Structure of the V_H and V_L segments of polyreactive and monoreactive human natural antibodies to HIV-1 and Escherichia coli β-galactosidase

Abstract

B lymphocytes committed to the production of antibodies binding to antigens on pathogenic bacteria and viruses (natural antibodies) are common components of the normal human B cell repertoire. A major proportion of natural antibodies Is capable of binding multiple antigens (polyreactive antibodies). Using B cells from three HIV-1 seronegative healthy subjects, and purified HIV-1 and β-galactosldase from Escherichia coll as selecting antigen, we generated three natural IgM mAb to HIV-1 and a natural IgM mAb to β-galactosldase. The three HIV-1-selected antibodies (mAb102, mAb103, and mAb 104) were polyreactive. They bound with different affinities (K_d = 10⁻⁶ to 10⁻⁸ M) to the HIV-1 envelope gp160, the p24 core protein, and the p66 reverse transcriptase, but not to the 120 glycosilated env protein. They also bound to β-galactosidase (K_d ~ 10⁻⁷ M), tetanus toxold, and various self antigens. In contrast, the natural mAb selected for binding to β-galactosldase (mAb207.F1) was monoreactive, In that It bound with a high affinity (K_d < 10⁻⁸ M) to this antigen, but to none of the other antigens tested, Including HIV-1. Structural analysis of the V_H and V_L segments revealed that the natural mAb utilized three segments of the V_HIV gene family and one of the V_HIII family, in conjunction with V_L segments of the V_λI, V_λII, V_λIII, or V_χIV subgroups. In addition, the natural mAb V_H and V_L segments were In unmutated or virtually unmutated (germllne) configuration, Including those of the monoreactive mAb207.F1 to β-galactosldase, and were identical or closely related to those utilized by specific autoantibodles or specific antibodies to viral and/or bacterial pathogens. Thus, the present data show that both polyreactive and monoreactive natural antibodies to foreign antigen can be Isolated from the normal human B cell repertoire. They also suggest that the V_H and V_L segments of not only polyreactive but also monoreactive natural antibodies can be encoded in unmutated or minimally mutated genes, and possibly provide the templates for the specific high affinity antibodies elicited by self or foreign antigens.

Introduction

The presence in normal animals and humans of a population of circulating molecules reactive with a variety of antigens on viruses, bacteria, parasites, and fungi, and displaying physicochemical and biological properties similar to those of the antibodies induced by immunization has been recognized for almost a century (reviewed in ^1-5). Because they arise independently of known and/or deliberate specific immunization, these Ig have been referred to as ‘natural antibodies’. Natural antibodies also bind to heterotogous plasma and tissue antigens, and include the antibodies responsible for blood group incompatibility and hyperacute rejection of xenografts (1,2,6,7). In fact, their reactivity is not restricted to foreign antigens, but includes a wide spectrum of self antigens, such as hormones, e.g. insulin and thyroglobulin, and cell surface glycoproteins and phospholipids, e.g. surface class II molecules and phosphatidylcholine respectively, or structural cell proteins, e.g. myosin, actin, and tubulin (reviewed in ^1-5). mAb with similar antigen-binding properties have been generated by Epstein – Barr virus (EBV)-transformation and/or somatic cell hybridization techniques using B cells from normal humans and mice (reviewed in ^{3 – 5}). Because of their ability to bind microbial antigens and self antigens, natural antibodies may play a major role in the primary line of defense against infections and may contribute, under certain conditions, to the establishment of autoimmune phenomena (reviewed in ^{3 – 5}).

Analysis of natural mAb has shown that these antibodies are in general polyreactive, i.e. a single antibody molecule can bind (with various affinities) multiple antigens, even very different in nature (3–5,8–11). For instance, the natural human mAb we selected for binding to the rabies virus ribonucleoprotein, also bound to ssDNA, human insulin, thyroglobulin, human IgG Fc fragment, bacterial potysaccharides and lipopotysaccharide (LPS) (12). Conversely, natural mAb we selected for binding to ssDNA or insulin also bound to viral and/or bacterial proteins, polysaccharides and LPS (9–11). The polyreactivity of natural antibodies contrasts with the monoreactivity and specificity of ‘antigen-induced’ antibodies (10–14). Whereas monoreactivity and high affinity are ‘acquired’ features of clonotypes and Ig V genes that underwent a process of somatic point-mutation and clonal selection driven by antigen (15–25), polyreactivity is thought to be a function, as we and others have suggested, of unmutated Ig V genes (3–5,16,26–30).

In the present studies, using B cells from healthy subjects and HIV-1 or β-galactosidase from Escherichia coli as selecting antigen, we generated four natural IgM mAb. We found that the three HIV-1 -selected natural mAb were polyreactive while the β-galactosidase-selected mAb was monoreactive and displayed a relatively high affinity for antigen. In addition, we found that not only the polyreactive but also the monoreactive mAb V_H and V_L segments were encoded in genes in the unmutated or virtually unmutated configuration.

Methods

Generation of human mAb-secreting cell lines

Peripheral blood mononuclear cells were obtained from three different healthy (28 – 35 years old) subjects. B lymphocytes were purified, infected with EBV, and cultured in 96-microwell Falcon 3077 plates (Becton-Dickinson, Lincoln Park, NJ) containing irradiated feeder cells (9 – 13,25,29,31,32). After 2 weeks, culture fluids were tested for content of antibodies binding to HIV-1 or E.coli β-galactosidase using specific ELISA (see below). EBV-transformed B cells producing antibodies binding to HIV-1 (mAb102, mAb103, and mAb104) and those producing antibody binding to β-galactosidase (mAb2O7.F1) were selected by sequential subculturing and eventually stabilized by fusion with F3B6 human–mouse hybrid cells (9–13,25,29,31,32). The resulting EBV-transformed B cell hybrids were doned three times and then amplified to produce amounts of mAb suitable for immunochemical analysis.

Analysis of mAb binding to whole HIV-1, HIV-1 components, β-galactosidase, and other antigens

HIV-1 was purified using zonal ultracentrifugation with ribonuclease-free sucrose gradients (33) and heat-inactivated (56°C, 30 min) (MicroGeneSys, Meriden, CT). The HIV-1 gp120 external grycosilated env protein, the gp160 external glycoprotein, the p24 gag polyprotein (derived from a HIV-1 gene fragment which codes for a gag polyprotein that includes all p24 plus 12 amino acids of the C-terminus of p17 and 57 amino acids of the N-terminus of p15), and the p66 reverse transcriptase (derived from an HIV-1 pol gene fragment which codes for all the reverse transcriptase plus 13 amino acids of the C-terminus of protease and 34 amino acids of the N-terminus of the endonuclease) were produced in insect cells using the baculovirus expression system and purified under conditions designed to preserve its biological activity and tertiary structure (MicroGeneSys). β-Galactosidase from E. coli, purified tetanus toxoid, ssDNA, recombinant human insulin, purified thyroglobulin, Fc fragment of human IgG, and BSA were as reported (9–13). HIV-1 and HIV-1 components were used to coat ELISA plates at a protein concentration of 3 – 1 0 μg/ml in 0.05 M carbonate buffer, pH 9.6. The ELISA plate coating conditions for all other antigens were as reported (9–13). Antigen-bound antibodies were detected using affinity-purified horseradish peroxidase-conjugated goat antibodies to human IgM (9–13). Competitive inhibition studies involving solid-phase and soluble antigen were performed to calculate the dissociation constant (K_d) values (9–13).

Sequences of the mAb V_H and V_L genes

Cellular mRNA was isolated from EBV-transformed B cell hybrids using the Fast Track kit (Invitrogen, San Diego, CA). cDNA libraries were constructed from 5 μg of poly(A⁺) mRNA using the λgt11 phage vector (12,23,24,29). Each cDNA library was screened by filter hybridization using the previously reported V_H, C_H, C_λ, and C_χ, probes labeled with deoxycytidine 5′-[α-³²P]-triphosphate (dCTP) (spec, act., 3000 Ci/mmol, Amersham, Arlington Heights, IL) by random priming or γ^-32P-end labeled (C_λ oligonucleotJde) (12,23,24,29). After doning of the recombinant phages, the V_H and V_L gene segments were amplified by polymerase chain reaction (PCR) from the isolated phages using the forward and reverse λgt11 primers. cDNA inserts were digested with EcoRI, purified, and ligated into EcoRI-digested, dephosphorylated pUC18 vector. Recombinant pUC18 plasmids were amplified in DH5α competent E.coli cells and purified (12,23,29). The cloned V_H and V_L genes were sequenced by the Sanger's dideoxy chain termination method, using the Taq polymerase (12,23–25,29).

Analysis of the nucleotide and deduced amino acid sequences

DNA sequences were analyzed using the software package of the Genetic Computer Group of the University of Wisconsin, Release 6, and a Model 6000-410 VAX computer (Digital Equipment Corp., Marlboro, MA). Ig V_H and V_L gene sequence identity searches were performed using the GenBank Database and the FASTA program (34).

Results

Generation and analysis of the human natural mAb

Using B lymphocytes from three different healthy subjects and purified HIV-1 or E. coli β-galactosidase as selecting antigen, we generated three HIV-binding IgM mAb (mAb 102,103, and 104) and one β-galactosidase-binding IgM mAb (mAb207.F1) (Table 1). Three mAb utilized λ and one χL chains. The HIV-1-binding mAb were polyreactive. mAb 102, mAb104, and, at much lower degree, mAb 103 bound with different affinities to different HIV-1 components, including the p66 reverse transcriptase, the p24 gag polyprotein, and the full-length gp160 external env glyco-protein (Table 1). The three mAb bound with very low affinity to the external gp120 env gtycoprotein and failed to neutralize HIV-1 in vitro (data not shown). However, they all bound in a dosesaturable fashion to other foreign antigens, including tetanus toxoid and E. coli β-galactosidase, and self antigens, including human insulin, thyroglobulin, ssDNA, and, much less efficiently, to human IgG Fc fragment (Fig. 1A–C). BothmAb102and 104 bound β-galactosidase with a moderate affinity (K_d = 10⁻⁷ M) (Table 1). In contrast to the polyreactivity of the HIV-1-selected mAb, the β-galactosidase-selected mAb207.F1 bound with a high affinity to this antigen (K_d < 10⁻⁸ M), but to none of the other antigens tested, including HIV-1 (Fig. 1D and Table 1).

Table 1.

Nucleotide and amino acid differences in V_H and V_L regions of natural human mAb

Clone	Donor	mAb chains	Binding
			HIV-1 components				β-galactosidase
		H,L	gp160	gp120	p66	p24
mAb102	267	μ,λ	ND	>10−4	4.7 × 10−7	>10−4	3.0 × 10−7
mAb103	267	μ,λ	ND	>10−4	>10−4	>10−4	ND
mAb104	259	μ,χ	1.0 × 10−7	>10−4	1.0 × 10−8	6.3 × 10−7	2.0 × 10−7
mAb207.F1	307	μ,λ	>10−4	>10−4	>10−4	>10−4	< 10 × 10−8

VH gene				Df gene	JH gene	VL gene				JL gene
Closest segment [family]	Identity (%)	Framework region	Complimentarity determining region			Closest segment [subgroup]	Identity (%)	Framework region	Complimentarity determining region
4.33a [VHIV]	99.0 (99.0)	3 (1)	0 (0)	D21/10	JH4	DSC VλIIg [VλII]	99.3 (98.0)	0 (0)	2 (1)	Jλ1
VH26cb [VHIII]	1000 (100.00)	0 (0)	0 (0)	DN2	JH4	VλIII.1H [VλIII]	100.0 (100.0)	0 (0)	0 (0)	Jλ2
VH4.21c [VHIV]	98.7 (99.0)	3 (1)	1 (0)	DXP’1	JH1	VχIVj [VχIV]	97.7 (96.1)	3 (2)	4 (2)	Jχ4
3d279dd [VHIV]	99.7 (99.0)	1 (1)	(0) (0)	DK4	JH2	T2:C5i [VλI]	100.0 (100.0)	0 (0)	0 (0)	Jλ2

Genomic germline V_H gene sequences as reported by.

^aWeng et al (35) and van Es et al. (36)

^bChen et al (39)

^cSanz et al. (37); and

^dvan der Maarel et al. (38). The only nucleotide difference, a G instead of an A in position 213 (resulting in a substitution of an I with an M at position 71 of the deduced amino acid sequence), displayed by the mAb207.F1 sequence when compared with that of the germline 3d279d gene is shared by at least 10 members and/or alleles of the V_HIV family, including 4.35 (see Fig. 2).

^fThe expressed D gene sequences displayed only partial identity with those of the reported germline genes (see Results).

Genomic germlme V_L gene sequences as reported by. ^h Cambriato and Kobeck (48)

ⁱKlobeck et al. (44). Expressed (possibly unmutated) V_L genes as reported by

^gPaul et al. (46) and

^jBerinstein et al. (50). The DSC V_λII sequence displays 18 nucleotide differences compared with the (only available V_λII) genomic germlme V_λII 2.1 sequence reported by Brockly et al (47) The mAb207 F1 and T2:C5 V_λ1 gene sequences are identical and displayed only two nucleotide differences with the sequence of the germline Hum1v117 gene (28) (see Fig. 2).

Percentages and number of substitutions in parentheses refer to deduced amino acid sequences.

Fig. 1.

Binding of natural human mAb to HIV-1 (◇), tetanus toxod (■), β-galactosidase from E. coli (+), ssDNA (△), insulin (●), thyroglobulin (▲), Fc fragment of human IgG (○), and BSA (□).

Human natural mAb V_H segments

The nucleotide and deduced amino acid sequences of the V_H segments of the four natural mAb are depicted in Fig. 2(A and B respectively). Their differences compared with those of the closest known germline genes are summarized in Table 1. The criterion used for assignment to a given V_H gene family was 80% sequence identity. mAb102 V_H nucleotide and deduced amino acid sequences were virtually identical with those of the 4.33 (35) and related H10 genes (36), members of the V_HIV family. Of the three nucleotide differences, only one (in FR2) resulted in a deduced amino acid difference. When compared with that of the germline V_H4-21 gene, another member of the V_HIV family (37, 38), the mAb104 v_H nucleotide sequence displayed only three differences throughout the coding area. These resulted in a single difference in the (FR3) deduced amino acid sequence. Thus, it is conceivable that both mAb102 and mAb104 V_H genes represented the expression of minimally polymorphic alleles of the 4.33 V_H and V_H4-21 genes respectively, although the possibility that they consisted of slightly mutated forms of the above germline genes could not be ruled out. The mAb103 V_H gene sequence was identical with that of the germline V_H26c gene, a member of the V_HIII family (38). Finally, the monoreactive anti-β-galactosidase mAb207.F1 V_H gene sequence was identical to that of the germline 3d279d gene, also a member of the V_HIV family (39), except for a G instead of an A in position 213, resulting in a substitution of an I with an M at position 71 of the deduced amino acid sequence. Such a unique difference is in fact shared by at least 10 members and/or alleles of the V_HIV family, including 4.35 (Fig. 2A and B) (35 – 37,39). This is highly consistent with the hypothesis that the mAb207.F1 V_H segment was encoded in a yet unidentified minimally polymorphic allele of the 3d279d gene.

Fig. 2.

Nucteotide (A) and deduced amino acid (B) sequences of the V_H genes utilized by the natural human mAb. In each duster, the top sequence is given for comparison and represents the germline V_H gene displaying the highest degree of identity to the expressed V_H genes of the cluster The 4.33, V_H4.2 1 , and 3d279d genes belong to the V_HIV family. The leader region sequence of the V_H4.21 is not available, that of the V58 gene leader region is provided instead. The V_H26c gene is a member of the V_HIII family. Dashes indicate identities. Solid lines on the top of each duster depict CDR. Small letters denote untranslated sequences Asterisks indicate translation initiation codons. The present sequences are available form EMBL/GenBank/DDBJ under accession numbers L25291, L25292, L25293, and L25294.

Human natural mAb D and J_H genes and configuration of the CDR3

The D segments utilized by the natural mAb were heterogeneous and very different in length (18–51 nucleotides) (Fig. 3A). The mAb102 and mAb104 D segment sequences displayed a high degree of similarity to portions of those of the D21/10 and DxP'1 germline D genes (40,41) respectively. The mAb103 and mAb207.F1 D segment sequences displayed stretches of similarity to those of the germline DN2 (42) and Dk4 (41) genes respectively. All mAb D gene sequences with the exception of that of mAb207.F1 were flanked by unencoded (N) nucleotide additions.

Fig. 3.

Nudeotide (A) and deduced amino acid (B) sequences of the natural mAb D and J_H segments. Germline D genes are given for comparison. Dashes indicate identities. The present sequences are available from EMBUGenBank/DDBJ under accession numbers L25291, L25292, L25293, and L25294.

mAb102 and mAb103 utilized intact and truncated forms respectively of the J_H4 gene (Fig. 3A). The two expressed J_H4 sequences were in germline configuration and displayed an identical nucleotide variation from the germline J_H4 gene sequence originally reported by Ravetch et al. (43). This variation has been previously reported in other expressed Ig genes (24,25,29) and is consistent with the prototypic sequence reported by Yamada et al. (44). mAb104 utilized a truncated form of J_H1 gene, displaying only one nucleotide difference when compared with the original sequence reported by Ravetch et al. (43). Finally, mAb207.F1 utilized an intact germline J_H2 gene.

Figure 3 (B) depicts the deduced amino acid sequences of the joined D – J genes. The first portion of each sequence encodes the CDR3 segment, as defined by Kabat et al. (45). The CDR3-encoding sequence encompasses the whole D gene and the first ‘non-conserved’ portion of the J_H gene, up to the invariant W codon (TCG). The remaining ‘conserved’ sequence of the J_H gene encodes the FR4. Using these criteria, the expressed CDR3 varied in length from 11 (mAb103) to 23 (mAb104) amino acids. The expressed FR4 sequences were conserved and invariable in length.

Human natural mAb V_L – J_L segments

The nucleotide and deduced amino acid sequences of the V_L segments of the four natural mAb are depicted in Fig. 4(A and B respectively). Their drfferences when compared with the closest known germline genomic or expressed gene sequences are summarized in Table 1. The mAb102 V_λ nudeotide and deduced amino acid sequences were virtually identical with those of the expressed DSC V_λII (V_λII subgroup) gene (46), which differs by 18 nudeotide from the only genomic germline V_λII gene sequenced, V_λII 2.1 (47). The mAb103 V_λ gene sequence was identical with that of the germline V_λII.1 gene (V_λIII subgroup) (48). The mAb104 V_χ nudeotide and deduced amino acid sequences displayed the highest degrees of identity with those of the germline V_χ4 gene (49). Finally, the mAb207.F1 V_λ gene sequence was identical with that of the expressed and T2:C5 gene (V_λI subgroup) (50). The two identical sequences differed by only two nucleotides from that of the germline Hum1v117 V_λ1 gene (28) and likely constituted the expression of a polymorphic allele or a Hum1v117-like gene.

Fig. 4.

Nucleotide (A) and deduced amino add (B) sequences of the V_L genes utilized by the natural human mAb. In each duster, the top sequence is given for comparison and represents the published V_L gene displaying the highest degree of homotogy to the expressed V_L genes of the cluster. V_λIII.1 (V_λIII subgroup), V_λIV, and Hum1v117 (V_λI subgroup) are sequences of genomic germline genes. T2:C5 (V_λI subgroup) and DSC V_λII are sequences rearranged, but possibly unmutated genes. Dashes ridicate identities. Solid lines on the top of each duster depict complimentarily determining regions. Small letters denote untranslated sequences. Asterisks indicate translation initiation codons. The present sequences are available from EMBL/GenBank/DDBJ under accession numbers L25295, L25296, L25297, and L25298.

The nudeotide and deduced amino acid sequences of the J_L segments of the four natural mAb are depicted in Fig. 5(A and B respectively). The mAb102 utilized an intact and unmutated J_λ1 gene. The mAb103 and mAb207.F1 utilized untruncated forms of the J_λ2 gene incomplete and virtually complete germline configuration respectively. Finally, the mAb104 utilized an untruncated J_χ4 gene with one silent change. Both mAb102 and mAb207.F1 J_λ gene sequences were 5’ flanked by three nucleotides, unencoded in the respective V_λ genes and likely representing N additions. Such unencoded additions might also account for the first two nucleotides of the mAb207.Fi J_λ2 gene sequence.

Fig. 5.

Nucleobde (A) and deduced amino acid (B) sequences of the J_L segments of the natural mAb. In each cluster, the top sequence is given for comparison and represents the published germline J_L gene displaying the highest degree of homology to the expressed J_L genes of the cluster. Dashes indicate identities. The present sequences are available from EMBUGenBank/DDBJ under accession numbers L25295, L25296, L25297, and L25298.

Overall configuration of the natural mAb V_H and V_L segments

The above structural analysis showed that the mAb 103 V_H and V_λ segment sequences were identical with the deduced amino acid sequences of the germline V_H26c and V_λIII genes respectively; and that the mAb207.F1 V_H and V_λ segment sequences were identical with the deduced amino acid sequence of a putative minimally polymorphic allele of the germline 3d279 gene and that of the putative T2:C4 variant of the Hum1v117 V_λI gene respectively. The mAb102 V_H and V_λ segment sequences displayed each only one difference compared with the deduced amino acid sequences of the germline 4.33 and DSC V_λII genes respectively, suggesting that they represented the expression of minimally polymorphic and/or mutated germline genes. The mAb104 V_H sequence displayed only one difference (in FR3) compared with the deduced amino acid sequence of the germline V_H4-21 gene, while the V_χ segment sequence displayed four differences compared with the deduced amino acid sequence of the germline V_χIV gene. The well documented polymorphism of the V_HIV family genes (35–37,39,51), the unmutated configuration of the V_H and V_χ FR4 sequences, and the fact that the R and D amino add variations found at position 83 and 99 respedively of the mAb104 V_χIV segment are displayed by other expressed V_χIV segments (45) are consistent with the hypothesis that the mAb104 V segments constituted unmutated forms of V_H4-21 and V_χIV polymorphic alleles. Alternatively, the mAb104 V segments could represent the expression of V_H4-21 and V_χIV genes that accumulated two to three somatic point-mutations. Such putative somatic point-mutations, however, did not display a nature and a distribution characteristic of those resulting from an antigen-dependent selection process.

Discussion

We analyzed the V_H and V_L segment structure and reactivity of three natural IgM mAb we generated by selection for binding to HIV-1, a foreign viral mosaic antigen, and those of a natural IgM mAb we generated by selection for binding to β-galactosidase from E. coli, a foreign bacterial antigen, using B cells from three healthy HIV-1 negative subjects. We found that: (i) consistent with the polyreactive nature of a major proportion of natural antibodies was the reactivity with multiple foreign and self antigen displayed by the three HIV-1-selected mAb; (ii) in spite of the β-galactosidase-binding activity displayed by a large proportion of natural polyreactive antibodies (4,11), including mAb102, mAb103, and mAb104, selection for binding to β-galactosidase led to the generation of the ‘specific’ IgM mAb207.F1; and, finally, (iii) not only the three polyreactive mAb, but also the monoreactive mAb207.F1 displayed unmutated or virtually unmutated V_H and V _L segments.

This is one of the first reports providing the complete structure of the V_H and V_L segments of human polyreactive and mono-reactive natural antibodies selected for binding to biologically relevant foreign (viral and bacterial) antigens, and showing that these V segments are unmutated or virtually unmutated. The putatively unmutated configuration of the V segments of the three polyreactive natural IgM is consistent with: (i) the germline configuration of both V_H and V_L chains of another human polyreactive natural IgM mAb, Kim 4.6, with well characterized antigen-binding activity (52). This mAb was generated in a healthy subject by selection for binding to ssDNA, and also reacts with cardiolipin, synthetic polynucleotides, and RNA, and utilizes a germline V_HIII gene, 1.9111, in conjunction with a germline V_λ I gene, designated Hum1v117(28); and (ii) the nearly unmutated configuration of the V_H and V_L segments of two natural polyreactive mAb binding xenoantigen, self antigen, and chemical haptens, generated from spleen cells of a normal, non-immune BALB/c mouse (26). Thus, utilization of germline genes is associated with potyreactivity of the encoded Ig V_H and V_L segments in both human and murine natural antibodies. Application of antjgenic pressure to polyreactive natural antibody-producing cell precursors may lead through somatic hypermutation and sequential selection to affinity maturation (3–5).

For instance, in A/J mice, sequential selection by p-azophenyl arsonate of polyreactive natural antibody-producing cells displaying the dominant V_H-ldCR (or CRI-A) idiotype results eventually in the generation of high affinity anti-arsonate IgG (16). Early stages of a similar antigen driven selection process may underlie the high proportion of somatically mutated IgM-producing cell precursors found in the human peripheral blood (53–55). Somatic hypermutation and antigen-driven clonal selection of polyreactive antibody-producing cell precursors can lead to an increased antibody affinity for the selecting antigen, but not necessarily to loss of polyreactivity, as suggested by our recent demonstration of antigen-selected point-mutations in polyreactive human IgG mAb (56). As shown in one of the present natural antibodies (mAb207.F1), monoreactivity and relatively high affinity for an exogerious antigen can be encoded in unmutated V_H and V_L segments. This contention is further supported by the unmutated configuration of the V_H and V_L segments of the monoreactive mAb to self antigen, particularly DNA, we recently generated in healthy subjects (11 and Chai et al., in preparation). The unmutated or minimally mutated configuration of the present not only polyreactive but also monoreactive mAb V_H and V_L segments would be consistent with the definition, proposed here, of natural antibodies as Ig binding to foreign antigens, including those on bacteria and viruses, and/or self antigen and produced by clones emerging in an antigen-independent fashion.

Findings in humans and mice have suggested that the antigen-binding properties of individual polyreactive antibodies are distinctive (reviewed in ^{3 – 5}). Accordingly, the present three polyreactive IgM mAb display discrete antigen-binding activities for different self and foreign antigens. The functional ‘uniqueness’ of each IgM likely reflects the structural heterogeneity of the antigen-binding sites of these mAb, which display different assortments of V_H, D, and J_H segments, as well as V_L and J_L segments, and different junctional V_H – D – J_H and V_L – J_L sequences. Although some unmutated V segments display an inherent affinity for certain antigens in mice, e.g. V_HOx1 for 2-phenyl-5-oxazolone (18), S107 V_H1 for phosphorylcholine (19), V_H11 and V_H12 for phosphatidylcholine (reviewed in ⁵⁷), and S107 V_H11 for DNA (58), or in humans, e.g. V_H4-21 and V_χIIb for the red cell i/l antigen and IgG Fc fragment respectively (59–61), the somatically generated H chain CDR3 appears to be the major contributor to the overall function of the antigen-binding site, at least in the case of relatively large proteinic antigens (62,63). The primary structures of the three polyreactive IgM mAb H chain CDR3 were highly divergent in composition and length, and did not allow for the identification of obvious common motifs possibly responsible for polyreactivity. Thus, if, consistent with what has been suggested by the in vitro expression of human IgM rheumatoid factor (RF) H and L chain genes (64), and by a census of 84 natural polyreactive and antigen-induced monoreactive murine mAb (65), the H chain CDR3 provides the correlate for polyreactivity, it does so by virtue of a discrete primary structure in each of the polyreactive mAb reported here.

The present findings further substantiate the hypothesis that the spectra of natural reactivities with foreign and self antigens vastly overlap as a result of the polyreactivity of a major proportion of the natural antibody clonotypes. They also further suggest that natural antibodies provide the templates for the generation of disease-related autoantibodies as well as specific high affinity antibodies induced by foreign antigens, particularly viruses and bacteria (reviewed in ^66–70). For instance, while a completely unmutated copy of the V_H26c gene, utilized by the mAb103 and the polyreactive naturaJ IgM mAb18 we reported elsewhere (27), was originally identified as encoding the V_H segment of an anti-DNA autoantibody from a SLE patient (71), mutated copies of the V_H26c gene are utilized by other anti-DNA (21,88) and RF (22,72,73) IgG and IgM autoantibodies from with SLE and rheumatoid arthritis patients respectively. In the case of foreign antigens, mutated V_H26c segments are utilized by high affinity specific antibodies to pathogenic viruses, including IgG mAb elicited through affinity maturation in healthy subjects by rabies virus (12,24) or herpes simplex virus (74), and by specific antibodies to pathogenic bacteria, including IgG and IgM mAb induced in healthy subjects by Haemophilus influenzas type b polysaccharide (75). Analogously, unmutated and mutated forms of the V_H4-21 gene are utilized by a variety of RF autoantibodies generated in healthy subjects, rheumatoid patients, as well as patients with chronic lymphocytic leukemia (reviewed in ^{67 – 69}). Also, mostly unmutated copies of the V_H4-21 gene are utilized by the vast majority of anti-i and anti-l red blood cell IgM auto-antibodies (cold agglutinins) generated in patients with idiopathic cold agglutinin disease (CAD) or CAD associated with B cell lymphomas (59,60). A gene consisting of a somatically mutated form of 4.33, utilized by the polyreactive anti-HIV-1 natural mAb102 or closely related to 4.33, encodes the V_H segment of the high affinity lgG2 mAb 12-116 (74), which is specific for a segment of the HIV-1 gp41 mapping amino acids 644 – 663 (V3 loop), and was generated using B cells from a HIV-1 seropositive subject. Finally somatically mutated forms of the V71-2 (V_HIV) segment, a gene highly related to 3d279d (38,51), utilized by the mAb207.F1, are utilized by some RF and other autoantibodies (reviewed in ^66–70).

Despite the smaller number of reported V_H gene sequences compared with that of V_H gene sequences, it was possible to establish that V_H segments identical or similar to those utilized by the present four natural mAb are utilized also by different RF and/or anti-DNA autoantibodies derived from autoimmune patients, as well as by specific high affinity antibodies to microbial pathogens. For instance, a virtually identical copy of the V_λII gene, utilized by mAb102, is utilized by an anti-DNA IgM auto-antibody bearing the characteristic 18.2 idiotype (46), and minimal variants of the mAb102 V_λII gene as well as other V_λII genes are utilized by specific IgG and IgA antibodies elicited by H. influenzae type b polysaccharide (77). Segments consisting of genes related to or, more likely, mutated copies of the V_λIII and V_λI genes utilized by mAb103 and mAb207.F1 respectively, are utilized by different IgM, IgG, and IgA RF (22,25,29,72). Finally, mutated copies of the single V_χIV gene (mAb104) are utilized by an IgA RF (78) and an anti-DNA IgG autoantibody (21), as well as by specific IgM mAb to Pseudomonas aeruginosa exotoxin A (79) and to Streptococcus group A carbohydrate N-acetylgucosammine (80,81).

Although limited in number, the assortment of the V_H genes utilized by the present natural antibodies is consistent with the hypothesis postulated by Walter et al. (82) and van Djik et al. (83), on the basis of DNA deletional mapping in a large panel of EBV-transformed B cells, that the elements located within 800–1000 kb 5’ of the J_H locus account for the majority of the ‘functional’ human V_H genes. Two of the natural mAb V_HIV genes, V_H4-21 and 3d279d, have been mapped to <650 and 850 kb respectively of the J_H locus (38,73). The third natural mAb V_HIV gene, 4.33, has not been located by the most recent analyses of the chromosome 14 long arm 14q32.3 region, but what it is likely an allelic form of it, the 4.34 gene, has been mapped to <500 kb 5’ of the J_H locus (38). Finally, the fourth natural mAb V_H gene, V_H26c, has been mapped to ~350 kb 5’ of the J_H locus (84). The high frequency of expression in the normal and autoimmune B cell repertoires of at least two of these ‘J_H-proximal’ V_H genes, i.e. V_H4.21 and V_H26c, has been documented (66–69, 85–87). If confirmed, the notion that a significant fraction of the expressed human Ig V_H gene repertoire arises from a preimmune repertoire that is dominated by relatively few V_H genes would suggest a significant evolutionary advantage for the expression of polyreactive B clonotypes which are encoded in germlme Ig V genes. A limited number of polyreactive (natural) clonotypes could provide the precursors form which a diverse repertoire arises by clonal selection.

Abbreviations

CAD

cold agglutinin disease

EBV

Epstein – Barr virus

PCR

polymerase chain reaction

RF

rheumatoid factor