Genetic and clinical features of patients with pRD

The pRD cohort included a 5-month-old asymptomatic male (P1)21 and 15 patients with a CID or CID-G/AI phenotype (P2–16) (Supplementary Table 1). Eleven patients carried RAG1 (7 compound heterozygous and 4 homozygous) and 5 patients carried RAG2 (3 compound heterozygous and 2 homozygous) variants, for a total of 16 RAG1 and 8 RAG2 distinct mutant alleles (Extended Data Fig. 1a). Pathogenicity was assigned based on combined assessment of in vitro recombination activity assays2,3 and curated data obtained from ClinVar database and/or analysis following guidelines of American College of Medical Genetics and Genomics23 and the Association for Molecular Pathology (Extended Data Fig. 1b and Supplementary Table 2).

All patients displayed profound lymphopenia as compared to aged-matched healthy ranges24 (Extended Data Fig. 1c–f), except P13 who had chronic Epstein–Barr virus (EBV) and Cytomegalovirus (CMV) viremia with lymphoproliferation25. Asymptomatic P1 was considered to be antigen-naive (pRD-N), whereas symptomatic P2–16 were grouped as antigen-experienced patients (pRD-Ag). P1 remained CMV negative and showed no clinical or laboratory signs of infections until successful hematopoietic stem cell transplantation21. All patients with pRD-Ag had upper and/or lower respiratory tract infections, including bronchiectasis in eight cases. Seven patients had one or a combination of adenovirus or herpesvirus infections (EBV, CMV, HSV and varicella). Ten patients had autoimmunity and three patients had granulomas affecting lung, skin and/or spleen (Extended Data Fig. 1b).

Diverse autoantibody profiles in patients with pRD

We detected autoantibodies in 11 plasma samples of 13 patients tested by immunofluorescent analysis (IFA) on HEp-2 cell slides with various intensities (very bright, bright, intermediate and low positive; 1, 4, 4 and 2 patients, respectively), whereas healthy controls (HCs) showed no or low positive staining (Extended Data Fig. 2a). Staining patterns included homogeneous nuclear (P11 and P14), speckled cytoplasmic (P1, P2, P4, P8, P9, P12 and P13) and cytoplasmic reticular/mitochondrial (P3 and P16). Quantitative image analysis confirmed stronger cytoplasmic and nuclear staining in patients compared to HCs (Extended Data Fig. 2b,c, respectively). Eleven patients from 15 produced IgG autoantibodies against at least one specific antigen as tested by ELISA (defined as z score >2) and autoantibody levels against nine antigens were significantly higher in patients compared to HCs (Extended Data Fig. 2d). IgG autoantibodies to interferon (IFN)-α, IFN-ω and/or interleukin (IL)-12, a hallmark of pRD5,26, were present in ten patients with pRD-Ag (newly detected in four patients and previously published in six cases5) but not in pRD-N (Extended Data Fig. 2e). We also detected VH4-34-encoded IgM 9G4 antibodies that recognize polysaccharide antigens in red blood cells27 in eight patients (Extended Data Fig. 2f); hence, the presence of a wide range of IgM and IgG autoantibodies indicates a failed B cell tolerance in pRD.

Dysregulated peripheral B cell maturation and activation

To assess peripheral B cell subsets, we performed unsupervised high-dimensional analysis from the peripheral blood of patients with pRD (Extended Data Fig. 3a–f). We identified 16 distinct metaclusters (M0–15), each corresponding to a unique B cell population with a specific surface marker expression pattern (Fig. 1a). Visualization of the B cell compartment composition by the 16 metaclusters with t-distributed stochastic neighbor embedding (t-SNE) indicated remarkable reduction in M15 and M10, whereas the M7–9 and M11–13 populations were expanded in pRD when compared to HCs (Fig. 1b).

Fig. 1: Immunophenotyping of peripheral blood B cells.

a, Automated B cell subset identification by FlowSOM. Minimal spanning trees of a representative HC and a patient with pRD are shown. Metaclusters, corresponding to individual B cell subsets are indicated by the background color of the nodes, numbers (M0–15) and marker expression pattern. b, B cell subset composition. Metaclusters from concatenated live B cells of four HCs and four patients with pRD were projected onto t-SNE space and assigned as in a. c, Frequencies of the specific B cell subsets. Circles represent antigen-experienced donors, HCs (HC-Ag, n = 18) and patients with pRD (pRD-Ag, n = 13); triangles represent antigen-naive infant donors (HC-N, n = 4; pRD-N n = 1). Subsets comprising Trans, CD38int, CD38+, CD38−, CD27+, DN and ASC populations are depicted with different backgrounds. Statistical analyses were performed on individuals with antigens (HC-Ag versus pRD-Ag) (Mann–Whitney U-test with Holm–Šídák multiple comparison). Asterisks indicate new B cell subsets for a–c. d, Activation of peripheral blood B cells. Surface expression of CD69, CD25, HLA-DR, CD80, CD86, CD21 and CXCR5 in CD38int, CD38− and CD27+ B cells is shown as histograms. Filled gray histograms depict the expression of each marker on the total B cells from an HC. Dark gray and red lines represent the expressions of each marker in the indicated compartments (CD38int, CD38− or CD27+) from an HC and a patient with pRD, respectively. Geometric mean fluorescence intensities (gMFIs) are indicated for each HC (black) and pRD B cell subsets (red) and HC total B cells (gray). e, Expression of activation markers. gMFI of each marker from CD38int, CD38− and CD27+ compartments were normalized by donors to the gMFIs of the total B cells from the HC(s) used in each experiment. Data are shown as mean ± s.e.m. with individual values. Therefore, changes in the expression by compartments in HC-Ag (gray, n = 14) and patients with pRD-Ag (red, n = 6–8) are shown as relative to 1 (green dashed line), which represents marker expression level in the total B cells of HCs. Data were analyzed by Mann–Whitney U-test.

Source data

Quantitative analysis confirmed the profound reduction of M15 (CD27−IgD+CD38hiCD24+IgMhiCD21int corresponding to transitional B cells) and M10 (CD27−IgD+CD38intCD24intIgMhiCD21hi, corresponding to mature resting naive B cells in healthy individuals28, referred to as ‘CD38int’) in pRD when compared to HCs (Fig. 1c).

Notably, three CD27−IgD+CD38lo/− metaclusters, M7–9, were expanded in patients. Specifically, M7 and M8 were identified as new specific B cell subsets (CD27−IgDintCD38loCD24+IgMintCD21hi and CD27loIgD+CD38−CD24−IgM+CD21int, respectively) as these populations have not been described in humans, neither in healthy nor pathological conditions (Fig. 1c). The third CD27−IgD+CD38lo/− population (M9) was phenotypically identical to the recently described ‘activated naive’ B cells (CD27−IgDhiCD38−CD24−IgM+CD21lo)28,29. As M7–9 became the predominant fraction of the CD27−IgD+ compartment in pRD, we collectively referred to them as CD38− in our further investigations.

Among CD27-expressing metaclusters (conventional memory compartment and collectively referred as CD27+) we identified M11–14 that expressed IgM, therefore representing non-switched memory (NSM) B cells. M12 corresponded to NSM resting cells (CD27+IgDintCD38−CD24hiIgM+CD21hi) and was expanded in pRD, whereas M14 represented ‘pre-switched memory B cells’ (CD27+IgDloCD38intCD24hiIgM+CD21hi)28. In addition, we defined M11 and M13 as new NSM cells (CD27+IgD+CD38−CD24hiIgM+CD21− and CD27+IgD+CD38loCD24loIgM+CD21−, respectively) that were significantly expanded in pRD (Fig. 1c). CD27+IgDlo/−IgMlo/− switched memory (SM) fraction harbored the resting SM B cells (M1, CD27+IgDloCD38intCD24+IgMloCD21hi)28, which were significantly decreased in pRD and M2–3 (CD27+IgDloCD38−CD24+IgM−CD21hi and CD27+IgDloCD38−CD24intIgM−CD21lo, respectively) (Fig. 1c). These alterations in the CD27+ populations resulted in the dominance of NSM over SM B cells in pRD when compared to HCs.

We identified M6 (CD27−IgD−CD38intCD24intIgMintCD21hi) and M4 (CD27−IgD−CD38−CD24−IgMintCD21lo), as CD27−IgD− double negative (DN) subsets, corresponding to DN1 and DN2, respectively28. Notably, DN1:DN2 (M6:M4) ratio in pRD was skewed toward DN2 (M4) similarly to what is seen in systemic lupus erythematosus30 (Fig. 1c). Finally, a decrease in pRD M0 (antibody-secreting cells (ASCs), CD27hiIgD−CD38hiCD24−IgMloCD21int) was also noticed, although the difference was not statistically significant.

In absolute counts, most B cell subsets were decreased in pRD compared to HCs, reflecting their B cell lymphopenia; however, M9 ‘activated naive’ B cells were significantly elevated (Extended Data Fig. 3g). Analysis of B cell subsets with conventional standard gating (Extended Data Fig. 4a) confirmed that transitional, mature resting naive and SM B cells, were reduced, whereas atypical CD38− naive B cells and marginal zone-like B cells were greatly expanded in pRD (Extended Data Fig. 4b–d).

In addition, all three B cell subsets (CD38int, CD38− and CD27+) defined above displayed increased activation status in pRD as assessed by the expression of CD69, CD80 and CD86 (Fig. 1d,e). We also found significantly elevated levels of CD25 and HLA-DR in the CD38int and the CD27+ compartments of the patients, respectively (Fig. 1d,e). Decrease in the expression of CD21, indicating previous history of B cell activation31, was detected in the patients, offering further evidence of promiscuous B cell activation in pRD (Fig. 1d,e). Finally, we detected significantly lower expression of the follicle homing receptor, CXCR5, on each B cell subset of patients with pRD compared to HCs (Fig. 1d,e).

In summary, the B cell compartment in patients with pRD-Ag displayed remarkable activation and subset dysregulation with expansion of non-conventional B cells. Notably, asymptomatic P1 (pRD-N) did not show these changes, thus, dysregulated B cell maturation with promiscuous activation in pRD is likely a dynamic process that worsens with age and disease state, implying the role of environmental triggers.

Expansion of non-conventional T-bet+ B cells

As a clinically established hallmark of immune dysregulation32, we documented substantial expansion of CD19hiCD21lo B cells in pRD (Fig. 2a,b), which were equally distributed among CD38int, CD38− and CD27+ B cells (Fig. 2c). The CD11chiCXCR5lo population, resembling murine age-associated B cells (ABCs)33, was also present at a higher frequency in pRD and were equally enriched in the CD38int, CD38− and CD27+ compartments (Fig. 2d–f). Although both HC and pRD ABCs contribute to a smaller fraction of total B cells than CD19hiCD21lo cells, they fully overlap with the latter, indicating that ABCs are part of the CD19hiCD21lo fraction (Fig. 2g), hence, they expand in parallel in pRD (Fig. 2h).

Fig. 2: T-bet+ B cells.

a, Detection of CD19hiCD21lo B cells. b,c, Gating for CD19hiCD21lo B cells in a representative HC-Ag and patient with pRD-Ag. Fraction of CD19hiCD21lo B cells in total (b) and CD38int, CD38− and CD27+ B cells (c) in HC-Ags (n = 14) and patients with pRD-Ag (n = 6). d, Detection of ABCs. Gating for ABCs in a representative HC-Ag and patient with pRD-Ag. e,f, Fraction of ABCs in total (e) and CD38int, CD38− and CD27+ B cells (f) in HC-Ags (n = 14) and patients with pRD-Ag (n = 6). g, Distribution of ABCs in conventional and CD19hiCD21lo B cells. Contour plots show total B cells (gray), gate depicts CD19highCD21low B cells and ABCs are overlayed (green) in a representative HC-Ag and patient with pRD-Ag. h, Correlation between CD19hiCD21lo B cells and ABCs. Linear regression line is shown with 95% confidence intervals, Pearson r and P values are shown. Data were obtained from HC-Ags (gray, n = 14) and patients with pRD-Ag (red, n = 6). i, T-bet expressions in total B cells are shown as gMFIs. j,k, Detection (j) and fraction (k) of CD21+CD11c−, CD21loCD11c− and CD21loCD11c+ B cell populations. Gating is shown in a representative HC-Ag and patient with pRD-Ag. l,m, T-bet expression in CD21+CD11c−, CD21loCD11c− and CD21loCD11c+ B cells in a representative HC-Ag and patient with pRD-Ag (l) and shown as gMFIs (m). n–p, Expressions of FcRL4, FcRL5, CD85j and CXCR3 (n), CD95 (o) and BAFF-R (p) in total B cells are shown as gMFIs. Data for i–p were obtained from HC-Ags (n = 7–14) and patients with pRD-Ag (n = 4–5). Data for b,c,e,f,k–o are shown as mean ± s.e.m. with individual values depicted. Statistical analyses were performed using two-sided unpaired Student’s t-test for data on b,c,e,f,I–o or Mann–Whitney U-test with multiple comparison (Holm–Šídák method) for k. q, Overlayed markers (CD19, CD4, T-bet and 4,6-diamidino-2-phenylindole (DAPI)) in a HC-Ags and patients with pRD-Ag. Magnification ×10. Follicular (Fo) and extrafollicular (EF) areas are indicated that were used for higher magnifications. r, T-bet expression by Fo and EF areas. Magnification ×60. s, T-bet expression in B cells by Fo and EF areas. Arrow shows T-bet+ B cells. Magnification ×60.

Source data

In addition to these surface markers, transcription factor T-bet serves an ultimate marker for ABCs33,34. T-bet was indeed expressed at a higher level in pRD B cells than in HCs (Fig. 2i and Extended Data Fig. 4e). Although CD21loCD11c− and CD21loCD11c+ cells were more abundant in pRD than in HCs (Fig. 2j,k) and T-bet expression was increased toward the latter, it was equally elevated in HCs and pRD, indicating that T-bet is induced at similar level in the corresponding populations of healthy individuals or patients (Fig. 2l,m). By assessing T-bet in CD38int, CD38−, CD27+ and DN B cells, we found increased expression in earlier B cell stages (CD38int and CD38−) in pRD, whereas its levels were comparable in CD27+ and DN B cells with those of HCs (Extended Data Fig. 4f–h) in pRD when measured.

In addition, FcRL4, FcRL5, CD85j and CXCR3 ABC-specific marker expressions were higher in total B cells (Fig. 2n) and CD38int, CD38− and CD27+ populations (but not in DN cells) in pRD compared to HCs (Extended Data Fig. 4i). Of note, CD95 expression was increased, whereas B cell-activating factor receptor (BAFF-R) expression was decreased in total B cells (Fig. 2o,p) and in each individual subset (Extended Data Fig. 4j,k) in pRD compared to HCs.

Examination of the spleen biopsy sample of a pRD patient (P5) confirmed white pulp hyperplasia and the presence of previously described non-necrotic epithelioid granulomas35, revealed remarkable reactive and giant follicles, marginal zone hyperplasia and periarteriolar T-zone hyperplasia, which scored high (12) in comparison to what observed in a previous cohort with common variable immunodeficiency and splenectomy36 (Extended Data Fig. 5a,b). Lymphoid hyperplasia was notable for expansion of B cells, especially CD21lo cells outside the germinal centers (Extended Data Fig. 5c,d) along with follicular helper T (TFH) cell accumulation (Extended Data Fig. 5e,f). Compared to the HC, spleen follicles from P5 were hyperplastic and irregularly shaped and they occupied a larger proportion of the splenic tissue (Extended Data Fig. 5g). Notably, we detected T-bet+ B cells in the follicular areas of the patient, whereas they were absent in those of the HC (Fig. 2q–s and Extended Data Fig. 5h). Although T-bet+ B cells were detected in the extrafollicular regions of both P5 and the HC, they were more abundant in the patient (Fig. 2q–s and Extended Data Fig. 5i). Thus, expansion of T-bet+ B cells in the circulation, follicular and extrafollicular spaces is a unique feature of pRD and represents an additional marker of B cell dysregulation.

Restricted BCR repertoire diversity

The prominent alterations in the subset distribution and activation of B cells in pRD prompted us to evaluate the BCR repertoire composition in sorted CD38int, CD38− and CD27+ populations (Extended Data Fig. 6a,b and Supplementary Table 3). As visualized on treemaps, CD38int B cells from pRD expressed oligoclonal repertoires with prominent expansion of 1 to 5 specific VH gene families (Fig. 3a). These expanded VH families were usually dominated by a few expanded unique clones. In contrast, CD38− and CD27+ repertoires were more similar to those of HCs except P13, whose repertoires were dominated VH1-2 genes. Restricted diversity with oligoclonality of the CD38int subset in pRD was further supported by the low ratio of unique to total sequences, lower Shannon’s H and increased Gini–Simpson indices in the CD38int BCR repertoires of the patients, compared to HCs (Fig. 3b–d). Of note, fewer than 5% of the unique sequences accounted for 50% of the total CD38int repertoires in patients P1, 9, 13, 14 and 16 (Fig. 3e) and the most common ten clones comprised a substantial fraction of their CD38int repertoires (mean ± s.e.m., 17.13 ± 6.071), in contrast to HCs (mean ± s.e.m., 1.45 ± 0.39) (Fig. 3f). Together, these data demonstrate decreased diversity and expansion of certain clones in the CD38int repertoire of patients with pRD, whereas their CD38− and CD27+ compartments seemed to be substantially more diverse. Regarding VH, DH and JH gene segment utilization, we found diversified repertoires in the patients (except in P13) (Extended Data Fig. 7a). In addition, despite the high plasma titer of IgM-type 9G4 antibodies in patients, we did not find elevated frequency of VH4–34-carrying clones in the pRD-Ag repertoires; however, it was expanded in the patient with pRD-N (20.92%). As previously published14, distal JH5 and JH6 gene segments were significantly less frequent in all three compartments repertoires reflecting the RAG activity impairment in the patients (Extended Data Fig. 7a).

Fig. 3: BCR repertoire characteristics in patients with pRD.

a, Treemap representation of the diversity and clonality of immunoglobulin heavy-chain repertoires.Each rectangle represents an immunoglobulin heavy-chain clonotype and the size of rectangles is proportional to the relative frequency of each clone in the entire repertoire. Distinct clones sharing the same VH gene are displayed using altered brightness of the same hue indicated by the color scale. Thick borders and colors assign clones into the corresponding VH gene families. Each treemap represents the most abundant 2,500 clones. One antigen-naive healthy infant (HC-N), three representative antigen-exposed HCs (HC-Ag), the antigen-naive infant patient (pRD-N) and the antigen-exposed patients with pRD (pRD-Ag, n = 5) are shown. Labels on the left show participant IDs. Labels on the top indicate B cell compartments. No sufficient amount of CD38− cells was obtained from the infant donors and P9 for sequencing. b, Ratio of unique and total sequence counts. c–e, Shannon’s H (c), Gini–Simpson (1 − D) (d) and Diversity 50 (D50) (e) diversity indices. f, Cumulative frequency of the ten most abundant clones. g, Frequency of SHM in the V segment. Bars represent mean ± s.e.m. for each group with the individual values indicated. h, Mutation distribution. Graph shows replacement mutation frequencies in CDR and FR regions in CD38int, CD38− and CD27+ B cells. i, Isotype distribution. Stacked columns show the percentage of each Ig class in each repertoire by individuals and B cell compartments. Numbers depict participant IDs. Data for each analysis were obtained from HC-Ags (n = 5), HC-N (n = 1), pRD-Ag (n = 5) and pRD-N (n = 1). Statistical analyses were performed only on individuals with antigens (HC-Ag versus pRD-Ag) using a Mann–Whitney U-test. For i, unswitched (IgD + IgM) and switched (IgG + IgA + IgE) sequences were compared.

Source data

Hence, in connection to their altered phenotype and activation of pRD B cells we identified severe diversity restrictions in the pre-immune repertoire of the patients.

Aberrant somatic diversification

Next, we identified a uniformly increased level of somatic hypermutation (SHM) in the CD38int repertoire of patients with pRD-Ag despite that these cells should express the germline version of their V segments with nearly no mutations, as found in the HCs (Fig. 3g). This finding indicates that although CD38int B cells in patients with pRD were phenotypically defined as mature naive resting B cells (IgD+CD21+CD24intCD27−CD38int), some may represent Ag-experienced B cells and cannot be considered bone fide naive cells. Of note, continuum expression of CD27 did not segregate naive and memory B cells with high confidence in pRD (Extended Data Fig. 3g; CD27-IgD plots). Notably, although the patient with pRD-N already acquired some SHM in the CD38int compartment, it did not reach the level seen in patients with pRD-Ag, suggesting that the unusually early-onset SHM of CD38int B cells is a progressive phenomenon in pRD. Notably, although the CD38− compartment represent phenotypically naive B cells (IgD+CD27−), they displayed a similarly elevated level of SHM whether they were obtained from HCs or patients with pRD, suggesting that despite their different abundance in the peripheral blood of HCs and pRD, they share comparable mutational diversification characteristics. Although SHM was elevated in the CD38int and CD38− compartments of patients, its level remained lower in their CD27+ repertoire compared to HCs, indicating impaired mutational diversification and affinity maturation in pRD. As expected, SHM levels in the CD27+ repertoire of both the infant patient and the age-matched control (pRD-N and HC-N, respectively) were similarly lower than in Ag-experienced individuals. SHM frequency was substantially higher in the complementarity-determining regions (CDRs) than in the framework regions (FRs) of the pRD CD38int B cells, suggesting antigen-driven selection (and remained close to zero in the CDRs and FRs of HCs) (Fig. 3h). This pattern was preserved in the CD38− and CD27+ compartments; however, SHM frequency in the latter did not reach that of corresponding HCs, further confirming the absence of proper memory response in pRD (Fig. 3h). Of note, mutations were more frequent in the activation-induced cytidine deaminase (AID) WRCY/RGYW hotspot of the CD38int and CD38− compartments in pRD than in HCs (Extended Data Fig. 7b,c).

Regarding immunoglobulin class-switch recombination (CSR), we detected an abundance of switched transcripts in CD38int repertoires in pRD although with substantial inter-individual variability, whereas it remained negligible in HCs (Fig. 3i). This implies the unforeseen inclusion of antigen-experienced B cells in the phenotypically naive compartment in pRD. In connection to this, we detected unusual co-presence of IgM and IgG on the surface of the patients’ B cells (Extended Data Fig. 7d,e). On the other hand, CSR was significantly lower in the CD27+ repertoires of patients compared to HCs (Fig. 3i).

In summary, we found somatic diversification in the CD38int compartment and subpar SHM and CSR in the CD27+ compartment of the patients with pRD; hence, these findings provide additional evidence for early widespread and dysregulated B cell activation in pRD, with predisposition to impaired humoral effector function, consistent with our previous findings of decreased fraction and number of SM B cells.

Defective tolerance that worsens with age

Frequencies of the polyreactive clones in new emigrant (NE) transitional and CD38int B cell compartments are measures of the efficiency of central and peripheral B cell tolerance, respectively6. Thus, we assessed the abundance of polyreactive clones by analyzing the reactivity of recombinant antibodies cloned from single-sorted NE and CD38int B cells (Extended Data Fig. 8a). Due to the scarcity of circulating NE cells in the patients with pRD-Ag, we managed to assess the central tolerance efficiency only in the asymptomatic infant (P1). We detected remarkably higher frequency of polyreactive clones compared to published age-matched individuals with functional central tolerance37 (Extended Data Fig. 8b,c). We also found decreased usage of distal Jκ gene segments (Jκ4 and 5) in the patients compared to HCs (Extended Data Fig. 8d), (but not in the distal Jλ gene segments (Jλ3); Extended Data Fig. 8e). Besides impaired V(D)J recombination, this observation also illustrates defective receptor editing providing additional evidence for impaired central tolerance in pRD. In contrast, we did not find significant disturbance in the proportion of κ and λ light-chain expression (another potential marker of receptor editing) in either of the B cell compartments of the patients, except transitional cells (Extended Data Fig. 8f–h).

Next, we found elevated fraction of polyreactive clones in the CD38int compartment in all patients compared to HCs, demonstrating a uniform peripheral tolerance defect in pRD (Fig. 4a–c and Supplementary Table 3). We noted significant difference (P = 0.031; Student’s t-test) in the frequency of polyreactive clones between children (n = 3) and adults (n = 3) with pRD (21.1 ± 2.0% mean ± s.e.m.; range 18.8–25.0% and 32.7 ± 3.0% mean ± s.e.m.; range 27.3–37.5%, respectively). Of note, a fraction of the analyzed clones from P12, P14 and P16 carried mutations in their V segment, indicating ongoing somatic diversification and therefore, cannot be considered as resting mature B cells. Nevertheless, the proportion of polyreactivity in unmutated clones was still uniformly higher in all patients compared to HCs, confirming the profound peripheral tolerance defect. HEp-2 reactivity of the expressed clones confirmed peripheral tolerance impairment in pRD (Fig. 4d,e). In addition, HEp-2 cell-based IFA revealed distinct binding of the polyreactive clones (cytoplasmic fibrillar filamentous (P12.1D3), multiple nuclear dots (P13.2G5), cytoplasmic dense speckled (P14.1A10) and homogenous nucleolar (P14.5B7)), suggesting that although they are polyreactive, they bear preferential self-specificities (Fig. 4f).

Fig. 4: Impaired peripheral B cell tolerance in patients with pRD.

a, Polyreactivity of CD38int B cells. Antibodies cloned from CD38int B cells were tested for anti-double-stranded DNA (dsDNA), insulin and lipopolysaccharide (LPS) reactivity in serial dilution. Binding of 20 randomly selected clones is shown on each graph. Thick black lines with blue circles show binding of the two positive controls used in each assay to determine the threshold for positive reactivity (mean of positive controls minus 2 s.d. at 1.0 µg ml−1). Threshold for positive reactivity is shown by a green dashed line. b, Frequency of polyreactive clones by individuals. Pie charts depict the frequency of non-polyreactive (white) and polyreactive (red) clones for each individual and the percentage of polyreactive clones are shown. The fraction of sequences carrying at least two mutations compared to their corresponding germline versions is depicted by dotted patterns. c, Frequency of polyreactivity. d, HEp-2 reactivity of CD38int B cell clones. Cloned antibodies were tested in 5.0 µg ml−1 concentration for anti-HEp-2 cell line lysate with the two positive controls used in each assay. Blank corrected absorbance values were normalized with the mean of positive controls minus 2 s.d. and values above 1 (green dashed line) were considered positive for HEp-2 reactivity. e, Frequency of HEp-2 reactivity. For c–e data are shown as mean ± s.e.m. with individual values and statistical analyses performed on individuals with antigens (HC-Ag versus pRD-Ag) using a two-sided unpaired Student’s t-test. Antibodies were cloned and tested from HC-Ags (n = 3), pRD-N (n = 1) and patients with pRD-Ag (n = 5). The total number of B cell clones tested is indicated for each individual in b. f, HEp-2 immunofluorescence. Cloned antibodies were used in HEp-2 immunofluorescence assays to show target antigen distribution. Representative images for negative and positive stainings are shown from one HC-Ag and three pRD-AG individuals. Individual and antibody clone IDs are shown. Magnification ×16.

Source data

In summary, these data demonstrate that B cell tolerance is defective in pRD, it worsens with age and likely with chronic antigen exposure.

Diversification and persistence of polyreactive clones

Because our findings imply that polyreactive B cell clones escape both central and peripheral tolerance checkpoints and survive, we aimed to evaluate their trajectories and determine whether they reach effector compartments. Therefore, we traced each expressed sequence and their related descendants in the CD38int, CD38− and CD27+ repertoires. Notably, we identified related clones in the CD38int BCR repertoires of P1, P12, P14, P16 (n = 24.8 ± 5.4 clones per donor) (Fig. 5a). Two of these patients (P1 and P14) had persistent polyreactive and non-polyreactive clones that contributed to the entire sequence pool at a remarkable level (Fig. 5b). As expected, no related clones were identified in the naive or effector compartments of the HCs. In contrast, we detected simultaneous presence of related clones in CD38int, CD38− and CD27+ compartments of three patients with pRD-Ag, whereas sister clones were only present in the CD38int compartment of the individual with pRD-N (Fig. 5c). Furthermore, longitudinal analysis of relatedness revealed durable presence of three distinct clones in P14 (1B5, 1C11 and 6E11) 1 year apart, with higher abundance in CD38− or CD27+ compartments at the second time point. Collectively, these data indicate that certain polyreactive (and non-polyreactive) CD38int clones persist chronically in pRD.

Fig. 5: Expansion and diversification of polyreactive clones.

a, Clones with descendants in total repertoire. Graph shows the number of the in vitro expressed clones by donors (gray bars). Number of clones with identified descendants in total repertoire are depicted with deep red. Data are shown from HC-Ags (n = 3), pRD-N (n = 1) and patients with pRD-Ag (n = 5) with individual IDs indicated. b, Contribution of descendants to repertoire. Fraction of the descendants of clones in total repertoires are shown by individuals. Descendants of the polyreactive and non-polyreactive clones are depicted with filled and striped bars, respectively. c, Compartmental distribution of the descendants.