Molecular structure of eight human autoreactive monoclonal antibodies

Introduction

The presence of natural antibodies (Abs) able to react, generally with moderate intrinsic affinity, with multiple and dissimilar self as well as foreign antigens (Ags), such as proteins, nucleic acids, polysaccharides, cytoskeletal and tissue components, polypeptidic hormones and IgG, in the sera of normal non-immunized individuals is known.^1,2 Such multi-reactive Abs are thought to be involved in the elimination of cellular debris and toxic substances, and to contribute to the homeostasis and/or competence of the primary humoral immune system. The majority of natural autoAbs are primarily polyreactive immunoglobulin M (IgM) encoded by a relatively small set of immunoglobulin V genes in near germ-line configuration. Because of their reactivity with various self Ags, it has been postulated that natural Abs can provide the templates for specific high-affinity autoAbs or Abs induced by Ags as found, for instance, in patients with autoimmune diseases. If natural polyreactive Abs provide the templates on which the pressure of an Ag selection process is exerted, they must use immunoglobulin gene segments similar to those used by high-affinity Abs and be able to accumulate somatic mutations of characteristic nature and distribution.

Several studies have indicated that the repertoires of V genes used for natural polyreactive Abs and for ‘regular’ Abs against foreign Ags overlap considerably, a property that may not be attributed only to the expression of certain V genes, but that may depend on other diversification mechanisms.^3–5 The characteristic spectra of Ag-binding activities of polyreactive Abs presumably reflects fundamental differences in the structure of their Ag-binding sites, as compared with those of Ag-induced monoreactive specific Abs. The heavy-chain third complementarity-determining region (H-CDR3) is encoded by the D and flanking N regions and by the 5′-end of the J_H gene segments, and is generally idiosyncratic to each V_H gene rearrangement. The H-CDR3 forms the centre of the Ag-binding site and thus plays a prominent role in Ag binding. Moreover, previous sequence comparisons have pointed towards the critical role played by the H-CDR3 in distinguishing polyspecific from monospecific Ag-binding sites in natural and Ag-induced Abs.^6–8

In the present work, we report the complete nucleotide sequence of V_H and V_L genes encoding eight IgM human autoreactive monoclonal antibodies (mAbs). Their production, characterization and binding to diverse Ags have been reported elsewhere.^9–12 Analysis of sequence homologies led us to determine their germline counterparts, to detect mutations (if any) and to assess the alterations produced by these mutations in the amino acid sequence. We have specifically focused the analysis on H-CDR3 given its importance in Ag binding, as well as in the correlation between V-gene usage and Ab specificity.

Materials and methods

Results

Discussion

In the present work we determined the nucleotide and predicted amino acid sequences of V_H and V_L genes from a panel of eight human IgM autoreactive mAbs, and investigated whether any of the common molecular features related to the functions of mAbs. The results obtained support the notion that, in some cases, anti-DNA specificity is germline encoded. We did not find an increase in the number of positively charged amino acid residues either in H-CDR or in L-CDR of the four mAbs that exhibited anti-DNA reactivity, although these mAbs shared some features with other IgM anti-DNA Abs that have been associated with this reactivity.²⁰ Both IRM-7 and BY-12 reacted with dsDNA and ssDNA and they contained unmutated copies of the V_H26 and DP-14 genes, respectively. Both also had in common L-chains with V_λ2 gene segments (DPL10 and DPL12, respectively). This is an interesting characteristic since the 8.12 anti-DNA-associated idiotype is exclusively encoded by members of the V_λ2 gene family.²⁰ Two other DNA-binding mAbs reacted with either ssDNA (BY-4) or dsDNA (BY-7). BY-4 contained a DXP′1 gene found in a previously described anti-DNA IgM Ab, and a V_κ1 gene which is also an anti-DNA-associated 3I⁺ idiotype.²¹

RF shares important characteristics with natural autoAbs and seems to be encoded by the same V_H gene repertoire.⁸ The L-CDR3 is typically nine amino acids long, the first seven residues being contributed by a particular V_L segment, and the last two residues encoded by the J_L segment.²² Several authors have described a number of autoAbs that contain one or two additional amino acids (mostly Pro and Gly) within the L-CDR3, particularly in RF.^23,24 Stüber et al.²⁵ hypothesized that those B cells that inappropriately regulate the insertion of N regions may be at increased risk for the production of autoAbs. The presence of additional amino acids at the V_κ–J_κ junction is rare, and the expansion of B-cell clones with unusual sequences at the V_κ–J_κ junction is often associated with Ag selection.²⁴ Five mAbs exhibited RF-like activity: BY-12, IRM-7, BY-4, BY-7 and IRM-10. The L-chain of mAb IRM-10 was an unmutated copy of the Humkv325 gene of the Vκ3b family, frequently found in RF. These five mAbs with RF-like activity had longer L-CDR3 segments (10–11 residues) than those not presenting this reactivity. In BY-4 and IRM-10, this was due to N-region nucleotide additions coding for Pro and Arg, and Gly in the case of BY-7 (Table 5). The other three mAbs included in this study not exhibiting RF-like activity did not present extra codons at the junctions, and therefore had the usual L-CDR3 length of nine amino acids. An IgM RF (SSH23) from a normal individual has been described to be encoded by a V_H2 segment and a V_κ2 chain;²⁵ the H-chain of mAb BY-4, with RF-like activity, contained the DP-28 gene segment of the V_H2 family, and a V_κ1 L-chain. The L-CDR3 of BY-4 was 10 amino acids long and contained two Pro residues; SSH23 also has a 10 amino acid long L-CDR3 with two Pro residues at the same position.

A number of human lymphocytotoxic autoAbs have been described, among them Abs that react against the i Ag of red blood cells, and in some cases also against lipid A of bacterial lipopolysaccharide.^26,27 The J_H6 gene (rearranged in 18% of normal human circulating B-cells) was used by 79% of a group of anti-Rh blood group Abs.²⁸ This may reflect structural requirements for Ag binding, since the majority of H-CDR3 segments encoded by J_H6 result in the amino acid sequence Tyr-x-Met-Glu-Val. MAbs IRM-3, IRM-7 and IRM-10 have a similar sequence (Tyr-Gly-Met-Asp-Val) in common, encoded by the J_H6 gene segment. On the other hand, IRM-8 did not exhibit this feature since it had a J_H3 gene element (Table 4).

The localization of the contact sites for SPA binding in immunoglobulin-variable regions is not yet resolved. All the SPA-binder Abs were found to use V_H3 family genes and certain residues showed a high degree of conservation.^29,30 The V_H3 sequences of the mAbs included in this study were in agreement with previous data indicating that eight residues at positions 9, 19, 27 (FR1), 48 (FR2), 67, 73, 78 and 82 (FR3), unique to members of the V_H3 family, are important for SPA binding. Some of these residues located outside the classical Ab binding site might directly interact with SPA, thus providing a distinct binding site that includes two regions, one located within FR1, the other within H-CDR2/FR3 (Fig. 3).

Polyclonal anti-PC Abs of IgG and IgM isotypes have been isolated from human sera and tested for binding to the T15H^50–73 peptide. These Abs share the property of self-binding with their murine counterparts.³¹ It is interesting to point out that Halpern et al.³² searched the sequence database for human immunoglobulin with homology to the T15H^50–73 region and did not find any significant sequence similarities. The mAbs under study were tested for PC specificity and binding to T15H^50–73 peptide: mAbs BY-4, BY-7, BY-12, IRM-7 and IRM-10 bound this peptide, whereas BY-12 and IRM-10 reacted against PC as well. Interestingly, none of these human mAbs had the amino acid sequence of the T15H^50–73 peptide, which determines the murine T15 idiotype, in their V_H segments. This implies that the primary structure responsible for binding the T15H^50–73 peptide, and therefore for self-binding properties, can differ and thus the self-binding property can be maintained in evolution by different primary structures.

As some authors have suggested polyreactive Abs should have longer H-CDR3 (30–60 bases) than monoreactive Abs.² We have analysed the length of the H-CDR3 regions of the mAbs used in this study and all of them were in general terms long (27–60 nucleotides), including those of the monoreactive Abs such as IRM-8 and CDC-1. It has been reported that the presence of Tyr and Trp residues in the H-CDR3 confers flexibility upon the Ab molecule and produces a malleable Ag-binding site capable of accomodating diverse Ags^1·9 It is important to point out that BY-4, the mAb in this study with the shortest H-CDR3, contained one Trp residue and a variable number of Tyr, as was also the case for BY-12, IRM-7 and IRM-10, all of them able to bind multiple compounds. The two monoreactive mAbs, IRM-8 and CDC-1, did not contain Trp and only contained one Tyr residue. Taken together, our data support the importance of H-CDR3 for polyreactivity, although how individual amino acid residues in this region make one Ab polyreactive and another monoreactive is still not clear.