Identification of the substrate-binding exosites of two snake venom serine proteinases: molecular basis for the partition of two essential functions of thrombin.

The venom of the South American snake Bothrops jararaca contains two serine proteinases, bothrombin and the platelet-aggregating enzyme PA-BJ, which share 66% sequence identity. Each of these proteinases possesses one of the two essential procoagulant functions of thrombin-the clotting of fibrinogen and platelet aggregation. Thus, bothrombin clots fibrinogen but has no direct effect on platelets, unless in the presence of exogenous fibrinogen. PA-BJ induces platelet aggregation by interacting with the protease-activated platelet receptor without clotting fibrinogen. On the other hand, thrombin possesses two extended surfaces. One is composed of basic and hydrophobic residues (exosite I) and the other one of basic residues only (exosite II). These exosites are involved in the recognition of physiological macromolecular substrates. In order to identify the corresponding exosites in bothrombin and PA-BJ and understand the molecular basis of the partition of the two procoagulant functions of thrombin among the two snake venom enzymes, we used molecular modeling to obtain models of their complexes with their natural substrates fibrinogen and a fragment of the protease-activated platelet receptor, respectively. In analogy to thrombin, each of the enzymes presents two exosites. Nonetheless, the exosites contain a smaller proportion of basic residues than thrombin does (45-72%), reducing thus the functional diversity of the enzymes. In addition, the composition of exosite I is different in both enzymes. We identify those residues in exosite I that could contribute to the differences in specificity. Finally, allostery does not seem to mediate macromolecular substrate recognition by these enzymes.

Toxicity evolution of Vipera aspis aspis venom: identification and molecular modeling of a novel phospholipase A(2) heterodimer neurotoxin.

We report the simultaneous presence of two phospholipase A(2) (PLA(2)) neurotoxins in the venom of Vipera aspis aspis, the first such observation. One is monomeric and identical to ammodytoxin B of Vipera ammodytes ammodytes. Its presence may result from gene flux after interbreeding between V. aspis aspis and V. ammodytes ammodytes. The second, a novel heterodimer named vaspin, is very similar to vipoxin of Vipera ammodytes meridionalis and to PLA(2)-I of Vipera aspis zinnikeri. It may result from expression of preexisting genes, the acidic subunit evolving from an ancestor common to ammodytin I2 from V. ammodytes ammodytes, which we also found in V. aspis aspis.

Identification and molecular structural prediction analysis of a toxicity determinant in the Bacillus sphaericus crystal larvicidal toxin.

The operon containing the genes encoding the subunits of the binary crystal toxin of Bacillus sphaericus strain LP1-G, BinA and BinB (41.9 kDa and 51.4 kDa, respectively), was cloned and sequenced. Purified crystals were not toxic to Culex pipiens larvae. Comparison of the amino-acid sequences of this strain (Bin4) with those of the three other known toxin types (Bin1, Bin2 and Bin3) revealed mutations at six positions, including a serine at position 93 of BinA4, whereas all other types of BinA toxin from B. sphaericus had a leucine at this position. Reciprocal site-directed mutagenesis was performed to replace this serine in BinA4 from LP1-G with a leucine and the leucine in the BinA2 protein from strain 1593 with a serine. Native and mutated genes were cloned and overexpressed. Inclusion bodies were tested on C. pipiens larvae. Unlike the native Bin4 toxin, the mutated protein was toxic, and the reciprocal mutation in Bin2 led to a significant loss of toxicity. In vitro receptor-binding studies showed similar binding behaviour for native and mutated toxins. In the absence of any experimental data on the 3D structure of these proteins, sequence analysis and secondary-structure predictions were performed. Amino acid 93 of the BinA polypeptide probably belongs to an alpha helix that is sensitive to amino-acid modifications. Position 93 may be a key element in the formation of the BinA-BinB complex responsible for the toxicity and stability of B. sphaericus Bin toxins.

Molecular basis for the partition of the essential functions of thrombin among snake venom serine proteinases: the case of thrombin-like enzymes.

Thrombin is a mammalian serine proteinase that plays a prominent role in the maintenance and regulation of hemostasis through its interaction with various substrates and/or ligands. The venoms of several snakes contain glycosylated serine proteinases that have been recognized to possess one or more of the essential activities of thrombin on fibrinogen (Fg) and/or platelets. These proteinases share about 60% sequence identity. One class of snake venom serine proteinases are those known as thrombin-like (TLE), named after their ability to directly clot Fg in order to preferentially produce fibrinopeptide A, fibrinopeptide B or both. To understand the molecular basis of this phenomenon, the corresponding amino acid sequences and molecular structures need to be analyzed. Given the absence of experimentally determined tertiary structures of snake venom, TLEs, three-dimensional molecular models should prove useful in this context. Towards this goal, we obtained models of snake venom TLEs that used TSV-PA as template, TSV-PA being the only snake venom serine proteinase whose crystal structure is known to date. Along with a comparative sequence analysis the models contribute to the identification and description of thrombin-homologous or alternative binding sites, helping thus to understand differences in specificity.

Combining phage display and molecular modeling to map the epitope of a neutralizing antitoxin antibody.

Crotoxin is a potent presynaptic neurotoxin from the venom of the rattlesnake Crotalus durissus terrificus. It is composed of the noncovalent and synergistic association of a weakly toxic phospholipase A2, CB, and a nontoxic three-chain subunit, CA, which increases the lethal potency of CB. The A-56.36 mAb is able to dissociate the crotoxin complex by binding to the CA subunit, thereby neutralizing its toxicity. Because A-56.36 and CB show sequence homology and both compete for binding to CA, we postulated that A-56.36 and CB had overlapping binding sites on CA. By screening random phage-displayed libraries with the mAb, phagotopes bearing the (D/S)GY(A/G) or AAXI consensus motifs were selected. They all bound A-56.36 in ELISA and competed with CA for mAb binding, although with different reactivities. When mice were immunized with the selected clones, polyclonal sera reacting with CA were induced. Interestingly, the raised antibodies retained the crotoxin-dissociating effect of A-56.36, suggesting that the selected peptides may be used to produce neutralizing antibodies. By combining these data with the molecular modeling of CA, it appeared that the functional epitope of A-56.36 on CA was conformational, one subregion being discontinuous and corresponding to the first family of peptides, the other subregion being continuous and composed of amino acids of the second family. Phage-displayed peptides corresponding to fragments of the two identified regions on CA reacted with A-56.36 and with CB. Our data support the hypothesis that A-56.36 and CB interact with common regions of CA, and highlight residues which are likely to be critical for CA-CB complex formation.

Molecular modeling of an active loop structure in lysozyme. Sequence effects or crystal packing?

The Val99-Gly 104 variable region in egg white lysozyme is part of the active site cleft and of the epitope recognized by some monoclonal antibodies. In general, this loop is found in a conformation inflected towards the active site (proximal conformational) such as in free hen lysozyme (HEL). But in a lysozyme such as Japanese quail's (JEL), the loop turns away from the active site cleft (distal conformation). In order to differentiate sequence effects from crystal packing, we generated and refined loop conformations for the 99-104 variable region in lysozyme, then estimated their relative conformational free energies. Some of the results indicate that (i) the flexibility of the 99-104 segment is much greater for HEL than for JEL sequences when unconstrained by the crystal lattice, (ii) for JEL, only distal structures are favored, while for HEL the states span the zone between proximal and distal regions, and (iii) epitopes elucidated from crystal structures may not always be conserved in solution. For the JEL loop, model building shows that an energy-costly distal to proximal transition appears necessary. Finally, analysis of available structural data indicates that changes of humidity, temperature and pressure on loop conformation are negligible.

Structural basis of the gp120 superantigen-binding site on human immunoglobulins.

B cell superantigens (SAg) interact with normal human nonimmune Igs (Igs), independently of the light chain isotype, and activate a large proportion of the B cell repertoire. Recently, the major envelope protein of HIV-1, gp120, was found to exhibit SAg-like properties for B cells with potential pathologic consequences for the infected host. This unconventional mode of interaction contrasts with its binding to immunization-induced Abs, which requires the tertiary structure of the heavy and light chain variable regions. In this report, we have examined the structural basis of the interaction between human Igs and gp120. We found that gp120 binding is restricted to Igs from the V(H)3 gene family and that the two V(H) genes 3-23 and 3-30, known to be overutilized during all stages of B cell development, frequently impart gp120 binding. We also provide evidence that the viral gp120 SAg can interact with only a subset of the human V(H)3+ Igs that can convey binding to the prototypic bacterial B cell SAg protein A from Staphylococcus aureus. Finally, we have identified amino acid positions present primarily in the first and third framework regions of the Ig heavy chain variable region, outside the conventional hypervariable loops, which correlate with gp120 binding. In a three-dimensional sequence-homology model, these residues partially overlap with the predicted SAg protein A binding site for V(H)3+ Igs.

Trimeresurus stejnegeri snake venom plasminogen activator. Site-directed mutagenesis and molecular modeling.

The specific plasminogen activator from Trimeresurus stejnegeri venom (TSV-PA) is a serine proteinase presenting 23% sequence identity with the proteinase domain of tissue type plasminogen activator, and 63% with batroxobin, a fibrinogen clotting enzyme from Bothrops atrox venom that does not activate plasminogen. TSV-PA contains six disulfide bonds and has been successfully overexpressed in Escherichia coli (Zhang, Y., Wisner, A., Xiong, Y. L., and Bon, C. (1995) J. Biol. Chem. 270, 10246-10255). To identify the functional domains of TSV-PA, we focused on three short peptide fragments of TSV-PA showing important sequence differences with batroxobin and other venom serine proteinases. Molecular modeling shows that these sequences are located in surface loop regions, one of which is next to the catalytic site. When these sequences were replaced in TSV-PA by the equivalent batroxobin residues none generated either fibrinogen-clotting or direct fibrinogenolytic activity. Two of the replacements had little effect in general and are not critical to the specificity of TSV-PA for plasminogen. Nevertheless, the third replacement, produced by the conversion of the sequence DDE 96a-98 to NVI, significantly increased the Km for some tripeptide chromogenic substrates and resulted in undetectable plasminogen activation, indicating the key role that the sequence plays in substrate recognition by the enzyme.

Induction of DNA bending by bifunctional intercalating agents of the 7H-pyridocarbazole family.

The structures and binding energetics of selected complexes formed between the deoxynucleotides d(CpGpGpCpG).d(CpGpCpCpG), d(CpGpApTpCpG)2, d(GpCpGpCpCpG).d(CpGpGpCpGpC), and d(CpGpCpCpCpG)2 with the DNA bifunctional intercalating agent ditercalinium and three of its rigid linking chain derivatives have been investigated theoretically by means of a molecular mechanics approach that takes into account nucleic acid flexibility, ligand flexibility and solvent dielectric effects (R. Lavery, in: Unusual DNA structures, eds S. Harvey and R. Wells (Pergamon, New York, 1988) p. 189; R. Lavery, in: DNA bending and curvature, eds W.K. Olson et al. (Adenine Press, New York, 1988) p. 191). The piperidinium chains of the bis-intercalating ligands are always located in the major groove of DNA. For the energy-minimized complexes the ligand proceeds to bind following preferentially the 5'-pyrimidine-purine-3' alternating sequence, thus dictating the number of internal exclusion sites. The complexes with three exclusion sites will present (i) a bending of the structure towards the major groove, and (ii) a non-ideal distribution of unwinding angles; complexes with less than three exclusion sites will remain essentially linear. The absence of a bend does not preclude other types of local deformations of the base-pairs such as inclination, buckle and tip. The proposed structures of the d(CpGpApTpCpG)2 complexes are in agreement with NMR structural results. The possible relevance of these findings to a previously proposed mode of interaction for ditercalinium-like DNA ligands is discussed.

Influence of L-cystinyl side-chain configurations on the melting of crosslinked alpha-tropomyosin dimers.

The experimental melting profile reported by Holtzer, M.E., Holtzer, A. and Skolnick, J. (Macromolecules 16 (1983) 173-180) for the rabbit alpha-tropomyosin dimer crosslinked at cysteine residue 190 has been analyzed using matrix methods. The configuration partition function employed includes a term arising from interactions at the crosslink site. This term, denoted by omega, is found to be smaller than 1, implying that events at the crosslink site resist helix formation by dimer. A theoretical analysis of the conformational restrictions imposed on the crosslink provides a satisfactory estimate of omega at high temperatures. Agreement deteriorates at lower temperatures, perhaps as a consequence of difficulty in establishing a reliable value for omega from analysis of the low-temperature circular dichroism data.