Variable region primary structures of a high affinity anti-fluorescein immunoglobulin M cryoglobulin exhibiting oxazolone cross-reactivity.

Previous studies of murine IgM hybridoma protein 18-2-3, derived from an (NZB/NZW)F1 mouse following hyperimmunization with fluorescein (Fl)-conjugated keyhole limpet hemocyanin, demonstrated a high affinity for Fl (Ka = 2.9 x 10(10) M-1) and cryoprecipitation that was abrogated upon Fl binding to the antibody-combining site. V region sequences of 18-2-3 were determined by Edman degradation and nucleotide sequence analysis. The VH region of 18-2-3 was encoded by a gene VHI(B) of the Q52 VH family with 96% homology to anti-oxazolone antibody NQ7.5.3 but utilized a larger D region (DQ52 plus N region). The V kappa region of 18-2-3 was encoded by a gene V kappa IV with an amino acid sequence 97% homologous to that of anti-oxazolone antibody NQ11.1.18. Although monoclonal anti-Fl antibodies 18-2-3 and 4-4-20 possessed similar binding affinities and quenched bound fluorescein to the same extent (Qmax greater than 96%), they utilized different VH, D, V kappa, and J kappa genes, but the same JH gene segment (JH4). Solid-phase analyses showed that 18-2-3 was not idiotypically related to 4-4-20 and 9-40, prototypic anti-Fl antibodies. Fine specificity binding patterns of Fl analogues by 18-2-3 IgM and IgMs were distinct from other anti-Fl antibodies. Monoclonal antibody 18-2-3 bound phenyloxazolone bovine serum albumin with a lower affinity than for Fl-bovine serum albumin. The first hypervariable region of the 18-2-3 light chain showed homology to human cryoglobulins. This is the first variable region sequence of a murine IgM which self-aggregates at low temperature.

Previous studies of murine IgM hybridoma protein 18-2-3, derived from an (NZB/NZW)Fl mouse following hyperimmunization with fluorescein (Fl)-conjugated keyhole limpet hemocyanin, demonstrated a high affinity for F1 (K. = 2.9 X 10" M-') and cryoprecipitation that was abrogated upon F1 binding to the antibody-combining site. V region sequences of 18-2-3 were determined by Edman degradation and nucleotide sequence analysis. The VH region of 18-2-3 was encoded by a gene VHI(B) of the 6 5 2 VH family with 96% homology to anti-oxazolone antibody NQ7.5.3 but utilized a larger D region (De62 plus N region). The V, region of 18-2-3 was encoded by a gene VJV with an amino acid sequence 97% homologous to that of antioxazolone antibody NQ 11.1.18. Although monoclonal anti-F1 antibodies 18-2-3 and 4-4-20 possessed similar binding affinities and quenched bound fluorescein to the same extent (Q-> 96%), they utilized different VH, D, V., and J. genes, but the same JH gene segment ( J H~) .
Solid-phase analyses showed that 18-2-3 was not idiotypically related to 4-4-20 and 9-40, prototypic anti-F1 antibodies. Fine specificity binding patterns of F1 analogues by 18-2-3 IgM and IgM. were distinct from other anti-F1 antibodies. Monoclonal antibody 18-2-3 bound phenyloxazolone bovine serum albumin with a lower affinity than for F1-bovine serum albumin. The first hypervariable region of the 18-2-3 light chain showed homology to human cryoglobulins. This is the first variable region sequence of a murine IgM which self-aggregates at low temperature.
Cryoglobulins reversibly precipitate at temperatures below 37 "C and have been classified into three types based on the molecular composition of the aggregate (Brouet et al., 1974). Murine IgM 18-2-3 is a Type I cryoglobulin consisting solely of the monoclonal 18-2-3 component. Cryoglobulins have been observed in normal BALB/c mice, but occur at increased levels in autoimmune-prone strains (NZB, NZB/NZW, MRL/ 1) correlated with age and disease severity (Andrews et al., 1978). Data suggest that autoreactive B cell precursors are in a proliferative state in these autoimmune strains since they * This work was supported in part by National Institutes of Health Grant AI 20960 (to E. W. V.). 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.
The nucleotide sequence(s) reported in thispaper has been submitted and 504610.

to the GenBankTM/EMBL Data Bank with accession number(s) 504609
ll To whom correspondence should be sent.
show abnormally high spontaneous polyclonal B cell activation . A high affinity (& = 2.9 X 10" M") murine monoclonal anti-fluorescein IgM antibody 18-2-3 displaying low temperature insolubility in the absence of bound ligand has served as a model to study Type I cryoprecipitation. Antigen binding site involvement was indicated since the presence of fluorescein prevented cryoprecipitation (Ballard et al., 1983). Antibody 18-2-3 was originally derived from an (NZB/NZW)FI mouse, a strain showing a high incidence of autoimmunity. Studies by Ballard et al. (1985) suggested that 18-2-3 was derived from a relatively rare B cell progenitor since examination of 37 IgM and IgG monoclonal antibodies of similar origin and specificity &d not reveal low temperature insolubility or high binding affinity for fluorescein. Previous results indicated that cryoprecipitation occurred via electrostatic interactions involving 18-2-3 antibody-combining sites with interactive sites in the Fc region of the homologous IgM (Dombrink-Kurtzman and Voss, 1988).
In the present study, variable region sequences of heavy and light chains derived from 18-2-3 have been determined through cloned cDNA synthesized from mRNA templates. Three DNA segments (VH,' DH, and J H ) encode the VH region, while two DNA segments (V, and J.) encode the V, region (Seidman et al., 1978;Sakano et al., 1979;Schilling et al., 1980). VH and V. polypeptides both contribute to antigenic binding specificity of antibodies. The V, of 18-2-3 was nearly identical to that of BALB/c anti-oxazolone antibodies (Berek et al., 1985), whereas the VH of 18-2-3 was highly homologous to other BALB/c anti-oxazolone antibodies (Griffiths et al., 1984). Although anti-fluorescein antibodies 18-2-3 and 4-4-20 had similar high binding affinity and fluorescence quenching of bound fluorescein, they differed in VH and V, gene usage. idiotypic and metatypic relatedness, and fine specificity. CDRl of the 18-2-3 V, gene segment closely resembled the human V, sub-subgroup IIIb, which has been preferentially used by a group of human monoclonal IgM-RF cryoglobulins (Kunkel et al., 1973).

Determination of NH2-terminal Amino Acid Sequences-
The NHz-terminal amino acid sequence (43 residues) of 18-2-3 light chain was determined by repetitive Edman degradation. After deblocking the amino-terminal residue of the heavy chain of 18-2-3 with pyroglutamate aminopeptidase, the NHz-terminal sequence (30 residues) was identified. Monoclonal antibody 18-2-3 utilized a VH gene of the Q52 family and a V, gene from the V,5 subgroup of Potter et al. (1982) or V,IV subgroup of Kabat et al. (1987). Southern blot hybridization was utilized to identify J, usage. Restriction fragments obtained from separate digests of 18-2-3 DNA with CfoI and EcoRII indicated J,5 was being used (data not shown).
Heavy C h i n Sequences-Six oligonucleotides were synthesized as primers in sequencing the heavy and light chain variable regions. Fig. 1 describes the oligonucleotides and locations to which they hybridized. The nucleotide and amino acid sequences of the VH segment of 18-2-3 are presented in Figs 31 determined by amino acid sequencing corresponded to those deduced from the nucleotide sequence. The VH region of 18-2-3 was encoded by a gene segment from subgroup VHI(B) (Kabat et al., 1987), belonging to the relatively complex 652 VH family that contains approximately 15 genes (Brodeur and Riblet, 1984). Members of this family also encode the VH region of antibodies specific for oxazolone. Anti-F1 antibody 18-2-3 possessed a D segment closely resembling the DqS2 gene segment (Sakano et al., 1981;Kurosawa and Tonegawa, 1982), as shown in Fig. 4, but did not express the germ line sequence Gln-Leu-Gly since it differed at two bases, resulting in a sequence of Arg-Leu-Glu. Additionally 18-2-3 exhibited variation in the length of the CDR3 segment. Eight noncoded bases (N region; Alt and Baltimore, 1982) appeared to be present between D and J H~, resulting in a D region of six amino acids (Fig. 4). The precise boundaries between V H and D gene segments and D and J gene segments will only be known when germ line D and J gene segments are cloned. A comparison of heavy chain variable region amino acid sequences is also shown in Fig. 3. Interestingly, the 18-2-3 sequence had much greater homology with antibodies of another specificity, anti-oxazolone (VH-Oxl and NQ7.5.3) than with anti-F1 antibodies (4-4-20 and 3-13). Monoclonal antibody 4-4-20 had an affinity comparable to that of 18-2-3 and quenched bound fluorescein to the same degree (Q-> 96%).

18-2-3 . A G G N T N Y N S A L M S R L S I S K D N S K S Q V
VH' Oxl * -" s -" " " " " " " " " " -  (Bothwell et al., 1981;Roth et al., 1985). Q5W is a myeloma clone (Sakano et al., 1981), 04-01 is an anti-ssDNA autoantibody (Smith et al., 1988), and VH-Oxl is an anti-oxazolone antibody. to those obtained from the nucleotide sequence. The V, region was encoded by a member of the V,IV gene family (Kabat et al., 1987) which also encodes the V, regions of antibodies specific for oxazolone, dinitrophenol, arsonate, and a particular anti-idiotype. The J region of 18-2-3 was encoded by J,5.
Binding of 18-2-3 to Fl-BSA and phOx-BSA-A direct binding assay was used in studying the binding of 18-2-3 to phOx-BSA because in preliminary inhibition assays it was not possible to inhibit the high affinity binding of 18-2-3 to solid-phase FI-BSA with fluid-phase phOx-BSA. Binding of 18-2-3 to phOx-BSA (Fig. 8, in the Miniprint) appeared to be low affinity. Although the heavy and light chain variable region sequences utilized by 18-2-3 have been observed in different anti-oxazolone antibodies, no single anti-phOx antibody used both variable sequences found in 18-2-3. Structurally, the epitopes differ. The xanthenone portion of fluorescein is planar (Voss et al., 1976), whereas phenyoxazolone has two nonrigid aromatic rings. Chemical structures of F1-BSA and phOx-BSA are shown in Fig. 9 (in the Miniprint).

DISCUSSION
Data reported here represent primary structural determinations of gene segments encoding the variable domains of murine IgM 18-2-3, an antibody that self-aggregates in the absence of its cognate antigen F1. Antibodies 18-2-3 and 4-4-20 had similar high intrinsic binding affinities (KO = 2-3 X 10" M-') and quenched bound fluorescein to the same extent (Qm.. > 96%), yet they utilized different VH, D, V., and J, gene segments. The only similarity in gene segment usage between the two antibodies was that they contained J H~. Gene segments utilized by the 18-2-3 heavy chain variable region appeared to be VHI(B) (a member of the v~Q 5 2 family), Dqsz, and J H~. Antibody 18-2-3 appeared to be using VH and D gene segments from families residing most proximal to D and J, respectively. Preferential utilization of D-proximal VH gene families has been observed in murine pre-B-cell lines (Yancopoulos et al., 1984) and in hybridomas derived from nonimmunized 6-day-old BALB/c mice (Holmberg, 1987). A comparison of heavy chain variable region sequences indicated that 18-2-3 and anti-phenyloxazolone antibodies used homologous VH genes. The highest degree of homology (96% at the nucleotide and amino acid levels) with known sequences was with anti-phOx antibody NQ7.5.3 which had been obtained 14 days following primary immunization. Meek et al. (1987) suggested that novel mechanisms were involved in generation of D segments in autoantibodies. Although 18-2-3 utilized a gene segment resembling D Q~~ there were eight additional noncoded nucleotides between the D and J H~ segments. The GAATCTTT sequence was probably not attributable to imprecise joining since it did not represent flanking regions nor was D-D joining indicated. The D segment was interesting in that the N segment was A,T-rich rather than G,C-rich. Extra nucleotides may be a product of the activity of terminal deoxynucleotidyltransferase (Alt and Baltimore, 1982), although this enzyme, which polymerizes random deoxynucleotides at 3' ends show a preference for dG residues. Alternatively, the possibility exists that the N segment was germ line-encoded since germ line D genes have not yet been isolated.
The J H~ segment of 18-2-3 was identical to the J H~ germ line gene segment of the BALB/c strain (Sakano et al., 1980)) except for one silent substitution at the third base of codon 102 (T for C). Since 18-2-3 was derived from a NZB/NZW mouse, this difference may be an allelic form of the J H~ gene. The entire J H~ segment was used by 18-2-3. Because anti-F1 antibodies 18-2-3, 4-4-20, and 3-13 all showed a high degree of fluorescence quenching of bound fluorescein (Table 11, in the Miniprint), this property may be related to utilization of the J H~ gene segment.
The sequence of 18-2-3 light chain indicated that the VL segment was encoded by a gene belonging to the V,5 subgroup (Potter et al., 1982). The deduced amino acid sequence of the 18-2-3 VL segment was almost identical to the sequence of light chains in antibodies against 2-phenyloxazolone from BALB/c mice.
The light chain of anti-phOx antibody NQ11.18.1 utilized a V, gene that was 98% homologous at the nucleotide level and 97% at the protein level to the VL gene used by 18-2-3. Hybridoma NQ11.18.1 had been obtained following a secondary immunization 8 weeks after the primary with phenyloxazolone-conjugated chicken serum albumin. Both gene segments could have been derived from two germ line genes in the same family or they could be related by somatic mutation of the same gene. Somatic mutations have been observed in IgM, but are more restricted than in IgG or IgA (Chua et al., 1987). Alternatively, slight differences in homology may indicate that different allelic genes were being utilized since 18-2-3 was derived from an (NZB/NZW)Fl mouse and the anti-phOx antibody NQ11.18.1 was from a BALB/c mouse.
The ability of monoclonal antibody 18-2-3 to bind to phOx may correlate with self-aggregation at low temperatures (Dombrink-Kurtzman and Voss, 1988). The structure of phOx may simulate dipeptidyl conformational or sequential epitopes (e.g. Phe-His) in the Fc region of 18-2-3 to which the antigen binding region of 18-2-3 can bind. Chemical modification studies have indicated that histidines are involved in the Fc region of human IgG and tyrosines on both antigenic and antibody sides of the interactions of two IgG-rheumatoid factors (RFs) (Nardella et al., 1985).
Although RFs are typically IgM and form immune complexes by binding Fc determinants on IgG molecules, selfassociation of IgM (Tsai et al., 1977), and IgG antibodies (Pope et al., 1974;Nardella et al., 1981) has been observed. In such cases each molecule serves as an antibody as well as an antigen, as with 18-2-3. Antibody 18-2-3 did not appear to have RF activity since it did not bind to IgG molecules, but bound to both human and murine IgM molecules.4 Interestingly, there was a high degree of homology between the CDRl region of 18-2-3 V, and that of human V, sub- subgroup IIIb, which is utilized by cryoimmunoglobulin RFs having anti-IgG activity. Ten of twelve amino acid residues were the same. Differences were at positions 27 (Ser and Gln) and 34 (His and Ala) for 18-2-3 and V,IIIb, respectively. Additionally, molecular modeling of the antigen binding site of 18-2-3 has indicated the presence of aromatic residues (tyrosines and tryptophan)! Tyrosine residues have been shown to be involved in the combining site of RFs (Nardella et al., 1985). The sequence VH listed as a murine V,IV gene in Kabat et al. (1987) appears to have been a mistaken classification as a murine gene because VH had been reported by Pech and Zachau (1984) to be related to human V. subsubgroup IIIb. Thus, in accord with the proposed evolution of human V, genes and murine V, genes (Barker et al., 1972), human V. sub-subgroups IIIB and murine V,IV could be considered related phylogenetically. Moynihan et al. (1985) observed restricted association of the VJIIB light chain subsubgroup with &-heavy chain in normal human serum and suggested that the KIIIb-p combination could represent a signal to prevent class switching. This may represent a way of generating high affinity IgM antibodies, as seen with 18-2-3.
X-ray crystallographic analyses of F(ab'L fragments (Amzel and Poljak, 1979) have indicated that the tertiary and quaternary structures of the antigen binding site can be significantly influenced by the chemical nature of the amino acid at position 96 of the light chain. This residue occurs at the V-J junction in CDR3 and is encoded by V and J genes. A conserved leucine was located at position 96 in both antioxazolone antibodies and 18-2-3.
Genes utilized by autoantibodies appear to be present in normal individuals as well as in autoimmune patients. It has been suggested that differences in the complex regulatory pathways of the immune system are responsible for the expansion in autoimmune patients of clones that would be down-regulated in normal individuals (Sanz and Capra, 1988). Findings indicate that autoantibody production in NZB mice results because NZB marrow-derived immature B cells abnormally resist tolerance induction due to defective clonal inactivation (Cowdery et al., 1987). During the secondary immune response in normal humans and animals, IgM RFs are regularly synthesized. Rheumatoid factors may have been maintained during evolution because they have the ability to remove opsonized bacterial and parasites (Clarkson and Mellow, 1981).
Studies investigating the genetic origin of murine autoantibodies have indicated that autoimmune mice do not possess unique IgVH genes (Kofler et al., 1985b). The genetic elements (V, D, J segments) used to encode autoantibodies and antibodies against foreign antigens are not obviously different (Kofler et al., 1985a;Manheimer-Lory et al., 1986). Although somatic mutations can be a contributory factor (Diamond and Scharff, 1984), germ line genes can encode autoantibodies (Naparstek et al., 1986). Recent findings indicate that unmodified or scarcely modified human VH germ line genes encode systemic lupus erythematosus-derived anti-DNA autoantibodies (Dersimonian et al., 1987). Studies based on idiotypic and structural characteristics of human monoclonal cryoglobulins with RF activity have indicated that different VH genes are utilized, but only a limited set of VL genes are present (Kunkel et dl., 1973). An inherent restriction in the immune response to self-antigens was suggested by the preferential association of KIIIb light chains with monoclonal human IgM. RF autoantibodies (Ledfordet al., 1983). The high degree of primary structure homology and cross-reacting idiotypes indicated that the majority of human IgM RF light chains were derived from a single germ line V, gene or a family of closely related VJII germ line genes (Goiii et aL, 1985).
For murine RFs, no clear consensus exists. Part of the divergence is due to the variety of strains and conditions (e.g. unmanipulated, polyclonally activated, antigen-injected) used in the different studies. An additional consideration is the actual number of genes that comprise a family. Originally the 5558 VH gene family was thought to have -60% of the approximately 100 germ line VH genes (Brodeur and Riblet, 1984). Recent evidence indicated that 500-1000 genes exist in the 5558 family (Livant et al., 1986).
Since the anti-fluorescein response is diverse, it was not surprising that antibodies 18-2-3 and 4-4-20 used different VH and V, genes, had unrelated idiotypic and metatypic structures  and demonstrated different fine specificities regarding structural analogues. Yet similarity exists between these two antibodies since they both exhibit high instrinsic affinity for fluorescein and >96% quenching of bound ligand. Although differing in primary structure, the three-dimensional structure of their respective antigen binding sites may be similar. X-ray crystallographic studies are in progress to determine such correlations.