Somatic mutations can lead to a loss of superantigenic and polyreactive binding

Abbreviation:

1 Introduction

Immunoglobulins in their monomeric form are four-chain macromolecules composed of two heavy (H) and two light (L) chains. Each chain contains variable domains, which define the antigen (Ag)-combining site of the antibody, and several constant domains aimed to define effector function. Both H and L chain variable domains contain three regions of extensive sequence variability, termed the complementarity-determining regions (CDR), and four regions of relative sequence stability, termed framework regions (FR). The three L chain CDR and the three H chain CDR are juxtaposed to form the antigen-combining site of the antibody, as classically defined. In turn, the FR creates a scaffold that surrounds, supports, and influences the conformation and structure of the CDR 1–4. Although the historical and canonical basis of immunology was the antibody monospecificity dogma, the binding of a single antibody to different unrelated antigens has been extensively documented 5–7. In addition, data derived from crystallographic studies elegantly visualized the simultaneous binding of two different antigens with a single paratope.

Together with their ability to bind antigens via CDR, lg can react with unconventional antigens via the FR 1. These unconventional Ag are the B cell superantigens (SAg), andmost of them include microbial proteins, such as protein A (SpA) from Staphylococcus aureus 4, gp120 from HIV-1 8, and the staphylococcal enterotoxin A 9. SpA contains five highly homologous extracellular Ig-binding domains, which can bind Fcγ and Fab fragments 4. The Fab specificity involves a site on the variable region of the Ig H chain 8 and is restricted to products of the human V_H3 family that represent as many as 43% of inherited human V_H genes 3, 9–11. It has been proposed that in vitro stimulation with SpA can contribute to selection of these B cells and promote the production of antibodies that may include rheumatoid factor autoantibodies 12, whereas in vivo exposure to recombinant SpA can result in supraclonal suppression and deletion of B lymphocytes expressing a given V_H family 13, 14. The structural basis for these Fab-binding interactions was recently revealed in crystallographic analysis of the complex between a V_H3-expressing Fab and domain D of SpA 15. A V_H3-binding human SAg has also been reported 9, 16.

In a previous study, we analyzed four different human monoclonal antibodies (mAb) displaying different isotypes (IgA1κ, IgG1κ, IgG2κ, and IgG4κ) and anti-tubulin activity that were isolated from the serum of a patient (patient PER) suffering from an immunocytic lymphoma 17. The complete sequence of these mAb showed that they shared the same V_H and V_L domains associated with mutations compatible with an antigen-driven process. However, despite the identity in their variable domains, IgA₁κ, IgG₁κ, and their Fab fragments bound to a common motif of β-tubulin with striking differences in affinity 18, 19. Interestingly, despite their expression of a V_H3 gene family, the different Ig from patient PER were unable to bind SpA, suggesting that the mutational process undergone by this Ig gene resulted in the loss of this activity.

To better assess the effect of somatic mutations in the V_H3–73 gene, we constructed two recombinant antibodies, one containing the complete mutated κ chain from IgA₁κ-PER and the corresponding V_H3–73 region in its germ-line configuration (IgMκ-Germ) and the other containing the same mutated κ chain and the mutated form of the V_H3–73 gene expressed by patient PER (IgMκ-PER). We compared the antigen and SAg binding affinities of the recombinant Ig, and our findings demonstrate that the germ-line transformation of V_H3–73 results in recovery of SpA-binding activity, does not introduce significant modifications in the affinity for tubulin, and induces the appearance of a polyreactive-binding pattern to the different autoantigens. This polyreactive binding was also observed when IgMκ-PER was submitted to a dissociating process, providing new evidence that conformational changes could play an important role in the genesis of polyreactivity.

2 Results

2.1 Comparison of recombinant mutated (IgMκ-PER) and germ-line (IgMκ-Germ) V_H3-encoded antibodies

To evaluate the role of the somatic mutation process in the generation of IgMκ-PER antibody specificity and to avoid misinterpretation related to V gene polymorphisms, we isolated and sequenced the germ-line V_H3–73 from autologous genomic DNA of patient's T cells 18. The differences between the germ-line 3–73 V_H domain and the mutated form expressed by patient PER are depicted (Fig. 1a, b). There were eight and four codons which underwent replacement (R) and silent (S) mutations, respectively, in the FR (R:S ratio of 2), as compared to nine replacement and three silent mutations in CDR 1 and 2 (R:S ratio of 3). Mutations and replacements predominant in CDR induced a significant hydrophobicity shift in these domains (Fig. 1a, b). In addition, the structural model of an Ab with mutated V_H3–73 is shown, with the mutated residues highlighted (Fig. 1c). Somatic mutations resulting in multiple replacements in the CDR with preservation of the peptide sequence in the FR are typical of Ab that have undergone selection by Ag. In addition, mutations in FR predominated in the FR3 domain of both H and L chains, a region that may itself contribute to Ag recognition 18.

2.2 Expression and purification of IgMκ-PER and IgMκ-Germ antibodies

Expression vectors containing the mutated human μ H chain (pRTM1μ-PER, Fig. 2a) and the germ-line-encoded human μ H chain (pRTM1μ-Germ, Fig. 2b) were constructed and co-transfected into the murine myeloma X-63 cell line together with a vector for the expression of the mutated human κ L chain (pSVκ-PER, Fig. 2c). Transfected clones were selected on the basis of their resistance to neomycin, and supernatants were tested by ELISA and Western blot for secretion of complete IgMκ. "Transfectomas" secreting complete pentameric human IgMκ-PER or IgMκ-Germ were expanded, and culture supernatants were purified by affinity chromatography (Fig. 2d). The presence of pentameric IgMκ recombinant Ig was substantiated by gel filtration chromatography for both recombinant Ig (data not shown).

2.4 Germ-line recovery of IgMκ-PER results in a moderate decrease in the affinity for tubulin

The affinity constants of both antibodies with the tubulin antigenic determinant (QQYQDATAEEEEDFGEEAE peptide motif) were also analyzed: K_D =9.8×10^–9 M (k_ASS=5.57×10⁴M^–1s^–1 and k_DISS =5.46×10^–4s^–1) for recombinant IgMκ-PER and K_D =1.7×10^–8 M (k_ASS = 5.04 × 10⁴ M^–1s^–1 and k_DISS =8.41×10^–4 s^–1) for recombinant IgMκ-Germ (Fig. 3 and Table 1).

Table 1. Affinity constants of IgMκ-PER and IgMκ-Germ against SpA and the β-tubulin peptide (QQYQDATAEEEEDFGEEAE)

	Protein A		Tubulin peptide
	IgMκ-PER	IgMκ-Germ	IgMκ-PER	IgMκ-Germ
kASS (M–1s–1)	2.48 × 103	8.32 × 104	5.57 × 104	5.04 × 104
kDISS (s–1)	3.66 × 10–4	5.71 × 10–4	5.46 × 10–4	8.41 × 10–4
KD (M)	1.47 × 10–7	6.86 × 10–9	9.8 × 10–9	1.7 × 10–8

2.5 Germ-line recovery of IgMκ-PER results in the appearance of polyreactive binding activity

To analyze the effects of somatic mutations on polyreactivity, we carried out ELISA with a panel of known autoantigens (actin, myosin, tubulin, and DNA). The utilization of the germ-line V_H3–73 segments in the IgMκ-Germ construction resulted in the appearance of significant binding to actin, myosin, and DNA, in addition to the binding to tubulin (Fig. 4). Because the two recombinant antibodies exhibited a similar reactivity to anti-κ Ig, differences in the antibody concentrations were disregarded (Fig. 4).

2.6 Dissociating treatment of IgMκ-PER and IgMκ-Germ results in the appearance of polyreactive binding activity

To determine whether the observed differences in the binding of the recombinant Ig could be related to conformational modifications, we used a method of mild denaturation of the antigen-combining site 20. Both recombinant Ig were treated with 6 M urea as a dissociating agent and renaturated by dialysis in PBS. The binding activities of IgMκ-PER and IgMκ-Germ against the autoantigen panel were measured by ELISA before and after the dissociating treatment. While no significant binding changes occurred with IgMκ-Germ (Fig. 5A), the dissociating process induced the appearance of significant polyreactive binding activity by IgMκ-PER, without introduction of a significant change in the binding to tubulin (Fig. 5B).

2.7 Temperature variations are able to reconstitute the polyreactive binding specificity of IgMκ-PER

Previous work indicated that treatments such as acid pH 21 or denaturing agents 20 are able to increase polyreactive binding. To further evaluate the role of conformation modifications, both recombinant IgMκ-PER and IgMκ-Germ were tested in ELISA at two different temperatures (37°C and 4°C). Incubation of ELISA at 4°C did not modify the binding specificity of IgMκ-Germ (Fig. 6A), whereas diminution of temperature induced a significant increase in the binding of IgMκ-PER to actin, myosin, and DNA (Fig. 6B).

3 Discussion

Antigen binding by the B cell receptor is an important event in B cell activation and selection. It has long been assumed that only the hypervariable regions of the Ab are involved in ligand binding. However, it has been demonstrated that together with their ability to bind antigens via CDR, lg can react with SAg via the FR 1. The Fab specificity of the human V_H3 family against SpA involves a site on the variable region of the Ig H chain 8. One current opinion is that microbial SAg are involved in pathogenicity by stimulation of a high percentage of B cells, leading either to expansion or cellular apoptosis of these B cells 22–24.

To analyze the role of somatic mutations in the binding to SpA, we constructed two recombinant IgM antibodies (IgMκ-PER and IgMκ-Germ). The only difference between these two IgM constructions was the replacement of the mutated V_H3 segment in IgMκ-PER with its non-mutated germ-line counterpart in IgMκ-Germ. Recovery of the V_H3–73 germ-line sequence resulted in a striking recovery of SpA binding, suggesting that somatic mutations in the V_H domain are responsible for this phenomenon. Graille et al. determined the crystal structure of the SpA domain D complexed with the Fab fragment of a human IgM antibody and described the key contact residues from both partners in this interaction 15. In the Fab, the contact residues are located in the FR and interact with helix II and helix III of the SpA domain D. The site of SpA-Fab interaction is remote from the antigen-combining site and the H and L chain constant domains. The following variable H chain residues mediate this interaction: Gly-H15, Ser-H17, Arg-H19, Lys-H57, Tyr-H59, Lys-H64, Gly-H65, Arg-H66; Thr-H68, Ser-H70; Gln-H81; Asn-H82a and Ser-H82b. Among the 13 V_H residues involved as SpA contacts, V_H3 genes include frequent germ-line sequence variations only at position H57. It is not a core residue interface, whereas Lys, Ile, and Thr are permissive at this position 25, 26. The other 12 V_H residues in contact with domain D are highly conserved in members of this family. Indeed, amongthe 22 potentially functional human V_H3 gene segments, there are only two germ-line variations: the conservative change to Lys at H19 in V_H3–73 and the nonconservative change to Gly at H82a in V_H3–64. However, each of these genes is the source of less than 2% of the V_H3 Ig repertoire expressed in adults 27. From a structural point of view, a core of seven critical residues can be defined: Arg/Lys-H19, Gly-H65, Arg-H66, Thr-H68, Ser-H70, Gln-H81, and Asn-H82a. Apart from the requisite salt bridge at position H19, all other residues are virtually unexposed to solvents, making their replacement by a larger residue highly disruptive. Hence, this core of seven V_H residues forms the structural motif for SpA binding, and this conveys the restricted specificity for V_H3-encoded Ig and their homologues.

Despite important variations in binding affinities to SpA, the discrepancy between the germ-line and mutated domains in terms of SpA reactivity depends on a single mutation from Lys to Arg. These two residues are alkaline, and Arg is the residue observed at this position in the V_H3 sequences. It is presently unclear whether this single change could, by itself, account for the different binding affinities.

We simultaneously compared the anti-tubulin activity of IgMκ-PER and IgMκ-Germ. Although, the mutation pattern observed for the PER clone is in agreement with a selective process driven by the antigen, we could not substantiate significant affinity differences between these two constructions. In the case of IgMκ-PER, important changes occurred in the CDR2 segment: Ser to Arg, Lys to Asn, Asn to Glu, Tyr to Asp, and Val to Arg. This suggests that a significant proportion of the CDR is not involved in the binding to tubulin and remains free for other Ag binding. Thepossibility that hydrophobic charges could hide the contact with the antigen is ruled out by the hydrophilic nature of the new amino acids (Fig. 1A). Thus, the somatic mutations present in the V_H segment did not appear to play a major role in the binding to tubulin. In agreement with this possibility, Carayannopoulos et al. 28 showed that the affinities of recombinant antibodies using somatically mutated parental rheumatoid factor (RF) V_H segments displayed a decreased affinity for IgG-Fcγ compared to their closest germ-linecounterparts (germ-line revertants). This indicates that the antibodies posses the highest affinity in their germ-line configuration. However, if we assume that tubulin is the true antigen that selected this clone, it is likely that the Vκ domain and/or the V_H-CDR3 could play a central role in the anti-tubulin reactivity.

Interestingly, replacement of the mutated V_H3–73 segment with its germ-line counterpart resulted in the acquisition of a polyreactive binding specificity absent in the mutated form. To date it is unclear whether somatic mutations induce conformational changes sufficient to preclude polyreactive binding specificities. In agreement with this possibility, Wademeyer et al. compared the crystal structures of two antigen-antibody complexes; the first included the antibody in its germ-line configuration and the bound hapten, whereas the second comprised the mutated affinity-matured antibody and the same bound hapten 29. As a result of nine replacements by somatic mutations, affinity maturation induced a 30,000 times greater affinity for the hapten than the germ-line-encoded antibody. Significant changes in the configuration of the combining site occurred upon binding of the hapten to the germ-line-encoded antibody, whereas the hapten bound to the mature antibody by a key-and-lock mechanism. The reorganization of the combining site nucleated by hapten binding is further optimized by somatic mutations that occur as far as 15 Å from the bound hapten. These results suggest that the binding potential of the primary antibody repertoire may be largely expanded by the ability of germ-line antibodies to adopt several configurations of the combiningsite, with both antigen binding and somatic mutation stabilizing the configuration with optimal hapten fit. When the germ-line antibody binds the hapten, its structure is substantially altered. In contrast, there is only a slight alteration of the structure of the mature antibody upon hapten binding. Interestingly, the structure of the mature antibody, in either the bound or unbound state, closely resembles that of the germ-line antibody in the bound state. The same mechanism could be evoked to explain the binding variation of Ig with SpA.

It appears that the germ-line-encoded antibody is pluripotent with respect to the conformations it can assume upon contacting various antigens. The mutations may consolidate a specific conformation that is especially well suited for binding the antigen or SAg that drives the immune response. The set of proteins that constitutes the germ-line antibody repertoire must recognize an almost limitless array of potential antigens. If each of these proteins can adopt a variety of conformations, then the effective diversity of the germ-line repertoire is greatly expanded. Mutations that preserve antigen binding while modifying some of the alternative conformations are expected to occur more frequently at positions that lie distant from the site of hapten binding.

There is consistent evidence demonstrating that humoral immune diversity is mainly obtained through combinatorial and somatic mutation processes. However, crystallographic evidence of antibodystructures obtained with or without the ligands, showing the presence of conformation dimorphisms, favors the existence of an additional source of diversity at the level of antibody tertiary structure 30, 31. The presence of conformational isomers of the same antibody binding to the hapten with different affinities has also been suggested by the kinetic studies of Foote and Milstein 32 and has recently been substantiated by x-ray crystallographic studies by James et al. 33. In addition, Labrousse et al. 34 suggested that polyreactive antibodies possess a more "plastic" structure than monoreactive species and can thus more easily undergo alternative conformations when they are subjected to unusual microenvironments. Overall, these data indicate that the binding site of a single antibody is able to adopt different conformations and thereby bind unrelated antigens. The fact that temperature variation and dissociating treatment, which are both able to provoke conformational changes, induce the appearance of polyreactive binding is in agreement with this hypothesis.

In conclusion, our results suggest that the monoclonal IgMκ-PER was derived by somatic mutation from a natural polyreactive antibody, which lost its SAg and polyreactive binding capacities following this process. This finding lends support to the hypothesis that polyreactive antibodies are a product of the germ-line repertoire and constitute a template upon which Ag-driven selection and somatic mutation operate to derive highly specific immune antibodies 35. The study of Naparstek et al. 36 provided evidence supporting this view. In the present case, the key mutation may have randomly occurred during the development of a malignant clone. Alternatively, the mutation may have preceded the malignant transformation in a clone already engaged in a specific immune response. A fascinating hypothesis would be that a staphylococcal infection was involved in this process and that this clone escaped SAg-induced apoptosis through a mutational process. If true, this case could constitute a new type of immune selection of B cells induced by SAg.

4 Materials and methods