Diffuse large B-cell lymphoma (DLBCL) comprises two major cell-of-origin subtypes, germinal center B-cell (GCB) type and activated B-cell (ABC) type. ABC-DLBCL is characterized by chronic active B-cell receptor (BCR) signalling and NFκB activation, which is explained by activating mutations of the BCR signalling cascade in a minority of cases. We here demonstrate that autonomous BCR signalling, akin to its essential pathogenetic role in chronic lymphocytic leukemia (CLL), can explain chronic active BCR signalling in DLBCL. We show that 13 of 18 tested DLBCL-derived BCR induced spontaneous calcium flux in murine triple knock-out pre-B cells10 in the absence of antigenic stimulation or external BCR crosslinking. Autonomous BCR signalling was associated with IgM isotype, dependent on somatic BCR mutations, and largely restricted to non-GCB DLBCL. Autonomous BCR signaling represents a novel immunological driver mechanism originating from individual BCR sequences and adds a new dimension to currently proposed genetics- and transcriptomics-based DLBCL classifications.

Cell lines and biopsies
Fresh-frozen biopsies of histologically confirmed DLBCL samples were identified in the pathology archive at Leiden University Medical Center (LUMC). The study was approved by the Scientific Review Committee of the LUMC Dept of Hematology under an applicable waiver of consent by the LUMC Ethical Committee (B16.048).

Genetic analyses
Whole exome sequencing (WES) was performed on fragmented DNA with the SureSelect Human All Exon V7 kit (Agilent) capture on the HiSeq2000 (Illumina) platform to an average coverage of 50x. For the variant calling analysis, FASTQ files were processed using the Sarek workflow v2.7 and aligned to the human reference genome GRCh38 using the Burrows-Wheeler Algorithm (BWA) v0.7.17. 35,36 Duplicated mapped reads were marked, local realignment of regions flanking indels, and recalibration of base quality scores were performed to obtain more accurate bases according to the Genome Analysis ToolKit (GATK) best practices version v4.1.7.0. 37 Single-nucleotide variants (SNV) and short insertions and deletions (INDELS) were called using Strelka2 v2.9.10. 38 Only high confidence variants defined by quality scores (GQX) of at least 15 for SNV and 30 for INDELS were kept. The resulting variant call files were annotated by Ensembl-VEP (v103) with four filtering steps. Variants were filtered for the NFkB signalling pathway of the Kyoto Encyclopedia of Genes and Genomes (www.kegg.jp/entry/map04064) and the most frequently mutated genes in DLBCL. Thereafter, variants were filtered by consequences, i.e. frameshift, in-frame deletion, missense, missense variant & splice region variant, splice region variant & synonymous variant, synonymous, in-frame insertion, stop gained, stop lost, frameshift variant & stop lost, missense variant & splice region variant, and coding sequence variant. Finally, variants were annotated for predicted effects by CADD phred, SIFT, and POLYPHEN scores, and according to clinical impact. Benign variants annotated in ClinVar 202008 were discarded. Workflow quality control metrics were calculated and aggregated by MultiQC v1.8. 41 Recalibrated bam files from the variant calling workflow were used for the detection of copy number variations (CNV) by the somatic copy number variation workflow following the Broad’s recommended best practices using GATK4 CNV 37 with a panel of 24 normal tissue samples. The modelled segment files were filtered by genomic coordinates of the genes/regions of interest. Methods for processing the data: For whole exome sequencing (WES), DNA was amplified if necessary by isothermal alkaline genome amplification with Phi29 polymerase and random hexamer priming (REPLI-g kit; Qiagen). WES was performed on fragmented DNA with the SureSelect Human All Exon V7 kit (Agilent) capture on the HiSeq2000 (Illumina) platform to an average coverage of 50x.