The role of rearrangement at the second immunoglobulin heavy chain locus in maintaining B-cell tolerance to DNA

Abstract

In recently generated B6.56R anti-DNA autoantibody transgenic mice, it was noted that a substantial faction of the B-cells that had “avoided” DNA-reactivity had done so through the rearrangement and usage of the endogenous, non-targeted HC allele. This suggested that rearrangement at the second HC locus might be an important mechanism through which self-reactive B-cells might successfully revise their initial antigen specificity. To test the importance of this mechanism in B-cell tolerance, we generated B6.56R/56R mice that possessed the “56R” anti-DNA heavy chain transgene inserted into both HC loci. These transgenic homozygotes developed higher titers of anti-DNA antibodies, with an expanded population of B220^low, MHC-Class II^lo B-cells, enriched for CD21^low, CD23^low pre-plasmablasts. The analysis of hybridomas from these mice revealed that the only avenue by which these B-cells could avoid DNA-reactivity was through the use of the editor light chains, V_k20 or V_k21. Hence, in addition to LC editing, rearrangement and usage of the second HC locus/allele constitutes an important “safety valve” for B-cells whose primary BCR confers DNA-reactivity. In contrast to these tolerance mechanisms, editing the first rearranged HC locus (through HC replacement) and somatic mutations appear to be less frequently employed to edit/revise self-reactive B cells.

INTRODUCTION

Newly generated B-lymphocytes are known to undergo a meticulous censoring program through which self-reactive B-cells are removed from the immune system. This is an important ritual since the persistence of potentially self-reactive B-cells and their subsequent activation can lead to autoimmunity. The mechanisms through which self-reactive B-cells are censored have been extensively studied, particularly through the use of BCR transgenic (Tg) models. It is now apparent that several checkpoints exist to maintain B-cell tolerance, as reviewed elsewhere (1).

Studies in BCR transgenic models have demonstrated that B-cells with strong self-reactivity are censored in the bone marrow largely through receptor–editing and deletion (1–5). Elegant work by several investigators has shown that secondary rearrangement at the light chain (LC) locus is a particularly effective mechanism that B-cells commonly employ to veto self-reactivity (6–9). Examples of receptor revision at the heavy chain (HC) locus have also been documented (10–12). B-cells that have weaker degrees of self-reactivity have been shown to be rendered “anergic” or functionally incapacitated (1, 13, 14). Additional tolerance events in the germinal center have also been uncovered (15–17). Hence, a rich panoply of central and peripheral mechanisms exists to censor self-reactive B-cells from the re-circulating immune system.

It has also become apparent that the genetic background can substantially impact how effectively anti-self B-cells are censored. Of relevance to this report, studies using the “3H9/56R” anti-DNA immunoglobulin (Ig) Tg model have shed light on how different genetic backgrounds might impact B-cell tolerance to DNA. Thus, whereas anti-DNA B-cells are effectively tolerized in the normal BALB/c background, tolerance is breached when the same Tg is bred onto the MRL/lpr background (18, 19). Interestingly, on the C57BL/6 (B6) background, an intermediate degree of breach in tolerance is noted, as reported elsewhere (20–22).

Thus, when the “56R” anti-DNA Ig Tg (with an arginine residue at position 56 in the CDR2 region of the “3H9” anti-DNA Ig HC) is bred onto the B6 background, these mice develop modest titers of autoantibodies to DNA as early as 4–6 mo of age. Potentially, B-cells from these anti-DNA HC-only Tg mice could potentially preclude DNA-reactivity in a couple of different ways – rearrangement and usage of the endogenous HC allele, pairing of the Tg HC with a LC partner that vetoes DNA-binding, mutation of the Tg HC so that it loses affinity for DNA, and replacement of the targeted HC by an endogenous VH gene. When hybridomas were previously derived from B6.56R mice to determine the molecular makeup of anti-DNA and non-anti-DNA Abs, it was intriguing to note that a substantial faction of the B-cells that had “avoided” DNA-reactivity had done so through the rearrangement and usage of endogenous (i.e., non-targeted) HC locus (21). Most of the remaining B-cells were apparently not DNA-reactive because they had paired the Tg 56R HC with a non-DNA binding LC partner, a mechanism that has been well documented in this Tg model (6, 20).

The above findings suggested that rearrangement at the second (non-targeted) HC locus might be an important mechanism through which self-reactive B-cells might successfully revise their initial antigen specificity. However, these observations did not indicate if this particular molecular mechanism was absolutely essential for thwarting autoimmunity. Conversely, when B-cells did not have the option of rearranging/using a second HC locus, it was not apparent whether LC editing alone was sufficient for “revising” all self-reactivity. To investigate this, we generated a mouse strain, referred to as B6.56R/56R that possessed the “56R” anti-DNA HC Tg inserted into both HC loci. If these B6.56R/56R mice were to exhibit a greater degree of autoreactivity compared to Tg-hemizygous B6.56R/+ mice, this would suggest that rearrangement at the alternate HC locus and usage of the second HC allele plays an additional role in maintaining self-tolerance. On the other hand, if both the hemizygous and homozygous Tg mice exhibited similar degrees of autoreactivity, this would indicate that other mechanisms (e.g., LC editing, mutation of Tg HC, etc) may be sufficient to thwart autoimmunity, without the need for any additional contribution from rearrangement at the alternate HC locus.

Materials and Methods

Mice

“56R” anti-DNA HC-only site-directed Tg mice were a kind gift from Dr. Martin Weigert (University of Chicago) (6). These mice were backcrossed onto the C57BL/6 (B6) background, as recently described (20, 21). “B6.56R/+” mice are hemizygous for the 56R anti-DNA Ig HC Tg, whereas “B6.56R/56R” mice bear the Tg knocked into both the HC loci. Double Tg mice were identified by PCR analysis using J_H1/J_H2 intron specific PCR primers that will amplify DNA from the non-targeted locus, but not the Tg, and confirmed by flow cytometry (based on the absence on any staining for surface IgM^b). B6.56R/56R mice were bred to B6.56R/+ mice to generate both the hemizygous and Tg homozygous mice. All mice used for this study were bred and housed in a specific pathogen free colony, at UT Southwestern Medical Center Department of Animal Resources, Dallas, TX. Comparable numbers of male and female mice were pooled for all experiments, as no significant sex differences were noted in the phenotypes studied.

Immunophenotyping of mice

Splenocytes were depleted of red blood cells using tris ammonium chloride, and single-cell suspensions were prepared for culture or flow cytometric analysis, as described previously (21, 23). Sera and monoclonal antibody (mAb) culture supernatants were screened for antibody-reactivity to ssDNA, dsDNA, and histone/DNA complexes by ELISA, as described elsewhere (21, 23). All serum samples were diluted with serum dilution buffer (2% BSA, 3mM EDTA, 0.05% Tween20, 0.1% gelatin). A positive control serum sample derived from a B6.Sle1^z.Sle3^z mouse (23; which bears the IgM^b/IgG2a^b allotype), or a serum sample derived from a seropositive B6.Sle1^z.56R mouse (which bears the IgM^a/IgG2a^a allotype) was included as a test standard. Sera with reactivities stronger than the test standard were diluted further and re-assayed one more time. For the allotype-specific ELISA, the above assays were repeated with one modification: instead of using anti-IgM or anti-IgG second-step antibodies, biotin-coupled anti-IgM^a, anti-IgM^b, anti-IgG2a^a, or anti-IgG2a^b antibodies were used, followed by avidin-coupled alkaline phosphatase.

B-cell sorting and culture

Spleens were removed from anesthetized mice. Splenocytes were depleted of red blood cells using ACK lysis buffer (Biosource, Rockville, MD) and single-cell suspensions were prepared as for Flow cytometry. Cells were stained with anti-B220, and sorted based on the level of B220 expression, using BD FACSAria™ (BD Bioscience). B220^high and B220^intermediate cells were collected and cultured at 1 × 10⁶ cells/ml for 48 hours in DMEM medium supplemented with 10% fetal bovine serum (FBS), 25 mM Hepes, 2 mM L-glutamine, antibiotics, 50 µM 2-mercaptoethanol, with or without LPS (20 ug/ml). Culture supernatants were analyzed for antibody levels by ELISA.

Assessment of nephritis

Mice were monitored at 6 mo and 9–12 mo of age for evidence of clinical nephritis. Urine was collected over a 24-h period using metabolic cages, and the total amount of urinary protein was assayed using a Comassie-based kit (Pierce, Rockford, IL). The level of blood urea nitrogen (BUN) was measured using a commercially available assay (Sigma Chemicals, St. Louis, MO). For the analysis of kidney sections, at least 100 glomeruli were examined per section by light microscopy, for evidence of hypertrophy, proliferative changes, crescent formation, hyaline deposits, fibrosis/sclerosis, and basement membrane thickening, and graded semi-quantitatively, as detailed previously (23).

Hybridomas

Spleens were removed aseptically from 9–12 mo old seropositive Tg mice, and the splenocytes were fused to the SP2/0 fusion partner, as described before (21, 24). All wells with single colonies were tested for Ab production 7–10 days post-fusion. Cells from positive wells were subcloned once more. Ab concentration was determined using a Coomassie PLUS protein assay kit (Pierce, Rockford, IL), and an isotype specific ELISA, as described above.

Sequence analysis and statistics

The HC and LC sequences were amplified and sequenced from the cDNA of the hybridomas. The primers and conditions for HC and LC RT-PCR were determined as described previously (21, 24). Specifically, the HC was amplified using a 5' primer that hybridizes to all V_H genes (5'-AGGT(G/C)(A/C)A(A/G)CTGCAG(G/C)AGTC(A/T)GG-3') and 3’ primers to Cµ (5'-CAGGGGGCTCTCGCAGGAGACGAGG-3') and Cγ (5'-GGACAGGGATCCAGAGTTCC-3'), as described previously (24). Hence, this should allow the amplification of Tg as well as endogenous sequences, including those with non-productive rearrangements and edited sequences with V_H replacements or D-invasions. The Ig-κ LC primers used have the potential to amplify most Ig-κ gene families including Vκ1, Vκ2, Vκ4, Vκ9, Vκ19, Vκ20, Vκ21, Vκ23, and Vκ38c, based on our previous work (21, 24 and data not shown). All antibody sequences were “blasted” against the public “IgBlast” database of mouse Ig sequences at NIH/NCBI (http://www.ncbi.nlm.nih.gov/igblast/), to determine the closest germline gene of origin. Comparing these sequences to the deposited repertoire of germline genes allowed us to identify residues that were potentially somatically mutated. The CDR position and numbering scheme adopted matched that used by the NCBI IgBlast database. The usage frequencies of individual genes were compared using Chi square tests. Where applicable, the Fisher’s exact test was applied. Inter-group comparisons of phenotypes between the two Tg strains were carried out using a one-tailed Student’s t-test (SigmaStat; Jandel Scientific). Complete sequence information is available from the authors.

RESULTS

Interestingly, B6.56R/56R homozygous mice exhibited elevated levels of IgM^a anti-ssDNA and anti-dsDNA Abs, relative to B6.56R/+ hemizygous controls, as displayed in Fig. 1. This serological difference was apparent as early as 3–6 mo of age, and became more pronounced with age. However, there were no significant differences in the titers of IgG (IgG2A^a) Tg-encoded anti-DNA Abs between the hemizygous and homozygous Tg mice (Fig. 1). There were also no gender differences in these phenotypes (data not shown). Finally, there was no evidence of renal disease in these strains, reproducing earlier results in B6.56R/+ mice (21; data not shown).

Figure 1. Serum levels of IgM and IgG Abs to nuclear antigens in B6.56R/+ and B6.56R/56R Tg mice.

Allotype-specific IgM and IgG Abs to ssDNA, dsDNA were measured by ELISA. Both 56R homozygous and hemizygous Tg male and female mice (pooled) were analyzed at the ages of 3–6 mo and 9–12 mo. Each dot represents a single mouse. The horizontal lines represent the mean levels of serum autoantibodies in each group of mice. The indicated P values represent the result of a Student’s t-test comparison of B.56R/56R versus B6.56R/+ mice. OD₄₀₅ readings for the positive control sera ranged from 1.2–1.5. NS = not significant

Next, we examined the cellular makeup of the lymphoid organs in B6.56R/56R and B6.56R/+ mice. Both strains exhibited similar splenic weights and cell numbers (Table 1). Likewise, both strains exhibited similar numbers of CD4 and CD8 T-cells, as well as B-cells (Table 1). Most of the B-cells in B6.56R/+ spleens exhibited high surface levels of IgM^a (i.e., with high mean fluorescent intensity), and also exhibited low grade staining for the endogenous HC, IgM^b (Fig. 2, Table 1). In contrast, almost all the B-cells in B6.56R/56R Tg mice expressed higher staining intensity for IgM^a with the absence of staining for endogenous IgM^b, as one might expect (Fig. 2). If the Tg B-cells were segregate based on their surface levels of IgM^a into IgM^a-high, IgM^a-medium and IgM^a-low B-cells, as demonstrated in Fig. 2, B6.56R/56R mice possessed significantly higher percentages of IgM^a-high B-cells but lower percentages of IgM^a-low B-cells, compared to B6.56R/+ mice (Table 1, Fig. 2).

Table 1.

Lymphocyte composition of spleen and peritonal cavity¹

	B6.56R/+ (N=11)	B6.56R/56R (N=13)	P value
Spleen
Weight (mg)	109 ± 15.9	104.5 ± 27.6	NS
Total cell numbers (×106)	68 ± 20.9	66.3 ± 14.5	NS
Total CD4+ (×106)	18.7 ± 2.4	20.9 ± 4.5	NS
% of all splenocytes	12.7 ± 0.6	13.6 ± 0.9	NS
Total CD8+ (×106)	18.4 ± 1.9	14.9 ± 3.9	0.017
% of all splenocytes	12.5 ± 0.6	10.1 ± 0.8	0.015
CD4:CD8 ratio	1.0	1.4	NS
CD4:%CD69+	17.2 ± 3.4	17.5 ± 7.3	NS
CD8:%CD69+	18.4 ± 1.9	14.9 ± 3.9	0.003
Total B cells (×106)	41.9 ± 0.5	44.6 ± 1.9	NS
% of all splenocytes	28.5 ± 3.1	29.4 ± 2	NS
IgMa (mfi)	218.4 ± 23.9	215 ± 23.7	NS
%IgMa+	69.1 ± 6.2	95.4 ± 1.7	0.0001
%IgMa high	48.7 ± 4.6	56.6 ± 4.9	0.0009
%IgMa medium	21.6 ± 2.7	26.9 ± 2.8	0.0003
%IgMa low	19.7 ± 2.9	11.9 ± 2.4	0.0001
%IgMb+	14.2 ± 2.7	0.8 ± 0.7	0.0001
% B1a cells	3.9 ± 1.9	4.7 ± 2.2	NS
Among IgMa cells
% follicular B-cells	42.5 ± 4	39.8 ± 1.5	NS
% MZ B-cells	44.3 ± 4.2	48.2 ± 1.6	NS
% CD21− CD23− B-cells	17 ± 4	10.8 ± 1.3	NS
% Igλ +	8.2 ± 1.2	10.4 ± 1.3	NS
I-Ab (mfi)	803 ± 100.6	687.9 ± 60.7	NS
CD86 (mfi)	46.3 ± 2.7	143.8 ± 12.9	0.0001
Peritoneal cavity
Among B220+ cells
% B1a	31.9 ± 11.8	8.9 ± 6.4	0.0001
% B1b	37.1 ± 8.2	44.41 ± 17.7	NS
% B2	31 ± 7.4	46.8 ± 19.8	0.023

¹Mice of each strain were examined at 9–12-mo age. Total cell numbers and percentage of splenocyte subsets shown are mean ± SEM.

The P-values shown pertain to one-tailed t-test comparisons of the 2 strains. NS means "not significant". mfi = geometric mean fluorescence intensity.

Figure 2. Splenic B-cell phenotypes in B6.56R/+ and B6.56R/56R Tg mice.

Splenocytes from 9–12 mo old B6.56R/+ (TOP) and B6.56R/56R mice (BOTTOM) were analyzed by flow cytometry for the expression of B220 vs IgM^a, IgM^b, CD86 and I-A^b, respectively. IgM^a B-cells were further divided arbitrarily into three groups as demarcated by the dotted lines: IgM^a-high B-cells, IgM^a-medium B-cells and IgM^a-low B-cells; the frequencies of B-cells with different surface IgM^a levels are summarized in Table 1. The boxes drawn indicate the expanded population of B-cells in B6.56R/56R Tg mice that were B220^int, I-Ab^int, CD86^hi, relative to the B6.56R/+ controls. The profiles shown are representative of data obtained from 8 mice per stain.

When activation markers were examined, it was interesting that B6.56R/56R B-cells expressed significantly higher levels of CD86 (B7-2) as illustrated in Fig. 2, and quantitated in Table 1. Interestingly, 2 clusters of B-cells were discerned based on their surface levels of B220 and I-A^b – B220^int, I-A^b-int B-cells (bearing intermediate levels of both surface molecules) and B220^hi, I-A^b-hi B-cells, with the former being boxed in Fig. 2. Interestingly, B220^int, I-A^b-int B-cells were significantly expanded in B6.56R/56R mice (Table 2). When these 2 populations of B-cells were further sub-gated, it was apparent that the expanded population of B220^int I-A^b-int B-cells in B6.56R/56R spleens were skewed towards being CD21⁻CD23⁻ in surface phenotype, compared to B220^hi I-A^b-hi B-cells from the same spleens (Table 2, Fig. 3). The possibility that the expanded population of B-cells in the Tg homozygous mice might include pre-plasmablasts is supported by their increased production of Abs in culture (Fig. 4). When all B-cells were pooled, the B6.56R/56R and B6.56R/+ spleens did not differ significantly in their overall fractions of B-cells that were follicular, marginal-zone, or B1a in phenotype (Table 1). When peritoneal B-cells were examined, almost all B6.56R/56R B-cells lacked CD5; in contrast, a third of the peritoneal B-cells in B6.56R/+ mice were CD5-positive, i.e., B1a in phenotype (Table 1), suggesting that endogenous IgM^b expression was required for selection into the B1a pool, as noted in our earlier work (21).

Table 2.

Comparsion of B220^hi and B220^int B cells¹

	B220 hi	B220 int	P value
B6.56R/56R (n=12)
% of all B-cells	57 ± 4.3	42.9 ± 4.32
mean size	72.5 ± 1.0	64.5 ± 1.3	0.0001
% FO	39.0 ± 1.2	21 ± 2.2	0.0001
% MZ	56.2 ± 1.3	36.7 ± 3.3	0.0001
% CD21−,CD23−	4.1± 0.7	41.4 ± 4.7	0.0001
I-Ab (mfi)	883.1 ± 48.9	405.5 ± 39.6	0.0001
CD86 (mfi)	133.9 ± 12	122.3 ± 9.9	NS
B6.56R/+ (n=11)
% of all B-cells	65.2 ± 3.7	34 ± 3.82
mean size	71.7 ± 6.6	71.4 ± 0.4	NS
% FO	39.8 ± 2.2	21.7 ± 2.8	0.0001
% MZ	46.9 ± 2.0	37 ± 2.0	0.0012
% CD21−,CD23−	11.8 ± 0.9	38.5 ± 3.8	0.0001
I-Ab (mfi)	1050 ± 78.4	407.4 ± 39.6	0.0001
CD86 (mfi)	24.9 ± 4.9	37.5 ± 8	NS

¹Splenic cells from both strains were assayed by flow cytometry and B-cells were distinguished based on surface B220 levels. Tabulated are the surface phenotype and subset distribution data pertaining to these 2 B-cell types. P values pertain to comparisons of these 2 B-cell types using the Student’s t-test. NS means not significant.

²The percentage of B-cells that typed as B220^int I-A^b-int were significantly different in the 2 strains (P < 0.01).

Figure 3. Surface phenotypes of B220^int I-A^b-int and B220^hi I-A^b-hi B-cells from B6.56R/56R and control spleens.

Splenic B-cells from 56R Tg mice were first gated based on surface levels of B220 and I-A^b, as indicated in Figure 2, and then analyzed from their surface CD21/CD23 profiles. Shown plots are representative of data obtained from 11–12 mice each, as detailed in Table 3.

Figure 4. In vitro antibody production by B220^int and B220^hi B-cells from B6.56R/56R spleens.

Splenic B-cells were sorted based on surface B220 levels from B6.56R/56R mice and cultured at indicated cell numbers (per well) with LPS (20 ug/ml) for 48 h, after which culture supernatants were assayed for antibody levels by ELISA. Each bar represents the mean ± SEM of three experiments.

Shown in Table 3 are the HC and LC gene compositions noted in B6.56R/56R B-cells, juxtaposed to previously reported data from B6.56R/+ control (21). As is evident from Table 3, DNA-binding B-cells from both the B6.56R/+ and B6.56R/56R mice utilized the Tg HC, predominantly or exclusively, paired with a spectrum of LC partners, notably V_k20, V_k21 and V_k38c. When non-ANA B-cells were next examined (Table 3, bottom), interesting molecular differences were noted. More than 80% of the non-DNA binding B-cells from B6.56R/56R spleens possessed BCRs composed of the 56R HC Tg paired with either V_k20 or V_k21, both of which have previously been shown to be effective at vetoing DNA-binding (6). In contrast, about 70% of the non-ANA binding B-cells from B6.56R/+ spleens apparently “avoided” DNA-binding through the use of endogenous HC genes (21, Table 3). Hence, in the absence of an alternative (non targeted) HC locus, B-cells expressing a potentially self-reactive BCR HC appear to “avoid” self-reactivity exclusively through the use of DNA-vetoing LC partners. Very few of these B-cells showed evidence of mutations in the 56R Tg or evidence of HC replacement or D-invasion into the Tg (Table 3).

Table 3.

Heavy chain and light chain uasge analysis in hybridomas¹

		B6.56R/+	B6.56R/56R	P value (Chi-sqr)
ANA	Total cells (N)	62	69
	Heavy Chain amplified
	Endogenous HC (%)	29	0	0.001
	56R Tg HC (%)	71	100	0.001
	% of 56R HC Tg with mutations	2	5	ns
	Tg HC invasion (N)	2	1	ns
	IgG in isotype (%)	4	13	0.001
	LC usage among B-cells with Tg HC
	%Vk4	5	5	ns
	%Vk9	21	3	0.002
	%Vk19	20	5	0.004
	%Vk20	6	36	0.001
	%Vk21	0	18	0.003
	%Vk38c	28	20	ns
	%Other Vk	20	13	ns
	Jκ Usage among B-cells with Tg HC
	%Jk1,Jk2	40	28	ns
	%Jk4,Jk5	60	72	ns
non-ANA	Total cells (N)	27	21
	Heavy Chain amplified
	Endogenous HC (%)	70	0	0.001
	56R Tg HC (%)	30	100	0.001
	% of 56R HC Tg with mutations	4	0	ns
	Tg HC invasion (N)	1	0	ns
	IgG in isotype (%)	11	5	ns
	LC usage among B-cells with Tg HC
	%Vk9	19	6	ns
	%Vk20	52	31	ns
	%Vk21	0	56	0.001
	%Vk38c	6	0	ns
	%Other Vk	23	7	ns
	Jκ Usage among B-cells with Tg HC
	%Jk1,Jk2	25	63	0.009
	%Jk4,Jk5	75	37	0.009

¹Whereas the fusion data for the B6.56R/56R spleen was generated in this study, the data shown for the B6.56R/+ mouse has been recently reported elsewhere (21). "ns" = not significant

It was somewhat perplexing to note that whereas some B-cells expressing V_k20 paired with the Tg HC were DNA-reactive, others were not (Table 3). Interestingly, when the bt20 V_k20 germline gene was recombined to J_k4 it was associated with DNA-reactivity (when paired with the 56R HC); when the same V_k germline gene was recombined to J_k2, it was apparently not associated with DNA-reactivity (Fisher’s exact test P < 0.002). However, this difference alone is unlikely to account for the differential DNA reactivity, based on previous reports (20, 21). Likewise, it was intriguing that whereas some of the B-cells bearing the 56R HC Tg paired with the V_κ21 LC gene did not bind DNA, others clearly did (Table 3). When we compared these 2 specificities of B-cells, there were no significant differences in J_κ or V_κ germline gene usage between the DNA-binding and non-DNA binding B-cells. The specificity differences were not due to any somatic mutations since all V_κ21-expressing clones exhibited no mutations in the Tg HC or the LC partners (data not shown). Also, we could not amplify any Ig lambda from these B-cells. Currently, we do not know if these V_κ21-expressing B-cells differ in the expression of any additional LC partners that our PCR primers might have failed to amplify.

DISCUSSION

Developing pre-B-cells first re-arrange their HC locus during early B-cell development. In the event that the first V_H gene randomly selected for rearrangement confers self-reactivity when paired with a LC partner, several molecular events occurring at different stages of B-cell development have the potential to revise the molecular makeup of the initial BCR so that it is no longer self-reactive. These non-mutually exclusive options include:

1. rearrangement at the alternate HC locus and usage of the alternate HC allele

2. pairing of the HC with an appropriate LC partner (either generated through primary recombination or through LC editing), so that DNA-reactivity is precluded, as discussed previously (6, 19)

3. editing the first HC allele so that it is no longer self-reactive (e.g., through V_H replacement), as reported by Weigert and colleagues (10), and

4. mutation of the HC or the paired LC gene so that DNA-reactivity is abrogated.

When we recently derived B-cell hybridomas from B6 and B6.Sle2^z congenics bearing the 56R anti-DNA Tg, the first 2 of the above mechanisms were evidently more commonly employed, compared to the latter options (21). Surprisingly, in that study, the vast majority of the B6.56R-derived non-DNA binding B-cells had apparently “avoided” DNA-reactivity through the rearrangement of the non-targeted HC locus (i.e., by adopting the first mechanism listed above). Though it was clear that this mechanism was commonly employed to maintain B-cell tolerance (at least on the B6 background), it was not apparent if this mechanism was absolutely essential. This study, designed to directly test this, illustrates that B-cells that are engineered to lack a non-rearranged alternate HC locus are more likely to remain autoreactive, presumably because the only available mechanism through which they can revise their initial BCR is through LC editing.

Previous studies using Tg-hemizygous 3H9/56R/76R mice (with progressively increasing avidity of the Tg for DNA) have shown that (a) increasing the B-cell avidity for DNA has a profound influence on the degree and nature of B-cell editing, and (b) the availability of a non-targeted HC locus is not sufficient to prevent autoimmunity in there mice (25, 26). Although the latter mechanism of tolerance by itself may not be sufficient at re-establishing tolerance, it appears that it may contribute incrementally in this direction, based on findings from the present work. When operative, the non-DNA binding HC substitute appears to originate predominantly from de novo rearrangement of the non-targeted locus, rather than through V_H replacement at the targeted HC locus (21), and this difference may relate to the relative efficiencies of these two processes.

The phenotype of the B-cells observed in the Tg-homozygous mice points to a couple of interesting observations. First, the co-presence of the pre-arranged HC Tg at both loci did not alter total B-cell numbers, consistent with previous reports in HC Tg-homozygous mice (27). Also, the total IgM levels (i.e., IgM^a + IgM^b) on B-cells did not differ between the Tg-hemizygous and Tg-homozygous mice (Fig. 2, Table1), again in-keeping with previous reports in the literature (27). Presently, we do not know if the increased numbers of IgM^a-hi B-cells in Tg-homozygous mice might arise in part from expression of HC from both alleles, but this appears to be unlikely based on earlier reports in other Ab-Tg-homozygous mice (27).

Though Tg-hemizygous and Tg-homozygous mice did not differ significantly in B-cell numbers or surface levels of total IgM, it was interesting that the Tg-homozygous mice exhibited an expanded population of B220^int I-A^b-int B-cells. Since these B-cells were skewed toward being CD21^−ve CD23^−ve (Table 2), they may comprise of memory B-cells and/or pre-plasmablasts. Expansions of similar populations of B-cells have been noted in other autoimmune mouse models (28, 29). Consistent with this, sorted B220^int I-A^b-int B-cells from Tg-homozygous mice tended to elaborate more ANAs in culture (Fig 4). Based on these findings, we envisage the following model. Precluding B-cells from rearranging and using the alternate HC allele may augment the pool of autoreactive B-cells, as observed in the Tg-homozygous mice. Continuous stimulation of these B-cells by self DNA might skew these cells towards plasmacytoid and memory B-cell differentiation, a scenario which is consistent with the B-cell phenotypes and elevated autoantibody levels seen in these mice.

Based on these observations and previous reports in the literature, we propose that newly developing B-cells (in a non-Tg, physiological setting) may employ 2 main mechanisms to revise their BCR if the initial HC/LC combination turned out to be self-reactive - rearrangement at the second HC locus or LC editing. In a context where anti-self B-cells have the luxury of rearranging a second HC locus, as well as editing the LC partner, as in the case in B6.56R/+ mice, the former mechanism is used by 70% of the B-cells that might have potentially been DNA-reactive (21, Table 3). Precluding the former tolerance mechanism from operating (as engineered in the B6.56R/56R double knock-in mice) may amplify the degree of autoimmunity that ensues, as demonstrated in the present study. In these mice, editing at the LC locus using strong DNA-vetoing LC such as V_k20 and V_k21 appear to be the only mechanism available for precluding DNA-reactivity, a finding that resonates well with previous reports in Tg-hemizygous mice by Weigert and colleagues (6). Hence, in addition to LC editing, rearrangement and usage of the second HC locus constitutes an important “safety valve” for B-cells whose primary BCR confers DNA-reactivity. In contrast to these mechanisms, editing the first rearranged HC locus (through HC replacement) and somatic mutation do not appear to be commonly employed for editing/revising self-reactive B cells.