0014-2980/03/0808-2469$17.50+.50/0© 2003 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

Autoimmune NZB/NZW F1 mice utilize B cellreceptor editing for generating high-affinity anti-dsDNA autoantibodies from low-affinity precursors

Nurit Yachimovich-Cohen, Ruth Fischel, Neta Bachar, Yuval Yarkoni and Dan Eilat

Department of Medicine, Hadassah University Hospital, Faculty of Medicine, Hebrew University,Jerusalem, Israel

We have previously constructed knock-in (C57BL/6×BALB/c) F1 mice, each expressing ananti-DNA heavy (H) chain (D42), combined with one of three different light (L) chains, namelyV ‹ 1-J ‹ 1, V ‹ 4-J ‹ 4 or V ‹ 8-J ‹ 5. All of these H/L chain combinations bind DNA with similaraffinity and fine specificity. However, while mice carrying V ‹ 1-J ‹ 1-transgenic L chain weretolerized almost exclusively by L chain receptor editing, the mice expressing V ‹ 8-J ‹ 5 Lchains utilized clonal anergy as their principal mechanism of B cell tolerance. V ‹ 4-J ‹ 4 tar-geted mice exhibited an intermediate phenotype. In the present study, these three H/L chaincombinations were backcrossed onto the autoimmune NZB/NZW F1 mice. We find that themechanism of clonal anergy is abrogated in these mice, but that receptor editing is main-tained. Moreover, diseased NZB/NZW mice utilize L chain secondary rearrangements for thegeneration of high-affinity, anti-dsDNA-producing B cells from low-affinity precursors. Theedited B cell clones are not deleted or anergized in the autoimmune animal; rather they areselected for activation, class-switching and affinity maturation by somatic mutation. Theseresults suggest that B cell receptor editing plays an important role not only in toleranceinduction, but also in generating high-affinity autoreactive B cells in autoimmune diseases.

Key words: Anti-DNA / Receptor editing / Anergy / Lupus / Knock-in mice

Received 13/3/03Revised 13/6/03Accepted 1/7/03

[DOI 10.1002/eji.200324025]

Abbreviations: BCR: B cell receptor MFI: Mean fluores-cence intensity NZB: New Zealand Black NZW: New Zea-land White CFSE: Carboxyfluorescein diacetate succinimi-dyl ester RAG: Recombinase activating gene

1 Introduction

Mouse strains with an inherited predisposition to autoim-mune disorders provide an approach for studying thebreakdown of self-immunological tolerance and the pro-duction of autoantibodies. The NZB/NZW F1 mice areconsidered to be the murine model most closely resem-bling human systemic lupus erythematosus (SLE) [1].The lupus-like disease in these mice is more severe infemales and is accompanied by high-affinity anti-dsDNAautoantibodies that are believed to play a role in thedevelopment of fatal immune glomerulonephritis [1, 2].Both NZB and NZW parents contribute multiple suscep-tibility genes to the immune abnormalities of the F1hybrid mouse [3]. These include polyclonal B cell hyper-activity at an early age [4], the appearance of IgM anti-DNA Ab at 3–5 months, and a spontaneous switch from

IgM to IgG anti-DNA Ab at 5–7 months concomitantlywith the onset of severe SLE and renal disease [5].

Several recent studies on the pathogenesis of humanand murine SLE have suggested that intrinsic B celldefects may be the primary cause of the loss of self-tolerance and development of autoimmune disease [6,7]. Therefore, the mechanisms of B cell tolerance, partic-ularly those involving clonal deletion of autoreactive Bcells [8], clonal anergy [9], and B cell receptor (BCR) edit-ing by secondary rearrangements of H and L chain V-region genes [10] have become the focus of intensiveinvestigation.

We have previously established several transgenicmouse lines in which a somatically mutated H chain vari-able region gene (D42H), derived from an anti-DNAhybridoma (D42), or its unmutated counterpart (glD42H)were targeted to the IgH locus of the mouse [11]. On anon-autoimmune (C57BL/6×BALB/c) F1 genetic back-ground, these H-chain-only knock-in mice were shownto be tolerant in that they did not produce significanttiters of anti-DNA antibodies in their serum [11]. How-ever, when transferred to the autoimmune NZB/NZW F1genetic background, the targeted glD42H chain gave

Eur. J. Immunol. 2003. 33: 2469–2478 B cell receptor editing and anergy in autoimmune anti-DNA knock-in mice 2469

rise to age-dependent high titer DNA-binding antibodies,due to an apparent loss of tolerance and to the associa-tion of particular endogenous light chains with the trans-genic heavy chain [12]. The selected H/L chain combina-tions were expressed as somatically mutated IgG anti-bodies with high affinity for dsDNA. Indeed, the majorityof anti-DNA hybridomas derived from 7-month-old H-chain-targeted NZB/NZW F1 mice expressed a singlemouse VL gene product (V ‹ RF), rearranged to a single Jminigene (J ‹ 5) [12]. This L chain configuration was foundto be the same as that of the original NZB/NZW D42hybridoma (D42L).

To gain a better understanding of the different tolerancemechanisms and their dependence on the specificityand affinity for the tolerogen, we have constructed sev-eral knock-in (C57BL/6×BALB/c) F1 mouse lines withdiscrete Ig H/L combinations [13]. Specifically, theglD42H and D42H-transgenic mice were crossed withknock-in mice expressing V ‹ 1-J ‹ 1, V ‹ 4-J ‹ 4 or V ‹ 8-J ‹ 5L chains. In vitro transfection experiments have shownthat all of these combinations have similar fine specifici-ties for DNA and that the apparent DNA affinities are˚ 3×106 M-1/bp and ˚ 1×107 M-1/bp for each of the three

L chains, combined with glD42H and D42H, respectively.For comparison, the apparent DNA affinities of glD42Hand D42H, combined with V ‹ RF-J ‹ 5 (D42L) are 6.5 ‰ 107

and 7.8×108 M-1/bp, respectively. None of the double-transgenic mouse strains exhibited significant levels ofserum anti-DNA activity when expressed on the non-autoimmune mouse genetic background. However, whileB cells from mice carrying the V ‹ 1-J ‹ 1-transgenic Lchains were tolerized almost exclusively by L chainreceptor editing in an affinity-independent manner, themice expressing V ‹ 8-J ‹ 5 L chains utilized affinity-dependent clonal anergy as their principal mechanism ofB cell tolerance. This was probably due to inefficient sec-ondary rearrangements on the transgenic L chain alleleswith V ‹ -J ‹ 5 configuration. V ‹ 4-J ‹ 4 targeted mice exhib-ited an intermediate phenotype with respect to these twomechanisms of B cell tolerance [13].

In order to see which, if any, of these tolerance mecha-nisms is abrogated in the autoimmune, lupus-pronemouse, we have backcrossed the D42H chain transgeneand its combination with each of the three transgenic Lchains onto the NZB/NZW genetic background. We findthat B cell anergy and deletion are largely defective inthese autoimmune mice, while receptor editing is main-tained. However, BCR editing in diseased NZB/NZW F1mice can not only allow B cells to escape from autoreac-tivity, but it can also generate high-affinity anti-DNAautoantibodies, as part of the autoimmune process inlupus-prone mice.

2 Results

2.1 B cell receptor editing in anti-DNA transgenicNZB/NZW F1 mice

Anti-DNA H/L chain double-transgenic NZB/NZW F1mice were prepared by backcrossing D42H C57BL/6knock-in mice onto the NZB strain for 7–8 generations.This was followed by one cross with mice of the NZWstrain that had been made L chain-transgenic by back-crossing BALB/c mice with targeted V ‹ 1-J ‹ 1, V ‹ 4-J ‹ 4or V ‹ 8-J ‹ 5 L chains for 7–8 generations. Splenic hybrid-omas were prepared from individual diseased, 7-month-old female NZB/NZW mice with or without prior stimula-tion with LPS in vitro (Table 1). LPS activation is requiredfor obtaining significant numbers of fusion hybrids fromunimmunized mice of non-autoimmune strains or fromyoung, prediseased NZB/NZW mice [12]. However, dis-eased mice from lupus-prone strains readily yield fusablesplenic B cells, probably due to their spontaneously acti-vated state [12, 14]. Conversely, following LPS stimula-tion for 72 h in vitro, high-affinity anti-DNA hybridomasare only rarely recovered from splenic B cells of diseasedNZB/NZW mice [12], either because these cells are rela-tively rare in the non-activated B cell compartment orbecause LPS stimulation of activated, mature B cells issuppressed [15, 16]. Alternatively, splenic B cells that areactivated by extensive cross-linking of membrane IgMare induced to die by apoptosis, and LPS does not inter-fere significantly with this process [17]. Hybridomaclones derived from LPS-activated B cells produced IgMantibodies almost exclusively, while significant numbersof hybridomas that were apparently derived from spon-taneously activated B cells in the diseased animalsecreted IgG antibodies (Table 1).

Receptor editing by secondary rearrangements ofupstream V ‹ gene segments to downstream J ‹ elementswas investigated by sequencing of the expressed Lchain mRNA in the H/L chain- transgenic hybridomasand by the relative number of peripheral B cells express-ing lambda L chains (Fig. 1). Analysis of J ‹ usage inhybridomas derived from LPS-stimulated B cells showsthat kappa L chains from a non-transgenic, 7 month-oldfemale NZB/NZW F1 mouse have a similar J ‹ distribu-tion to that of a non-autoimmune (C57BL/6 ‰ BALB/c) F1mouse (Fig. 1A). In both strains, J ‹ 1+J ‹ 2 are utilized in70-80% of the expressed L chains. In contrast, H-chain-only, transgenic D42H NZB/NZW F1 mouse hybridomasexhibit a dramatic shift in J ‹ usage of endogenous Lchains, where J ‹ 4+J ‹ 5 are now utilized in the over-whelming majority ( ˚ 85%) of kappa L chains. This is asimilar distribution to that of D42H chain-transgenic,non-autoimmune (C57BL/6×BALB/c) F1 mice [11]. It

2470 N. Yachimovich-Cohen et al. Eur. J. Immunol. 2003. 33: 2469–2478

Table 1. Properties of hybridoma clones derived from H/L double-transgenic NZB/NZW F1 mice

a) Fusion of splenocytes was performed on day 3 after 40 ? g/ml LPS activationb) Specificity for DNA was determined by ELISA. Optical density values of 10-times greater than background or higher were

considered positive.c) All clones in this category express IgM isotype, except one IgG hybridoma generated from the spontaneous (non-LPS) D42H/

V ‹ 8-J ‹ 5 fusion experiment.d) All clones in this category express IgG isotypes.

Fig. 1. Light chain editing in H/L chain, double-transgenic NZB/NZW mice. (A) J ‹ usage in LPS hybridomas was determined bydirect sequencing of hybridoma RNA following RT-PCR amplification. Empty bars represent expression of the original V ‹ -J ‹ -transgenic L chain. The black bar on top of an empty bar in hybridomas derived from D42H/V ‹ 8-J ‹ 5 NZB/NZW mice representsedited (non-V ‹ 8) L chains rearranged to J ‹ 5 on the endogenous allele. (B) Lambda L chain-expressing B cells from circulatingblood were analyzed by FACS [13]. *p X 0.01, by Student’s t test.

indicates that NZB/NZW L chains are subject to exten-sive secondary rearrangements and that BCR editing isnot abrogated in diseased NZB/NZW mice.

Essentially the same conclusion was reached by analyz-ing the expressed kappa L chains from LPS hybridomas,obtained from the different Ig H/L double-transgenicNZB/NZW mice (Fig. 1A and Table 1). Practically allof the D42/V ‹ 1-J ‹ 1 B cells were shown to edit theirkappa L chains by secondary rearrangements to down-stream J ‹ minigenes, particularly to J ‹ 5 (85%). Theseextensive secondary rearrangements are similar to thoseseen in non-autoimmune H/L double-transgenic (C57BL/

6 ‰ BALB/c) F1 mice (90% editing, 60% to J ‹ 5) [13], andagain indicate that BCR editing is essentially intact indiseased NZB/NZW mice. The distribution of V ‹ -geneeditors in edited B cells from D42/V ‹ 1-J ‹ 1-transgenicNZB/NZW mice was somewhat different from that ofnon-autoimmune mice. While the secondary rearrange-ments in normal mice were not biased towards any par-ticular V ‹ segment [13], there was an over-represen-tation of one V ‹ gene, V ‹ at4 [18], in edited B cells oftransgenic NZB/NZW mice (23/34, 67%). D42/V ‹ at4 is apolyreactive IgM antibody that binds DNA with low affin-ity, as well as phosphatidylcholine, phosphatidylserineand actin (data not shown).

Eur. J. Immunol. 2003. 33: 2469–2478 B cell receptor editing and anergy in autoimmune anti-DNA knock-in mice 2471

Like their non-autoimmune counterparts [13], the over-whelming majority (90%) of NZB/NZW double-transgenic B cells expressing D42H/V ‹ 8-J ‹ 5 L chainretained their original H/L combination (Fig. 1A andTable 1), possibly due to the unavailability of J ‹ seg-ments for secondary rearrangements on the active Lchain allele, or alternatively, due to the accelerated matu-ration of immature B cells with a productive J ‹ 5 rear-rangement [13]. In transgenic NZB/NZW mice express-ing D42H/V ‹ 4 BCR, the level of L chain editing was inter-mediate between the D42/V ‹ 1 and D42/V ‹ 8 knock-inmice. Of the hybridomas originally expressing V ‹ 4-J ‹ 4,55% retained their H/L chain combination; the restedited their L chains exclusively to J ‹ 5 (Fig. 1A).

Consistent with the results of L chain secondary rear-rangements, the fraction of LPS hybridoma clonessecreting anti-DNA autoantibodies correlated inverselywith the efficiency of L chain editing (Table 1), indicatingthat many of the edited B cells have changed their anti-DNA specificity to a non-autoreactive phenotype. Thissuggests that the editing process represents an attemptby the autoimmune mice to establish immunological tol-erance. Anti-DNA activity of antibodies secreted by thenon-edited LPS hybridomas was of low affinity (data notshown), consistent with the measured DNA affinities ofthe three transgenic H/L combinations [13]. The degreeof BCR editing in the three H/L double-transgenic mouselines is also reflected in the number of B cell clones thathave inactivated their transgenic D42 heavy chain andswitched to endogenous H chain expression (Table 1)[13]. Thus, the fraction of clones which have retained thetransgenic D42H chain was higher in spontaneous andLPS-induced hybridomas derived from V ‹ 8-J ‹ 5-transgenic mice than in those obtained from V ‹ 1-J ‹ 1and V ‹ 4-J ‹ 4 mice. This probably reflects the level ofRAG expression in bone marrow B cells of the threeknock-in mouse strains [13]. Some of the hybridomasthat switched to the endogenous (non-transgenic) H-chain secreted DNA-binding immunoglobulins ( ˚ 10% inLPS hybridomas and 20–50% in spontaneous hybrid-omas, data not shown).

A direct evaluation of BCR editing in unmanipulatedmouse lymphocytes, derived from the three double-transgenic mouse strains, was carried out by FACS anal-ysis of peripheral blood B cells expressing lambdachains. These lambda chains must be the result of sec-ondary rearrangement and activation of lambda locibecause the transgenic L chains belong to the kappaisotype. As shown in Fig. 1B, the level of lambda chainexpression is highest in targeted D42H/V ‹ 1-J ‹ 1 NZB/NZW mice and is similar to that of H-chain-only D42HNZB/NZW mice, suggesting a similar degree of L chainediting in these two transgenic strains. Lambda chain

expression was lowest in D42/V ‹ 8-J ‹ 5 and intermediatein D42/V ‹ 4-J ‹ 4, as would be predicted from the J ‹ utili-zation analysis of splenic LPS hybridomas derived fromthese mice.

The picture emerging from the analysis of presumablyactivated, spontaneous hybridomas derived from H/Ldouble-transgenic NZB/NZW mice is different from thatof the LPS-activated hybridomas in three majorrespects: (i) the higher abundance of non-edited, trans-genic V ‹ 1-J ‹ 1, and V ‹ 4-J ‹ 4 L chains. (ii) the switch of afraction of IgM-producing B cells to IgG producers, and(iii) the selection of the unique V ‹ RF-J ‹ 5 rearrangementfor the expressed D42 L chain (Table 1 and Fig. 2A). Wehave previously shown that the association of D42H withthis L chain results in a particularly high-affinity anti-DNAantibody, which gives a homogeneous nuclear stainingin the ANA test and is typical of active SLE [12, 19]. Asshown in Fig. 2A, D42H/V ‹ 1-J ‹ 1 double-transgenicNZB/NZW mice generated the highest number of D42H/D42L anti-DNA-expressing clones (all of which were IgGsecretors and all were expressed from the targetedallele, since the endogenous kappa alleles remainedunrearranged, data not shown), while D42H/V ‹ 4-J ‹ 4-transgenic mice produced fewer B cells with this combi-nation, and D42H/V ‹ 8-J ‹ 5 mice produced none. Thesefindings closely parallel the efficiency of BCR editing inthese three mouse strains (Fig. 1). Consequently, thegeneration of B cells producing high-affinity anti-DNAantibodies from low-affinity precursors seems to bedependent largely on the efficiency of L chain receptorediting in these mice.

2.2 Affinity maturation of anti-DNA antibodiesderived from edited NZB/NZW B cells

Sequence analysis of expressed H and L chains derivedfrom “spontaneous” (non-LPS-stimulated) NZB/NZWhybridomas has shown (Table 2) that somatic mutationwas closely associated with the production of IgG anti-DNA antibodies. This suggests that immunoglobulinclass-switch and somatic hypermutation are closelylinked processes in this autoimmune response. Addition-ally, all D42H/V ‹ RF-J ‹ 5 IgG antibodies comprising theedited, high-affinity anti-DNA B cell products were foundto be somatically mutated. In contrast, and unlike the sit-uation in MRL/lpr mice [14], spontaneously activated IgMhybridomas, including the non-editing D42H/V ‹ 8-J ‹ 5IgM immunoglobulins, were mostly unmutated (Table 2).

2472 N. Yachimovich-Cohen et al. Eur. J. Immunol. 2003. 33: 2469–2478

Fig. 2. Frequency of high-affinity anti-DNA hybridomas andserum anti-DNA binding activity in transgenic NZB/NZWmice. (A) Frequency of V ‹ RF-J ‹ 5-expressing hybridomas asdetermined by direct sequencing of hybridoma RNA follow-ing RT-PCR amplification. (B) Serum anti-DNA bindingcapacity in individual wild-type and transgenic NZB/NZWmice, determined by the nitrocellulose filter assay. The com-bined DNA binding capacity of all isotypes is measured inthis solution assay [11–13]. The serum anti-DNA bindingdata of non-transgenic (C57BL/6×BALB/c) F1 and of D42H/V ‹ 8-J ‹ 5 (C57BL/6 ‰ BALB/c) F1 Tg mice are reproducedfrom [13]. The DNA binding capacity of IgM (C) and IgG (D)NZB/NZW serum antibodies was measured in ELISA, rela-tive to the corresponding IgM and IgG DNA binding of nor-mal mouse serum. *p X 0.001, by Student’s t test.

2.3 Serum anti-DNA antibodies in H/L chaindouble-transgenic NZB/NZW F1 mice

The generation of high-affinity anti-DNA antibodies byBCR editing in the three transgenic NZB/NZW strainshas a direct influence on the DNA binding capacity of

serum antibodies in these mice. Fig. 2B shows that theaverage DNA binding capacity of D42H/V ‹ 1-J ‹ 1-transgenic NZB/NZW mice is almost as high as that ofD42 H chain-only mice, in which anti-DNA activity wasassociated with the D42H/D42L chain combinationalmost exclusively [12]. D42H/V ‹ 4-J ‹ 4-transgenic micehave only slightly lower serum DNA binding, while theaverage DNA binding capacity of D42H/V ‹ 8-J ‹ 5 sera ismore than an order of magnitude lower than that of V ‹ 1-J ‹ 1 mice, and is similar to DNA binding of non-transgenic NZB/NZW mice, in which only a fraction ofthe B cell population secretes anti-DNA Ab. It appears,therefore, that the efficiency of B cell receptor editingcorrelates directly with the production of high-affinityanti-DNA autoantibodies in the serum of double-transgenic NZB/NZW F1 mice. The difference in serumDNA binding capacity between D42H/V ‹ 1-J ‹ 1 andD42H/V ‹ 8-J ‹ 5–transgenic NZB/NZW mice was morepronounced for IgG (30-fold, Fig. 2D) than for IgM (9-fold, Fig. 2C) anti-DNA antibodies. This again correlateswell with the much greater selection of edited D42H/V ‹ 1-J ‹ 1-expressing B cells for class switching and affin-ity maturation.

2.4 Light chain secondary rearrangements indiseased NZB/NZW F1 mice result fromtolerance-induced receptor editing

The finding that secondary L chain rearrangements in Bcells of autoimmune mice can lead to the production ofhigh-affinity, somatically mutated anti-DNA autoanti-bodies raises the question of receptor editing versusreceptor revision as the underlying mechanism of thesesecondary rearrangements. It has been suggested that Lchain replacement may not be restricted to immature Bcells in the bone marrow, but may also occur in germinalcenters (GC) of peripheral organs, where somatic hyper-mutation takes place and mature B cells with high affinityfor antigen are selected. This receptor “revision” hasbeen reported to be accompanied by re-expression ofRAG proteins in mature B cells [20–22]. However, laterwork has suggested that RAG expression in the periph-ery was largely restricted to immature B cells [23, 24]. Todistinguish between these possibilities, purified popula-tions of bone marrow cells, containing “pre-B” andimmature B cells (pre-B cells may be phenotypicallyindistinguishable from BCR-editing immature B cells[13]) as well as splenic mature B cells were isolated fromD42H/V ‹ 1-J ‹ 1 non-autoimmune and NZB/NZW-transgenic mice. The purified cellular fractions were ana-lyzed for RNA expression levels of RAG-2 and V ‹ RF bysemi-quantitative RT-PCR. As shown in Fig. 3A, RAG-2expression is somewhat higher in bone marrow B cells ofdiseased transgenic NZB/NZW mice compared with nor-

Eur. J. Immunol. 2003. 33: 2469–2478 B cell receptor editing and anergy in autoimmune anti-DNA knock-in mice 2473

Table 2. Mutational analysis of “spontaneous” B cell hybridomas from transgenic NZB/NZW mice

a) No hybridomas were found in this category.

mal mice. However, RAG-2 gene expression was unde-tectable in peripheral mature B cells of non-autoimmuneas well as of NZB/NZW-transgenic mice. This suggeststhat receptor editing in immature B cells, but not recep-tor revision in peripheral mature B cells, is likely to beresponsible for the formation of high-affinity anti-DNA,D42H/D42L Ig combinations in diseased NZB/NZWmice. Consistent with this interpretation is the findingthat V ‹ RF gene expression was found to be sevenfoldhigher in the bone marrow immature B cell population ofD42H/V ‹ 1 NZB/NZW mice than in immature B cells fromnon-autoimmune mice (data not shown). This indicatesthat B cells carrying edited, autoreactive receptors withhigh affinity for DNA are already present in the bone mar-row of the diseased animal at the immature stage.

Finally, we have been able to analyze one edited D42H/D42L IgG hybridoma clone (#38), in which the replacedV ‹ 1-J ‹ 1 gene has been retained in the cellular genomicDNA. Since the endogenous kappa allele was found tobe in the germ-line configuration (data not shown), reten-tion of the V ‹ 1-J ‹ 1 on the targeted allele was probably arare event. This is because the V ‹ RF gene is in the sameorientation as the J ‹ gene cluster [18] and should replacethe targeted V ‹ 1-J ‹ 1 gene by a deletional rearrange-ment mechanism. The replaced V ‹ 1-J ‹ 1 gene can beretained in the cellular genomic DNA, only, if the V ‹ RFrearrangement was preceded by another editing step, inwhich a putative V ‹ gene located upstream of V ‹ RF inthe V ‹ gene cluster had replaced the targeted V ‹ 1-J ‹ 1gene by an inversional rearrangement event (Fig. 3B).Sequence analysis of these two L chains in clone #38shows that the replaced V ‹ 1-J ‹ 1 gene is unmutated,while the expressed V ‹ RF-J ‹ 5 gene has several muta-tions (Fig. 3C). This finding is consistent with the notion[25] that the editing step took place before the activatedB cells entered the GC reaction and should be regarded

as tolerance-induced receptor editing rather thanactivation-induced receptor revision.

2.5 B cell clonal anergy is abrogated in diseasedNZB/NZW mice

On a non-autoimmune genetic background, autoreactiveD42H/V ‹ 8-J ‹ 5 B cells that do not edit their BCR are con-trolled by clonal anergy and do not secrete anti-DNAautoantibodies [13]. To test whether this tolerance mech-anism is effective in diseased female NZB/NZW mice,BCR density [13, 26] and B cell proliferation in responseto LPS activation [13, 27] were examined in non-editingD42/V ‹ 8-J ‹ 5-transgenic mice. As shown in Fig. 4A, theBCR density of splenic B cells from NZB/NZW-transgenic mice was intermediate between the BCRdensities of non-anergized, wild-type C57BL/6 mouse Bcells, and fully anergized D42H/V ‹ 8-J ‹ 5-transgenic Bcells from non-autoimmune (C57BL/6 ‰ BALB/c) F1mice. This suggests that B cell clonal anergy is inefficientin diseased NZB/NZW mice. Furthermore, the extent ofLPS-induced proliferation of D42/V ‹ 8-J ‹ 5-transgenic Bcells from NZB/NZW mice was similar to that of wild-type C57BL/6 cells and different from anergized (C57BL/6×BALB/c) F1-transgenic B cells (Fig. 4B). This also indi-cates that B cell clonal anergy in diseased NZB/NZWmice is abrogated. Taken together, the absence of Lchain receptor editing in D42/V ‹ 8-J ‹ 5 B cells, combinedwith at least partially non-functional B cell anergy may bereflected in the anti-DNA serum titers (Fig. 2B) of D42/V ‹ 8-J ‹ 5 NZB/NZW mice (3.0±4.0 ng/ml) that are con-siderably higher than those of non-autoimmune double-transgenic mice (0.6±0.6 ng/ml), but are much lowerthan the anti-DNA serum titers of editing D42/V ‹ 1-J ‹ 1NZB/NZW mice (17.2±5.8 ng/ml, Fig. 2B).

2474 N. Yachimovich-Cohen et al. Eur. J. Immunol. 2003. 33: 2469–2478

Fig. 3. BCR editing vs. BCR revision in D42H/V ‹ 1-J ‹ 1-transgenic mice. (A) RAG-2 expression in purified populations of bonemarrow immature and splenic mature B cells from 7-month-old, female D42H/V ‹ 1-J ‹ 1 NZB/NZW and normal mice. For detailsof B cell purification and measurement of RAG expression see Sect. 4 and [13]. (B) A suggested scheme for a two-step L chainediting on the targeted allele, leading to expression of V ‹ RF and retention of V ‹ 1-J ‹ 1 in genomic DNA. (C) Nucleotide sequencesof expressed V ‹ RF and non-expressed V ‹ 1-J ‹ 1 in clone #38 hybridoma, derived from a 7-month-old, female D42H/V ‹ 1-J ‹ 1NZB/NZW Tg mouse. Dashes indicate nucleotide identity with residues in the top line. Somatic mutations are given in bold lettersand amino acid changes are indicated above the germ-line sequence. V ‹ RF is a one-membered gene family situated in the 5’-portion of the kappa locus [18].

Eur. J. Immunol. 2003. 33: 2469–2478 B cell receptor editing and anergy in autoimmune anti-DNA knock-in mice 2475

Fig. 4. B cell anergy in D42H/V ‹ 8-J ‹ 5 double-transgenicmice. (A) Splenic (B220+) B cells were measured for BCRdensity, using anti-mouse IgM antibody in flow cytometricanalysis. (B) Splenic (B220+) B cells were studied for LPS-induced proliferation, as analyzed by CFSE dilution. LPSstimulation of cultured splenocytes was carried out for5 days. The MFI of IgM density (A) and of dividing cells andtheir percentage of total transgenic B cells (B) are desig-nated. The results represent several experiments in whichthe same relative MFI were obtained.

3 Discussion

In this report, we present evidence that a protectivemechanism of self-immunological tolerance such as Bcell receptor editing may actually be deleterious to theanimal, if the delicate balance between B cell toleranceand activation is disrupted, as may be the case in lupus-prone mouse strains and humans with SLE. Experimentswith the MRL/lpr mouse model of SLE [14, 28–30] havesuggested that the loss of immunological tolerance inautoimmune mice may be antigen-specific. Brard et al.[14] have studied the breakdown of B cell tolerance toDNA in MRL/lpr knock-in mice that express one of thetargeted L chains (V ‹ 8-J ‹ 5) used in this study, in combi-nation with a different anti-DNA H chain (3H9). Interest-ingly, the majority of spontaneous hybridomas express-ing the V ‹ 8-J ‹ 5 targeted L chain retained their original Hand L chain transgenes in both MRL/lpr [14] and NZB/NZW (this study) mice. However, the acquisition ofdsDNA and antinuclear activity in the MRL/lpr miceresulted principally from extensive somatic mutation inboth the H and L chain V-region genes. In contrast, mostof the V ‹ 8-J ‹ 5-targeted NZB/NZW mouse cells in thisstudy were not subjected to significant somatic mutationon either H or L chains, and their low-affinity anti-DNAactivity was due to the loss of B cell anergy in theselupus-prone mice. It is not clear whether this differencebetween the two studies results from the different trans-

genic H chains, which in combination with V ‹ 8-J ‹ 5 Lchain have slightly different fine specificity for DNA(ssDNA for 3H9/V ‹ 8, [14] and ssDNA+dsDNA for D42H/V ‹ 8, [13]). Alternatively, different modes of B cell activa-tion are operating in lupus-prone MRL/lpr and NZB/NZWmice [2]. Sekiguchi et al. [31] have used the anti-DNA3H9-transgenic H chain in a chronic graft-versus-hostmouse model. Their genetic analysis showed evidencefor secondary rearrangements of the L chain, associatedwith anti-dsDNA reactivity. However, a direct proof forconversion of non-autoreactive to anti-DNA-producing Bcells was lacking, because the original, unedited L chaincomponent was not known.

Strikingly, the major force behind the generation of high-affinity anti-DNA antibodies in our H/L chain double-transgenic NZB/NZW mice was receptor editing itself.The efficiency of secondary rearrangement in the differ-ent transgenic mouse strains was similar in the non-autoimmune and lupus-prone mice and was dependenton the rearrangement status of the transgenic L chain.However, while receptor editing in non-autoimmunemice has led to the complete elimination of anti-DNAspecificity of the edited clones, this same mechanism onthe lupus genetic background, was responsible for thegeneration of high-affinity autoreactive B cell clones. Asshown in Table 1, most of the V ‹ 1-J ‹ 1 L chains changetheir specificity and are presumably non-activated.Some of the autoreactive cells apparently escape editingand are recruited to the activated B cell compartment,which is represented by the “spontaneous” hybridomas.In the healthy animal, these clones would be subject toalternative mechanisms of tolerance such as clonal dele-tion and anergy [11, 13]. However, as shown in this study,these mechanisms are abrogated in NZB/NZW autoim-mune mice. Consequently, the activated B cell compart-ment in the diseased Tg mouse includes two types ofautoreactive B cells, namely, unedited low-affinity anti-DNA, and edited high-affinity anti-DNA-producing Bcells. While the majority of the low-affinity, activated cellssecrete IgM autoantibodies that have relatively fewsomatic mutations, all of the high-affinity (V ‹ RF contain-ing) anti-DNA B cells switch to the IgG isotype andundergo extensive affinity maturation (Table 2).

Our analysis of RAG expression, V ‹ RF expression andsomatic mutation suggests that the high-affinity D42H/D42L anti-DNA-expressing B cells were generated at theimmature B cell stage in the bone marrow as part of Bcell receptor editing and not in the process of mature Bcell receptor revision in the periphery [10, 20, 22]. Thisinitial analysis of bone marrow immature and splenicmature B cells does not exclude the possibility thatimmature B cells which immigrate to peripheral tissues,possibly as transitional B lymphocyte subsets are

2476 N. Yachimovich-Cohen et al. Eur. J. Immunol. 2003. 33: 2469–2478

induced to edit their receptors and provide a new BCRrepertoire, including self-reactive specificities [32]. Theappearance of highly autoreactive B cells in the periph-ery of NZB/NZW mice probably results from a function-ing editing mechanism and a concomitant loss of otherprotective tolerance mechanisms. Since the underlying,disease-related susceptibility genes in New Zealandmice are only partly characterized, it is not possible tosay whether these genes are responsible for functionalabnormalities in specific tolerance mechanisms. Alterna-tively, they may induce a general B cell hyperactivity [6]or a relative resistance of autoreactive B cells to apopto-sis [33], which tip the balance between tolerance andactivation.

Selection for high-affinity autoreactive B cells is clearly amajor disease-related event, because the edited D42H/D42L-expressing B cells comprise the majority of clonesselected for class-switch and affinity maturation in thetargeted NZB/NZW mice. We propose that this BCRediting mechanism is physiologically significant in thepathogenesis of SLE, because it may greatly increasethe fraction of high-affinity, autoreactive B cell clonesand the level of anti-DNA autoantibodies in the circula-tion of individuals with SLE. This is particularly relevantwhere the availability of a critical component of an auto-reactive antibody, such as the V ‹ RF-J ‹ 5 gene product,is entirely dependent on the precise rearrangement of arelatively rare immunoglobulin gene (e.g. V ‹ RF) to a par-ticular J ‹ segment. In this case, the likelihood of obtain-ing the desired product should increase several fold bysecondary rearrangements.

4 Materials and methods

4.1 Knock-in mice

NZB and NZW mice were purchased from Harlan (Oxon, UK)and were bred at the animal facility of the Hebrew UniversityMedical School (Jerusalem, Israel). Knock-in mice, trans-genic for D42H, V ‹ 1-J ‹ 1, V ‹ 4-J ‹ 4 or V ‹ 8-J ‹ 5 were con-structed as described previously [11, 13]. The D42H trans-gene was backcrossed onto the NZB background, and eachof the targeted L chains was backcrossed onto the NZWbackground for 7–8 generations. This was followed by a sin-gle cross between an H chain-transgenic NZB female andan L chain-transgenic NZW male to obtain the lupus-proneNZB/NZW F1 hybrid. Seven month-old female NZB/NZW F1mice were used in all the experiments described in thisstudy.

4.2 Isolation of B cell subsets

Bone marrow 493+ cells containing mostly pre-B and imma-ture B cells, but not recirculating mature B cells, were iso-

lated by labeling the cells with biotinylated mAb 493 [34] andseparation with streptavidin magnetic beads (Miltentyi Bio-tec, Bergisch Gladbach, Germany). Splenic CD43–, 493–

mature B cells were isolated by depletion of T cells with anti-CD43 magnetic beads (Miltenyi Biotec), followed by deple-tion of immature B cells with biotinylated mAb 493 andstreptavidin magnetic beads. The purity of the two B cellfractions was over 90% as analyzed by FACS.

4.3 Measurement of RAG-2 and V ‹ RF gene expression

Semi-quantitative measurements of RAG-2 expressionusing RAG-2 and control CD19 cDNA have been describedpreviously [13]. A similar procedure was used for mea-surement of V ‹ RF gene expression using cDNA randompriming and PCR amplification (15 cycles) with the forwardprimer 5’-GCTCTAGAATGAGGTTCCAGGTTCAGGTTCTG-GGGCTC-3’ and the reverse primer 5’-CCATCGATCGG-GTATTCATTATGCTGTTGACAGTAATAC-3’. The V ‹ RF PCRproduct was ethanol-precipitated and digested at theunique EcoRI site. The resulting fragments were separatedby agarose gel electrophoresis, transferred onto a nylonmembrane and hybridized with a V ‹ RF-specific DNA probe(obtained by PCR amplification of V ‹ RF cDNA with the sameoligonucleotide primers).

4.4 Other experimental procedures

Flow cytometric analysis, hybridoma production, directsequence determinations, Ab binding tests and measure-ment of RAG-2 expression were carried out as describedpreviously [12, 13]. For analysis of somatic mutations inexpressed Ig chains, DNA prepared by RT-PCR was A-Tcloned into pGEM-T Easy (Promega, Madison, WI, USA) andsequenced by an automatic sequencer. Analysis of germ-line configuration in untargeted, endogenous kappa alleleswas carried out by PCR, using specific primers upstreamand downstream of genomic J ‹ 1 [13]. The apparent affinityof the various Ig ? ‹ chain combinations for dsDNA wasmeasured by the nitrocellulose filter assay and calculatedfrom Scatchard plots [19].

Acknowledgments: This study was supported by the IsraelScience Foundation administered by the Israeli Academy ofScience and Humanities, and by the German-Israeli Foun-dation for Scientific Research and Development (GIF). Wethank Dr. Martin Weigert for providing the non-autoimmuneL chain-targeted mice and Dr. Eline Luning Prak for her criti-cal reading of the manuscript and very helpful discussions.

References

1 Hahn, B., Animal models of systemic lupus erythematosus. InWallace, D. and Hahn, B. (Eds.) Dubois’ Lupus Erythematosus.Lea & Febigen, Philadelphia. 1993. pp. 157.

Eur. J. Immunol. 2003. 33: 2469–2478 B cell receptor editing and anergy in autoimmune anti-DNA knock-in mice 2477

2 Theofilopoulos, A. N. and Dixon, F. J., Murine models of sys-temic lupus erythematosus. Adv. Immunol. 1985. 37: 269–390.

3 Wakeland, E. K., Liu, K., Graham, R. R. and Behrens, T. W.,Delineating the genetic basis of systemic lupus erythematosus.Immunity 2001. 15: 397–408.

4 Izui, S., McConahey, P. J. and Dixon, F. J., Increased spontane-ous polyclonal activation of B lymphocytes in mice with sponta-neous autoimmune disease. J. Immunol. 1978. 121: 2213–2219.

5 Papoian, R., Pillarisetty, R. and Talal, N., Immunological regula-tion of spontaneous antibodies to DNA and RNA. II. Sequentialswitch from IgM to IgG in NZB/NZW F1 mice. Immunology 1977.32: 75–79.

6 Lipsky, P. E., Systemic lupus erythematosus: an autoimmunedisease of B cell hyperactivity. Nat. Immunol. 2001. 2: 764–766.

7 Reininger, L., Winkler, T. H., Kalberer, C. P., Jourdan, M., Mel-chers, F. and Rolink, A. G., Intrinsic B cell defects in NZB andNZW mice contribute to systemic lupus erythematosus in(NZB×NZW)F1 mice. J. Exp. Med. 1996. 184: 853–861.

8 Klinman, N. R., The “clonal selection hypothesis” and currentconcepts of B cell tolerance. Immunity 1996. 5: 189–195.

9 Nossal, G. J., Clonal anergy of B cells: a flexible, reversible, andquantitative concept. J. Exp. Med. 1996. 183: 1953–1956.

10 Nemazee, D. and Weigert, M., Revising B cell receptors. J. Exp.Med. 2000. 191: 1813–1817.

11 Pewzner-Jung, Y., Friedmann, D., Sonoda, E., Jung, S.,Rajewsky, K. and Eilat, D., B cell deletion, anergy, and receptorediting in “knock-in” mice targeted with a germline-encoded orsomatically mutated anti-DNA heavy chain. J. Immunol. 1998.161: 4634–4645.

12 Friedmann, D., Yachimovich, N., Mostoslavsky, G., Pewzner-Jung, Y., Ben-Yehuda, A., Rajewsky, K. and Eilat, D., Produc-tion of high-affinity autoantibodies in autoimmune New ZealandBlack/New Zealand white F1 mice targeted with an anti-DNAheavy chain. J. Immunol. 1999. 162: 4406–4416.

13 Yachimovich, N., Mostoslavsky, G., Yarkoni, Y., Verbovetski, I.and Eilat, D., The efficiency of B cell receptor (BCR) editing isdependent on BCR light chain rearrangement status. Eur. J.Immunol. 2002. 32: 1164–1174.

14 Brard, F., Shannon, M., Luning Prak, E., Litwin, S. and Wei-gert, M., Somatic mutation and light chain rearrangement gener-ate autoimmunity in anti-single-stranded DNA transgenic MRL/lpr mice. J. Exp. Med. 1999. 190: 691–704.

15 Kearney, J. F., Cooper, M. D. and Lawton, A. R., B lymphocytedifferentiation induced by lipopolysaccharide. III. Suppression ofB cell maturation by anti-mouse immunoglobulin antibodies. J.Immunol. 1976. 116: 1664–1668.

16 Andersson, J., Bullock, W. W. and Melchers, F., Inhibition ofmitogenic stimulation of mouse lymphocytes by anti-mouseimmunoglobulin antibodies. I. Mode of action. Eur. J. Immunol.1974. 4: 715–722.

17 Scott, D. W., Lamers, M., Köhler, G., Sidman, C. L., Maddox,B. and Carsetti, R., Role of c-myc and CD45 in spontaneous andanti-receptor-induced apoptosis in adult murine B cells. Int.Immunol. 1996. 8: 1375–1385.

18 Thiebe, R., Schäble, K. F., Bensch, A., Brensing-Küppers, J.,Heim, V., Kirschbaum, T., Mitlöhner, H., Ohnrich, M., Pourra-jabi, S., Röschenthaler, F., Schwendinger, J., Wichelhaus, D.,Zocher, I. and Zachau, H. G., The variable genes and gene fami-lies of the mouse immunoglobulin kappa locus. Eur. J. Immunol.1999. 29: 2072–2081.

19 Pewzner-Jung, Y., Simon, T. and Eilat, D., Structural elementscontrolling anti-DNA antibody affinity and their relationship to anti-phosphorylcholine activity. J. Immunol. 1996. 156: 3065–3073.

20 Hikida, M. and Ohmori, H., Rearrangement of lambda light chaingenes in mature B cells in vitro and in vivo. Function of reexpre-ssed recombination-activating gene (RAG) products. J. Exp.Med. 1998. 187: 795–799.

21 Han, S., Zheng, B., Schatz, D. G., Spanopoulou, E. and Kel-soe, G., Neoteny in lymphocytes: Rag1 and Rag2 expression ingerminal center B cells. Science 1996. 274: 2094–2097.

22 Papavasiliou, F., Casellas, R., Suh, H., Qin, X. F., Besmer, E.,Pelanda, R., Nemazee, D., Rajewsky, K. and Nussenzweig, M.C., V(D)J recombination in mature B cells: a mechanism for alter-ing antibody responses. Science 1997. 278: 298–301.

23 Yu, W., Nagaoka, H., Jankovic, M., Misulovin, Z., Suh, H.,Rolink, A., Melchers, F., Meffre, E. and Nussenzweig, M. C.,Continued RAG expression in late stages of B cell developmentand no apparent re-induction after immunization. Nature 1999.400: 682–687.

24 Monroe, R. J., Seidl, K. J., Gaertner, F., Han, S., Chen, F., Seki-guchi, J., Wang, J., Ferrini, R., Davidson, L., Kelsoe, G. andAlt, F. W., RAG2:GFP knock-in mice reveal novel aspects ofRAG2 expression in primary and peripheral lymphoid tissues.Immunity 1999. 11: 201–212.

25 Goossens, T., Brauninger, A., Klein, U., Küppers, R. andRajewsky, K., Receptor revision plays no major role in shapingthe receptor repertoire of human memory B cells after the onsetof somatic hypermutation. Eur. J. Immunol. 2001. 31: 3638–3648.

26 Nguyen, K. A., Mandik, L., Bui, A., Kavaler, J., Norvell, A.,Monroe, J. G., Roark, J. H. and Erikson, J., Characterization ofanti-single-stranded DNA B cells in a non-autoimmune back-ground. J. Immunol. 1997. 159: 2633–2644.

27 Lyons, A. B. and Parish, C. R., Determination of lymphocytedivision by flow cytometry. J. Immunol. Methods 1994. 171:131–137.

28 Rathmell, J. C. and Goodnow, C. C., Effects of the lpr mutationon elimination and inactivation of self-reactive B cells. J. Immu-nol. 1994. 153: 2831–2842.

29 Kench, J. A., Russell, D. M. and Nemazee, D., Efficient periph-eral clonal elimination of B lymphocytes in MRL/lpr mice bearingautoantibody transgenes. J. Exp. Med. 1998. 188: 909–917.

30 Roark, J. H., Kuntz, C. L., Nguyen, K. A., Caton, A. J. and Erik-son, J., Breakdown of B cell tolerance in a mouse model of sys-temic lupus erythematosus. J. Exp. Med. 1995. 181: 1157–1167.

31 Sekiguchi, D. R., Jainandunsing, S. M., Fields, M. L., Maldo-nado, M. A., Madaio, M. P., Erikson, J., Weigert, M. and Eisen-berg, R. A., Chronic graft-versus-host in Ig knock-in transgenicmice abrogates B cell tolerance in anti-double-stranded DNA Bcells. J. Immunol. 2002. 168: 4142–4153.

32 Magari, M., Sawatari, T., Kawano, Y., Cascalho, M., Wabl, M.,Kanayama, N., Hikida, M. and Ohmori, H., Contribution of lightchain rearrangement in peripheral B cells to the generation ofhigh-affinity antibodies. Eur. J. Immunol. 2002. 32: 957–966.

33 Rozzo, S. J., Allard, J. D., Choubey, D., Vyse, T. J., Izui, S.,Peltz, G. and Kotzin, B. L., Evidence for an interferon-induciblegene, Ifi202, in the susceptibility to systemic lupus. Immunity2001. 15: 435–443.

34 Rolink, A. G., Andersson J. and Melchers, F., Characterizationof immature B cells by a novel monoclonal antibody, by turnoverand by mitogen reactivity. Eur. J. Immunol. 1998. 28: 3738–3748.

Correspondence: Dan Eilat, Department of Medicine,Hadassah University Hospital, P. O. Box 12000, Jerusalem,91120, IsraelFax: +972-2-6437775e-mail: eilatd — md.huji.ac.il

2478 N. Yachimovich-Cohen et al. Eur. J. Immunol. 2003. 33: 2469–2478