Dynamics of Ibrutinib Response in CLL

André F Rendeiro et al. Nat Commun. .
Free PMC article


The Bruton tyrosine kinase (BTK) inhibitor ibrutinib provides effective treatment for patients with chronic lymphocytic leukemia (CLL), despite extensive heterogeneity in this disease. To define the underlining regulatory dynamics, we analyze high-resolution time courses of ibrutinib treatment in patients with CLL, combining immune-phenotyping, single-cell transcriptome profiling, and chromatin mapping. We identify a consistent regulatory program starting with a sharp decrease of NF-κB binding in CLL cells, which is followed by reduced activity of lineage-defining transcription factors, erosion of CLL cell identity, and acquisition of a quiescence-like gene signature. We observe patient-to-patient variation in the speed of execution of this program, which we exploit to predict patient-specific dynamics in the response to ibrutinib based on the pre-treatment patient samples. In aggregate, our study describes time-dependent cellular, molecular, and regulatory effects for therapeutic inhibition of B cell receptor signaling in CLL, and it establishes a broadly applicable method for epigenome/transcriptome-based treatment monitoring.

Conflict of interest statement

The authors declare no competing interests.


Fig. 1
Fig. 1. Multi-omics analysis of ibrutinib time courses reveals broad changes among immune cells.
a Schematic representation of the study design. Peripheral blood from patients with CLL undergoing single-agent ibrutinib therapy was collected at defined time points and assayed by flow cytometry (cell composition and immunophenotype), single-cell RNA-seq (gene expression), and ATAC-seq (chromatin accessibility). b Cell type abundance over the ibrutinib time course, as measured by flow cytometry. Triangles represent the mean for each time point and dashed lines indicate the 75% confidence interval around the mean, calculated across seven patients. c Flow cytometry scatterplots showing the abundance of T cell subsets for one representative patient at three time points (day 0: before the initiation of ibrutinib therapy, day 30 (120): 30 (120) days after the initiation of ibrutinib therapy). Cells positive for CD3 or CD8 were gated as indicated by the black rectangles and quantified as percentages of live PBMCs. d Flow cytometry histograms showing CD5 and CD38 expression on CLL cells (pre-gated for live, single CD19+CD5+ cells) for a representative patient and three time points. e Two-dimensional similarity map (UMAP projection) showing all 43,049 single-cell transcriptome profiles that passed quality control. Cells are color-coded according to their assigned cell types based on the expression of known marker genes. f DNA copy number profiles for CLL cells, as inferred from single-cell RNA-seq data. Three genetic aberrations common in CLL are indicated. For illustration, 2500 randomly selected CLL cells are shown for each patient. g Clustered single-cell transcriptome heatmap for the most differentially expressed genes between time points. For illustration, 20,000 randomly selected from a total of 43,049 cells are displayed. h Violin plots showing the distribution of gene expression levels for selected differentially expressed genes over the time course. i Differential gene expression signatures in four cell types, comparing each sample to the matched pre-treatment sample and averaging across patients. eg, i Based on scRNA-seq data for 12 samples obtained from four patients.
Fig. 2
Fig. 2. Changes in chromatin accessibility define an ibrutinib-induced regulatory program in CLL cells.
a Heatmap showing changes in chromatin accessibility for CLL cells over the time course of ibrutinib treatment, based on ATAC-seq data for 33 samples obtained from seven patients. b Mean chromatin accessibility across patients plotted over the ibrutinib time course in dynamically changing regulatory regions. Crosses represent samples from a single patient at a specific time point, and 95% confidence intervals are shown as colored shapes. c Region set enrichments for clusters of dynamic regions, calculated using the LOLA software. Enrichment p-values were Z-score transformed per column. d Heatmaps showing mean chromatin accessibility of regulatory regions overlapping with putative binding sites, expression of the corresponding transcription factor, and total number of its binding sites. Clustering was performed on the mean chromatin accessibility values. e Scatterplot showing differential regulation of transcription factors upon ibrutinib treatment. The x-axis displays the enrichment of transcription factors enriched in the LOLA analysis, and the y-axis displays the enrichment of their target genes among the differentially expressed genes. f Gene expression histogram across CLL cells in one patient, demonstrating the decline of a B cell-specific expression signature over the time course of ibrutinib therapy. For illustration, data are shown for the patient with most time points in the single-cell RNA-seq analysis (CLL5).
Fig. 3
Fig. 3. Non-malignant immune cells acquire a shared quiescence-like gene signature upon ibrutinib therapy.
a Mean chromatin accessibility across patients plotted over the ibrutinib time course for clusters of dynamically changing regulatory regions in five immune cell types, based on ATAC-seq data for 122 samples obtained from seven patients. b Heatmap of chromatin accessibility for CD4+ cells, illustrating dynamic regulation over the ibrutinib time course. c Stacked bar plots indicating the percentage of dynamically changing regions in each cluster. d Region set enrichments for the clusters of dynamically changing regions, calculated using the LOLA software and publicly available region sets as reference (mainly based on ChIP-seq data). Enrichment p-values were Z-score transformed per column. e Heatmap showing mean expression levels for genes that were differentially expressed over the ibrutinib time course when combining the data for CLL cells and for the five non-malignant immune cell types. Values represent column Z-scores of gene expression. f Gene set enrichments for genes downregulated across cell types, using WikiPathways as reference (Fisher’s exact test, left: FDR-corrected p-value, right: odds ratio as a measure of effect size). g Expression score for the quiescence-like gene signature (as shown in e) in an independent cohort, calculated from bulk RNA-seq data for PBMCs collected before the start of ibrutinib therapy and at two subsequent time points. Significance was assessed using a paired t-test. h ROC curves illustrating the prediction performance of the gene signature (from e) for classifying samples in the independent validation cohort (solid lines). As negative controls, each prediction was repeated 100 times with permuted class labels, and the mean ROC curves across iterations are shown (dotted lines).
Fig. 4
Fig. 4. Heterogeneity across patients reflects and predicts the patient-specific temporal response to ibrutinib.
a Computational approach to quantify changes in genetic diversity based on copy number profiles inferred from the single-cell RNA-seq data. Shifts in the distribution of pairwise distance similarities between time points indicate changes in the genetic diversity of the cell population. b Scatterplot comparing across patients the change in genetic diversity between day 0 and day 120/150 of ibrutinib treatment (x-axis) with the change in the CLL cell percentage on day 120/150 of ibrutinib treatment compared to day 0 as measured by flow cytometry (y-axis). c Clustered heatmap for chromatin accessibility profiles of CLL cells, based on ATAC-seq data for 33 samples obtained from seven patients. The heatmap shows the top 1000 genomic regions that at day 0 associate with the second principal component (Supplementary Fig. 11a), annotated on the left with the change in CLL cell fraction (as in b). d Scatterplot comparing across patients the average chromatin accessibility for regions linked to the second principal component (as in c, x-axis) with the change in CLL cell fraction (as in b, y-axis). e Stacked bar charts showing the number and direction of deviations from the actual collection time point when predicting time points in each patient after training the classifier in all other patients. f Violin plots showing the predicted (x-axis) and actual (y-axis) number of days under ibrutinib therapy in each patient. Predictions are derived from regression models trained on all other patients. g Scatterplot comparing the predicted time under ibrutinib therapy (from f, x-axis) with the change in CLL cell fraction (as in b, y-axis).

Similar articles

See all similar articles

Cited by 2 articles


    1. Byrd, J. C., Stilgenbauer, S. & Flinn, I. W. Chronic lymphocytic leukemia. Hematol. Am. Soc. Hematol. Educ. Prog. 163–183 (2004). - PubMed
    1. Stevenson FK, Krysov S, Davies AJ, Steele AJ, Packham G. B-cell receptor signaling in chronic lymphocytic leukemia. Blood. 2011;118:4313–4320. doi: 10.1182/blood-2011-06-338855. - DOI - PubMed
    1. Puente XS, et al. Whole-genome sequencing identifies recurrent mutations in chronic lymphocytic leukaemia. Nature. 2011;475:101–105. doi: 10.1038/nature10113. - DOI - PMC - PubMed
    1. Quesada V, et al. Exome sequencing identifies recurrent mutations of the splicing factor SF3B1 gene in chronic lymphocytic leukemia. Nat. Genet. 2011;44:47–52. doi: 10.1038/ng.1032. - DOI - PubMed
    1. Puente XS, et al. Non-coding recurrent mutations in chronic lymphocytic leukaemia. Nature. 2015;526:519–524. doi: 10.1038/nature14666. - DOI - PubMed

MeSH terms