Most mature B lymphocytes express one BCR L chain, κ or λ, but recent work has shown that there are exceptions in that some B lymphocytes express both κ and λ and some even bear two different κ L chains. Using the anti-DNA H chain-transgenic mouse, 56R, we find that B cells with pre-existing autoreactivity are especially subject to L chain inclusion. Specifically, we show that isotypic and allelic inclusion enables autoreactive B cells to bypass central tolerance giving rise to B cells that retain dangerous features. One receptor in dual receptor B cells is an editor L chain, i.e., neutralizes or alters self-reactivity of the 56R H chain transgene. We compare the 56R mouse when on the C57/BL/6 background, a strain prone to autoimmunity, with that of 56R when on the BALB/c background, a strain that resists autoimmunity. In the B6.56R mouse, polyreactive B cells with dual L chain move to the follicular B cell compartment. Their localization in the follicular compartment may explain the ease with which B cells in the B6.56R differentiate into autoantibody-producing plasma cells. Likewise, in the BALB/c.56R mouse, dual L chain B cells are found in the follicular B cell compartment. Yet, the lack of autoantibody-producing plasma cells in the BALB/c.56R suggests that postfollicular tolerance checkpoints are intact. The Jκ usage in dual κ L chain B cells suggests increased receptor editing activity and is consistent with the expected distribution of Jκ genes in our computational model for random selection of Jκ.

Anti-DNA Abs are a major pathogenic factor in autoimmune diseases such as lupus erythematosus. In healthy individuals, anti-DNA Abs are regulated by deletion, anergy, and receptor editing (1, 2, 3, 4, 5, 6). All three mechanisms lead to a self-tolerant peripheral B cell repertoire with few if any remaining autoreactive B cells. Among the regulatory mechanisms, receptor editing deserves special attention because the peripheral B cell repertoire is widely shaped by receptor editing (7). Editing is based on secondary Ig gene rearrangement mainly at the L chain locus by replacing an autoreactive with a nonautoreactive receptor (8, 9). Another way of dealing with autoreactive B cells is the expression of two different L chains, κλ and κκ; such isotypically and allelically included dual receptor B cells are part of the peripheral repertoire in humans and mice (10, 11, 12, 13, 14).

In our anti-DNA 56R H chain-transgenic mouse (15, 16, 17), B cells bind to dsDNA in association with most L chains, but the 56R H chain can be silenced when paired with “editor” L chains. Examples of such editor κ L chains are Vκ20, Vκ21D, and Vκ38C (17, 18). Among λ L chain, Vλx edits the 56R H chain transgene (19). Association of the 56R H chain with an editor L chain sufficiently alters anti-DNA reactivity allowing B cells to leave the bone marrow. Also κλ dual L chain B cells are found in the 56R mouse (17, 20). λ1 in association with the 56R is self-reactive, and self-reactive λ1 B cells are thought to survive by coexpressing a κ L chain (21). In the 56R, dual κλ B cells have been found in the marginal zone (MZ)3 of the spleen (20, 22, 23).

We have recently shown that failure to edit or incomplete BCR editing in the 56R strain gives rise to polyreactive autoantibodies (20). This autoantibody polyreactivity is mediated by the editor L chains Vκ38C and Vκ20, which incompletely edit the 56R H chain transgene. 56R with Vκ38C binds to dsDNA and phosphatidylserine as well as a series of (self-) proteins such as myelin basic protein, thyroglobulin, cytochrome c, histone, β-galactosidase, and insulin. 56R associated with Vκ20 binds to dsDNA and phosphatidylserine (20). Polyreactive features have also been shown for the editor L chain Vλx (24). During the course of the chronic graft-vs-host reaction (cGvH), a model of induced autoimmunity, incompletely edited B cells and κλ dual receptor B cells produce high titers of polyreactive autoantibodies that contribute to cGvH disease (20). Hence, these findings suggest that dysregulation of anti-DNA Abs can occur by incomplete L chain editing and at the level of dual receptor κλ B cells (20). In humans, the frequency of polyreactive autoantibodies is increased in autoimmune patients (8, 9). Although, to date, a mechanistic link between polyreactive autoantibodies and disease pathogenesis is still missing.

Recent findings showed that the genetic background causes differences in the peripheral L chain repertoire in the 56R mouse: differences in the L chain repertoires correlate with differences in susceptibility to autoimmunity (25, 26). Hybridoma panels showed that the 56R-transgenic mouse on a C57BL/6 (B6) background is permissive for polyreactive Abs and mainly expresses Vκ38C. In contrast, 56R on a BALB/c background mainly expresses Vκ21D (17, 25, 26), an efficient editor of anti-DNA activity (17, 20). These differences in the L chain repertoires are correlated with the observation that the B6 strain is prone to autoimmunity, whereas BALB/c is more resistant (27, 28, 29, 30, 31). Thus, in lupus-susceptible mice the L chain repertoire is shifted toward incompletely edited polyreactive B cells and suggesting a role of polyreactive Abs in autoimmune disease.

To investigate the regulation of incompletely edited and dual receptor B cells in autoimmunity, we examined peripheral B cell subsets in detail by single-cell PCR. Specifically, B cells from the MZ and follicular compartment in the B6.56R and in BALB/c.56R were analyzed to define their L chain repertoires and the localization of incompletely edited B cells as well as dual L chain B cells. We determined the frequency and the type of κλ and κκ dual L chain B cells in the two strains and were interested to find whether receptor editing contributes to the generation of κλ or κκ dual L chain B cells.

The L chain repertoire was skewed toward the polyreactive L chain Vκ38C in B6.56R, whereas in BALB/c.56R, we saw a bias toward Vκ21D. This was in agreement with previous results (17, 25, 26). In this study, we show that the differences in the L chain repertoires between both strains were most notable in B cells of the MZ and less prominent in the follicular zone compartment. We found B cells with dual κκ L chain in the MZ. We also demonstrate that B cells with dual κκ L chain locate to the follicular zone B cell compartment of both strains, B6.56R and BALB/c.56R. This suggests that the lack of autoantibodies in the BALB/c.56R (25, 26, 32) is due to intact postfollicular tolerance checkpoints as compared with B6.56R. In B cells with dual κκ L chain, one of the two κ L chains was the editor Vκ21D or Vκ38C. B cells with in-frame λ message always had an in-frame κ L chain rearrangement, hence in the 56R mouse, B cells with λ L chain represent a population in which isotypic exclusion is broken.

The generation of site-directed 3H9/56R knockin mice (56R mice) has been described previously (16, 17) and resulted in the BALB/c.3H9/56R mice. The 3H9/56R has been backcrossed onto the C57/BL/6 background for 14 generations to engender B6.3H9/56R. The 3H9/56R transgene was determined by PCR amplification of tail DNA (15, 17). All mice were maintained in our mouse colony at the University of Chicago. All animal care and procedures were conducted in accordance with the Animal Welfare Act.

Before staining, cells were preincubated with excess mouse anti FcγIII/IIR (2.4G2; BD Pharmingen) to block FcRs. Dead cells were excluded by propidium iodide staining (0.33 μg/ml). The following mAb were used for staining: RA3-6B2-allophycocyanin (anti-B220), 7G6-FITC (anti-CD21), B3B4-PE (anti-CD23) (all obtained from BD Pharmingen). Single-cell PCR was done as described (8). MZ B cells, B220+CD21intCD23high, from one B6.56R mouse were sorted on a FACSVantage (BD Biosciences) into 96-well PCR plates containing 4 μl of lysis solution (0.5× PBS containing 10 mM DTT, 8 U of RNA in (Promega) and 0.4 U of 5′-3′ Prime RNase inhibitor (Eppendorf) and were immediately frozen on dry ice. All samples were stored at −70°C. The purity of single-sorted B cells was 98.2% for B cells from the MZ and 98.8% for B cells from the follicular zone compartment (data not shown). cDNA was synthesized in a total volume of 14 μl in the original 96-well PCR plate. RNA from single cells was reverse-transcribed at 37°C for 55 min with 150 ng of random hexamer primer (pd(N)6; Amersham Pharmacia Biotech), 0.5 μl of dNTP mix (10 mM each), 1 μl of 0.1 M DTT, 0.5% (v/v) Nonidet P-40, RNase inhibitors (4 U of RNasin and 6 U of prime RNase inhibitor), and 50 U of Superscript III reverse transcriptase (Invitrogen Life Technologies). Ig gene rearrangement was tested using the degenerate primer Vκ(S) (33) with Cκ reverse primers, and specific primers for Vκ21D, Vκ38C with Cκ reverse primers (for primers, see Ref. 26), and primers specific for the 56R H chain transgene and reverse primers for IgM (26), and primers for Vλ and reverse primers for Cλ (20). Products were amplified by nested PCR in 40-μl reactions containing 20 pM primers and 1.2 U of HotStar TaqDNA polymerase (Qiagen). The PCR conditions for were as follows: denaturation for 4 min at 94°C, then 30 s at 94°C, 30 s at 55°C (for Vκ21D and Vκ38C and 56R H chain) or at 50°C (for Vκ(S) (as previously suggested in Ref. 34 and Vλ1/2) and 55 s (first PCR) or 45 s (nested PCR) at 72°C for 50 cycles, and elongation was allowed for 10 min at 72°C. As a template, 2 μl of cDNA were used for the first PCR and 3 μl of first PCR product were used for the nested PCR. The amount of cDNA from a single cell allowed for five PCR assays that were exhausted with the above-listed PCRs. PCR products were sequenced using primers for Vκ20, Vκ21D, Vκ38C, Vκ(S) and Cκ, and Vλ1/2 and Cλ as indicated in Results. Sense and antisense strands of each L chain sequence were aligned and the type of L chain was identified (supplemental table IB-F)4 using a National Center for Biotechnology Information Blast search for Ig genes. Alignments of sense and antisense strand were checked for base ambiguities using Sequencher DNA Software (Gene Codes).

The computation of random and sequential Jκ usage in κ L chain rearrangements (see Fig. 3, B and C) was conducted using The Mathematica Software Package (Wolfram Research). The calculations somewhat follow Louzoun et al. (35) and were computed under certain assumptions to get the best fit to our data. See detailed description of formulas online (www.phys.huji.ac.il/∼eldadb/Supplement.pdf).

Values of p were calculated using the χ2 test and the Student t test for small numbers (see Fig. 2B).

Analysis of the mismatch score of the degenerate primer Vκ(S) to framework region (FR) 3 of all mouse Vκ genes (supplemental table IA) was done using the Fuzznuc program as part of the Emboss software package. Alignment of Vκ and Vλ L chain genes (supplemental table I, B–F) was conducted using Sequencher Software and Clustal X.

To determine the BCR repertoires of peripheral B cell subsets in the 56R mouse, single B cells from the splenic MZ (B220posCD21highCD23low) and from the follicular zone compartment (B220posCD21intCD23high) were sorted and analyzed by nested RT-PCR. In both strains, between 30 and 45% of all splenic peripheral B cells display the phenotype of MZ B cells (20, 23). κ L chain transcripts were amplified using specific primers for the κ editor L chains Vκ38C and Vκ21D, and using the degenerate primer Vκ(S) (33). The degenerate primer Vκ(S) binds to FR3 of most κ L chains and amplifies >80% of all κ L chains (M. Schlissel, personal communication) including Vκ38C and Vκ21D. We now estimate, based on our primer alignment analysis to all 98 Vκ genes, that only very few κ L chains will be ignored by the Vκ(S) primer (supplemental table IA). Sequencing analysis is necessary to determine the type of κ L chain amplified by this PCR. The presence of λ L chain was tested using specific primers for Vλ1 and Vλ2 L chain. We compared B cells from the MZ and follicular zone compartment in 56R mice on the B6 and on the BALB/c background.

The sum of all analyzed peripheral B cells confirmed previous results and showed that the ratio of Vκ38C and Vκ21D L chain was reversed in the B6.56R and the BALB/c.56R strain (25, 26, 32) (Fig. 1,A). In the B6.56R, the L chain repertoire was skewed toward the polyreactive incomplete editor L chain Vκ38C, whereas in BALB/c.56R, the B cell repertoire was biased toward the complete editor L chain Vκ21D, a L chain that neutralizes binding to self (17, 20). Such skewing of the L chain repertoire is not found in nontransgenic B6 and BALB/c mice: Vκ38C in the B6 and all eight Vκ21 family members in the BALB/c are represented with 0.7 and 4%, respectively (26, 36). In the majority of B cells in B6.56R (79.1%) and in BALB/c.56R (95.7%), we could identify a κ L chain rearrangement amplified with the degenerate primer Vκ(S). We also found that λ L chain was amplified in a similar number of B cells in B6.56R (9.9%) and in BALB/c.56R (10.3%) (Fig. 1 A). The differences in the L chain repertoires held true and reached statistical significance for Vκ21D and Vκ38C when only B cells were analyzed in which we could identify the presence of the transgenic H chain 56R using specific primers for the IgM-56R transgene (supplemental figure 1A, data online at www.phys.huji.ac.il/∼eldadb/Supplement.pdf).

FIGURE 1.

B cells with L chain editors are sequestered in the MZ compartment of the spleen. A, The L chain repertoire in the 56R H chain-transgenic mouse depends on the strain background. 56R on a C57/BL/6 background (B6.56R) uses mainly the polyreactive (20) Vκ38C (black bars) and few non-anti-nuclear Ab (ANA) (20) Vκ21D (dark gray bars) L chain. 56R on a BALB/c background (BALB/c. 56R) uses mainly Vκ21D and less Vκ38C L chain. The percentage of the ANA (21) λ1 and λ2 L chain (light gray bars) is similar in both mouse strains. Because of the overlap between Vκ(S) with Vκ21D and Vκ38C, and because this representation contains B cells with dual L chain message, the bars shown here do not add up to 100%. Numbers in parentheses represent numbers of total B cells as determined by PCR product-positive cells. For the B6.56R mouse, only 184 cells are included in the representation of Vκ21D L chain. B, The strain-specific repertoire differences of B6.56R and BALB/c. 56R are most apparent in MZ B cells. The L chain editors Vκ21D and Vκ38C are increased in the MZ compartment of the spleen and are found at a smaller percentage in B cells of the follicular compartment. The white bars show L chain products from a PCR with the degenerate primer Vκ(S) which amplifies most Vκ genes including Vκ21D and Vκ38C. Numbers in parentheses represent numbers of total B cells. For the MZ/B6.56R group, only 90 cells are included in the representation of Vκ21D L chain. The graph shows all L chain PCR products prior to sequencing analysis. In-frame vs out-of-frame products are detailed in Fig. 3 D and out-of-frame products did not change the L chain repertoire shift of Vκ38C and Vκ21D in B6.56R and BALB/c.56R mice. C, A large percentage of B cells with the editor L chains Vκ21D or Vκ38C are associated with the IgM-56R transgene (56R+, black bars). In some edited B cells with Vκ21D or Vκ38C the IgM-56R transgene could not be detected (56R, gray bars). These 56R B cells could have been subject to VH replacement (16) or they could be switched to IgG. The H chain status of 56R B cells was not analyzed further. The sum of all other κ L chains positive with Vκ(S) but negative for Vκ21D and Vκ38C also showed association with the IgM-56R transgene. The bar graph includes B cells some of which have dual L chain rearrangements and for the MZ/B6.56R group the number of cells with Vκ38C is greater (179 cells), therefore the sum of all bars in each group may not add up to 100%.

FIGURE 1.

B cells with L chain editors are sequestered in the MZ compartment of the spleen. A, The L chain repertoire in the 56R H chain-transgenic mouse depends on the strain background. 56R on a C57/BL/6 background (B6.56R) uses mainly the polyreactive (20) Vκ38C (black bars) and few non-anti-nuclear Ab (ANA) (20) Vκ21D (dark gray bars) L chain. 56R on a BALB/c background (BALB/c. 56R) uses mainly Vκ21D and less Vκ38C L chain. The percentage of the ANA (21) λ1 and λ2 L chain (light gray bars) is similar in both mouse strains. Because of the overlap between Vκ(S) with Vκ21D and Vκ38C, and because this representation contains B cells with dual L chain message, the bars shown here do not add up to 100%. Numbers in parentheses represent numbers of total B cells as determined by PCR product-positive cells. For the B6.56R mouse, only 184 cells are included in the representation of Vκ21D L chain. B, The strain-specific repertoire differences of B6.56R and BALB/c. 56R are most apparent in MZ B cells. The L chain editors Vκ21D and Vκ38C are increased in the MZ compartment of the spleen and are found at a smaller percentage in B cells of the follicular compartment. The white bars show L chain products from a PCR with the degenerate primer Vκ(S) which amplifies most Vκ genes including Vκ21D and Vκ38C. Numbers in parentheses represent numbers of total B cells. For the MZ/B6.56R group, only 90 cells are included in the representation of Vκ21D L chain. The graph shows all L chain PCR products prior to sequencing analysis. In-frame vs out-of-frame products are detailed in Fig. 3 D and out-of-frame products did not change the L chain repertoire shift of Vκ38C and Vκ21D in B6.56R and BALB/c.56R mice. C, A large percentage of B cells with the editor L chains Vκ21D or Vκ38C are associated with the IgM-56R transgene (56R+, black bars). In some edited B cells with Vκ21D or Vκ38C the IgM-56R transgene could not be detected (56R, gray bars). These 56R B cells could have been subject to VH replacement (16) or they could be switched to IgG. The H chain status of 56R B cells was not analyzed further. The sum of all other κ L chains positive with Vκ(S) but negative for Vκ21D and Vκ38C also showed association with the IgM-56R transgene. The bar graph includes B cells some of which have dual L chain rearrangements and for the MZ/B6.56R group the number of cells with Vκ38C is greater (179 cells), therefore the sum of all bars in each group may not add up to 100%.

Close modal

Next, we analyzed the compartmental distribution of the Vκ21D and Vκ38C editor L chain. We found that in both strains, B6.56R and BALB/c.56R, the majority of B cells with Vκ38C or Vκ21D showed a MZ B cell phenotype (Fig. 1,B); 52.2% of B cells with Vκ38C L chain are likely to belong to the MZ compartment of B6.56R as compared with 26.6% that were found in the follicular compartment. Similarly, 77.3% of B cells with Vκ21D were sequestered in the MZ compartment in the BALB/c.56R as compared with 15.9% in the follicular compartment. Vκ21D in B6.56R and Vκ38C in BALB/c.56R, even though found less in these strains, were also slightly increased in the MZ relative to the follicular compartment. In the B6.56R, λ was found more often in the MZ than in the follicular compartment, whereas in the BALB/c.56R, the distribution of λ was the same (Fig. 1,B). These findings held true upon analysis of the fraction of IgM-56R+ B cells only, with the exception of a slight increase of Vκ38C cells in the MZ as compared with the follicular compartment of B6.56R that did not reach statistical significance (supplemental figure 1B, data online at www.phys.huji. ac.il/∼eldadb/Supplement.pdf). The data suggest that the peripheral BCR repertoire in the B6.56R is prone to (poly-) autoreactivity due to the high frequency of Vκ38C. This finding is in contrast to the BALB/c.56R where the high representation of the innocuous L chain Vκ21D may help to maintain a tolerant BCR repertoire. Sequestration of L chain-edited B cells in the MZ of both strains may be an important mechanism of tolerance. Yet, a considerable proportion of polyreactive B cells with Vκ38C locate to the follicle in B6.56R (Fig. 1 B) and can therefore be available for the production of isotype-switched autoantibodies. Plasma cell differentiation in accordance with the production of autoantibodies is well-documented in the B6.56R (25, 26, 32).

We analyzed the presence of the 56R-transgenic H chain in B cells from the MZ and follicular zone compartment of B6.56R and BALB/c.56R. A large percentage of B cells with the editor L chain Vκ21D or Vκ38C was associated with the 56R transgene (Fig. 1,C). Some B cells with Vκ21D or Vκ38C were 56R. This fraction of 56R B cells can still carry the transgene as the PCR assay for IgM-56R does not amplify the transgene in isotype-switched IgG B cells. Another possible explanation for 56R B cells is that the already rearranged VDJ 56R H chain transgene can be edited by replacement with an upstream VH gene. Such VH replacement is possible through a recombination signal embedded in VH genes (16, 37, 38). This gene replacement is comparable to heptamer-/nonamer-mediated rearrangement in L chain genes. Thus, 56R B cells can be cells that have been isotype switched to IgG or they may have been subject to VH replacement. In both cases, cells in the IgM-56R fraction are likely to be derived from transgenic 56R B cells. This notion is supported by the resemblance of the L chain repertoire in IgM-56R+ and IgM-56R B cells (data not shown). The H chain status of 56R B cells could not be analyzed further due to a limited amount of cDNA isolated from a single cell. However, the results obtained from B cells in which the transgene could be detected (supplemental figures 1–4, data online at www.phys.huji.ac.il/∼eldadb/Supplement.pdf) did not differ considerably from the results that included all cells (Figs. 1–4).

FIGURE 2.

Allelic inclusion at the κ locus depends on the self-reactivity of the κ L chain. A, B cells with κκ dual L chain are found in the MZ and follicular compartment of both mice, B6.56R and BALB/c. 56R. B cell with κκ dual L chain are decreased in the follicular compartment of BALB/c.56R. Numbers in parentheses represent numbers of total B cells analyzed. B, B cells with the polyreactive Vκ38C L chain are more likely to have a second κ L chain in-frame as compared with B cells with Vκ21D L chain. Numbers in parentheses represent numbers of total B cells with in-frame Vκ21D and Vκ38C. C, Frequency of B cells with message for dual κλ L chain. B cells with a dual κλ L chain are found in both mice B6.56R and BALB/c.56R and no difference was found between B cells from the MZ and follicular compartment. B cells with λ1 L chain have rearranged a κ L chain. The majority of λ products found in the 56R mouse are dual L chain B cells.

FIGURE 2.

Allelic inclusion at the κ locus depends on the self-reactivity of the κ L chain. A, B cells with κκ dual L chain are found in the MZ and follicular compartment of both mice, B6.56R and BALB/c. 56R. B cell with κκ dual L chain are decreased in the follicular compartment of BALB/c.56R. Numbers in parentheses represent numbers of total B cells analyzed. B, B cells with the polyreactive Vκ38C L chain are more likely to have a second κ L chain in-frame as compared with B cells with Vκ21D L chain. Numbers in parentheses represent numbers of total B cells with in-frame Vκ21D and Vκ38C. C, Frequency of B cells with message for dual κλ L chain. B cells with a dual κλ L chain are found in both mice B6.56R and BALB/c.56R and no difference was found between B cells from the MZ and follicular compartment. B cells with λ1 L chain have rearranged a κ L chain. The majority of λ products found in the 56R mouse are dual L chain B cells.

Close modal
FIGURE 3.

Experiment and computational model to explain the usage of Jκ genes in Vκ/Jκ gene rearrangements. A, Experimental results of Jκ usage in noneditor κ L chains, Vκ38C and Vκ21D L chain. Left group, Jκ usage in a group of 64 other κ (noneditor) L chains consisting of 29 different Vκ genes. All noneditor κ L chains were identified from dual κ L chain B cells. Middle group, Jκ usage in 80 Vκ38C L chains; 59% of the shown sequences were identified from dual κ L chain B cells. Right group, Jκ usage in 106 Vκ21D L chains; 26% of the shown sequences were identified from dual κ L chain B cells. In-frame products are shown in black, out-of-frame products are shown in gray. B, Computation of a random distribution of Jκ genes under certain assumptions to best fit the experimental data in A for noneditor κ L chains. These are the probability for an in-frame product = one-third, the probability for another attempt being made after an out-of-frame product has been produced as α = 0.5, a factor to describe the instability of an out-of-frame product as β = 0.25. C, Computation of a strictly sequential distribution under the same assumptions as in B. Computational models were inspired by previous work by Louzoun et al. (35 ) and modified to best fit our experimental data. Detailed formulas are available at www.phys.huji.ac.il/∼eldadb/Supplement.pdf. D, Experimental data for the frequency of κ and λ L chain in frame vs out of frame.

FIGURE 3.

Experiment and computational model to explain the usage of Jκ genes in Vκ/Jκ gene rearrangements. A, Experimental results of Jκ usage in noneditor κ L chains, Vκ38C and Vκ21D L chain. Left group, Jκ usage in a group of 64 other κ (noneditor) L chains consisting of 29 different Vκ genes. All noneditor κ L chains were identified from dual κ L chain B cells. Middle group, Jκ usage in 80 Vκ38C L chains; 59% of the shown sequences were identified from dual κ L chain B cells. Right group, Jκ usage in 106 Vκ21D L chains; 26% of the shown sequences were identified from dual κ L chain B cells. In-frame products are shown in black, out-of-frame products are shown in gray. B, Computation of a random distribution of Jκ genes under certain assumptions to best fit the experimental data in A for noneditor κ L chains. These are the probability for an in-frame product = one-third, the probability for another attempt being made after an out-of-frame product has been produced as α = 0.5, a factor to describe the instability of an out-of-frame product as β = 0.25. C, Computation of a strictly sequential distribution under the same assumptions as in B. Computational models were inspired by previous work by Louzoun et al. (35 ) and modified to best fit our experimental data. Detailed formulas are available at www.phys.huji.ac.il/∼eldadb/Supplement.pdf. D, Experimental data for the frequency of κ and λ L chain in frame vs out of frame.

Close modal
FIGURE 4.

Downstream Jκ genes in B cells with dual L chain message. B cells with only one Vκ message use mainly Jκ2 whereas allelically κκ- and isotypically κλ-included B cells use more often Jκ4 and Jκ5. Three groups are shown, excluded B cells in which just one L chain could be identified, included B cells with dual κκ L chain and included B cells with dual κλ or κκλ L chain. Note that ∼89.6% the excluded cells with single Vκ gene message use Vκ21D in which Jκ2 is the predominant. In this representation, only B cells are included of which all κ L chain products have been sequenced.

FIGURE 4.

Downstream Jκ genes in B cells with dual L chain message. B cells with only one Vκ message use mainly Jκ2 whereas allelically κκ- and isotypically κλ-included B cells use more often Jκ4 and Jκ5. Three groups are shown, excluded B cells in which just one L chain could be identified, included B cells with dual κκ L chain and included B cells with dual κλ or κκλ L chain. Note that ∼89.6% the excluded cells with single Vκ gene message use Vκ21D in which Jκ2 is the predominant. In this representation, only B cells are included of which all κ L chain products have been sequenced.

Close modal

56R+ B cells using other κ L chains were sequestered in the MZ compartment in B6.56R, yet the majority of 56R+ B cells using other κ L chains were found in the follicular compartment in BALB/c.56R (Fig. 1 C). We conclude that in the B6.56R, sequestration in the MZ B cell compartment leads to a potentially dangerous situation, because MZ B cells are easily activated and release autoantibodies (20, 25, 26, 32). In contrast, in the BALB/c.56R, potentially autoreactive B cells make it into the follicular compartment, yet they do not differentiate into plasma cells. This finding indicates that postfollicular checkpoints in the BALB/c.56R mouse are intact and sequestration in the follicular B cell compartment does not allow plasma cell differentiation. This has been well-established by the lack of serum autoantibodies in the BALB/c.56R strain (25, 26, 32).

To identify B cells with dual L chain message, we further analyzed B cells showing two κ L chain PCR products when amplifying Vκ21D, Vκ38C, or κ L chain using the degenerate primer Vκ(S). All PCR products from these dual κ/κ L chain B cells were sequenced using both nested primers from the 5′ and from the 3′ end of the L chain transcript. Sense and antisense strands were aligned and the type of κ L chain was identified. We screened 364 cells which included 56R+ and 56R B cells from B6.56R and BALB/c.56R mice (Table Iand Fig. 2,A). Among these 356 cells, the vast majority (97.8%) had a κ L chain product (Table I). Two different in-frame κκ L chain messages were found in 48 cells (12.9%) (Table I). Two κ PCR products were found in a larger fraction of cells, yet, in most cases, the two κ PCR products were identical, in that Vκ21D L chain was often amplified using both the specific primer for Vκ21D and the degenerate primer Vκ(S). These cells, therefore, yielded B cells with only one κ L chain (Table I). The frequency of B cells with in-frame dual κκ L chain was higher in B6.56R than in BALB/c.56R among B cells from the follicular zone (p = 0.021) but not from the MZ compartment. A significant increase of B cells with in-frame dual κκ L chain was also found in B6.56R as compared with BALB/c.56R among B cells from the follicular zone (p = 2.3 × 10−3) when only IgM-56R+ B cells were analyzed (supplemental figure 2A, data online at www.phys.huji.ac.il/∼eldadb/Supplement.pdf). In the B6.56R, in-frame κκ B cells were distributed equally in the MZ (15.5%) and the follicular compartment (16%). In contrast, in the BALB/c.56R, we found a significantly (p = 0.036) higher percentage of in-frame dual κκ B cells in the MZ (14.7%) as compared with the follicular compartment (5.4%) (Fig. 2 A). This increase also held true and was statistically significant (p = 0.04) when only IgM-56R+ B cells were analyzed (supplemental figure 2A, data online at www.phys.huji.ac.il/∼eldadb/Supplement.pdf).

Table I.

Single B cells with message for dual κκ L chaina

Cells fromCells with κ PCR Products from Vκ21D/Vκ38Cor Vκ(S)Cells withκκ DualPCR ProductCells with κ/κ Dual In Frame (% of Total Cells inEach Group)Vκ38C/Vκ21D Dual In Frame (Numbers in Parentheses Depict In-Frame Vκ38C and Vκ21D)κκ Dual In Frame with λ In Frame (% of Total Cells)
MZ/B6.56R (90) 85 34 14 (15.5%) 4 (38C n = 28, 21D n = 15) 
 56R+ (67) 62 32 14 (20.9%) 
 56R (23) 23 0 (0%) 
Fo/B6.56R (94) 93 28 15 (16.0%) 4 (38C n = 25, 21D n = 6) 
 56R+ (35) 35 17 9b (25.7%) 
 56R (59) 58 11 7c (11.8%) 
MZ/BALB/c.56R (88) 88 67 13 (14.7%) 6 (38C n = 12; 21D n = 65) 
 56R+ (52) 52 41 8 (15.3%) 
 56R (36) 36 26 5 (13.8%) 
Fo/BALB/c.56R (92) 90 15 5 (5.4%) 1 (38C n = 4; 21D n = 14) 
 56R+ (62) 59 10 2 (3.2%) 
 56R (30) 31 3 (3.3%) 
Total cells analyzed from B6.56R and BALB/c.56R (364) 356 (97.8%) 144 (39.5%) 48 (12.9%) 15 (4.1%) 2 (0.5%) 
Cells fromCells with κ PCR Products from Vκ21D/Vκ38Cor Vκ(S)Cells withκκ DualPCR ProductCells with κ/κ Dual In Frame (% of Total Cells inEach Group)Vκ38C/Vκ21D Dual In Frame (Numbers in Parentheses Depict In-Frame Vκ38C and Vκ21D)κκ Dual In Frame with λ In Frame (% of Total Cells)
MZ/B6.56R (90) 85 34 14 (15.5%) 4 (38C n = 28, 21D n = 15) 
 56R+ (67) 62 32 14 (20.9%) 
 56R (23) 23 0 (0%) 
Fo/B6.56R (94) 93 28 15 (16.0%) 4 (38C n = 25, 21D n = 6) 
 56R+ (35) 35 17 9b (25.7%) 
 56R (59) 58 11 7c (11.8%) 
MZ/BALB/c.56R (88) 88 67 13 (14.7%) 6 (38C n = 12; 21D n = 65) 
 56R+ (52) 52 41 8 (15.3%) 
 56R (36) 36 26 5 (13.8%) 
Fo/BALB/c.56R (92) 90 15 5 (5.4%) 1 (38C n = 4; 21D n = 14) 
 56R+ (62) 59 10 2 (3.2%) 
 56R (30) 31 3 (3.3%) 
Total cells analyzed from B6.56R and BALB/c.56R (364) 356 (97.8%) 144 (39.5%) 48 (12.9%) 15 (4.1%) 2 (0.5%) 
a

The percentage of κκ dual L chain message-included B cells is higher within the transgenic B cells carrying the 56R-IgM transgene (56R+) than in B cells in which the 56R transgene could not be detected (56R). This increase was observed in B6.56R but not in BALB/c.56R. In the B6.56R, some B cells have, in addition to a dual κ message, a λ L chain dual L chain B cells (numbers in parentheses represent numbers of total B cells analyzed.)

A total of bthree, and cfour, dual L chain B cells have Vκ12-40 (pseudogene) as a second rearrangement.

The follicular B cell compartment in B6.56R contained a considerable number of κ L chain that had highest similarity to Vκ12-40, which has been suggested to be a pseudogene (Table I, Fig. 2 A) (accession number AJ235952). However, in this group of seven Vκ12-40 transcripts, all κ L chains were in frame, and we did not identify a stop codon. Therefore, we conclude that this Vκ12-40 was expressed and we have no evidence that this L chain is nonfunctional. Among IgM-56R+ B cells (supplemental figure 2A, data online at www.phys. huji.ac.il/∼eldadb/Supplement.pdf), the difference between BALB/c and B6 follicular zone B cells in the amount of allelic inclusion remained significant (p = 0.03, not depicted) when the Vκ12-40 L chain was not considered.

One way by which editing functions is by inclusion of L chain genes. It has been proposed that tolerance is accomplished by “diluting” the autoreactivity of one receptor by coexpressing a second receptor (12, 13). We were interested in testing whether polyreactivity as conferred by Vκ38C L chain leads to a breakdown of allelic exclusion in κ L chain genes. We focused our attention to B cells with in-frame Vκ38C L chain and analyzed the frequency of dual κκ L chain B cells in this group. We compared the results for dual κκ L chain B cells with Vκ38C to dual κκ L chain B cells with Vκ21D (Fig. 2 B).

In the 364 analyzed B cells, which included 56R+ and 56R B cells, we found a significant difference in the frequency of allelic inclusion in B cells with Vκ38C as compared with B cells with Vκ21D (Fig. 2,B). In all analyzed cells, 69 B cells had the Vκ38C L chain and, among these, 41 (59%) had a second in-frame κ L chain rearrangement. In contrast, 100 B cells had a Vκ21D L chain, but only 26 (26%) had a second in-frame L chain rearrangement (Fig. 2,B). The same finding held true and reached statistical significance (p = 9 × 10−4) for the fraction of 56R+ B cells only (supplemental figure 2B, data online at www.phys.huji.ac.il/∼ eldadb/Supplement.pdf). The strain-specific differences were small but there was a slightly higher percentage of dual Vκ38C B cells in BALB/c.56R (69%) as compared with B6.56R (57%) and this difference was amplified in B cells from the MZ compartment (p = 0.124; not significant) (Fig. 2,B). Interestingly, we found a large group of dual κκ L chain B cells that had rearranged both editor L chains Vκ38C and Vκ21D (Table I). Of the dual κκ L chain B cells examined, Vκ38C together with Vκ21D was found in 15 of 48 dual κκ L chain B cells from both mice (31.1%) (Table I). The percentage was slightly higher in 56R BALB/c (38.9%) than in B6.56R (27.5%). Additionally, we found strain-specific differences in the type of κ L chain that was coexpressed with Vκ38C in dual L chain B cells. In the B6.56R mouse, B cells with Vκ38C mainly coexpressed a noneditor κ L chain: 22 of 30 (73.3%) dual Vκ38C cells expressed a noneditor κ L chain and only 8 of 30 (26.6%) coexpressed Vκ21D (data not depicted). In the BALB/c.56R, only 4 of 11 (36.3%) B cells with Vκ38C coexpressed a noneditor L chain and 7 of 11 (63.6%) B cells coexpressed Vκ21D L chain (data not depicted). In sum, we found that B cells with BALB/c.56R rarely use the polyreactive incomplete editor Vκ38C, and if Vκ38C is found, it is allelically included with the complete editor L chain Vκ21D. In contrast, B6.56R has a more polyreactive Vκ38C L chain but has limited ways to neutralize them because of the low frequency of Vκ21D L chain. We interpret our findings such that efficient neutralization of polyreactive Vκ38C BCRs by effective allelic inclusion with Vκ21D L chain might be in part responsible for the healthy state of BALB/c.56R that is characterized by only few autoantibodies as compared with the B6.56R (25, 26, 32). In contrast, in the B6.56R, higher titers of serum autoantibody might be the result of limited L chain allelic inclusion of incompletely edited receptors with Vκ38C.

L chain isotypic inclusion has been described for λ L chain in the 56R mouse (17, 20, 22, 23); however, the detailed analysis of these cells and their compartmental distribution is missing. To determine the scope of isotypic inclusion in B cells with λ L chain, we analyzed cells with λ L chain and determined the frequency and type of κ L chains that were rearranged in the same cell. In the pool of 46 cells with λ L chain from B6.56R and BALB/c.56R, 43 λ L chains were accompanied by a κ L chain (Table II). Sequencing analysis of all λ L chain identified 11 in-frame λ L chain products. In the group of 11 λ L chains, we could identify eight cells that had an in-frame κ product. The coexpressed κ L chain often was the “editor” Vκ38C or Vκ21D (data not shown). A large percentage of λ L chains was out of frame (Fig. 3,D) and also these B cells had a κ L chain. We found an equal distribution of κλ dual L chain B cells in the tested compartments of B6.56R and BALB/c.56R and counted the same number of in-frame κλ dual B cells in the 56R+ and 56R fraction of cells (Table II). Hence, B cells with λ1 L chain represent a population in which isotypic exclusion is broken. These κλ dual L chain B cells were also found in the follicular compartment (Fig. 2 C) and, as we have shown before, become activated and secrete autoantibodies in cGvH-induced autoimmunity (20).

Table II.

Single B cells with message for dual λκ L chaina

Cells with λ PCR Products from Vλ1/2/Cλ PCRCells with κλ or κκλ Dual PCR ProductsCells with λ In FrameCells with κλ Dual In Frame
MZ/B6.56R (179) 24 22 3 (1.7%) 
 56R+ cells (124) 16 15 
 56R cells (55) 
Fo/B6.56R (94) 2 (2.1%) 
 56R+ cells (35) 
 56R cells (61) 
MZ/BALB/c.56R (88) 2 (2.3%) 
 56R+ cells (52) 
 56R cells (36) 
Fo/BALB/c.56R (92) 11 10 1–3 (1.1–3.3%) 
 56R+ cells (62) 
 56R cells (30) 1 in frame (3.3%) and 2 nd 
Total cells analyzed from B6.56R and BALB/c.56R (453) 46 (10.1%) 43 (9.5%) 11 (2.4%) 8 (1.8%) 
Cells with λ PCR Products from Vλ1/2/Cλ PCRCells with κλ or κκλ Dual PCR ProductsCells with λ In FrameCells with κλ Dual In Frame
MZ/B6.56R (179) 24 22 3 (1.7%) 
 56R+ cells (124) 16 15 
 56R cells (55) 
Fo/B6.56R (94) 2 (2.1%) 
 56R+ cells (35) 
 56R cells (61) 
MZ/BALB/c.56R (88) 2 (2.3%) 
 56R+ cells (52) 
 56R cells (36) 
Fo/BALB/c.56R (92) 11 10 1–3 (1.1–3.3%) 
 56R+ cells (62) 
 56R cells (30) 1 in frame (3.3%) and 2 nd 
Total cells analyzed from B6.56R and BALB/c.56R (453) 46 (10.1%) 43 (9.5%) 11 (2.4%) 8 (1.8%) 
a

B cells with dual λκ L chain are found in both mice B6.56R and BALB/c.56R and no difference was found between the compartments MZ and follicular B cells. B cells with in-frame λ 1 message have in most cases rearranged a κ L chain dual L chain B cells. The majority of λ products found in the 56R mouse are out of frame. B cells with λ have also rearranged a κ L chain (numbers in parentheses represent numbers of total B cells analyzed); nd, not determined.

Several works have shown that Jκ genes 3′ of the Jκ cluster are more likely to associate with the Vκ gene during L chain rearrangement (39) (M. Schlissel, personal communication). A computational model for such a sequential order of Jκ genes has been described before and is shown in this work by introducing parameters that consider in-frame and out-of-frame κ L chain products to be formed (Fig. 3,C). It has been described that L chain receptor editing leads to a bias toward downstream Jκ genes (40, 41). Thus, we were interested in whether incompletely edited B cells and B cells with dual L chain receptor were especially subject to receptor editing. We analyzed the Jκ genes in noneditor L chains, the Vκ38C L chain, and the Vκ21D L chain (Fig. 3,A and supplemental table I, B–E). In the group of noneditor L chains consisting of 29 different Vκ genes (supplemental table ID), we found an increase in the frequency from Jκ1, Jκ2, Jκ4 to Jκ5 (Fig. 3,A, left). Note that in this group, all noneditor L chains were identified from dual κκ L chain B cells. In the Vκ38C L chain, we also found a bias to downstream Jκ genes (Fig. 3,A, middle). The majority of B cells with Vκ38C L chain (59%) expressed an additional κ L chain. In contrast, Vκ21D L chain almost exclusively used Jκ2, very few Jκ1 and Jκ4, and no Jκ5 (Fig. 3,A, right). Twenty-six percent of the B cells with Vκ21D L chain expressed an additional κ L chain. The usage of upstream Jκ1/2 vs downstream Jκ4/5 in B cells with Vκ38C is significantly different from the usage in B cells with Vκ21D L chain which means that B cells with Vκ38C exhibit more distal Jκ4/5 usage, indicative of increased receptor editing. The results held true when only IgM-56R+ B cells were analyzed (supplemental figure 3A, data online at www.phys.huji.ac.il/ ∼eldadb/Supplement.pdf). We conclude that B cells with autoreactive properties such as nonedited B cells and incompletely edited B cells with Vκ38C L chain are forced to undergo more rearrangement cycles than B cells that use the complete editor L chain Vκ21D. Another possibility is that in B cells with dual κ L chain message, κ L chain is rearranged at random as represented in the computational model (Fig. 3 B).

To exclude the possibility that certain Vκ genes rearrange preferentially to a Jκ, we looked at the distribution of Jκ in the diverse group of noneditor κ L chains. The κ L chains were classified into groups of related V genes (supplemental table IE). We found an increased usage of Jκ4 in Vκ12 (7 Vκ12–40/Jκ4, 2 Vκ12–46/Jκ4, and 2 Vκ12–46/Jκ2) and a bias to Jκ4 and Jκ5 in Vκ4 (2 Vκaj4/Jκ5, 1 Vκad4/Jκ4, 1 Vκah4/Jκ5, 1 Vκkb4/Jκ4, 5 Vκkk4/Jκ5, 1 Vκkk4/Jκ4, 1 Vκkh4/Jκ4, 2 Vκkj4/Jκ5, and 1 Vκ4–57/Jκ5), yet, in all other κ L chains, Jκ usage was mixed or very few examples were found for each Vκ gene. Hence, the data do not allow for the general conclusion that certain Vκ genes preferentially rearrange to a particular Jκ gene. Instead, because of the increased usage of Jκ2, Jκ4, Jκ4, and Jκ5 genes in Vκ21D, Vκ38C, Vκ12, and Vκ4 genes, respectively, we think that other processes such as selection influence the Jκ usage of a particular Vκ gene.

To test whether B cells with dual L chains behave like B cells with only one L chain with respect to Jκ gene usage, we analyzed the usage of Jκ genes in B cells that had a single in-frame L chain (cells with a single L chain are referred to as κ L chain-excluded cells) and compared these to the Jκ usage in B cells with dual in-frame κκ or λκ L chain (cells with dual L chain are referred to as κκ or λκ L chain-included cells). We found a strong bias to Jκ2 usage in L chain-excluded B cells and little usage of Jκ1, Jκ4, and Jκ5. The distribution of Jκ in L chain-excluded B cells strongly resembled the distribution of Jκ genes in cells with Vκ21D (Fig. 3,A), and, indeed, 89.6% of κ L chain-excluded cells had rearranged Vκ21D (see supplemental table IB). In contrast, the usage of Jκ genes in L chain-included B cells showed less Jκ2 and more Jκ4 and Jκ5 compared with L chain-excluded cells when all cells (IgM-56R and IgM-56R+) were analyzed (Fig. 4) and the same pattern was confirmed when only IgM-56R+ B cells were analyzed (supplemental figure 4, data online at www.phys.huji.ac.il/∼eldadb/Supplement.pdf). The distribution of Jκ genes in L chain-included cells did not resemble the distribution of Jκ genes in noneditor L chains from κ L chain-included B cells (Fig. 3 A). Further analysis showed that L chain-included B cells are “hybrid cells” in that they carry one noneditor L chain and one editor L chain: of 56 L chain-included B cells, 25 cells (44.6%) had one Vκ21D rearrangement that was mostly rearranged to Jκ2, and 39 cells (69.6%) used Vκ38C (data not depicted). In summary, we find that most L chain-excluded B cells in the 56R-transgenic mouse use the complete editor L chain Vκ21D. In L chain-included B cells, the incomplete editor L chain Vκ38C or a noneditor L chain is often accompanied by Vκ21D.

Receptor editing is a mechanism by which autoreactive receptors are replaced by nonautoreactive receptors. An example of an efficiently edited autoreactive receptor is the 56R H chain transgene paired with Vκ21D L chain. Vκ21D vetoes binding to dsDNA and other self-Ags (20). Yet, receptor editing has a dangerous side effect because it can be incomplete and generate polyreactive, anti-self Abs (20). An example of an incompletely edited receptor is 56R paired with Vκ38C L chain. This combination has high relative affinity to self-proteins as well as to dsDNA (20).

Incomplete editors such as Vκ38C raise the question of whether it is appropriate to call Vκ38C an “editor” L chain. However, the tolerance threshold for incomplete editor L chains is defined by genetic predisposition as seen by the L chain repertoire shift in BALB/c.56R and B6.56R and the “lupus-resistant” strain background, BALB/c, is stringent whereas the more “lupus-susceptible” strain background, B6, is less efficient (27, 28, 29, 30, 31). In BALB/c.56R, lupus resistance concurs with a high percentage of B cells with the complete editor L chain Vκ21D. In contrast, in the lupus-susceptible strain, B6.56R, a high percentage of B cells with the incomplete editor Vκ38C L chain is found (Fig. 1). The shift in κ L chain usage is not a natural property of the B6 and BALB/c mouse but is caused by the 56R H chain transgene and the genetic differences between the strains B6 and BALB/c subsequently lead to an L chain repertoire shift. Analysis of Vκ38C in B6 and Vκ21D L chain in BALB/c nontransgenic mice showed no overrepresentation of these L chains as compared with other Vκ genes. In the nontransgenic B6, Vκ38C is found in 0.7% of peripheral B cells (26), and in the BALB/c mouse, the whole of the Vκ21 family (eight members) is represented with no more than 4% among splenic B cells (36). Further evidence for the genetic influence on the L chain repertoire comes from B6.56R mice which lack the Fcγr2b. In this mouse, the peripheral L chain repertoire is further shifted toward the usage of polyreactive Vκ38C L chain as compared with the B6.56R mouse. Interestingly, B6.56R/Fcγr2b−/− mice show IgG and complement C3 deposition in the kidney glomeruli and succumb to autoimmune disease (26). Increased numbers of polyreactive B cells have also been observed in human lupus erythematosus as compared with healthy individuals. It has been shown that checkpoints to eliminate polyreactive B cells upon emigration from the bone marrow and maturation of B cells in the periphery are defect (8, 9). Even though coincidence of polyreactive B cells with autoimmune diseases is shown, proof is missing for a direct involvement of polyreactive B cells in autoimmunity or in disease onset. Some work even suggests the contrary in that polyreactive B cells prevent the onset of experimental autoimmune disease (42). Polyreactive or natural (auto) Abs are considered to have beneficial functions and are important in the first line of defense against pathogens through the activation of immune complement and the formation of immune complexes which are delivered to follicular dendritic cells (reviewed in Ref. 43). Natural autoantibodies are also involved in the clearance of apoptotic cells, thereby limiting the circulation of autoantigens and reducing the risk of autoimmunity (44). Thus, the concurrence of polyreactive autoantibodies and autoimmunity does not show polyreactivity to be causal for the onset of disease. Yet, deregulation and overexpression of polyreactive anti-DNA autoantibodies may bear dangerous features and promote autoimmune phenotypes.

A way of creating tolerance is the coexpression of two L chains (11, 13), and we show here that at least one of these L chains is an editor κ L chain. Coexpression of two L chains leads to a partially edited B cell. Incompletely edited receptors can be further regulated by allelic inclusion and, here again, the genetic background of the mouse is important: in the BALB/c.56R mouse, the presence of Vκ21D L chain may facilitate the silencing of autoreactive/polyreactive receptors. In contrast, in the B6.56R mouse, the main editor L chain Vκ38C has polyreactive/self-reactive properties itself.

How do these partially edited (or included) B cells come about? Do both κ alleles rearrange simultaneously or do they rearrange sequentially? BrdU labeling experiments showed that allelically included B cells take longer to develop and appear later in the immature B cell compartment than allelically excluded B cells (11). The time the first dual L chain B cell appears in the immature B cell pool takes twice as long as a B cell needs to edit a single receptor (45, 46, 47). This favors subsequent rearrangement of κ alleles and suggests that dual receptor B cells arise by receptor editing. Simultaneous rearrangement is unlikely. It has been shown that locus activation by Ig enhancers is quite low (around 10%), and therefore the probability that a cell has two active loci is 1%. Only one-third of these rearrangements are in frame, which results in a frequency of cells with two in-frame gene rearrangements of 0.3% (48). Therefore, it is unlikely that dual L chain receptor B cells occur simply by accident. In our study, the frequency of κκ dual L chain receptor B cells in the 56R mouse is 12% (Table I).

We find that the generation of dual L chain receptor B cells is directed in that it occurs primarily when receptor editing is incomplete as is the case in 56R/Vκ38C B cells. We hypothesize that receptor inclusion may be a self-Ag-driven tolerance mechanism because B cells with Vκ38C L chain often coexpress a second κ L chain, and two explanatory models are available for the generation of dual L chain B cells (1). The B cell could have first rearranged a noneditor L chain in an attempt to silence the 56R H chain thereby using up all available Jκ genes; to escape deletion the B cell rearranges an editor κ L chain on the other allele resulting in the generation of a second receptor (2). Alternatively, the B cell first rearranged an editor κ L chain, Vκ21D or Vκ38C, to silence the 56R H chain. This can result in the successful neutralization of 56R as is the case in Vκ21D B cells, yet this first editing attempt can also be incomplete and result in the incompletely edited polyreactive receptor 56R/Vκ38C. In case of incomplete editing, rearrangement of a second L chain rescues the cell from clonal deletion. We favor the latter. First, the frequency of dual L chain B cells is higher in B cells with Vκ38C than in B cells with Vκ21D, and we interpret this to be the result of autoreactivity of B cells with Vκ38C (Fig. 2,B). Second, B cells with Vκ38C are biased to downstream Jκ genes such as if Vκ38C B cells already muddled through several attempts in search of a complete editor L chain (Fig. 3 A). The bias to downstream Jκ genes has been shown to be a property of receptor editing (40, 41). Caution is warranted in interpreting the finding that polyreactivity infers κ L chain inclusion in the cell, because the degenerate primer Vκ(S) might not amplify all κ L chains amplified equally well (supplemental table IA). However, a competition effect would skew the relative frequencies uniformly across all compartments in both mice, B6.56R and BALB/c.56R. Consequently, differences observed between different compartments and between both mice would be retained even after compensating for this skewing. Therefore, we argue that κ L chain inclusion is truly enhanced in incompletely edited B cells.

Another way of creating dual L chain receptor B cells is by isotype inclusion as recognized by the coexpression of κ and λ L chain on the cell surface. It has been shown that 56R with Vλ1 has high relative affinity for dsDNA (21), and the self-reactivity of this combination was thought to lead to expression of a κ L chain (22, 23). In the 56R mouse, such κλ dual receptors are expressed on the cell surface (17, 20, 22, 23), and hybridoma studies showed that Vκ38C was a common κ L chain accompanying Vλ1 (17). In light of our new findings, we think that Vλ1 rearrangement has taken place after rearrangement of the incomplete editor Vκ38C. In support of this idea, the rearrangement of λ L chain is the ultimate bias in strongly edited B cells (4, 49, 50). Even though λ L chain lacks the ability of secondary rearrangement due to the lack of rearrangement signals, receptor editing may be acting in parallel on the κ locus and could explain the abundant expression of “editor” L chains such as Vκ21D and Vκ38C with Vλ1.

Coexpressed, autoreactive, and a nonautoreactive receptor both deliver signals into the cell, one Ag-induced and one tonic signal (31) and these B cells undergo differentiation as well as receptor editing (13). This also true for κλ dual receptor B cells in the 56R mouse, except that probably both receptors deliver Ag-induced signals. The existence of B cells with λ1 and Vκ38C therefore demonstrates that B cells with two self-reactive receptors can become part of the splenic B cell compartment.

Our work suggests that partially edited B cells “dilute” anti-self-specificity of one receptor by coexpressing a second κ L chain, for example, if a Vκ38C B cell rearranges Vκ21D as the second L chain. Expression of two κ L chains can be dangerous for the immune system and help self-reactive B cells to bypass central tolerance mechanisms. As discussed above, dual receptor B cells can have two self-reactive receptors. In this case, the differences in the specificities of the two receptors may lead to sufficiently weak avidity for a given (self-) Ag so that the B cell can leave the bone marrow. Each receptor contributes to selection and reduces the selective pressure as is the case if the B cell had just one receptor. Beneficial effects of dual receptors have been shown in T cells in which dual α-chain expression extends the available repertoire of TCR specific for foreign Ags (51), a process suggested to contribute to allorecognition (52). The second TCR rescues the cell from death by neglect (53).

Self-reactive B cells that reach the spleen can be rendered harmless for the immune system in other ways: unresponsiveness to autoantigens by ignorance and anergy (1, 6) have been well-documented mechanisms and prevent autoimmunity in newly generated B cells (1, 6). Another path to tolerance is peripheral sequestration of potentially pathogenic autoreactive B cells into immunological niches such as the MZ. Weakly self-reactive B cells are found in the MZ and sequestration in the MZ was suggested as a mechanism of peripheral tolerance (54, 55, 56). We demonstrate here that incompletely edited B cells are also regulated by sequestration into the MZ (Fig. 1 B).

In the B6.56R, the majority of B cells with Vκ38C was found in the MZ (Fig. 1, B and C) and serum autoantibody titers to dsDNA and histone in the B6.56R (32) may be due to the location and easy activation of MZ B cells. Yet, in our study, we observe that a fraction of 56R/Vκ38C B cells also locate to the follicular B cell compartment (Fig. 1, B and C). Even though a direct transition to the follicle was previously suggested for MZ B cells upon activation (57), proof of this hypothesis is still missing. The location of 56R B cells in the follicular zone compartment could explain the ease with which 56R B cells differentiate into plasma cells and produce autoantibodies of the IgM and the IgG isotype (25, 26, 32). In sum, susceptibility to develop an autoimmune phenotype in the B6.56R is accompanied by at least three findings: 1) inefficient ability to control the expression of Vκ38C with the resulting poor ability to silence unedited or incompletely edited receptors by coexpression of the complete editor Vκ21D (Fig. 1 and Table I); 2) the location of 56R/Vκ38C B cells to the follicular B cell compartment (Fig. 1, B and C); and 3) plasma cell differentiation with the production of autoantibodies (25, 26, 32).

In the BALB/c.56R mouse, we find that B cells with Vκ21D reside in the MZ B cell compartment (Fig. 1,B). This could be due to the following reasons: a large percentage of MZ B cells in the 56R mouse are not self-reactive. The MZ has been shown to serve as a bottleneck container in many transgenic mouse models in which B cell development is impaired due to a transgenic receptor (as reviewed in Ref. 58). In these transgenic models, the enlargement of the MZ compartment is not related to self-specificity of the BCR transgene. An example is the anti-HEL H chain-transgenic mouse (6). Another possibility is that 56R B cells with Vκ21D are weakly self-reactive to an unknown autoantigens. If 56R/Vκ21D has weak self-reactive features, the distribution of BCRs in the BALB/c.56R could also be explained by the signal-strength-hypothesis. According to this theory, B cells with weak self-reactive properties are selected into the MZ, whereas B cells with strong self-reactivity are selected into the follicle (6, 59, 60). We indeed see noneditor κ L chain in the follicular compartment of BALB/c.56R mice (Fig. 1 C). Yet, the lack of autoantibodies in the serum of BALB/c.56R mice suggests that postfollicular tolerance checkpoints are intact and do not allow autoreactive B cells to differentiate into Ab-producing plasma cells (61, 62).

Another way to bypass central tolerance in the bone marrow is intracellular retention of the incompletely edited receptor 56R/Vκ38C. We found recently that 56R/Vκ38C bound to an intracellular Ag in the Golgi and therefore surface expression was delayed. Yet, 56R/Vκ38C was eventually expressed on the cell surface. B cells with 56R/Vκ38C were sequestered in the MZ (S. N. Khan, E. J. Witsch, N. G. Goodman, A. K. Panigrahi, C. Chen, Y. Yang, A. M. Cline, J. Erikson, M. Weigert, E. T. L. Prak, and M. Radic, submitted for publication). Also other work has shown that B cells reactive to an ubiquitously present self-Ag hide their autoreactivity by intracellular retention of the autoreactive receptor. Upon emigration from the bone marrow these receptors are invisible to the immune system. The autoreactive receptor is internalized but seems to be ready to go as these mice show autoantibody titers with increased age (13).

A large fraction of newly rearranged Ig genes are nonfunctional and the necessity to shape the peripheral B cell repertoire by receptor editing is obvious, assuming that two-thirds of the rearrangements are out of frame and a considerable number of pseudogenes lead to nonfunctional L chain rearrangements. In the 56R mouse, the anti-DNA H chain transgene adds to the low frequency of functional rearrangements, and it has been estimated that at least 50% of the B cells are subject to L chain receptor editing (63, 64). Usage of downstream Jκ genes is thought to be a general sign of receptor editing (40, 41).

Some studies addressed the question of whether DNA rearrangement occurs with equal frequency at each of the Jκ genes and have shown that the proximal Jκ gene 5′ of the Jκ cluster is more likely to rearrange than the 3′ Jκ gene. In B cells that were unperturbed by antigenic selection, the usage of Jκ genes followed a sequential order, starting with Jκ1. The data sets analyzed in this work, however, were rather small (39). Confirmation for a sequential usage of Jκ genes comes from experiments in which rearrangement was induced in a bulk of B cells that are germline at both κ alleles. This resulted in a striking bias of dsDNA recombination signal sequence breaks at Jκ1 (M. Schlissel, personal communication). In none of these experiments did pre-existing autoreactivity or Ag specificity influence the usage of Jκ. A bias to upstream Jκ could also be observed in the 56R mouse. This upstream bias was found in hybridoma panels from BALB/c.56R mice. Yet, most of the hybridomas used the editor κ L chain Vκ21D which almost always rearranged to Jκ2 (N. Prak, unpublished observation). To explain the distribution of Jκ genes in the BCR repertoire in BALB/c.56R, a mathematical model was suggested previously (35). According to this model, Vκ and Jκ genes are rearranged in a “semisequential” order, and by this model the strong bias of Jκ2 in the L chain repertoire of the BALB/c.56R was explained. Generally, in a sequential process downstream Jκ genes are less likely to participate in a rearrangement because they are excluded once a Jκ gene upstream has rearranged in frame (Fig. 3,C). Yet, as a consequence of receptor editing the bias to upstream Jκ genes can be reversed and shifted to downstream Jκ genes because downstream Jκ genes are still available during secondary rearrangement. In this study, we find a shift to downstream Jκ4 and Jκ5 in incompletely edited B cells with Vκ38C as well as in B cells with noneditor κ L chains. We conclude that B cells with Vκ38C and with noneditor κ L chains are preferentially subject to L chain receptor editing (Fig. 3 A, left and middle group). The same results were obtained when only 56R+ B cells were analyzed (supplemental figure 3A, data online at www.phys.huji.ac.il/∼eldadb/Supplement.pdf).

Yet, we would like to suggest another explanation for the observed downstream usage of Jκ genes. The observed bias to downstream Jκ genes in Vκ38C B cells and in B cells with a noneditor L chain (Fig. 3,A) is consistent with random rearrangement as it is represented by a computational model (Fig. 3,B). In the random process, Jκ genes upstream from a Jκ gene, which has participated in a rearrangement, are excluded from participating again and making them less likely to form products (Fig. 3,B). It is possible that B cells with Vκ38C B cells and B cells with a noneditor L chain have a restricted time window in which they have to seclude the rearrangement process of a second L chain. As discussed above, rearrangement of a second allele takes time, and we propose that this time restriction can lead to random rearrangement, as computed in Fig. 3 B.

A possible explanation for the Jκ2 skewing in Vκ21D L chain is the rate of B cell generation. Cells that successfully express an editor on the first try of κ rearrangement are rare, but because they are generated in a single step, they are produced at a higher rate than cells that must muddle through several rounds of editing rearrangement. In the case of Vκ21D, rearrangement to downstream Jκ may be even less frequent because of the placement of the gene proximal to the Jκ cluster (65).

We thank Drs. Martin Weigert, Chandra Mohan, Gudrun Debes, Hidehiro Fukuyama, Yoram Louzoun, and Salar Khan for critical reading of the manuscript and helpful discussions. We thank the Sequencing Facility of the University of Chicago (Chicago, IL) for sequencing and the Bioinformatics Unit of the Weizmann Institute (Rehovot, Israel).

The authors have no financial conflict of interest.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

1

This work was supported by National Institutes of Health Grant GM02964-32, by a grant from the Lupus Research Institute, and by the Israeli Science Foundation.

3

Abbreviations used in this paper: MZ, marginal zone; cGvH, chronic graft-vs-host reaction; FR, framework region.

4

The online version of this article contains supplemental material.

1
Nossal, G. J., B. L. Pike.
1980
. Clonal anergy: persistence in tolerant mice of antigen-binding B lymphocytes incapable of responding to antigen or mitogen.
Proc. Natl. Acad. Sci. USA
77
:
1602
-1606.
2
Adams, E., A. Basten, C. C. Goodnow.
1990
. Intrinsic B-cell hyporesponsiveness accounts for self-tolerance in lysozyme/anti-lysozyme double-transgenic mice.
Proc. Natl. Acad. Sci. USA
87
:
5687
-5691.
3
Gay, D., T. Saunders, S. Camper, M. Weigert.
1993
. Receptor editing: an approach by autoreactive B cells to escape tolerance.
J. Exp. Med.
177
:
999
-1008.
4
Tiegs, S. L., D. M. Russell, D. Nemazee.
1993
. Receptor editing in self-reactive bone marrow B cells.
J. Exp. Med.
177
:
1009
-1020.
5
Nemazee, D., K. Buerki.
1989
. Clonal deletion of autoreactive B lymphocytes in bone marrow chimeras.
Proc. Natl. Acad. Sci. USA
86
:
8039
-8043.
6
Goodnow, C. C., J. Crosbie, S. Adelstein, T. B. Lavoie, S. J. Smith-Gill, R. A. Brink, H. Pritchard-Briscoe, J. S. Wotherspoon, R. H. Loblay, K. Raphael, et al
1988
. Altered immunoglobulin expression and functional silencing of self-reactive B lymphocytes in transgenic mice.
Nature
334
:
676
-682.
7
Nemazee, D..
2006
. Receptor editing in lymphocyte development and central tolerance.
Nat. Rev. Immunol.
6
:
728
-740.
8
Wardemann, H., S. Yurasov, A. Schaefer, J. W. Young, E. Meffre, M. C. Nussenzweig.
2003
. Predominant autoantibody production by early human B cell precursors.
Science
301
:
1374
-1377.
9
Yurasov, S., H. Wardemann, J. Hammersen, M. Tsuiji, E. Meffre, V. Pascual, M. C. Nussenzweig.
2005
. Defective B cell tolerance checkpoints in systemic lupus erythematosus.
J. Exp. Med.
201
:
703
-711.
10
Giachino, C., E. Padovan, A. Lanzavecchia.
1995
. κ+λ+ dual receptor B cells are present in the human peripheral repertoire.
J. Exp. Med.
181
:
1245
-1250.
11
Casellas, R., Q. Zhang, N. Y. Zheng, M. D. Mathias, K. Smith, P. C. Wilson.
2007
. Igκ allelic inclusion is a consequence of receptor editing.
J. Exp. Med.
204
:
153
-160.
12
Huang, H., J. F. Kearney, M. J. Grusby, C. Benoist, D. Mathis.
2006
. Induction of tolerance in arthritogenic B cells with receptors of differing affinity for self-antigen.
Proc. Natl. Acad. Sci. USA
103
:
3734
-3739.
13
Liu, S., M. G. Velez, J. Humann, S. Rowland, F. J. Conrad, R. Halverson, R. M. Torres, R. Pelanda.
2005
. Receptor editing can lead to allelic inclusion and development of B cells that retain antibodies reacting with high avidity autoantigens.
J. Immunol.
175
:
5067
-5076.
14
Sirac, C., C. Carrion, S. Duchez, I. Comte, M. Cogne.
2006
. Light chain inclusion permits terminal B cell differentiation and does not necessarily result in autoreactivity.
Proc. Natl. Acad. Sci. USA
103
:
7747
-7752.
15
Erikson, J., M. Z. Radic, S. A. Camper, R. R. Hardy, C. Carmack, M. Weigert.
1991
. Expression of anti-DNA immunoglobulin transgenes in non-autoimmune mice.
Nature
349
:
331
-334.
16
Chen, C., Z. Nagy, E. L. Prak, M. Weigert.
1995
. Immunoglobulin heavy chain gene replacement: a mechanism of receptor editing.
Immunity
3
:
747
-755.
17
Li, H., Y. Jiang, E. L. Prak, M. Radic, M. Weigert.
2001
. Editors and editing of anti-DNA receptors.
Immunity
15
:
947
-957.
18
Chen, C., M. Z. Radic, J. Erikson, S. A. Camper, S. Litwin, R. R. Hardy, M. Weigert.
1994
. Deletion and editing of B cells that express antibodies to DNA.
J. Immunol.
152
:
1970
-1982.
19
Li, Y., Y. Louzoun, M. Weigert.
2004
. Editing anti-DNA B cells by Vλx.
J. Exp. Med.
199
:
337
-346.
20
Witsch, E. J., H. Cao, H. Fukuyama, M. Weigert.
2006
. Light chain editing generates polyreactive antibodies in chronic graft-versus-host reaction.
J. Exp. Med.
203
:
1761
-1772.
21
Radic, M. Z., J. Mackle, J. Erikson, C. Mol, W. F. Anderson, M. Weigert.
1993
. Residues that mediate DNA binding of autoimmune antibodies.
J. Immunol.
150
:
4966
-4977.
22
Li, Y., H. Li, D. Ni, M. Weigert.
2002
. Anti-DNA B cells in MRL/lpr mice show altered differentiation and editing pattern.
J. Exp. Med.
196
:
1543
-1552.
23
Li, Y., H. Li, M. Weigert.
2002
. Autoreactive B cells in the marginal zone that express dual receptors.
J. Exp. Med.
195
:
181
-188.
24
Doyle, C. M., J. Han, M. G. Weigert, E. T. Prak.
2006
. Consequences of receptor editing at the lambda locus: multireactivity and light chain secretion.
Proc. Natl. Acad. Sci. USA
103
:
11264
-11269.
25
Sekiguchi, D. R., L. Yunk, D. Gary, D. Charan, B. Srivastava, D. Allman, M. G. Weigert, E. T. Prak.
2006
. Development and selection of edited B cells in B6.56R mice.
J. Immunol.
176
:
6879
-6887.
26
Fukuyama, H., F. Nimmerjahn, J. V. Ravetch.
2005
. The inhibitory Fcγ receptor modulates autoimmunity by limiting the accumulation of immunoglobulin G+ anti-DNA plasma cells.
Nat. Immunol.
6
:
99
-106.
27
Rozzo, S. J., T. J. Vyse, C. G. Drake, B. L. Kotzin.
1996
. Effect of genetic background on the contribution of New Zealand Black loci to autoimmune lupus nephritis.
Proc. Natl. Acad. Sci. USA
93
:
15164
-15168.
28
Vidal, S., D. H. Kono, A. N. Theofilopoulos.
1998
. Loci predisposing to autoimmunity in MRL-Fas lpr and C57BL/6-Faslpr mice.
J. Clin. Invest.
101
:
696
-702.
29
Santiago, M. L., C. Mary, D. Parzy, C. Jacquet, X. Montagutelli, R. M. Parkhouse, R. Lemoine, S. Izui, L. Reininger.
1998
. Linkage of a major quantitative trait locus to Yaa gene-induced lupus-like nephritis in (NZW × C57BL/6)F1 mice.
Eur. J. Immunol.
28
:
4257
-4267.
30
Morel, L., X. H. Tian, B. P. Croker, E. K. Wakeland.
1999
. Epistatic modifiers of autoimmunity in a murine model of lupus nephritis.
Immunity
11
:
131
-139.
31
Bolland, S., Y. S. Yim, K. Tus, E. K. Wakeland, J. V. Ravetch.
2002
. Genetic modifiers of systemic lupus erythematosus in FcγRIIB−/− mice.
J. Exp. Med.
195
:
1167
-1174.
32
Liu, Y., L. Li, K. R. Kumar, C. Xie, S. Lightfoot, X. J. Zhou, J. F. Kearney, M. Weigert, C. Mohan.
2007
. Lupus susceptibility genes may breach tolerance to DNA by impairing receptor editing of nuclear antigen reactive B cells.
J. Immunol.
179
:
1340
-1352.
33
Schlissel, M. S., D. Baltimore.
1989
. Activation of immunoglobulin κ gene rearrangement correlates with induction of germline κ gene transcription.
Cell
58
:
1001
-1007.
34
Paul, E., J. Lutz, J. Erikson, M. C. Carroll.
2004
. Germinal center checkpoints in B cell tolerance in 3H9 transgenic mice.
Int. Immunol.
16
:
377
-384.
35
Louzoun, Y., T. Friedman, E. Luning Prak, S. Litwin, M. Weigert.
2002
. Analysis of B cell receptor production and rearrangement. Part I. Light chain rearrangement.
Semin. Immunol.
14
:
169
-190. discussion 221–122.
36
Kalled, S. L., P. H. Brodeur.
1991
. Utilization of Vκ families and Vκ exons: implications for the available B cell repertoire.
J. Immunol.
147
:
3194
-3200.
37
Reth, M., P. Gehrmann, E. Petrac, P. Wiese.
1986
. A novel VH to VHDJH joining mechanism in heavy-chain-negative (null) pre-B cells results in heavy-chain production.
Nature
322
:
840
-842.
38
Reth, M. G., S. Jackson, F. W. Alt.
1986
. VHDJH formation and DJH replacement during pre-B differentiation: non-random usage of gene segments.
EMBO J.
5
:
2131
-2138.
39
Wood, D. L., C. Coleclough.
1984
. Different joining region J elements of the murine κ immunoglobulin light chain locus are used at markedly different frequencies.
Proc. Natl. Acad. Sci. USA
81
:
4756
-4760.
40
Radic, M. Z., J. Erikson, S. Litwin, M. Weigert.
1993
. B lymphocytes may escape tolerance by revising their antigen receptors.
J. Exp. Med.
177
:
1165
-1173.
41
Nemazee, D., M. Weigert.
2000
. Revising B cell receptors.
J. Exp. Med.
191
:
1813
-1817.
42
Bruley-Rosset, M., L. Mouthon, Y. Chanseaud, F. Dhainaut, J. Lirochon, D. Bourel.
2003
. Polyreactive autoantibodies purified from human intravenous immunoglobulins prevent the development of experimental autoimmune diseases.
Lab. Invest.
83
:
1013
-1023.
43
Ochsenbein, A. F., R. M. Zinkernagel.
2000
. Natural antibodies and complement link innate and acquired immunity.
Immunol. Today
21
:
624
-630.
44
Logtenberg, T., F. M. Young, J. H. Van Es, F. H. Gmelig-Meyling, F. W. Alt.
1989
. Autoantibodies encoded by the most JH-proximal human immunoglobulin heavy chain variable region gene.
J. Exp. Med.
170
:
1347
-1355.
45
Casellas, R., T. A. Shih, M. Kleinewietfeld, J. Rakonjac, D. Nemazee, K. Rajewsky, M. C. Nussenzweig.
2001
. Contribution of receptor editing to the antibody repertoire.
Science
291
:
1541
-1544.
46
Allman, D. M., S. E. Ferguson, V. M. Lentz, M. P. Cancro.
1993
. Peripheral B cell maturation. II. Heat-stable antigenhigh splenic B cells are an immature developmental intermediate in the production of long-lived marrow-derived B cells.
J. Immunol.
151
:
4431
-4444.
47
Rocha, B., C. Penit, C. Baron, F. Vasseur, N. Dautigny, A. A. Freitas.
1990
. Accumulation of bromodeoxyuridine-labeled cells in central and peripheral lymphoid organs: minimal estimates of production and turnover rates of mature lymphocytes.
Eur. J. Immunol.
20
:
1697
-1708.
48
Schlissel, M..
2002
. Allelic exclusion of immunoglobulin gene rearrangement and expression: why and how?.
Semin. Immunol.
14
:
207
-212;. discussion 225–206.
49
Retter, M. W., D. Nemazee.
1998
. Receptor editing occurs frequently during normal B cell development.
J. Exp. Med.
188
:
1231
-1238.
50
Hertz, M., D. Nemazee.
1997
. BCR ligation induces receptor editing in IgM+IgD bone marrow B cells in vitro.
Immunity
6
:
429
-436.
51
He, X., C. A. Janeway, Jr, M. Levine, E. Robinson, P. Preston-Hurlburt, C. Viret, K. Bottomly.
2002
. Dual receptor T cells extend the immune repertoire for foreign antigens.
Nat. Immunol.
3
:
127
-134.
52
Padovan, E., G. Casorati, P. Dellabona, S. Meyer, M. Brockhaus, A. Lanzavecchia.
1993
. Expression of two T cell receptor α chains: dual receptor T cells.
Science
262
:
422
-424.
53
Hardardottir, F., J. L. Baron, C. A. Janeway, Jr.
1995
. T cells with two functional antigen-specific receptors.
Proc. Natl. Acad. Sci. USA
92
:
354
-358.
54
Qian, Y., H. Wang, S. H. Clarke.
2004
. Impaired clearance of apoptotic cells induces the activation of autoreactive anti-Sm marginal zone and B-1 B cells.
J. Immunol.
172
:
625
-635.
55
Chen, X., F. Martin, K. A. Forbush, R. M. Perlmutter, J. F. Kearney.
1997
. Evidence for selection of a population of multi-reactive B cells into the splenic marginal zone.
Int. Immunol.
9
:
27
-41.
56
Fulcher, D. A., A. B. Lyons, S. L. Korn, M. C. Cook, C. Koleda, C. Parish, B. Fazekas de St. Groth, A. Basten.
1996
. The fate of self-reactive B cells depends primarily on the degree of antigen receptor engagement and availability of T cell help.
J. Exp. Med.
183
:
2313
-2328.
57
Cinamon, G., M. Matloubian, M. J. Lesneski, Y. Xu, C. Low, T. Lu, R. L. Proia, J. G. Cyster.
2004
. Sphingosine 1-phosphate receptor 1 promotes B cell localization in the splenic marginal zone.
Nat. Immunol.
5
:
713
-720.
58
Martin, F., J. F. Kearney.
2002
. Marginal-zone B cells.
Nat. Rev. Immunol.
2
:
323
-335.
59
Cariappa, A., M. Tang, C. Parng, E. Nebelitskiy, M. Carroll, K. Georgopoulos, S. Pillai.
2001
. The follicular versus marginal zone B lymphocyte cell fate decision is regulated by Aiolos, Btk, and CD21.
Immunity
14
:
603
-615.
60
Pillai, S..
1999
. The chosen few: positive selection and the generation of naive B lymphocytes.
Immunity
10
:
493
-502.
61
Ekland, E. H., R. Forster, M. Lipp, J. G. Cyster.
2004
. Requirements for follicular exclusion and competitive elimination of autoantigen-binding B cells.
J. Immunol.
172
:
4700
-4708.
62
Mandik-Nayak, L., S. Seo, A. Eaton-Bassiri, D. Allman, R. R. Hardy, J. Erikson.
2000
. Functional consequences of the developmental arrest and follicular exclusion of anti-double-stranded DNA B cells.
J. Immunol.
164
:
1161
-1168.
63
Novobrantseva, T., S. Xu, J. E. Tan, M. Maruyama, S. Schwers, R. Pelanda, K. P. Lam.
2005
. Stochastic pairing of Ig heavy and light chains frequently generates B cell antigen receptors that are subject to editing in vivo.
Int. Immunol.
17
:
343
-350.
64
Nemazee, D..
1996
. Antigen receptor “capacity” and the sensitivity of self-tolerance.
Immunol. Today
17
:
25
-29.
65
Brekke, K. M., W. T. Garrard.
2004
. Assembly and analysis of the mouse immunoglobulin κ gene sequence.
Immunogenetics
56
:
490
-505.

Supplementary data