L-Glutamic acid decarboxylase (GAD) exists as both membraneassociated and soluble forms in the mammalian brain. Here, we propose that there is a functional and structural coupling between the synthesis of ␥-aminobutyric acid (GABA) by membraneassociated GAD and its packaging into synaptic vesicles (SVs) by vesicular GABA transporter (VGAT). This notion is supported by the following observations. First, newly synthesized [ 3 H]GABA from [ 3 H]L-glutamate by membrane-associated GAD is taken up preferentially over preexisting GABA by using immunoaffinity-purified GABAergic SVs. Second, the activity of SV-associated GAD and VGAT seems to be coupled because inhibition of GAD also decreases VGAT activity. Third, VGAT and SV-associated Ca 2؉ ͞ calmodulin-dependent kinase II have been found to form a protein complex with GAD. A model is also proposed to link the neuronal stimulation to enhanced synthesis and packaging of GABA into SVs.T he rate-limiting enzyme L-glutamic acid decarboxylase (GAD, EC 4.1.1.15) is involved in the synthesis of ␥-aminobutyric acid (GABA), a major inhibitory neurotransmitter in the mammalian brain. There are two well-characterized GAD isoforms in the human brain, namely GAD 65 and GAD 67 (referring to GAD with a molecular mass of 65 kDa and 67 kDa, respectively) (1). Both GAD 65 and GAD 67 are present as homodimers or heterodimers in soluble GAD (SGAD) and membrane-associated GAD (MGAD) pools (2-4). The ratio of GAD 65 to GAD 67 is higher in synaptic vesicle (SV) fractions than in the cytosol (5). Some studies suggest that GAD 65 binds to the membranes (6, 7) and that GAD 67 subsequently interacts with MGAD 65 (2, 6). However, the nature of anchorage of GAD to membranes and its physiological significance is still not well understood. GAD is not considered to be an integral membrane protein because it lacks a stretch of hydrophobic amino acids long enough to span the membrane. Subpopulations of GAD 65 and GAD 67 remain firmly anchored to membranes despite various ionic extraction methods (2,4,8). The interaction of GAD with membranes was reported to be through ionic (9-11), hydrophobic (12, 13), protein phosphorylation (14), or proteinprotein interaction (15). Previously, we reported that MGAD is activated by phosphorylation that requires an electrochemical gradient across the SV membrane (7). A model for the anchoring mechanism of GAD to SV and its role as a link between GABA synthesis and storage in nerve terminals was also proposed (15). The evidence presented here will demonstrate that GABA synthesized by SV-associated GAD is preferentially transported into the SV by vesicular GABA transporters (VGATs). We have also demonstrated that VGAT, a 10-transmembrane helix protein (16), forms a protein complex with GAD on the SV and could be involved in the anchorage of MGAD to the SV. The formation of this GAD protein complex ensures an efficient coupling between GABA synthesis and packaging into the SV. Materials and MethodsPreparation of SVs. SVs were purified from whole rat brain (Sprague-Dawle...

Previously, we reported that protein phosphorylation plays an important role in regulating soluble l-glutamic acid decarboxylase (GAD) [Bao, J. (1995) J. Biol. Chem. 270, 6464-6467] and membrane-associated GAD activity [Hsu, C. C. (1999) J. Biol. Chem. 274, 24366-24371]. Here, we report the effect of phosphorylation on the two well-defined GAD isoforms, namely, GAD65 and GAD67, using highly purified preparations of recombinant human brain GAD65 and GAD67. GAD65 was activated by phosphorylation, while GAD67 was inhibited by phosphorylation. The effect of phosphorylation on GAD65 and GAD67 could be reversed by treatment with protein phosphatases. We further demonstrate that protein kinase A (PKA) and protein kinase C isoform epsilon are the protein kinases responsible for phosphorylation and regulation of GAD67 and GAD65, respectively. Direct phosphorylation of GAD65 and GAD67 was demonstrated by incorporation of [(32)P] from [gamma-(32)P]ATP into purified GAD65 and GAD67 and immunoblotting assay using anti-phosphoserine/threonine antibodies. We have identified one specific phosphorylation site, threonine 91 (T91), in hGAD67 that can be phosphorylated by PKA using MALDI-TOF. Site-directed mutation of T91 to alanine abolished PKA-mediated phosphorylation and inhibition of GAD activity. Furthermore, mutation of T91 to aspartic acid or glutamic acid mimics the effect of phosphorylation. A model depicting the effect of phosphorylation on GAD activity upon neuronal stimulation is also proposed.