Frederick A. Saul

Structural and functional insights into the malaria parasite moving junction complex.

Members of the phylum Apicomplexa, which include the malaria parasite Plasmodium, share many features in their invasion mechanism in spite of their diverse host cell specificities and life cycle characteristics. The formation of a moving junction (MJ) between the membranes of the invading apicomplexan parasite and the host cell is common to these intracellular pathogens. The MJ contains two key parasite components: the surface protein Apical Membrane Antigen 1 (AMA1) and its receptor, the Rhoptry Neck Protein (RON) complex, which is targeted to the host cell membrane during invasion. In particular, RON2, a transmembrane component of the RON complex, interacts directly with AMA1. Here, we report the crystal structure of AMA1 from Plasmodium falciparum in complex with a peptide derived from the extracellular region of PfRON2, highlighting clear specificities of the P. falciparum RON2-AMA1 interaction. The receptor-binding site of PfAMA1 comprises the hydrophobic groove and a region that becomes exposed by displacement of the flexible Domain II loop. Mutations of key contact residues of PfRON2 and PfAMA1 abrogate binding between the recombinant proteins. Although PfRON2 contacts some polymorphic residues, binding studies with PfAMA1 from different strains show that these have little effect on affinity. Moreover, we demonstrate that the PfRON2 peptide inhibits erythrocyte invasion by P. falciparum merozoites and that this strong inhibitory potency is not affected by AMA1 polymorphisms. In parallel, we have determined the crystal structure of PfAMA1 in complex with the invasion-inhibitory peptide R1 derived by phage display, revealing an unexpected structural mimicry of the PfRON2 peptide. These results identify the key residues governing the interactions between AMA1 and RON2 in P. falciparum and suggest novel approaches to antimalarial therapeutics.

Structures of synthetic O-antigen fragments from serotype 2a Shigella flexneri in complex with a protective monoclonal antibody

The anti-LPS IgG mAb F22-4, raised against Shigella flexneri serotype 2a bacteria, protects against homologous, but not heterologous, challenge in an experimental animal model. We report the crystal structures of complexes formed between Fab F22-4 and two synthetic oligosaccharides, a decasaccharide and a pentadecasaccharide that were previously shown to be both immunogenic and antigenic mimics of the S. flexneri serotype 2a O-antigen. F22-4 binds to an epitope contained within two consecutive 2a serotype pentasaccharide repeat units (RU). Six sugar residues from a contiguous nine-residue segment make direct contacts with the antibody, including the nonreducing rhamnose and both branching glucosyl residues from the two RUs. The glucosyl residue, whose position of attachment to the tetrasaccharide backbone of the RU defines the serotype 2a O-antigen, is critical for recognition by F22-4. Although the complete decasaccharide is visible in the electron density maps, the last four pentadecasaccharide residues from the reducing end, which do not contact the antibody, could not be traced. Although considerable mobility in the free oligosaccharides can thus be expected, the conformational similarity between the individual RUs, both within and between the two complexes, suggests that short-range transient ordering to a helical conformation might occur in solution. Although the observed epitope includes the terminal nonreducing residue, binding to internal epitopes within the polysaccharide chain is not precluded. Our results have implications for vaccine development because they suggest that a minimum of two RUs of synthetic serotype 2a oligosaccharide is required for optimal mimicry of O-Ag epitopes.

Genome-wide regulon and crystal structure of BlaI (Rv1846c) from Mycobacterium tuberculosis

Comparative genomics with Staphylococcus aureus suggested the existence of a regulatory system governing beta‐lactamase (BlaC) production in Mycobacterium tuberculosis. The crystal structure of Rv1846c, a winged helix regulator of previously unknown function, was solved thus revealing strong similarity to the BlaI and MecI repressors of S. aureus, which both respond to beta‐lactam treatment. Using chromatin immunoprecipitation and hybridization to microarrays (ChIP‐on‐chip), the Rv1846c regulon was shown to comprise five separate genomic loci. Two of these mediate responses and resistance to beta‐lactam antibiotics (rv1845c, rv1846c–rv1847; blaC–sigC); two encode membrane proteins of unknown function (rv1456c, rv3921c) while the last codes for ATP synthase (rv1303–atpBEFHAGDC–rv1312). The ChIP‐on‐chip findings were confirmed independently using electrophoretic mobility shift assays, DNAse footprinting and transcript analysis leading to Rv1846c being renamed BlaI. When cells were treated with beta‐lactams, BlaI was released from its operator sites causing derepression of the regulon and upregulation of ATP synthase transcription. The existence of a potential regulatory loop between cell wall integrity and ATP production was previously unknown.

Crystal structure of Urtica dioica agglutinin, a superantigen presented by MHC molecules of class I and class II.

BACKGROUND Urtica dioica agglutinin (UDA), a monomeric lectin extracted from stinging nettle rhizomes, is specific for saccharides containing N-acetylglucosamine (GlcNAc). The lectin behaves as a superantigen for murine T cells, inducing the exclusive proliferation of Vbeta8.3(+) lymphocytes. UDA is unique among known T cell superantigens because it can be presented by major histocompatibility complex (MHC) molecules of both class I and II. RESULTS The crystal structure of UDA has been determined in the ligand-free state, and in complex with tri-acetylchitotriose and tetra-acetylchitotetraose at 1.66 A, 1.90 A and 1.40 A resolution, respectively. UDA comprises two hevein-like domains, each with a saccharide-binding site. A serine and three aromatic residues at each site form the principal contacts with the ligand. The N-terminal domain binding site can centre on any residue of a chito-oligosaccharide, whereas that of the C-terminal domain is specific for residues at the nonreducing terminus of the ligand. We have shown previously that oligomers of GlcNAc inhibit the superantigenic activity of UDA and that the lectin binds to glycans on the MHC molecule. We show that UDA also binds to glycans on the T cell receptor (TCR). CONCLUSIONS The presence of two saccharide-binding sites observed in the structure of UDA suggests that its superantigenic properties arise from the simultaneous fixation of glycans on the TCR and MHC molecules of the T cell and antigen-presenting cell, respectively. The well defined spacing between the two binding sites of UDA is probably a key factor in determining the specificity for Vbeta8.3(+) lymphocytes.

Comparative structural studies of two natural isoforms of ammodytoxin, phospholipases A2 from Vipera ammodytes ammodytes which differ in neurotoxicity and anticoagulant activity.

Ammodytoxin A (AtxA) and its natural isoform AtxC from the venom of Vipera ammodytes ammodytes belong to group IIA-secreted phospholipases A(2) which catalyze the hydrolysis of glycerophospholipids and exhibit strong neurotoxic and anticoagulant effects. The two isoforms, which differ in sequence by only two amino acid residues (Phe124>Ile and Lys128>Glu), display significant differences in toxicity and anticoagulant properties and act on protein targets including neurotoxic proteic receptors and coagulation factor Xa with significantly different strengths of binding. In order to characterize the structural basis of these functional differences, we have determined the crystal structures of the two isoforms. Comparison of the structures shows that the mutation Lys128>Glu in AtxC could perturb interactions with FXa, resulting in lower anticoagulant activity, since the side chain of Glu128 is partly buried, making a stabilizing hydrogen bond with the main-chain nitrogen atom of residue Thr35. This interaction leads to a displacement of the main polypeptide chain at positions 127 and 128 (identified by mutagenesis as important for interaction with FXa), and a different orientation of the side chain of unmutated Lys127. The mutation Phe124>Ile in AtxC induces no significant conformational changes, suggesting that the differences in toxicity of the two isoforms are due essentially to differences in surface complementarity in the interaction of the toxin with the neurotoxic protein receptor. The crystal structures also reveal a novel dimeric quaternary association involving significant hydrophobic interactions between the N-terminal alpha-helices of two molecules of ammodytoxin related by crystallographic symmetry. Interactions at the dimer interface include important contributions from Met7, which is unique to ammodytoxin. Equilibrium sedimentation experiments are consistent with the crystallographic model. Competition experiments using SPR technology show complete inhibition of AtxA binding to FXa by calmodulin (CaM). The crystal structure shows that the C-terminal region, important for binding to FXa and CaM, is fully exposed and accessible for interaction with proteic receptors in both the monomeric and dimeric forms of ammodytoxin described here.

Crystal structure of a defective folding protein

Maltose‐binding protein (MBP or MalE) of Escherichia coli is the periplasmic receptor of the maltose transport system. MalE31, a defective folding mutant of MalE carrying sequence changes Gly 32→Asp and Ile 33→Pro, is either degraded or forms inclusion bodies following its export to the periplasmic compartment. We have shown previously that overexpression of FkpA, a heat‐shock periplasmic peptidyl‐prolyl isomerase with chaperone activity, suppresses MalE31 misfolding. Here, we have exploited this property to characterize the maltose transport activity of MalE31 in whole cells. MalE31 displays defective transport behavior, even though it retains maltose‐binding activity comparable with that of the wild‐type protein. Because the mutated residues are in a region on the surface of MalE not identified previously as important for maltose transport, we have solved the crystal structure of MalE31 in the maltose‐bound state in order to characterize the effects of these changes. The structure was determined by molecular replacement methods and refined to 1.85 Å resolution. The conformation of MalE31 closely resembles that of wild‐type MalE, with very small displacements of the mutated residues located in the loop connecting the first α‐helix to the first β‐strand. The structural and functional characterization provides experimental evidence that MalE31 can attain a wild‐type folded conformation, and suggest that the mutated sites are probably involved in the interactions with the membrane components of the maltose transport system.

Crystal structure of a recombinant form of the maltodextrin-binding protein carrying an inserted sequence of a B-cell epitope from the preS2 region of hepatitis B virus

We report the crystal structure of MalE‐B133, a recombinant form of the maltodextrin‐binding protein (MBP) of Escherichia coli carrying an inserted amino‐acid sequence of a B‐cell epitope from the preS2 region of the hepatitis B virus (HBV). The structure was determined by molecular replacement methods and refined to 2.7 Å resolution. MalE‐B133 is an insertion/deletion mutant of MBP in which residues from positions 134 to 142, an external α helix in the wild‐type structure, are replaced by a foreign peptide segment of 19 amino acids. The inserted residues correspond to the preS2 sequence from positions 132 to 145 and five flanking residues that arise from the creation of restriction sites. The conformation of the recombinant protein, excluding the inserted segment, closely resembles that of wild‐type MBP in the closed maltose‐bound form. MalE‐B133 was shown by previous studies to display certain immunogenic and antigenic properties of the hepatitis B surface antigen (HBsAg), which contains the preS2 region. The crystal structure reveals the conformation of the first nine epitope residues (preS2 positions 132 to 140) exposed on the surface of the molecule. The remaining five epitope residues (preS2 positions 141 to 145) are not visible in electron density maps. The path of the polypeptide chain in the visible portion of the insert differs from that of the deleted segment in the structure of wild‐type MBP, displaying a helical conformation at positions 134 to 140 (preS2 sequence numbering). A tripeptide (Asp‐Pro‐Arg) at the N terminus of the helix forms a stable structural motif that may be implicated in the cross‐reactivity of anti‐HBsAg antibodies with the hybrid protein. Proteins 27:1–8

Structural Mimicry of O-Antigen by a Peptide Revealed in a Complex with an Antibody Raised against Shigella flexneri Serotype 2a

The use of carbohydrate-mimicking peptides to induce immune responses against surface polysaccharides of pathogenic bacteria offers a novel approach to vaccine development. Factors governing antigenic and immunogenic mimicry, however, are complex and poorly understood. We have addressed this question using the anti-lipopolysaccharide monoclonal antibody F22-4, which was raised against Shigella flexneri serotype 2a and shown to protect against homologous infection in a mouse model. In a previous crystallographic study, we described F22-4 in complex with two synthetic fragments of the O-antigen, the serotype-specific saccharide moiety of lipopolysaccharide. Here, we present a crystallographic and NMR study of the interaction of F22-4 with a dodecapeptide selected by phage display using the monoclonal antibody. Like the synthetic decasaccharide, the peptide binds to F22-4 with micromolar affinity. Although the peptide and decasaccharide use very similar regions of the antigen-binding site, indicating good antigenic mimicry, immunogenic mimicry by the peptide was not observed. The F22-4-antigen interaction is significantly more hydrophobic with the peptide than with oligosaccharides; nonetheless, all hydrogen bonds formed between the peptide and F22-4 have equivalents in the oligosaccharide complex. Two bridging water molecules are also in common, adding to partial structural mimicry. Whereas the bound peptide is entirely helical, its structure in solution, as shown by NMR, is helical in the central region only. Moreover, docking the NMR structure into the antigen-binding site shows that steric hindrance would occur, revealing poor complementarity between the major solution conformation and the antibody that could contribute to the absence of immunogenic mimicry.

Low Levels of Polymorphisms and No Evidence for Diversifying Selection on the Plasmodium knowlesi Apical Membrane Antigen 1 Gene

Infection with Plasmodium knowlesi, a zoonotic primate malaria, is a growing human health problem in Southeast Asia. P. knowlesi is being used in malaria vaccine studies, and a number of proteins are being considered as candidate malaria vaccine antigens, including the Apical Membrane Antigen 1 (AMA1). In order to determine genetic diversity of the ama1 gene and to identify epitopes of AMA1 under strongest immune selection, the ama1 gene of 52 P. knowlesi isolates derived from human infections was sequenced. Sequence analysis of isolates from two geographically isolated regions in Sarawak showed that polymorphism in the protein is low compared to that of AMA1 of the major human malaria parasites, P. falciparum and P. vivax. Although the number of haplotypes was 27, the frequency of mutations at the majority of the polymorphic positions was low, and only six positions had a variance frequency higher than 10%. Only two positions had more than one alternative amino acid. Interestingly, three of the high-frequency polymorphic sites correspond to invariant sites in PfAMA1 or PvAMA1. Statistically significant differences in the quantity of three of the six high frequency mutations were observed between the two regions. These analyses suggest that the pkama1 gene is not under balancing selection, as observed for pfama1 and pvama1, and that the PkAMA1 protein is not a primary target for protective humoral immune responses in their reservoir macaque hosts, unlike PfAMA1 and PvAMA1 in humans. The low level of polymorphism justifies the development of a single allele PkAMA1-based vaccine.

Crystal structure of Plasmodium knowlesi apical membrane antigen 1 and its complex with an invasion-inhibitory monoclonal antibody

The malaria parasite Plasmodium knowlesi, previously associated only with infection of macaques, is now known to infect humans as well and has become a significant public health problem in Southeast Asia. This species should therefore be targeted in vaccine and therapeutic strategies against human malaria. Apical Membrane Antigen 1 (AMA1), which plays a role in Plasmodium merozoite invasion of the erythrocyte, is currently being pursued in human vaccine trials against P. falciparum. Recent vaccine trials in macaques using the P. knowlesi orthologue PkAMA1 have shown that it protects against infection by this parasite species and thus should be developed for human vaccination as well. Here, we present the crystal structure of Domains 1 and 2 of the PkAMA1 ectodomain, and of its complex with the invasion-inhibitory monoclonal antibody R31C2. The Domain 2 (D2) loop, which is displaced upon binding the Rhoptry Neck Protein 2 (RON2) receptor, makes significant contacts with the antibody. R31C2 inhibits binding of the Rhoptry Neck Protein 2 (RON2) receptor by steric blocking of the hydrophobic groove and by preventing the displacement of the D2 loop which is essential for exposing the complete binding site on AMA1. R31C2 recognizes a non-polymorphic epitope and should thus be cross-strain reactive. PkAMA1 is much less polymorphic than the P. falciparum and P. vivax orthologues. Unlike these two latter species, there are no polymorphic sites close to the RON2-binding site of PkAMA1, suggesting that P. knowlesi has not developed a mechanism of immune escape from the host’s humoral response to AMA1.