A disease-related rheumatoid factor autoantibody is not tolerized in a normal mouse: implications for the origins of autoantibodies in autoimmune disease.

We have analyzed B cell tolerance in a rheumatoid factor (RF) transgenic mouse model. The model is based on AM14, a hybridoma, originally isolated from an autoimmune MRL/lpr mouse that has an affinity and specificity typical of disease-related RFs from this strain. AM14 binds to immunoglobulin (Ig)G2a of the "a" allotype (IgG2aa) and not to IgG2ab. Thus, by crossing the transgenes onto an IgHa (BALB/c) background or to a congenic IgHb (CB.17) background, we could study the RF-expressing B cells when they were self-specific (IgHa) or when they were not self-specific (IgHb). These features make the AM14 model unique in focusing on a true autoantibody specificity while at the same time allowing comparison of autoreactive and nonautoreactive transgenic B cells, as was possible in model autoantibody systems such as anti-hen egg lysozyme. Studies showed that AM14 RF B cells can make primary immune responses and do not downregulate sIgM, indicating that the presence of self-antigen does not induce anergy of these cells. In fact, IgHa AM14 transgenic mice have higher serum levels of transgene-encoded RF than their IgHb counterparts, suggesting that self-antigen-specific activation occurs even in the normal mouse background. Since AM14 B cells made primary responses, we had the opportunity to test for potential blocks to self-reactive cells entering the memory compartment. We did not find evidence of this, as AM14 B cells made secondary immune responses as well. These data demonstrate that a precursor of a disease-specific autoantibody can be present in the preimmune repertoire and functional even to the point of memory cell development of normal mice. Therefore, immunoregulatory mechanisms that normally prevent autoantibody production must exert their effects later in B cell development or through T cell tolerance. Conversely, the data suggest that it is not necessary to break central tolerance, even in an autoimmune mouse, to generate pathologic, disease-associated autoantibodies.

The mechanism of autoantibody production in an autoimmune MRL/lpr mouse.

Rheumatoid factors (RF) and anti-DNA Abs from MRL/lpr mice have features similar to Abs directed toward foreign Ags, indicating a role of specific activation by Ags during disease. But our previous studies and analogous studies from others concentrated on a limited subset of hybridomas selected on the basis of Ag binding to well characterized target autoantigens. Thus, it has been unclear to what extent clonal expansion is restricted to identifiable autospecificities. To obtain a more complete picture of disease-associated autoantibody production, we designed the following experiment. A large number of B cell hybridomas were generated from the spleen of an MRL/lpr mouse and then analyzed for self-specificity, sequence, and clonal relationship. Surprisingly, we found that clonal expansion was limited to only a few autospecificities, implying a unique property of this response. In addition, we used Southern blotting with heavy and L chain constant region probes to screen both RF and non-RF hybridomas for membership in clones, one of which was first identified among RF hybridomas. We found no non-RF members of this clone. The size and number of mutations of this clone were sufficient for us to conclude that nonspecific (i.e., non-RF) mutant members are rapidly lost. Had an Ag other than IgG2a been driving clonal expansion, we should have seen mutants that retained spectificity for that Ag but that lost specificity for IgG2a. This observation, along with the restriction of clonal expansion to a few autospecificities, provides strong evidence that normal autoantigens themselves drive autoantibody clonal expansion.

An isotype switched and somatically mutated rheumatoid factor clone isolated from a MRL-lpr/lpr mouse exhibits limited intraclonal affinity maturation.

Employing site-directed mutagenesis we have reconstructed and expressed the germ-line precursor of an expanded rheumatoid factor (RF) clone. This RF clone, designated clone F, was isolated from an autoimmune MRL/MpJ-lpr/lpr mouse. Most of the clone members were extensively mutated and isotyped-switched. The predominant isotype of clone F was gamma 3. The RF bound specifically to the MRL gamma 2 a allotype (Igh-1j) but not to the B6 gamma 2a allotype (Igh-1b). The germ-line antibody was also found to bind gamma 2a in an RF assay. The affinities of the germ-line RF and representative members of the clone were measured in an ELISA-based equilibrium binding assay. The dissociation constant (Kd) of the germ-line RF was 2.5 x 10(-6) M. All of the expressed clone members had affinities within a two- to sixfold range of the germ line, indicating that the mechanisms of somatic hypermutation and selection resulted in only limited affinity maturation of this autoantibody clone.

A rheumatoid factor transgenic mouse model of autoantibody regulation.

To address whether B cells expressing a disease-associated autospecificity are regulated in normal mice, we have established a rheumatoid factor (RF) transgenic model of autoimmunity, using V genes derived from an IgA anti-IgG2a RF isolated from an autoimmune MRL/lpr mouse. As we wished to study induction of tolerance during B cell development, we cloned the VH gene into an IgM expression vector. The RF we chose binds only IgG2a of the 'a' allotype (IgG2a) but not IgG2ab allowing us to produce transgenic animals on IgHa and IgHb backgrounds, which either express or lack the self-antigen. Two transgenic lines were studied. Using mice which lack the self-antigen, we show by fluorescence activated cell sorting and hybridoma analysis that the H and L transgenes are expressed to the exclusion of endogenous genes in most splenic B cells. In spite of good allelic exclusion, transgenic mice which are genetically capable of expressing IgG2aa have reduced but significant (approximately 50 micrograms/ml) serum levels. Nonetheless, the frequency and numbers of transgene-expressing B cells in peripheral lymphoid organs of such mice which have the self-antigen are similar to those which lack it (IgHb mice). Thus, B cells expressing an anti-self IgG2a surface receptor can develop in this system. Whether such B cells are anergic or otherwise regulated in autoantigen-expressing mice is discussed.

Influence of a V kappa 8 L chain transgene on endogenous rearrangements and the immune response to the HA(Sb) determinant on influenza virus.

A rearranged murine V kappa 8/J kappa 5 L chain gene that codes for the L chain of most antibodies generated in the primary response of BALB/c mice to the antigenic site, Sb, of the hemagglutinin (HA) molecule of influenza virus A/PR/8/34 (PR8) has been cloned. Three transgenic lines were generated by microinjecting the gene. Lines Ga and L each contain a single copy of the transgene whereas line Gb contains three complete copies. Mice of the Ga lineage showed increased V kappa 8-specific mRNA levels only in spleen, but not in nonlymphoid organs and therefore displayed apparently normal lymphoid-specific regulation of the Ig transgene. B cell hybridomas generated from these mice were analyzed for rearrangements of endogenous V kappa genes. Greater than 90% of the C kappa alleles were retained in germ-line configuration in the Ga line, compared with only 0 to 18% in the L line. Thus, a wide variation in the frequency of endogenous rearrangements is seen among mice of different lineages using the same transgene construct. None of more than 150 hybridomas derived from LPS-stimulated splenic B cells of Ga mice exhibited HA-binding activity although they expressed the transgene and, in most cases, excluded endogenous V kappa rearrangements. In contrast, a large fraction of hybridomas isolated after primary immunization with PR8 were HA(Sb)-specific. This indicated that the transgene was functional but formed HA-specific antibodies with a more restricted set of H chains than previously hypothesized. The primary anti-HA response to immunization with PR8 was diminished in all lines compared with normal mice except for a slightly accelerated but transient burst of anti-HA antibody formation in two out of three lines (Ga and Gb). This early response in G lineage mice was largely specific for HA(Sb) and thus appeared to be composed of transgene-expressing antibodies. No differences in serum titers were observed in the secondary anti-HA responses to booster inoculation with PR8 between transgenic and normal mice.

V region gene usage and somatic mutation in the primary and secondary responses to influenza virus hemagglutinin.

Most (80 to 90%) primary antibodies specific for the Sb site of influenza virus (A/PR/8/34) hemagglutinin share an Id (designated C4). Secondary antihemagglutinin(Sb) antibodies also exhibit the C4 Id although less frequently (10 to 15%). We have analyzed the V region nucleotide sequences of primary and secondary antibodies with the C4 Id. Primary C4 antibodies are encoded by the same Vk gene, belonging to the Vk8 group, usually rearranged to Jk5. The H chains are diverse, encoded by VH genes belonging to at least four different VH families, a variety of DH genes, and either JH2, JH3, or JH4. There is only one somatic mutation among seven Vk and two VH genes encoding primary C4 antibodies. Secondary C4 antibodies are also encoded by the same Vk8-Jk5 gene segment and by diverse VH genes. Additional heterogeneity in the secondary response is caused by somatic mutation.

Analysis of VH gene replacement events in a B cell lymphoma.

We have analyzed a series of recombinational events at the IgH chain locus of the B cell lymphoma, NFS-5. Each of these recombinational events results in the replacement of the VH gene segment of the rearranged H chain gene (VhDJh) with that of an upstream germline gene segment. Replacements on the productive and nonproductive alleles have been observed. In each case, the recombination occurs in close proximity to a highly conserved heptameric sequence (5'TACTGTG3') which is located at the 3' end of the VH coding region. In the two examples of recombination on the productive allele that have been analyzed, the initial VHQ52 gene is replaced by different VH7183 genes. On the non-productive allele, sequential replacement events have been analyzed: the initial VHQ52 rearrangement is first replaced by a closely related VHQ52 gene, followed by a second replacement using a VHQ52 pseudogene. Southern blot analysis using VH probes indicates that these recombinations may be accompanied by the deletion of germline VH genes belonging to both the VHQ52 and VH7183 families, suggesting that these gene families are interspersed in the NFS/N mouse.

Interspersion of the VHQ52 and VH7183 gene families in the NFS/N mouse.

Deletion mapping analysis has shown that members of the VH7183 and VHQ52 gene families are interspersed in the NFS/N mouse. To obtain direct evidence that members of these gene families are physically linked, an NFS/N liver library was constructed and genomic clones were analyzed for hybridization to both VHQ52 and VH7183 gene probes. Four clones were identified which contained both VHQ52 and VH7183 hybridizable restriction fragments. Two clones containing rearranged VHQ52 genes were also found to hybridize with the VH7183 gene probe. Sequence analysis of three of the VH7183-containing restriction fragments indicate that all are pseudogenes which contain interruptions at either the 5' and/or 3' ends of the VH coding region. Given the D-proximal location of at least a portion of the VHQ52 gene family relative to VH7183 in NFS/N mice, and the known correlation between D proximity and the frequency of VH gene utilization, 22 NFS/N-derived pre-B cell lines were analyzed for VHQ52 gene utilization. More than 40% of the identified H chain (VHDJH) rearrangements in this survey used members of this gene family. Furthermore, analysis of poly(A)+ RNA from NFS/N fetal liver and adult spleen also indicates preferential utilization of VHQ52 family in fetal liver. Kinetic studies show, however, that there are no changes in relative utilization throughout fetal ontogeny. The implications of these findings for the expression and randomization of the VH repertoire are discussed.

Structure and function of anti-DNA autoantibodies derived from a single autoimmune mouse.

Four monoclonal anti-DNA antibodies derived from a single autoimmune MRL/lpr mouse were studied. Three of these antibodies showed similarities in DNA binding; the fourth had a much higher specific activity for single-stranded DNA and, in addition, was unique in binding double-stranded DNA and cardiolipin. Complete nucleotide sequences of heavy- and light-chain variable regions demonstrated that all four antibodies are clonally related. The sequences also showed numerous somatic mutations, the distribution of which suggests that positive selection by antigen operated on these clonally related autoantibodies.

The role of clonal selection and somatic mutation in autoimmunity.

Polyclonal activation has been proposed as the reason that autoantibodies are produced during autoimmune disease. This model denies a role for specific antigen selection of B cells and predicts instead a multiclonal population of unmutated or randomly mutated autoantibodies. We have found that the genetic features and clonal composition of spontaneously derived immunoglobulin G (IgG) antiself-IgG (rheumatoid factor (RF] autoantibodies derived from the autoimmune MRL/lpr mouse strain are inconsistent with both the predictions of this model and the actual outcome of experimental polyclonal activation. Instead we have found that MRL/lpr RFs are oligoclonal or even monoclonal in origin. They harbour numerous somatic mutations which are distributed in a way that suggests immunoglobulin-receptor-dependent selection of these mutations. In this sense, the MRL/lpr RFs resemble antibodies elicited by exogenous antigens after secondary immunization. The parallels suggest that, like secondary immune responses, antigen stimulation is important in the generation of MRL/lpr RFs.

Variable region sequences of murine IgM anti-IgG monoclonal autoantibodies (rheumatoid factors). A structural explanation for the high frequency of IgM anti-IgG B cells.

The nucleotide sequences of heavy and light chains from 10 monoclonal IgM anti-IgG1 (RF) antibodies were determined and reported here as translated amino acid sequences. Only three families of VK light chains were used in these antibodies: VK1 (two examples), VK8 (three examples), and VK19 (four examples). This represents a significant nonrandom selection of light chains. In contrast, all other variable region gene segments (i.e., VH, DH, JH, and JK) were used in a pattern consistent with random selection from the available pool of germline genes. In two cases, the same anti-IgG1 specificity was generated by a combination of very homologous light chains with unrelated heavy chains. We infer from this that the light chain is the segment used by these antibodies to bind IgG1. The nature of these sequences provides an explanation for the curious observation that as many as 15% of splenic B cells in normal mice may be expressing IgM anti-IgG; if, as our data suggest, certain light chains in combination with many different heavy chains can be used in assembling the anti-IgG specificity, then, because of combinatorial association in which the heavy chain is not relevant for specificity, the fraction of IgM-producing B cells expressing these light chains should approximate the fraction of B cells making IgM anti-IgG. We calculate, based on data presented in several other studies, that 5-17% of B cells express one of the VK types observed in monoclonal RF. This agrees well with estimates for the number of B cells making IgM anti-IgG. In addition, our findings could rule out other explanations of the high percentage of B cells making RF, such as constant stimulation by antigen or presence of numerous antigenic epitopes since it was shown that IgM anti-IgG1 antibodies are not somatically mutated and that they are structurally homogeneous. We aligned the VK sequences of the RF in hopes of finding some primary sequence homology between the represented VK families which might point to residues involved in the binding interaction. Although we found no such homology in the hypervariable regions, we did find significant and unexpected homology in the FR2 and FR3 of these light chains. We noted that these regions are exposed in the Ig structure and postulate that they may be involved in a unique type of binding interaction between two Ig family domains, i.e., VK binding to a constant region domain of IgG.

Immunochemical studies on dextran-specific and levan-specific myeloma proteins from NZB mice.

Two dextran-specific (PC 3858 and PC 3936) and one levan-specific (PC 3660) NZB myeloma proteins were studied by quantitative precipitin and precipitin-inhibition assays. Both myeloma antidextrans were alphaD-(1 leads to 6) specific and precipitated strongly with a synthetic, linear dextran, molecular weight 35,500, and with other dextrans. The two myeloma antidextrans differed with respect to their relative reactivities with dextrans containing various proportions of alpha-D-(1 leads to 6), alpha-D-(1 leads to 4)-like, and alpha-D-(1 leads to 3)-like linkages. In inhibition assays, the two antidextran myeloma proteins behaved differently from each other, from alpha-D-(1 leads to 6)-specific BALB/c myeloma antidextrans, and from the human antidextrans previously studied. Isomalto-oligosaccharides IM3, IM4, and IM5 were all equal in inhibitory power but were only about 60% as potent as IM6 and IM7, which also inhibited equally on a molar basis. Although precipitation with linear dextran suggests that both may have groove-type sites, as previously inferred for QUPC 52, the size of their combining sites is uncertain. It is not clear whether the sites are only as big as three glucose residues with the increased inhibition by six and seven glucose residues being attributable to partial bivalence and to their ability to combine in several ways along the chain, or whether the site is as big as six glucose residues with the increment in binding by the fourth and fifth glucose residues being minimal and the sixth contributing considerable additional binding-energy. The fructan-specific myeloma protein did not react with inulin, but reacted with many levans and with perennial rye-grass levan containing only beta-D-(2 leads to 6) links. The levan-antilevan reaction was not inhibited by beta-D-(2 leads to 1)- linked oligosaccharides. The findings suggest that PC 3660 has a specificity for (2 leads to 6)-linked chains.