The role of the VL- and VH-segments in the preferential reassociation of immunoglobulin subunits

Molecular Immunology, Vol. 23, No. 5, pp. 5033510, 1986 Printed in Great Britain

0161-5890/86 $3.00 +O.OO Pergamon Press Ltd

THE ROLE OF THE V,- AND V,-SEGMENTS IN THE PREFERENTIAL REASSOCIATION OF IMMUNOGLOBULIN SUBUNITS* PAUL A. HAMEL, MICHEL H. KLEIN and KEITH J. DORRINGTONt Department

of Biochemistry,

University

of Toronto,

Toronto,

Canada

M5S IA8

(Received 29 July 1985; accepted in reviesed form 15 October 1985) Abstract-Competitive reassociation experiments, in which equimolar amounts of two different L-chains were allowed to compete for a limiting amount of H-chain, were performed to assess the role of the V,and _I,-segments on the ability of an L-chain to compete. Using H- and L-chains from the murine anti-phosphorylcholine (PC) myelomas, TEPC15, MOPC167 and MCPC603, and a series of V,21 L-chains, it was found that the V,21 L-chains competed uniformly better than the anti-PC L-chains, when the anti-PC H-chains were used, despite any differences in the J-segments of the competing L-chains. In addition, when the anti-PC L-chains, which all employ identical J,-segments but very diverse V,-segments, were in competition against each other, a hierarchy of competitive ability existed which was independent of whether the chains were autologous or heterologous and independent of antigen binding activity. Competitive reassociation experiments between the V,21 and anti-PC L-chains were also performed using the heterologous anti-lysozyme monoclonal HyHEL-10 H-chain or the anti-galactan 5539 H-chain, and it was found that the relative competitive ability of the V,21 L-chains with respect to the anti-PC L-chains was dependent on which H-chain was employed. The results suggested that the main factor favouring preferential reassociation by any particular L-chain was the V,-segment and that the effects of the J,-segment could not be observed where a high degree of diversity in the V-segments existed. Furthermore, while the results implied that specific pairs of V,- and V,-domains had a higher affinity for each other, this was not a necessary criterion in the formation of autologous pairs of H- and L-chains as demonstrated by the preferential heterologous reassociation of the V,21 L-chains over the autologous anti-PC L-chains. These results were consistent with the independent, random rearrangement of immunoglobulin H- and L-chain V-domain gene segments and predict that the hypothetical repertoire of antibodies is not limited by the selection of specific pairs of high-affinity V,-V, domains.

INTRODUCTION The complete transcriptional unit coding for immunoglobulin heavy (H) and light (L) chain variable domains arises by the juxtaposition of several gene segments (V, D and J in the H-chain, V and J in the L-chain) in the pre-B cell (Tonegawa et al., 1978; Weigert et al., 1978; Early et al., 1980). While the order in which the segments rearrange appears to be sequential, the H-chain locus followed by the kappa and subsequently the lambda locus (Alt et al., 1981, 1984; Coffman and Weissman, 1983), there is no direct evidence that the selection of the genetic elements used for the Vn- and V,-domains is linked. However, in competitive reassociation experiments in which equimolar amounts of an autologousf and a heterologous L-chain were allowed to compete in vitro for a limiting amount of H-chain, it has been previously found that the autologous L-chain preferentially recombined in 80% of the cases (Roholt et al., 1964; Grey and Mannik, 1965; P&al and Fouge*Supported by the Medical Research Council of Canada (grant MT 4259). tAuthor to whom correspondence should be addressed. SAutologous refers to H- and L-chains derived from the same immunoglobulin molecule. Heterologous refers to H- and L-chains from different immunoglobulin molecules.

reau, 1976). This was apparently not due to differences in the rate of reassociation of the competing chains (Bunting et al., 1977; Hamel et al., 1984) nor to differences in the L-chain constant regions (Grey and Mannik, 1965; Prival and Fougereau, 1976). Further, in studies using either the murine anti-DNP myeloma (Olander and Little, 1975) or anti-fluorescein hybridoma (Kranz and Voss, 1981) systems, it was suggested that preferential autologous reassociation correlated with antigenbinding activity. The specificity of autologous interactions has also been demonstrated using human myeloma Vn- and V,-domain fragments in which it was found that in the absence of Cnl or C, domains, respectively, heterologous V-domains had an affinity for the L-chains or Fd’ fragments several orders of magnitude less than the autologous V-domains (Klein et al., 1979; Horne et al., 1982). Taken together, these results might suggest that specific sets of Vn- and V,-gene segments are selected during rearrangement to produce pairs of domains with higher affinity than would be achieved if random rearrangement occurred. It would also appear that these unique pairs are a necessary condition for the formation of an antigen-binding site. If this were true however, there would be a large reduction in the possible repertoire of antibodies and the requirement for a novel and, as yet, unknown mechanism for the

503

PAUL A. HAMEL et ul.

504

linked expression of these high-affinity variable segments. Using a series of murine myelomas, all of which employed the V,21 gene segment, it has previously been shown that the nature of the residue at position 96, the first amino acid of the J-segment, was primarily responsible for the competitive ability of these L-chains (Hamel et al., 1984). Specifically, L-chains with a hydrophobic side chain at this position recombined better with H-chains than with L-chains with polar groups in a competitive situation, regardless of the limited number of amino acid differences in the rest of the variable region. It was also shown that L-chains with Phe or Tyr side chains at residue 96 competed better than L-chains with Leu or Trp at this position. In the present study, these investigations have been extended to assess the influence of the J,-segment and very diverse V,-segments on the competitive ability of different L-chains in chain reassociation experiments. It was found that the ability of an L-chain to compete was determined primarily by the V,-segment of the variable domain and that the influence of the Jsegment was much less important. Furthermore, the relative binding ability of each L-chain appeared to be dependent on which Vu-segment was used in the H-chain variable domain. It was also found that, for the anti-PC antibodies, no correlation between the competitive ability of an L-chain and antigen-binding activity could be made. MATERIALS AND METHODS

Cell lines Cell lines producing the anti-phosphorylcholine IgA myelomas TEPClS (Tl5), MOPC167 (M167) and MCPC603 (M603) were kindly provided by Dr Michael Potter (NIH, Bethesda). Immunoglobulins The purified IgG immunoglobulins 2485, 4050, 7210 and Cl01 were kindly provided by Dr Martin Weigert (Fox Chase Cancer Center, Philadelphia). The purified IgGl HyHEL-10 and IgA 5539 were obtained from Dr Sandra Smith-Gill (NIH). Myeloma tumors were grown in pristane-primed, BALB/c mice (Charles River). The myeloma proteins were affinity purified by passing the ascites fluid over para-aminophenylphosphorylcholine (Sigma) coupled to Affigel-10 (Bio-Rad) according to the method of Young et al. (1985). The bound proteins were then eluted with 0.1 M phosphorylcholine (Sigma) in 20mM Tri-HCI (Sigma), pH 8.5, dialyzed exhaustively against 20 mM Tris, pH 8.5, to remove all traces of the ligand and the purity of the proteins was confirmed by SDS-polyacrylamide electrophoresis (Laemmli, 1970). Competitive recombinations H- and L-chains were prepared by mild reduction with 2 mM dithioerythritol (Sigma) for 1 hr at room

temp followed by alkylation with 22 mM iodoacetamide (Sigma) for 2 hr. To label L-chains, alkylation was performed using either t3H]- or [‘4C]iodoacetic acid (Amersham). The sp. acts were typically 20.0 mCi/mmol for the ‘H L-chains and 2.0 mCi/mmole for the 14C L-chains. The proteins were dialyzed against a dissociating buffer, 1.0 M propionic acid, 4.5 M urea, 50 mM NaCl (PUS), and the chains separated by size-permeation chromatography using a Waters HPLC system on TSK 4000SW and 2000SW columns run in series (each 7.5 x 600 mm, LKB) and equilibrated in PUS. Competitive reassociations were performed by adding equimolar amounts of an H-chain, labelled autologous L-chain and labelled heterologous L-chain together in the dissociating buffer (typically 1.5 nmol of each, total vol. of 750~1) and allowed to recombine by dialyzing exhaustively against distilled water, followed by a 20mM sodium acetate buffer, pH 7.2. Finally, cont. PBS was added dropwise to bring the final concn to 50 mM phosphate, 0.5 M NaCl, pH 8.3. The reassociated immunoglobulins were separated from the excess L-chain by immunoadsorption as follows. An excess of rabbit antiserum, raised against murine a-H-chain (20 ~1, Daymar Laboratories), was added to the recombinant mixture and incubated for 1.5 hr at room temp. Then 200 ~1 of a 10% formaldehyde-fixed Staphylococcus aureus (Staph A) suspension (kindly provided by Dr G. A. Levy) was added and incubated for 1 hr at room temp. The Staph A was centrifuged and washed 4 times in PBS, 0.5% Triton X-100 to remove trapped and non-specifically bound L-chain. When y-Hchains were used in the competitive recombination experiments, the Staph A suspension was added directly without using a primary anti-mouse y-Hchain antiserum. It was determined that the amount of contaminating free L-chain in the Staph A pellet was less than 3% of the total L-chain analyzed in each experiment. The radioactivity in the supernatant and the pellet was determined by liquid scintillation counting. Molar amounts of the 3H and 14C L-chains in the recombinant were determined from the sp. act. of the isolated L-chains and the amount of autologous L-chain was expressed as a percentage of the total L-chain in the recombinant fractions. As previously determined, preferential reassociation was judged not to have occurred if the amount of the autologous L-chain in the recombinant represented between 45 and 55% of the total L-chain in the recombinant (Hamel et al., 1984). Heterologous antigen-binding activity Unlabelled H- and L-chains were prepared as described above. Heterologous or autologous recombinant molecules were then prepared by adding one H-chain and one L-chain together (typically 0.5 mg of each) in the dissociating buffer and allowed to reassociate as described above. Anti-PC binding activity was assessed by passing the reassociated

505

The role of the V,- and V,-segments

mixture over a PC column (see above) and washing with PBS. Bound molecules, those which showed anti-PC activity, were eluted using 0.1 M PC in PBS and the protein fractions monitored by absorbance at 28Onm. Yields of the reassociated molecules were typically in the range from 50 to 65%, due mainly to aggregation and precipitation of the H-chains.

Table 2. Anti-PC activity of recombinant immunoglobulins H-chain ~_.__ Tl5 Ml67 M603 715 TIS Ml67 Ml67 M603 M603

RESULTS To determine if differences in the relative competitive ability of L-chains which had identical J,segments but very diverse V,-segments could be observed, competitive reassociation experiments were performed between the anti-PC L-chains T15, Ml67 and M603 (see Fig. 1 for sequences). The results of these experiments are summarized in Table 1. The results indicated that a hierarchy existed in the chains to compete (i.e. ability of these T15 > M603 > M167) and that the relative competitive ability of each L-chain was independent of whether it was the autologous or heterologous Lchain. These results might be expected, despite the use of identical J,-segments in the competing Lchains, due to the high degree of diversity in the different V,-segments (T15 uses the V,22 L-chain isotype, Ml67 the V,24 isotype and M603 the V,8 isotype). The differences between the V,-segments extend not only through the CDR regions but also into framework regions which have been shown to participate in the non-covalent association of the Vuand V,-domains (Saul et al., 1978; Novotny et al., 1983).

It has been suggested previously that there was a correlation between the preferential reassociation of autologous H- and L-chains and the ability to bind antigen. If this could be extended generally then the results of the competitive reassociation assays between the anti-PC L-chains might suggest that some heterologous recombinants, particularly those produced from the T15 L-chain and the Ml67 or M603 H-chain, might be able to bind PC. Thus, recombinant immunoglobulins were produced using all combinations of anti-PC H- and L-chains and their anti-PC binding activity tested by assessing their retention on a PC column. The results are summarized in Table 2. Table

I. Competitive reassociations” between anti-PC L-chains

L-chain TIS Ml67 M603 Ml67 M603 Tl5 M603 T15 MI67

The recombinant molecules were applied to a PC column, washed with buffer and any bound reassociated molecules were then eluted in 0.1 M PC. Protein in the collected fractions were monitored by absorbance at 280 nm.

As expected, all of the autologous recombinants bound to the PC column and were specifically eluted when PC was added. However, none of the heterologous recombinants showed any binding to the PC column. This lack of binding activity could not be attributed to inactive H- or L-chains since they were prepared in the same manner as the autologous recombinants, which were shown to have anti-PC activity, although the production of a very low affinity site may have gone undetected by this method. Thus, for this set of proteins, there does not appear to be any correlation between the relative competitive ability of a particular L-chain and the ability to form an antigen-binding site. It has been shown previously that, in L-chains which have a limited amount of diversity in the V,-segment, the nature of residue 96, the first residue of the J-segment, was important in dictating the competitive ability of an L-chain (Hamel et al., 1984). In order to assess the contribution of this residue when a high degree of diversity existed in the V,-segment, a number of V,21 L-chains were used, which had different J,-segments (see Fig. 1 for sequences), in competitive reassociation experiments against the anti-PC L-chains. The results of these experiments are summarized in Table 3. The results clearly indicate that the V,21 L-chains were better competitors than any of the anti-PC L-chains when the anti-PC H-chains were employed. Furthe~ore, in contrast to our previous results, differences in the J,-segments of the V,21 L-chains Table 3. Competitive reassociations” between anti-PC and V,21 L-chains Heterologous

Heterologous L-chain Autolcgous L-chain T15 Ml67 M603

T15

17.8 * 9.9 21.2 & 10.3

Ml67

Binding -, + + + _ _ _

M603

._ ._-_______ 83.0 + 13.2 84.0 * 8.2 41.3 69.0

“Values represent the percentage of autologous L-chain in the recombinant and are the mean of three experiments, where standard deviations are indicated, or the average of two experiments. Values between 45 and 55% suggest that no significant preference for either L-chain exists.

Autologous L-chain Tl5 Ml67 M603

L-chain

4050

Cl01

7210

34.0 + 8.4 34.1 + 0.4 36.4 + 3.7

33.5 I 9.6 25.4 + 3.9 32.0 & 2.8

35.4 38.0 41.3

“Values represent the percentage of the autologous L-chain in the recombinant and are the mean (f standard deviation) of three or four experiments. Values for experiments with the 7210 L-chain are the average of two experiments. Values between 45 and 55% suggest that no significant preference for either L-chain exists.

506

PAUL A. HAMELet al.

The role of the V,- and V,-segments

507

were not reflected by any significant differences in the competitive ability of these L-chains. Specifically, the 4050, Cl01 and 7210 L-chains, which employ Leu, Tyr and Trp, respectively, at residue 96, all competed to the same relative degree against the anti-PC Lchains. To rule out the possibility that the differences in the competitive ability of these chains were due to differences in the rates of refolding of the competing chains, the competition experiments between M 167 and 4050 or Cl01 were repeated but the chains were pre-dialyzed into PBS before being mixed. When the recombinants were analyzed, the same percentage of autologous L-chain in the recombinant was obtained as with the usual procedure (data not shown). In addition, sedimentation velocity experiments demonstrated that the murine kappa L-chains employed were monomeric at the concn used in the recombination experiments (data not shown), thus ruling out the possibility that differences in competitive ability were due to certain L-chains forming dimers. Thus, it appeared that when a high degree of diversity existed in the V,-segment of the competing L-chains, the effect of differences in the J,-segment was not significant. Furthermore, in every case, the heterologous V,21 L-chain competed better than any of the autologous anti-PC L-chains, suggesting that there may not be any correlation between the preferential reassociation of autologous H- and L-chains and the clonal expansion of a particular immunoglobulin or the ability of an immunoglobulin molecule to bind an antigen. The results in Table 3 also suggested that the Dand J-segments of the H-chain had little or no influence on the relative competitive ability of these L-chains. Figure 2 shows that the D- and J-segments of the anti-PC H-chains were quite diverse. However, the diversity was not reflected by any significant differences in the relative competitive abilities of the V,21 L-chains with respect to the anti-PC L-chains. The preferential heterologous reassociation demonstrated in Table 3 suggested to us that the Vsegment of the Vu-domain might have an influence on the relative competitive ability of two competing L-chains. Thus, the same pairs of V,21 and anti-PC L-chains were tested in the competitive assay using H-chains which showed a high degree of diversity in their V-segments. Recently, it has been demonstrated that the entire known repertoire of V, gene segments in mouse could be divided into seven major families based on their ability to bind to a panel of genetic probes (Brodeur and Riblet, 1984) and based on amino acid homology (Dildrop, 1984). Thus two H-chains were chosen, derived from the antilysozyme hybridoma HyHEL-10 and the antigalactan myeloma J539, which each had Vu-segments from different families than the Vu-segment of the anti-PC H-chain (for sequences see Fig. 2). The results of the competitive reassociation experiments

PAUL A. HAMEL etal.

508

Table 4. Competitive reassociations” between anti-PC and V,Zl L-chains mine H-chains derived from HvHEL-IO and 5539 HyHEL-IO

TIS Ml67 M603

H-Chain

5539 H-Chain

2485

Cl01

2485

Cl01

20.5 2.2 7.3

15.4 2.5 I.5

69.8 33.3 51.5

64.2 24.6 36.7

“Values represent the percentage of either the Tl5, Ml67 or M603 L-chain in the recombinant and are the average of two experiments. Values between 45 and 55% suggest that no significant preference for either L-chain exists.

between the V,21 and anti-PC L-chains using either the HyHEL-10 or 5539 H-chain are summarized in Table 4. The results in Table 4 demonstrated dramatically that the relative competitive ability of two different L-chains was dependent on which Vn-segment was present in the H-chain. When the HyHEL-10 H-chain was used in the competition experiment between any of the V,21 L-chains and any of the anti-PC L-chains, the V,21 L-chains showed a very strong preferential reassociation. However, when the 5539 H-chain was used, the relative competitive ability of the same pairs of L-chains was quite different. Specifically, the T15 L-chain, which competed poorly against either 2485 or Cl01 when the HyHEL-10 H-chain was used, showed almost a complete reversal in its relative competitive ability when the 5539 H-chain was used. Furthermore, both the Ml67 and M603 L-chains, which showed a striking lack of competitive ability when the HyHEL-10 H-chain was used, had a significantly better competitive ability when the 5539 H-chain was employed. The reversal in the competitive ability of these chains also demonstrated that preferential reassociation of these L-chains was not due to differences in the kinetics of refolding of the competing chains. It was also observed that both the Cl01 (Tyr-96) and 2485 (Leu-96) L-chains competed to the same degree, against any one of the anti-PC L-chains, despite differences in their J,-segments. This was consistent with the data in Table 3, where the presence of a Leu, Tyr or Trp residue at position 96 of 4050, Cl01 and 7210, respectively, had no observable effect on the relative competitive ability of these chains. DISCUSSION

In this study, the influence of discrete gene products of the V,- and V,-domains on the competitive ability of L-chains to bind to a H-chain in a chain reassociation assay has been assessed. It was apparent that the major influence dictating the preferential reassociation of a particular L-chain was the V,-segment. The results in Table 1 demonstrated that both autologous or heterologous preferential reassociation of a particular L-chain occurred despite identical J,-segments in both of the competing Lchains. The major influence of the V,-segment was further demonstrated in Tables 3 and 4, where

differences in the J,-segments of the competing chains were not reflected by significant differences in the competitive ability of those chains. These results were consistent with the three-dimensional structure of the Fv portion of immunoglobulins (Saul et al., 1978; Novotny et al., 1983). It has been shown that the vast majority of contacts between the interacting surfaces of the V,- and V,-domains were found in the betapleated sheets of the V-segments of the H- and L-chains. Thus, it might have been predicted that the large number of differences between the different L-chain V-segments would be predominant in dictating the competitive ability of a particular L-chain. This may appear to be at variance with our previous results where, using a series of murine V,21 L-chains, it was shown that the J,-segment was dominant in dictating the preferential reassociation of a particular L-chain. However, the majority of differences between the competing chains in that series of competitive reassociations were found in the CDR regions which have little or no participation in the interaction between beta-pleated sheets of the V,and Vn-domains. Since the sequences in the regions of the beta-pleated sheets of the different V,21 Lchains were virtually identical, it would have been expected that the L-chains would behave similarly and that the influence of a single change in the J-segment could be observed by a difference in the L-chains’ competitive ability. It has also been demonstrated that the relative competitive ability of an L-chain is dependent on the nature of the V-domain of the H-chain. Specifically, the Vn-segments of either the anti-PC H-chains, the HyHEL-10 H-chain or the 5539 H-chain, which have less than 54% homology at the protein level (Dildrop, 1984) caused dramatic changes in the relative competitive ability of the competing chains. Furthermore, the results in Tables 1 and 3 demonstrated that differences in the D- and J-segments of the V,-domain were not reflected by differences in the competitive ability of these L-chains. Since the number of sequence differences in the V,-segments of the T15, M 167 and M603 H-chains were very small it was expected that differences in the relative competitive ability of the L-chains would have been observed if the D- and J-segments had important or extensive contacts with the V,-domain. However, no significant differences were observed suggesting that the D- and J-segments in the H-chain have little or no role in the specific interactions between V,- and V,-domains. Based on two previous studies using murine antigen-binding myeloma (Olander and Little, 1975) or hybridoma (Kranz and Voss, 1981) systems, it has been suggested that a correlation existed between the preferential reassociation of the autologous L-chains and the ability to bind antigen. Expressed differently, one of the conditions for the formation of a specific, functional antigen-binding site was a unique high affinity between the Vi,- and V,-domains. This model may be generalized further, to include earlier data

The role of the V,- and Vu-segments

which employed human myeloma proteins, to suggest that there was a correlation between preferential reassociation of the autologous L-chain and the clonal expansion, either by normal or neoplastic mechanisms, of an immunoglobulin producing cell. A mechanism to recognize and expand only pairs of Hand L-chains which exhibited preferential autologous reassociation (i.e. a unique high affinity between autologous V,- and V,-domains) is difficult to conceive unless selection on the basis of antigen-binding activity was included, as was suggested by Kranz and Voss (1981). However, the data presented in this paper suggest that this model may not be generally applicable. While the Vu-domain of the anti-PC H-chains were nearly identical, T15, M 167 and M603 all had unique V,-segments which were predicted, and subsequently shown, to compete quite differently. It was also found that there was strict specificity between the Vu- and V,-domains with respect to the formation of an antigen-binding site, as was demonstrated by the lack of activity of the heterologous recombinants (Table 2) analogous to the results of Kranz and Voss (1981). Thus, in the anti-PC antigen-binding system, there was no correlation found between antigen-binding activity and the preferential reassociation of the autologous Lchains. Furthermore, when completely unrelated Lchains (i.e. the V,21 L-chains) were competed against any of the autologous anti-PC L-chains, preferential heterologous reassociation was observed. Thus in this paper a unique high-affinity association between autologous Vu- and V,-domains is suggested, reflected by preferential autologous reassociation in a competitive L-chain reassociation assay, was not a necessary condition for the formation of an antigenbinding site or the clonal expansion of a particular immunoglobulin producing cell. These results were also consistent with the unlinked selection and expression of V, and V, gene segments and, furthermore, would not limit the repertoire of potential antibodies to those which showed preferred association of autologous chains. The fact that pairs of V-domains have different mutual affinities implies an important role for the C-domain adjacent to the V-domains (C, and C,l). Previous studies from this laboratory have demonstrated that isolated V,- or Vn-domains which have very low affinities for either H- or L-chains respectively, would bind with considerably higher affinity in the presence of the C-domains (Klein et al., 1979; Horne et al., 1982). Recently, it has also been demonstrated that idiotypic expression in a human myeloma was modulated by the C,l domain (Rinfret et al., 1985). Taken together with the results presented here, a major function of the constant domains adjacent to the V-domain would be to modulate the interaction of the V-domains, enhancing their association but maintaining or facilitating the expression of unique structural characteristics exhibited by different families of Vnand

509

V,-domains. The ability to enhance the association of the V-domains is particularly important when somatic mutation of the rearranged variable-region genes is considered (Gearhart et al., 1981; Tonegawa, 1983). The structural changes produced by introducing essentially random mutations in the rearranged variable gene would be predicted to require some compensatory mechanism to overcome any decreases in V,-V, affinity. This mechanism could lie in the ability of the constant domains to enhance the association of the variable domains, as long as gross aberrations in the three-dimensional structure of the domains were not introduced as a result of the somatic mutation, thereby allowing a potentially greater repertoire of immunoglobulins to exist through the somatic-mutation mechanism. Acknowledgements-The authors would like to thank Dr Heather Boux and John Hagan (Connaught Research Institute, Toronto) for producing the ascites fluid from the T15, Ml67 and M603 myelomas. They would also like to thank Dr David Isenman for helpful discussions.

REFERENCES Ah F., Rosenberg N., Lewis S., Thomas E. and Baltimore D. (1981) Organization and reorganization of immunoglobulin genes in A-MuLV-transformed cells: rearrangement of heavy but not light chain genes. Cell 27, 381-390. Ah F. W., Yancopoulos G. D., Blackwell T. K., Wood C., Thomas E., Boss M., Coffman R., Rosenberg N., Tonegawa S. and Baltimore D. (1984) Ordered rearrangement of immunoglobulin heavy chain variable segments. EMBO J. 3, 1209-1219. Brodeur P. H. and Riblet R. (1984) The immunoglobulin heavy chain variable region (Igh-V) locus in the mouse. I. One hundred Igh-V genes comprise seven families of homologous genes. Eur. J. Immun. 14, 922-930. Bunting P. S., Kells D. I. C., Kortan C. and Dorrington K. J. (1977) Non-covalent interactions between heavy and light chains of immunoglobulin G. Role of light chain variable-region subgroup. Immunochemistry 14, 45-52. Coffman R. L. and Weissman 1. L. (1983) Immunoglobulin gene rearrangement during pre-B cell differentiation. J. molec. cell. Immun. 1, 31-38. C. de Preval and Fougereau M. (1976) Specific interaction between Vu and V, regions of human monoclonal immunoglobulins. J. molec. Biol. 102, 657-678. Dildrop R. (1984) A new classification of mouse V, sequences. Immun. Today 5, 85-86. Early P., Huang M., Davis M., Calame K. and Hood L. (1980) An immunoglobulin heavy chain variable region is generated from three segments of DNA: Vu, D and J,. Cell 19, 981-992. Gearhart P. J.. rohnson N. D., Douglas R. and Hood L. (1981) IgG antibodies to phosphorylchohne exhibit more diversity than their IgM counterparts. Nature, Lond. 291, 29-34. Grey H. M. and Mannik M. (1965) Specificity of recombination of H and L chains from human yG-myeloma proteins. J. exp. Med. 122, 619-632. Hamel P. A., Isenman D. E., Klein M. H., Luedtke R. and Dorrington K. J. (1984) Structural basis for the preferential association of autologous immunoglobulin subunits: role of the J region of the light chain. Molec. Immun. 21, 271-283. Horne C., Klein M., Polidoulis I. and Dorrington K. J. (1982) Noncovalent association of heavy and light chains

510

PAUL A. HAMEL et al.

of human immunoglobulins. III. Specific interactions between Vu and V,. J. Immun. 129, 66&664. Kabat E. A., Wu T. T., Bilofsky H., Reid-Miller M. and Perry H. (1983) Sequences of Proteins ofImmunological Interest. U.S. Department of Health and Human Services. Klein M., Kortan C., Kells D. I. C. and Dorrington K. J. (1979) Equilibrium and kinetic aspects of the interaction of isolated variable and constant domains of light chain with the Fd’ fragment of immunoglobulin G. Biochemistry 18, 1473-1481. Kranz D. M. and Voss E. W. (1981) Restricted reassociation of heavy and light chains from hapten specific monoclonal antibodies. Proc. natn. Acad. Sci. U.S.A. 78. 5807-5811. Laemmli U. K. (1970) Cleavage of structural proteins during the assembly of the head of bacteriophage T4. Nature, Lond. 227, 179-246. Novotny J., Bruccoleri R., Newell J., Murphy D., Haber E. and Karplus M. (1983) Molecular anatomy of the antibody binding site. J. biol. Chem. 258, 14433314437. Olander J. and Little J. R. (1975) Preferential homologous recombination of H- and L-chains from mouse myeloma proteins which bind DNP ligands. Immunochemistry 12, 383-387. Rinfret A., Home C., Dorrington K. J. and Klein M. (1985) Noncovalent association of heavy and light chains of human immunoglobulins. IV. The role of C,l and C,

domains in idiotypic expression. J. Immun. 135, 25742581. Roholt 0. A., Onoue K. and Pressman D. (1964) Specific combination of H and L chains of rabbit y-globulins. Proc. natn. Acad. Sci. U.S.A. 51, 173-178. Rudikoff S. (1983) Immunoglobulin structure-function correlates: antigen binding and idiotypes. In Contemporary Topics in Molecular Immunology (Edited by F. P. Inman and T. J. Kindt), Vol. 9, pp. 1699209. Plenum, New York. Saul F. A., Amzel L. M. and Poljak R. J. (1978) Preliminary refinement and structural analysis of the Fab fragment from human immunoglobulin New at 2.0 A resolution. J. biol. Chem. 253, 585-597. Tonegawa S., Maxam A. M., Tizard R., Bernard 0. and Gilbert W. (1978) Sequence of a mouse germ-like gene for a variable region of an immunoglobulin light chain. Proc. natn. Acad. Sci. U.S.A. 75, 1485-1489. Tonegawa S. (1983) Somatic generation of antibody diversity. Nature, Lond. 302. 575-581. Weigert M., Gatmaitan L.,‘Loh E., Schilling J. and Hood L. (1978) Rearrangement of genetic information may produce immunoglobulin diversity. Nature, Land. 276, 785-790. Young N. M., Williams R. E. and Claflin J. L. (1985) The circular dichroism of phosphocholine-specific mouse hybridoma and myeloma proteins: unusual properties of the hybridoma protein 101.6G6. Molec. Immun. 22, 305-3 I I,