Deletion of IgG-switched Autoreactive B Cells and Defects in Fas^lpr Lupus Mice

Abstract

During a T cell-dependent antibody (Ab) response, B cells undergo Ab class-switching and variable region hypermutation, with the latter process potentially rendering previously innocuous B cells autoreactive. Class switching and hypermutation are temporally and anatomically linked with both processes dependent on the enzyme, activation-induced deaminase, and occurring principally, but not exclusively, in germinal centers. To understand tolerance regulation at this stage, we generated a new transgenic (Tg) mouse model expressing a membrane-tethered γ2a-reactive superantigen (γ2a-macroself Ag) and assessed the fate of emerging IgG2a-expressing B cells that have, following class switch, acquired self-reactivity of its Ag receptor to the macroself-Ag. In normal mice, self-reactive IgG2a-switched B cells were deleted, leading to the selective absence of IgG2a memory responses. These findings identify a novel negative selection mechanism for deleting mature B cells that acquire reactivity to self-Ag. This process was only partly dependent on the Bcl-2 pathway, but markedly inefficient in MRL-Fas^lpr lupus mice, suggesting that defective apoptosis of isotype-switched autoreactive B cells is central to Fas mutation-associated systemic autoimmunity.

INTRODUCTION

Foreign Ags trigger the growth and differentiation of Ag-specific memory B cells in peripheral lymphoid organs (1). Memory B cells upon rechallenge produce high affinity isotype-switched Abs with diverse effector functions and thus prime the immune system for rapid and effective responses. Somatic hypermutation (SHM), required for generating high affinity changes, however, may also lead to the acquisition of self-reactivity (2-4), thereby raising the question of how this potentially harmful event is controlled. Receptor editing, clonal deletion, and anergy in early B cell development constitute the first level by which the precursor frequency of autoreactive B cells is restrained (5-9), but the mechanism (s) involved in censoring mature B cells that acquire self-reactivity through SHM is, to a large extent, unknown.

Early experiments using an in vitro 4-hydroxy-3-nitrophenyl acetyl (NP)-hapten model that mimics responses to T-dependent Ags suggested that memory B cells are susceptible to tolerance induction (10). This postulate was later reinforced by several in vivo studies, some of which demonstrated apoptosis-mediated deletion of germinal center (GC) B cells that had acquired heightened capacity to bind the immunizing Ags (11-14). These studies suggested two plausible mechanisms for censoring newly created autoreactive GC B-cells during ongoing immune responses (15). The first proposes that soluble self-Ag interferes with the interaction between centrocytes and Ag bound to follicular dendritic cells (FDCs), a process critical for the survival of the developing nascent memory B cells and also for selecting B cells that have acquired higher affinities. The second hypothesizes that, at a later stage, autoreactive centrocytes do not obtain the T-cell help necessary for continued survival and undergo apoptosis. Since self-reacting T cells are deleted or rendered incapable of responding to self-Ag through central and peripheral tolerance mechanisms this provides a second layer of protection. Other studies on Ab responses to self-Ags in normal or BCR Tg mice have also suggested that memory B cells are biased away from self-reactivity, but the mechanisms involved and the developmental stages at which this bias may occur was not defined (16-20).

The extreme heterogeneity of lymphocyte specificity and the low precursor frequency of Ag specific lymphocytes make Ag receptor Tg mouse models indispensable in the study of lymphocyte biology and tolerance. However, memory B cell tolerance cannot be easily assessed experimentally in these quasi-monoclonal immune systems, since tolerance is already imposed at early B cell developmental stages. Tolerance during an ongoing immune response can, however, be analyzed in polyclonal immune systems in which, with appropriate experimental tools, the fate of autoreactive B cells can be assessed precisely when SHM and class switch recombination (CSR) occur. We recently developed such a novel experimental system based on the transgenic expression of a membrane-bound synthetic superAg, dubbed “macroself Ag (Ag)” (21, 22). This model exploits the striking propensity of B cells to exchange the genes encoding immunoglobulin constant regions during developmental progression. With the macroself Ag, engineered to react with a defined constant region of mouse immunoglobulin, BCRs can be specifically triggered when the reactive class or subclass appears on the cell surface.

In this study, a membrane-bound macroself Ag specific to an allelic variant of mouse IgG2a was generated by single chain Fv Ab engineering technology and expressed as a transgene. Since B cells require antigenic stimulation to undergo CSR, we reasoned that an IgG2a specific macroself Ag should only engage Ag-experienced B cells, thereby allowing normal preimmune development. Furthermore, SHM and CSR occur concurrently during the Ag-driven immune response, such that the cells that undergo CSR, are often those that hypermutate their Ab genes (23). We show here that ubiquitously expressed anti-IgG2a-macroself Ag promotes tolerance and rapid deletion of autoreactive isotype-switched B cells emerging after immunization with a T-dependent Ag. These and additional results provide direct evidence for the negative selection of mature B cells that acquire self-reactivity during ongoing immunity, suggest a novel mechanism for deleting GC B cells that bind surface-expressed self-Ags, and show that lupus-prone MRL-Fas^lpr mice are defective in this process.

MATERIALS AND METHODS

Mice and immunization protocol

C57BL/6 (B6), B6-Igh^a congenic, B6-Cd45.1 and MRL/ Fas^lpr mice were obtained from Jackson Laboratories (Bar Harbor, Maine). MRL-Fas^lpr pURF-Tg mice were generated by backcrossing B6-Igh^a pURF-Tg mice with MRL-Fas^lpr mice for at least 7 generations. EmuBcl-2-22 Tg (Bcl2-Tg) mice (24) were kindly provided by Drs. Strasser and Harris (WEHI, Melbourne). Mice were immunized with 400μg of NP₂₀-KLH in RIBI adjuvant (Sigma-Aldrich, St. Louis, MO). All mice were bred and maintained in The Scripps Research Institute Animal Resources facility according to Institutional Animal Care and Use Committee guidelines.

Generation of γ2a-macroself Ag gene constructs

The VJ light chain and VDJ heavy chain variable genes were amplified by polymerase chain reaction using as templates the plasmids containing genomic DNA of the anti-mouse IgG2a^{a, d, e, f, g, h, j, n, o} monoclonal Ab (a kind gift of Mark Shlomchik) derived from the 20.8.3 hybridoma (25). To generate a single chain Ab gene a PCR sewing approach was taken using the following oligonucleotide primers: primer 1 (5′V_L) 5′-TCGCGAATCGCCGACAGGTGCGATGGACATGAGGGCCCATGCTC-3′; primer 2 (3′V_L) 5′-CCTCCCGAGCCACCGCCTCCGCTGCCTCCGCCTCCCCGTTTTATTTCCAACTTCGTCCCG-3′; primer 3 (5′V_H) 5′-GCAGCGGAGGCGGTGGCTCGGGAGGCGGAGGCTCGCAGGTACAGCTGAAAGAGTCAGG-3′; primer 4 (3′V_H) 5′-CCCGGGTTTCTGGGGGCTGTTGTTTCAGCTGAGGAGAC. The “overlapping” sequences are underlined; those recognized by restriction endonucleases are in italics. PCR products corresponding to the LVJ_L and the VDJ_H region were initially amplified with the primers 1, 2 and 3, 4, respectively. The single chain Fv gene, including the intervening flexible peptide codons (Gly₄-Ser)₃, was assembled by a second PCR step with the outer primers 1 and 4, using as templates the LV_L and V_H PCR products. The PCR product was further cloned as a NruI/XmaI restriction fragment downstream the human ubiquitin C promoter and upstream the rat-IgG1 Fc cDNA fused with the transmembrane, cytoplasmic exons and 3′-UTR regions of the H-2K^b genomic DNA in the pBluescript II SK plasmid (21).

Transient transfection of human embryonic kidney 293T-cells

HEK 293T cells were co-transfected with PIRES-EGFP plasmid (Clontech, Mountain View, CA) and the plasmid containing the pURF transgene using Lipofectamine/Plus reagent (Invitrogen) on six well plates according the manufacturer’s recommendations. Transfected cell were harvested after two days of growth in complete IMDM medium for flow cytometry analysis.

Production of pURF Tg mice

The 4 kb pURF transgene construct was separated from bacterial vector sequences by a digestion with HindIII/Not1 and agarose gel electrophoresis. The fragment was isolated and purified for microinjection as previously described (21). Tg mice were produced by classical microinjection techniques at the TSRI Mouse Genetics Core Facility.

Spleen transplantation chimeras

Recipient mice were pURF-Tgs or littermate controls; all carried the CD45.1 allele while spleen donors were CD45.2⁺. Recipients received 750 rads gamma radiation from a Cs source then 1 h later were injected i.v. with 30 million donor spleen cells. The next day, mice were immunized with 400μg of NP₂₀-KLH in RIBI and 2 wk later spleen cells were studied. Chimeras with ≥ 98% donor-derived spleen cells were analyzed.

Flow cytometry analysis

HEK 293T transfected cells were harvested using 1X PBS, 0.5mM EDTA, washed twice and incubated with either a biotin-conjugated mouse anti-rat IgG1, a mouse IgG2a^a, or a mouse IgG2a^b monoclonal Abs. Cells were incubated with a biotin-coupled rat anti-mouse IgG2a,b to assess Fv binding specificity. Biotin-coupled Abs were revealed with streptavidin-phycoerythrin. For the analysis of mouse cells ex-vivo, nucleated cell suspensions were prepared from the spleen as previously described (21). Five million cells were surface staining with the following monoclonal Abs: FITC-conjugated anti-CD4, CD8, TcRβ, F4/80, Gr1, IgM, IgD (Dump channel) and PerCP-coupled anti-CD45R/B220 (RA3-6B2). Surface stained cells were fixed and permeabilized using a kit (Cytofix/Cytoperm™, BD Biosciences, San Jose, CA) and stained according to the manufacturer’s instructions with one of the following biotin-conjugated monoclonal Ab: anti-IgG2a^a (8.3), anti-IgG2a^b (5.7), anti-mouse IgG2a (RMG2a-62, Biolegend, San Diego, CA), anti-mouse IgG2b (R12-3) and anti-mouse-IgG1 (A85-1). After two washes, spleen cells were incubated with a phycoerythrin-conjugated rat anti-mouse Igκ (187.1) and allophycocyanin-conjugated streptavidin. Stained cells were acquired on a FACSCalibur flow cytometer (BD Biosciences) and results were analyzed using the FlowJo software package using 5% or 2% contour plot on logarithmic graphic displays. For direct analysis of Ag-specific B cells, ex vivo cells were labeled for 45min with FITC-A85.1 (anti-IgG1), biotin-8.3 (anti-IgG2a^a) or biotin-5.7 (anti-IgG2a^b), washed and blocked with rat and mouse serum for 15min before addition of the appropriate labeling mix combination containing FITC- or biotin-11.26 (anti-IgD), NP-APC (provided by McHeyser-Williams), Cy7PE-6B2 (anti-B220, Biolegend), PE-281.2 (anti-CD138) Cy5PE-H129.19 (anti-CD4), 53-6.7 (anti-CD8), and then streptavidin-Cy7APC. Stained cells were analyzed on a FACS Vantage SE (BD Biosciences). All reagents and Abs were purchased from BD Biosciences Pharmingen, unless indicated.

RT PCR and quantitative real time PCR

Total RNA was purified from spleen of immunized IgH^a non-Tg and pURF Tg mice using the RNeasy Mini kit and RNase-Free DNase Set (Qiagen) and cDNA was generated with the SuperScript III First-Strand Synthesis System for RT-PCR kit (Invitrogen) according to the manufacturers’ instructions. PCR reactions were done in a final volume of 50 μl containing four fold cDNA serial dilutions. V_H-γ2a PCR products were generated using a sense primer specific for the majority of mouse V_H genes (5′-GAGGTGCAGCTGCAGGAGTCTGG) and an IgG2a specific anti-sense located in the hinge region (5′-ATCCTAGAGTCACCGAGGAGCC). For the detection of IgG2a Iμ-γ2a germline and Iγ2a-γ2a transcripts, the following forward primer used were 5′-CTCTGGCCCTGCTTATTGTTG and 5′-GGCTGTTAGAAGCACAGTGACAAAG, respectively. The reverse IgG2a primer was 5′-GCCACATTGCAGGTGATGGA and the PCR products size were 472 bp for the Iμ-γ2a and 910, 670 and 560 base pairs and Iγ2a-γ2a germline transcripts. The specificity of these PCR reactions was verified by sequencing the amplification products. Samples were amplified 25 cycles, 1 min 94°C, 1 min 60°C, and 1 min 72°C. PCR products were electrophoresed in 1.5% agarose gels, blotted on nylon membranes, and probed with a 500 bp DNA probe containing the mouse IgG2a CH1 exon. Quantitation was performed using ImageQuant TL software (GE Healthcare, Piscataway, NJ). Data were analyzed by first normalizing specific RNA products to corresponding actin levels (specific product/actin) and then expressing the amounts as a fraction or percent of non-Tg naïve mice (1 or 100%). For real time PCR analysis, cDNA amplification was performed with SYBR Green PCR buffer containing 0.1μM of IgG2a specific primers using the 7900 HT (Applied Biosystems, Foster City, CA). IgG2a (CH1-hinge) mRNA expression was normalized to actin using the formula 2exp(Ct_actin - Ct_γ2a) as previously described (26).

Serum Ig determinations and NP-specific ELSIPOT

For total IgG2a, IgG1 and IgG1^a serum Ab, polyvinylchloride microplates (Falcon, BD Biosciences) were coated with the following Abs: polyclonal goat-anti-mouse IgG2a or anti-mouse IgG1. After washing and blocking, sera (diluted in PBS plus 1% BSA) were incubated 3h at room temperature and bound Abs were detected using biotinylated anti-mouse IgG2a, IgG1 or anti-mouse IgG1a^a. To assess the titer of total IgG2a^a and IgG2a^b, competition assays were performed as follows: microplates were coated with the anti-mouse IgG2a^a (20.8.3) or the anti-mouse IgG2a^b (5.7) and incubated with serial serum dilutions mixed with biotinylated IgG2a^a or IgG2a^b monoclonal Abs. For NP-specific Ab titers, microplates were coated with NP₇-BSA (10μg/ml). After incubation with serial serum dilutions, the plates were treated with biotinylated the anti-IgG2a^a, anti-anti-IgG2a^b, anti-IgG2a, anti-IgG1, or anti-IgG1^a monoclonal Abs above mentioned. Biotinylated Abs were revealed using streptavidin-peroxidase (Sigma-Aldrich) followed by addition of the chromogenic substrate 2, 2′-azino-bis (3-ethylbenzthiazoline-6-sulfonic acid) (Sigma-Aldrich) in McIlvain’s buffer (84 mM Na₂PO₄/48mM citrate, pH 4.6). Absorbance was measured on a Molecular Devices Spectra MAX250 model reader. Enzyme-linked immunospot (ELISPOT) for detection of NP₇-specific IgG2a^a and IgG2a^b was performed with 96-well multiscreen membrane filtration plates (Millipore, Billerica, MA) coated with NP₇-BSA, washed, blocked and 50,000 to 500,000 spleen cells/well were cultured at 37°C in 5% CO₂ for 18 h in RPMI 1640 supplemented with 5% FCS. The wells were washed before addition of biotinylated anti-mouse IgG2a^a or IgG2a^b. After washing, the wells were incubated with streptavidin-HRP (BioLegend). ELISPOTs were developed using a kit (BD™ ELISPOT AEC Substrate Set, BD Biosciences). The plates were washed with water and positives were scored manually using a dissection microscope. Each assay was done in triplicate from at least three separate animals.

Statistical analysis

Group comparisons were analyzed by 2-tailed Student’s t-test unless otherwise indicated. P<0.05 was considered significant.

RESULTS

Generation of the γ2a^a-macroself Ag constructs and Tg mice

To express an artificial autoAg reactive to mouse IgG2a (γ2a-macroself Ag, Fig. 1A), we engineered the Tg construct (pURF) depicted in Fig. 1B. The antigenic specificity was obtained by generating a single chain Fv gene derived from the heavy and light chain variable genes of the 20.8.3 monoclonal Ab, which binds specifically to mouse IgG2a^{a, d, e, f, g, h, j, n, o}, but not to mouse IgG2a^{b, c, p} (25). The construct also includes the hinge and Fc domains of the rat IgG1, and the transmembrane and cytoplasmic regions of the H2-K^b gene, which promote ubiquitous cell surface expression (21). A plasmid carrying the chimeric gene under the control of the human ubiquitin C promoter was transiently transfected in human embryonic kidney 293T cells, and the chimeric gene encoded the predicted cell surface protein with binding specificity for the mouse IgG2a^a immunoglobulin (Fig. 1C).

Figure 1.

Design, in vitro testing and generation of γ2a-macroself Ag Tg mice (pURF-Tg). (A) Schematic representation of anti-mouse Igγ2a membrane-bound macroself Ag. A single chain Fv generated from the heavy and light chain variable genes of the 20.8.3 hybridoma is linked to the hinge and membrane proximal domains of rat IgG1 followed by transmembrane and cytoplasmic tail region (Tm/Cy) of H-2K^b gene. (B) Schematic representation of the DNA construct encoding the γ2a^a-macroself Ag used for transfection and microinjection to generate Tg mice. Introns are represented as thin lines. (G₄S)₃ refers to linker codons in one letter amino acid code: GGGGSGGGGSGGGGS. Li stands for the Ig light chain leader exon and first intron. The human ubiquitin C promoter controls the γ2a-macroself Ag transcription. (C) Flow cytometry analysis of transiently transfected HEK 293T cell line with pURF Tg and EGFP-containing vectors. GFP⁺ gated cells were analyzed. (Left), Staining with an anti-rat IgG1 monoclonal Ab compared with empty vector transfected cells. (Right) Binding specificity of the macroself Ag to mouse IgG2a^a (bold line) and IgG2a^b (thin line). γ2a-macroself Ag transfected cells were incubated first with IgG2a^a or IgG2a^b mouse monoclonal Abs followed by a biotin-conjugated anti-mouse IgG2a,b step and binding was revealed with streptavidin conjugate. (D) Flow cytometry analysis of γ2a-macroself Ag expression in bone marrow, spleen and lymph nodes of pURF Tg mice detected with an anti-rat IgG1 monoclonal Ab (thin line, non-Tg cells; bold line, pURF Tg cells).

To analyze the effect of the ubiquitously expressed γ2a-macroself Ag on IgG2a-expressing B cells in vivo, Tg mice expressing the pURF construct were generated. Five B6 Tg lines were derived with distinct γ2a-macroself Ag expression levels ranging from high to intermediate, as measured by flow cytometry (data not shown). One Tg line was selected for further studies because this line expressed high levels of the γ2a-macroself Ag on the surface of virtually all cells tested (Fig. 1D).

Absence of IgG2a^a response in γ2a-macroself Ag Tg mice

To follow tolerance in B cells whose self-reactive receptors emerged after activation, we immunized mice carrying or lacking γ2a-macroself Ag with NP-KLH in RIBI adjuvant (27), boosted on d28, and assessed the IgG2a^a response by ELISA assay and flow cytometry. Mice used in these experiments were IgH^a/b, providing the following control IgG isotypes that do not react with the superAg: IgG2a^b, IgG1, IgG2b, IgG3. In the sera of macroself Ag Tg mice, total IgG2a^a immunoglobulin levels were essentially at background levels at all time points post-immunization, whereas, as expected, IgG2a^a levels in littermate control C57BL/6-Igh^a/b (B6-Igh^a/b) mice were high and further elevated upon immunization (Fig. 2A). By contrast, IgG2a^b and IgG1^a as well as IgG1 Ab levels were comparable to those of littermate control mice. Similar results were obtained in assays of NP-specific Abs (Fig. 2B). These results indicated that γ2a-macroself Ag specifically affected the generation of serum IgG2a^a.

Figure 2.

Absence of IgG2a^a Ab response in IgH^a/b γ2a-macroself Ag Tg mice after priming and recall immunization with NP-KLH in RIBI adjuvant. (A) Mean±SD of total IgG2a^a, IgG2a^b, IgG1^a, IgG2a and IgG1 Ab titers measured by ELISA assays of serum derived from B6 (open lozenges), B6-Igh^a congenic (grey lozenges), IgH^a/b littermate control mice (non-Tg, open squares) and IgH^a/b pURF-Tg mice (black circles) after priming and recall immunization. (B) Mean±SD of NP₇-specific-IgG2a^a, -IgG2a^b, -IgG1^a, -IgG2a and -IgG1 Ab titers measured by ELISA assays on serum derived from IgH^a/b pURF-Tg (black circles) and littermate control mice (open squares) as indicated. The average concentrations of serum Ab and days after priming and recall immunization are indicated on the y- (logarithmic scale) and x-axis, respectively. Three to five mice were analyzed per group at each time point.

Cell surface staining was used initially to follow B cells responding to immunogen. We gated on viable splenocytes that were negative for CD4, CD8, and IgD, and assessed their binding to NP along with other markers of interest. Total NP-specific B cell numbers were similar in immunized Tg and littermate mice. Similarly, the total numbers of NP-specific IgG1 cells were comparable between the macroself Ag Tgs and control mice after priming and recall immunization (Fig. 3A). Contrastingly, we observed a significant reduction of NP-specific-, B220^high-, IgG2a^a-expressing cells after Ag priming and this was even more pronounced after recall, consistent with a block in the development of memory B cells (Fig. 3B). Collectively, these results support the notion that GC IgG2a^a B cells were selectively deleted in γ2a-macroself Ag Tg mice after immunization with a T-dependent Ag.

Figure 3.

Absence of IgG2a^a memory B cell formation in IgH^a/b γ2a-macroself Ag Tg mice. (A) Flow cytometry analysis of spleen cells derived from pURF-Tg and non-Tg littermate control mice 7d post-priming and 5d after recall immunization with NP-KLH in RIBI adjuvant. NP-specific B220^high/CD138⁻ and B220^low/CD138⁺ cells were visualized by successive gating on a broad forward scatter versus side scatter window to include B cell blasts, then on PI⁻, Dump⁻ (CD4⁻, CD8⁻, IgD⁻) to exclude dead cells, T cells and naive B cells. NP-binding IgG1-positive cells were visualized by gating on B220^high cells. The histograms on the right represent the mean numbers of NP-specific and NP-specific B220^high/IgG1⁺ cells in non-Tg (grey bars) and pURF-Tg mice (black bars). (B) Flow cytometry analysis of NP-specific, B220^high/IgG2a^a+ and B220^high/IgG2a^b+ cells in pURF-Tg and non-Tg mice after priming and recall immunization. The histograms on the bottom represent the average numbers of NP-specific IgG2a^b-, IgG2a^a-positive cells and the IgG2a^b:IgG2a^a cell ratio in littermate control (grey bars) and pURF-Tg (black bars) mice. The percentage of positive cells (± SEM) is indicated in each gate. Three mice were analyzed per group at each time point except for the non-Tg mice after Ag recall (n=2).

IgG2a^a B cells are deleted in γ2a-macroself Ag Tg mice

Because the absence of Ag specific IgG2a^a B cells in γ2a-macroself Ag Tg mice immunized with NP-KLH could have been the result of Ag receptor down-regulation rather than cellular depletion, we used a flow cytometry approach to detect both surface and intracellular isotype-switched immunoglobulins (Fig. 4A). Spleen cells, obtained at 14 days post-immunization from IgH^a/b mice, were first surface-stained for B220 and with a pool of Abs against non-B cells as well as non-switched B cells (CD4⁺, CD8⁺, F4/80⁺, Gr1⁺, IgM⁺ and IgD⁺; denoted as “Dump”) and then, to detect cytoplasmic Ig (cIg), permeabilized and stained with Abs to Igκ and IgG isotypes. Among the Dump⁻/cIgκ⁺ spleen cells, two populations of class-switched cells were detected: B220^high/cIgG^intermediate and B220^low/cIgG^high, which correspond to developing memory B cells and plasma cells, respectively (Fig. 4A). In γ2a-macroself Ag Tg mice, virtually no Dump⁻/cIgκ⁺/B220⁺ spleen B cells expressed IgG2a^a, whereas IgG2a^b-, IgG2b-, and IgG1-positive B cells were readily detected. When cell numbers were examined, although the average number of IgG2a^b B cells was not significantly affected, IgG2b and IgG1 B cells were 2- to 3- fold more abundant in Tg mice compared to littermate controls (Fig. 4B).

Figure 4.

Absence of cytoplasmic IgG2a^a-expressing B cells after Ag priming in γ2a-macroself Ag Tg mice. (A) Flow cytometry analysis of B220⁺, cytoplasmic IgG2a^a, IgG2a^b, IgG1 and IgG2b expressing spleen B cells identified by successive gating on side scatter versus forward scatter, Dump⁻ (CD8⁻, CD4⁻, F4/80⁻, Gr1⁻, IgM⁻, IgD⁻) and cytoplasmic Igκ⁺ (cIgk) cells, 14 days after priming with NP-KLH in RIBI adjuvant. The frequency of positive cells, rounded to the nearest 1%, is indicated next to each gate. (B) Histograms representing the mean±SE percentages and numbers of Dump⁻, cIgκ⁺, cIgG2a^a+; cIgG2a^b+; cIgG1⁺ and cIgG2b⁺ cells in littermate control (open bars) and pURF Tg (dark bars) mice. The numbers of mice analyzed are indicated in parenthesis.

These experiments were extended to investigate whether autoreactive B cells in γ2a-macroself Ag Tgs were present at any time after priming with NP-KLH. In these experiments, Tg and littermate control mice of homozygous IgH^a/a allotype were used because on this genetic background, the γ2a-macroself Ag reacts with all IgG2a-expressing B cells, thereby allowing detection of IgG2a^a B cells with a monoclonal Ab specific to an epitope different from that recognized by the macroself Ag. This approach revealed that IgG2a-expressing B cells were rare in the spleens of Tg mice at all time points (Supplemental Fig. 1). In contrast, the average numbers of IgG1 and IgG2b B cells were increased 2- to 3-fold in Tg mice. Taken together, these results indicate that polyclonal and Ag specific B cells reactive to the γ2a-macroself Ag are promptly deleted in the spleen after priming and CSR.

Molecular analysis of IgG2a CSR

To test further the notion that γ2a-macroself Ag led to B cell clonal elimination, we quantified by RT-PCR germline and mature IgG2a transcripts present in RNA from spleens of unimmunized and NP-KLH immunized mice. Transcripts containing the heavy chain variable region, the CH1 and the hinge regions of mouse IgG2a gene were greatly reduced in Tg mice (Fig. 5A and B). Consistent with these results, little or no γ2a mRNA was detected by quantitative real time PCR in the spleens of Tg mice (Fig. 5C). We also quantified the germline γ2a-specific transcripts driven by the Iμ promoter (Iμ-γ2a), which is a product of CSR. Since the Iμ promoter is still active after isotype switch, the amount of transcripts composed of the Iμ exon spliced onto the γ2a 5′-exon correlates with the number of cells that survive after CSR to the IgG2a isotype, despite some transcripts arising from switching of the non-productive allele. After Ag priming, increasing levels of germline γ2a transcripts driven by the Iμ promoter were detected in littermate control mice but, in contrast, these transcripts remained poorly induced in γ2a-macroself Ag Tg mice (Fig. 5A and D). Seven days after Ag priming, when γ2a post-switched transcripts reached their highest levels in spleens of littermate control mice, a 6-fold lower amount was observed in Tg mice. Finally, we quantified γ2a germline transcripts driven by the Iγ2a promoter (Iγ2a-γ2a), which is induced in B cells before CSR (Fig. 5A and E). The Iγ2a-γ2a germline transcripts were induced in the spleen of Ag-primed γ2a-macroself Ag Tg mice with a pattern similar to non-Tg littermates, though somewhat lower at the peak of the response. These results indicate that the artificial autoAg selectively induces deletion of reactive B cells soon after CSR and transcription of functional γ2a mRNA.

Figure 5.

Quantification of V_H-γ2a transcripts and molecular parameters of γ2a CSR in pURF Tg and littermate control mice (non-Tg) after Ag priming. (A) RNA samples purified from total spleen cells were subjected to reverse transcription and PCR reaction. Amplification products were quantified by Southern blot using an IgG2a specific probe. RNA samples without reverse transcription were used as negative control (“-“ labeled lines). Four-fold serial template dilutions were tested. (Top row) PCR detection of V_H-γ2a (V_H-CH1-hinge) transcripts. (Middle row) Detection of post switched germline transcripts driven by the Iμ promoter (Iμ-γ2a post-switch). (Third row) Detection of Iγ2a germline transcripts driven by the Iγ2a promoter (Iγ2a-γ2a germline). (Bottom row) Actin PCR is used as a DNA loading control. Representative blots are shown. Numbers above blots indicate days after Ag priming. Blots were cropped for clarity. No additional bands were detected in the full-length blots. (B) Relative quantification of V_H-γ2a transcripts as detected in A. (C) Quantification of γ2a (CH1-hinge) transcripts by quantitative real time PCR in RNA isolated from spleen cells of the indicated mouse types. A minimum of three mice was analyzed per group at each time point. (D) Relative quantification of Iμ-γ2a post-switched and (E) Iγ2a-γ2a germline transcripts as detected in A from three experiments. The results are representative of three to four independent experiments.

Adoptive transfer studies

Because the γ2a-macroself Ag is expressed on all cells tested, including B cells, we assessed tolerance induction after primary immunization with a T-cell-dependent Ag in an adoptive transfer model in which γ2a-macroself Ag Tg mice served as hosts of wild-type IgH^a/b donor spleen cells. As with non-manipulated Tg mice, two weeks after NP-KLH priming, surface IgG2a^a B cells were largely depleted, whereas non-reactive IgG2a^b B cells were readily detected in the spleen of Tg mice (Fig. 6A and B). Similarly, using ELISPOT assays, NP-specific IgG2a^a Ab-forming cells (AFC) were not detected in the spleen of Tg hosts, whereas the numbers of IgG2a^b AFC were similar to those found in non-Tg chimeras (Fig. 6C). These results supported the notion that wild-type isotype switched autoreactive B cells developing in γ2a-macroself Ag Tg mice undergo rapid deletion, presumably by an apoptotic mechanism.

Figure 6.

Flow cytometry and NP-specific ELISPOT analyses of tolerance induction in radiation chimeras using γ2a-macroself Ag Tg hosts. IgH^a/b/CD45.2, wild type or Bcl2 Tg spleen cells were used to reconstitute sub-lethally irradiated IgH^b/CD45.1 pURF-Tg or non-Tg control recipients. Spleen donor cells are indicated on the left side of each arrow. Mice were analyzed 14 day post reconstitution and priming with NP-KLH in RIBI adjuvant. (A) Analysis of Igκ⁺ spleen cells for surface expression of B220 and IgG2a^a (top plots) or IgG2a^b (bottom plots). Plasma cells were identified as falling in the lower right boxes, as indicated. The percentage in each gate, rounded to the nearest 1%, is indicated in the upper right corner of each plot. (B) Average numbers (± SE) of IgG2a^a- and IgG2a^b-expressing B cells in the spleen of the indicated radiation chimeras. (C) Mean±SE numbers of IgG2a^a and IgG2a^b NP-specific Ab-forming cells (AFC) per 10⁶ spleen cells counted by ELISPOT assay. The numbers of mice analyzed per group are indicated in parentheses.

Enforced expression of Bcl2 reduces the deletion of autoreactive IgG2a-switched B cells

Splenocytes overexpressing Bcl2 in the B cell compartment (24) were tested by adoptive transfer to determine whether enforced expression of this anti-apoptotic molecule would suppress the deletion of macroself Ag-reactive IgG2a switched B cells. Indeed, IgG2a^a cells were substantially rescued in the macroself Ag recipients that received Bcl2 Tg cells (Fig. 6A-B). Moreover, although the average numbers of NP-specific IgG2a^a AFC were 1/3 of those in macroself Ag-free hosts of Bcl2 Tg cells, they were similar to those obtained with non-Tg recipients of wild-type spleen cells (Fig. 6C). In the absence of macroself Ag, chimeras reconstituted with Bcl2 Tg cells had elevated numbers of surface IgG2a^a and IgG2a^b B cells compared to wild type mice. These differences also correlated well with the increased numbers of NP-specific IgG2a^a and IgG2a^b AFC by ELISPOT assays. We conclude that the survival-enhancing effects of Bcl2 partially inhibit deletion of autoreactive isotype-switched B cells, thereby promoting the development of Ag-specific autoreactive plasma cells.

MRL-Fas^lpr mice fail to delete autoreactive IgG2a B cells

We next tested whether lupus-prone MRL-Fas^lpr mice are capable of eliminating γ2a-macroself Ag-reactive memory B-cells by introducing the Tg to this background. Similar to above, 6-wk-old mice were injected with NP-KLH and analyzed 14 days later for cIgG2a- and cIgG2b-expressing B cells. In non-Tg controls, MRL-Fas^lpr had up to 20-fold greater numbers of cIgG2a- and cIgG2b-expressing B cells compared to B6-Igh^a mice, consistent with Fas^lpr-associated B cell expansion (Fig. 7A, Table I). Furthermore, as previously shown, the numbers of IgG2a-expressing B cells in both the memory B (B220^hi/cIgG2a^low) and plasma cell (B220^low/cIgG2a^high) compartments were markedly reduced in macroself Ag Tg compared to non-Tg B6-Igh^a mice. In striking contrast, cIgG2a B cells were abundantly present in MRL-Fas^lpr macroself Ag Tg mice with a similar number of memory B cells and a 2- to 3-fold greater number of plasma cells detected. Consistent with these results, IgG2a anti-NP and anti-chromatin were detected in the sera of NP-KLH-immunized MRL-Fas^lpr γ2a^a-macroself Ag Tg mice (Fig. 7B). Moreover, total IgG2a levels in Tg were similar to non-Tg MRL Fas-deficient mice suggesting that loss of tolerance to the γ2a-macroself Ag occurred prior to immunization.

Figure 7.

MRL-Fas^lpr γ2a-Tg mice fail to delete IgG2a-expressing B cells after NP-KLH priming. Six week-old MRL-Fas^lpr and MRL-Fas^lpr pURF-Tg mice were analyzed by flow cytometry for cytoplasmic IgG2a and IgG2b expression two weeks after immunization with NP-KLH in RIBI adjuvant. Age-matched B6-Igh^a pURF-Tg and littermates (non-Tg) were used as controls. (A) Spleen cells from the indicated mice were analyzed by flow cytometry for B220⁺, cytoplasmic IgG2a, and IgG2b-expressing spleen B cells as described in Fig. 4 except that an anti-TcRβ was added in the Dump channel to counter select double-negative T-cells. Top panels are Dump⁻ cells. The percentage in each gate, rounded to the nearest 1%, is indicated. (B) Mean±SE of total, NP-specific and anti-chromatin IgG2a Abs in the serum of naïve and immunized littermate B6-Igh^a and MRL-Fas^lpr controls and pURF-Tg mice, 14 days after Ag priming. Three to five mice were analyzed in each group. (C) Allotype-specific serum IgG2a in (B6-Fas^lprxMRL-Fas^lpr)F1 pURF mice born to IgG2a^b allotype (B6-Fas^lpr) mothers. IgG2a levels below the detection limit are indicated as <1 μg/ml. *P<0.0001 between nTg and Tg mice. (D) Allotype-specific serum IgG2a and IgG1 in Fas-deficient B6-Igh^a pURF Tg mice. Littermates for this experiment were (B6-Igh^a Fas^lpr/+ × B6-Igh^a Fas^lpr/+ pURF Tg)F1 mice. IgG2a concentrations <1 μg/ml were below detection. n=5-9 mice, except 35 d Tg lpr/+ where n=2. *P<0.05.

Table I.

IgG2a-switched B cells escape tolerance in MRL-Fas^lpr γ2a-macroself Ag mice.

Genotype		Number of positive cells/104 B cells (Mean ± SE)
		B220high/ cIgG2alow	B220low/ cIgG2ahigh	B220high/ cIgG2blow	B220low/ cIgG2bhigh
Non-Tg	B6-Igha (3)	1±0.2	0.1±0.05	0.4±0.06	0.08±0.02
	MRL-Faslpr (4)	15 ± 4.5	7 ± 2.1	3.1 ± 1	3.7 ± 2
pURF-Tg	B6-Igha (3)	0.03±0.01*	0.007±0.001*	3.3±0.7*	0.11±0.04
	MRL-Faslpr (5)	25 ± 6	18 ± 3.5 *	4.7±2	2.9 ± 0.1

^*P < 0.05 between γ2a-macroself Ag (pURF) transgenic compared to wild-type littermate of the same genetic background. Bold type indicates significant (P < 0.05) differences between MRL-Fas^lpr compared to B6-Igh^a mice in non-transgenic and transgenic mice. Number of mice analyzed indicated in parenthesis.

A possible explanation not requiring loss of tolerance, however, is that maternal transfer of IgG2a^a through the placenta or milk could have bound to and neutralized the γ2a-macroself Ag in offspring of MRL-Fas^lpr mice. Indeed, low levels of cell surface γ2a-macroself Ag not bound by IgG2a were detected when peripheral blood cells from adult MRL-Fas^lpr γ2a-macroself Ag Tg mice were incubated with fluorescent-conjugated IgG2a^a and fluorescent-labeled-anti-rat IgG1 (data not shown). To address this issue and to document spontaneous loss of tolerance to the γ2a-macroself Ag, we measured serum allotype-specific IgG2a in 1 to 8 week-old F1 littermates from female B6-Fas^lpr (IgG2a^b) crossed with MRL-Fas^lpr (IgG2a^a) males heterozygous for the γ2a^a-macroself Ag transgene; whereby mothers are IgG2a^b and non-Tg, while offspring are IgG2a^a/b with about half being γ2a^a-macroself Ag positive (Fig. 7C). In the non-Tg (B6-Fas^lprxMRL-Fas^lpr)F1 pups, IgG2a^a was initially detected at 3 weeks of age with levels progressively increasing at 4 and 8 weeks. Similarly, IgG2a^a was first detected at 3 weeks in γ2a^a-macroself Ag Tg mice, but initially at lower concentrations (3 and 4 weeks) until 8 weeks when levels were comparable to non-Tg littermates. In contrast, IgG2a^b levels in both groups were similar throughout and, as expected because of maternal transfer of IgG, were present even at the earliest time point (1 week). During this same period, surface expression of the γ2a^a-macroself Ag, detected by a monoclonal anti-rat IgG1, was present at similar or greater levels as controls at 1-2 weeks and in peripheral blood cells, but were reduced at 8 weeks in the thymus (Supplemental Fig. 2A-B). In contrast, the capacity of macroself Ag to bind IgG2a^a in the (B6xMRL)F1-Fas^lpr Tg mice, although similar to B6-Tg mice at 1 and 2 weeks, was significantly reduced by 4 weeks and essentially nil by 8 weeks (Supplemental Fig. 2C), consistent with the initial detection and subsequent rise of serum IgG2a^a from weeks 3 to 8. A similar breach of tolerance to the γ2a^a-macroself Ag was also observed in Fas-deficient B6-Igh^a macroself Ag Tg mice, but appeared less complete; serum IgG2a although present, was at significantly lower concentrations in Tg compared to non-Tg B6-Fas^lpr mice at 80 days (Fig. 7D and Supplemental Fig. 3). Taken together these findings indicate that loss of tolerance to the γ2a^a-macroself Ag in Fas-deficiency occurs spontaneously despite high levels of macroself Ag (similar to levels in Tg B6 mice), can be detected at the earliest appearance of IgG2a^a, is not caused by blocking of the γ2a-macroself Ag by maternal transmission of IgG2a, and is enhanced by MRL background genes.

DISCUSSION

Memory B cell differentiation in response to T-dependent Ags poses a great challenge for self/non-self discrimination because the V-gene somatic mutations that promote affinity maturation may also render cells self-reactive. Somatic mutation and class switching in response to T cell-dependent immune reactions are believed to occur mainly, but not exclusively, in GCs (28-33). Regardless of where such reactions take place, however, their consequences for self-reactivity need to be controlled. The dynamic nature of the GC reaction, the low frequency of lymphocyte precursors, and the extraordinary heterogeneity of B cell receptor (BCR) specificities have made in vivo assessment of memory B cell tolerance difficult. Perhaps, the greatest difficulty has been to target tolerogens to newly acquired B cell specificities following SHM and CSR.

We describe here a novel Tg mouse model carrying an anti-IgG2a-reactive artificial autoAg (γ2a-macroself Ag) that induces memory B cell tolerance in normal mice during immune response to a T-dependent Ag. Ubiquitous expression of the membrane-bound γ2a-macroself Ag or even on non-bone marrow-derived cells induces specific deletion of Ag-specific IgG2a^a B cells soon after primary immunization. Moreover, IgG2a^a B cells are effectively absent at any time after Ag priming even when assayed by intracellular staining. From these results it can be concluded that the absence of Ag-specific IgG2a^a B cells is not due to Ag receptor down regulation, that tolerance induction is solely dependent on expression of the IgG2a^a heavy chain, and finally, that autoreactive primary plasma cells, post-GC memory B cells and long-lived plasma cells are absent after Ag priming. Furthermore, RNA analysis shows that, in the presence of the macroself Ag, deletion of IgG2a^a B cells appears to be a very rapid process, occurring early after Ag priming and CSR likely within or soon after emigration from the GC.

As a result of the macroself Ag expression, the memory B cell and plasma cell compartments are virtually completely purged of autoreactive B cells without compromising the amplitude of the Ag-specific Ab response. In fact, we observed compensatory increases of other non-autoreactive isotype-switched B cells presumably because of a relative excess in B cell trophic cytokines and reduced competition for T-cell help and/or BCR binding to Ags on FDCs as a result of the deletion of the entire IgG2a^a isotype subset. This dual effect of autoreactive B cell deletion combined with simultaneous expansion of non-autoreactive B cells may be central to efficient memory responses for foreign Ags. Indeed, tolerance is sustained over time by the exclusion of autoreactive Ab-forming cells from the long-lived bone marrow plasma B cell pool, which is responsible for the continuous production of serum Abs (34, 35).

As noted, previous studies have suggested potential mechanisms by which newly formed autoreactive B cells might be deleted in GCs. One is the blocking of BCR-mediated B cell contact with FDCs by an excess of soluble self-Ag (14) and the other, insufficient cognate T cell interaction because of tolerance of T cells to self-Ags (15, 36). In both instances, autoreactive B cells succumb to insufficient survival signals. Here we show that despite adequate FDC and T cell function, as indicated by the normal development of other Ag-specific IgG isotype-expressing B cells following immunization, B cells that bind to macroself Ag are nonetheless deleted. Our findings thus identify a selection mechanism that deletes autoreactive B cells that have acquired self-reactivity to surface-expressed self-Ags during the process of class switching. Since class switch to IgG2a occurs in the GC and during the same time frame as SHM, our findings suggest a model for studying the fate of B cells that acquire reactivity to membrane bound self-Ags during SMH and CSR. There are several reasons to believe that deletion induced by the γ2a^a-macroself Ag might take place in the germinal center. First, the absence of IgG2a^a-expressing B cells following immunization indicates censorship by some process of which deletion appears the most likely and this process must affect B cells that class switched in GCs. Second, the inability to detect IgG2a^a B cells in normal mice indicates that the censoring occurs rapidly. Third, the dependence of this process on Fas^lpr is consistent with this occurring in the GC, where B cells express high levels of Fas. This Fas-dependent deletion of B cells that have acquired self-reactivity would seem to cover a potential gap in the other aforementioned tolerance mechanisms since it is conceivable that self-reactive B cells through engagement of membrane-expressed self-molecules could interact with both FDCs and activated T cells, and escape deletion.

Several possible ways in which deletion might occur can be envisioned. First, since engagement of IgG2a^a class-switched receptors to the IgG2a-macroself Ag will occur regardless of affinity maturation, B cells that have class-switched to IgG2a^a prior to affinity maturation will be at a competitive disadvantage with B cells expressing other isotypes in acquiring Ag for presentation to T cells. Thus, deletion will occur not because of a lack of T cell helper cells, per se, but because of the inability of self-reactive B cells to generate high affinity receptors. A another related possibility is that B cells might be unable to adequately internalize antigen because of antigen receptor binding to macroself Ag on surrounding cells, making them incapable of receiving T cell help. A third possibility is that B cells engaging cell surface self-Ags are deleted by negative selection. This contrasts with the FDC and T cell-mediated tolerance mechanisms whereby GC B cells undergo apoptosis because of insufficient survival signals (lack of positive selection). Moreover, while negative selection in early B cell development in the bone marrow and in the periphery for immature B cells (T1 subset) is determined solely by affinity threshold, selection in GCs must depend on an additional factor or factors since BCRs with high affinity to foreign Ags are not only maintained, but positively selected. Of possible relevance is a previous in vitro study showing that prolonged BCR cross-linking induced death of CD40-activated GC, but not similarly activated naïve or memory B cells (37). Although the basis for this differential effect was not determined, it nevertheless suggests an intrinsic difference in the response of GC B cells to BCR engagement. It is also possible that recent GC emigrants might be susceptible to BCR-mediated cell death similar to naive B cells (8, 38). Peripheral deletion of IgM⁺IgD⁺ B cells, however, in marked contrast to our findings with GC B cells, is not dependent on Fas (39).

Our findings contrast with an earlier study involving Tg mice expressing influenza hemagglutinin (HA) as a membrane-bound “neo-self” Ag, which concluded that specificity for self-Ags does not prevent differentiation of autoreactive memory B-cells after viral infection (36). However, subsequent analyses by these investigators showed that expression of the HA neo-self Ag on lymphoid cells was very low in this Tg line, thereby suggesting Ag ignorance. Indeed, when a line with high HA neo-self Ag expression was assessed, absence of HA-specific IgG Abs after immunization was observed (20), leading to reformulation of the previous conclusion, now in support of clonal deletion. The model used, however, did not allow firm conclusions in this regard, since tolerance was also imposed at early B cell developmental stages.

The evidence that isotype switched IgG2a^a B cells were deleted in the macroself Ag Tg mice included lack of cytoplasmic IgG2a^a B cells, absence of IgG2a^a AFC and NP-binding B cells, and reduction of IgG2a^a and post-switched germline Iμ–γ2a transcripts in immunized mice. Accordingly, Bcl2 Tg B cells transferred into macroself Ag hosts showed increased survival of IgG2a^a switched B cells following immunization. Thus, it appears that apoptosis is directly involved in the deletion of autoreactive isotype-switched B cells. This finding is consistent with previous observations that Bcl2 overexpression promoted the survival of B cells expressing a Tg BCR with dual reactivity for the hapten Arsenate and nuclear autoAgs (18, 19), as well as with the development of a lupus-like syndrome in Bcl2 Tg mice (24). Contrastingly, however, others showed that apoptotic death of GC B cells provoked by high doses of soluble Ags was not affected by Bcl2 overexpression (13, 14), raising the possibility that Bcl2 may instead act by promoting the survival of self-reactive primary plasma cells. In this case, primary Ab-forming cells and memory B cells were believed to derive from two distinct precursors (40). There is, however, no a priori reason for such differential effects of Bcl2, and indeed our findings support pro-survival effects for both self-reactive IgG2a primary plasma and memory B cells.

Apoptosis of autoreactive B cells can also be induced by an extrinsic pathway primarily through the interaction of the Fasl with the Fas receptor (41). Indeed, Fas (lpr) or Fasl (gld) gene mutations in mice are associated with lupus-like manifestations, including high titers of somatically-mutated IgG autoAbs (42, 43). Similarly, autoimmunity accompanies Fas or Fasl mutations in human with the autoimmune lymphoproliferative (ALPS) or Canale-Smith syndromes (44). Functional complementation studies have clearly shown that absence of Fas expression on B cells is responsible for most of the serologic and histologic abnormalities in MRL-Fas^lpr mice (45, 46).

Earlier studies with MRL-Fas^lpr (H-2^k) Tg for anti-H-2K^k (39, 47) or Fas^lpr double Tg for HEL (soluble or membrane bound) and anti-HEL BCR (48) showed no clear defects in B cell tolerance, though receptor editing and anergy are predicted to be less efficient (49, 50). These findings suggested that the Fas-FasL apoptotic pathway plays no major role in tolerance during the early B cell developmental stages. By contrast, in these same mice, high levels of IgG autoAbs to lupus associated Ags developed with time, suggesting that tolerance defects mapped to the class switched compartment. Consistent with this conclusion, previous studies with Tg anti-ssDNA BCRs showed that in normal mice these B cells were functionally silenced by central tolerance-mediated anergy (51) and by an undefined checkpoint in the GC (52), whereas in MRL-Fas^lpr mice, such self-reactive B cells were present in the periphery and, through SHM and secondary light chain rearrangement, gained the ability to bind other autoAgs such as dsDNA and cell nuclei and thus become pathogenic (53). More recently, specific targeted deletion of Fas in B cells during class-switch recombination was found to be sufficient to induce lymphoproliferation in (B6xMRL)F1-Fas^+/lpr mice (54). Although this suggests that B cell tolerance during class switch might be defective, B cell tolerance was not examined (54). Other studies, however, focused on tolerance induction in B cells undergoing a T cell dependent Ab response have concluded that Fas played no role (12, 55, 56). The present findings, made possible by the development of a new macroself Ag model, clearly document that the Fas pathway is the main mechanism responsible for the elimination of autoreactive switched B cells shortly after their generation. Furthermore, since Fas is highly expressed in GC B cells (55) and participates in memory B cell repertoire selection (57) it can be concluded that escape from censoring of autoreactive IgG-switched B cells in Fas-mutant mice most likely occurs at this site (51), and also outside GCs at the extra-follicular T zone-red pulp border where autoreactive B cell proliferation and SHM predominate in MRL-Fas^lpr mice (58).

Importantly, our findings show that tolerance to the γ2-macroself Ag is affected not only by Fas, but also by MRL genes and Bcl-2, and furthermore suggest that the Fas^lpr and MRL background effects on tolerance are additive. Based on this, we believe that defective elimination of newly generated autoreactive B cells post-CSR/SHM may be a major underlying cause of lupus and other autoimmune diseases.

An increased frequency of SHM-related low-affinity poly- and self-reactive B cells in the circulating IgG⁺, but not IgM⁺ memory pools in normal human individuals was reported (59, 60). This suggests differences in how these two compartments are established and raises the possibility that abnormalities in the activation of self-reactive IgG⁺ memory B cells might contribute to the development of autoimmunity in susceptible individuals. Here, we show that lupus-associated Fas^lpr mutation results in loss of tolerance to autoreactive class-switched B cells, suggesting that autoimmunity, at least in this case, is more likely mediated by loss of tolerance to newly generated self-reactive B cells rather than the aberrant activation of self-reactive IgG⁺ memory B cells.

In the γ2-macroself Ag model, a potential limitation is that interaction of the macroself-Ag with the BCR occurs at the constant region rather than at the variable region where normal binding of antigen occurs. Antibody binding to the constant region, however, activates B cells similar to normal antigen-BCR interactions (61) and has been used to successfully study B cell development and tolerance in other mouse models (62-65). Thus, it is likely that the findings in this study are applicable to B cells that acquire self reactivity to membrane-bound self-Ags during SHM/CSR.

In summary, the present study with a novel system documents that autoreactive memory B cells emerging after an immune reaction to T-dependent Ags are normally tolerized following isotype CSR and SMH, a finding consistent with the existence of a “window of tolerance” during memory B cell repertoire formation (10, 60). It further demonstrates that Fas mutation and associated humoral autoimmunity is characterized by defects in late checkpoints pertaining to memory B cell tolerance. It is likely that proliferation and differentiation of otherwise innocuous low affinity autoreactive B cells in the periphery of Fas-defective lupus-prone mice are induced through the simultaneous engagement of BCR and TLR by self-nucleic acids and related immune complexes in concert with DC and T cell activation (reviewed in (66-68)). Considering that B cell tolerance defects are central to the pathogenesis of lupus in predisposed strains of mice (69) and humans (70, 71), the macroself Ag approach offers a versatile system by which these defects, from the earliest to the latest checkpoints, can accurately be identified.

ABBREVIATIONS USED

Ab

antibody

AFC

Ab-forming cell

Ag

antigen

B6 (C57BL/6)

BCR

B cell receptor

CSR

class-switch recombination

FDC

follicular dendritic cell

GC

germinal center

HA

hemagglutinin

NP, KLH

keyhole limpet hemocyanin; 4-hydroxy-3-nitrophenyl acetyl

pURF

γ2a-macroself Ag Tg construct

SMH

somatic hypermutation

Tg

transgene or transgenic