Introduction: The enzyme malate dehydrogenase, which is present in various subcellar locations such as cytoplasm (cMDH), mitochondria (mMDH), glyoxysomes (gMDH), peroxisomes and chloroplasts, catalyses the oxidation of malate into oxaloacetate using NAD+/NADP+ as cofactor. The primary structure of soybean gMDH has been determined from sequencing data of severa! eDNA, clones (isolated from a library constructed with mRNA from 3-day-old cotyledons). In order to predict the effect of mutations on the activity or specificity of this enzyme, we built up its 3D structure starting from data obtained with crystallized isozymes.
Results and Discussion: From the Brookhaven Protein Database, it appears that only the pig cytoplasmic MDH (cMDH) structure has been solved with an acceptable resolution. Despite the low identity (26%) between pig cMDH and soybean gMDH, we used the former as a template for the 3D modelling of soybean gMDH. Using the BioPolymer module of the SYBYL Software together with the Amber force field and Pullman partial atomic charges, a 3D madel of gMDH is proposed. Soybean gMDH was first aligned onto water melon gMDH, water melon mMDH and pig mMDH, which was possible without introducing gaps. Pig cMDH (starting point of the 3D modelling) was aligned onto the former enzymes using the conserved amino acids with pig mMDH as anchor points, while the rest of the sequence was aligned manually and with the help of computer programs to maximise the number of similar amino acids and to minimize the number of gaps. After this alignment, the 3D structure of pig cMDH was progressively changed to soybean gMDH by modifications of the side chains, by elimination or insertion of amino acids, and by preparing Ioops. Local geometry optimizations were conducted for most of these changes. Final optimizations were conducted in the absence or presence of NAD+ and/or substrate (i.e., malate), allowing comparison of the active site with respect to those of related enzymes (pig cMDH, dogfish LDH). The conservation of shape and functionality of the active site was quite impressive and can be used to propose a slight revision of amino acids involved in malate binding as suggested previously. Our computer model of soybean gMDH identified three amino acids which are directly involved in the binding of malate: Arg87, Arg159 and Ser229. In pig cMDH the corresponding Arg91, Arg 161 and Ser240 bind malate together with a fourth amino acid (Ser 241). This Ser residue is not conserved in other MDHs. The lack of this H-bond would not significantly affect the binding of malate, as Ser241 only adds a fourth H-bond to the malate carbonyl group which is already tightly bound to Arg161 (Arg 159 in gMDH).