Structural analysis of mutants of high-affinity and low-affinity p-azophenylarsonate-specific antibodies generated by alanine scanning of heavy chain complementarity-determining region 2.

Alanine scanning was used to determine the affinity contributions of 10 side chain amino acids (residues at position 50-60 inclusive) of H chain complementarity-determining region 2 (HCDR2) of the somatically mutated high-affinity anti-p-azophenylarsonate Ab, 36-71. Each mutated H chain gene was expressed in the context of mutated (36-71L) and the unmutated (36-65L) L chains to also assess the contribution of L chain mutations to affinity. Combined data from fluorescence quenching, direct binding, inhibition, and capture assays indicated that mutating H:Tyr(50) and H:Tyr(57) to Ala in the 36-71 H chain results in significant loss of binding with both mutated (36-71L) or unmutated (36-65L) L chain, although the decrease was more pronounced when unmutated L chain was used. All other HCDR2 mutations in 36-71 had minimal effect on Ab affinity when expressed with 36-71 L chain. However, in the context of unmutated L chain, of H:Gly(54) to Ala resulted in significant loss of binding, while Abs containing Asn(52) to Ala, Pro(53) to Ala, or Ile(58) to Ala mutation exhibited 4.3- to 7.1-fold reduced affinities. When alanine scanning was performed instead on certain HCDR2 residues of the germline-encoded (unmutated) 36-65 Ab and expressed with unmutated L chain as Fab in bacteria, these mutants exhibited affinities similar to or slightly higher than the wild-type 36-65. These findings indicate an important role of certain HCDR2 side chain residues on Ab affinity and the constraints imposed by L chain mutations in maintaining Ag binding.

Interactions of human NKG2D with its ligands MICA, MICB, and homologs of the mouse RAE-1 protein family.

NKG2D is an activating receptor that is expressed on most natural killer (NK) cells, CD8 alphabeta T cells, and gammadelta T cells. Among its ligands is the distant major histocompatibility complex class I homolog MICA, which has no function in antigen presentation but is induced by cellular stress. To extend previous functional evidence, the NKG2D-MICA interaction was studied in isolation. NKG2D homodimers formed stable complexes with monomeric MICA in solution, demonstrating that no other components were required to facilitate this interaction. MICA glycosylation was not essential but enhanced complex formation. Soluble NKG2D also bound to cell surface MICB, which has structural and functional properties similar to those of MICA. Moreover, NKG2D stably interacted with surface molecules encoded by three newly identified cDNA sequences (N2DL-1, -2, and -3), which are identical to the human ULBP proteins and may represent homologs of the mouse retinoic acid-early inducible family of NKG2D ligands. Because of the substantial sequence divergence among these molecules, these results indicated promiscuous modes of receptor binding. Comparison of allelic variants of MICA revealed large differences in NKG2D binding that were associated with a single amino acid substitution at position 129 in the alpha2 domain. Varying affinities of MICA alleles for NKG2D may affect thresholds of NK-cell triggering and T-cell modulation.

Complex structure of the activating immunoreceptor NKG2D and its MHC class I-like ligand MICA.

The major histocompatibility complex (MHC) class I homolog, MICA, is a stress-inducible ligand for NKG2D, a C-type lectin-like activating immunoreceptor. The crystal structure of this ligand-receptor complex that we report here reveals an NKG2D homodimer bound to a MICA monomer in an interaction that is analogous to that seen in T cell receptor-MHC class I protein complexes. Similar surfaces on each NKG2D monomer interact with different surfaces on either the alpha1 or alpha2 domains of MICA. The binding interactions are large in area and highly complementary. The central section of the alpha2-domain helix, disordered in the structure of MICA alone, is ordered in the complex and forms part of the NKG2D interface. The extensive flexibility of the interdomain linker of MICA is shown by its altered conformation when crystallized alone or in complex with NKG2D.

Ligand preference inferred from the structure of neutrophil gelatinase associated lipocalin.

Neutrophil gelatinase associated lipocalin (NGAL), a constituent of neutrophil granules, is a member of the lipocalin family of binding proteins. NGAL can also be highly induced in epithelial cells in both inflammatory and neoplastic colorectal disease. NGAL is proposed to mediate inflammatory responses by sequestering neutrophil chemoattractants, particularly N-formylated tripeptides and possibly leukotriene B(4) and platelet activating factor. The crystal structures of NGAL display a typical lipocalin fold, albeit with an unusually large and atypically polar binding site, or calyx. The fold of NGAL is most similar to the epididymal retinoic acid-binding protein, another lipocalin, though the overall architecture of the calyces are very different. The crystal structures also reveal either sulfate ions or an adventitiously copurified fatty acid bound in the binding site. Neither ligand is displaced by added N-formylated tripeptides. The size, shape, and character of the NGAL calyx, as well as the low relative affinity for N-formylated tripeptides, suggest that neither the copurified fatty acid nor any of the proposed ligands are likely to be the preferred ligand of this protein. Comparisons between the crystal structures and the recently reported solution structure of NGAL reveal significant differences, in terms of both the details of the structure and the overall flexibility of the fold.

Crystal structure of the MHC class I homolog MIC-A, a gammadelta T cell ligand.

The major histocompatibility complex (MHC) class I homolog MIC-A functions as a stress-inducible antigen that is recognized by a subset of gammadelta T cells independent of beta2-microglobulin and bound peptides. Its crystal structure reveals a dramatically altered MHC class I fold, both in detail and overall domain organization. The only remnant of a peptide-binding groove is a small cavity formed as the result of disordering a large section of one of the groove-defining helices. Loss of beta2-microglobulin binding is due to a restructuring of the interaction interfaces. Structural mapping of sequence variation suggests potential receptor binding sites on the underside of the platform on the side opposite of the surface recognized by alphabeta T cell receptors on MHC class I-peptide complexes.

Expression, purification, crystallization and crystallographic characterization of dimeric and monomeric human neutrophil gelatinase associated lipocalin (NGAL).

Crystals of the monomeric and dimeric forms of human neutrophil gelatinase associated lipocalin have been grown in hanging-drop vapor-diffusion trials using PEG as a precipitating agent with recombinant protein expressed in a baculovirus-based system. Crystals of monomeric NGAL belong to the cubic space group P432 with lattice constants a = b = c = 126.6 A; crystals of dimeric NGAL belong to the tetragonal space group P41212 (or its enantiomorph P43212) with lattice constants a = b = 54.14 and c = 121.56 A. Isomorphous crystals of the NGAL dimer can be grown in the presence of ligand: the tripeptide N-formyl-Met-Leu-Phe.

Expression, purification, crystallization and crystallographic characterization of the human MHC class I related protein MICA.

Crystals of the human MHC-encoded molecule MICA, a homologue of MHC class I proteins, have been grown in hanging-drop vapor-diffusion trials using ammonium sulfate as a precipitating agent with recombinant protein expressed in a baculovirus-based system. Cryo-preserved crystals of MICA belong to the cubic space group F4132 with lattice constants a = b = c = 260.7 A and diffract to a resolution limit of 3.0 A when cryo-preserved. These crystals do not diffract when handled conventionally.

The carboxyl terminus of the human foamy virus Gag protein contains separable nucleic acid binding and nuclear transport domains.

The Gag protein of human foamy virus (HFV) lacks Cys-His boxes present in the nucleocapsid (NC) domains of other retroviruses; instead it contains three glycine-arginine-rich motifs (GR boxes). We have expressed the carboxyl end of HFV Gag containing the GR boxes (the NC domain equivalent) and analyzed its nucleic acid binding properties. Our results show that the NC domain of HFV Gag binds with high affinity to both RNA and DNA, in a sequence-independent manner, as determined by filter binding assays. Analysis of a mutant containing a heterologous sequence in place of GR box I indicates that this motif is required for nucleic acid binding and for viral replication. A mutant in GR box II still binds to RNA and DNA in vitro, but virus containing this mutation does not replicate and no nuclear staining of the Gag protein is found in transfected cells. Surprisingly, a revertant from this mutant that completely lacks GR box II and exhibits very little nuclear transport of Gag can readily replicate in tissue culture. This finding thus provides a direct evidence that although the sequences in GR box II can serve as a nuclear transport signal, they are not required for HFV replication and it is unlikely that nuclear localization of Gag protein plays any critical role during viral infection. Taken together, our results suggest that the Gag protein of HFV may be more analogous to the core protein of the hepatitis B virus family than to conventional retroviral Gag protein.

Engineering and expression of a secreted murine TCR with reduced N-linked glycosylation.

Structural studies of TCR-alpha beta heterodimers would be greatly aided by the ability to produce nonchimeric, secreted material with less carbohydrate heterogeneity. Here, we report the engineering and expression of variants of the murine TCR 2B4 in which many of the potential N-linked glycosylation sites were eliminated. Specific truncations proximal to the transmembrane region were also introduced that result in a secreted heterodimer. Although elimination of N-linked oligosaccharide on the beta-chain does not significantly affect the expression levels of 2B4 heterodimers, ablation of N-linked oligosaccharide on the alpha-chain results in a measurable reduction in expression levels of membrane-associated molecules. Secreted forms of 2B4 heterodimers in which the N-linked glycosylation of the beta-chain has been eliminated can be expressed. The secreted receptor is shown by a variety of Ab determinants to be indistinguishable from native material.

Expression and crystallization of a soluble form of Drosophila fasciclin III.

A truncated form of Drosophila fasciclin III has been engineered by site-directed mutagenesis. Secreted fasciclin III is expressed at 35 to 40 mg/l in insect cells with baculovirus carrying the recombinant gene. Single crystals of purified soluble fasciclin III have been grown by vapor diffusion versus polyethylene glycol 8000/sodium citrate at low pH. The space group is P6(1)22 or its enantiomorph P6(5)22, with unit cell dimensions a = b = 140 A, c = 260 A. Cryo-preserved crystals diffract to reciprocal lattice spacings beyond 3.0 A.

26-10 Fab-digoxin complex: affinity and specificity due to surface complementarity.

We have determined the three-dimensional structures of the antigen-binding fragment of the anti-digoxin monoclonal antibody 26-10 in the uncomplexed state at 2.7 A resolution and as a complex with digoxin at 2.5 A resolution. Neither the antibody nor digoxin undergoes any significant conformational changes upon forming the complex. Digoxin interacts primarily with the antibody heavy chain and is oriented such that the carbohydrate groups are exposed to solvent and the lactone ring is buried in a deep pocket at the bottom of the combining site. Despite extensive interactions between antibody and antigen, no hydrogen bonds or salt links are formed between 26-10 and digoxin. Thus the 26-10-digoxin complex is unique among the known three-dimensional structures of antibody-antigen complexes in that specificity and high affinity arise primarily from shape complementarity.

Conservation of binding site geometry among p-azophenylarsonate-specific antibodies.

Murine A/J anti-p-azophenylarsonate mAb that express a dominant cross-reactive Id are encoded by a single set of germ-line VH and VL region genes. The crystal structure of the Fab of antibody 36-71, which uses this canonical set of genes but is somatically mutated, was previously determined. An Fab 36-71:phenylarsonate complex was modeled, identifying amino acid side chains that were proposed as contact residues to hapten. The remarkable conservation of these residues among canonical anti-p-azophenylarsonate antibodies suggested that the overall binding site geometry was maintained among somatically mutated antiarsonate monoclonal antibodies. To test this hypothesis, we used the germ-line-encoded antibody 36-65 to construct mutant antibodies, using oligonucleotide-directed mutagenesis, which differed only at the putative H chain hapten-contacting residues, and measured their hapten binding. A framework residue at H chain position 47 involved in a hydrogen bond network with CDR residues was also mutated. Substitution of several amino acids at each position permitted evaluation of the stereochemical requirements for binding. The results indicate the importance of aromatic stacking of two H chain tyrosine residues against the phenyl ring of the hapten in maintaining affinity, as well as strict complementarity at H chain position 35. The results are consistent with the crystal model of the combining site, and provide further evidence for conservation of the three-dimensional binding site motif among antiarsonate antibodies that bear a dominant heritable ld.

Crystallization of Sindbis virus and its nucleocapsid.

Crystals of Sindbis virus, which contains a lipid-bilayer membrane, have been grown using polyethylene glycol. The space group is R32, a = b = 640 A, c = 1520 A. The crystals are highly mosaic, and recorded diffraction is therefore restricted to spacings of about 30 A. The crystals show that the packing of glycoproteins E1 and E2 in the icosahedral outer shell is sufficiently precise that it permits regular and repeated interactions between virus particles in the lattice. Crystals of Sindbis nucleocapsids have also been grown. The limited diffraction data are consistent with close packing of nucleocapsids 404 A in diameter.

Three-dimensional structure of murine anti-p-azophenylarsonate Fab 36-71. 1. X-ray crystallography, site-directed mutagenesis, and modeling of the complex with hapten.

The structure of the antigen-binding fragment (Fab) of an anti-p-azophenylarsonate monoclonal antibody, 36-71, bearing a major cross-reactive idiotype of A/J mice has been refined to an R factor of 24.8% at a resolution of 1.85 A. The previously solved partial structure of this Fab at a resolution of 2.9 A (Rose et al., 1990) was used as an initial model for refinement against the high-resolution data. The complex with hapten has been modeled by docking the small-molecule crystal structure of phenylarsonic acid into the structure of the native Fab on the basis of a low-resolution electron density map of the complex. In this model, residue Arg-96 in the light chain and residues Asn-35, Trp-47, and Ser-99 in the heavy chain contact the arsonate moiety of the hapten; an additional bond is found between the arsonate group and a tightly bound water molecule. The phenyl moiety of the hapten packs against two tyrosine side chains at positions 50 and 106 in the heavy chain. Residue Arg-96 in the light chain had been implicated as involved in hapten binding on the basis of previous experiments, and indeed, this residue appears to play a crucial role in this model. Experiments employing site-directed mutagenesis directly support this conclusion. The heavy-chain complementarity-determining regions have novel conformations not previously observed in immunoglobulins except for the recently solved anti-p-azophenylarsonate Fab R 19.9 (Lascombe et al., 1989).