The three-dimensional structure of a phosphorylcholine-binding mouse immunoglobulin Fab and the nature of the antigen binding site

A complementarity-determining region peptide of an anti-desmosome autoantibody may interact with the desmosomal plaque through molecular mimicry with a cytoplasmic desmoglein 1 sequence

The sera of patients with pemphigus, a group of autoimmune blistering skin diseases, contain autoantibodies directed against components of adhering junctions termed desmosomes. F12, a human monoclonal antibody derived from a pemphigus patient, recognizes an unknown polypeptide of the desmosomal and hemidesmosomal plaques. The third complementarity-determining region of the F12 heavy chain (VH-CDR3) was shown to share a four-amino-acid sequence (GSSG) with the intracellular domains of desmoglein 1 and bullous pemphigoid antigen 2 which interact with components of, respectively, the desmosomal and hemidesmosomal plaques. Computer modeling of F12 showed that the GSSG sequence protudes inside the antigen-combining site and thus might be involved in antigen interactions. The GSSG sequence is essential to F12 function, since a peptide containing the VH-CDR3 inhibited its binding to target antigens while VH-CDR3 peptides with specific modifications of the GSSG sequence did not. These data allow us to hypothesize that certain autoantibodies produced during the course of an autoimmune disease can behave as adhesion molecules, through the molecular mimicry of the motif involved in protein/protein adhesion, and to propose a new self-antigen binding mechanism for some autoantibodies.

The response of protein structures to amino-acid sequence changes

The general response of protein structures to mutations, insertions and deletions is conformational change. Comparisons of related proteins show that a large part of the polypeptide chain retains its basic folding pattern. This ‘core’ of the structures comprises major elements of secondary structure and residues flanking them, including active-site peptides. The core may amount to as little as 40% of the structure for distantly related proteins, but is 90 % or more for proteins with amino-acid sequence homologies of 50 % or more. There is a direct relation between the root mean square deviation of the main-chain atoms of the core residues of a pair of proteins and the overall amino-acid sequence homology. The deviation primarily reflects the shifts and rotations of packed secondary structures with respect to one another. For distantly related proteins, shifts of 3-5 Å (1 Å = 10 -1 nm = 10 -10 m) are typical and shifts of up to 7 Å have been observed. For proteins with sequence homologies of 50% or more the shifts are much smaller, lying in the range 0.3- 1.5 Å. Such closely related proteins are also characterized by a conservation of over 85% of the conformational angles of the backbone and of the side chains of unmutated residues. These observations suggest that successful model building of an unknown protein structure depends on knowing the structure of a reasonably close relative. As an application of these results we propose a model for the V L and V H domains of the antilysozyme antibody D1.3, the crystal structure determination of which is in progress.

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.

Symmetry of Fv architecture is conducive to grafting a second antibody binding site in the Fv region

Molecular modeling studies on antibody Fv regions have been pursued to design a second antigen-binding site (chi-site) in a chimeric single-chain Fv (chi sFv) species of about 30 kDa. This analysis has uncovered an architectural basis common to many Fv regions that permits grafting a chi-site onto the Fv surface that diametrically opposes the normal combining site. By using molecular graphics analysis, chimeric complementarity-determining regions (chi CDRs) were defined that comprised most of the CDRs from an antibody binding site of interest. The chain directionality of chi CDRs was consistent with that of specific bottom loops of the sFv, which allowed for grafting of chi CDRs with an overall geometry approximating CDRs in the parent combining site. Analysis of 10 different Fv crystal structures indicates that the positions for inserting chi CDRs are very highly conserved, as are the corresponding chi CDR boundaries in the parent binding site. The results of this investigation suggest that it should be possible to generally apply this approach to the development of chimeric bispecific antibody binding site (chi BABS) proteins.

Proline-rich tandem repeats of antibody complementarity-determining regions bind and neutralize human immunodeficiency virus type 1 particles

The proline-rich tandem repeat domain of human mucin MUC1 forms an extended structure containing large repeating loops that are crested by a turn. We show that the repeating-loop structure of MUC1 can be replaced by an antibody complementarity-determining region loop of a human immunodeficiency virus type 1 (HIV-1)-specific neutralizing antibody to create a chimeric, multivalent, mucin-like, anti-HIV-1 compound. We used 8 residues of an antibody molecule to replace 8 of 20 residues of the MUC1 tandem-repeat sequence. The antiviral peptide discussed here contains three copies of a 20-residue tandem repeat, (IYYDYEEDPAPGSTAPPAHG)3, for a total of 60 residues. We demonstrate that the mucin-antibody chimera retains the binding specificity of the parent antibody (monoclonal antibody F58), GPGR of the HIV-1 gp120 V3 neutralizing epitope, and the ability to neutralize virus particles. In inhibition enzyme-linked immunosorbent assay, the mucin-antibody chimeric peptide could inhibit 71 to 84% of binding to a V3 loop peptide by monoclonal antibodies known to be specific for GPGR in the V3 loop. The mucin-antibody chimeric peptide could also inhibit monoclonal antibody binding to native gp120 captured from virus particles. In addition, the chimeric peptide neutralized the homologous HIV-IIIB virus in a standard neutralization assay. The methods of antiviral peptide design and construction presented here are general and theoretically limited only by the size of the antibody repertoire. This approach could be used to synthesize peptides for a variety of therapeutic applications.

Comparative docking studies on ligand binding to the multispecific antibodies IgE-La2 and IgE-Lb4

A large comparative study is presented in which the binding of approximately 30 different ligands to two IgE antibodies (La2 and Lb4) is analyzed by means of an automated-docking procedure based on simulated annealing. The method is able to reproduce experimentally verified binding orientations, as shown by application to the Ig-AN02-hapten complex. The main address of the study is to investigate the concept of antibody multispecificity. Problems and usefulness of docking in this context are discussed. The results indicate reasons for multispecific binding properties and how they can be understood from the topology of the binding site. Though similar in general behaviour, the two antibodies show interesting differences in their binding characteristics. The binding sites of both antibodies are described and the main interacting residues revealed.

Insights into antibody catalysis: structure of an oxygenation catalyst at 1.9-angstrom resolution

The x-ray crystal structures of the sulfide oxidase antibody 28B4 and of antibody 28B4 complexed with hapten have been solved at 2.2-angstrom and 1.9-angstrom resolution, respectively. To our knowledge, these structures are the highest resolution catalytic antibody structures to date and provide insight into the molecular mechanism of this antibody-catalyzed monooxygenation reaction. Specifically, the data suggest that entropic restriction plays a fundamental role in catalysis through the precise alignment of the thioether substrate and oxidant. The antibody active site also stabilizes developing charge on both sulfur and periodate in the transition state via cation-pi and electrostatic interactions, respectively. In addition to demonstrating that the active site of antibody 28B4 does indeed reflect the mechanistic information programmed in the aminophosphonic acid hapten, these high-resolution structures provide a basis for enhancing turnover rates through mutagenesis and improved hapten design.

The crystal structures of complexes formed between lysozyme and antibody fragments

Type c lysozymes, and hen egg lysozyme in particular, have been extensively used to study the immune response because of their strong immunogenicity, the availability of many natural variants to study cross-reactivity, and the possibility to correlate these results with the known three-dimensional structure of lysozymes from several species. To date, the structure of six different murine monoclonal anti-lysozyme antibodies has been studied as a complex between the Fab fragment and antigen. In some cases, the structure of the uncomplexed Fab is also available, giving detail at the atomic level of the changes which take place during the formation of the antibody-antigen complex. The bacterially-expressed Fv molecule, the simplest fragment of an immunoglobulin retaining an intact antigen-binding site, has been studied for three of the monoclonal anti-lysozyme antibodies. Recombinant Fv fragments have opened up the possibility of using site-directed mutagenesis to study the effect of amino acid changes at the antibody-antigen interface. The six monoclonal antibodies appear to recognize epitopes which are localised on three different regions of the lysozyme surface.

Fetal versus adult PreB or B cells: the human VH repertoire

At the preB stage, when only the IGH locus has rearranged, mu chains become expressed in association with the psi L chains, lambda-like and VpreB, thus forming the preB receptor. By the use of a monoclonal anti VpreB antibody, preB cells were isolated from two adult bone marrow samples, and the VH repertoire was analyzed and compared to fetal, XLA (X-linked agammaglobulinemia), and adult B repertoires. Most VH genes identified were also expressed in fetal liver, XLA bone marrow, and adult PBLs, with similar predominant usage of certain germline genes. Multiple D/D fusions, limited N diversity, and preferential use of JH4 with a low level of DQ52 usage were also identified. Few mutations could be observed, not specifically localized in CDR regions, that could be interpreted as not positively selected. Conversely, a shorter length of CDR3 appeared to be the hallmark of the preB step. Thus, the association of psi L chains with mu does not bring about a bias in the VH gene usage, but a first selection on the CDR3 region could be the result of recognition by given autoantigens or ligands different for preB cells and B cells.

Antibody engineering

Antibody engineering has been an extremely intensive research area for many years. Recent achievements discussed in this review include: (i) significant improvements in the field of selection of antigen-specific antibody fragments on bacteriophages; (ii) new structural work, in particular using NMR; (iii) the cloning of essentially the complete set of human VH genes; (iv) the use of antibodies to catalyze complicated chemical reactions; and (v) novel antibody fusion proteins to potentiate immune therapy. An interesting new development is the replacement of antibodies with more stable protein scaffolds for many future biotechnological applications.

A human anti-insulin IgG autoantibody apparently arises through clonal selection from an insulin-specific "germ-line" natural antibody template. Analysis by V gene segment reassortment and site-directed mutagenesis

We analyzed the structural correlates underlying the insulin-dependent selection of the specific anti-insulin IgG1 kappa mAb13-producing cell clone, derived from a patient with insulin-dependent diabetes mellitus treated with recombinant human insulin. First, we cloned the germ-line genes that putatively gave rise to the expressed VH and V kappa segments and used them to generate the full (unmutated) "germ-line revertant" of the "wild-type" (somatically mutated) mAb13, using recombinant PCR methods and an in vitro human C gamma 1 and C kappa expression system. The full "germ-line revertant" bound insulin specifically and in a dose-saturable fashion, but with a relative avidity (AVrel) more than three-fold lower than that of its wild-type counterpart (Avrel, 1.69 x 10(-8) vs 4.91 x 10(-9) g/microliters). Second, we established, by reassorting wild-type and germ-line revertant forms of the mAb13 VH and V kappa segments, that the increased Avrel for insulin of mAb13 when compared with its full "germ-line revertant" counterpart was entirely dependent on the mutations in the VH not those in the V kappa chain. Third, we determined, by site-directed mutagenesis experiments, that of the three mutations in the mAb13 VH segment (Ser-->Gly, Ser-->Thr, and Ser-->Arg at positions 31, 56, and 58, respectively), only Arg58 was crucial in increasing the mAb13 Avrel (from 1.44 x 10(-8) to 5.14 x 10(-9) g/microliters) and affinity (Kd, from 189 to 59 nM) for insulin. The affinity enhancement mediated by the VH segment Arg58 residue reflected about a threefold decrease in dissociation rate constant (Koff, from 4.92 x 10(-3) to 1.54 x 10(-3) s-1) but not an increase in association rate constant (Kon, from 2.60 x 10(4) to 2.61 x 10(4) M-1 s-1), and it contrasted with the complete loss of insulin binding resulting from the substitution of the VH segment Asn52 by Lys. The present findings suggest that human insulin, a self Ag, has the potential to recruit a natural autoantibody-producing cell precursor expressing a specific surface receptor for Ag in unmutated configuration, and drive it through affinity maturation. They also show that binding of insulin by such a receptor can be enhanced or completely abrogated by a single amino acid change.

Evolution of specific antigen recognition: size reduction and restricted length distribution of the CDRH3 regions in the rainbow trout

The immunoglobulin heavy chain repertoire in fish was investigated by cloning a total of 88 rearranged VDJ junctions from the head kidney B cell mRNA of a salmonid, the rainbow trout (Oncorhynchus mykiss). Trout DH segments are short and cannot be classified into independent DH families. Several of the ten identified putative DH segments had stretches of nucleotide sequence identity with mouse (DQ52, DFL 16.2 and Dsp 2.1), human (DM1) and chicken (DH4) DH. There was a clear preference for one or two of the three putative DH reading frames and a stop codon is often present in the less used reading frame. Four of the six JH segments are preferentially used, and analysis of the VH-DH and DH-JH junctions suggest the presence of N-nucleotides. The absolute size and size heterogeneity of the rainbow trout CDRH3 are smaller than those of the Xenopus, mouse and human CDRH3. About 75% of the 84 in-frame trout CDRH3 have 8, 9 or 10 residues and none of them have more than 11 residues. This homogeneization of the CDRH3 loop size may partly explain the restricted antibody diversity in lower vertebrates.

Analysis of the structural correlates for antibody polyreactivity by multiple reassortments of chimeric human immunoglobulin heavy and light chain V segments

Polyreactive antibodies (Abs) constitute a major proportion of the early Ab repertoire and are an important component of the natural defense mechanisms against infections. They are primarily immunoglobulin M (IgM) and bind a variety of structurally dissimilar self and exogenous antigens (Ags) with moderate affinity. We analyzed the contribution of Ig polyvalency and of heavy (H) and light (L) chain variable (V) regions to polyreactivity in recombinatorial experiments involving the VH-diversity(D)-JH and V kappa-J kappa gene segments of a human polyreactive IgM, monoclonal antibody 55 (mAb55), and those of a human monoreactive anti-insulin IgG, mAb13, in an in vitro C gamma l and C kappa human expression system. These mAbs are virtually identical in their VH and V kappa gene segment sequences. First, we expressed the VH-D-JH and V kappa-J kappa genes of the IgM mAb55 as V segments of an IgG molecule. The bivalent recombinant IgG Ab bound multiple Ags with an efficiency only slightly lower than that of the original decavalent IgM mAb55, suggesting that class switch to IgG does not affect the Ig polyreactivity. Second, we coexpressed the mAb55-derived H or kappa chain with the mAb13-derived kappa or H chain, respectively. The hybrid IgG Ab bearing the mAb55-derived H chain V segment paired with the mAb13-derived kappa V segment, but not that bearing the mAb13-derived H chain V segment paired with the mAb55-derived kappa V segment, bound multiple Ags, suggesting that the Ig H chain plays a major role in the Ig polyreactivity. Third, we shuffled the framework 1 (FR1)-FR3 and complementarity determining region 3 (CDR3) regions of the H and kappa chain V segments of the mAB55-derived IgG molecule with the corresponding regions of the monoreactive IgG mAb13. The mAb55-derived IgG molecule lost polyreactivity when the H chain CDR3, but not the FR1-FR3 region, was replaced by the corresponding region of mAb13, suggesting that within the H chain, the CDR3 provides the major structural correlate for multiple Ag-binding. This was formally proved by the multiple Ag-binding of the originally monoreactive mAb13-derived IgG molecule grafted with the mAb55-derived H chain CDR3. The polyreactivity of this chimeric IgG was maximized by grafting of the mAb55-derived kappa chain FR1-FR3, but not that of the kappa chain CDR3.(ABSTRACT TRUNCATED AT 400 WORDS)

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.

Molecular dynamics simulation of the docking of substrates to proteins

A simple method is described to perform docking of substrates to proteins or probes to receptor molecules by a modification of molecular dynamics simulations. The method consists of a separation of the center-of-mass motion of the substrate from its internal and rotational motions, and a separate coupling to different thermal baths for both types of motion of the substrate and for the motion of the receptor. Thus the temperatures and the time constants of coupling to the baths can be arbitrarily varied for these three types of motion, allowing either a frozen or a flexible receptor and allowing control of search rate without disturbance of internal structure. In addition, an extra repulsive term between substrate and protein was applied to smooth the interaction. The method was applied to a model substrate docking onto a model surface, and to the docking of phosphocholine onto immunoglobulin McPC603, in both cases with a frozen receptor. Using translational temperatures of the substrate in the range of 1300-1700 K and room temperature for the internal degrees of freedom of the substrate, an efficient nontrapping exploratory search ("helicopter view") is obtained which visits the correct binding sites. Low energy conformations can then be further investigated by separate search or by dynamic simulated annealing. In both cases the correct minima were identified. The possibility to work with flexible receptors is discussed.

Antigenicity of the N8 influenza A virus neuraminidase: existence of an epitope at the subunit interface of the neuraminidase

To locate antigenic epitopes on the N8 neuraminidase (NA), we generated a panel of 97 monoclonal antibodies (MAbs), 66 of which inhibited NA activity (NI antibodies). Three groups of NI MAbs were identified from their different reactivities with escape mutants. Group 1 antibodies recognized the peptide loop containing residues 344 to 346, which appears to be an immunodominant region on the rim of the enzyme center of the N8 NA. Group 2 antibodies recognized a novel epitope containing residues 150, 199, 367, 399, and 400 (N2 numbering). From the location of these residues on the three-dimensional structure of the N8 NA, the epitope appears to be located at the interface of two adjacent monomers in the tetrameric NA, one contributing residues 150 and 199 and the other contributing residues 367 and 399 to 400. The available evidence indicates that the MAbs of this group react with the NA only after it is fully assembled. The third group of antibodies recognized the peptide loops containing residues 367 and 399 to 400. All of the amino acid substitutions in N8 escape mutants which affect the NI activity of antibodies were located in the peptide loops known to form epitopes in the N2 and N9 subtypes, indicating that antigenic regions in the NA head inducing NI antibodies appear to be similar among different subtypes of influenza A viruses. The MAbs used in this study will be valuable in studying the role of each N8 NA epitope in host immune defense systems and in the kinetics analysis of the biosynthesis of the enzyme.

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.

A designed metal-binding protein with a novel fold

A major challenge in protein design is to create stable scaffolds into which tailored functions can be introduced. Here we present the design, synthesis and characterization of a 61-residue all-beta protein: the minibody. We used a portion of the heavy chain variable domain of an immunoglobulin as a template, obtaining a molecule with a novel beta-sheet scaffold and two regions corresponding to the hypervariable loops H1 and H2. To exploit the potential for creating functional centres in the minibody, we engineered a metal-binding site into it. This site is formed by one histidine in H1 and two in H2. The protein is folded, compact and able to bind metal, thus representing the first designed beta-protein with a novel fold and a tailored function. By randomizing the sequence of the hypervariable loops, we are using the minibody scaffold to construct a conformationally constrained peptide library displayed on phage.

Fine specificity and VJ usage of light chains in antibodies to the phosphorylcholine hapten

In the memory response to the phosphorylcholine hapten (PC) two major groups of anti-PC antibodies with different fine specificities are elicited. Group I antibodies are mainly PC specific, whereas Group II antibodies are comprised of two specificities directed against the phenyl-PC and the phenyl moiety of the PC hapten. The VL gene usage of 17 monoclonal memory anti-PC antibodies were investigated by Southern blot analysis and nucleotide sequencing. Six out of eight Group I memory PC-specific antibodies used the same VK22-JK5 rearrangement as the major T15 primary response idiotype. One expressed a mutated JK1 and one employed another VK22 gene family member. A shift in specificity from PC (Group I) towards phenyl-PC (Group II) was accompanied with the usage of either VK1C-JK1 or VK1A-JK5 rearrangements. The phenyl-specific Group II antibodies expressed the V lambda 1-J lambda 1 L chain rearrangement in combination with VH M141 expressing H chains. In this specific segment of Group II antibodies most mutations were found. Thus four different VL genes were found to contribute to the fine specificity of memory response antibodies to the PC hapten in a clear structure-function relationship. The diversified fine specificity in the memory response derives mainly from the usage of different L chains with particular VJ rearrangements in combination with VH of the dominant initial response clonotype and is not primarily due to mutational events.

Modeling the anti-CEA antibody combining site by homology and conformational search

A model for an antibody specific for the carcinoembryonic antigen (CEA) has been constructed using a method which combines the concept of canonical structures with conformational search. A conformational search technique is introduced which couples random generation of backbone loop conformations to a simulated annealing method for assigning side chain conformations. This technique was used both to verify conformations selected from the set of known canonical structures and to explore conformations available to the H3 loop in CEA ab initio. Canonical structures are not available for H3 due to its variability in length, sequence, and observed conformation in known antibody structures. Analysis of the results of conformational search resulted in three equally probable conformations for H3 loop in CEA. Force field energies, solvation free energies, exposure of charged residues and burial of hydrophobic residues, and packing of hydrophobic residues at the base of the loop were used as selection criteria. The existence of three equally plausible structures may reflect the high degree of flexibility expected for an exposed loop of this length. The nature of the combining site and features which could be important to interaction with antigen are discussed.

Generation and analysis of random point mutations in an antibody CDR2 sequence: many mutated antibodies lose their ability to bind antigen

We have investigated the impact of mutations on the binding functions of the phosphocholine (PC)-specific T15 antibody in the absence of antigen selection pressure. The H chain complementarity determining region 2 (CDR2) sequence of T15 antibody was saturated with point mutations by in vitro random mutagenesis. From the mutant library, 289 clones were screened by direct DNA sequencing. The point mutations generated by this method were randomly distributed throughout the CDR2 region and included all kinds of substitutions. 46 unique mutant antibodies, containing one to four point mutations each, were expressed in SP2/0 myeloma cells. Functional analysis on these antibodies has provided insights into several aspects of somatic mutation. (a) The majority (26/46) of mutant antibodies either lost (20/46) or had reduced (6/46) ability to bind PC-protein conjugates or R36a, a PC-expressing strain of Streptococcus pneumoniae. In contrast, none of the mutant antibodies displayed increased binding for these PC antigens. Taken together with calculations of destructive mutations elsewhere in the V region, the data suggest that somatic mutation may cause extensive wastage among B cells during clonal expansion after antigen stimulation. (b) The frequency of binding-loss mutants increased sharply when a second mutation was introduced into the CDR2 sequence; it appears that, in some cases, two or more mutations are needed to destroy binding. (c) The mutant antibodies were tested for their reactivity to 11 non-PC antigens as well as to three PC analogues. None of the mutants gained new reactivity or changed their ability to discriminate structural analogues, supporting the notion that the major role of somatic mutation is to increase or decrease affinity rather than to create new specificities. (d) Mutations in at least five different positions in CDR2 were deleterious, suggesting that these residues may be essential for antigen binding. Three of these positions are novel in that they had not been identified to be important for binding PC by previous crystallographic analysis. (e) Introduction of mutations into two highly conserved residues in CDR2 did not alter the overall conformation of the V region as judged by antiidiotypic analysis, and, in some cases, did not affect the antigen binding function. The results thus indicate that even nonconservative substitutions of invariant residues need not be deleterious, suggesting that their conservation may be due to reasons other than maintaining antibody structure or specificity.

Common features of the conformations of antigen-binding loops in immunoglobulins and application to modeling loop conformations

Using database screening techniques we have examined the relationship between antigen-binding loops in immunoglobulins, and regions of similar conformation in other protein families. The conformations of most antigen-binding loops are not unique to immunoglobulins. But in many cases, the geometrical relationship between the loop and the peptides flanking it differs between the immunoglobulins and other structures with the same loop. We assess model building by data base screening, compared with that based on canonical structures.

The disulfide bonds in antibody variable domains: effects on stability, folding in vitro, and functional expression in Escherichia coli

The formation of the disulfide bonds in the variable domains VH and VL of the antibody McPC603 was found to be essential for the stability of all antigen binding fragments investigated. Exposure of the Fv fragment to reducing conditions in vitro resulted in irreversible denaturation of both VH and VL. In vitro refolding of the reduced Fv fragment was only possible when the disulfide bonds were allowed to form under oxidizing conditions. The analysis of a series of mutants of the Fv fragment, the Fab fragment and the single-chain Fv fragment, all secreted into the periplasm of Escherichia coli, in which each of the cysteine residues of the variable domains was replaced by a series of other amino acids, showed that functional antigen binding fragments required the presence of both the disulfide bond in VH and the one in VL. These results were also used to devise an alternative expression system based on the production of insoluble fusion proteins consisting of truncated beta-galactosidase and antibody domains, enzymatic cleavage, and refolding and assembly in vitro. This strategy should be useful for providing access to unstable antibody domains and fragments.

Anti-insulin antibody structure and conformation. I. Molecular modeling and mechanics of an insulin antibody

A knowledge-based three-dimensional model of an anti-insulin antibody, 125, was constructed using the structures of conserved residues found in other known crystallographic immunoglobulins. Molecular modeling and mechanics were done with the 125 amino acid sequences using QUANTA and CHARMm on a Silicon Graphics 4D70GT workstation. A minimal model was made by scaffolding using crystallography coordinates of the antibody HyHEL-5, because it had the highest amino acid sequence homology with 125 (84% light chain, 65% heavy chain). The three hypervariable loop turns that are longer in 125 than in HyHEL-5 (L1, L3, and H3) were modeled separately and incorporated into the HyHEL-5 structure; then other amino acid substitutions were made and torsions optimized. The 125 model maintains all the structural attributes of an antibody and the structures conserved in known antibodies. Although there are many polar amino acids (especially serines) in this site, the overall van der Waals surface shape is determined by positions of aromatic side chains. Based on this model, it is suggested that hydrogen bonding may be key in the interaction between the human insulin A chain loop antigenic epitope and 125.

Comparison of junctional diversity in the neonatal and adult immunoglobulin repertoires

Junctional diversity in immunoglobulin (Ig) from an adult mouse contributes significantly to the size of the final Ig repertoire. In adult pre-B cells, N region addition and deletion of nucleotides form coding regions produces very heterogenous CDR3 sequences. In contrast, Ig from fetal and newborn mice show very restricted junctional diversity. The reasons for this are: (a) the lack of N regions; and (b) the predominance of certain junctional sequences. These common junctional sequences all appear to occur by targeted rearrangement to short stretches of sequence homology near the ends of the segments to be joined. Targeted rearrangement may play a role in the overexpression of certain Vh genes early in ontogeny. These non-random junctional sequences in the neonate will reproducibly create certain Ig, for example, the dominant T15 anti-PC antibodies. Thus the immune system first creates a small repertoire of predictable Ig sequences. To the extent that these Ig are expressed in long-lived B cells, these early Ig sequences may persist in the adult. Superimposed upon this early repertoire is an enormously diverse adult Ig repertoire.

Protein engineering of antibodies

This article reviews the technical advances in antibody engineering and the clinical applications of these molecules. Recombinant DNA technology facilitates the construction and expression of engineered antibodies. These novel molecules are designed to meet specific applications. Although genomic and cDNA cloning have been used widely in the past to isolate the relevant antibody V domains, at present, the PCR-based cloning is the preferred system. Bacterial and mammalian expression systems are used commonly for the production of antibodies, antibody fragments, and antibody fusion proteins. A range of chimeric antibodies with murine V domains joined to C regions from human and other species have been produced and found to exhibit the expected binding characteristics and effector functions. Humanized antibodies have been developed to minimize the HAMA response, and bifunctional immunoglobulins are being used in tumor therapy and diagnosis. Single chain antibodies and fusion proteins with antibody specificities jointed to nonimmunoglobulin sequences provide a source of antibody-like molecules with novel properties. The potential applications of minimal recognition units and antigenized antibodies are described. Combinatorial libraries produced in bacteriophage present an alternative to hybridomas for the production of antibodies with the desired antigen binding specificities. Future developments in this field are discussed also.

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.

Information-theoretical entropy as a measure of sequence variability

We propose the use of the information-theoretical entrophy, S = -sigman pi log2 pi, as a measure of variability at a given position in a set of aligned sequences. pi stands for the fraction of times the i-th type appears at a position. For protein sequences, the sum has up to 20 terms, for nucleotide sequences, up to 4 terms, and for codon sequences, up to 61 terms. We compare S and Vs, a related measure, in detail with Vk, the traditional measure of immunoglobulin sequence variability, both in the abstract and as applied to the immunoglobulins. We conclude that S has desirable mathematical properties that Vk lacks and has intuitive and statistical meanings that accord well with the notion of variability. We find that Vk and the S-based measures are highly correlated for the immunoglobulins. We show by analysis of sequence data and by means of a mathematical model that this correlation is due to a strong tendency for the frequency of occurrence of amino acid types at a given position to be log-linear. It is not known whether the immunoglobulins are typical or atypical of protein families in this regard, nor is the origin of the observed rank-frequency distribution obvious, although we discuss several possible etiologies.

Molecular recognition in antibodies and its application

The structure and function of immunoglobulins, and the nature of the antibody-antigen interaction are described. Applications of the molecular recognition properties of antibodies are discussed in the areas of immunotherapy, immunoassay, immunotargeting and catalytic antibodies.

The functional expression of antibody Fv fragments in Escherichia coli: improved vectors and a generally applicable purification technique

We have previously demonstrated that the expression of fully functional Fv and Fab fragments in E. coli is possible by the simultaneous secretion of both chains to the periplasm. To increase production levels and facilitate engineering and random mutagenesis, we improved our previous vectors by introducing a resident repressor gene and a filamentous phage origin. We also developed a new purification strategy based on immobilized metal ion chromatography, with which a single-chain Fv fragment can be purified to homogeneity in a single step. We investigated the most efficient tail constructions and found that only a minimal structural change of three additional C-terminal amino acids is necessary. This modification has no deleterious effect on in vivo transport and folding or antigen affinity.

Design and synthesis of a mimetic from an antibody complementarity-determining region

A technique for producing non-peptide compounds (mimetics) of designed specificities was developed that permitted the synthesis of a conformationally restricted molecule that mimicked the binding and functional properties of monoclonal antibody (MAb) 87.92.6, which recognizes the reovirus type 3 cellular receptor. Binding of either MAb 87.92.6, peptide analogs, or 87.1-mimetic to the cellular receptor inhibited cellular proliferation. The mimetic was a synthetic beta-loop structure that mimics the second complementarity-determining region of the MAb. These studies may lead to strategies for the synthetic design of antibody complementarity regions, ligands, and other pharmacologically active agents that are water soluble, resistant to proteolysis, and nonimmunogenic.

Antibody engineering: advances from the use of Escherichia coli expression systems

Synthesis in Escherichia coli of correctly folded antibody fragments that bind antigen with the same affinity as the whole antibody is now possible. Here I review the techniques for achieving this and the physical properties of the various fragments produced. This technology not only facilitates antibody engineering but is also the basis of screening libraries for binding activity. Although the immunization of animals has not been made unnecessary in the production of monoclonal antibodies, steps toward this goal are now feasible.

Analysis of T cell reactivities to phosphorylcholine-conjugated hen egg lysozyme in C57BL/6 mice: hapten-conjugate specificity reflects an altered expression of a major carrier epitope

We have selected several different T cell hybridoma clones reactive to hen egg lysozyme (HEL) conjugated to phosphorylcholine (PC) after fusion of PC-HEL-primed C57BL/6 lymphocytes with BW5147 parent cells. These hybridoma clones preferentially recognize PC-HEL over unconjugated HEL, but not other carrier molecules conjugated with the same hapten. All the PC-HEL-reactive clones are similarly responsive to not only p-azobenzenearsonate (ABA)-conjugated HEL (ABA-HEL) but also to a variety of other diazotized hapten-HEL conjugates. However, these clones are not stimulated by fluoresceinated or dinitrophenylated HEL beyond the level of HEL carrier alone. Therefore, the type of hapten linkage (diazonium) to the carrier molecule appears to affect T cell recognition. The hybridoma clones apparently recognize the carrier molecule alone, although the level of stimulation is relatively low compared to that induced by either PC-HEL or ABA-HEL. Interestingly, HEL unfolded by S-carboxymethylation is capable of stimulating the hybridomas to a level comparable to that obtained with PC-HEL. T cell recognition of the unfolded HEL is independent of antigen processing, which is different from that of PC-HEL. The peptide sequence corresponding to the amino acids 81-93 of HEL appears to contain the epitope region for the hybridoma clones based on testing stimulation activity with synthetic peptides. Previously, the peptides including this region (81-96) have been reported as the determinant recognized by T cells derived from C57BL/6 mice after immunization with an HEL peptide (HEL 13-105) but not with native HEL. These results suggest that a hapten conjugation via diazonium linkage modifies antigen presentation and consequently the presentation of the major T cell epitopes similar to that of the HEL fragment.

Characterization of somatically mutated S107 VH11-encoded anti-DNA autoantibodies derived from autoimmune (NZB x NZW)F1 mice

We have studied 19 S107 heavy chain variable region gene (VH11)-encoded monoclonal antibodies from NZBWF1 mice. These studies show that a single VH gene can encode both antibodies to foreign antigens (anti-phosphorylcholine) and to self antigens (anti-double-stranded DNA) in the same animal. All of the anti-DNA antibodies contain many somatic mutations compared with the relevant germline genes. Since the anti-DNA antibodies were extensively somatically mutated and had undergone isotype switching, the response seems to be T cell dependent. While some of the antibodies appear to be the products of an antigen-driven and antigen-selected response, a number of characteristics of the antibodies suggest that forces other than antigen are contributing to the stimulation and selection of this response.

Conformational sampling using high-temperature molecular dynamics

High-temperature molecular dynamics as a method for conformational search was explored on the antigen combining site of McPC 603, a phosphorylcholine binding immunoglobulin. Simulations at temperatures of 500, 800, and 1500 K were run for 111.5, 101.7, and 76.3 ps, respectively. The effectiveness of the search was assessed using a variety of methods. For the shorter hypervariable loops, molecular dynamics explored an appreciable fraction of the conformational space as evidenced by a comparison to a simple theoretical model of the size of the conformational space. However, for the longer loops and the antigen combining site as a whole, the simulation times were too short for a complete search. The simulations at 500 and 800 K both generated conformations that minimized to energies 200 kcal/mole lower than the crystal structure. However, the 1500 K simulation produced higher energy structures, even after minimization; in addition, this highest temperature run had many cis-trans peptide isomerizations. This suggests that 1500 K is too high a temperature for unconstrained conformational sampling. Comparison of the results of high temperature molecular dynamics with a direct conformational search method, [R. E. Bruccoleri & M. Karplus (1987) Biopolymers 26, 137-168]. showed that the two methods did not overlap much in conformational space. Simple geometric measures of the conformational space indicated that the direct method covered more space than molecular dynamics at the lower temperature, but not at 1500 K. The results suggest that high-temperature molecular dynamics can aid in conformational searches.

Antibodies from Escherichia coli

Use of Escherichia coli as an expression host has opened up new possibilities in antibody research and its applications. It greatly facilitates rational engineering and random mutagenesis.

Structural features of the McPC603 Fab fragment not defined in the X-ray structure

The proteolytic Fab fragment of the well characterized antibody McPC603 was compared to the recombinant Fab fragment, which was obtained in functional form from an Escherichia coli expression system [(1989) Methods Enzymol. 178, 497-515]. We found evidence that the proteolytic fragment is glycosylated at Asn H160 in the CH1 domain, where additional electron density had been observed in the crystal structure [J. Mol. Biol. 190, 593-604]. In addition, its heavy chain is about 30 amino acids longer than visible in the electron density and thus contains the complete hinge region. These structural differences between the recombinant Fab fragment, which had been designed exactly according to the defined electron density, and the proteolytic Fab fragment of McPC603 had no effect on the hapten binding properties of these antigen binding fragments. Yet, it may be important to be aware of these structural features of McPC603 in folding studies and some comparative analyses of antibody structures.

A single amino acid substitution in the variable region of the light chain specifically blocks immunoglobulin secretion

Although immunoglobulin light chains are usually secreted in association with heavy chains, free light chains can be secreted by lymphocytes. To identify the structural features of light chains that are essential for their secretion, we mutated a conserved sequence in the variable domain of a lambda I light chain. The effects of the mutations on secretion were assayed by transient expression in COS-1 cells. One mutant (AV60), which replaced Ala-60 with Val, was secreted as efficiently as wild-type lambda I by transfected COS-1 cells. This result was not surprising because secreted lambda II chains contain valine in this position. However, a second lambda I mutant (AV60FS62), which replaced Phe-62 with Ser as well as Ala-60 with Val, was not secreted. This mutant was arrested in the endoplasmic reticulum, as judged by immunofluorescence and by its association with a lumenal endoplasmic reticulum protein, immunoglobulin heavy chain binding protein (BiP). The defect in secretion was not due to gross misfolding of the lambda I chain, since cells cotransfected with AV60FS62 and an immunoglobulin heavy chain gene produced functional antigen-binding antibodies. These assembled IgM molecules were still not secreted. Hence, the replacement of Phe-62 with Ser specifically affects a determinant on the lambda I light chain that is necessary for the intracellular transport of this molecule.