{"paper_id":"18bd41ab-9764-4aae-91ec-3e51e668d686","body_text":"Title: Distributed Clonal Deletion Prevents Autoimmune Disease Progression \nAuthors: Anna M. Newen1, †, Uzair A. Ansari1, †, Mikala J. Simpson1, †, Fiona Flynn1, Sukriti \nSharma1, Michael Ivanov1, Ishaan Antani1, Dylan Pfannenstiel1, Baktiar Karim2, Laura Bassel2, \nBrian Capaldo3, Qingrong Chen3, Daoud Meerzaman3, Indu Raman4, Chengsong Zhu4, Cornelius \nY. Taabazuing5, Hamid Kashkar6, and Christian T. Mayer1* \nAffiliations: \n1Experimental Immunology Branch, Center for Cancer Research, National Cancer Institute, \nNational Institutes of Health; Bethesda, MD 20892, USA \n2Molecular Histopathology Laboratory, National Cancer Institute, National Institutes of Health, \nFrederick, MD 21702, USA \n3Computational Genomics and Bioinformatics Branch, National Cancer Institute, National \nInstitutes of Health, Frederick, MD 20850, USA \n4Microarray and Immune Phenotyping Core Facility, University of Texas, Southwestern Medical \nCenter, Dallas, TX 75235, USA  \n5Department of Biochemistry and Biophysics, University of Pennsylvania, Philadelphia, PA \n19104, USA \n6Institute for Molecular Immunology, Center for Molecular Medicine Cologne, CECAD \nResearch Center, Medical Faculty, University of Cologne, 50935 Cologne, Germany \n*Corresponding author. Email: christian.mayer@nih.gov   \n† These authors contributed equally to this work \n105 and is also made available for use under a CC0 license. \n(which was not certified by peer review) is the author/funder. This article is a US Government work. It is not subject to copyright under 17 USC \nThe copyright holder for this preprintthis version posted May 16, 2026. ; https://doi.org/10.64898/2026.05.12.724341doi: bioRxiv preprint \n\nAbstract:  \nSelf-reactive B cells are generated during normal development and can acquire increased \npathogenicity through activation-induced cytidine deaminase (AID)-mediated diversification \nfollowing activation. Clonal deletion is thought to eliminate these cells, yet how deletion is \ndistributed across developmental and activation stages to prevent autoimmune disease remains \nunclear. Here, we show that clonal deletion is enforced through temporally distinct mitochondrial \napoptosis (MOMP) checkpoints that differentially regulate autoreactive B cell fate and disease \nprogression. Using conditional Bcl-2 expression to inhibit MOMP either before or after B cell \nactivation, we find that early inhibition permits the survival and maturation of autoreactive B \ncells after peripheral egress, expanding the pool of cells available for activation. These cells \nsubsequently undergo AID-dependent diversification, producing class-switched IgG \nautoantibodies with expanded antigen breadth that target a wider range of self-antigens and drive \nlethal, female-biased autoimmune disease characterized by complement activation and kidney \npathology. In contrast, inhibition of MOMP only after activation allows the accumulation of \ngerminal center, switched memory, and plasma cells and promotes autoantibody production, but \nresults in more restricted IgG autoreactivity, limited complement activation and limited tissue \ndamage, and normal survival. Notably, early MOMP inhibition does not expand immature bone \nmarrow B cells, indicating that a major clonal deletion checkpoint operates in the periphery \nrather than during initial B cell generation. Together, these findings support a Distributed Clonal \nDeletion Model in which early checkpoints restrict the entry of autoreactive B cells into \ndiversification pathways, while later checkpoints limit the persistence of diversified autoreactive \nclones, thereby constraining autoimmune disease progression. \n \n105 and is also made available for use under a CC0 license. \n(which was not certified by peer review) is the author/funder. This article is a US Government work. It is not subject to copyright under 17 USC \nThe copyright holder for this preprintthis version posted May 16, 2026. ; https://doi.org/10.64898/2026.05.12.724341doi: bioRxiv preprint \n\nOne Sentence Summary: Distributed clonal deletion prevents autoimmune disease progression \nby restricting the breadth of autoreactive clones entering immune responses, with early MOMP \ncheckpoints limiting diversification and later checkpoints constraining persistence. \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n105 and is also made available for use under a CC0 license. \n(which was not certified by peer review) is the author/funder. This article is a US Government work. It is not subject to copyright under 17 USC \nThe copyright holder for this preprintthis version posted May 16, 2026. ; https://doi.org/10.64898/2026.05.12.724341doi: bioRxiv preprint \n\nMain Text:  \nINTRODUCTION \nSystemic autoimmune diseases such as systemic lupus erythematosus (SLE) arise when immune \ntolerance fails and B cells generate antibodies that target self-tissues. These conditions \ndisproportionately affect women and can lead to chronic inflammation, renal failure, and organ \ndamage. Autoreactive B cells are abundant in healthy individuals ( 1, 2), yet most never produce \nautoreactive antibodies. Additionally, some individuals produce autoantibodies without \ndeveloping disease ( 3-5). These three levels of tolerance - autoreactive B cells, autoantibodies, \nand autoimmune disease - remain incompletely understood. Defining how tolerance mechanisms \noperate at each level and cooperate to prevent disease progression is essential for understanding \nwhy tolerance breaks down in autoimmunity. \n \nB cell development generates a diverse preimmune repertoire through recombination-activating \ngene (RAG)–mediated V(D)J recombination in the bone marrow. Pro-B cells (Fraction B/C) \nrearrange immunoglobulin heavy chains and proliferate as large pre-B cells (Fraction C\n′ ), \nfollowed by light-chain rearrangement in small pre-B cells (Fraction D), and progression to \nimmature B cells (Fraction E) ( 6-8). After leaving the bone marrow, immature B cells mature \nthrough transitional stages and enter the follicular (FO) or marginal zone compartments in the \nspleen (8-10). A longstanding but unresolved question is whether immature B cells destined for \ndeletion die within the bone marrow or after peripheral egress (11-13), a distinction that has been \ndifficult to resolve because dying cells are rapidly cleared in vivo. \n \n105 and is also made available for use under a CC0 license. \n(which was not certified by peer review) is the author/funder. This article is a US Government work. It is not subject to copyright under 17 USC \nThe copyright holder for this preprintthis version posted May 16, 2026. ; https://doi.org/10.64898/2026.05.12.724341doi: bioRxiv preprint \n\nUpon antigen encounter in T cell–dependent immune responses, FO B cells express activation-\ninduced cytidine deaminase (AID). AID mediates somatic hypermutation (SHM), primarily \nwithin germinal centers (GCs), which can generate or enhance autoreactivity, and it mediates \nclass-switch recombination (CSR), which frequently occurs early during immune responses and \ncan take place outside GCs ( 14-17). SHM alters antigen specificity, whereas CSR changes \nantibody isotype and thereby modifies effector properties. Because both RAG and AID can \ngenerate or modify autoreactivity, multiple apoptosis checkpoints censor autoreactive B cells \n(12, 13, 18-25 ), and  genetic models that attenuate apoptosis are associated with autoimmune \ndisease ( 26-32). Yet, autoreactivity and autoantibody production do not always progress to \ndisease ( 1-5, 33 ). Whether checkpoints before and after activation operate independently or \ncooperatively remains unclear, in part because prior studies of individual tolerance checkpoints \nhave not directly examined how the timing of checkpoint failure across B cell compartments \ninfluences progression from autoreactivity to autoimmune disease.  \n \nWe hypothesized that clonal deletion is distributed across developmental and activation stages \nand that checkpoint timing determines disease outcomes by controlling which autoreactive B \ncells undergo AID-dependent diversification. Early checkpoints acting before B cell activation \nwould restrict low-avidity autoreactive clones from undergoing CSR and SHM, whereas late \ncheckpoints acting after activation would limit the persistence of already-diversified clones but \nact on an already-restricted substrate. Under this model, failure of early checkpoints would \nbroaden the class-switched autoreactive repertoire and drive autoimmune disease progression, \nwhereas failure of late checkpoints would permit autoreactive cell accumulation and \nautoantibody production without causing organ damage. \n105 and is also made available for use under a CC0 license. \n(which was not certified by peer review) is the author/funder. This article is a US Government work. It is not subject to copyright under 17 USC \nThe copyright holder for this preprintthis version posted May 16, 2026. ; https://doi.org/10.64898/2026.05.12.724341doi: bioRxiv preprint \n\nTo test this hypothesis, we used identical Rosa26 LSL-Bcl2 alleles driven by distinct Cre \nrecombinases to express Bcl-2 either beginning early in B cell development (Bcl2 Early) or only \nafter activation (Bcl2 Late). Bcl-2 inhibits mitochondrial outer membrane permeabilization \n(MOMP), enabling direct comparison of the consequences of disrupting early versus late \nMOMP-sensitive tolerance checkpoints. We combined these models with in vivo EdU pulse–\nchase labeling, comprehensive disease assessment, and autoantibody profiling to define how \ncheckpoint timing regulates B cell progression, AID-dependent diversification, and autoimmune \ndisease outcomes. \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n105 and is also made available for use under a CC0 license. \n(which was not certified by peer review) is the author/funder. This article is a US Government work. It is not subject to copyright under 17 USC \nThe copyright holder for this preprintthis version posted May 16, 2026. ; https://doi.org/10.64898/2026.05.12.724341doi: bioRxiv preprint \n\n \nRESULTS  \nMOMP inhibition before activation, but not after activation, causes autoimmune disease \nWe generated mice expressing human Bcl-2 (hBcl-2) and GFP from the Rosa26 LSL-Bcl2 allele \nusing Mb1 Cre or Aicda Cre to inhibit mitochondrial outer membrane permeabilization (MOMP) \neither from early B cell development (Bcl2 Early) or only after activation (Bcl2 Late). Flow \ncytometry confirmed stage-specific Cre recombination (fig. S1, S2A), and germinal center (GC), \nswitched memory B (swMem), and plasma cells (PC) expressed comparable hBcl-2 levels (fig. \nS2B, C), enabling direct comparison. \nTo determine how the timing of MOMP inhibition influences autoimmune disease progression, \nwe monitored disease development in Bcl2Early, Bcl2Late, and control cohorts. Only Bcl2Early mice \ndeveloped clinical illness, with reduced survival and a higher incidence of disease in females \n(Fig. 1A, B). Renal histopathology revealed widespread glomerulonephritis and tubular injury in \nBcl2\nEarly mice, characterized by glomerular hypercellularity, capillary wall thickening, and \ncrescent formation, together with inflammatory infiltrates (Fig. 1C). Quantitative scoring \nconfirmed significantly greater pathology in this group than in Bcl2\nLate and control mice, with \nmore severe changes in females (Fig. 1D–F). Immunoglobulin (Ig) deposition within glomeruli \nwas comparable between Bcl2 Early and Bcl2 Late mice, whereas cortical Ig deposition was \nincreased in Bcl2 Early kidneys (Fig. 1G), indicating that differences in disease severity are not \nexplained by glomerular Ig deposition alone. In contrast, complement C3d deposition was \nincreased only in Bcl2\nEarly mice (Fig. 1H). Serum urea measurements indicated impaired renal \nfunction in a subset  of Bcl2 Early m i c e  ( 2 5 %  o f  f e m a l e s ;  0 %  o f  m a l e s )  ( F i g .  1 I ,  J ) ,  w h e r e a s  \nBcl2Late mice maintained normal urea levels despite the presence of autoreactive antibodies (Fig. \n105 and is also made available for use under a CC0 license. \n(which was not certified by peer review) is the author/funder. This article is a US Government work. It is not subject to copyright under 17 USC \nThe copyright holder for this preprintthis version posted May 16, 2026. ; https://doi.org/10.64898/2026.05.12.724341doi: bioRxiv preprint \n\n1I). Together, these findings demonstrate that disruption of early, but not post-activation, MOMP \ncheckpoints drives autoimmune disease progression, whereas post-activation checkpoint failure \npermits autoreactivity and autoantibody production without progression to organ damage. \n \nBcl-2 expression during development restricts IgG autoantibody breadth  \nTo determine whether checkpoint timing affects autoreactive antibody specificity, we performed \nautoantigen array analysis (Fig. 2A-C). Bcl2\nEarly mice exhibited a marked expansion of IgG \nautoreactivity, with 23 significantly increased autoantibodies compared to 5 in Bcl2 Late mice \nrelative to controls (Fig. 2A, C and Supplementary Data 1). In contrast, IgM autoreactivity was \nincreased but more comparable between Bcl2 Early and Bcl2 Late groups (Fig. 2B, C and \nSupplementary Data 2). Consistent with this, 13 IgG autoantibodies were significantly increased \nin Bcl2Early compared with Bcl2Late mice, whereas only a single IgM reactivity differed (Fig. 2C), \nindicating that early checkpoint failure selectively expands class-switched autoreactive breadth \nrather than overall autoreactivity. Notably, among the most expanded IgG reactivities in Bcl2Early \nmice were antibodies targeting complement C3 and PCNA, which have been associated with \nsevere nephritis in subsets of patients with systemic lupus erythematosus ( 34, 35 ). Total \nautoreactive IgG titers and dsDNA/ssDNA ELISAs were similar between Bcl-2 models (fig. \nS3A–E), with ELISA confirmation of anti-C3 and anti-PCNA IgG autoantibody differences (fig. \nS3F, G).  \nTo independently assess autoreactive repertoires, we performed phage immunoprecipitation \nsequencing (PhIP-seq).  Although numerous peptide reactivities were detected across all groups, \nincluding controls (fig. S3H), gene set enrichment analysis (GSEA) of peptide-associated genes \nrevealed qualitative differences between genotypes. Bcl2\nEarly mice showed selective enrichment \n105 and is also made available for use under a CC0 license. \n(which was not certified by peer review) is the author/funder. This article is a US Government work. It is not subject to copyright under 17 USC \nThe copyright holder for this preprintthis version posted May 16, 2026. ; https://doi.org/10.64898/2026.05.12.724341doi: bioRxiv preprint \n\nfor gene sets related to lipid translocation, intracellular ligand-gated ion channel activity, \ntranscriptional coactivator function, and immune response–associated transcriptional programs \n(Supplementary Data 3), whereas Bcl2 Late mice did not show enrichment of these categories. \nThese findings indicate distinct functional profiles of autoreactive IgG repertoires despite \nbroadly comparable numbers of individual peptide reactivities across groups. \nFocusing on peptides enriched in at least two of three Bcl2Early mice but absent from Bcl2Late and \ncontrol mice identified 12 Bcl2 Early-specific reactivities, including Prkcz, which is associated \nwith autoimmune myopathies in humans (36)(Fig. 2D).  \nTogether, these data demonstrate that early checkpoint failure selectively widens class-switched \nautoreactive IgG breadth, whereas late checkpoint failure permits autoreactivity without broad \nclass-switched diversification.  \n \nAID is required for autoimmune pathology when MOMP is inhibited before activation \nGiven the increased breadth of class-switched autoantibodies in Bcl2 Early mice, we next tested \nwhether AID-dependent processes are required for autoimmune disease in this setting. To do so, \nwe analyzed Bcl2Early mice lacking AID (encoded by Aicda), thereby preventing SHM and CSR. \nStrikingly, Bcl2 EarlyAicda-/- mice remained free of autoimmune disease and showed no renal \ninjury (Fig. 3A–G), despite exhibiting the same early checkpoint disruption as Bcl2 Early mice. \nHistopathologic evaluation revealed infection-associated mammary gland abscesses (Fig. 3B, C), \nbut no autoimmune pathology (Fig. 3D-G). One Bcl2 Early mouse developed B cell lymphoma \n(Fig. 3B, C). These findings demonstrate that AID-dependent diversification is required for \nautoimmune disease in the setting of early MOMP-regulated checkpoint failure, and that \n105 and is also made available for use under a CC0 license. \n(which was not certified by peer review) is the author/funder. This article is a US Government work. It is not subject to copyright under 17 USC \nThe copyright holder for this preprintthis version posted May 16, 2026. ; https://doi.org/10.64898/2026.05.12.724341doi: bioRxiv preprint \n\nautoreactive B cell survival without CSR/SHM is insufficient to cause autoimmune organ \ndamage. \nMOMP inhibition before activation expands peripheral B cell populations \nTo determine how early MOMP checkpoint disruption alters the B cell pool available for \nactivation, we analyzed B cell development in Bcl2Early and Bcl2Late mice (Fig. 4A). Despite Bcl-\n2 expression beginning at the pro-B stage, bone marrow pro-B, pre-B, and immature B cells were \nnot expanded in Bcl2\nEarly mice, whereas a reduction in large pre-B cells (Fraction C ′ ) was noted, \nconsistent with decreased proliferation (Fig. 4B). \nIn contrast, Bcl2 Early mice displayed significant expansion of splenic transitional (T1, T2), \nanergic (T3), and follicular (FO) B cells, as well as increased recirculating mature B cells in the \nbone marrow (Fig. 4B, C). Bcl2Late mice resembled controls across these compartments (Fig. 4B, \nC). Both Bcl2 Early and Bcl2Late mice accumulated post-activation germinal center (GC) B cells, \nswitched memory B cells (swMem), and plasma cells (PC). These findings indicate that early, \nbut not post-activation, MOMP inhibition selectively expands peripheral immature and mature B \ncell populations. \n \nImmature B cell deletion occurs predominantly after bone marrow egress \nThe absence of immature B cell expansion in Bcl2\nEarly mice suggested either limited apoptosis at \nthis stage or accelerated marrow egress. To test the latter, we performed vascular labeling of \nbone marrow sinusoids ( 11)(Fig. 4D). Immature B cells labeled by intravenously injected anti-\nCD45R/B220-PE-Cy7 did not increase in proportion or number in Bcl2 Early mice relative to \ncontrols (Fig. 4E, F), excluding enhanced marrow egress as a cause of the normal immature pool. \n105 and is also made available for use under a CC0 license. \n(which was not certified by peer review) is the author/funder. This article is a US Government work. It is not subject to copyright under 17 USC \nThe copyright holder for this preprintthis version posted May 16, 2026. ; https://doi.org/10.64898/2026.05.12.724341doi: bioRxiv preprint \n\nIncreased numbers of labeled mature B cells (Fraction F) were detected, reflecting expansion and \nincreased bone marrow egress of this compartment (Fig. 4B, E, F). \nTo determine whether non-apoptotic programmed cell death contributes to immature B cell loss, \nwe analyzed Gsdmd -/-Gsdme-/-Mlkl-/- triple-deficient mice, which are unable to undergo \npyroptosis and necroptosis (Fig. 4G). Bone marrow and splenic B cell subsets were comparable \nto controls with only modest changes in Fraction C\n′ , T3, and swMem cells (Fig. 4H, I), while \nmacrophages resisted pyroptotic and necroptotic stimuli (fig. S4A). These findings rule out \npyroptosis and necroptosis, supporting a MOMP-sensitive peripheral deletion mechanism. \n \nTransitional B cells are the first peripheral stage regulated by MOMP  \nTo define how MOMP inhibition influences developmental progression, we performed EdU \npulse–chase labeling (Fig. 5A). In control mice, EdU labeled Fraction C\n′  within two hours and \nappeared in Fraction E, T1, T2, and T3 by day 3, with labeled transitional and anergic \npopulations contracting by day 7 and labeled FO and Fraction F cells persisting (Fig. 5B, C). \nIn Bcl2 Early mice, EdU labeling of Fraction C ′  cells was reduced at two hours (Fig. 5B), \nconsistent with decreased proliferation. Fraction E labeling on day 3 was unchanged, whereas \nsplenic EdU/i2  T1, T2, and T3 cells were significantly increased at day 3 and remained elevated \nat day 7 (Fig. 5C). Early FO cells appeared similarly at day 3; however, by day 7, FO and \nFraction F B cells were significantly expanded in Bcl2 Early mice (Fig. 5C). These findings \nidentify transitional and anergic B cells as the first peripheral populations whose survival is \nlimited by MOMP. \n \n105 and is also made available for use under a CC0 license. \n(which was not certified by peer review) is the author/funder. This article is a US Government work. It is not subject to copyright under 17 USC \nThe copyright holder for this preprintthis version posted May 16, 2026. ; https://doi.org/10.64898/2026.05.12.724341doi: bioRxiv preprint \n\nFO-intrinsic MOMP inhibition is insufficient to account for peripheral expansion  \nTo assess the contribution of FO-intrinsic MOMP inhibition, we examined CD21 CreRosa26LSL-\nBcl2 (Bcl2Int) mice, in which recombination is efficient in FO B cells and partial in T2 and anergic \nT3 cells ( 37)(fig. S4B). FO B cells were increased in Bcl2 Int mice relative to controls but \nremained substantially fewer than in Bcl2Early mice (Fig. 5D). Transitional and anergic T3 B cell \nnumbers in Bcl2 Int mice did not differ from controls (Fig. 5E and fig S4C, D). These findings \nindicate that FO-intrinsic MOMP inhibition contributes to but is insufficient to account for the \nBcl2Early phenotype, highlighting the importance of early peripheral checkpoints acting at \ntransitional stages. \nTogether, these findings suggest that early MOMP checkpoint disruption permits the survival of \nautoreactive B cells after peripheral egress, leading to expansion of transitional and follicular B \ncell populations, increased class-switched IgG breadth, and AID-dependent autoimmune disease, \nwhereas post-activation checkpoint failure permits autoreactivity without progression to \npathology. \n \n \n \n \n \n \n \n105 and is also made available for use under a CC0 license. \n(which was not certified by peer review) is the author/funder. This article is a US Government work. It is not subject to copyright under 17 USC \nThe copyright holder for this preprintthis version posted May 16, 2026. ; https://doi.org/10.64898/2026.05.12.724341doi: bioRxiv preprint \n\n \n \n \n \nDISCUSSION  \nThis study demonstrates that the timing of MOMP-sensitive checkpoints determines whether \nautoreactive B cells progress to autoimmune disease. Inhibition of MOMP before B cell \nactivation permits the survival and expansion of peripheral B cell populations, leading to \nincreased class-switched autoreactive IgG breadth and severe, AID-dependent autoimmune \npathology. In contrast, MOMP inhibition restricted to activated and post-germinal center stages \nallows autoreactive B cell accumulation and autoantibody production but does not result in organ \ndamage. These findings provide a mechanistic explanation for how autoreactivity and \nautoantibodies can be present in the absence of clinical disease (1, 3, 4). \n \nOur results extend prior work on apoptosis checkpoints in B cells by directly comparing the \nconsequences of disrupting early versus late MOMP-sensitive stages. Previous studies \ndemonstrated apoptosis in developing and germinal center B cells ( 11, 38-40) and identified a \npost-germinal center checkpoint that limits autoreactive plasma cell and memory B cell survival \n(33, 41). However, how checkpoint timing influences progression from autoreactivity to \nautoimmune disease remained unresolved. Here, we show that early checkpoint disruption \nuniquely permits disease progression, whereas late checkpoint disruption alone is insufficient in \n105 and is also made available for use under a CC0 license. \n(which was not certified by peer review) is the author/funder. This article is a US Government work. It is not subject to copyright under 17 USC \nThe copyright holder for this preprintthis version posted May 16, 2026. ; https://doi.org/10.64898/2026.05.12.724341doi: bioRxiv preprint \n\nour model, establishing that these checkpoints make non-equivalent contributions to immune \ntolerance. \n \nMechanistically, early MOMP inhibition expands transitional and follicular B cell populations \nwithout altering immature bone marrow subsets, indicating that a major clonal deletion \ncheckpoint operates after peripheral egress. This extends prior apoptotic cell quantification ( 11) \nby defining functional consequences through genetic loss-of-function approaches. EdU tracing \nfurther identifies transitional B cells as the first stage at which survival is MOMP-limited, with \ndownstream expansion of follicular populations. These findings suggest that early peripheral \ncheckpoints constrain the size and composition of the B cell pool entering immune responses, \nwhereas later checkpoints primarily regulate the persistence of activated clones. \n \nThe selective expansion of class-switched autoreactive IgG breadth in Bcl2\nEarly mice provides a \nfunctional link between checkpoint timing and disease outcome. Despite comparable IgM \nautoreactivity and similar total autoreactive IgG titers, early checkpoint disruption broadened \nIgG specificity, including reactivities associated with severe SLE manifestations ( 34, 35 ). In \ncontrast, Bcl2Late mice exhibited autoreactive antibodies without comparable IgG diversification \nor pathology. Together with the complete protection observed in AID-deficient Bcl2 Early mice, \nthese findings demonstrate that AID-dependent diversification is required for disease and that \nsurvival of autoreactive B cells alone is insufficient to drive pathology in this model. AID has \nalso been implicated in B cell tolerance ( 42-45), and AID deficiency has been reported to either \nattenuate (46, 47) or exacerbate (48, 49) autoimmune pathology depending on the model. In our \nsystem, AID deficiency completely prevented disease despite early checkpoint disruption, \n105 and is also made available for use under a CC0 license. \n(which was not certified by peer review) is the author/funder. This article is a US Government work. It is not subject to copyright under 17 USC \nThe copyright holder for this preprintthis version posted May 16, 2026. ; https://doi.org/10.64898/2026.05.12.724341doi: bioRxiv preprint \n\nindicating that the pathogenic effects of AID-dependent diversification outweigh potential \ntolerance-promoting functions under these conditions. These findings suggest that the impact of \nAID on autoimmunity is context-dependent and shaped by the availability of autoreactive \nprecursors. \n \nNotably, immunoglobulin deposition within glomeruli was observed in both Bcl2 Early and \nBcl2Late mice, yet complement activation and renal pathology occurred only in Bcl2 Early animals. \nThis dissociation suggests qualitative differences in the autoreactive antibody response rather \nthan differences in total antibody deposition. One possibility is that the antigenic targets differ, as \nBcl2Early mice exhibit broader IgG reactivity, including antibodies such as anti-C3 that have been \nshown to enhance complement activation by interfering with regulatory mechanisms or \nstabilizing C3 convertases (50, 51). Alternatively, differences in antibody isotype may influence \ncomplement fixation and pathogenic potential. These considerations indicate that the \ncomposition of the autoreactive repertoire, rather than the presence of immune complexes alone, \nis a key determinant of downstream tissue injury. \n \nThese observations support a model in which clonal deletion is distributed across pre-activation \nand post-activation stages. Because receptor editing efficiently removes high-avidity autoreactive \nB cells in the bone marrow with minimal reliance on deletion ( 11, 52, 53), we propose that early \nperipheral checkpoints limit the entry of low-affinity or low-avidity autoreactive B cells into \nimmune responses, thereby restricting the breadth of class-switched autoreactive repertoires, \nwhereas later checkpoints constrain the persistence and expansion of autoreactive clones once \nactivated (Fig. 6). Failure of early checkpoints expands the pool of cells available for \n105 and is also made available for use under a CC0 license. \n(which was not certified by peer review) is the author/funder. This article is a US Government work. It is not subject to copyright under 17 USC \nThe copyright holder for this preprintthis version posted May 16, 2026. ; https://doi.org/10.64898/2026.05.12.724341doi: bioRxiv preprint \n\ndiversification and enables pathogenic autoantibody responses, while failure of late checkpoints \npermits autoreactivity without disease progression (Fig. 6). This framework, which we term the \nDistributed Clonal Deletion Model , provides a unifying explanation for how substantial \nautoreactivity can be tolerated in healthy individuals yet lead to disease when early tolerance \nmechanisms fail.  \n \nThese findings have implications for human autoimmune disease, where defects in early \ntolerance checkpoints or altered survival of autoreactive precursors may expand the repertoire \navailable for pathogenic diversification. Targeting pathways that promote autoreactive B cell \ndeletion or limit class-switched autoantibody diversification and complement activation, rather \nthan global B cell depletion, may therefore represent a strategy to reduce autoimmune pathology \nwhile preserving immune competence. \n \nLimitations of the study \nThis study has several limitations. Bcl-2 inhibits MOMP but may also affect additional cellular \nprocesses, and thus our findings define MOMP-sensitive checkpoints rather than apoptosis-\nspecific mechanisms. In addition, although AID dependence is demonstrated, the relative \ncontributions of SHM and CSR were not resolved. \n \n \n \n \n105 and is also made available for use under a CC0 license. \n(which was not certified by peer review) is the author/funder. This article is a US Government work. It is not subject to copyright under 17 USC \nThe copyright holder for this preprintthis version posted May 16, 2026. ; https://doi.org/10.64898/2026.05.12.724341doi: bioRxiv preprint \n\n \n \n \n \nMATERIALS AND METHODS \n \nStudy Design \nThis study compared the effect of MOMP inhibition before or after B cell activation on B cell \naccumulation, autoantibody formation, and autoimmune pathology. Experiments were designed \nto analyze B cell subset composition, developmental kinetics, serological, and pathological \noutcomes in Bcl2\nEarly, Bcl2 Late, and control mice. Both male and female mice were analyzed. \nSample sizes were based on prior studies demonstrating sufficient statistical power to detect \ndifferences in B cell subset frequencies and autoantibody reactivity. No animals or data points \nwere excluded. \n \nAnimals \nAicdaCre (54), C57Bl/6J, Gsdmd -/- (55), Gsdme-/- (56) and Mb1 Cre (57) mice were from Jackson \nLaboratories. Aicda -/- (58) mice were provided by Dr. Michel Nussenzweig (The Rockefeller \nUniversity). CD21 Cre ( 37) mice by Dr. Jagan Muppidi (NCI). Mlkl -/- (59) mice by Dr. James \nMurphy (WEHI) and Dr. Alan Sher (NIAID), and Rosa26LSL-Bcl2-IRES-GFP mice (60) by Dr. Hamid \nKashkar (University of Cologne). All mice were on C57Bl/6 background. Some mice were \ninjected intravenously with 1mg 5-ethynyl-2'-deoxyuridine (EdU) (ThermoFisher, A10044) 2h, \n105 and is also made available for use under a CC0 license. \n(which was not certified by peer review) is the author/funder. This article is a US Government work. It is not subject to copyright under 17 USC \nThe copyright holder for this preprintthis version posted May 16, 2026. ; https://doi.org/10.64898/2026.05.12.724341doi: bioRxiv preprint \n\n3d or 7d prior to analysis, or with 2µg anti-CD45R/B220-PE-Cy7 (ThermoFisher, 25-0452-82) \n2min prior to analysis. Mice were maintained under specific pathogen-free conditions at NCI \n(Frederick and Bethesda). All procedures were approved by the NCI Animal Care and Use \nCommittee (ACUC) and conformed with federal regulatory requirements and standards. The \nintramural NIH ACU program is accredited by AAALAC International.  \n \nPathology  \nTissues were fixed in buffered 10% formalin (Azer scientific, PFNBF-120) for 48 hours at room \ntemperature and stored in 70% ethanol prior to paraffin embedding. Sections were stained with \nhematoxylin & eosin and evaluated by a board-certified veterinary pathologist. All slides were \nevaluated in a blinded fashion using an Olympus BX46 microscope at 200× magnification. \nImages were captured using a Nikon DS-Ri2 camera. Two pathologists, blinded to genotype, \ngraded glomerulopathy on a scale from 0 to 4 based on the extent and severity of glomerular \nchanges. Grade 0 represented normal glomeruli, without abnormalities. Grade 1 indicated \nminimal hypercellularity involving less than 25% of glomeruli. Grade 2 was defined by mild \nhypercellularity affecting 25-50% of the glomerular tufts. Grade 3 denoted moderate \nhypercellularity in greater than 50 to 75% of glomeruli, with or without crescent formation and \nthickening of basement membrane. Grade 4 represented marked hypercellularity in greater than \n75% of glomeruli, accompanied by crescent formation and thickening of basement membrane. \nCortical tubular lesions were graded on a scale from 0 to 4 based on the extent of cortical \ninvolvement. Grade 0 (normal) indicated normal tubules without any pathological changes. \nGrade 1 (minimal) represented tubular lesions, including degeneration and/or regeneration, \naffecting less than 10% of the cortical area. Grade 2 (mild) involved tubular lesions affecting \n105 and is also made available for use under a CC0 license. \n(which was not certified by peer review) is the author/funder. This article is a US Government work. It is not subject to copyright under 17 USC \nThe copyright holder for this preprintthis version posted May 16, 2026. ; https://doi.org/10.64898/2026.05.12.724341doi: bioRxiv preprint \n\nmore than 10% but less than 25% of the cortical tubules. Grade 3 (moderate) indicated lesions \ninvolving more than 25% but less than 50% of the cortical tubules. Grade 4 (severe) denoted \ntubular lesions affecting more than 50% of the cortex. Additionally, lymphoplasmacytic \ninfiltrates in the kidney were semi-quantitatively graded on a scale from 0 to 4, where 0 was \nnormal, 1 was minimal, 2 was mild, 3 was moderate, and 4 was severe.  \n \nEnzyme-linked immunosorbent assay (ELISA) \nSerum IgG autoantibodies were measured by ELISA as described ( 41) using peroxidase-\nconjugated goat anti-mouse IgG Fc (Jackson ImmunoResearch, 115-035-164). Self-antigens \nincluded dsDNA (Sigma, D4522), single stranded DNA (ssDNA, prepared from dsDNA), mouse \nC3 (Complement Technology, M113), and human PCNA (Novus Biologicals, NBC1-18428). \nSerum was tested at a 1:640 dilution and at three four-fold dilutions (dsDNA, ssDNA), or at a \n1:160 dilution (C3, PCNA). Monoclonal control antibodies (mGO53, non-reactive; ED38, highly \npolyreactive) with mouse IgG1 constant regions were tested at 4µg/ml and at three four-fold \ndilutions. Absorbance at 405nm was measured with a SpectraMax iD3 Multi-Mode Microplate \nReader (Molecular Devices). A\n405 values were corrected for PBS-only signal and area under the \ncurve (AUC) values were calculated using GraphPad Prism.  \n \nKidney function \nSerum urea concentrations were measured with the Urea Nitrogen (BUN) Colorimetric Detection \nKit (Thermo Fisher Scientific, EIABUN) according to the manufacturer’s instructions. \nMeasurements were performed with a SpectraMax iD3 Multi-Mode Microplate Reader \n(Molecular Devices). \n105 and is also made available for use under a CC0 license. \n(which was not certified by peer review) is the author/funder. This article is a US Government work. It is not subject to copyright under 17 USC \nThe copyright holder for this preprintthis version posted May 16, 2026. ; https://doi.org/10.64898/2026.05.12.724341doi: bioRxiv preprint \n\n \nFlow cytometry  \nFlow cytometry was performed as described ( 11). EdU was detected using the Click-iT™ Plus \nEdU Pacific Blue™ Flow Cytometry Assay Kit (Cat. C10636, Thermo Fisher). For antibody \ndetails see Supplementary Table 1. \n \nFLOWMIST assay for autoreactivity \nThe flow cytometry assay was performed as described ( 11). Serum was diluted 1:160 and \nautoreactive antibodies detected with 1µg/ml AlexaFluor647-conjugated F(ab')2 goat anti-mouse \nIgG Fc (Jackson ImmunoResearch, 115-606-071). End titers were the last dilution producing a \nmean fluorescence intensity (MFI) ratio >3 over negative control serum. \n \nAutoantigen array \nAutoantigen arrays were performed as described (33) with the following differences: Serum was \ntested at 1:640 dilution in PBS. Autoantibodies were detected with Cy3-conjugated goat anti-\nmouse IgG(H+L) (Invitrogen, Cat. A10521) and with AlexaFluor633-conjugated goat anti-\nmouse IgM (Invitrogen, Cat. No. A21046) secondary antibodies.  \n \nPhage Immunoprecipitation Sequencing (PhIP-Seq) \nPhIP-Seq was performed by CDI Labs as described ( 61, 62 ). Counts provided by the vendor \nwere analyzed using edgeR ( 63). Modifications to the edgeR workflow were followed as \ndescribed (64). Gene set enrichment analyses and over representation analyses were performed \n105 and is also made available for use under a CC0 license. \n(which was not certified by peer review) is the author/funder. This article is a US Government work. It is not subject to copyright under 17 USC \nThe copyright holder for this preprintthis version posted May 16, 2026. ; https://doi.org/10.64898/2026.05.12.724341doi: bioRxiv preprint \n\nusing the clusterProfiler package with the msigdbr package ( https://CRAN.R-\nproject.org/package=msigdbr) used for the gene set database (65). \n \nComplement staining \nStaining was performed on 5 μ m FFPE sections using a manual benchtop method. Antigen \nretrieval was performed with EDTA buffer for 10mins at 100°C.  Nonspecific binding was \nblocked with an incubation of 2% normal donkey serum for 20 mins.  This was followed by an \novernight incubation at 4°C with the C3d primary antibody (R&D Systems, AF2655) at a 1:250 \ndilution. Antibody detection was accomplished by incubation with Donkey anti-Goat AlexaFluor \n594 (Invitrogen, A11058), followed by incubation in DAPI.  Slides were digitally imaged at 20x \nusing a Leica Aperio FL fluorescence digital scanner. \n \nImmunoglobulin (Ig) staining \nFFPE blocks were sectioned at 5\nμ m in preparation for IgM, IgG, IgA IHC.  Staining was  \nperformed using a Leica Bond RX autostainer (Leica Biosystems). Antigen retrieval was \nperformed using EDTA for 20 mins at 100°C on the Bond autostainer. Endogenous biotin was \nblocked using the Avidin-Biotin blocking kit (Vector Labs, SP-2001) per the manufacturer’s \ninstructions. The primary antibody for IgM, IgG, and IgA (Southern Biotechnology, 1012-08) \nwas diluted 1:100, with a 30-minute incubation time.  Antibody detection was accomplished \nusing AlexaFluor 660 conjugated Streptavidin (Invitrogen, S21377) followed by DAPI \ncounterstain. Slides were digitally imaged at 20x using a Leica Aperio FL fluorescence digital \nscanner.   \n \n105 and is also made available for use under a CC0 license. \n(which was not certified by peer review) is the author/funder. This article is a US Government work. It is not subject to copyright under 17 USC \nThe copyright holder for this preprintthis version posted May 16, 2026. ; https://doi.org/10.64898/2026.05.12.724341doi: bioRxiv preprint \n\nImage analysis \nImage analysis was conducted using the Area Quantification FL v2.3.3 algorithm in HALO \n(Indica Labs, Albuquerque, NM). Analyses were conducted for immunoglobulin / complement \nstaining within the cortex and glomeruli to quantify the percentage of positive area. \n \nBone marrow derived macrophage stimulation \nBone marrow derived macrophages (BMDM) were differentiated in RPMI1640 with 10% heat-\ninactivated fetal bovine serum, penicillin/streptomycin, and recombinant Fc-tagged human M-\nCSF (amino acids 33 to 190; produced and purified in-house) for 6 days, then primed with \n1\nμ g/ml Pam3CSK4 (InvivoGen, tlrl-pms) for 5h. To induce pyroptosis, BMDM were stimulated \nwith 10μ M nigericin (Invivogen, tlrl-nig) for 1h. To induce necroptosis, BMDM were stimulated \nwith 100ng/ml TNFα  (Sigma, T7539-10UG), 500nM birinapant (Apexbio, A4219) and 20μ M Z-\nVAD-FMK (Tocris, 2163) for 16h.  \n \nStatistical analysis \nStatistical significance was determined using GraphPad Prism software. Data were first \nevaluated for normal distribution using the Anderson-Darling test, D’Agostino & Pearson test, \nShapiro-Wilk test and Kolmogorov-Smirnov test. If any test reported that N is too small for \nevaluation or if data were normally distributed, a two-tailed unpaired t-test was used. If data did \nnot pass normality tests, Mann-Whitney test was used. Survival curves were evaluated using the \nLog-rank (Mantel-Cox) test. Test results are indicated in the Figures and Figure legends. \n \n \n105 and is also made available for use under a CC0 license. \n(which was not certified by peer review) is the author/funder. This article is a US Government work. It is not subject to copyright under 17 USC \nThe copyright holder for this preprintthis version posted May 16, 2026. ; https://doi.org/10.64898/2026.05.12.724341doi: bioRxiv preprint \n\n \n \n \n \n \n \nSupplementary Materials \nfig S1. Gating strategy for B cell subset identification \nfig S2. Validation of GFP and hBcl-2 expression in Bcl2\nEarly and Bcl2Late mice \nfig S3. Serum IgG autoantibody reactivity \nfig S4. Control experiments and validation of Bcl2\nInt mice \nSupplementary Table 1. Details for monoclonal antibodies used in this study \nSupplementary Data 1. IgG autoantigen array data \nSupplementary Data 2. IgM autoantigen array data \nSupplementary Data 3. Gene-set enrichment analysis of IgG autoreactivity assessed by phage \nimmunoprecipitation sequencing \n \nDuring the preparation of this work the authors used Claude and ChatGPT provided by the U.S. \nDepartment of Health and Human Services to optimize language. After using these tools, the \n105 and is also made available for use under a CC0 license. \n(which was not certified by peer review) is the author/funder. This article is a US Government work. It is not subject to copyright under 17 USC \nThe copyright holder for this preprintthis version posted May 16, 2026. ; https://doi.org/10.64898/2026.05.12.724341doi: bioRxiv preprint \n\nauthors reviewed and edited the content as needed and take full responsibility for the content of \nthe published article. \n \n \n \n \nReferences and Notes \n1. H. Wardemann  et al., Predominant autoantibody production by early human B cell \nprecursors. Science 301, 1374–1377 (2003). \n2. T. Tiller  et al., Autoreactivity in human IgG+ memory B cells. Immunity 26, 205–213 \n(2007). \n3. R. J. Ludwig  et al., Mechanisms of Autoantibody-Induced Pathology. Front Immunol 8, \n603 (2017). \n4. M. Shome  et al., Serum autoantibodyome reveals that healthy individuals share common \nautoantibodies. Cell Rep 39, 110873 (2022). \n5. W. Egner, The use of laboratory tests in the diagnosis of SLE. J Clin Pathol 53, 424–432 \n(2000). \n6. M. R. Clark, M. Mandal, K. Ochiai, H. Singh, Orchestrating B cell lymphopoiesis \nthrough interplay of IL-7 receptor and pre-B cell receptor signalling. Nat Rev Immunol \n14, 69–80 (2014). \n7. S. Herzog, M. Reth, H. Jumaa, Regulation of B-cell proliferation and differentiation by \npre-B-cell receptor signalling. Nat Rev Immunol 9, 195–205 (2009). \n8. R. R. Hardy, K. Hayakawa, B cell development pathways. Annu Rev Immunol  19, 595–\n621 (2001). \n9. D. Allman  et al., Resolution of three nonproliferative immature splenic B cell subsets \nreveals multiple selection points during peripheral B cell maturation. J Immunol 167, \n6834–6840 (2001). \n10. F. Loder  et al., B cell development in the spleen takes place in discrete steps and is \ndetermined by the quality of B cell receptor-derived signals. J Exp Med 190, 75–89 \n(1999). \n11. M. J. Simpson  et al., Peripheral apoptosis and limited clonal deletion during physiologic \nmurine B lymphocyte development. Nat Commun 15, 4691 (2024). \n12. D. M. Allman, S. E. Ferguson, V. M. Lentz, M. P. Cancro, Peripheral B cell maturation. \nII. Heat-stable antigen(hi) splenic B cells are an immature developmental intermediate in \nthe production of long-lived marrow-derived B cells. J Immunol 151, 4431–4444 (1993). \n105 and is also made available for use under a CC0 license. \n(which was not certified by peer review) is the author/funder. This article is a US Government work. It is not subject to copyright under 17 USC \nThe copyright holder for this preprintthis version posted May 16, 2026. ; https://doi.org/10.64898/2026.05.12.724341doi: bioRxiv preprint \n\n13. A. G. Rolink, J. Andersson, F. Melchers, Characterization of immature B cells by a novel \nmonoclonal antibody, by turnover and by mitogen reactivity. Eur J Immunol 28, 3738–\n3748 (1998). \n14. G. D. Victora, M. C. Nussenzweig, Germinal Centers. Annu Rev Immunol 40, 413–442 \n(2022). \n15. C. Young, R. Brink, The unique biology of germinal center B cells. Immunity 54, 1652–\n1664 (2021). \n16. O. Bannard, J. G. Cyster, Germinal centers: programmed for affinity maturation and \nantibody diversification. Curr Opin Immunol 45, 21–30 (2017). \n17. J. A. Roco  et al., Class-Switch Recombination Occurs Infrequently in Germinal Centers. \nImmunity 51, 337–350 e337 (2019). \n18. D. A. Nemazee, K. Burki, Clonal deletion of B lymphocytes in a transgenic mouse \nbearing anti-MHC class I antibody genes. Nature 337, 562–566 (1989). \n19. D. M. Russell  et al., Peripheral deletion of self-reactive B cells. Nature 354, 308–311 \n(1991). \n20. R. Brink, T. G. Phan, Self-Reactive B Cells in the Germinal Center Reaction. Annu Rev \nImmunol 36, 339–357 (2018). \n21. T. Ota, M. Ota, B. H. Duong, A. L. Gavin, D. Nemazee, Liver-expressed Igkappa \nsuperantigen induces tolerance of polyclonal B cells by clonal deletion not kappa to \nlambda receptor editing. J Exp Med 208, 617–629 (2011). \n22. J. J. Taylor  et al., Deletion and anergy of polyclonal B cells specific for ubiquitous \nmembrane-bound self-antigen. J Exp Med 209, 2065–2077 (2012). \n23. S. Han, B. Zheng, J. Dal Porto, G. Kelsoe, In situ studies of the primary immune response \nto (4-hydroxy-3-nitrophenyl)acetyl. IV. Affinity-dependent, antigen-driven B cell \napoptosis in germinal centers as a mechanism for maintaining self-tolerance. J Exp Med \n182, 1635–1644 (1995). \n24. B. Pulendran, G. Kannourakis, S. Nouri, K. G. Smith, G. J. Nossal, Soluble antigen can \ncause enhanced apoptosis of germinal-centre B cells. Nature 375, 331–334 (1995). \n25. K. M. Shokat, C. C. Goodnow, Antigen-induced B-cell death and elimination during \ngerminal-centre immune responses. Nature 375, 334–338 (1995). \n26. P. Bouillet  et al., Proapoptotic Bcl-2 relative Bim required for certain apoptotic \nresponses, leukocyte homeostasis, and to preclude autoimmunity. Science 286, 1735–\n1738 (1999). \n27. A. Egle, A. W. Harris, M. L. Bath, L. O'Reilly, S. Cory, VavP-Bcl2 transgenic mice \ndevelop follicular lymphoma preceded by germinal center hyperplasia. Blood 103, 2276–\n2283 (2004). \n28. F. Ewald  et \n al., Constitutive expression of murine c-FLIPR causes autoimmunity in aged \nmice. Cell Death Dis 5, e1168 (2014). \n29. A. Strasser  et al., Enforced BCL2 expression in B-lymphoid cells prolongs antibody \nresponses and elicits autoimmune disease. Proc Natl Acad Sci U S A 88, 8661–8665 \n(1991). \n30. J. A. Wright  et al., Impaired B Cell Apoptosis Results in Autoimmunity That Is \nAlleviated by Ablation of Btk. Front Immunol 12, 705307 (2021). \n31. O. Takeuchi et al. , Essential role of BAX,BAK in B cell homeostasis and prevention of \nautoimmune disease. Proc Natl Acad Sci U S A 102, 11272–11277 (2005). \n105 and is also made available for use under a CC0 license. \n(which was not certified by peer review) is the author/funder. This article is a US Government work. It is not subject to copyright under 17 USC \nThe copyright holder for this preprintthis version posted May 16, 2026. ; https://doi.org/10.64898/2026.05.12.724341doi: bioRxiv preprint \n\n32. R. Watanabe-Fukunaga, C. I. Brannan, N. G. Copeland, N. A. Jenkins, S. Nagata, \nLymphoproliferation disorder in mice explained by defects in Fas antigen that mediates \napoptosis. Nature 356, 314–317 (1992). \n33. C. T. Mayer  et al., An apoptosis-dependent checkpoint for autoimmunity in memory B \nand plasma cells. Proc Natl Acad Sci U S A 117, 24957–24963 (2020). \n34. V. V. Vasilev  et al., Functional Characterization of Autoantibodies against Complement \nComponent C3 in Patients with Lupus Nephritis. J Biol Chem 290, 25343–25355 (2015). \n35. R. Asero  et al., Autoantibody to proliferating cell nuclear antigen (PCNA) in SLE: a \nclinical and serological study. Clin Exp Rheumatol 5, 241–246 (1987). \n36. S. Megremis  et al., Analysis of human total antibody repertoires in TIF1gamma \nautoantibody positive dermatomyositis. Commun Biol 4, 419 (2021). \n37. M. Kraus, M. B. Alimzhanov, N. Rajewsky, K. Rajewsky, Survival of resting mature B \nlymphocytes depends on BCR signaling via the Igalpha/beta heterodimer. Cell 117, 787–\n800 (2004). \n38. C. T. Mayer  et al., The microanatomic segregation of selection by apoptosis in the \ngerminal center. Science 358,  (2017). \n39. Y. J. Liu  et al., Mechanism of antigen-driven selection in germinal centres. Nature 342, \n929–931 (1989). \n40. R. A. Sater, P. C. Sandel, J. G. Monroe, B cell receptor-induced apoptosis in primary \ntransitional murine B cells: signaling requirements and modulation by T cell help. Int \nImmunol 10, 1673–1682 (1998). \n41. A. D. Gitlin  et al., Independent Roles of Switching and Hypermutation in the \nDevelopment and Persistence of B Lymphocyte Memory. Immunity 44, 769–781 (2016). \n42. T. Cantaert  et al., Decreased somatic hypermutation induces an impaired peripheral B \ncell tolerance checkpoint. J Clin Invest 126, 4289–4302 (2016). \n43. T. Cantaert  et al., Activation-Induced Cytidine Deaminase Expression in Human B Cell \nPrecursors Is Essential for Central B Cell Tolerance. Immunity 43, 884–895 (2015). \n44. M. Kuraoka  et al., Activation-induced cytidine deaminase mediates central tolerance in B \ncells. Proc Natl Acad Sci U S A 108, 11560–11565 (2011). \n45. G. Meyers  et al ., Activation-induced cytidine deaminase (AID) is required for B-cell \ntolerance in humans. Proc Natl Acad Sci U S A 108, 11554–11559 (2011). \n46. J. Zhu  et al., Abrogated AID Function Prolongs Survival and Diminishes Renal \nPathology in the BXSB Mouse Model of Systemic Lupus Erythematosus. J Immunol 204, \n1091–1100 (2020). \n47. C. Jiang  et al., Abrogation of lupus nephritis in activation-induced deaminase-deficient \nMRL/lpr mice. J Immunol 178, 7422–7431 (2007). \n48. Q. Tan  et al., Activation-induced cytidine deaminase deficiency accelerates autoimmune \ndiabetes in NOD mice. JCI Insight 3,  (2018). \n49. L. Chen, L. Guo, J. Tian, B. Zheng, S. Han, Deficiency in activation-induced cytidine \ndeaminase promotes systemic autoimmunity in lpr mice on a C57BL/6 background. Clin \nExp Immunol 159, 169–175 (2010). \n50. V. V. Vasilev  et al., Autoantibodies Against C3b-Functional Consequences and Disease \nRelevance. Front Immunol 10, 64 (2019). \n51. F. Corvillo  et al., Nephritic Factors: An Overview of Classification, Diagnostic Tools and \nClinical Associations. Front Immunol 10, 886 (2019). \n105 and is also made available for use under a CC0 license. \n(which was not certified by peer review) is the author/funder. This article is a US Government work. It is not subject to copyright under 17 USC \nThe copyright holder for this preprintthis version posted May 16, 2026. ; https://doi.org/10.64898/2026.05.12.724341doi: bioRxiv preprint \n\n52. D. Ait-Azzouzene  et al., An immunoglobulin C kappa-reactive single chain antibody \nfusion protein induces tolerance through receptor editing in a normal polyclonal immune \nsystem. J Exp Med 201, 817–828 (2005). \n53. R. Halverson, R. M. Torres, R. Pelanda, Receptor editing is the main mechanism of B \ncell tolerance toward membrane antigens. Nat Immunol 5, 645–650 (2004). \n54. D. F. Robbiani  et al., AID is required for the chromosomal breaks in c-myc that lead to c-\nmyc/IgH translocations. Cell 135, 1028–1038 (2008). \n55. I. Rauch  et al., NAIP-NLRC4 Inflammasomes Coordinate Intestinal Epithelial Cell \nExpulsion with Eicosanoid and IL-18 Release via Activation of Caspase-1 and -8. \nImmunity 46, 649–659 (2017). \n56. Y. Wang  et al., Chemotherapy drugs induce pyroptosis through caspase-3 cleavage of a \ngasdermin. Nature 547, 99–103 (2017). \n57. E. Hobeika  et al., Testing gene function early in the B cell lineage in mb1-cre mice. Proc \nNatl Acad Sci U S A 103, 13789–13794 (2006). \n58. M. Muramatsu  et al., Class switch recombination and hypermutation require activation-\ninduced cytidine deaminase (AID), a potential RNA editing enzyme. Cell 102, 553–563 \n(2000). \n59. J. M. Murphy  et al., The pseudokinase MLKL mediates necroptosis via a molecular \nswitch mechanism. Immunity 39, 443–453 (2013). \n60. G. Knittel  et al., B-cell-specific conditional expression of Myd88p.L252P leads to the \ndevelopment of diffuse large B-cell lymphoma in mice. Blood 127, 2732–2741 (2016). \n61. E. Rackaityte  et al., Validation of a murine proteome-wide phage display library for \nidentification of autoantibody specificities. JCI Insight 8,  (2023). \n62. D. Mohan  et al., PhIP-Seq characterization of serum antibodies using oligonucleotide-\nencoded peptidomes. Nat Protoc 13, 1958–1978 (2018). \n63. Y. Chen, L. Chen, A. T. L. Lun, P. L. Baldoni, G. K. Smyth, edgeR v4: powerful \ndifferential analysis of sequencing data with expanded functionality and improved \nsupport for small counts and larger datasets. Nucleic Acids Res 53,  (2025). \n64. A. Chen, K. Kammers, H. B. Larman, R. B. Scharpf, I. Ruczinski, Detecting antibody \nreactivities in Phage ImmunoPrecipitation Sequencing data. BMC Genomics 23, 654 \n(2022). \n65\n. S. Xu  et al., Using clusterProfiler to characterize multiomics data. Nat Protoc 19, 3292–\n3320 (2024). \n \n \n \n \n \n \n105 and is also made available for use under a CC0 license. \n(which was not certified by peer review) is the author/funder. This article is a US Government work. It is not subject to copyright under 17 USC \nThe copyright holder for this preprintthis version posted May 16, 2026. ; https://doi.org/10.64898/2026.05.12.724341doi: bioRxiv preprint \n\n \n \n \n \n \n \nAcknowledgments: We thank all members of the Experimental Immunology Branch, and Drs. \nBen Afzali, Avinash Bhandoola, Didier Portilla and Jagan Muppidi for discussions and advice; \nJeffrey Chiang and Jie Mu for technical support; Assiatu Crossman, Kheem Bisht, Don Plugge, \nWilliam Hajjar, and Tengfei Zhang for flow cytometry support; all staff at the NCI Frederick and \nNCI Bethesda animal facilities for their critical help, particularly Jennifer Wise.  \n \nFunding:  \nNational Cancer Institute, Center for Cancer Research, National Institutes of Health, ZIA BC  \n011975 and Contract No. HHSN26120150003I. The content of this publication does not  \nnecessarily reflect the views or policies of the Department of Health and Human Services, nor  \ndoes mention of trade names, commercial products, or organizations imply endorsement by the \n U.S. Government. \n \n105 and is also made available for use under a CC0 license. \n(which was not certified by peer review) is the author/funder. This article is a US Government work. It is not subject to copyright under 17 USC \nThe copyright holder for this preprintthis version posted May 16, 2026. ; https://doi.org/10.64898/2026.05.12.724341doi: bioRxiv preprint \n\nDisclaimer: This research was supported by the Intramural Research Program of the National \nInstitutes of Health (NIH). The contributions of the NIH author(s) were made as part of \ntheir official duties as NIH federal employees, are in compliance with agency policy \nrequirements, and are considered Works of the United States Government. However, the \nfindings and conclusions presented in this paper are those of the author(s) and do not \nnecessarily reflect the views of the NIH or the U.S. Department of Health and Human \nServices. \n \nAuthor contributions:  \nConceptualization: CTM \nMethodology: CTM, MJS, CYT, HK \nInvestigation: CTM, AMN, UAA, MJS, FF, SS, MI, IA, BK, LB, IR, CZ, BC, QC, DM \nVisualization: CTM, AMN, UAA, MJS, FF, DP, BK, LB, BC, QC, DM, IR, CZ \nFunding acquisition: CTM \nProject administration: CTM \nSupervision: CTM \nWriting – original draft: CTM \nWriting – review & editing: CTM, AMN, UAA, MJS, FF, SS, MI, IA, DP, BK, LB, BC, \nQC, DM, IR, CZ, CYT, HK \n \nCompeting interests: Authors declare that they have no competing interests. \n105 and is also made available for use under a CC0 license. \n(which was not certified by peer review) is the author/funder. This article is a US Government work. It is not subject to copyright under 17 USC \nThe copyright holder for this preprintthis version posted May 16, 2026. ; https://doi.org/10.64898/2026.05.12.724341doi: bioRxiv preprint \n\nData and materials availability: All data are available in the main text or the supplementary  \nmaterials. Autoantigen array data are available at GEO accession GSE296597. PhIP-seq data are  \navailable at NCBI accession PRJNA1271546 \n \n \n \n \nFigures:  \n105 and is also made available for use under a CC0 license. \n(which was not certified by peer review) is the author/funder. This article is a US Government work. It is not subject to copyright under 17 USC \nThe copyright holder for this preprintthis version posted May 16, 2026. ; https://doi.org/10.64898/2026.05.12.724341doi: bioRxiv preprint \n\n \nFig. 1. Early, but not post-activation, MOMP inhibition drives autoimmune disease \n105 and is also made available for use under a CC0 license. \n(which was not certified by peer review) is the author/funder. This article is a US Government work. It is not subject to copyright under 17 USC \nThe copyright holder for this preprintthis version posted May 16, 2026. ; https://doi.org/10.64898/2026.05.12.724341doi: bioRxiv preprint \n\nCohorts of Bcl2 Early (n=24), Bcl2 Late (n=21), and Control mice (n=33) were monitored. ( A, B ) \nSurvival analysis of (A) all mice combined or (B) according to sex. ** p=0.0036 (both sexes), ** \np=0.0063 (females), NS=not statistically significant (Log-rank (Mantel-Cox) test). ( C-F) \nAnalysis of kidney pathology at 40 weeks. (C) Representative micrographs depict hematoxylin \nand eosin-stained kidney sections of females of the indicated genotypes. Top, glomeruli (scale \nbar: 20µm). Bottom, cortex (scale bar: 200µm). ( D-F) Pathology scores for ( D) \nglomerulonephritis or ( E) tubular lesion or (F ) lymphoplasmacytic infiltrates according to sex. \n(G, H) Immunofluorescence analysis of kidney sections. Micrographs to the left depict staining \n(green) for (G) Immunoglobulin (Ig) or (H) complement C3d in glomeruli (top, scale bar: 50µm) \nor cortex (bottom, scale bar: 100µm). DAPI staining is shown in blue. Graphs on the right show \nIg+ and C3d+ area quantification. (I, J) Serum urea quantification for (I) all mice at 36 weeks or \n(J) longitudinally for two Bcl2 Early mice showing urea elevation at 36 weeks of age. Cross \nindicates terminal disease. Dotted lines represent twice the mean serum urea concentration of \ncontrol mice, and measurements above this line are considered abnormal. (D-H) **** p<0.0001, \n*** p<0.001, ** p<0.01, * p<0.05, NS=not statistically significant (unpaired two-tailed \nStudent’s t-test). Comparisons relative to controls and between Bcl2\nEarly and Bcl2 Late mice are \nshown. Horizontal bars represent mean values. ( A-J) Results are from one cohort sequentially \nrecruited from independent litters. \n \n \n \n \n105 and is also made available for use under a CC0 license. \n(which was not certified by peer review) is the author/funder. This article is a US Government work. It is not subject to copyright under 17 USC \nThe copyright holder for this preprintthis version posted May 16, 2026. ; https://doi.org/10.64898/2026.05.12.724341doi: bioRxiv preprint \n\n \nFig. 2. Early MOMP inhibition broadens class-switched IgG autoreactivity \n(A, B) Autoantigen array heatmaps depict normalized antibody scores (red, high reactivity; \ngreen, low reactivity) for (A) IgG or (B ) IgM serum antibody binding to indicated self-antigens. \n105 and is also made available for use under a CC0 license. \n(which was not certified by peer review) is the author/funder. This article is a US Government work. It is not subject to copyright under 17 USC \nThe copyright holder for this preprintthis version posted May 16, 2026. ; https://doi.org/10.64898/2026.05.12.724341doi: bioRxiv preprint \n\nEach column represents one 40-week-old mouse of the indicated genotypes. ( C) Quantification \nof autoantibody breadth from autoantigen array data (Supplementary Data 1-2). Left: Number of \nstatistically significant IgG-reactive autoantigens per genotype. Bcl2 Early m i c e  s h o w  2 3  \nsignificant IgG autoantigens compared to 5 in Bcl2Late mice relative to controls. Middle: Number \nof statistically significant IgM-reactive autoantigens per genotype. Bcl2 Early m i c e  s h o w  1 8  \nsignificant IgM autoantigens compared to 11 in Bcl2Late mice relative to controls. Right: Number \nof statistically significant autoantigens in Bcl2 Early mice relative to Bcl2 Late mice according to \nisotype. Bcl2 Early mice show 13 statistically significant IgG-reactive autoantigens, but only 1 \nIgM-reactive autoantigen. Early checkpoint failure specifically expands IgG breadth while IgM \nremains comparable. ( D) Phage immunoprecipitation sequencing (PhIP-seq) analysis of IgG \nautoantibodies. Heatmaps display normalized signal intensity (red, high; blue, low). Each row \nrepresents one 40-week-old mouse of the indicated genotypes. Dots in the center of a cell depict \np<0.05. Only phages yielding significant hits in at least two Bcl2 Early mice were included in the \nheatmap. Antigens encoded by phages yielding no significant hits in control and Bcl2 Late mice \nare labeled in red. \n105 and is also made available for use under a CC0 license. \n(which was not certified by peer review) is the author/funder. This article is a US Government work. It is not subject to copyright under 17 USC \nThe copyright holder for this preprintthis version posted May 16, 2026. ; https://doi.org/10.64898/2026.05.12.724341doi: bioRxiv preprint \n\n \nFig. 3. AID is required for autoimmune disease following early MOMP inhibition \n105 and is also made available for use under a CC0 license. \n(which was not certified by peer review) is the author/funder. This article is a US Government work. It is not subject to copyright under 17 USC \nThe copyright holder for this preprintthis version posted May 16, 2026. ; https://doi.org/10.64898/2026.05.12.724341doi: bioRxiv preprint \n\nCohorts of female Bcl2 Early (n=5), Bcl2 EarlyAicda-/- (n=5), and Control mice (n=4) were \nmonitored. (A) Fraction of diseased mice at 52 weeks (white: healthy; red: diseased). ( B) Type \nof disease (Autoimmune/other: magenta; B cell malignancy: black; abscess: green). ( C) \nMicrographs depict hematoxylin and eosin-stained abdominal mass sections of indicated mice \n(scale bar: 50µm).  Diagnosis is shown below micrographs. (D) Representative micrographs \ndepict hematoxylin and eosin-stained kidney sections of 38-52 weeks old females of the \nindicated genotypes. Glomeruli are shown in the center (scale bar: 100µm). ( E-G) Pathology \nscores for (E) glomerulonephritis or (F) tubular lesion or (G) lymphoplasmacytic infiltrates. *** \np<0.001, ** p<0.01, * p<0.05, NS=not statistically significant (unpaired two-tailed Student’s t-\ntest). Comparisons relative to controls and between Bcl2\nEarly and Bcl2 EarlyAicda-/- mice are \nshown.  Horizontal bars represent mean values.  \n \n \n105 and is also made available for use under a CC0 license. \n(which was not certified by peer review) is the author/funder. This article is a US Government work. It is not subject to copyright under 17 USC \nThe copyright holder for this preprintthis version posted May 16, 2026. ; https://doi.org/10.64898/2026.05.12.724341doi: bioRxiv preprint \n\n \nFig. 4. Early MOMP inhibition expands peripheral B cell populations without increasing \nbone marrow immature B cells \nMice aged 9-13 weeks were analyzed by flow cytometry. ( A) Genotypes and color coding. ( B, \nC) Total numbers of the indicated B cell subsets in ( B) bone marrow and ( C) spleen. Data are \n105 and is also made available for use under a CC0 license. \n(which was not certified by peer review) is the author/funder. This article is a US Government work. It is not subject to copyright under 17 USC \nThe copyright holder for this preprintthis version posted May 16, 2026. ; https://doi.org/10.64898/2026.05.12.724341doi: bioRxiv preprint \n\ncombined from six independent experiments (Controls, n=16; Bcl2 Early, n=11 for bone marrow \nand n=13 for spleen; Bcl2 Late, n=12). ( D-F) To assess bone marrow egress, mice aged 14-20 \nweeks were injected intravenously with anti-CD45R/B220-PE-Cy7 prior to analysis by flow \ncytometry. ( D) Experimental scheme. ( E, F ) Quantification of ( E) the percentage of B220-\nlabeled cells within the indicated B cell subsets and (F) the total number of B220-labeled B cell \nsubsets, indicating cells exposed to the bone marrow sinusoids. Data are combined from three \nindependent experiments (Controls, n=6; Bcl2\nEarly, n=6). ( G-I) To test the contribution of non-\napoptotic programmed cell death, mice aged 7-17 weeks were analyzed by flow cytometry. ( G) \nGenotypes and color coding. ( H, I) Total numbers of the indicated B cell subsets in ( H) bone \nmarrow and ( I) spleen of Gsdmd -/-Gsdme-/-Mlkl-/- mice and controls. Data are combined from \nfive independent experiments (Controls, n=10; Gsdmd -/-Gsdme-/-Mlkl-/-, n=11). **** p<0.0001, \n*** p<0.001, ** p<0.01, * p <0.05, NS=not statistically significant (two-tailed Mann-Whitney \ntest for panels B, C, H and I; two-tailed unpaired Student’s t-test for panels E and F). Horizontal \nbars represent mean values. Abbreviations: swMem, class-switched memory B cells; PC, plasma \ncells; T1, transitional 1 B cells; T2, transitional 2 B cells; Anergic/T3, anergic B cells; FO, \nmature follicular B cells; MZ, marginal zone B cells; GC, germinal center B cells. \n \n \n \n105 and is also made available for use under a CC0 license. \n(which was not certified by peer review) is the author/funder. This article is a US Government work. It is not subject to copyright under 17 USC \nThe copyright holder for this preprintthis version posted May 16, 2026. ; https://doi.org/10.64898/2026.05.12.724341doi: bioRxiv preprint \n\n \nFig. 5. Transitional B cell selection limits follicular B cell expansion \nMice aged 8-18 weeks were injected intravenously with 1mg EdU and analyzed by flow \ncytometry at the indicated chase times. (A) Experimental schematic. (B, C) Total numbers of the \nindicated EdU+ B cell subsets in ( B) bone marrow and ( C) spleen. Data are combined from two \n(2h, 7d) or five (3d) independent experiments (2h: Controls, n=5; Bcl2 Early, n=5; 3d: Controls, \nn=12; Bcl2Early, n=14; 7d: Controls, n=9; Bcl2 Early, n=7; Fr. C’, n=5 per genotype for all time \npoints). (D, E) Steady-state total numbers of (D) follicular (FO) and (E) anergic T3 B cells in the \nspleen, analyzed by flow cytometry in mice aged 9–13 weeks. Bcl2 Int mice are combined from \n105 and is also made available for use under a CC0 license. \n(which was not certified by peer review) is the author/funder. This article is a US Government work. It is not subject to copyright under 17 USC \nThe copyright holder for this preprintthis version posted May 16, 2026. ; https://doi.org/10.64898/2026.05.12.724341doi: bioRxiv preprint \n\nfour independent experiments (n = 11) run in parallel with the same Control and Bcl2Early cohorts \nshown in Fig. 4. **** p<0.0001, *** p<0.001, ** p<0.01, * p <0.05, NS=not statistically \nsignificant (two-tailed unpaired Student’s t-test). Horizontal bars represent mean values. \nAbbreviations: T1, transitional 1 B cells; T2, transitional 2 B cells; T3, anergic B cells; FO, \nmature follicular B cells. \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n105 and is also made available for use under a CC0 license. \n(which was not certified by peer review) is the author/funder. This article is a US Government work. It is not subject to copyright under 17 USC \nThe copyright holder for this preprintthis version posted May 16, 2026. ; https://doi.org/10.64898/2026.05.12.724341doi: bioRxiv preprint \n\n \nFig. 6. Distributed Clonal Deletion Model of B Cell Tolerance. Schematic illustrating the \nregulation of B cell tolerance by apoptosis. Receptor editing in the bone marrow efficiently \neliminates high-affinity or high-avidity autoreactive immature B cells, independent of genotype, \nallowing the release of low-affinity or low-avidity autoreactive B cells (light red) and non-\nautoreactive B cells (white) into the periphery.  \nIn wild-type mice, these cells undergo apoptosis prior to activation, representing a key tolerance \ncheckpoint that prevents autoimmunity (top left). In addition, autoreactive B cells generated \nduring immune responses through somatic hypermutation (blue) are deleted after activation, \nlimiting autoantibody production (bottom left). \nEarly Bcl-2 expression disrupts this process by preventing the deletion of low-affinity or low-\navidity autoreactive B cells, enabling their survival, entry into the follicular compartment, and \nparticipation in immune responses. These cells undergo AID-mediated diversification, \ngenerating class-switched, high-affinity autoreactive B cells that produce autoantibodies. The \nresulting immune complexes deposit in the kidney, activate complement, and drive severe renal \npathology and loss of function (top right). \n105 and is also made available for use under a CC0 license. \n(which was not certified by peer review) is the author/funder. This article is a US Government work. It is not subject to copyright under 17 USC \nThe copyright holder for this preprintthis version posted May 16, 2026. ; https://doi.org/10.64898/2026.05.12.724341doi: bioRxiv preprint \n\nIn contrast, late Bcl-2 expression does not affect pre-activation tolerance checkpoints but \npromotes the survival of autoreactive B cells generated by somatic hypermutation. This leads to \nclass-switched autoantibody production and immune complex deposition in the kidney, but with \nminimal complement activation, minimal pathology, and no significant loss of renal function \n(bottom right). This model is conceptual and not quantitative. \n \n \n \n \n \n \n105 and is also made available for use under a CC0 license. \n(which was not certified by peer review) is the author/funder. This article is a US Government work. It is not subject to copyright under 17 USC \nThe copyright holder for this preprintthis version posted May 16, 2026. ; https://doi.org/10.64898/2026.05.12.724341doi: bioRxiv preprint","source_license":"Public-Domain","license_restricted":false}