A predicted three-dimensional structure for the human immunodeficiency virus binding domains of CD4 antigen

A predicted three-dimensional structure of the two N-terminal extracellular domains of human CD4 antigen, a cell surface glycoprotein, is reported. This region of CD4, particularly the first domain, has been identified as containing the binding region for the envelope gp120 protein of the human immunodeficiency virus. The model was predicted based on the sequence homology of each domain with the variable light chain of immunoglobulins. The framework beta-sheet regions were taken from the crystal coordinates of REI. For one region in the first domain of CD4 there was an ambiguity in the alignment with REI and two alternate models are presented. Loops connecting the framework were modelled from fragments selected from a database of main chain coordinates from all known protein structures. Residues identified as involved in binding gp120 have been located in several other studies within the first domain of CD4. Epitopes from eight monoclonal antibodies have been mapped onto residues in both domains. Competition of these antibodies with each other and with gp120 can be interpreted from the structural model.     

Structural models for carcinoembryonic antigen and its complex with the single-chain Fv antibody molecule MFE23

MFE23 is a single chain Fv antibody that has a high affinity for carcinoembryonic antigen (CEA). A full homology model for CEA based on V-type, I-type and C2-type immunoglobulin folds, 28 oligosaccharides and the interdomain angle of CD2 was validated using solution scattering data. The superimposition of the intermolecular contacts observed in our recent crystal structure of MFE23 with the N-terminal domain of CEA permitted the MFE23-CEA complex to be modelled. Good surface and electrostatic complementarity and carbohydrate-unhindered access of MFE23 with the indentation between the first two CEA domains was observed. The model is supported by biochemical data and provides insight on the high affinity of MFE23 for CEA.     

Homophilic adhesion between Ig superfamily carcinoembryonic antigen molecules involves double reciprocal bonds

Both carcinoembryonic antigen (CEA) and neural cell adhesion molecule (NCAM) belong to the immunoglobulin supergene family and have been demonstrated to function as homotypic Ca(++)-independent intercellular adhesion molecules. CEA and NCAM cannot associate heterotypically indicating that they have different binding specificities. To define the domains of CEA involved in homotypic interaction, hybrid cDNAs consisting of various domains from CEA and NCAM were constructed and were transfected into a CHO-derived cell line; stable transfectant clones showing cell surface expression of CEA/NCAM chimeric-proteins were assessed for their adhesive properties by homotypic and heterotypic aggregation assays. The results indicate that all five of the Ig(C)-like domains of NCAM are required for intercellular adhesion while the COOH-terminal domain containing the fibronectin-like repeats is dispensable. The results also show that adhesion mediated by CEA involves binding between the Ig(V)-like amino-terminal domain and one of the Ig(C)-like internal repeat domains: thus while transfectants expressing constructs containing either the N domain or the internal domains alone were incapable of homotypic adhesion, they formed heterotypic aggregates when mixed. Furthermore, peptides consisting of both the N domain and the third internal repeat domain of CEA blocked CEA-mediated cell aggregation, thus providing direct evidence for the involvement of the two domains in adhesion. We therefore propose a novel model for interactions between immunoglobulin supergene family members in which especially strong binding is effected by double reciprocal interactions between the V-like domains and C-like domains of antiparallel CEA molecules on apposing cell surfaces.     

Crystal structure of a soluble form of the human T cell coreceptor CD8 at 2.6 A resolution.

A secreted fragment of the extracellular portion of human CD8 alpha has been expressed in CHO cells, and a deglycosylated and proteolyzed form of this fragment has been crystallized. We report here the crystal structure of this fragment as refined at 2.6 A resolution. The structure was solved by molecular replacement using a superposition of ten variable domains from immunoglobulin light chains as the search model. Only the N-terminal 114 amino acids of CD8 alpha are visible in the electron density maps. The domain formed by these residues possesses a fold typical of immunoglobulin variable domains and associates to form Fv-like homodimers.     

The three-dimensional structure of the tenth type III module of fibronectin: an insight into RGD-mediated interactions.

The solution structure of the tenth type III module of fibronectin has been determined using nuclear magnetic resonance techniques. The molecule has a fold similar to that of immunoglobulin domains, with seven beta strands forming two antiparallel beta sheets, which pack against each other. Both beta sheets contribute conserved hydrophobic residues to a compact core. The topology is more similar to that of domain 2 of CD4, PapD, and the extracellular domain of the human growth hormone receptor than to that of immunoglobulin C domains. The module contains an Arg-Gly-Asp sequence known to be involved in cell adhesion. This tripeptide is solvent exposed and lies on a conformationally mobile loop between strands F and G, consistent with its cell adhesion function.     

The carcinoembryonic antigen (CEA) contains multiple immunoglobulin-like domains

The amino acid sequences of human carcinoembryonic antigen deduced from the cDNA sequences have been analysed. This antigen contains seven extracellular domains (previously recognized three highly repetitive domains are further divided into A and B subdomains each) which are strikingly homologous to each other and to immunoglobulin variable regions, poly-Ig receptor and Thy 1.1. The N-terminal domain lacks immunoglobulin-like fold but the other six domains have, suggesting that the CEA belongs to immunoglobulin superfamily.     

The structure of an entire noncovalent immunoglobulin kappa light-chain dimer (Bence-Jones protein) reveals a weak and unusual constant domains association.

Monoclonal free light chains secreted in immunoproliferative disorders are frequently involved in renal complications, including a specific proximal tubule impairment, Fanconi's syndrome. The latter is characterized in most cases by intracellular crystallization including a light-chain variable-domain fragment which resists lysosomal proteases. Bence-Jones protein (BJP) DEL was isolated from a patient with myeloma-associated Fanconi's syndrome. The crystal structure of this human kappa immunoglobulin light-chain noncovalent dimer was determined using molecular replacement with the structure of molecule REI, as the variable domain, and that of BJP LOC as the constant domain. To our knowledge, DEL is the first complete kappa BJP structure described to date. The R-factor is 20.7% at 2.8 A resolution. The BJP DEL dimer was compared with other light-chain dimers and with Fab fragments with a kappa light chain. Although the domain-folding pattern was similar, the relative positions of the constant domains differed. BJP DEL showed a noncanonical quaternary structural arrangement which may be attributable to the poor CL-CL affinity and lack of an interchain disulfide bridge, combined with the conformational editing effect of the crystal-packing forces. Our results suggest that, in the absence of a disulfide bridge, most BJP CLs are probably mobile in solution. This may explain their high susceptibility to proteases and the absence of naturally occurring crystals for these dimers. Furthermore, these findings of an unusual quaternary structure of an immunoglobulin light-chain association extend our knowledge about the large and highly diverse structures of the immunoglobulin superfamily.     

Characterization of the surface topography and putative tertiary structure of the human CD7 molecule

The CD7 gp40 molecule is a member of the Ig gene superfamily and is expressed on T cell precursors before their entry into the thymus during fetal development. N-terminal amino acids 1-107 of CD7 are highly homologous to Ig kappa-L chains whereas the carboxyl-terminal region of the extracellular domain of CD7 is proline-rich and has been postulated to form a stalk from which the Ig domain projects. To define potential functional regions of CD7, we have studied the surface topography of the CD7 Ag by synthesizing peptides corresponding to linear sequences within the CD7 extracellular domains, by raising polyclonal anti-CD7 rabbit sera against these peptides, and by computer analysis of the primary CD7 amino acid sequence. Polyclonal anti-CD7 sera were studied using indirect immunofluorescence, RIA, radioimmunoprecipitation, and Western blot assays. Computer analysis was performed comparing the CD7 sequence with all other known protein sequences. We found that three CD7 epitopes defined by peptides CD7-1A (AA 1-38), CD7-4 (AA 48-74), and CD7-7 (AA 129-146) were available for binding antibody on the surface of the CD7 molecule. Using computer analysis, we transposed the amino acid sequence of the CD7 Ig kappa-like N-terminal domain of CD7 onto the spatial coordinates of REI, a previously reported Ig kappa-molecule highly homologous (48%) to the CD7 N-terminal Ig-like region. Based on computer analysis of this putative CD7 three-dimensional structure, both the CD7-1A and CD7-4 regions protruded from the surface of the N-terminal domain of the CD7 molecule. Finally, comparison of the CD7 transmembrane sequence with CD4 and HIV transmembrane sequences and with respiratory syncytial virus fusion sequences demonstrated similar sequence motifs among these molecules.     

Insights into function of PSI domains from structure of the Met receptor PSI domain

PSI domains are cysteine-rich modules found in extracellular fragments of hundreds of signaling proteins, including plexins, semaphorins, integrins, and attractins. Here, we report the solution structure of the PSI domain from the human Met receptor, a receptor tyrosine kinase critical for proliferation, motility, and differentiation. The structure represents a cysteine knot with short regions of secondary structure including a three-stranded antiparallel beta-sheet and two alpha-helices. All eight cysteines are involved in disulfide bonds with the pattern consistent with that for the PSI domain from Sema4D. Comparison with the Sema4D structure identifies a structurally conserved core comprising the N-terminal half of the PSI domain. Interestingly, this part links adjacent SEMA and immunoglobulin domains in the Sema4D structure, suggesting that the PSI domain serves as a wedge between propeller and immunoglobulin domains and is responsible for the correct positioning of the ligand-binding site of the receptor.     

The retinal specific CD147 Ig0 domain: from molecular structure to biological activity

CD147 is a type I transmembrane protein that is involved in inflammatory diseases, cancer progression, and multiple human pathogens utilize CD147 for efficient infection. CD147 expression is so high in several cancers that it is now used as a prognostic marker. The two primary isoforms of CD147 that are related to cancer progression have been identified, differing in their number of immunoglobulin (Ig)-like domains. These include CD147 Ig1-Ig2, which is ubiquitously expressed in most tissues, and CD147 Ig0-Ig1-Ig2, which is retinal specific and implicated in retinoblastoma. However, little is known in regard to the retinal specific CD147 Ig0 domain despite its potential role in retinoblastoma. We present the first crystal structure of the human CD147 Ig0 domain and show that the CD147 Ig0 domain is a crystallographic dimer with an I-type domain structure, which maintained in solution. Furthermore, we have utilized our structural data together with mutagenesis to probe the biological activity of CD147-containing proteins, both with and without the CD147 Ig0 domain, within several model cell lines. Our findings reveal that the CD147 Ig0 domain is a potent stimulator of interleukin-6 and suggest that the CD147 Ig0 domain has its own receptor distinct from that of the other CD147 Ig-like domains, CD147 Ig1-Ig2. Finally, we show that the CD147 Ig0 dimer is the functional unit required for activity and can be disrupted by a single point mutation.     

Molecular Modeling of Immunoglobulin Superfamily Proteins: Predicting the Three Dimensional Structure of the Extracellular Domain of CTLA-4 (CD152)

The interactions between CD28/CTLA-4 (CD152) on T cells and their ligands CD80/CD86 on antigen presenting cells provide costimulatory signals critical for T cell activation. CD28/CTLA-4 and CD80/CD86 are members of the immunoglobulin superfamily (IgSF). CD28 and CTLA-4 both contain a single extracellular immunoglobulin (Ig) domain which binds CD80/CD86. Here we report modeling studies on the three-dimensional (3D) structure of the CTLA-4 binding domain. Since CTLA-4 displays only very weak sequence homology to proteins with known 3D structure, conventional modeling techniques were difficult to apply. Structure-oriented sequence comparison, consensus residue analysis, conformational searching, and inverse folding calculations were employed to aid in the generation of a comparative CTLA-4 model. Regions of high and low prediction confidence were identified, and the sequence-structure compatibility of the model was determined. Characteristics of the modeled structure, which resembles an Ig V domain, were analyzed, and the model was used to map N-linked glycosylation sites and residues critical for CTLA-4 function. The modeling approach described here can be applied to predict 3D structures of other IgSF proteins.     

Structure of a novel Bence-Jones protein (Rhe) fragment at 1.6 A resolution

The crystal structure of Rhe, a lambda-type Bence-Jones protein fragment, has been solved and refined to a resolution of 1.6 A. A model fragment consisting of the complete variable domain and the first three residues of the constant domain yields a crystallographic residual RF value of 0.149. The protein exists as a dimer both in solution and in the crystals. Although the "immunoglobulin fold" is generally preserved in the structure, there are significant differences in both the monomer conformation and in the mode of association of monomers into dimers, when compared to other known Bence-Jones proteins or Fab fragments. The variations in conformation within monomers are particularly significant as they involve non-hypervariable residues, which previously were believed to be part of a "structurally invariant" framework common to all immunoglobulin variable domains. The novel mode of dimerization is equally important, as it can result in combining site shapes and sizes unobtainable with the conventional mode of dimerization. A comparison of the structure with other variable domain dimers reveals further that the variations within monomers and between domains in the dimer are coupled. Some possible functional implications revealed by this coupling are greater variability, induced fitting of the combining site to better accommodate antigenic determinants, and a mechanism for relaying binding information from one end of the variable domain dimer to the other. In addition to providing the most accurate atomic parameters for an immunoglobulin domain yet obtained, the high resolution and extensive refinement resulted in identification of several tightly bound water molecules in key structural positions. These water molecules may be regarded as integral components of the protein. Other water molecules appear to be required to stabilize the novel conformation.     

Deglycosylation to obtain stable and homogeneous Pichia pastoris-expressed N-A1 domains of carcinoembryonic antigen

Carcinoembryonic antigen (CEA) is a seven domain membrane glycoprotein widely used as a tumour marker for adenocarcinomas and as a target for antibody-directed therapies. Structural models have proposed that the first two domains of CEA (the N terminal and adjoining A1 domains) bind MFE-23, a single chain Fv antibody in experimental clinical use. We aimed to produce recombinant N-A1 to test this hypothesis. The N-A1 domains were expressed as soluble protein with a C-terminal hexahistidine tag (His6-tag) in the yeast Pichia pastoris. His6-tagged N-A1 was captured from the supernatant by batch purification with copper-loaded Streamline Chelating, an immobilised metal affinity chromatography (IMAC) matrix usually utilised in expanded bed techniques. Purified N-A1 was heterogeneous with a molecular weight range from 38 to 188 kDa. Deglycosylation with endoglycosidase H (Endo H) resulted in three discrete molecular weight forms of N-A1, one partially mannosylated, one fully Endo H-digested and one fully Endo H-digested but lacking the His6-tag. These were separated by concanavalin A chromatography followed by HiTrap IMAC. The procedure resulted in single-band-purity, mannose-free N-A1. The binding interaction of MFE-23 to N-A1 was analysed by surface plasmon resonance. The affinity constants retrieved were KD = 4.49 x 10(-9)M for the P. pastoris expressed, native N-A1, and 5.33 x 10(-9) M for the Endo H-treated N-A1. To our knowledge this is the first time that two consecutive domains of CEA have been stably expressed and purified from P. pastoris. This work confirms that the CEA epitope recognised by MFE-23 resides in N-A1.     

Molecular Modeling of Immunoglobulin Superfamily Proteins: Predicting the Three Dimensional Structure of the Extracellular Domain of CTLA-4 (CD152)

The interactions between CD28/CTLA-4 (CD152) on T cells and their ligands CD80/CD86 on antigen presenting cells provide costimulatory signals critical for T cell activation. CD28/CTLA-4 and CD80/CD86 are members of the immunoglobulin superfamily (IgSF). CD28 and CTLA-4 both contain a single extracellular immunoglobulin (Ig) domain which binds CD80/CD86. Here we report modeling studies on the three-dimensional (3D) structure of the CTLA-4 binding domain. Since CTLA-4 displays only very weak sequence homology to proteins with known 3D structure, conventional modeling techniques were difficult to apply. Structure-oriented sequence comparison, consensus residue analysis, conformational searching, and inverse folding calculations were employed to aid in the generation of a comparative CTLA-4 model. Regions of high and low prediction confidence were identified, and the sequence-structure compatibility of the model was determined. Characteristics of the modeled structure, which resembles an Ig V domain, were analyzed, and the model was used to map N-linked glycosylation sites and residues critical for CTLA-4 function. The modeling approach described here can be applied to predict 3D structures of other IgSF proteins.Supplementary material to this paper is available in electronic form at http://dx.doi.org/10.1007/s008940050025     

NMR structure of the natural killer cell receptor 2B4 (CD244): implications for ligand recognition

2B4, a transmembrane receptor expressed primarily on natural killer (NK) cells and on a subset of CD8(+) T cells, plays an important role in activating NK-mediated cytotoxicity through its interaction with CD48 on target cells. We report here the atomic-resolution structure of the ligand-binding (D1) domain of 2B4 in solution determined by nuclear magnetic resonance (NMR) spectroscopy. The overall main chain structure resembles an immunoglobulin variable (V) domain fold, very similar to that seen previously for domain 1 of CD2 and CD4. The structure contains nine beta-strands assembled into two beta-sheets conventionally labeled DEB and AGFCC'C' '. The six-stranded sheet (AGFCC'C' ') contains structural features that may have implications for ligand recognition and receptor function. A noncanonical disulfide bridge between Cys2 and Cys99 stabilizes a long and parallel beta-structure between strand A (residues 3-12) and strand G (residues 100-108). A beta-bulge at residues Glu45 and Ile46 places a bend in the middle of strand C' that orients two conserved and adjacent hydrophobic residues (Ile46 and Leu47) inside the beta-sandwich as seen in other V domains. Finally, the FG-loop (implicated in ligand recognition in the CD2-CD58 complex) is dynamically disordered in 2B4 in the absence of a ligand. We propose that ligand binding to 2B4 might stabilize the structure of the FG-loop in the ligand complex.     

Structural model of human endoglin, a transmembrane receptor responsible for hereditary hemorrhagic telangiectasia

Endoglin is a type I membrane protein expressed as a disulphide-linked homodimer on human vascular endothelial cells whose haploinsufficiency is responsible for the dominant vascular dysplasia known as hereditary hemorrhagic telangiectasia (HHT). Structurally, endoglin belongs to the zona pellucida (ZP) family of proteins that share a ZP domain of ∼ 260 amino acid residues at their extracellular region. Endoglin is a component of the TGF-β receptor complex, interacts with the TGF-β signalling receptors types I and II, and modulates cellular responses to TGF-β. Here, we have determined for the first time the three-dimensional structure of the ∼ 140 kDa extracellular domain of endoglin at 25 Å resolution, using single-particle electron microscopy (EM). This reconstruction provides the general architecture of endoglin, which arranges as a dome made of antiparallel oriented monomers enclosing a cavity at one end. A high-resolution structure of endoglin has also been modelled de novo and found to be consistent with the experimental reconstruction. Each subunit comprises three well-defined domains, two of them corresponding to ZP regions, organised into an open U-shaped monomer. This domain arrangement was found to closely resemble the overall structure derived experimentally and the three modelled de novo domains were tentatively assigned to the domains observed in the EM reconstruction. This molecular model was further tested by tagging endoglin's C terminus with an IgG Fc fragment visible after 3D reconstruction of the labelled protein. Combined, these data provide the structural framework to interpret endoglin's functional domains and mutations found in HHT patients.     

Identification and mapping on human chromosome 19 of chromosome-specific repeated element preferentially binding with nuclear matrix]

From a library of sequences binding preferentially to nuclear matrix (matrix attachment regions, MARs), a fragment of about 300 bp in length (CEA (carcinoembryonic antigen)-MAR) was isolated and characterized. The CEA-MAR sequence was found in more than ten loci of chromosome 19 containing elements similar to genes of the CEA family. No sequences of this group were found on other human chromosomes. Two CEA-MAR-containing loci were sequenced, and sequences for another seven loci were found in GenBank. A comparative analysis of CEA-MARs and the flanking sequences is reported. Based on the sequence of the CEA-containing chromosome 19 loci, a hypothetical model of the domain structure of a 2-Mb chromosome region was constructed and the mutual arrangement of CEA-MARs and genes of CEA family was elucidated. The CEA-MARs were located 5-20 kb downstream of the CEA genes. These results suggest that the duplication unit of the CEA family may coincide with chromatin domains containing these genes.     

Enzymatic cleavage of a CD4 immunoadhesin generates crystallizable, biologically active Fd-like fragments

CD4, the cell-surface receptor for the human immunodeficiency virus (HIV), is a member of the immunoglobulin (Ig) gene superfamily. It contains four extracellular sequences homologous to Ig VL domains. The first of these (V1) is sufficient for binding to HIV; however, the structural basis for this binding has yet to be elucidated. While several models for the structure of Ig-like domains in CD4 have been proposed on the basis of crystal structures of Ig VL domains, direct evidence that CD4 and VL domains fold similarly has not been obtained. To produce individual domains of CD4 for structural studies, we used molecular fusions of such domains with Ig heavy chain (CD4 immunoadhesins), which are very efficiently expressed and secreted in mammalian cells and can be easily isolated in single-step purification with protein A. Since these fusion molecules are antibody-like homodimeric proteins, we investigated the possibility that they might be cleaved enzymatically to produce Fd-like and Fc fragments. We found that cleavage with papain releases an Fd-like fragment containing the V1 and V2 CD4 domains; this fragment fully retains the ability to bind to the HIV-1 envelope glycoprotein gp120 and to block HIV infection in vitro. Moreover, folding of the CD4 domains in the Fd-like fragment and in the parent immunoadhesin is indistinguishable, as indicated by circular dichroism. Spectral analysis of the Fd-like fragment suggests that secondary structure content is identical with that predicted from the known structure of Ig VL domains; this directly supports the hypothesis that the V1 and V2 domains of CD4 fold similarly to Ig VL domains.(ABSTRACT TRUNCATED AT 250 WORDS)     

High level expression in Chinese hamster ovary cells of soluble forms of CD4 T lymphocyte glycoprotein including glycosylation variants

The CD4 cell surface antigen is of interest as a marker of T lymphocytes that recognize foreign antigens in the context of MHC Class II antigen, as a receptor for the human immunodeficiency virus (HIV) and as a member of the immunoglobulin superfamily (IgSF) with four Ig-like domains present in the extracellular domain. In order to produce large amounts of soluble CD4 for x-ray crystallography and other molecular studies, a recently developed expression system based on selection via glutamine synthetase was used. Expression was attempted for rat CD4 corresponding to the full extracellular sequence (sCD4; domains 1-4), the NH2-terminal half (domains 1 and 2) and the first domain alone. Stable transfected Chinese hamster ovary cell lines were obtained that expressed sCD4 and sCD4 (half) at typical maximal levels in spent tissue culture supernatant of greater than 80 and 25 mg/liter, respectively. Domain 1 alone was not expressed and introduction of a N-linked glycosylation site did not facilitate expression. The role of glycosylation in the expression of sCD4 was investigated by mutagenesis of the constructs to remove each of the two N-linked glycosylation sites in turn and both together. All three forms were expressed at 60-120 mg/liter. The sCD4 (half) was not expressed after deletion of its N-linked site. The disulfide bonds of sCD4 were determined to be within domains 1, 2, and 4 and isolation of glycopeptides showed that both N-linked sites were glycosylated. Analysis of the hydrodynamic properties of sCD4 suggested that the molecule adopted an extended conformation in solution rather than folding to form a compact structure like an Fab. The possibility of dimerisation of CD4 was investigated but sCD4 dimers were not seen at an affinity cut-off of about 4 x 10(5) M-1.