MECHANISMS OF INACTIVATION OF GAMMA-AMINOBUTYRIC-ACID AMINOTRANSFERASE BY THE ANTIEPILEPSY DRUG GAMMA-VINYL GABA (VIGABATRIN)

The molecular mechanism of inactivation of pig brain y-aminobutyric acid (GABA) aminotransferase by the anticonvulsant drug vigabatrin (1, gamma-vinyl GABA) has been investigated. Inactivation of GABA aminotransferase that had been reconstituted with [H-3]pyridoxal 5'-phosphate (PLP) produced two tritiated metabolites upon denaturation, pyridoxamine 5'-phosphate and a cofactor adduct in the ratio of about 2:1. Inactivation with 16-C-14]-gamma-vinyl GABA led to the incorporation of 1.02 equiv of radioactivity under nondenaturing conditions and 0.60-0.76 equiv, depending upon the pH, under denaturating conditions. This indicates the formation of two different adducts, one that is stable to denaturation and one that is not. The identity of the structure of the stable adduct was determined with the use of chemical model studies on hypothesized adducts and the application of this chemistry to the [C-14]-labeled enzyme. Base treatment and sodium borohydride reduction followed by sodium periodate oxidation indicated that the structure of the stable adduct was 6-X-4-oxohexanoic acid (10, Scheme 1). The adduct that was released upon denaturation (18, Scheme IV) was identified as the product of an enamine rearrangement onto the PLP. Therefore, two distinct mechanistic pathways are implicated during inactivation. Azaallylic isomerization of the Schiff base of gamma-vinyl GABA with PLP (that is, the normal catalytic mechanism) leads to a reactive intermediate (6, Scheme 1), which is attacked by an active-site nucleophile to give adduct 10 after Schiff base hydrolysis. Alternatively, hydrolysis of 6 to 6a followed by active-site nucleophilic attack (prior to release of 6a from the active site) also could give 10. This adduct is stable to denaturation and accounts for 70-75% of the total active site labeling. The other pathway, allylic isomerization (Scheme II), leads to reactive intermediate 12; however, transimination by the active-site lysine residue (Scheme IV), which, apparently, is faster than Michael addition to 12, leads to enamine formation. The enamine reacts with the lysine-bound PLP to give adduct 17. This adduct, which accounts for 25-30% of the total inactivation adducts, is unstable to denaturation and is released as 18. Since no product of hydrolysis of the enamine could be detected, it appears that when the enamine is generated, it does not escape the active site. Although 1.8 transamination events appear to be occurring for every inactivation event, the product of transamination could not be detected.