B-cell lymphocytosis and reprogramming due to biallelic CARD11 mutations

Abstract

Background:

Adaptive immune responses are tightly controlled by proteins including CARD11 that regulate signaling events downstream of the T- and B-cell receptors. Germline mutations in CARD11 cause several distinct monogenic inborn errors of immunity with early childhood onset and potentially fatal prognoses. Somatic CARD11 gain-of-function mutations are associated with B-cell malignancies. Precisely how various CARD11 mutations culminate in unique clinical entities, and the mechanisms of CARD11-driven B-cell proliferation, is not fully understood.

Objective:

We sought to identify the genetic basis of disease and characterize immune-cell phenotypes and functions in a patient with apparent BENTA (B-cell expansion with nuclear factor-κB [NF-κB] and T-cell anergy) disease and a history of sibling death in early childhood.

Methods:

We used whole-exome sequencing, flow and mass cytometry, whole blood mRNA analyses, in vitro T-cell proliferation and B-cell differentiation assays, and single-cell RNA sequencing of patient-derived samples to identify the genetic basis of disease and define its molecular and cellular mechanisms. We also ectopically expressed wild-type and mutant forms of CARD11 in T and B cells and assessed their functional impacts on NF-κB–dependent responses.

Results:

We report a surprising new genetic basis of BENTA caused by homozygosity for the novel CARD11 mutation R331P and characterized by massive expansion of B cells with a naive surface phenotype and aberrant transcriptional program. This autosomal-recessive form of BENTA features exaggerated B-cell lymphocytosis relative to monoallelic BENTA. Furthermore, we have identified patterns of gene expression that distinguish B cells of patients with the autosomal-recessive form of BENTA from those of healthy controls and from monoallelic BENTA. We found that ectopic CARD11 R331P expression induced constitutive NF-κB activity in T and B cells. These data suggest that R331P is a gain-of-function mutation and causes BENTA in homozygosity.

Conclusions:

These results define a novel autosomal-recessive form of BENTA disease. Additional analysis of mutation-driven changes in B-cell function may shed light on the mechanisms of B lymphomagenesis in patients with germline or somatic CARD11 variants.

CARD11 transmits signals downstream of antigen receptor engagement in B and T lymphocytes to activate the nuclear factor-κB (NF-κB), c-Jun N-terminal kinase, and mTOR pathways^1,2 and is critical to appropriate triggering and resolution of adaptive immune responses. The pivotal role of CARD11 in regulating lymphocyte function is underscored by the diversity of human diseases caused by CARD11 mutations. Biallelic null mutations in CARD11 result in marked immunodeficiency.^3,4 However, various other mutations that partially preserve CARD11 expression and function underlie inborn errors of immunity (IEI) with a spectrum of clinical manifestations. Heterozygous gain-of-function (GOF) CARD11 mutations underlie B-cell expansion with NF-κB and T-cell anergy (BENTA), a disease characterized by B-cell lymphocytosis and poorly functional T cells.⁵ This condition was recognized clinically in 1971⁶ and its genetic basis was discovered in 2012 through sequencing of the 2 daughters of the index case, both of whom also presented with B lymphocytosis.⁵ Since then, additional patients with BENTA with CARD11 GOF variants have been described, with clinical features including B-cell expansion, splenomegaly, lymphadenopathy, and recurrent bacterial and viral infections.¹ Somatic GOF variants in CARD11 are also frequent in B-cell lymphoma, emphasizing the key role for these mutations in pathogenic B-cell lymphoproliferation. Diffuse large B-cell lymphoma is an aggressive non-Hodgkin lymphoma, and CARD11 mutations have been reported in approximately 10% of such cases.⁷ In contrast, heterozygous dominant-negative CARD11 variants can lead to severe atopic dermatitis.⁸ The precise mechanisms by which specific CARD11 mutations perturb NF-κB and other signaling pathways, alter lymphocyte development and function, and culminate in these diverse phenotypes are incompletely understood.

METHODS

Human subjects

Written informed consent was obtained from all participants before participation, and this study was conducted in accordance with the Declaration of Helsinki and approved by the Institutional Review Board at Vanderbilt University Medical Center (Nashville, Tenn).

Whole-exome sequencing

Genomic DNA was prepared from blood samples of the index patient (P1) and both parents and used for trio-design whole-exome sequencing. Exome capture was performed using the Comprehensive Exome Kit (Twist Biosciences, San Francisco, Calif). Paired-end sequencing was performed on NovaSeq (Illumina, San Diego, Calif). Whole-exome sequencing data quality control was performed, data were aligned to the reference genome GRCH38/hg38, and variant calling was done according to the Genome Analysis Toolkit⁹ best practices and the Burrows-Wheeler Aligner-Maximal Exact Matches (BWA-MEM)¹⁰ alignment algorithm. Next, we used Genome Analysis Toolkit/Picard tools to sort the resulting files, performed base quality score recalibration, and then used HaplotypeCaller to call variants. Variant annotation was done using ANNOVAR.¹¹ Variants were then filtered to retain only those with minor allele frequencies less than 0.01 in the gnomAD¹² database and which fit the autosomal-recessive inheritance pattern for the pedigree of P1.

Cytometry by time-of-flight analysis

PBMCs were isolated from blood samples of healthy donors and patients using Ficoll-Paque density gradient (17144002; Cytiva, Marlborough, Mass) and Sepmate tubes (85450; StemCell Technologies, Vancouver, British Columbia, Canada). For each sample, 5 × 10⁶ PBMCs were stained with Cell-ID Cisplatin-195Pt (Standard BioTools, San Francisco, Calif) to discriminate live and dead cells. Samples were then blocked with Human TruStain FcX (BioLegend, San Diego, Calif), stained with antibody mix (see Table E1 in this article’s Online Repository at www.jacionline.org), fixed with 1.6% paraformaldehyde, and labeled with Cell-ID intercalator-Ir (Standard BioTools). Stained PBMCs were analyzed in a Helios CyTOF (cytometry by time of flight) 3.0 instrument following the manufacturer’s instructions. Resulting data were exported as a flow cytometry standard file and normalized using EQ bead standards (Standard BioTools) following the manufacturer’s protocol. For multidimensional analysis, the data were pregated to remove dead cells and debris as well as select for leukocytes using FlowJo 10.7.2 (Becton Dickinson, Franklin Lakes, NJ). The pregated data were then imported into RStudio. We used the CATALYST¹³ package to arcsine transform marker intensities with a cofactor of 5. Unsupervised clustering was performed using the FlowSOM¹⁴ package, generating 20 clusters that were characterized using marker enrichment modeling¹⁵ and visualized using the R package ggplot2. Subsequent manual clustering generated 8 clusters on the basis of the original 20 cluster FlowSOM heatmap of marker intensities, and these manual clusters were represented through t-distributed stochastic neighbor embedding and Uniform Manifold Approximation and Projection (UMAP) plots, which were standardized at the lowest cell count across all samples (8700 cells). Frequencies of different populations on the basis of total cell counts for each sample were exported from RStudio and analyzed using GraphPad Prism 9 (GraphPad, La Jolla, Calif).

NanoString mRNA analysis

Blood samples were collected in PAXgene blood RNA tubes (762165; Beckton Dickinson). RNA was purified using the GeneJET RNA Purification Kit (K0732; Thermo Fisher Scientific, Waltham, Mass). Targeted gene arrays were performed using an nCounter Immunology Panel (XT-CSO-HIM2–12; NanoString, Seattle, Wash). Raw RCC files were imported into nSolver Analysis Software v4.0. Background thresholding value was estimated from negative control probes and set to 20. Data were normalized using 6 positive control parameters included on the NanoString chip, and the geometric mean of 11 housekeeping genes (EEF1G, G6PD, GAPDH, HPRT1, OAZ1, POLR2A, PPIA, RLP19, SDHA, TBP, and TUBB) as a normalization factor, as suggested by the manufacturer. Probes were excluded from further analysis if less than 75% of samples had values higher than the threshold. The average count for each probe was tabulated for healthy controls and patient samples and graphed using GraphPad Prism 9 (GraphPad). Selected genes, as shown, had a 1-way ANOVA P value less than .05, indicating significant differences between patients and healthy controls, but did not show significant differences when comparing healthy adult and healthy pediatric controls.

T-cell proliferation assay

T cells were purified from PBMCs isolated using the Dynabeads Untouched Human T Cells Kit (11344D; Thermo Fisher Scientific) following the manufacturer’s protocol. Purified T cells were stained with CellTrace Violet (C34571; Thermo Fisher Scientific) for 20 minutes and plated in 96-well round-bottom tissue culture plates (353077; Corning, Corning, NY) with the Dynabeads Human T-Activator αCD3/αCD28 (11161D; Thermo Fisher Scientific) in RPMI-1640 medium (R8758; Sigma-Aldrich, St Louis, Mo) supplemented with 10% FBS and 1X antibiotic and antimycotic for 96 hours at 37°C with 5% CO₂. After incubation, stimulated T cells were stained with CD4-phycoerythrin (357403; BioLegend) and CD8a-allophycocyanin (300912; BioLegend). Flow cytometry was performed using the BD LSR-II cytometer, and data were analyzed using FlowJo software (Becton Dickinson).

Ectopic expression and functional testing of CARD11 mutants

CARD11-knockout (KO) Jurkat T cells carrying a stable 6×κB-responsive Green Fluorescent Protein (GFP) reporter gene were kindly provided by Dr Daniel Krappmann (Helmholtz Munich, Neuherberg, Germany). JPM50.6 T cells¹⁶ were kindly provided by Dr Lawrence Kane (University of Pittsburgh, Pittsburgh, Pa). BJAB B cells were originally provided by Dr Michael Lenardo (National Institute of Allergy and Infectious Diseases/National Institutes of Health). All cell lines were cultured in complete RPMI 1640 medium (Thermo Fisher Scientific) supplemented with 10% heat-inactivated FCS (Millipore Sigma, Burlington, Mass) and 100 U/mL each of penicillin and streptomycin (Thermo Fisher Scientific). The human pUNO-CARD11 expression vector (InvivoGen, San Diego, Calif) was modified to include C-terminal 3X FLAG or V5 tags as previously described.¹⁶ Site-directed mutagenesis for mutant CARD11 constructs was performed using specific primers for linear amplification using 2× Pwo DNA polymerase (Roche, Basel, Switzerland), followed by digestion of methylated DNA using DpnI (Thermo Fisher Scientific). Primer sequences to generate the R331P mutation are as follows: R331P-FOR: 5′-GAGGAGGCCCCCC AGGCA GAG-3′; R331P-REV: 5′-CTCTGCCTGGGGGGCCTCCTC. All plasmids were purified from transformed DH5α Escherichia coli (New England Biolabs, Ipswitch, Mass) and selected using Blasticidin (InvivoGen). For T-cell transfections, CARD11-KO Jurkat or JPM50.6 T cells (3 × 10⁶/cuvette) were electroporated (260 V, 950 μF) with 3 μg DNA in 0.4 mL RPMI plus 10% FBS (no antibiotics) using a BTX electroporator (BTX Harvard Apparatus, Holliston, Mass). BJAB cells (3 × 10⁶/cuvette) were similarly electroporated (250 V, 950 μF) with 6 μg DNA. Transfected cells were then incubated overnight at 37°C in 2 mL RPMI + 10% FBS (no antibiotics). Transfected cells were then stimulated with 1 μg/mL anti-CD3/CD28 antibodies (Becton Dickinson) or 50 ng/mL phorbol myristate acetate (PMA) + 1 mg/mL ionomycin (Millipore Sigma) for 6 hours (CARD11-KO Jurkat) or 24 hours (JPM50.6 and BJAB). GFP reporter expression was quantified on an Accuri C6+ flow cytometer (Becton Dickinson) to measure NF-κB activation in transfected CARD11-KO and JPM50.6 cells. BJAB transfectants were stained on ice for 30 minutes with 2 μL phycoerythrin-conjugated anti-CD83 antibody (clone HB15; BioLegend) and subsequently washed twice in 2 mL FACS buffer (1× PBS, 1% BSA, and 0.1% sodium azide). NF-κB–dependent CD83 expression was quantified on a BD Symphony flow cytometer (Becton Dickinson). All resulting flow-cytometric data were analyzed using FlowJo software. Remaining transfectants (~1–2 million cell equivalents per transfection) were lysed on ice for 30 minutes in 30 to 50 μL NP-40 lysis buffer (50 mM Tris, pH 7.4, 0.5 mM EDTA, 150 mM NaCl, 1% NP-40, 0.5% deoxycholate, complete protease inhibitor cocktail; Millipore Sigma), and insoluble material was removed by centrifugation (14,000 rpm for 5 minutes). SDS-PAGE was used to separate 10 to 20 mL lysate on 4% to 20% TRIS-glycine gels (Bio-Rad); proteins were subsequently transferred to nitrocellulose using a Trans-Blot Turbo transfer system (Bio-Rad). After blocking for 15 to 30 minutes (EveryBlot, Bio-Rad, Hercules, Calif), membranes were incubated overnight with the following primary antibodies diluted in Intrepid blocking buffer (LiCOR Biosciences, Lincoln, Neb): anti-FLAG (clone M2, 1:2,000 dilution; Millipore Sigma), anti-V5 (clone D3H8Q, 1:2,000, Cell Signaling Technology, Danvers, Mass), and/or anti–β-actin (clone D6A8, 1:10,000; Cell Signaling Technology). Following 3 washes in TBS–Tween 20, membranes were incubated for 1 hour with goat anti-mouse or anti-rabbit secondary antibodies conjugated to IRDye680 or IRDye800, washed as earlier, and imaged on a LiCOR Odyssey CLx infrared imaging system (LiCOR Biosciences).

In vitro B-cell differentiation assay

Cryopreserved PBMC samples were rapidly thawed in a 37°C water bath and washed with 10 mL prewarmed media (RPMI with 10% FBS). Naive B cells were enriched using the Naive B Cell Isolation Kit II, human (130–091-150; Miltenyi Biotec), MS columns (130–042-201; Miltenyi Biotec), and OctoMACS separator (130–042-109; Miltenyi Biotec) in 1× PBS + 2% FBS + 2 mM EDTA. Postenrichment, purity of naive B cells was assessed by flow cytometry using antibodies against CD19, IgD, and CD27 (see Table E2 in this article’s Online Repository at www.jacionline.org) and was routinely more than 90%. These naive B cells were cocultured with irradiated (~40 Gy) NIH3T3 modified to stably express human CD40L, BAFF, and IL-21 (CRL-3571; ATCC, Manassas, Va) and used as feeder cells. Irradiated feeder cells (10,000 cells/well) were plated in 96-well flat-bottom plates (FB012931; Thermo Fisher Scientific) the day before purified naive B cells (25,000 cells/well) were added. These cocultures were maintained in ClonaCell-HY Medium A (03801; StemCell Technologies) supplemented with antibiotic/antimycotic (CA002–010; GenDEPOT, Altair, Tex). B-cell differentiation was assessed at various time points by spectral flow cytometry (Table E2) on a Cytek Northern Lights cytometer. Changes in relative frequencies of IgD⁻CD27⁺ and CD38⁺CD27⁺ cells were compared using 2-way repeated-measures ANOVA tests in GraphPad Prism (GraphPad).

Single-cell RNA sequencing

Cryopreserved PBMC samples were rapidly thawed in a 37°C water bath and washed with 10 mL prewarmed media (RPMI with 10% FBS). Cell suspensions were passed through 30 μM SmartStrainers (130–098-458; Miltenyi Biotec) and then centrifuged at 300g for 5 minutes. An aliquot of cells was counted using a Countess II instrument and Countess II FL Automated Cell Counting Chamber Slides (C10228; Thermo Fisher Scientific). Cells were resuspended in PBS with 1% BSA, and then dead cells were removed using the MACS Dead Cell Removal Kit (130–090-101; Miltenyi Biotec) and MS columns (130–042-201; Miltenyi Biotec) according to the manufacturer’s protocol. Viability of more than 95% was confirmed by automated cell counting (Countess II) with trypan blue. Cells were then washed with 3 mL PBS with 1% BSA and centrifuged at 300g for 5 minutes, supernatant was removed and discarded, and cells were resuspended in UltraPure PBS with 1% BSA at a concentration of 1200 cells/μL in low-bind tubes for single-cell capture. Up to 20,000 cells per sample were used for 5′ single-cell RNA-sequencing library preparation and sequencing at the VANTAGE Core Facility (Vanderbilt University Medical Center) using the 10X Genomics Chromium Controller and NovaSeq (Illumina). Resulting raw data were processed using CellRanger (10X Genomics, Pleasanton, Calif).

Quality control

Filtered CellRanger output HDF5 files were analyzed with the Seurat v5.0.3 package¹⁷ in R (version 4.4.1). Genes expressed in fewer than 3 cells and cells expressing fewer than 10 genes were removed from initial processing. Cells containing higher than 10% mitochondrial genes were removed from further analysis. Genes and total read counts less than 5% and greater than 95% of the quantile distribution across each sample were removed from further processing. Doublets were predicted using the scDblFinder package¹⁸ and removed from further processing.

Cell-type annotation

Cell cycle scores were calculated using the CellCycleScoring function and regressed out using the SCTransform function. Anchors between reference and query were precomputed using the annotated PBMC Cite-seq reference and FindTransfer-Anchors function using 50 dimensions and a precomputed principal-component analysis transformation.¹⁹ Cell-type labels and protein data from the reference to the query were transferred using the MapQuery function. Cells with prediction level 1 scores less than the mean – 1 SD were considered as having low confidence scores and removed from further analysis.²⁰

Characterization of B cells

Analysis of B-cell phenotypes was performed using unsupervised clustering with default parameters in standard Seurat workflow and visualized by UMAPs. UMAP density plots were created using the kde2d function from the MASS package. Differentially expressed genes (DEGs) across B clusters were identified with the FindMarkers function using default parameters. Genes with an adjusted P value less than .05 were considered statistically significant. Overrepresentation analysis from the FindMarkers output was done using the clusterProfiler package.²¹ The Wilcoxon rank-sum test was used to identify DEGs in healthy controls and patients.

Gene set enrichment analysis

Gene set enrichment analysis was conducted using fGSEA²² and the Hallmark gene set collection from the Molecular Signatures Database.²³

CD8 and CD4 T-cell differential expression analysis

CD8 and CD4 T-cell subsets were identified using reference-based level 1 annotation (as described earlier). DEGs between the patients with BENTA and the healthy donors were defined using Seurat FindMarkers function with the Wilcoxon rank-sum test method and then filtered to keep only genes with an adjusted P value less than .05 and an absolute value of the average log₂ fold change greater than 1.

CARD11 protein structural modeling

The structure of a CARD11 homodimer was predicted and visualized using AlphaFold3²⁴ on the basis of the amino acid sequence of human CARD11 provided by UniProt (accession no. Q9BXL7).

Data availability statement

CellRanger matrix files as well as R scripts for all analyses presented are available at https://github.com/JanetMarkleLaboratory/phs004115. Raw sequencing data generated for the present study are available in the National Institutes of Health dbGaP depository (accession no. phs004115.v1). Additional raw data are available on request.

RESULTS

We studied a young male patient (P1; Fig 1, A, individual I-5) born to healthy third-degree consanguineous parents and noted to have splenomegaly since birth. At age 2 months 19 days, blood samples from P1 were evaluated for major leukocyte frequencies and counts. B cells were strikingly elevated in frequency (79% of total cells) and absolute number (46,044/μL). Total IgM, IgA, and IgG were within normal ranges for age. P1 also showed signs of anemia and thrombocytopenia and had no history of infections. Blood test results for EBV by PCR were negative. Family history was significant for a sibling death in early childhood. A male sibling (Fig 1, A, individual I-3) had presented with anemia, thrombocytopenia, leukocytosis, and hepatosplenomegaly within the first months of life. A bone marrow aspirate showed an elevated frequency of lymphoid cells, with erythroid and myeloid cells relatively suppressed. This sibling sadly died of unknown causes at age 5 months. A second male sibling (Fig 1, A, I-4) had been healthy. Each parent was healthy, and complete blood cell counts showed that neither parent had B lymphocytosis. Despite no history of infections, P1 was treated with monthly intravenous immunoglobulin infusions and cotrimoxazole prophylaxis. He underwent a mismatched unrelated peripheral blood stem cell transplant (9/10 HLA match) at age 11 months with the following conditioning regimen: intravenous busulfan (targeted to 90 mg/L × h), thiotepa (10 mg/kg), fludarabine (180 mg/m²), ATG grafalon (30 mg/kg), and intravenous rituximab. Graft-versus-host disease prophylaxis consisted of posttransplant cyclophosphamide, cyclosporine, and mycophenolate mofetil. Neutrophil engraftment was noted on day 16 posttransplant. P1 developed cytomegalovirus reactivation on day 20 posttransplant. Unfortunately, he developed severe veno-occlusive disease with multiorgan dysfunction and succumbed to his illness on day 33 posttransplant.

FIG 1.

Genetic analysis of a patient with apparent BENTA disease identifies an unexpected biallelic CARD11 mutation. A, Patient 1 (P1, I-5) is a male infant from a family with consanguineous parentage and a history of sibling death in childhood. Whole-exome sequencing of P1, his father (I-1), mother (I-2), and healthy brother (I-4) revealed a novel mutation in CARD11 resulting in the amino acid change R331P. Genotypes for this variant are shown for all surviving family members. B, Sanger sequencing of the CARD11 variant confirmed its autosomal-recessive inheritance pattern. C, Patient 2 (P2, II-5) is a male child with classical BENTA symptoms and a previously reported CARD11 heterozygous mutation, G126D, with known GOF impact.²⁶ The G126D mutation is de novo in this patient because it was not present in his healthy parents (II-1 and II-2) nor in his 2 elder healthy brothers (II-3 and II-4).

To determine the genetic etiology of disease in P1, whole-exome sequencing was performed with blood samples from all 4 living family members when P1 was 4 months (Fig 1, A). We identified a novel single nucleotide variant at Chr7:2938704, CARD11 c.G992C, p.R331P that was homozygous in P1, heterozygous in each parent, and not present in his healthy brother (Fig 1, A). This mutation was confirmed by Sanger sequencing (Fig 1, B) and cosegregates with B lymphocytosis in an autosomal-recessive manner. The R331P mutation was not found in the gnomADv4.1.0 data set²⁵; however, other variants affecting the R331 residue (R331H and R331C) were present, with minor allele frequencies less than 0.0001 and never in homozygosis.²⁵ P1’s clinical history was consistent with BENTA, yet surprisingly our genetic results suggested a novel biallelic CARD11 mutation as its underlying etiology.

To compare and contrast immune-cell phenotypes of P1 with those caused by a known BENTA-causal monoallelic CARD11 mutation, we recruited patient 2 (P2) who was heterozygous for the BENTA-causal G126D mutation⁵ and had a clinical history of classical BENTA (Fig 1, C, individual II-5).²⁶ Sequencing of P2’s family members indicated that the G126D mutation was de novo (Fig 1, C). We performed CyTOF analysis using PBMC samples of 12 healthy adult donors, 2 healthy pediatric donors, I-1 (wild-type [WT]/R331P), P1 (R331P/R331P), and P2 (WT/G126D) and a 33-antibody panel to profile all major human leukocytes subsets (Table E1). We performed unsupervised clustering to generate 20 cell clusters (see Fig E1, A, in this article’s Online Repository at www.jacionline.org). On the basis of patterns of marker expression across these clusters (Fig E1, B), we reclustered these data and analyzed 8 leukocyte populations (Fig 2, A and B). Increased expression of B-cell–associated proteins, including CD19, CD20, and IgD in P1 and P2, but not in I-1 (WT/R331P), was apparent when assessing median marker intensities by sample (Fig E1, C). We observed marked B-cell expansion in P2, and to an even greater extent in P1, but not in I-1 (WT/R331P), indicating that the R331P mutation causes B lymphocytosis only in the homozygous state (Fig 2, C and D). Several other leukocyte subsets were concomitantly decreased in relative frequency because of B-cell expansion in P1 and P2 (see Fig E2, A, in this article’s Online Repository at www.jacionline.org). Next, we assessed the expression of more than 500 immune-related transcripts in whole blood samples of healthy adult controls (n = 7), healthy pediatric controls (n = 5), and patients with BENTA (P1 and P2). Both patients’ samples showed elevated expression of B-cell–related transcripts BTK, BLNK, CD22, CD40, and HLA-DRA and reduced expression of several genes critical to T-cell activation and proliferation (eg, CD7, IL7R, IL2RB, and STAT5A) (Fig E2, B), relative to healthy controls. We tested the capacity of T cells from healthy controls (n = 3), I-1 (WT/R331P), P1 (R331P/R331P), and P2 (WT/G126D) to proliferate in response to polyclonal T-cell receptor (TCR) stimulation. Like previously reported patients with BENTA⁵, CD4⁺ and CD8⁺ T cells from P1 showed impaired proliferation relative to healthy controls, whereas T cells from I-1 proliferated more similarly to healthy control T cells (Fig 3, A–D). Taken together, these results confirm that P1 has immune-cell phenotypes consistent with BENTA and suggest that heterozygosity for this variant is insufficient to markedly perturb lymphocyte function in these settings.

FIG 2.

Broad profiling of immune-cell frequencies using CyTOF analysis identifies extreme B-cell expansion in P1. A, CyTOF data analysis was used to define 8 major cell clusters and these clusters were visualized on a UMAP. B, Heatmap of marker expression levels across the 8 clusters. C, Density UMAPs showing 8700 cells per plot. Representative plots are shown for adult healthy controls (HC) and healthy pediatric (HP) controls. Results for P2 (CARD11 WT/G126D) and P1 (CARD11 R331P/R331P) as well as the healthy father of P1 (CARD11 WT/R331P) are shown. D, Quantitation of B-cell frequencies in healthy donors, including adult healthy controls (HC, n = 12; dark blue symbols), healthy pediatric controls (HP, n = 2; light blue symbols), and the healthy father of P1 (WT/R331P; yellow symbol) (left bar), and 2 patients with BENTA, P1 (R331P/R331P; red symbol) and P2 (WT/G126D; orange symbol) (right bar), by CyTOF analysis. *P < .05 (1-way ANOVA).

FIG 3.

A and B, Both dominant and recessive patients with BENTA show impaired T-cell proliferation. CD4⁺ (Fig 3, A) and CD8⁺ (Fig 3, B) T cells were polyclonally activated and dilution of CellTrace Violet was measured after 96 hours to assess T-cell proliferation. T cells from both P1 (R331P/R331P; red trace) and P2 (WT/G126D; orange trace) showed severe defects in proliferation, relative to T cells from 3 healthy controls (HC1, HC2, and HC3; blue trace) and the healthy father of P1 (WT/R331P; yellow trace). C and D, MFI of CellTrace Violet (CTV) (Fig 3, C) and expansion index (Fig 3, D) were determined for each sample using FlowJo software. MFI, Mean fluorescence intensity. *P < .05 (unpaired t tests).

To firmly establish the functional impact of the R331P mutation, we ectopically expressed WT or R331P CARD11-encoding plasmids in CARD11-KO Jurkat T cells containing an NF-κB–driven GFP reporter. The published BENTA-causing GOF mutation E134G²⁶ was included as a positive control. Relative to WT CARD11, both the E134G and R331P mutations resulted in higher constitutive activation of the NF-κB pathway (Fig 4, A–C). In response to stimulation with anti-CD3/CD28, E134G but not R331P displayed higher NF-κB activity than WT (Fig 4, A–C). However, with stronger stimulation using PMA and ionomycin, the R331P variant, but not the E134G variant, showed attenuated NF-κB activity relative to WT (Fig 4, A–C). Cotransfection of WT and R331P mutant CARD11 resulted in only a marginal increase in NF-κB activity relative to WT alone, both at steady state and after anti-CD3/CD28 stimulation, similar to effects observed with E134G. Responses to PMA/ionomycin stimulation were more variable but comparable with WT, indicating no dominant-negative effect and consistent with the lack of clinical phenotype in the heterozygous parents of P1 (Fig 4, D–F). Expression levels of mutant and WT CARD11 protein were validated by Western blotting (Fig 4, G and H). Similar results, including constitutive NF-κB activity and blunted response to PMI/ionomycin in cells with R331P expression, were obtained using JPM50.6 T cells containing an NF-κB–driven GFP reporter (see Fig E3, A–F, in this article’s Online Repository at www.jacionline.org). Collectively, these data suggest the R331P variant acts as a constitutive GOF in the absence of stimulation, but as a hypomorph on T-cell activation.

FIG 4.

Ectopic expression and in vitro functional testing of novel CARD11 mutation R331P. A, CARD11-KO Jurkat T cells containing an NF-κB–driven GFP reporter were transfected with an empty vector (EV) or with vectors encoding WT, E134G, or R331P mutant versions of CARD11. Cells were then either cultured without stimulation (No Stim) or stimulated with antibodies directed against CD3 and CD28 (CD3/CD28) or with PMI and ionomycin (PMA/I). Histograms show the percentage of GFP⁺ cells for each sample. B, Summary of percent GFP⁺ cells from 3 independent replicate experiments performed as in Fig 4, A. C, The MFI of the NF-κB–GFP reporter was measured by flow cytometry in 3 independent replicate experiments performed as in Fig 4, A. D, CARD11-KO Jurkat T cells containing an NF-κB–driven GFP reporter were cotransfected with a WT CARD11 vector alongside EV, E134G, or R331P mutant CARD11 plasmids, and then stimulated and measured by flow cytometry as in Fig 4, A. Histograms show the percentage of GFP⁺ cells for each sample. E, Summary of percent GFP⁺ cells from 3 independent replicate experiments performed as in Fig 4, D. F, The MFI of the NF-κB–GFP reporter was measured by flow cytometry in 3 independent replicate experiments performed as in Fig 4, B. *P < .05 relative to WT No Stim condition (1-way ANOVA). ⊕P < .05 relative to WT anti-CD3/CD28 condition (1-way ANOVA). #P < .05 relative to WT PMA/I condition (1-way ANOVA). G, Western blots confirmed expression of WT and mutant versions of CARD11 in cells used for panels A-C. H, Western blots confirmed expression of WT and mutant versions of CARD11 in cells used for panels D to F. MFI, Mean fluorescence intensity.

To assess how R331P influences NF-κB signaling in B cells, we also performed CARD11 transfections in the BJAB B-cell line, using CD83 expression as a surrogate for NF-κB activity.⁷ Similar to our results in T-cell lines, both R331P and E134G mutations induced higher CD83 expression in unstimulated BJAB transfectants when compared with WT, confirming GOF activity (Fig E3, G and H). In contrast, PMA/ionomycin stimulation induced significantly higher CD83 expression for both variants relative to WT, suggesting that in contrast to T cells, R331P does not attenuate NF-κB signaling in B cells under these conditions (Fig E3, G and H). No significant differences in CD83 expression were observed with ectopic coexpression of WT CARD11 (Fig E3, I and J). Comparable expression of WT and mutant CARD11 proteins was confirmed by immunoblotting (Fig E3, K and L). Collectively, these data confirm that the R331P mutation confers a GOF for basal NF-κB activation, consistent with the BENTA phenotype observed in P1. Moreover, these data suggest that the R331P variant selectively attenuates NF-κB–dependent transcription in T cells but not in B cells under strong activation conditions, suggesting cell-type–specific impacts of this variant.

Expansion of B cells with a naive and/or transitional phenotype and poor generation of antibody-secreting cells is a recognized consequence of GOF CARD11 mutations leading to BENTA. We used biaxial gating of CyTOF data generated with PBMCs from healthy adult controls (n = 12), healthy pediatric controls (n = 2), I-2 (WT/R331P), P1 (R331P/R331P), and P2 (WT/G126D) to assess the relative frequencies of B-cell subsets. Within the CD19⁺ B-cell compartment, P1 and P2 showed reduced frequencies of both switched (IgD⁻CD27⁺) and nonswitched (IgD⁺CD27⁺) memory cells and elevated frequencies of naive (CD27⁻IgD⁺) cells relative to healthy controls or R331P heterozygous samples (Fig 5, A and B), consistent with other patients with BENTA.⁵ We also observed a trend for increased transitional B cells in P1 and P2, consistent with previous findings in patients with BENTA.⁵ A previous study demonstrated intrinsic defects in B-cell differentiation in samples from patients with BENTA, with diminished generation of antibody-secreting cells.²⁷ To determine how the R331P CARD11 mutation affects this process in both the homozygous and heterozygous states, purified naive B cells from healthy WT donors, P1, P2, and I-2 were cocultured with irradiated NIH3T3 cells modified to stably express human CD40L, BAFF, and IL-21. This coculture system reproducibly supports the differentiation of healthy donor naive (IgD⁺CD27⁻) B cells into cells with a switched memory B-cell phenotype (IgD⁻CD27⁺) and the emergence of CD27⁺CD38⁺ antibody-secreting cells by day 7 (see Fig E4 in this article’s Online Repository at www.jacionline.org). We observed a decreased frequency of IgD⁻CD27⁺ memory B cells (Fig 5, C and D) and decreased development of CD27⁺CD38⁺ antibody-secreting cells (Fig 5, E and F) in samples from P1 and P2, in agreement with previous findings in patients with BENTA,²⁷ but not in I-2 (WT/R331P genotype). These data indicate that homozygosity, but not heterozygosity, for the R331P CARD11 mutation results in intrinsic defects in B-cell differentiation.

FIG 5.

Patients with BENTA have aberrant B-cell subset distributions and show B-cell–intrinsic differentiation defects. A, IgD vs CD27 expression (upper panel) and CD24 vs CD38 expression (lower panel) were assessed within gated CD19⁺ B cells in PBMC samples from a healthy control (HC), I-1 (WT/R331P), P1 (R331P/R331P), and P2 (WT/G126D) by CyTOF analysis. B, Quantitation of switched memory (IgD⁻CD27⁺), nonswitched memory (IgD⁺CD27⁺), naive (CD27⁻IgD⁺), and transitional (CD24^hiCD38^hi) B-cell frequencies in healthy donors, including adult healthy controls (HC, n = 12; dark blue symbols), healthy pediatric controls (HP, n = 2; light blue symbols), and the healthy father of P1 (WT/R331P; yellow symbol) (left bar), and 2 patients with BENTA, P1 (R331P/R331P; red symbol) and P2 (WT/G126D; orange symbol) (right bar), by CyTOF analysis. *P < .05 (1-way ANOVA). C, Naive B cells were enriched from PBMC samples of 3 healthy controls (HC1, HC2, and HC3), I-1 (WT/R331P), P1 (R331P/R331P), and P2 (WT/G126D), and expression levels of IgD and CD27 were assessed by flow cytometry immediately following isolation (day 0; right panel) and after 7 days of coculture with irradiated NIH3T3 cells modified to stably express human CD40L, BAFF, and IL-21 (day 7; left panel). D, The percentages of cells with a switched memory (IgD⁻CD27⁺) phenotype at days 0 and 7 of coculture are shown for healthy controls (HC; n = 5), I-1 (WT/R331P), P1 (R331P/R331P), and P2 (WT/G126D). E, Naive B cells were enriched as in Fig 5, C, and expression levels of CD38 and CD27 were assessed by flow cytometry. F, The percentages of CD27⁺CD38⁺ cells at days 0 and 7 of coculture are shown for healthy controls (HC; n = 5), I-1 (WT/R331P), P1 (R331P/R331P), and P2 (WT/G126D). NS, Not significant. *P < .05 (2-tailed t tests).

To more deeply characterize the phenotype and function of B cells in P1 and P2 in an unbiased manner, we performed single-cell RNA sequencing with PBMC samples from 2 pediatric healthy controls (HP1 and HP2), P1, and P2. Following supervised clustering and cell-type annotation (Fig 6, A), the expansion of B cells and relative depletion of other cell types in P1 and P2 were readily apparent (Fig 6, B and C). Across all samples, cell-type level 1 prediction scores were high (see Fig E5, A, in this article’s Online Repository at www.jacionline.org), indicating that cells from all donors could be accurately mapped to the 8 level 1 cell types annotated by the Seurat v5.0.3 R package.¹⁷ In contrast, the samples from the patients with BENTA, particularly cells from P1, had low prediction scores for level 2 cell-type annotations (Fig E5, B), suggesting a mismatch between the transcriptomes of reference (healthy control) samples and patients with BENTA for B-cell subset identities (eg, naive B and memory B cells). Thus, we instead performed unsupervised clustering within the B-cell compartment to characterize B-cell subpopulation structure in an unbiased manner (Fig 6, D) and identify genes that distinguish patient from control B cells. Although HP1 and HP2 showed similar B-cell subpopulation architectures, B cells from patients with BENTA were dominated by clusters distinct both from healthy control B-cell clusters and from each other (Fig 6, E). Clusters 0 and 1 predominated in B cells from P2 and P1, respectively, but were nearly absent from controls (Fig 6, F). Analysis of DEGs by cluster identified divergent transcriptional programs in clusters 0 and 1 (Fig 6, G), indicating that although both patients with BENTA have atypical and expanded B cells, each patient’s B cells are functionally dysregulated in unique ways. Taken together, these findings establish that homozygosity for the R331P variant in CARD11 is associated with dramatic B lymphocytosis (in excess of that typically described in monoallelic BENTA), B-cell–intrinsic defects in differentiation, and an aberrant B gene expression pattern.

FIG 6.

Identification of an aberrant B-cell transcriptional program in patients with BENTA. PBMCs from 2 healthy pediatric controls (HP1 and HP2) and 2 patients with BENTA, P1 (R331P/R331P) and P2 (WT/G126D), were used for single-cell RNA sequencing. A, Following data quality control, we performed supervised analysis using reference data and the Seurat v5.0.3 package in R (for details, see this article’s Methods section in the Online Repository at www.jacionline.org) to annotate 8 major cell types. B and C, Density UMAPs for both patients show relative enrichment for B cells and depletion of other cell types, which is particularly prominent in P1 (Fig 6, B), also summarized in Fig 6 C. D, Within the B-cell compartment, we performed unsupervised clustering to interrogate subpopulation structure. This analysis identified 7 clusters. E, UMAP reveals that clusters 0 and 1 are highly enriched in P2 and P1, respectively. F, Cluster frequencies across samples. G, DEGs by cluster (adjusted P < .05).

Although the occurrence of T-cell anergy in BENTA is well documented, the impact of GOF CARD11 mutations on T-cell phenotypes and functions in vivo remains incompletely understood. We measured the relative frequencies of naive (CD45RA⁺CCR7⁺), central memory (CD45RA⁻CCR7⁺), effector memory (CD45RA⁻CCR7⁻), and terminally differentiated effector memory cells reexpressing CD45RA (TEMRA cells and CD45RA⁺CCR7⁻) within the CD4⁺ and CD8⁺ T-cell compartments using PBMC samples of 12 healthy adult donors, 2 healthy pediatric donors, I-1 (WT/R331P), P1 (R331P/R331P), and P2 (WT/G126D). We observed moderately reduced frequencies of central memory cells in both the CD4⁺ and the CD8⁺ T-cell compartments of the patients with BENTA (P1 and P2), although this difference was statistically significant in CD4⁺ T cells only (see Fig E6, A–D, in this article’s Online Repository at www.jacionline.org). We observed an increased relative frequency of TEMRA cells in both the CD4⁺ and the CD8⁺ T-cell subsets of the patients with BENTA (Fig E6, A–D). The frequencies of naive and effector memory CD4⁺ and CD8⁺ T cells remained unchanged. In addition, we performed differential expression analysis comparing the transcriptional programs of CD4⁺ and CD8⁺ T cells from patients with BENTA versus healthy controls. Among the top 25% DEGs (ranked by adjusted P value) in both the CD4⁺ and the CD8⁺ T-cell analyses, we found several immune “checkpoints” with well-characterized roles in suppression of T-cell proliferation and effector function that showed elevated expression in patients with BENTA, relative to healthy controls (see Table E3 in this article’s Online Repository at www.jacionline.org). CTLA4 was upregulated in CD4⁺ and CD8⁺ T cells of patients with BENTA, relative to the same T-cell populations in healthy controls. CARD11-BCL10-MALT1 (CBM) complex signaling has been shown to induce CTLA-4 expression in T cells.²⁸ Once expressed, CTLA-4 competes with the costimulatory receptor CD28 for binding to B7 molecules on antigen-presenting cells, delivering an inhibitory signal that leads to long-term T-cell unresponsiveness.²⁹ CD8⁺ T cells of patients with BENTA were also found to express elevated LAG3, another negative regulator of T-cell activation and proliferation.³⁰ CD4⁺ T cells of patients with BENTA expressed elevated TIGIT, which clusters with the TCR in the immune synapse and delivers inhibitory signals that suppress TCR-dependent T-cell proliferation.^31–33 The heightened expression of these T-cell checkpoint genes suggests a potential mechanism for hyporesponsiveness to TCR stimulation observed in T cells of patients with BENTA. Additional work is needed to define possible roles of T-cell checkpoints as contributors to the T-cell–dependent portion of the BENTA disease phenotype. Collectively, our data identify an autosomal-recessive form of BENTA caused by the R331P allele, which exerts a GOF impact on CARD11-dependent NF-κB signaling and results in dramatic B-cell expansion, B-cell–intrinsic defects in differentiation, and T-cell anergy in the homozygous, but not in the heterozygous, state. These findings extend the diversity of IEI resulting from mutations in the CBM complex.

DISCUSSION

The dramatic B-cell expansion and aberrant B-cell transcriptomes observed in P1, together with the fatal disease in his elder sibling, strongly suggested that homozygosity for the CARD11 R331P mutation leads to life-threatening derangement of B-cell function. The R331P mutation lies within the CARD11 coiled-coil (CC) domain (amino acids 130–449). Interactions between the CC domain and the linker region, together with the inhibitory domain, are required to prevent constitutive activation and maintain CARD11 in an autoinhibitory state.^5,34 Similar to reported heterozygous GOF mutations that cause BENTA, our findings suggest that homozygosity for R331P results in spontaneous NF-κB signaling without requiring antigen receptor engagement.^1,35 These findings are consistent with a mutation that may bypass the typical “opening” step (loss of autoinhibition) for CBM complex activation yet hinders the formation of a fully functional CBM complex on maximal stimulation.³⁶ Structural modeling of a CARD11 homodimer suggests that the 2 R331 residues of each monomer’s CC domain are in close proximity to each other (see Fig E7 in this article’s Online Repository at www.jacionline.org). Thus, we speculate that the stability of homodimerization, which is critical to CARD11 autoinhibition,³⁴ may be disrupted by the presence of P331 on both monomers, but preserved if 1 WT CARD11 monomer is available. Although the stoichiometry of CARD11 oligomers within the mature CBM complex remains unknown, this model offers a plausible mechanism accounting for the development of BENTA in individuals who are homozygous for the R331P mutation, but not in heterozygotes.

GOF mutations in CARD11 disrupt B-cell differentiation in a cell-intrinsic manner.²⁷ Although we observed similar defects in memory B and plasma cell differentiation in P1 and P2, our data suggest that different CARD11 mutations may have subtly differential impacts on B-cell fates and functions. B-cell lymphocytosis was consistently higher in P1 than in P2 (Fig 2, C and D, and Fig 6, C), and B cells from these patients showed different patterns of gene expression (Fig 6, D–G). Among the DEGs specific to B-cell cluster 1 (Fig 6, G), which is overrepresented in P1 (Fig 6, E and F), are several genes with known or putative roles in B-cell proliferation and/or malignancy. For example, a recent study identified ARRDC3 as a regulator of p53-mediated suppression of B-cell lymphoma in a mouse model.³⁷ SIK3 expression has recently been shown to promote tumor cell survival and protect tumor cells from attack by cytotoxic T cells.³⁸ IGF2BP3, an RNA-binding protein, plays a crucial role in B-cell malignancies by regulating the expression of oncogenic targets such as MYC and CDK6, promoting B-cell survival and proliferation.³⁹ Gene set enrichment analysis of DEGs specific to B-cell cluster 1 revealed enrichment for E2F targets, MYC targets, and genes involved in oxidative phosphorylation (see Fig E8 in this article’s Online Repository at www.jacionline.org), consistent with broadly aberrant control of proliferation and activation in R331P homozygous B cells. These findings suggest that R331P homozygosity could strongly predispose to B-cell lymphomagenesis, consistent with profound B-cell lymphocytosis in P1 comparable with only 1 other reported patient with BENTA.⁴⁰ Furthermore, our findings suggest that GOF CARD11 variants such as R331P signal differently in T cells versus B cells and can actually attenuate T-cell (but not B-cell) NF-κB activation and consequent proliferation. Further work is needed to determine how specific CARD11 mutations differentially perturb lymphocyte proliferative and survival signals and result in aberrant B- and T-cell fates. Continued dissection of how diverse mutations affect CARD11 function will illuminate both mechanisms of IEI in rare patients and molecular drivers of B-cell malignancies.

Key messages.

• The first recessive form of BENTA disease was identified and homozygosity for the CARD11 mutation R331P was identified as its cause.

• The R331P mutation confers a GOF for basal NF-κB activation in both T and B cells, yet it reduces NF-κB activity in the setting of strong T-cell, but not B-cell, activation.

• Homozygosity for R331P results in extreme expansion of naive B cells and broad perturbations in B-cell transcriptomes in a manner distinct from that observed in monoallelic BENTA.

Abbreviations used

BENTA

B-cell expansion with NF-κB and T-cell anergy

CBM

CARD11-BCL10-MALT1

CyTOF

Cytometry by time of flight

DEG

Differentially expressed gene

GOF

Gain of function

IEI

Inborn errors of immunity

KO

Knockout

NF-κB

Nuclear factor-κB

PMA

Phorbol myristate acetate

TCR

T-cell receptor

UMAP

Uniform Manifold Approximation and Projection

WT

Wild type