Three-dimensional structure of Fab R19.9, a monoclonal murine antibody specific for the p-azobenzenearsonate group

Alanine Mutagenesis in the Complementarity Determining Region 3 of the MTB and HIV-1 Peptide-Bispecific T Cell Receptor Beta Chain Affects Ligand Recognition

Mycobacterium tuberculosis/human immunodeficiency virus (MTB/HIV) coinfection presents a special challenge to the prevention and treatment of tuberculosis and HIV/AIDS. Adoptive transfer of high-affinity T cell receptor (TCR) gene-modified T cells against MTB and HIV antigens is a promising approach to treating MTB/HIV coinfected patients whose cellular immunity is obviously disordered. We have previously successfully identified that a bispecific TCR screened out from peripheral blood mononuclear cells of a HLA-A*0201⁺ healthy individual using the complementarity determining region 3 (CDR3) spectratype analysis recognizes both MTB Ag85B_199–207 and HIV-1 Env_120–128 peptide. However, it has not been known how residues on CDR3 loops, which have been shown to play a leading role in antigen binding and specificity contribute to the bispecific TCR contact with the peptide–major histocompatibility complex (MHC) complexes. In this study, we provided an extensive investigation of residues in the predicted CDR3 of the bispecific TCR beta (β) chain using alanine scanning mutagenesis. Our data showed that three of the five substituted residues (G115A, T116A, A117G) in CDR3β of the bispecific TCR caused a significantly diminished T cell response to antigen, whereas the remaining two substituted residues (D114A, S118A) resulted in completely eliminated response, thus identifying the two residues that were particularly critical for the recognition of peptide–MHC in the bispecific TCR. These findings will provide an imperative foundation for generating an improved high-affinity bispecific TCR for use in T cell adoptive immunotherapy for MTB/HIV coinfected individuals.

Quantitative Analysis of the Association Angle between T-cell Receptor Vα/Vβ Domains Reveals Important Features for Epitope Recognition

T-cell receptors (TCR) play an important role in the adaptive immune system as they recognize pathogen- or cancer-based epitopes and thus initiate the cell-mediated immune response. Therefore there exists a growing interest in the optimization of TCRs for medical purposes like adoptive T-cell therapy. However, the molecular mechanisms behind T-cell signaling are still predominantly unknown. For small sets of TCRs it was observed that the angle between their Vα- and Vβ-domains, which bind the epitope, can vary and might be important for epitope recognition. Here we present a comprehensive, quantitative study of the variation in the Vα/Vβ interdomain-angle and its influence on epitope recognition, performing a systematic bioinformatics analysis based on a representative set of experimental TCR structures. For this purpose we developed a new, cuboid-based superpositioning method, which allows a unique, quantitative analysis of the Vα/Vβ-angles. Angle-based clustering led to six significantly different clusters. Analysis of these clusters revealed the unexpected result that the angle is predominantly influenced by the TCR-clonotype, whereas the bound epitope has only a minor influence. Furthermore we could identify a previously unknown center of rotation (CoR), which is shared by all TCRs. All TCR geometries can be obtained by rotation around this center, rendering it a new, common TCR feature with the potential of improving the accuracy of TCR structure prediction considerably. The importance of Vα/Vβ rotation for signaling was confirmed as we observed larger variances in the Vα/Vβ-angles in unbound TCRs compared to epitope-bound TCRs. Our results strongly support a two-step mechanism for TCR-epitope: First, preformation of a flexible TCR geometry in the unbound state and second, locking of the Vα/Vβ-angle in a TCR-type specific geometry upon epitope-MHC association, the latter being driven by rotation around the unique center of rotation.

The recognition of antigenic peptides by cytotoxic T-cells is one of the crucial steps during the adaptive immune response. Thus a detailed understanding of this process is not only important for elucidating the mechanism behind T-cell signaling, but also for various emerging new medical applications like T-cell based immunotherapies and designed bio-therapeutics. However, despite the fast growing interest in this field, the mechanistic basis of the immune response is still largely unknown. Previous qualitative studies suggested that the T-cell receptor (TCR) Vα/Vβ-interdomain angle plays a crucial role in epitope recognition as it predetermines the relative position of its antigen-recognizing CDR1-3 loops and thus TCR specificity. In the manuscript we present a systematic bioinformatic analysis of the structural characteristics of bound and unbound TCR molecules focusing on the Vα/Vβ-angle. Our results demonstrate the importance of this angle for signaling, as several distinct Vα/Vβ-angle based structural clusters could be observed and larger angle flexibilities exist for unbound TCRs than for bound TCRs, providing quantitative proof for a two-step locking mechanism upon epitope recognition. In this context, we could identify a unique rotational point, which allows a quantitative, yet intuitive description of all observed angle variations and the structural changes upon epitope binding.

Third complementarity-determining region of mutated VH immunoglobulin genes contains shorter V, D, J, P, and N components than non-mutated genes

The third complementarity-determining region (CDR3) of immunoglobulin variable genes for the heavy chain (VH) has been shown to be shorter in length in hypermutated antibodies than in non-hypermutated antibodies. To determine which components of CDR3 contribute to the shorter length, and if there is an effect of age on the length, we analysed 235 cDNA clones from human peripheral blood of VH6 genes rearranged to immunoglobulin M (IgM) constant genes. There was similar use of diversity (D) and joining (JH) gene segments between clones from young and old donors, and there was similar use of D segments among the mutated and non-mutated heavy chains. However, in the mutated heavy chains, there was increased use of shorter JH4 segments and decreased use of longer JH6 segments compared to the non-mutated proteins. The overall length of CDR3 did not change with age within the mutated and non-mutated categories, but was significantly shorter by three amino acids in the mutated clones compared to the non-mutated clones. Analyses of the individual components that comprise CDR3 indicated that they were all shorter in the mutated clones. Thus, there were more nucleotides deleted from the ends of VH, D, and JH gene segments, and fewer P and N nucleotides added. The results suggest that B cells bearing immunoglobulin receptors with shorter CDR3s have been selected for binding to antigen. A smaller CDR3 may allow room in the antibody binding pocket for antigen to interact with CDRs 1 and 2 as well, so that as the VDJ gene undergoes hypermutation, substitutions in all three CDRs can further contribute to the binding energy.

Immunoglobulin superfamily proteins: structure, mechanisms, and drug discovery

We review the recent progress made in our laboratories in structure-based drug design targeting proteins of the immunoglobulin superfamily (IgSF). We will focus on the CD4 protein, which is involved in T cell function, as a specific example of how the general concept and methodologies can be applied. Recent studies of CD4 structure and function have revealed new insight into possible mechanisms for CD4 self-association and its role in binding to major histocompatibility complex (MHC) class II molecules and initiation of T cell activation. This has led to the formulation of a hypothetical model of co-oligomerization of CD4, MHC class II, and T cell receptor (TCR). Such a basic understanding of CD4 structure and mechanisms has aided the development of a new generation of potential immunotherapeutics targeting specific CD4 surface functional sites. The design and discovery of small molecular inhibitors of CD4 and other IgSF proteins, in peptide, peptidomimetic, and nonpeptidic organic forms have opened new avenues for chemical research in which peptide, organic, and more recently combinatorial chemistry techniques can be used to further develop these promising lead analogs into a new generation of effective pharmaceuticals.

Antibody modeling: implications for engineering and design

Our understanding of the rules relating sequence to structure in antibodies has led to the development of accurate knowledge-based procedures for antibody modeling. Information gained from the analysis of antibody structures has been successfully exploited to engineer antibody-like molecules endowed with prescribed properties, such as increased stability or different specificity, many of which have a broad spectrum of applications both in therapy and in research. Here we describe a knowledge-based procedure for the prediction of the antibody-variable domains, based on the canonical structures method for the antigen-binding site, and discuss its expected accuracy and limitations. The rational design of antibody-based molecules is illustrated using as an example one of the most widely employed modifications of antibody structures: the humanization of animal-derived antibodies to reduce their immunogenicity for serotherapy in humans.

Insight into odorant perception: the crystal structure and binding characteristics of antibody fragments directed against the musk odorant traseolide

Monoclonal antibodies were elicited against the small hydrophobic hapten traseolide, a commercially available musk fragrance. Antibody variable region sequences were found to belong to different sequence groups, and the binding characteristics of the corresponding antibody fragments were investigated. The antibodies M02/01/01 and M02/05/01 are highly homologous and differ in the binding pocket only at position H93. M02/05/01 (H93 Val) binds the hapten traseolide about 75-fold better than M02/01/01 (H93 Ala). A traseolide analog, missing only one methyl group, does not have the characteristic musk odorant fragrance. The antibody M02/05/01 binds this hapten analog about tenfold less tightly than the original traseolide hapten, and mimics the odorant receptor in this respect, while the antibody M02/01/01 does not distinguish between the analog and traseolide. To elucidate the structural basis for the fine specificity of binding, we determined the crystal structure of the Fab fragment of M02/05/01 complexed with the hapten at 2.6 A resolution. The crystal structure showed that only van der Waals interactions are involved in binding. The somatic Ala H93 Val mutation in M02/05/01 fills up an empty cavity in the binding pocket. This leads to an increase in binding energy and to the ability to discriminate between the hapten traseolide and its derivatives. The structural understanding of odorant specificity in an antibody gives insight in the physical principles on how specificity for such hydrophobic molecules may be achieved.

The complement regulator C4b-binding protein analyzed by molecular modeling, bioinformatics and computer-aided experimental design

Molecular modeling and bioinformatics have gained recognition as scientific disciplines of importance in the field of biomedical research. Molecular modeling not only allows to predict the three-dimensional structure of a protein but also helps to define its function. Careful incorporation of the experimental findings in the structural/theoretical data provides means to understand molecular mechanisms for highly complex biological systems. C4b-binding protein (C4BP) is composed of one beta-chain and seven alpha-chains essentially built from three- and eight-complement control protein (CCP) modules, respectively, followed by a non-repeat carboxy-terminal region involved in polymerization of the chains. C4BP is involved in the regulation of the complement system and interacts with many molecules such as C4b, Arp, protein S and heparin. Here, we report experimental and computer data obtained for C4BP. Protein modeling together with site directed mutagenesis indicate that R39, R64 and R66 from the C4BP alpha-chain form a key binding site for heparin, suggesting that this region could be of major importance for interaction with C4b. We also propose that the first CCP of the C4BP beta-chain displays a key hydrophobic surface of major importance for the interaction with the coagulation cofactor protein S.

In vitro scanning saturation mutagenesis of all the specificity determining residues in an antibody binding site

For the first time, each specificity determining residue (SDR) in the binding site of an antibody has been replaced with every other possible single amino acid substitution, and the resulting mutants analyzed for binding affinity and specificity. The studies were conducted on a variant of the 26-10 antidigoxin single chain Fv (scFv) using in vitro scanning saturation mutagenesis, a new process that allows the high throughput production and characterization of antibody mutants [Burks,E.A., Chen,G., Georgiou,G. and Iverson,B.L. (1997) Proc. Natl Acad. Sci. USA, 94, 412-417]. Single amino acid mutants of 26-10 scFv were identified that modulated specificity in dramatic fashion. The overall plasticity of the antibody binding site with respect to amino acid replacement was also evaluated, revealing that 86% of all mutants retained measurable binding activity. Finally, by analyzing the physical properties of amino acid substitutions with respect to their effect on hapten binding, conclusions were drawn regarding the functional role played by the wild-type residue at each SDR position. The reported results highlight the value of in vitro scanning saturation mutagenesis for engineering antibody binding specificity, for evaluating the plasticity of proteins, and for comprehensive structure-function studies and analysis.

T-cell receptor structure and TCR complexes

The first crystal structures of intact T-cell receptors (TCRs) and their complexes with MHC peptide antigens (pMHC) were reported during the past year, along with those of a single-chain TCR Fv fragment and a beta-chain complexed with two different bacterial superantigens. These structures have shown the similarities and differences in the architecture of the antigen-binding regions of TCRs and antibodies, and how the TCR interacts with pMHC ligands as well as with superantigens.

Standard conformations for the canonical structures of immunoglobulins

A comparative analysis of the main-chain conformation of the L1, L2, L3, H1 and H2 hypervariable regions in 17 immunoglobulin structures that have been accurately determined at high resolution is described. This involves 79 hypervariable regions in all. We also analysed a part of the H3 region in 12 of the 15 VH domains considered here. On the basis of the residues at key sites the 79 hypervariable regions can be assigned to one of 18 different canonical structures. We show that 71 of these hypervariable regions have a conformation that is very close to what can be defined as a "standard" conformation of each canonical structure. These standard conformations are described in detail. The other eight hypervariable regions have small deviations from the standard conformations that, in six cases, involve only the rotation of a single peptide group. Most H3 hypervariable regions have the same conformation in the part that is close to the framework and the details of this conformation are also described here.

Site-specific mutagenesis of a recombinant anti-single-stranded DNA Fab. Role of heavy chain complementarity-determining region 3 residues in antigen interaction

The heavy chain complementarity-determining region 3 (HCDR3) of the anti-oligo(dT) recombinant antibody fragment, DNA-1, contributes significantly to antigen binding (Komissarov, A. A., Calcutt, M. J., Marchbank, M. T., Peletskaya, E. N., and Deutscher, S. L. (1996) J. Biol. Chem. 271, 12241-12246). In the present study, the role of separate HCDR3 residues of DNA-1 in interaction with oligo(dT) was elucidated. Based on a molecular model of the combining site, residues at the base (Arg98 and Asp108) and in the middle (Tyr101-Arg-Pro-Tyr-Tyr105) of HCDR3 were predicted to support the loop conformation and directly contact the ligand, respectively. Twenty-five site-specific mutants were produced as hexahistidine-tagged proteins, purified, and examined for binding to (dT)15 using two independent methods. All mutations in the middle of HCDR3 led to either abolished or diminished affinity. Tyr101 likely participates in hydrogen bonding, while Tyr104 and Tyr105 may be involved in aromatic-aromatic interactions with the ligand. The residues Arg102 and Pro103 were not as critical as the tyrosines. It is speculated that HCDR3 interacts with the thymines, rather than the phosphates, of the ligand. A 3-fold increase in affinity was observed by mutation of Asp108 to alanine. The highly conserved Arg98 and Asp108 do not appear to form a salt bridge.

Dual conformations of a T cell receptor V alpha homodimer: implications for variability in V alpha V beta domain association

The crystal structure of a mutant T cell receptor (TCR) V alpha domain containing a grafted third complementarity-determining region (CDR3) from a different V alpha was determined at 2.3 A resolution by molecular replacement using the wild-type V alpha structure as a search model. Like the wild-type V alpha domain, the mutant crystallized as a homodimer very similar to TCR V alpha V beta and antibody V(L)V(H) heterodimers, with the CDR loops disposed to form part of the antigen-binding site. However, the relative orientation of the two chains in the mutant V alpha homodimer differs from that in the wild-type by a rotation of 14 degrees such that the buried surface area in the dimer interface of the mutant is 140 A2 less than in the wild-type. While the residues forming the interface are essentially the same in the two structures, there are only four pairs of interface hydrogen bonds in the case of the mutant compared with eight for the wild-type. These results suggest that multiple relative orientations of the V alpha and V beta domains of TCRs may be possible, providing a significant contribution to TCR combining site diversity.

A computer screening approach to immunoglobulin superfamily structures and interactions: discovery of small non-peptidic CD4 inhibitors as novel immunotherapeutics

The interaction between CD4 and major histocompatibility complex (MHC) class II proteins is critical for the activation of CD4+ T cells, which are involved in transplantation reactions and a number of autoimmune diseases. In this study we have identified a CD4 surface pocket as a functional epitope implicated in CD4-MHC class II interaction and T-cell activation. A computer-based strategy has been used to screen approximately 150,000 non-peptidic organic compounds in a molecular data base and to identify a group of compounds as ligands of the proposed CD4 surface pocket. These small organic compounds have been shown to specifically block stable CD4-MHC class II binding, and exhibit significant inhibition of immune responses in animal models of autoimmune disease and allograft transplant rejection, suggesting their potential as novel immunosuppressants. This structure-based computer screening approach may have general implications for studying many immunoglobulin-like structures and interactions that share similar structural features. Furthermore, the results from this study have demonstrated that the rational design of small non-peptidic inhibitors of large protein-protein interfaces may indeed be an achievable goal.

Crystal structure of NMC-4 fab anti-von Willebrand factor A1 domain

We have solved the crystal structure of the Fab fragment of NMC-4, a mouse monoclonal antibody that binds to the A1 domain of von Willebrand factor (vWF). Two Asp and three Tyr residues in the complementarity determining regions 1 and 3 of the heavy chain exhibited a spatial orientation suggestive of a dominant role in establishing contact with the antigen. A cluster of Asp and Tyr residues occurs also in a region of the platelet glycoprotein (GP) Ib alpha amino terminal domain known to be critically involved in vWF binding. Thus, the structural information obtained with NMC-4 may prove relevant to understand the stereochemical bases of the GP Ib alpha-vWF interaction essential for thrombus formation at sites of vascular lesion.

Adaptive amino acid replacements accompanied by domain fusion in reverse transcriptase

Two basic processes are involved in protein evolution: One is amino acid replacement and another is reorganization of structural or functional units of proteins. Multidomain or multifunctional proteins are thought to have evolved by fusion of smaller structural units such as modules or domains. Reverse transcriptase (RT) is one of such fused proteins. The N-terminal part forms of globular domain with polymerase activity and the C-terminal part forms another globular domain with ribonuclease H activity (RNase H domain). There are single-domain enzymes which are homologous with the RNase H domain. The group of enzymes is called type I ribonuclease H (RNase HI). It is most likely that the ancestors of RNase HI and the polymerase domain were fused and became contemporary RT. At fusion, amino acid replacements presumably occurred at the interface of the domains to reinforce the interdomain interactions. Such replaced amino acid residues are conserved during evolution of the fused enzyme. We analyzed the pattern of amino acid replacement at each residue site in the free form, RNase HI group, and the integrated form, RNase H domain group. Then we compared the patterns between the two forms. Drastic fitting replacements of amino acid residues occurred at four of 29 residue sites involved in interdomain contact. Hydrophilic amino acid residues of the free form were substituted with hydrophobic or ambivalent ones in the integrated form. These substitutions aid in stabilizing the fused conformation by hydrophobic interactions at the interface of the domains. These observations imply that domain fusion could have occurred with only a relatively small number of adaptive amino acid substitutions.

The structural and functional basis of antibody catalysis

Ten years have passed since the initial reports that antibodies could be programmed to have enzymatic activity by immunization with a transition-site analog. Much of the research over the last decade has focused on defining the scope and generality of antibody catalysis; however, during the past two years the first few crystal structures of catalytic antibody transition-state analogs have been reported. This review analyzes four such structures of catalytic antibodies that catalyze markedly different reactions, including ester hydrolysis, sulfide oxidation, and a pericyclic rearrangement. Structure-function relations for these catalysts are discussed and compared to the structure and function of natural enzymes, as well as the chemistry that occurs in solution.

Structural model of the anti-snake-toxin antibody, M alpha 2,3

We present results of structural modeling of the variable fragment of M alpha 2,3, an antibody capable of neutralizing all short snake toxins. Three different methods were used to model the hypervariable loops: the conformational search algorithm CONGEN (Bruccoleri and Karplus, Biopolymers 26:137-168, 1987), high-temperature molecular dynamics (Bruccoleri and Karplus, Biopolymers 29:1847-1862, 1990), and a combined knowledge-based and energy-based algorithm (Martin et al., Proc. Natl. Acad. Sci. USA 86:9268-9272, 1989). Ninety plausible conformations were generated and were clustered into 13 classes. The clustering results indicate that there was little overlap of the conformational space explored by the different methods. Canonical loop structures were found by all methods for two of the loops, in agreement with previously established empirical modeling criteria. Nine of the 13 classes of structure were rejected on the ground of their lacking common features of antibody combining-site structure. The remaining four models were refined using restrained molecular dynamics. It was found that interconversion between the four resulting structures is possible with no significant energy barriers, suggesting that they are in thermodynamic equilibrium at 300 K. Features of the combining-site structure likely to be particularly important for antigen binding are discussed.

Equilibrium binding studies of recombinant anti-single-stranded DNA Fab. Role of heavy chain complementarity-determining regions

We previously isolated nucleic acid-binding antibody fragments (Fab) from bacteriophage display libraries representing the immunoglobulin repertoire of automimune mice to expedite the analysis of antibody-DNA recognition. In the present study, the binding properties of one such anti-DNA Fab, high affinity single-stranded (ss) DNA-binding Fab (DNA-1), were defined using equilibrium gel filtration and fluorescence titration. Results demonstrated that DNA-1 had a marked preference for oligo(dT) (100 nM dissociation constant) and required oligo(dT) >5 nucleotides in length. A detailed analysis of the involvement of the individual heavy chain (H) complementarity-determining regions (CDR) ensued using previously constructed HCDR transplantation mutants between DNA-1 and low affinity ssDNA-binding Fab (D5), a Fab that binds poorly to DNA (Calcutt, M. J. Komissarov, A. A., Marchbank, M. T., and Deutscher, S. L. (1996) Gene (Amst.) 168, 9-14). Circular dichroism studies indicated that the wild type and mutant Fab studied were of similar overall secondary structure and may contain similar combining site shapes. The conversion of D5 to a high affinity oligo(dT)-binding Fab occurred only in the presence of DNA-1 HCDR3. Results with site-specific mutants in HCDR1 further suggested a role of residue 33 in interaction with nucleic acid. The results of these studies are compared with previously published data on DNA-antibody recognition.

The immunological evolution of catalysis

The germline genes used by the mouse to generate the esterolytic antibody 48G7 were cloned and expressed in an effort to increase our understanding of the detailed molecular mechanisms by which the immune system evolves catalytic function. The nine replacement mutations that were fixed during affinity maturation increased affinity for the transition state analogue by a factor of 10(4), primarily the result of a decrease in the dissociation rate of the hapten-antibody complex. There was a corresponding increase in the rate of reaction of antibody with substrate, k(cat)/k(m), from 1.7 x 10(2)M(-1) min(-1) to 1.4 x 10(4)M(-1) min(-1). The three-dimensional crystal structure of the 48G7-transition state analogue complex at 2.0 angstroms resolution indicates that one of the nine residues in which somatic mutations have been fixed directly contact the hapten. Thus, in the case of 48G7, affinity maturation appears to play a conformational role, either in reorganizing the active site geometry of limiting side-chain and backbone flexibility of the germline antibody. The crystal structure and analysis of somatic and directed active site mutants underscore the role of transition state stabilization in the evolution of this catalytic antibody.

Crystal structure of the OPG2 Fab. An antireceptor antibody that mimics an RGD cell adhesion site

Cell surface receptors called integrins mediate diverse cell adhesion phenomena through recognition of the sequence arginine-glycine-aspartic acid (RGD) present in proteins such as fibronectin and fibrinogen. Platelet aggregation in hemostasis is mediated by the binding of fibrinogen to the gpIIb/IIIa integrin. The OPG2 antibody binds the gpIIb/IIIa receptor and acts as a ligand mimic due to the presence of an arginine-tyrosine-aspartic acid (RYD) sequence in the CDR3 loop of the heavy chain. The RYD loop and side chains are ordered in the 2.0-A resolution crystal structure of the Fab fragment from this antireceptor antibody. Moreover, the RYD loop assumes two clearly defined conformations that may correspond to the orientations of the loop in the free state or bound to integrin. This molecule will serve as a tool for understanding protein-integrin recognition in platelet aggregation and other RGD-mediated cell adhesion interactions.

Fv structure of monoclonal antibody II-481 against herpes simplex virus Fc gamma-binding glycoprotein gE contains immunodominant complementarity determining region epitopes that react with human immunoglobulin M rheumatoid factors

Human immunoglobulin M (IgM) rheumatoid factors (RFs) show primary direct enzyme-linked immunosorbent assay (ELISA) reactivity with Fab rather than Fc fragments of monoclonal antibody (mAb) II-481 directed against the Fc gamma-binding site of herpes simplex virus glycoprotein gE. This preferential anti-Fab specificity suggests that RFs react with antigen-binding portions of mAb II-481 as anti-idiotypic antibodies directed at the combining site regions of mAb reacting with the Fc gamma-binding region of gE. Analysis of this idiotype-anti-idiotype reaction employed polymerase chain reaction amplification and sequencing of the variable heavy and light (VH and VL) regions of mAb II-481. When VH and VL regions of mAb II-481 were synthesized as overlapping 7-mer peptides on polypropylene pins, a panel of 10 polyclonal and 6 monoclonal human IgM RFs reacted primarily with epitopes within the three solvent-exposed mAb II-481 complementarity determining regions (CDRs). Preincubation of single CDR heptamer peptides with IgM RFs in free solution, resulted in 63-100% inhibition of RF binding to mAb II-481 on the ELISA plate, confirming the antigenic importance of linear CDR regions for RF reactivity. Combinations of two or three CDR peptides frequently produced 94-100% inhibition of RF binding to whole mAb II-481. Control peptides, singly or in combination, showed no inhibition. Computer modeling suggested that the RF-reactive mAb II-481 Fv region and a previously demonstrated RF-reactive CH3 epitope displayed considerable three-dimensional similarities in conformation. These studies may provide insight into limited shape homologies possibly involved in an RF anti-idiotypic reaction.

Crystal structure of a catalytic antibody with a serine protease active site

The three-dimensional structure of an unusually active hydrolytic antibody with a phosphonate transition state analog (hapten) bound to the active site has been solved to 2.5 A resolution. The antibody (17E8) catalyzes the hydrolysis of norleucine and methionine phenyl esters and is selective for amino acid esters that have the natural alpha-carbon L configuration. A plot of the pH-dependence of the antibody-catalyzed reaction is bell-shaped with an activity maximum at pH 9.5; experiments on mechanism lend support to the formation of a covalent acyl-antibody intermediate. The structural and kinetic data are complementary and support a hydrolytic mechanism for the antibody that is remarkably similar to that of the serine proteases. The antibody active site contains a Ser-His dyad structure proximal to the phosphorous atom of the bound hapten that resembles two of the three components of the Ser-His-Asp catalytic triad of serine proteases. The antibody active site also contains a Lys residue to stabilize oxyanion formation, and a hydrophobic binding pocket for specific substrate recognition of norleucine and methionine side chains. The structure identifies active site residues that mediate catalysis and suggests specific mutations that may improve the catalytic efficiency of the antibody. This high resolution structure of a catalytic antibody-hapten complex shows that antibodies can converge on active site structures that have arisen through natural enzyme evolution.

Models of the serine protease domain of the human antithrombotic plasma factor activated protein C and its zymogen

Three-dimensional structural analysis of physiologically important serine proteases is useful in identifying functional features relevant to the expression of their activities and specificities. The human serine protease anticoagulant protein C is currently the object of many genetic site-directed mutagenesis studies. Analyzing relationships between its structure and function and between naturally occurring mutations and their corresponding clinical phenotypes would be greatly assisted by a 3-dimensional structure of the enzyme. To this end, molecular models of the protease domain of protein C have been produced using computational techniques based on known crystal structures of homologous enzymes and on protein C functional information. The resultant models corresponding to different stages along the processing pathway of protein C were analyzed for structural and electrostatic differences arising during the process of protein C maturation and activation. The most satisfactory models included a calcium ion bound to residues homologous to those that ligate calcium in the trypsin structure. Inspection of the surface features of the models allowed identification of residues putatively involved in specific functional interactions. In particular, analysis of the electrostatic potential surface of the model delineated a positively charged region likely to represent a novel substrate recognition exosite. To assist with future mutational studies, binding of an octapeptide representing a protein C cleavage site of its substrate factor Va to the enzyme's active site region was modeled and analyzed.

Crystal structure of an idiotype-anti-idiotype Fab complex

Anti-idiotypic monoclonal antibody 409.5.3 is raised against an antibody that neutralizes feline infectious peritonitis virus. This antibody, used as an immunogen, elicits the production of anti-anti-idiotypic antibodies that in turn neutralize the virus. The crystal structure of the complex between anti-idiotypic Fab 409.5.3 and idiotypic Fab fragment of virus-neutralizing antibody has been solved by molecular replacement using real-space Patterson search and filtering by Patterson correlation-coefficient refinement. The structure has been refined to an R value of 0.21 based on 21,310 unique reflections between 40.0 and 2.9 A. The three-dimensional structure reveals extensive, specific interactions that involve 118 van der Waals contacts and at least 9 probable hydrogen bonds. The two Fabs are rotated 61 degrees with respect to each other around the approximate long axis of the complex and are within 26 degrees being aligned along their major axes.

Bound water molecules and conformational stabilization help mediate an antigen-antibody association

We report the three-dimensional structures, at 1.8-A resolution, of the Fv fragment of the anti-hen egg white lysozyme antibody D1.3 in its free and antigen-bound forms. These structures reveal a role for solvent molecules in stabilizing the complex and provide a molecular basis for understanding the thermodynamic forces which drive the association reaction. Four water molecules are buried and others form a hydrogen-bonded network around the interface, bridging antigen and antibody. Comparison of the structures of free and bound Fv fragment of D1.3 reveals that several of the ordered water molecules in the free antibody combining site are retained and that additional water molecules link antigen and antibody upon complex formation. This solvation of the complex should weaken the hydrophobic effect, and the resulting large number of solvent-mediated hydrogen bonds, in conjunction with direct protein-protein interactions, should generate a significant enthalpic component. Furthermore, a stabilization of the relative mobilities of the antibody heavy- and light-chain variable domains and of that of the third complementarity-determining loop of the heavy chain seen in the complex should generate a negative entropic contribution opposing the enthalpic and the hydrophobic (solvent entropy) effects. This structural analysis is consistent with measurements of enthalpy and entropy changes by titration calorimetry, which show that enthalpy drives the antigen-antibody reaction. Thus, the main forces stabilizing the complex arise from antigen-antibody hydrogen bonding, van der Waals interactions, enthalpy of hydration, and conformational stabilization rather than solvent entropy (hydrophobic) effects.

X-ray structures of fragments from binding and nonbinding versions of a humanized anti-CD18 antibody: structural indications of the key role of VH residues 59 to 65

X-ray crystal structures of fragments from two different humanized anti-CD18 antibodies are reported. The Fv fragment of the nonbinding version has been refined in space group C2 with a = 64.2 A, b = 61.3 A, c = 51.8 A, and beta = 99 degrees to an R-value of 18.0% at 1.9 A, and the Fab fragment of the tight-binding version has been refined in space group P3 with a = 101. A and c = 45.5 A to an R-value of 17.8% at 3.0 A resolution. The very large difference in their binding affinity (> 1000-fold) is attributed to large and local structural differences in the C-terminal part of CDR-H2, and from this we conclude there is direct contact between this region and antigen when they combine. X-ray structures of antibody-antigen complexes available in the literature have yet to show this part of CDR-H2 in contact with antigen, despite its hypervariable sequence. Implications of this result for antibody humanization are discussed.

How the anti-(metal chelate) antibody CHA255 is specific for the metal ion of its antigen: X-ray structures for two Fab'/hapten complexes with different metals in the chelate

Antibodies with bound metal-chelate haptens provide new means for exploiting the diverse properties of metallic elements. The murine monoclonal antibody CHA255 (IgG1 lambda) binds the metal-chelate hapten indium (III)-4-[N'-(2-hydroxyethyl)thioureido]-L-benzyl-EDTA (designated In-EOTUBE) with high affinity (K(a) = 1.1 x 10(10) M-1). Antibody binding is highly specific for the indium chelate; the affinity decreases as much as 10(4) with other metals, even those having ionic radii close to indium. To better understand this selectivity, the crystal structure of the antigen-binding fragment (Fab') of CHA255 complexed with its hapten, In(III)-EOTUBE, was determined by molecular replacement and refined at 2.2-A resolution. The structure of CHA255 Fab' complexed with Fe(III)-EOTUBE was also determined and refined at 2.8-A resolution. In both structures, the hapten's EDTA moiety is half-buried near the center of the complementarity-determining regions (CDR's). Five of the six CDR's on the Fab' interact with the hapten through protein side-chain atoms (but not main-chain atoms). A novel feature of the In-EOTUBE/Fab' complex is coordination of the indium by N epsilon of one histidine from the heavy chain's third CDR (distance = 2.4 A). The histidine coordination is not observed in the Fe-EOTUBE/Fab' complex, due mainly to a slightly different hapten conformation that reduces metal accessibility; this may partially explain the 20-fold lower affinity of CHA255 for iron hapten. An unexpected feature of the Fab' overall is an elbow angle of 193 degrees (the angle between the pseudodyad axes of the Fab's constant and variable domains).

Principles and pitfalls in designing site-directed peptide ligands

An immunoglobulin light chain dimer with a large generic binding cavity was used as a host molecule for designing a series of peptide guest ligands. In a screening procedure peptides coupled to solid supports were systematically tested for binding activity by enzyme linked immunosorbent assays (ELISA). Key members of the binding series were synthesized in milligram quantities and diffused into crystals of the host molecule for X-ray analyses. These peptides were incrementally increased in size and affinity until they nearly filled the cavity. Progressive changes in binding patterns were mapped by comparisons of crystallographically refined structures of 14 peptide-protein complexes at 2.7 A resolution. These comparisons led to guidelines for ligand design and also suggested ways to modify previously established binding patterns. By manipulating equilibria involving histidine, for example, it was possible to abolish one important intramolecular interaction of the bound ligand and substitute another. These events triggered a change in conformation of the ligand from a compact to an extended form and a comprehensive change in the mode of binding to the protein. In dipeptides of histidine and proline, protonation of both imidazolium nitrogen atoms was used to program an end-to-end reversal of the direction in which the ligand was inserted into the binding cavity. Peptides cocrystallized with proteins produced complexes somewhat different in structure from those in which ligands were diffused into preexisting crystals. In such a large and malleable cavity, space utilization was thus different when a ligand was introduced before the imposition of crystal packing restraints.

Crystal structure to 2.45 A resolution of a monoclonal Fab specific for the Brucella A cell wall polysaccharide antigen

The atomic structure of an antibody antigen-binding fragment (Fab) at 2.45 A resolution shows that polysaccharide antigen conformation and Fab structure dictated by combinatorial diversity and domain association are responsible for the fine specificity of the Brucella-specific antibody, YsT9.1. It discriminates the Brucella abortus A antigen from the nearly identical Brucella melitensis M antigen by forming a groove-type binding site, lined with tyrosine residues, that accommodates the rodlike A antigen but excludes the kinked structure of the M antigen, as envisioned by a model of the antigen built into the combining site. The variable-heavy (VH) and variable-light (VL) domains are derived from genes closely related to two used in previously solved structures, M603 and R19.9, respectively. These genes combine in YsT9.1 to form an antibody of totally different specificity. Comparison of this X-ray structure with a previously built model of the YsT9.1 combining site based on these homologies highlights the importance of VL:VH association as a determinant of specificity and suggests that small changes at the VL:VH interface, unanticipated in modeling, may cause significant modulation of binding-site properties.

Length distribution of CDRH3 in antibodies

Sequences of the third complementarity determining region of antibody heavy chains (CDRH3s) are listed according to their length. Human sequences vary from 2 to 26 amino acids residues, but less extensively in other species. When combined with the other five complementarity determining regions, this enormous length variation of CDRH3, together with amino acid substitutions in their sequences, can provide a very large number of antibody specificities and can influence the shape of antibody combining sites.

Crystallographic analysis of the interaction between cyclosporin A and the Fab fragment of a monoclonal antibody

The structure of the complex between cyclosporin A and the Fab fragment of a monoclonal antibody has been established by crystallographic analysis to 2.65 A resolution. The structure has been solved by molecular replacement using a composite Fab model. The current R-factor after refinement is 0.179 between 8 and 2.65 A resolution. The antibody is one among three known structures with long H3 loops. This loop conformation is observed for the first time in the presence of the antigen. Residues from all six hypervariable loops interact with cyclosporin A. However, the 17 residues long loop H3 is the main contributor to the buried combining site area and to the van der Waals contacts made with cyclosporin A, with 52 and 63%, respectively, of the total contribution.

Antigen recognition properties of mutant V beta 3+ T cell receptors are consistent with an immunoglobulin-like structure for the receptor

We examined the effect of mutations in the V beta portion of a pigeon cytochrome c (cyto c)-specific V beta 3+/V alpha 11+ T cell receptor on its ability to recognize cyto c/IEk and various superantigens. The results were consistent with an immunoglobulin-like structure for the receptor V beta domain and with separate interaction sites on V beta for conventional antigen and superantigens. An amino acid predicted to lie in CDR1 was critical for cyto c/IEk but not superantigen recognition, while several amino acids predicted to lie in the hypervariable region 4 loop were critical for superantigen but not cyto c/IEk recognition.

Antigen recognition and targeted delivery by the single-chain Fv

The single-chain Fv (sFv) has proven attractive for immuno-targeting, both alone and as a targeting element within sFv fusion proteins. This chapter summarizes the features of sFv proteins that have sparked this interest, starting with the conservation of Fv architecture that makes general sFv design practical. The length and composition of linkers used to bridge V domains are discussed based on the sFv literature; special emphasis is given to the (Gly4Ser)3 15-residue linker that has proven of broad utility for constructing Fv regions of antibodies and other members of the immunoglobulin superfamily. The refolding properties of sFv proteins are summarized and examples given from our laboratory. Spontaneous refolding from the fully reduced and denatured state, typified by 26-10 sFv, is contrasted with disulfide-restricted refolding, exemplified by MOPC 315 and R11D10 sFv proteins, which recover antigen binding only if their disulfides have been oxidized prior to removal of denaturant. The medical value of sFv proteins hinges on their reliability in antigen recognition and rapidity in targeted delivery. Detailed analysis of specificity and affinity of antigen binding by the 26-10 antidigoxin sFv has demonstrated very high fidelity to the binding properties of the parent 26-10 sFv. These results gave confidence to the pursuit of more complex biomedical applications of these proteins, which is indicated by our work with the R11D10 sFv for the imaging of myocardial infarctions. Diagnostic imaging and therapeutic immunotargeting by sFv present significant opportunities, particularly as a result of their pharmacokinetic properties. Intravenously administered sFv offers much faster clearance than conventional Fab fragments or intact immunoglobulin with minimal background binding.

Three-dimensional structure of the Fab fragment of a neutralizing antibody to human rhinovirus serotype 2

The crystal structure of the antigen-binding fragment of a monoclonal antibody (8F5) that neutralizes human rhinovirus serotype 2 has been determined by X-ray diffraction studies. Antibody 8F5, obtained by immunization with native HRV2 virions, cross-reacts with peptides of the viral capsid protein VP2, which contribute to the neutralizing immunogenic site B in this serotype. The structure was solved by the molecular replacement method and has been refined to an R-factor of 18.9% at 2.8 A resolution. The elbow angle, relating the variable and constant modules of the molecule is 127 degrees, representing the smallest elbow angle observed so far in an Fab fragment. Furthermore, the charged residues of the epitope can be well accommodated in the antigen-binding site. This is the first crystal structure reported for an antibody directed against an icosahedral virus.

Functional analysis of the antigen binding site on the T cell receptor alpha chain

We have identified residues on a T cell receptor (TCR) alpha chain that are important for interaction with antigen/major histocompatibility complex (MHC). Using site-directed mutagenesis, we modified DNA encoding the postulated antigen/MHC binding loops on the TCR alpha chain expressed by the T cell clone D5, which recognizes p-azobenzenearsonate-conjugated antigens presented by cells bearing I-Ad. These variant TCR alpha chains were expressed in conjunction with the wild-type D5 TCR beta chain on the surface of hybridoma cells, and were tested for the ability to recognize hapten-conjugated antigens presented by I-Ad. Individual amino acid substitutions in each of the three antigen binding loops (alpha 1, alpha 2, alpha 3) of the D5 TCR alpha chain affected antigen recognition, demonstrating that all three loops are important in recognition of antigen/MHC. A subset of the single amino acid substitutions completely eliminated antigen recognition, thus identifying the residues that are particularly important in the recognition of antigenic peptide/MHC by the D5 TCR. Because the wild-type D5 TCR recognizes arsonate and certain structural analogues of arsonate conjugated to a variety of protein antigens, we were able to test whether the TCR substitutions affected the specificity of the D5 TCR for hapten or carrier antigen. One substitution introduced into antigen binding loop alpha 3 markedly altered the pattern of carrier recognition. Together, these results verify the Ig model for the TCR and are consistent with the proposition that residues forming the first and second antigen binding loops of the TCR contact the MHC, while those forming the third loop contact mainly antigenic peptides.

Distances between the antigen-binding sites of three murine antibody subclasses measured using neutron and X-ray scattering

For three different murine immunoglobulins (IgG subclasses 1, 2a, and 2b), the distances between their antigen-binding sites have been measured using neutron scattering from deuterated antigens complexed with proteated IgG. Neutron-scattering data were measured for each antibody-antigen complex in a 41% D2O solvent. Unlike the proteated antibody molecule, the deuterated antigens are strongly contrasted against the 41% D2O solvent and give rise to a scattering profile that contains an interference term related to the distance between the deuterated antigens. For all three subclasses, the damping of this interference term, which gives information on the relative flexibility of the antigen-binding sites, indicates that a single distance is inadequate to describe the observed scattering and a distribution of distances is needed. The scattering profile has been modeled for each subclass to give the mean distance between the antigens and the variance of this distance. For all three IgG subclasses, the mean distance is between 117 and 134 A, and the variance is large (approximately 40 A), indicating a high degree of flexibility of the Fab arms. Small-angle X-ray scattering measurements on the same samples are consistent with the neutron-scattering results.

A strong propensity toward loop formation characterizes the expressed reading frames of the D segments at the Ig H and T cell receptor loci

A compilation of murine and human Ig H and TcR beta D segment sequences was used to estimate the relative usage of the various reading frames and to look for associated sequence patterns. We confirm a strong bias in the expression of the Ig H D segments, with more than 90% (murine) and 85% (human) expressed peptides resulting from a preferred reading frame. Remarkably, 86% (mouse) and 90% (human) of those peptides contain at least one glycine residue. All but one of the atypical preferred D peptides contain serine or proline residues and are found in the immediate vicinity of glycine residues provided by specific JH segments. The presence of tyrosine residues is also a characteristic feature of expressed reading frames in both mouse (75%) and human (90%). These results suggest that the constraints of forming a flexible loop within the third complementarity-determining region, is a factor in the preference for a particular reading frame in Ig H D. For the TcR beta D segments, glycine is specified in most reading frames, and no significant preference is observed.

Molecular heterogeneity of auto-anti-idiotypic antibodies in MLR-lpr/lpr mice

The VH and V kappa gene families expressed by 20 monoclonal auto-anti-idiotypes (Ab2) derived from unmanipulated MLR-lpr/lpr mice were determined by Northern blotting. Complete variable region sequences of six Ab2, along with three additional V kappa-JH Ab2 sequences, were obtained. These auto-anti-idiotypes arose spontaneously in the animals, and they bound specifically to an idiotypic determinant (Id/r) on mAb 28/12, a monoclonal IgG2b MLR-lpr/lpr anti-small nuclear ribonucleoprotein antibody. The 16 Ab2 heavy chains belonged to 7 different VH gene families, and the 10 Ab2 light chains were derived from 8 V kappa families. The light chains of two Ab2 were approximately 99% identical; the remaining variable region sequences were highly heterogeneous. There was no correlation between primary amino acid sequence of either heavy or light chain and idiotypic properties of the auto-anti-idiotypes. Six Ab2 used VH or V kappa genes that are identical to known germ-line genes. A high proportion of the spontaneous auto-anti-idiotypes was shown to have autoantibody activity (anti-DNA, anti-ribonucleoprotein), or specific binding reactions with lipopolysaccharide of Salmonella RE, or both properties. The structural diversity of spontaneous MLR-lpr/lpr auto-anti-idiotypes differs sharply from the structural homogeneity reported for Ab2 induced in normal animals against syngeneic Ab1. Our results suggest that auto-anti-idiotypes might arise independently of an immunogenic stimulus from an Ab1.

Small rearrangements in structures of Fv and Fab fragments of antibody D1.3 on antigen binding

The potential use of monoclonal antibodies in immunological, chemical and clinical applications has stimulated the protein engineering and expression of Fv fragments, which are heterodimers consisting of the light and heavy chain variable domains (VL and VH) of antibodies. Although Fv fragments exhibit antigen binding specificity and association constants similar to their parent antibodies or Fab moieties, similarity in their interactions with antigen at the level of three-dimensional structure has not been investigated. We have determined the high-resolution crystal structure of the genetically engineered FvD1.3 fragment of the anti-hen egg-white lysozyme (HEL) monoclonal antibody D1.3, and of its complex with HEL. On comparison with the crystallographically refined FabD1.3-HEL complex, we find that FvD1.3 and FabD1.3 make, with minor exceptions, very similar contacts with the antigen. Furthermore, a small but systematic rearrangement of the domains of FvD1.3 occurs on binding HEL, bringing the contacting residues closer to the antigen by a mean value of about 0.7 A for VH (aligning on VL) or of 0.5 A for VL (aligning on VH). This is indicative of an induced fit rather than a 'lock and key' fit to the antigen.

Structural correlates of high antibody affinity: three engineered amino acid substitutions can increase the affinity of an anti-p-azophenylarsonate antibody 200-fold

The basis for the 200-fold difference in affinity between two hybridoma antibodies specific for the hapten p-azophenylarsonate (Ars) that have diversified by somatic hypermutation was examined. Oligonucleotide-directed mutagenesis was used to sequentially convert the nucleotide sequence of the lower-affinity antibody into that of the higher-affinity one, and the mutant antibodies generated by transfection of hybridoma cells were analyzed for affinity to Ars-tyrosine. The data showed that out of the 19 amino acid differences between the two hybridoma antibodies, the affinity increase could be reproduced by three heavy-chain substitutions that are present in the high-affinity antibody. The combined effect on affinity of amino acid substitutions was generally found to reflect their individual effects. Although the light chain of the high-affinity antibody did not seem to play a major role in the affinity increase, its contribution varied with the kind and number of heavy-chain substitutions. The results hold promise for antibody engineering and are consistent with a stepwise acquisition of somatic hypermutations in which the existing structural context of an antibody most likely influences the affinity-based selection of later substitutions. They further suggest that many substitutions may be tolerated in vivo during the antigen-driven selection process, even though they confer on the antibody no affinity increase.

Nucleotide sequences and three-dimensional modelling of the VH and VL domains of two human monoclonal antibodies specific for the D antigen of the human Rh-blood-group system

The nucleotide sequences were determined for the VH and VL domains of two human IgG1 antibodies, Pag-1 and Fog-B, specific for the D antigen of the Rh-blood-group system. The VH-region genes of the two antibodies were derived from separate germ-line genes within the VH-IV gene family, but both antibodies used the same JH6 gene. The D-region genes differed from each other, and no similarity was found to known D regions. The light chain of Fog-B belongs to the V lambda-I subgroup and that of Pag-1 probably belongs to the V lambda-V subgroup; both light chains used the J2/3 gene. Three-dimensional models of the variable domains were made, based on those of known crystallographic structure. The surface contours at the combining sites are clearly different, consistent with the evidence that the antibodies recognize different but overlapping epitopes. Some details of the molecular modelling of hypervariable regions have been deposited as Supplementary Publication SUP 50155 (6 pages) at the British Library Document Supply Centre, Boston Spa, Wetherby, West Yorkshire LS23 7BQ, U.K., from whom copies can be obtained on the terms indicated in Biochem. J. (1990) 265, 5.

Crystal structure of the antigen-binding fragment of the murine anti-arsonate monoclonal antibody 36-71 at 2.9-A resolution

The structure of the antigen-binding fragment (Fab) of an anti-phenylarsonate monoclonal antibody (36-71) bearing a major crossreacting idiotype of A/J mice has been solved and refined to an R factor of 19.3% at a resolution of 2.9 A. An initial electron density map was obtained with phase information from a total of six isomorphous heavy-atom derivatives (from two different compounds) and a molecular replacement solution using the HED10 Fab crystal structure as a model. The structure of the McPC603 Fab was used to provide an initial set of atomic coordinates. The electron density maps are clear and easily interpretable for the entire sequence except for sections from two of the heavy-chain complementarity-determining regions totaling 21 residues. These residues have been left out of the refinement and are not represented in our current model. The antigen-combining site was located by means of a difference Fourier synthesis with one of the heavy-atom derivatives, which contained arsanilic acid. It lies in a small pocket formed by residues from the hypervariable regions of both the heavy and the light chains. Interactions with the hapten from framework residues are also possible.

On the nature of antibody combining sites: unusual structural features that may confer on these sites an enhanced capacity for binding ligands

A detailed analysis of the structural aspects of antibody-antigen interactions has been made possible by the availability of X-ray structures for three complexes of antilysozyme Fabs to lysozyme (reviewed by Davies et al.: J. Biol. Chem. 263:10541-10544, 1988.) Examination of the antigen-contacting residues in the three antilysozyme Fabs reveals the occurrence of a large number of aromatics, particularly tyrosines, and the absence of apolar, aliphatic residues. Calculation of the frequency of occurrence of the various amino acid types reveals that tyrosines are three times, and histidines and asparagines eight times, more likely to be found in the complementarity-determining regions than in the framework of the variable domains. Analysis of the solvent accessibility of the residues in Fvs (the modules containing variable domains of the light and heavy chains) of known three-dimensional structure indicates that tyrosines and tryptophans are more exposed when they occur in the complementarity-determining regions than when in the framework. Furthermore, many more of the asparagines in the complementarity-determining regions than in the framework are buried. These asparagines appear to have a structural role in that they hydrogen-bond through their side chains to other side chains and, even more so, to the protein backbone. The stabilizing effect of the asparagines, plus the rigidity of the framework, may serve to allow the greater exposure of the aromatic residues to solvent. In view of the greater potential contribution of aromatic side chains to the total binding energy, these results suggest that antibody combining sites have structural features that make them especially suited for interacting with ligands.