Structural predictions of the binding site architecture for monoclonal antibody NC6.8 using computer-aided molecular modeling, ligand binding, and spectroscopy.

Monoclonal antibody NC6.8 binds the superpotent sweetener ligand N-(p-cyanophenyl)-N'-(diphenylmethyl) guanidineacetic acid with high affinity (Kd = 53 nM). Using computer-aided molecular modeling and several experimental techniques, such as competitive ligand binding, absorbance spectroscopy, and fluorescence spectroscopy, we have predicted the structure of the variable domain fragment (Fv) and identified the key residues in the combining site of the antibody. We have identified nine specific amino acids as being involved in ligand recognition and complexation. Most notable are H:33W, which is responsible for ligand-induced tryptophan fluorescence quenching, H:56R, which forms a salt bridge with the carboxylate moiety of the ligand, and L:34H, which, deep in the binding site, interacts with the cyanophenyl portion of the ligand. Two residues located deep in the putative binding pocket, H:35E and H:50E, provide the negatively charged potential for interaction with the protonated aryl nitrogen and the positive guanidinium group. These modeling predictions were made before the solution of high-resolution structures of the native Fab (2.6 A) and the Fab-ligand complex (2.2 A). Comparisons between the theoretical model and experimental native and liganded Fab structures are made.     

Computer-aided molecular modeling of the binding site architecture for eight monoclonal antibodies that bind a high potency guanidinium sweetener

Computer-aided molecular modeling of the antibody binding site of eight different monoclonal antibodies (mAb) that bind the intense sweetener ligand (N-(p-cyanophenyl)-N'-diphenylmethyl) guanidine acetic acid was completed using canonical loop structures and framework regions from known immuno globulins as “parent structures” for the molecular scaffoldings. The models of the fragment variable (Fv) region of the mAb were analyzed/or the presence and location of residues predicted to be involved in ligand binding. Several binding site tryptophan residues in these models were located in positions that support previous flurospectroscopic observations of the mAb-ligand complexation. Computer-aided renderings of the electrostatic potential at the van der Waals surface of the Fv region were compared and found to be consistent with the ligand binding specificity profiles for the different mAb. The Fv model of mAb NC6.8 was consistent with the binding site features determined in the Fab structure recently solved by X-ray diffraction techniques. These Fv models should provide an adequate basis for site-directed mutagenesis experiments in order to characterize interactive motifs in the mAb binding site.     

Immunological and structural characterization of a high affinity anti-fluorescein single-chain antibody

Single-chain antibody of the (NH2) VL-linker-VH (COOH) design, was constructed based on prototype high affinity anti-fluorescein monoclonal antibody (mAb) 4-4-20. Purified single-chain antibody (SCA) 4-4-20/212 was studied relative to Ig mAb 4-4-20 in terms of ligand binding, kinetics, idiotypy, metatypy, and stability in denaturing agents. Ligand-binding data correlated with metatypic relatedness of the liganded site. Anti-metatypic reagents reacted preferentially with the liganded conformer of the 4-4-20 antibody active site and were unreactive with free ligand and the non-liganded (idiotypic) state. All results were consistent with the conclusion that SCA 4-4-20/212, with a 14-amino acid linker folded into a native conformational state that closely simulated the prototypical mAb. Furthermore, GndHCl unfolding and refolding studies demonstrated H and L chain variable domain intrinsic stability between SCA 4-4-20/212 and a 50 kDa antigen-binding fragment were nearly identical. This suggested CH1 and CL domain interactions may be more prevalent in V region molecular dynamics than structure.     

The increased flexibility of CDR loops generated in antibodies by Congo red complexation favors antigen binding

The dye Congo red and related self-assembling compounds were found to stabilize immune complexes by binding to antibodies currently engaged in complexation to antigen. In our simulations, it was shown that the site that becomes accessible for binding the supramolecular dye ligand is located in the V domain, and is normally occupied by the N-terminal polypeptide chain fragment. The binding of the ligand disrupts the beta-structure in the domain, increasing the plasticity of the antigen-binding site. The higher fluctuation of CDR-bearing loops enhances antigen binding, and allows even low-affinity antibodies to be engaged in immune complexes. Experimental observations of the enhancement effect were supported by theoretical studies using L lambda chain (4BJL-PDB identification) and the L chain from the complex of IgM-rheumatoid factor bound to the CH3 domain of the Fc fragment (1ADQ-PDB identification) as the initial structures for theoretical studies of dye-induced changes. Commercial IgM-type rheumatoid factor (human) and sheep red blood cells with coupled IgG (human) were used for experimental tests aimed to reveal the dye-enhancement effect in this system. The specificity of antigen-antibody interaction enhanced by dye binding was studied using rabbit anti-sheep red cell antibodies to agglutinate red cells of different species. Red blood cells of hoofed mammals (horse, goat) showed weak enhancement of agglutination in the presence of Congo red. Neither agglutination nor enhancement were observed in the case of human red cells. The dye-enhancement capability in the SRBC-antiSRBC system was lost after pepsin-digestion of antibodies producing (Fab)2 fragments still agglutinating red cells. Monoclonal (myeloma) IgG, L lambda chain and ovoalbumin failed to agglutinate red cells, as expected, and showed no enhancement effect. This indicates that the enhancement effect is specific.     

Proteolytic dissection of a hapten binding site

IgG Gar, a human myeloma protein that binds riboflavin with a high affinity, was used to derive variable region fragments from the heavy chain and the light chain. Riboflavin binding ability of the active site generated by V(H) and light chain and the active site generated by V(H) and V(L) was compared to riboflavin binding by the F(ab) fragment. The riboflavin binding ability of the F(ab) fragment is the same as the intact molecule, while the binding ability of the active site formed by V(H) and light chain is lowered by two to three orders of magnitude, indicating that the removal of C(H1) domain decreases the interaction between riboflavin and the amino acids that is important in tight binding of riboflavin. Removal of the third hypervariable region and the constant region domain from the light chain further lowers the binding constant by one order of magnitude. The results indicate that the V(H) and V(L) segments of IgG Gar can reconstitute a riboflavin binding site. The decrease in affinity probably reflects a decrease in the rigidity with which the hypervariable loops are held together to place the contact amino acid residues in optimal contact with the hapten.     

The effect of isotype and the J kappa region on antigen binding and idiotype expression by antibodies binding alpha (1----6) dextran

Gene transfection and expression techniques have been used to produce three antibodies specific for alpha (1----6) linked dextran B512 with altered isotypes and J kappa regions. Expression of the L chain V region joined to J kappa 4 or J kappa 5 instead of to J kappa 2 reduced or abolished dextran binding. One antidextran with a reduced binding constant for dextran had the same combining site size as the parental mAb. Transfectoma Ig unreactive with dextran B512 did not bind to other class I or class II dextrans. Antibodies with J kappa 4-containing L chains expressed the 10.16.1 (anti-alpha(1----6) dextran) idiotype. In contrast variants expressing L chains with J kappa 5 lost idiotype expression, except when oligosaccharide is present on VH; all antibodies with J kappa 5 L chains continued to bind dextran but with reduced affinity. The presence of carbohydrate in VH may alter the conformation of both paratope and idiotope. Alteration of H chain isotype did not appear to significantly alter the ability of the antidextran to bind Ag; an exception may be that switching V regions to the IgM C region may decrease the apparent affinity for Ag.     

Binding of the neuroleptic drug haloperidol to a monoclonal antibody: refinement of the binding site molecular model using canonical structures

The ligand binding site of a monoclonal antibody (185), which binds the neuroleptic drug haloperidol, has been modelled using canonical structures and energy minimization techniques. This refined modelling protocol has allowed us to predict the variable region loop conformations. Three key residues, H:50(W), H:100a(D) and L:96(Y) appear to create the basis of the electrostatic, pi-pi stacking interactions and hydrogen bonding required for the high affinity binding site characteristics present in this antibody. The use of computer-aided graphics techniques and appropriate three-dimensional modelling permits inspection of the predicted molecular recognition features of the ligand binding site.     

Probing the human antibody repertoire to exogenous antigens. Characterization of the H and L chain V region gene segments from anti-hepatitis B virus antibodies.

Structural studies of human antibody V regions have been largely limited to those involving the fetal repertoire, autoantibodies, and malignant cell rearrangements, leaving the "normal" repertoire relatively unexplored. In this study we describe the nucleotide sequences of the H and L chain V regions of four antibodies specific for the surface Ag of the hepatitis B virus. Monoclonal cell lines were derived from healthy individuals who received standard immunizations with the serum-derived or recombinant hepatitis B virus vaccines by fusion of PBL to a heterohybridoma cell line, SPAZ-4. We utilized the polymerase chain reaction to amplify the H and L chain V regions for cloning and sequencing. The four antibodies express the following V region combinations: VHIII/V lambda V, VHIII/V kappa II, VHIV/V kappa I, VHV/V lambda III. When compared to germline genes with the closest sequence homology, all of the V regions appear to have undergone somatic mutation, ranging from 3.4 to 11.3% for the H chain, and 5.1 to 9.2% for the L chain. Analysis of the mutations shows them to be typical for an Ag-driven immune response.     

Molecular cloning of a bovine immunoglobulin lambda chain cDNA

A cDNA library of the bovine mammary gland constructed in pBR322 was screened by mRNA hybrid-selected translation and by differential hybridization. Several immunoglobulin (Ig) lambda light-chain clones were identified and sequenced. Nucleotide sequence comparison of bovine and human Ig lambda chains showed a high degree of homology for constant regions and for J regions. The amino acid (aa) sequence encoded by the constant region of the bovine Ig lambda chain cDNA contains 107 aa with differences at 24 aa positions from the human Ig lambda chain. Three complementarity-determining regions (CDR1,2,3) characteristic of the variable region of bovine Ig lambda chain cDNA can be distinguished. The bovine and human sequences display good homology in the framework region 3 (FR3) but only patches of homology throughout the FR2 region. The 5' end of the bovine Ig lambda chain cDNA fragment of clone 1-14E contains five stop codons: two in CDR1, one in FR1 and two in the hydrophobic prepeptide region. These data suggest that the Ig lambda mRNA of clone 1-14E is transcribed from the V lambda pseudogene.     

Molecular modeling of cardiac glycoside binding by the human sequence monoclonal antibody 1B3

The amino acid sequences of the heavy- and light-chain variable regions of the high-affinity human sequence antidigoxin monoclonal antibody 1B3 (mAb 1B3) were determined, and a structural model for the mAb's variable region was developed by homology modeling techniques. The structural model provided the basis for computationally docking digoxin and eight related cardiac glycosides into the putative binding site of mAb 1B3. Analysis of the consensus binding mode obtained for digoxin showed that the cardenolide moiety of digoxin is deeply embedded in a predominantly hydrophobic, narrow cavity, whereas the terminal, gamma-carbohydrate group is solvent-exposed. The docking results indicated that the primary driving forces for digoxin binding by mAb 1B3 are hydrophobic interactions with the digoxin steroid ring system and hydrogen bonds with the digitoxose groups. The binding model accounts for the experimentally observed variations in mAb 1B3 binding affinity for various structural analogs of digoxin used previously to develop a 3D structure-activity relationship model of drug binding (Farr CD, Tabet MR, Ball WJ Jr, Fishwild DM, Wang X, Nair AC, Welsh WJ. Three-dimensional quantitative structure-activity relationship analysis of ligand binding to human sequence antidigoxin monoclonal antibodies using comparative molecular field analysis. J Med Chem 2002;45:3257-3270). In particular, the hydrogen bond pattern is consistent with the unique sensitivity of mAb 1B3's binding affinity to the number of sugar residues present in a cardiac glycoside. The hydrophobic environment about the steroid moiety of digoxin is compatible with the mAb's reduced affinity for ligands that possess hydrophilic hydroxyl and acetyl group modifications in this region. The model also indicated that most of the amino acid residues in contact with the ligand reside in or about the three complementarity determining regions (CDRs) of the heavy chain and the third CDR of the light chain. A comparison of the 1B3 binding model with the crystal structures of two murine antidigoxin mAbs revealed similar binding patterns used by the three mAbs, such as a high frequency of occurrence of aromatic, hydrophobic residues in the CDRs and a dominant role of the heavy chain CDR3 in antigen binding.     

Mapping of a region of dengue virus type-2 glycoprotein required for binding by a neutralizing monoclonal antibody.

Envelope glycoprotein E of flaviviruses is exposed at the surface of the virion, and is responsible for eliciting a neutralizing antibody (Ab) response, as well as protective immunity in the host. In this report, we describe a method for the fine mapping of a linear sequence of the E protein of dengue virus type-2 (DEN-2), recognized by a type-specific and neutralizing monoclonal Ab (mAb), 3H5. First, an Escherichia coli expression vector containing a heat-inducible lambda pL promoter was used to synthesize several truncated, and near-full length E polypeptides. Reactivities of these polypeptides with polyclonal mouse hyperimmune sera, as well as the 3H5 mAb revealed the location of the 3H5-binding site to be within a region of 166 amino acids (aa) between aa 255 and 422. For fine mapping, a series of targeted deletions were made inframe within this region using the polymerase chain reaction (PCR). The hydrophilicity pattern of this region was used as a guide to systematically delete the regions encoding the various groups of surface aa residues within the context of a near-full-length E polypeptide by using PCR. The 3H5-binding site was thus precisely mapped to a region encoding 12 aa (between aa 386 and 397). A synthetic peptide containing this sequence was able to bind to the 3H5 mAb specifically, as shown by enzyme-linked immunosorbent assay. In addition, we show that rabbit Abs raised against the synthetic peptide of 12 aa were able to bind to the authentic E protein, and to neutralize DEN-2 virus in a plaque reduction assay.     

Identification of a binding site for ganglioside on the receptor binding domain of tetanus toxin

The carboxyl-terminal region of the tetanus toxin heavy chain (H(C) fragment) binds to di- and trisialylgangliosides on neuronal cell membranes. To determine which amino acids in tetanus toxin are involved in ganglioside binding, homology modeling was performed using recently resolved X-ray crystallographic structures of the tetanus toxin H(C) fragment. On the basis of these analyses, two regions in tetanus toxin that are structurally homologous with the binding domains of other sialic acid and galactose-binding proteins were targeted for mutagenesis. Specific amino acids within these regions were altered using site-directed mutagenesis. The amino acid residue tryptophan 1288 was found to be critical for binding of the H(C) fragment to ganglioside GT1b. Docking of GD1b within this region of the toxin suggested that histidine 1270 and aspartate 1221 were within hydrogen bonding distance of the ganglioside. These two residues were mutagenized and found also to be important for the binding of the tetanus toxin H(C) fragment to ganglioside GT1b. In addition, the H(C) fragments mutagenized at these residues have reduced levels of binding to neurites of differentiated PC-12 cells. These studies indicate that the amino acids tryptophan 1288, histidine 1270, and aspartate 1221 are components of the GT1b binding site on the tetanus toxin H(C) fragment.     

Isolation and sequence of sheep Ig H and L chain cDNA

Sheep lymphocyte poly(A+) RNA was used as a template for the enzymatic synthesis of cDNA before cloning into the expression vector lambda gt11. Screening of the cDNA library with mAb probes resulted in the isolation of two recombinant phages containing Ig coding sequences of 704 bp and 925 bp. These were inserted into the EcoRI site of pUC18 and named pSLC (sheep Ig L chain) and pSHC (sheep Ig H chain). The insert in pSLC revealed sequence homology by using GenBank to lambda L chain and pSHC revealed sequence homology to IgG sequences from various species. The L chain cDNA contained the full translation sequence and 5' and 3' nontranslating region while the H chain cDNA coded for the secreted form of IgG1 and lacked sequences upstream from the C region. The derived amino acid sequences showed significant homology with various Ig sequences already described for human, mouse, rabbit, pig, and chicken but the degree of homology showed no consistency with established phylogenetic relationships.     

B cell selection and affinity maturation during an antibody response in the mouse with limited B cell diversity

The quasi-monoclonal mouse has limited B cell diversity, whose major (approximately 80%) B cell Ag receptors are comprised of the knockin V(H) 17.2.25 (V(H)T)-encoded H chain and the lambda1 or lambda2 L chain, thereby being specific for 4-hydroxy-3-nitrophenylacetyl. The p-nitrophenylacetyl (pNP) was found to be a low affinity analog of nitrophenylacetyl. We examined affinity maturation of anti-pNP IgG by analyzing mAbs obtained from quasi-monoclonal mice that were immunized with this low affinity Ag. The results are: 1) Although V(H)T/lambda1 and V(H)T/lambda2 IgM were equally produced, V(H)T/lambda2 IgG almost exclusively underwent affinity maturation toward pNP. 2) A common mutation in complementarity-determining region 3 of V(H)T (T313A) mainly contributed to generating the specificity for pNP. 3) Because mutated V(H)T-encoded gamma-chains could form lambda1-bearing IgG in Chinese hamster ovary cells, apparent absence of V(H)T/lambda1 anti-pNP IgG may not be due to the incompatibility between the gamma-chains and the lambda1-chain, but may be explained by the fact that V(H)T/lambda1 B cells showed 50- to 100-fold lower affinity for pNP than V(H)T/lambda2 B cells. 4) Interestingly, a pNP-specific IgM mAb that shared common mutations including T313A with high affinity anti-pNP IgG was isolated, suggesting that a part of hypermutation coupled with positive selection can occur before isotype switching. Thus, even weak B cell receptor engagement can elicit an IgM response, whereas only B cells that received signals stronger than a threshold may be committed to an affinity maturation process.     

The role of interface framework residues in determining antibody V(H)/V(L) interaction strength and antigen-binding affinity

While many antibodies with strong antigen-binding affinity have stable variable regions with a strong antibody heavy chain variable region fragment (V(H))/antibody light chain variable region fragment (V(L)) interaction, the anti-lysozyme IgG HyHEL-10 has a fairly strong affinity, yet a very weak V(H)/V(L) interaction strength, in the absence of antigen. To investigate the possible relationship between antigen-binding affinity and V(H)/V(L) interaction strength, a novel phage display system that can switch two display modes was employed. We focused on the two framework region 2 regions of the HyHEL-10 V(H) and V(L), facing each other at the domain interface, and a combinatorial library was made in which each framework region 2 residue was mixed with that of D1.3, which has a far stronger V(H)/V(L) interaction. The phagemid library, encoding V(H) gene 7 and V(L) amber codon gene 9, was used to transform TG-1 (sup+), and the phages displaying functional variable regions were selected. The selected phages were then used to infect a nonsuppressing strain, and the culture supernatant containing V(H)-displaying phages and soluble V(L) fragment was used to evaluate the V(H)/V(L) interaction strength. The results clearly showed the existence of a key framework region 2 residue (H39) that strongly affects V(H)/V(L) interaction strength, and a marked positive correlation between the antigen-binding affinity and the V(H)/V(L) interaction, especially in the presence of a set of particular V(L) residues. The effect of the H39 mutation on the wild-type variable region was also confirmed by a SPR biosensor as a several-fold increase in antigen-binding affinity owing to an increased association rate, while a slight decrease was observed for the single-chain variable region.     

Spectroscopic evidence for charge-transfer complexation in monoclonal antibodies that bind opiates

Molecular complexes of four monoclonal anti-morphine antibodies (mAb) with the opiate ligands morphine, oxymorphone, and naloxone were studied using UV-VIS absorption spectroscopy. Although strong overlaps in the absorption spectra of the antibodies, ligands, and complexes were observed, a curve-fitting method was developed to correlate the absorbance with the concentration of the ligand-antibody complex. Using this technique, we determined the intrinsic association constants for the mAb with morphine to be in the nanomolar range, while association constants for oxymorphone and naloxone were in the micromolar range. These values were found to be in agreement with previous radioimmunoassay determinations. We also observed different changes in the absorbancy of the mAb upon complexation with different ligands and such changes were found to be different for all four mAb examined. Upon complexation with the ligand morphine, two of the mAb (clone numbers MOR368-21 and MOR10.5) displayed distinct charge-transfer spectral bands in the 320-nm region. These observations suggest that mAb binding site tryptophans may participate in the formation of the antibody-ligand complex and such complexation involves a charge-transfer interaction.Abbreviations CAMM computer-aided molecular modeling - CDR complementarity-determining region - corr correlation coefficient - K _A assocation constant - mAb monoclonal antibody - PBS phosphate buffered saline,pH 7.0.     

The natural autoantibody repertoire of nonobese diabetic mice is highly active

Analysis of spontaneous hybridomas generated from nonobese diabetic (NOD) mice indicates that the natural autoantibody repertoire of NOD mice is highly active compared with C57BL/6 and BALB/c mice. This property of increased B cell activity is present early in life (4 wk) and persists in older mice of both sexes. Even when selected for binding to a prototypic beta cell Ag, such as insulin, NOD mAb have characteristics of natural autoantibodies that include low avidity and broad specificity for multiple Ags. Analyses of the variable region of Ig H chain (V(H)) and variable region kappa L chain genes expressed by six insulin binding mAb show that V gene segments are often germline encoded and are identical with those used by autoantibodies, especially anti-dsDNA, from systemic autoimmune disease in MRL, NZB/W, and motheaten mice. V(H) genes used by four mAb are derived from the large J558 family and two mAb use V(H)7183 and V(H)Q52 genes. The third complementarity-determining region of Ig H chain of these mAb have limited N segment diversity, and some mAb contain DNA segments indicative of gene replacement. Genetic abnormalities in the regulation of self-reactive B cells may be a feature that is shared between NOD and conventional systemic autoimmune disorders. In NOD, the large pool of self-reactive B cells may fuel autoimmune beta cell destruction by facilitating T-B cell interactions, as evidenced by the identification of one mAb that has undergone Ag-driven somatic hypermutation.     

Autoimmune mice and stimulated normal mice make lambda 1 with increased V region diversity

Previous studies of the genetic bases of murine SLE have defined gene segments that encode the H chain and the kappa L chain of anti-DNA, anti-Sm, and anti-IgG autoantibodies. As a result of these studies, the genetic origins of autoantibody H chains and kappa L chains are better understood, but little remains known about the genetic bases of autoantibody lambda-chains. Thus, we have analyzed serologically the germ-line and somatic origins of lambda 1 L chains in antibodies of normal mice and in both antibodies and autoantibodies of autoimmune mice. This study finds an increased lambda 1 diversity in both Ag-stimulated mice and autoimmune mice. This study also finds that the lambda 1 L chains in antibodies of unstimulated normal mice have the gene segment-encoded variable region, V lambda 1. In contrast, additional genetic processes appear to make the lambda 1 V regions of antibodies in Ag-stimulated normal mice and the lambda 1 V regions of both antibodies and autoantibodies in autoimmune mice. The increased lambda 1 diversity that we found in both Ag-stimulated mice and autoimmune mice might be caused by mutational processes creating antibody diversities. Therefore, the same somatic processes might be able to make both antibody and autoantibody lambda 1 diversities.     

Mutation of a single conserved residue in VH complementarity-determining region 2 results in a severe Ig secretion defect

During an immune response, somatic mutations are introduced into the VH and VL regions of Ig chains. The consequences of somatic mutation in highly conserved residues are poorly understood. Ile51 is present in 91% of murine VH complementarity-determining region 2 sequences, and we demonstrate that single Ile51-->Arg or Lys substitutions in the PCG1-1 Ab are sufficient to severely reduce Ig secretion (1-3% of wild-type (WT) levels). Mutant H chains, expressed in the presence of excess L chain, associate with Ig binding protein (BiP) and GRP94 and fail to form HL and H2L assembly intermediates efficiently. The mutations do not irreversibly alter the VH domain as the small amount of mutant H chain, which assembles with L chain as H2L2, is secreted. The secreted mutant Ab binds phosphocholine-protein with avidity identical with that of WT Ab, suggesting that the combining site adopts a WT conformation. A computer-generated model of the PCG1-1 variable region fragment of Ig (Fv) indicates that Ile51 is buried between complementarity-determining region 2 and framework 3 and does not directly contact the L chain. Thus, the Ile51-->Arg or Ile51-->Lys mutations impair association with the PCG1-1 L chain via indirect interactions. These interactions are in part dependent on the nature of the L chain as the PCG1-1 VH single Ile51-->Arg or Ile51-->Lys mutants were partially rescued when expressed with the J558L lambda1 L chain. These results represent the first demonstration that single somatic mutations in V(H) residues can impair Ig secretion and suggest one reason for the conservation of Ile51 in so many Ig VH.     

Heat-induced formation of a specific binding site for self-assembled Congo Red in the V domain of immunoglobulin L chain lambda

Moderate heating (40-50 degrees C) of immunoglobulins makes them accessible for binding with Congo Red and some related highly associated dyes. The binding is specific and involves supramolecular dye ligands presenting ribbon-like micellar bodies. The L chain lambda dimer, which upon heating disclosed the same binding requirement with respect to supramolecular dye ligands, was used in this work to identify the site of their attachment. Two clearly defined dye-protein (L lambda chain) complexes arise upon heating, here called complex I and complex II. The first is formed at low temperatures (up to 40-45 degrees C) and hence by a still native protein, while the formation of the second one is associated with domain melting above 55 degrees C. They contain 4 and 8 dye molecules bound per L chain monomer, respectively. Complex I also forms efficiently at high dye concentration even at ambient temperature. Complex I and its formation was the object of the present studies. Three structural events that could make the protein accessible to penetration by the large dye ligand were considered to occur in L chains upon heating: local polypeptide chain destabilization, VL-VL domain incoherence, and protein melting. Of these three possibilities, local low-energy structural alteration was found to correlate best with the formation of complex I. It was identified as decreased packing stability of the N-terminal polypeptide chain fragment, which as a result made the V domain accessible for dye penetration. The 19-amino acid N-terminal fragment becomes susceptible to proteolytic cleavage after being replaced by the dye at its packing locus. Its splitting from the dye-protein complex was proved by amino acid sequence analysis. The emptied packing locus, which becomes the site that holds the dye, is bordered by strands of amino acids numbered 74-80 and 105-110, as shown by model analysis. The character of the temperature-induced local polypeptide chain destabilization and its possible role in intramolecular antibody signaling is discussed.