Molecular modeling of the amino-terminal domain of the mGluR1 glutamate metabotropic receptor by the threading method

Abstract

The glutamatergic system plays a key role in the functioning of the human central nervous system; therefore, knowledge of its structure at the molecular level is important not only for the understanding of its operation but also for the target-directed design of new drugs, in particular, new neuroprotective drugs based on selective agonists and antagonists of glutamate receptors. Two main approaches are currently used in the comparative modeling of protein structures. The first and the most widely used homology method is based on comparison of the amino-acid sequences of the protein being modeled with those of a protein whose spatial structure is known from experiments [1]. The main limitation of this approach is the requirement that at least 30% of the amino-acid sequences in the target and template proteins be homologous, which is usually fulfilled only in the series of evolutionally and functionally related proteins [2]. The use of this approach allowed us to model the closed forms of the glutamate [3] and glycine [4] sites of the ionotropic NMDA receptor; the ionic channel of the NMDA receptor with phencyclidine binding sites [5]; and the open form of the aminoterminal domain (ATD) of the first subtype of the metabotropic mGluR glutamate receptor [6]. In the last-mentioned case, we were able to demonstrate on the basis of multiple alignment that R78 is involved in binding of the distal carboxy group of agonists. Models of the open and closed forms of ATD of the mGluR1 receptor made by comparative analysis were also considered in [7, 8]. The second approach, called threading, is based on the idea that even unrelated proteins with non-homologous amino acid sequences can have identical spatial structures [9‐10]. In terms of this approach, a “thread” corresponding to the sequence of the target protein is pulled through a fold determined by the template protein and the position of the thread is found to maximize the function of the generalized empirical potential. The main advantage of this approach over the above-considered one is that it is often possible to model rather reliably the spatial structures of proteins even in the absence of any homology of the primary sequence to the template protein. The purpose of this work is to construct the spatial models for the open and closed forms of the amino acid domain of the mGluR1 receptor by the threading method and to estimate the potential of this method for the cases of low homology of amino acid sequences. The search performed using the Threader2 program [9], which contains more than 2000 different folds in the database, made it possible to identify suitable template proteins for the modeling of open and closed forms of these receptors. The results were confirmed by statistic tests, which are also implemented in the Threader2 program. As in the case of homology modeling, the template proteins are LBP (leucine-binding proteins) for the open form and PEA (amide-binding proteins) for the closed form. The homology of the amino acid sequences is about 17% in both cases. The probability that the spatial structure of the target protein is similar to that of the template protein is estimated quantitatively by a statistic value called the Z factor. The table presents the calculated Z factors for all sub