Comparative molecular model building of two serine proteinases from cytotoxic T lymphocytes

Abstract

Two genes that are expressed when precursor cytotoxic T lymphocytes are transformed to T killer cells have been cloned and sequenced. The derived amino acid sequences, coding for the cytotoxic cell protease 1 (CCP1) and Hannuka factor (HF) are highly homologous to members of the serine proteinase family. Comparative molecular model building using the known three‐dimensional structures and the derived amino acid sequences of the lymphocyte enzymes has been provided useful information, especially in predicting the conformations of the substrate binding sites. In applying this modelling procedure, we used the X‐ray structures of four serine proteinase to provide a structurally based sequences alignment: α‐chymotrypsin (CHT), bovine trypsin (BT), Streptomyces griseus trypsin (SGT), and rat must cell protease 2 (RMCP2). The root mean square differences in α‐carbon atom positions among these four structures when compared in a pairwise fashions range form 0.79 to 0.97 Å for structurally equivalent residues. Te sequences of the two lymphocyte enzymes were then aligned to these proteinase using chemical criteria and the superimposed X‐ray structures as guides. The alignment showed that the sequence of CCP1 was most similar to RMCP2, whereas HF has regions of homology with both RMCP2 and BT. RmCP2 and BT as templates for HF, the molecular models were constructed. Intermolecular steric clashes that resulted from replacement of amino acid chains of the templates by the aligned residues of CCP1 and HF were relieved by adjustment of the side chain conformational angels in an interactive computer graphics device. This process was followed by energy minimization of the enzyme model to optimize the stereochemical geometry and to relieve any remaining unacceptably close nonbonded contacts. The resulting model of CCP1 has an arginie residue at position 226 in the specificity pocket, thereby predicting a substrate preference for P₁ aspartate or glutamate residues. The model also predicts favorable binding for a small hydrophobic residue at the P₂ position of the substrate. The primary specificity pocket of HF resembles that of BT and therefore predicts a lysine or arginine preference for the P₁ residue. The arginine at position 99 in the model of HF suggests a preference for aspartate or glutamate side chains in the P₂ positions of the substrate. Both CCP1 and HF have a free cysteine in the segment of the polypeptide 88 to 93. Models of the dimeric form of these enzymes can be constructed by forming disulfide bridges between the suitably oriented monomers, ie., Cys88–Cys88′ for CCP1 and Cys93–Cys93′ for HF.