Dr. R. Sowdhamini - Specific systems
Modelling transforming growth factors (TGF) and their interaction with receptors
Sequence analysis and structure prediction of a protein in bioluminescence superfamily
Comparitive modelling and analysis of active site region of a non-functional serine protease domain
Most systems in signal transduction are composed of multi-domain components and involve protein-protein interactions and significant conformational changes. proteins involved in cellular growth and differentiation acts as effectors both at short-range and at long-range suggesting multiple pathways of interaction. Computational tools on structure-function prediction can play an important role in deciphering these mechanisms.
Particular emphasis will be paid to the study of specific proteins involved in cellular signaling where a combined use of computational analysis of sequence- structure-function relationships and knowledge of biochemical and biological properties can provide initial framework for the understanding of the structural basis of celluar signaling.Homologous signal transduction domains,though closely related,are mediated by high ligand specificity and accurate molecular recognition, often operating by multiple pathways. The proposed computational approach involves prediction and comparative modelling (Sali & Blundell, 1993) of the structures of specific proteins involved in signaling (if the experimental structure is not available),analysis of the experimental structures which are functionally and/or structurally related to the protein of interest and using these analyses to model the structures in different forms such as active and inactive. This approach can aid the recognition and characterisation of the function or biological roles of a protein. Under this category we would study specific proteins,in close interaction with molecular biologists,biochemists and developmental biologists who input exprimental results, to address the problem of function prediction and structural basis of regulation. These will form useful inputs in the development of methods on fold and function prediction.
To this end,collaboration with experimental groups have been initiated in the following areas :
- Structure prediction and modelling of transmembrane domain of inositol-triphosphate and ryanodine receptors
Previous sequence analyses on inositol-triphosphate receptors (Insp3R) and ryanodine receptors (ryR), undertaken in this laboratory, have led to a higher focus on the C-terminal, transmembrane (TM) domain for the following reasons:
(a) This domain is the most highly conserved part between the two families
(b) This domain acts as a calcium channel in the tetrameric form and
(c) due to the strong structural and functional resemblance to the potassium channels.
We have built a three-dimensional model of the C-terminal part of the TM domain corresponding to the last two predicted TM helices. Additionally, the tetramer co-ordinates were generated from the basis potassium channel structure, but moved away from the pore axis to suit the higher pore radius of calcium channels as compared to the potassium channels. The resultant tetrameric model shows a broad patch of negatively charged residues at the mouth of the pore.
Fig: Model of the tetrameric arrangement of the transmembrane domain of inositol-triphosphate receptor
TGF superfamily consists of secreted signalling proteins which mediate key diverse events in growth and development. TGFs in Drosophila are required for developmental events such as dorsoventral axis formation in early embryos and correct patterning of adult appendages. They can be grouped into subfamilies, such as TGF1-5, activin and 60A. Several representative TGFs have been modelled using the homology modelling program, MODELLER. Their interaction with the type I and type II receptors have been predicted with a view to understanding the basis of specificity. Graphics viewing packages like RASMOL and GRASP have enabled the comparison of residue distributions at the binding sites. For example, while one of the protruding ends of TGF-b2 is predominantly positive, that of bmp2 is acidic and that of dpp is hydrophobic, complimentary patches are observed in one of the 'variable fingers' of the receptors: type II-TGFR is negative, bmpR has a positive patch and punt (binds to dpp) is hydrophobic. Such residue complimentarities and the presence of solvent-exposed hydrophobics in the TGFs have given rise to convincing models of TGF-TGFR interactions.
Fig:Conserved and class-specific residues for TGF dimer and the TGF type-II receptor
Small molecules like pheromones act as sensory regulatory units for a variety of processes like bioluminescence in bacteria. By means of a lux-plasmid based bioluminescent sensor for a pheromone, OHHL, proteins that regulate the production of OHHL were identified. OHHL regulates different biological processes such as biolumeniscence and antibiotic production in a cell-density dependent manner. These can be grouped under the broad superfamily of N-acylhomoserine lactone synthases or the autoinducer synthases (AS superfamily). The recent crystal structure determination of EsaI, a key enzyme in this pathway [Watson, W.T., Minogue, T.D., Val, D.L., von Bodman, S.B. and Churchill, M.E.A. [2002] Mol.Cell, 9:685-694], shows that the NAHL synthase superfamily members adopt the fold of N-acetyltransferases (NAT). We suggest, by the identification of intermediate sequences, that AS and NAT superfamilies are evolutionarily related. Evolutionary trace analyses of aligned sequences, comparative modelling and docking studies have been used to discuss functionally important residues (FIRs) of EsaI homologues.
Fig: Models of members of bioluminescence superfamily - EagI model compared with LasI model
masquerade(mas) encodes a 1047-amino acid preproprotein that is processed by proteolytic cleavage to generate two polypeptides. The carboxy-terminal polypeptide is highly similar to the serine protease fold but lacks its proteolytic activity. Comparison of the C-terminal sequence with classical serine proteases shows that the catalytic Ser in the mas protein has been substituted by Gly. Since the C-terminal polypeptide of mas accumulates at somatic muscle attachment sites, it is proposed that this protein might act as a novel adhesion molecule and stabilize muscle and cell-matrix attachment (Murugasu-Oei et. al., 1995). On comparison with other serine proteases we find the mas sequence to be closely related to the mammalian serine proteases, in particular, the plasma kallikreins . We will be modeling the mas serine protease-like domain and examining the specificity pocket which should provide clues as to whether the mas mutant phenotype is likely to act as an antagonist of serine proteases.
- Fold prediction and comparative modeling of Bdm1: a probable alpha/beta hydrolase associated with hot water epilepsy.
Hot water epilepsy (HWE) is a benign and rare form of reflex epilepsy that occurs most commonly in humans. Bdm1 is one of the proteins whose mRNA transcript is overexpressed during HWE in a rat model. We show, by sequence analysis and fold recognition methods, that Bdm1 has strong structural similarities to alpha/beta hydrolases like the thioesterases. A three-dimensional model derived by comparative modeling methods allowed the search for catalytic residues using a flexible functional template characteristic of these enzymes. We predict that Bdm1 might be regulated by homocysteine levels by means of direct participation in degradation pathways.
Fig:Three-dimensional model of Bdm1 obtained by threading the sequence on the fold of BPHD