Rationalization and design of the complementarity determining region sequences in an antibody-antigen recognition interface

Protein-protein interactions are critical determinants in biological systems. Engineered proteins binding to specific areas on protein surfaces could lead to therapeutics or diagnostics for treating diseases in humans. But designing epitope-specific protein-protein interactions with computational atomistic interaction free energy remains a difficult challenge. Here we show that, with the antibody-VEGF (vascular endothelial growth factor) interaction as a model system, the experimentally observed amino acid preferences in the antibody-antigen interface can be rationalized with 3-dimensional distributions of interacting atoms derived from the database of protein structures. Machine learning models established on the rationalization can be generalized to design amino acid preferences in antibody-antigen interfaces, for which the experimental validations are tractable with current high throughput synthetic antibody display technologies. Leave-one-out cross validation on the benchmark system yielded the accuracy, precision, recall (sensitivity) and specificity of the overall binary predictions to be 0.69, 0.45, 0.63, and 0.71 respectively, and the overall Matthews correlation coefficient of the 20 amino acid types in the 24 interface CDR positions was 0.312. The structure-based computational antibody design methodology was further tested with other antibodies binding to VEGF. The results indicate that the methodology could provide alternatives to the current antibody technologies based on animal immune systems in engineering therapeutic and diagnostic antibodies against predetermined antigen epitopes.     

Length of the antibody heavy chain complementarity determining region 3 as a specificity-determining factor

The antigen binding site of an antibody is made up of residues residing in six hypervariable loops of the heavy and light chains. In most cases several or all of these loops are required for the establishment of the antigen-binding surface. Five of these loops display a limited diversity in length and sequence while the third complementarity determining region (CDR) of the heavy chain is highly different between antibodies not only with respect to sequence but also with respect to length. Its extensive diversity is a key component in the establishment of binding sites allowing for the recognition of essentially any antigen by humoral immunity. The relative importance of its sequence vs its length diversity in this context is however, not very well established. To investigate this matter further we have used an approach employing combinatorial antibody libraries and antigen-specific selection in the search for CDRH3 length and sequence diversity compatible with a given antigen specificity, the major antigenic determinant on the tumour-associated antigen mucin-1. In this way we have now defined heavy chain CDR3 length as a critical parameter in the creation of an antigen-specific binding site. We also propose that this may reflect a dependence of a particular structure of this hypervariable loop, the major carrier of diversity in the binding site, for establishment of a given specificity.     

Generation of humanized anti-DNA hydrolyzing catalytic antibodies by complementarity determining region grafting

Humanization of nonhuman antibodies (Abs) has been carried out mainly for Abs which bind to antigen without catalytic activity. Here we report humanization of mouse-originated 3D8 (m3D8) mAbs (scFv, VH, and VL) with DNA hydrolyzing catalytic activity by grafting the complementarity determining regions (CDRs) into the corresponding regions of a fixed human framework scaffold, generating humanized 3D8 (h3D8) Abs in the respective format of scFv, VH, and VL. h3D8 Abs retained comparable DNA binding and hydrolyzing activities to those of the corresponding m3D8 Abs. Our results suggest that CDRs of anti-DNA hydrolyzing Abs might possess the intrinsic properties of DNA binding and hydrolyzing activities.     

Thermodynamics of Fab-ssDNA interactions: contribution of heavy chain complementarity determining region 3.

The recombinant anti-ssDNA Fab, DNA-1, and 16 heavy chain complementarity determining region 3 (HCDR3) mutant variants were selected for thermodynamic characterization of ssDNA binding. The affinity of Fab to (dT)(15) under different temperatures and cation concentrations was measured by equilibrium fluorescence quenching titration. Changes in the standard Gibbs free binding energy (DeltaG degrees ), enthalpy (DeltaH degrees ), entropy (DeltaS degrees ), and the number of ionic pairs (Z) formed upon interaction were determined. All Fab possessed an enthalpic nature of interaction with ssDNA, that was opposite to the previously reported entropically driven binding to dsDNA [Tanha, J., and Lee, J. S. (1997) Nucleic Acids Res. 25, 1442-1449]. The contribution of separate residues of HCDR3 to ssDNA interaction was investigated. Analysis of the changes in DeltaH degrees and TDeltaS degrees, induced by substitutions in HCDR3, revealed a complete entropy/enthalpy compensation. Mutations R98A and D108A at the ends of the HCDR3 loop produced increases in TDeltaS degrees ( )()by 10.4 and 15.9 kcal/mol, respectively. Substitution of proline for arginine at the top of HCDR3 resulted in a new electrostatic contact with (dT)(15). The observed linear correlation of Z and DeltaG degrees ( )()of nonelectrostatic interactions (DeltaG degrees (nonel)) at the anti-ssDNA combining site was used for the estimation of the specific DeltaG degrees (nonel) [-20 to -25 cal/(mol.A(2))], the average contact area (450-550 A(2)), the maximal Z (6-7), and the limit in affinity under standard cation concentrations [(0.5-1) x 10(8) M(-)(1)] for this family of Fab. Results suggested that rational engineering of HCDR3 could be utilized to control the affinity and likely the specificity of Ab-DNA interactions.     

Programmed +1 frameshifting stimulated by complementarity between a downstream mRNA sequence and an error-correcting region of rRNA

Like most retroviruses and retrotransposons, the retrotransposon Ty3 expresses its pol gene analog (POL3) as a translational fusion to the upstream gag analog (GAG3). The Gag3-Pol3 fusion occurs by frameshifting during translation of the mRNA that encodes the two separate but overlapping ORFs. We showed previously that the shift occurs by out-of-frame binding of a normal aminoacyl-tRNA in the ribosomal A site caused by an aberrant codonoanticodon interaction in the P site. This event is unlike all previously described programmed translational frameshifts because it does not require tRNA slippage between cognate or near-cognate codons in the mRNA. A sequence of 15 nt distal to the frameshift site stimulates frameshifting 7.5-fold. Here we show that the Ty3 stimulator acts as an unstructured region to stimulate frameshifting. Its function depends on strict spacing from the site of frameshifting. Finally, the stimulator increases frameshifting dependent on sense codon-induced pausing, but has no effect on frameshifting dependent on pauses induced by nonsense codons. Complementarity between the stimulator and a portion of the accuracy center of the ribosome, Helix 18, implies that the stimulator may directly disrupt error correction by the ribosome.     

Display of somatostatin-related peptides in the complementarity determining regions of an antibody light chain

Peptide display in antibody complementarity determining regions (CDRs) offers several advantages over other peptide display systems including the potential to graft heterologous peptide sequences into multiple positions in the same backbone molecule. Despite the presence of six CDRs in an antibody variable domain, the majority of insertions reported have been made in heavy chain CDR3 (h-CDR3) which may be explained in part by the highly variable length and sequence diversity found in h-CDR3 in native antibodies. The ability to graft peptide sequences into CDRs is restricted by amino acids in these loops that make structural contacts to framework regions or are oriented towards the hydrophobic interior and are important for the proper folding of the antibody. To identify such positions in human kappa-light chain CDR1 (kappa-CDR1) and CDR2 (kappa-CDR2), we performed alignments of 1330 kappa-light chain variable region amino acid sequences and 19 variable region X-ray crystal structures. From analyses of these alignments, we predict insertion points where sequences can be grafted into kappa-CDR1 and kappa-CDR2 to prepare synthetic antibody molecules. We then tested these predictions by inserting somatostatin and somatostatin-related sequences into kappa-CDR1 and kappa-CDR2, and analyzing the expression and ability of the modified antibodies to bind to membranes containing somatostatin receptor 5. These results expand the repertoire of CDRs that can be used for the display of heterologous peptides in the CDRs of antibodies.     

Characterization of the human Ig heavy chain antigen binding complementarity determining region 3 using a newly developed software algorithm, JOINSOLVER

We analyzed 77 nonproductive and 574 productive human V(H)DJ(H) rearrangements with a newly developed program, JOINSOLVER. In the productive repertoire, the H chain complementarity determining region 3 (CDR3(H)) was significantly shorter (46.7 +/- 0.5 nucleotides) than in the nonproductive repertoire (53.8 +/- 1.9 nucleotides) because of the tendency to select rearrangements with less TdT activity and shorter D segments. Using criteria established by Monte Carlo simulations, D segments could be identified in 71.4% of nonproductive and 64.4% of productive rearrangements, with a mean of 17.6 +/- 0.7 and 14.6 +/- 0.2 retained germline nucleotides, respectively. Eight of 27 D segments were used more frequently than expected in the nonproductive repertoire, whereas 3 D segments were positively selected and 3 were negatively selected, indicating that both molecular mechanisms and selection biased the D segment usage. There was no bias for D segment reading frame (RF) use in the nonproductive repertoire, whereas negative selection of the RFs encoding stop codons and positive selection of RF2 that frequently encodes hydrophilic amino acids were noted in the productive repertoire. Except for serine, there was no consistent selection or expression of hydrophilic amino acids. A bias toward the pairing of 5' D segments with 3' J(H) segments was observed in the nonproductive but not the productive repertoire, whereas V(H) usage was random. Rearrangements using inverted D segments, DIR family segments, chromosome 15 D segments and multiple D segments were found infrequently. Analysis of the human CDR3(H) with JOINSOLVER has provided comprehensive information on the influences that shape this important Ag binding region of V(H) chains.     

Identification of a lymphocyte-activating pentapeptide sequence in the Fc region of human IgG1

The synthetic peptide p23, representing residues 335 to 357 in the Fc region of human IgG1, was previously shown to induce Ig secretion in murine spleen cell cultures. In this report, overlapping peptides based on the sequence of p23 were synthesized to further map the active site in this molecule. The results from these studies indicate that leu-pro-pro-ser-arg (residues 351 to 355) retained the B cell differentiation-inducing properties of p23; however, expression of activity by this sequence was markedly influenced by N-flanking sequences. By using T cell-depleted spleen cell cultures, it was determined that at least two signals are required for p23-induced Ig secretion: one supplied by p23 directly and one supplied by a T cell-replacing factor present in p23-conditioned spleen cell supernatants. Both signals were mapped into the sequence leu-pro-pro-ser-arg. However, the latter signal, but not the former signal, again appeared to be influenced by sequences proximal to the active site. These data indicate that although the leu-pro-pro-ser-arg sequence is able to provide both required signals for p23-induced Ig secretion in spleen cell cultures, there may be subtle differences in how the cell types involved in this response interact with and/or are activated by this sequence.     

Structure of oxidized thioredoxin to 4 with 5 A resolution

The structure of the oxidized form of Escherichia coli thioredoxin, space group C2, has been determined from X-ray crystallographic data, to a resolution of 4.5 Å using two heavy-atom derivatives, platinum diaminedichloride and 3-pyridyl mercuric chloride. The electron density maps show the molecular shape and the packing of the thioredoxin molecules as well as the positions of the cupric ions necessary for crystallization of thioredoxin. The shape of the thioredoxin molecule is ellipsoidal with approximate dimensions 25 Å × 34 Å × 35 Å. The two thioredoxin molecules in the asymmetric unit appear very similar. They are related by a translation vector with components (0, 0.1, 0.5) along the axis of the unit cell and not by a 2-fold rotation axis. Each of the two molecules in the asymmetric unit belongs to separate infinite layers of molecules parallel to the xy plane. The basic unit in these layers is a dimer formed by interaction of two thioredoxin molecules across the crystallographic 2-fold axis. The structural role of the cupric ions in the crystal lattice is to bridge these dimers within the layers.     

Synthesis and receptor binding of IgG1 peptides derived from the IgG Fc region

The IgG binding Fcgamma receptors (FcgammaRs) play a key role in defence against pathogens by linking humoral and cell-mediated immune responses. Impaired expression and/or function of FcgammaR may result in the development of pathological autoimmunity. Considering the functions of FcgammaRs, they are potential target molecules for drug design to aim at developing novel anti-inflammatory and immunomodulatory therapies. Previous data mostly obtained by X-ray analysis of ligand-receptor complexes indicate the profound role of the CH2 domain in binding to various FcgammaRs. Our aim was to localize linear segments, which are able to bind and also to modulate the function of the low affinity FcgammaRs, like FcgammaRIIb and FcgammaRIIIa. To this end a set of overlapping octapeptides was prepared corresponding to the 231-298 sequence of IgG1 CH2 domain and tested for binding to human recombinant soluble FcgammaRIIb. Based on these results, a second group of peptides was synthesized and their binding properties to recombinant soluble FcgammaRIIb, as well as to FcgammaRs expressed on the cell surface, was investigated. Here we report that peptide representing the Arg(255)-Ser(267) sequence of IgG1 is implicated in the binding to FcgammaRIIb. In addition we found that peptides corresponding to the Arg(255)-Ser(267), Lys(288)-Ser(298) or Pro(230)-Val(240) when presented in a multimeric form conjugated to branched chain polypeptide in uniformly oriented copies induced the release of TNFalpha, a pro-inflammatory cytokine from MonoMac monocyte cell line. These findings indicate that these conjugated peptides are able to cluster the activating FcgammaRs, and mediate FcgammaR dependent function. Peptide Arg(255)-Ser(267) can also be considered as a lead for further functional studies.     

Coevolution of RNA helix stability and Shine-Dalgarno complementarity in a translational start region

The initiation legion of the coat-protein gene of RNA bacteriophage MS2 adopts a well-defined hairpin structure with the start codon occupying the loop position, while the Shine-Dalgarno (SD) sequence is part of the stem. In a previous study, we introduced mutations in this hairpin that changed its thermo dynamic stability. The resulting phages evolved to regain the wild-type stability by second-site compensatory substitutions. Neither the original nor the suppressor mutations were in the SD region. In the present analysis, we have made changes in the SD region that shorten or extend its complementarity to the 3 end of 16S rRNA and monitored their evolution to a stable pseudorevertant species. Phages in which the SD complementarity was decreased evolved an initiator hairpin of lower stability than wild type while those in which the complementarity was extended evolved a hairpin with an increased stability. We conclude that weaker SD sequences still allow maximal translation if the secondary structure of the ribosome-landing site is destabilized accordingly. Alternatively, translation-initiation regions with a stronger secondary structure still allow maximal expression, if the SD complementarity is extended. These findings support a previously published model in which the SD Interaction helps the ribosome to melt the structure in a translation-initiation region.     

Identification and characterization of a -1 reading frameshift in the heavy chain constant region of an IgG1 recombinant monoclonal antibody produced in CHO cells

Frameshifts lead to complete alteration of the intended amino acid sequences, and therefore may affect the biological activities of protein therapeutics and pose potential immunogenicity risks. We report here the identification and characterization of a novel -1 frameshift variant in a recombinant IgG1 therapeutic monoclonal antibody (mAb) produced in Chinese hamster ovary cells during the cell line selection studies. The variant was initially observed as an atypical post-monomer fragment peak in size exclusion chromatography. Characterization of the fragment peak using intact and reduced liquid chromatography-mass spectrometry (LC-MS) analyses determined that the fragment consisted of a normal light chain disulfide-linked to an aberrant 26 kDa fragment that could not be assigned to any HC fragment even after considering common modifications. Further analysis using LC-MS/MS peptide mapping revealed that the aberrant fragment contained the expected HC amino acid sequence (1-232) followed by a 20-mer novel sequence corresponding to expression of heavy chain DNA sequence in the -1 reading frame. Examination of the DNA sequence around the frameshift initiation site revealed that a mononucleotide repeat GGGGGG located in the IgG1 HC constant region was most likely the structural root cause of the frameshift. Rapid identification of the frameshift allowed us to avoid use of a problematic cell line containing the frameshift as the production cell line. The frameshift reported here may be observed in other mAb products and the hypothesis-driven analytical approaches employed here may be valuable for rapid identification and characterization of frameshift variants in other recombinant proteins.     

Assembly of Trp1 in a signaling complex associated with caveolin-scaffolding lipid raft domains

Trp1 has been proposed as a component of the store-operated Ca(2+) entry (SOC) channel. However, neither the molecular mechanism of SOC nor the role of Trp in this process is yet understood. We have examined possible molecular interactions involved in the regulation of SOC and Trp1 and report here for the first time that Trp1 is assembled in signaling complex associated with caveolin-scaffolding lipid raft domains. Endogenous hTrp1 and caveolin-1 were present in low density fractions of Triton X-100-extracted human submandibular gland cell membranes. Depletion of plasma membrane cholesterol increased Triton X-100 solubility of Trp1 and inhibited carbachol-stimulated Ca(2+) signaling. Importantly, thapsigargin stimulated Ca(2+) influx, but not internal Ca(2+) release, and inositol 1,4,5-triphosphate (IP(3))-stimulated I(soc) were also attenuated. Furthermore, both anti-Trp1 and anti-caveolin-1 antibodies co-immunoprecipitated hTrp1, caveolin-1, Galpha(q/11), and IP(3) receptor-type 3 (IP(3)R3). These results demonstrate that caveolar microdomains provide a scaffold for (i) assembly of key Ca(2+) signaling proteins into a complex and (ii) coordination of the molecular interactions leading to the activation of SOC. Importantly, we have shown that Trp1 is also localized in this microdomain where it interacts with one or more components of this complex, including IP(3)R3. This finding is potentially important in elucidating the physiological function of Trp.     

Downscaling patterns of complementarity to a finer resolution and its implications for conservation prioritization

Given species inventories of all sites in a planning area, integer programming or heuristic algorithms can prioritize sites in terms of the site's complementary value, that is, the ability of the site to complement (add unrepresented species to) other sites prioritized for conservation. The utility of these procedures is limited because distributions of species are typically available only as coarse atlases or range maps, whereas conservation planners need to prioritize relatively small sites. If such coarse‐resolution information can be used to identify small sites that efficiently represent species (i.e., downscaled), then such data can be useful for conservation planning. We develop and test a new type of surrogate for biodiversity, which we call downscaled complementarity. In this approach, complementarity values from large cells are downscaled to small cells, using statistical methods or simple map overlays. We illustrate our approach for birds in Spain by building models at coarse scale (50 × 50 km atlas of European birds, and global range maps of birds interpreted at the same 50 × 50 km grid size), using this model to predict complementary value for 10 × 10 km cells in Spain, and testing how well‐prioritized cells represented bird distributions in an independent bird atlas of those 10 × 10 km cells. Downscaled complementarity was about 63–77% as effective as having full knowledge of the 10‐km atlas data in its ability to improve on random selection of sites. Downscaled complementarity has relatively low data acquisition cost and meets representation goals well compared with other surrogates currently in use. Our study justifies additional tests to determine whether downscaled complementarity is an effective surrogate for other regions and taxa, and at spatial resolution finer than 10 × 10 km cells. Until such tests have been completed, we caution against assuming that any surrogate can reliably prioritize sites for species representation.     

Modulation of the effector functions of a human IgG1 through engineering of its hinge region

We report here the engineering of a humanized anti-human EphA2 mAb (mAb 12G3H11) in an effort to explore the relationship between the hinge of a human IgG1 and its effector functions. mAb 12G3H11, used here as a model, is directed against the human receptor tyrosine kinase EphA2, which is an actively investigated target for cancer therapy due to its up-regulation in many cancer cells. Various rational modifications were introduced into the hinge region of mAb 12G3H11. These mutations were predicted to modulate the hinge's length, flexibility, and/or biochemical properties. We show that the upper and middle hinge both play important, although functionally distinct roles. In particular, middle hinge modifications predicted to decrease its rigidity or length as well as eliminating either one of its two cysteine residues had a strong negative impact on C1q binding and complement-dependent cytotoxicity. Disruption of covalent bonds between both H chains may account in part for these effects. We also describe middle hinge mutants with a significantly decreased ability to bind FcgammaRIIIA and trigger Ab-dependent cell-mediated cytotoxicity. Conversely, we also generated upper hinge mutants exhibiting an increase in C1q binding and complement-dependent cytotoxicity activity. Therefore, this approach represents a novel strategy to fine-tune the biological activity of a given human IgG1. We also define, for the first time in such a systematic fashion, the relationship between various characteristics of the middle and upper hinge and the corresponding effector functions.     

Correlation between splicing sites within an intron and their sequence complementarity with U1 RNA

We have determined the nucleotide sequence of two short introns (respectively 215 and 90 nucleotides) in the chick alpha 2-collagen (type I) gene as well as parts of the adjacent exons. For one of these introns we find that the 5' end of U1 RNA is complementary not only to the two ends of the intron but also to one end of the intron and sequences inside this intron. These complementarities predict three potential internal splicing sites. By S1 mapping experiments we find three discrete RNA precursors in which different portions of this intron have been deleted. The sizes of the deleted segments are in good agreement with the location of the predicted splicing points inside the intron. The DNA sequence indicates that removal of one portion of the intron should still allow the subsequent elimination of the rest of the intron and the correct splicing of the coding segments located at each end of the intron. The new introns created by the first splicing events contain sequences at each end which are also complementary to U1 RNA. Our data indicate that in the intron which we have examined the sequences at the 3' end of the intron are removed before those at the 5' end.     

Engineered glycoforms of an antineuroblastoma IgG1 with optimized antibody-dependent cellular cytotoxic activity

The glycosylation pattern of chCE7, an antineuroblastoma chimeric IgG1, was engineered in Chinese hamster ovary cells with tetracycline-regulated expression of beta(1,4)-N-acetylglucosaminyltransferase III (GnTIII), a glycosyltransferase catalyzing formation of bisected oligosaccharides that have been implicated in antibody-dependent cellular cytotoxicity (ADCC). Measurement of the ADCC activity of chCE7 produced at different tetracycline levels showed an optimal range of GnTIII expression for maximal chCE7 in vitro ADCC activity, and this activity correlated with the level of constant region-associated, bisected complex oligosaccharides determined by matrix-assisted laser desorption/ionization time-of-flight mass spectrometry. The new optimized variants of chCE7 exhibit substantial ADCC activity and, hence, may be useful for treatment of neuroblastoma. The strategy presented here should be applicable to optimize the ADCC activity of other therapeutic IgGs.