Newly Synthesized Phosphatidylinositol Phosphates Are Required for Synaptic Norepinephrine but Not Glutamate or γ-Aminobutyric Acid (GABA) Release*

Newly synthesized phosphatidylinositol phosphates have been implicated in many membrane-trafficking reactions. They are essential for exocytosis of norepinephrine in PC12 cells and chromaffin cells, suggesting a function in membrane fusion. We have now studied the role of phosphatidylinositol phosphates in synaptic vesicle exocytosis using synaptosomes. Under conditions where phosphorylation of phosphatidylinositols is blocked, norepinephrine secretion was nearly abolished whereas glutamate and GABA release was still elicited. Thus phosphatidylinositides are essential only for some membrane fusion reactions, and exocytotic release mechanisms differ between neurotransmitters.

Newly synthesized phosphatidylinositol phosphates have been implicated in many membrane-trafficking reactions. They are essential for exocytosis of norepinephrine in PC12 cells and chromaffin cells, suggesting a function in membrane fusion. We have now studied the role of phosphatidylinositol phosphates in synaptic vesicle exocytosis using synaptosomes. Under conditions where phosphorylation of phosphatidylinositols is blocked, norepinephrine secretion was nearly abolished whereas glutamate and GABA release was still elicited. Thus phosphatidylinositides are essential only for some membrane fusion reactions, and exocytotic release mechanisms differ between neurotransmitters.
At synapses, similar mechanisms of synaptic vesicle exocytosis are thought to effect the secretion of different neurotransmitters (e.g. glutamate, GABA, 1 and norepinephrine; reviewed in Refs. [1][2][3][4]. The same vesicle proteins are present in presynaptic nerve terminals independent of neurotransmitter type (e.g. see . In norepinephrine-secreting chromaffin cells and in PC12 cells, phosphorylation of phosphatidylinositols by PI 4-kinase is required for Ca 2ϩ -triggered exocytosis (9 -12). In these cells, PI 4-kinase can be potently inhibited by phenylarsine oxide (PAO), resulting in a block of exocytosis (11,12). Several proteins involved in synaptic vesicle exocytosis bind to the products of PI 4-kinase activity, PIP and PIP 2 (13)(14)(15)(16)(17). Furthermore, phosphoinositides have also been implicated in a number of other membrane trafficking reactions (reviewed in Refs. 18 -20). Together these findings suggested that phosphorylation of PI is essential for exocytosis and membrane fusion. To explore the role of PI phosphorylation in neurotransmitter release, we have now studied neurotransmitter release from synaptosomes. We demonstrate that inhibition of PIP and PIP 2 synthesis results in dramatically different effects for different neurotransmitters and that PIP and PIP 2 synthesis is not a requirement for most synaptic exocytosis.

EXPERIMENTAL PROCEDURES
Treatments and Phospholipid Analysis of Synaptosomes-Synaptosomes were prepared as described (21) and resuspended in aerated (95% O 2 , 5% CO 2 ) ice-cold Krebs-bicarbonate buffer, pH 7.4 (composition in mM: NaCl 118, KCl 3.5, CaCl 2 1.25, MgSO 4 1.2, KH 2 PO 4 1.2, NaHCO 3 25, glucose 11.5, and HEPES-NaOH 5) or phosphate-free Krebs-bicarbonate buffer (for 32 P-labeling experiments). Synaptosomes were 32 P-labeled for 1.5 h at 35°C in a 95% O 2 , 5% CO 2 atmosphere with [ 32 P]orthophosphate (1 mCi/ml). After labeling, PAO (from Aldrich) freshly made in Me 2 SO or Me 2 SO alone (Ͻ1% of total volume) was added, and synaptosomes were incubated for the indicated times in triplicate. Reactions were stopped on ice. Lipids were extracted with 3.75 volumes of chloroform:methanol:concentrated HCl (100:200:1). After 10 min on ice, 10 g of phosphatidylethanolamine (Sigma) was added as a carrier, and phase partitioning was induced by with 1.25 volumes of chloroform and of 0.1 N HCl. The chloroform phase with phospholipids was washed twice with cold methanol:0.1 N HCl (1:1). Equal amounts of the extracts were loaded on TLC plates with phospholipid standards (10 -20 g each). TLC plates were developed in 1-propanol:H 2 O:concentrated NH 4 OH (65:20:15) and analyzed by autoradiography; phospholipid standards were identified with iodine vapors. 32 P incorporation was quantified with a PhosphorImager ( 3 H-Loaded synaptosomes (0.1 ml) were trapped on glass fiber filters (GF/B, Whatman), overlaid with 50 l of a 50% Sephadex G-25 slurry, and superfused at 33°C with Krebs-bicarbonate buffer (flow rate, 0.8 ml/min) under continuous aeration with 95% O 2 , 5% CO 2 . For sucrosetriggered neurotransmitter release experiments, synaptosomes were superfused with Ca 2ϩ -free Krebs-bicarbonate buffer containing 0.1 mM EGTA. After 12 min of washing, two 1-min fractions were collected to determine base-line release. We then evoked release from synaptosomes by the following agents: 1) 25 mM KCl for 30 s; 2) 5 M ionomycin for 30 s; 3) 0.5-3 nM ␣-latrotoxin for 1 min; 4) 0.5 M sucrose for 30 s in Ca 2ϩ -free Krebs-bicarbonate buffer. All stimuli were applied by rapid switching of the superfusion lines between regular and stimulation buffers. The amounts of [ 3 H]glutamate, [ 3 H]norepinephrine, or [ 3 H]GABA released into the superfusate and remaining in the synaptosomes at the end of the experiment were quantified by liquid scintillation counting. Fractional neurotransmitter release was calculated by dividing the amount of neurotransmitter released during a time interval by the amount of transmitter remaining in the synaptosomes at that time. To obtain the total GABA, glutamate, and norepinephrine release induced by a given stimulus, the evoked release above baseline was integrated over the time of the experiment. To test the effect of PAO on release, synaptosomes were treated with PAO in Me 2 SO, Me 2 SO alone (control; Ͻ1% of total volume), or vanadyl hydroperoxide (VOOH; prepared as a complex with 1,10-phenanthroline as in Ref. 22) for 20 min at 35°C in Ca 2ϩ -containing aerated (95% O 2 , 5% CO 2 ) Krebs-bicarbonate buffer. During the last 5 min, 3 H-labeled neurotransmitters were added for loading the synaptosomes, which were then used for release measurements as described above. PAO treatment partly inhibited [ 3 H]norepinephrine and [ 3 H]GABA but not [ 3 H]glutamate uptake. To control for this and other possible indirect effects, we performed experiments in which synaptosomes were first loaded with 3 H-labeled neurotransmitters and then treated with PAO, with identical results to those shown here. For each experiment, results are expressed in percent release compared with control conditions. * This study was supported by National Institute of Mental Health Grant RO1-50824. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

PAO Blocks PIP and PIP 2 Synthesis in Synaptosomes-
We labeled synaptosomes with 32 P i and treated them with PAO or control buffer. The synaptosomes were then incubated under various control and stimulation conditions, and their phospholipids were analyzed by TLC. Three major 32 P-labeled phospholipids were observed and identified with unlabeled phospholipid standards as PIP, PIP 2 , and closely co-migrating PI and phosphatidic acid (data not shown). The preferential labeling of PI, PIP, and PIP 2 in synaptosomes agrees well with the high turnover rate of PIPs (23)(24)(25). We then tested the effect of PAO on PIP and PIP 2 synthesis in synaptosomes. PAO caused a major inhibition of 32 P-labeling of PIP and PIP 2 but induced only marginal changes in PI and phosphatidic acid ( Fig. 1 and data not shown). Stimulation of synaptosomes by ␣-latrotoxin or KCl depolarization elicited moderate decreases in the levels of 32 P-labeled PIP and PIP 2 in the absence of PAO but had no effect on the inhibition of PI and PIP phosphorylation by PAO (data not shown). We did not determine at which positions the phosphatidylinositol rings are phosphorylated in these experiments. However, similar studies in chromaffin cells showed that the majority of PIP and PIP 2 is phosphorylated at the 4and 5-positions (11,12). These results argue that, as in chromaf-fin cells, PAO inhibits phosphorylation of PIP and PIP 2 at the 4-and 5-positions of the inositol ring in synaptosomes (11,12).
To optimize the PAO treatment conditions, we measured the time and dose dependence of the effect of PAO (Fig. 1). Synaptosomes were incubated with different concentrations of PAO for various time periods. 32 P-Labeled PIP and PIP 2 were quantified using TLC and PhosphorImager detection. Inhibition of PI and PIP phosphorylation by PAO reached a maximum after 20 min and was almost complete at PAO concentrations of more than 3 M PAO (Fig. 1).
Ca 2ϩ -dependent and Ca 2ϩ -independent Release of Norepinephrine, Glutamate, and GABA from Synaptosomes-To study release, we loaded synaptosomes with 3 H-labeled glutamate, GABA, and norepinephrine, placed them in superfusion chambers, and monitored neurotransmitter secretion under continuous superfusion. Release was stimulated by four procedures that act on different components of the release machinery: moderate KCl depolarization (K) to open voltage-gated Ca 2ϩ channels (26), ionomycin (I), a Ca 2ϩ ionophore, hypertonic sucrose that induces exocytosis of docked vesicles from the readily releasable pool (27), and ␣-latrotoxin (L) that triggers release by an unknown mechanism. We found that KCl and ionomycin induced secretion of glutamate, GABA, and norepinephrine from synaptosomes in a strictly Ca 2ϩ -dependent manner (Fig. 2). KCl-evoked release was transient whereas ionomycin caused a prolonged effect, probably because as a Ca 2ϩ ionophore, ionomycin induces a long lasting increase in intracellular Ca 2ϩ that results in the continuous recruitment of vesicles for exocytosis. ␣-Latrotoxin triggered neurotransmitter release by a Ca 2ϩ -independent high affinity interaction (EC 50 Ϸ 3 nM). Its action was inhibited by low concentrations of La 3ϩ , indicating a specific mechanism (Fig. 2). Sucrose caused massive transient neurotransmitter release (data not shown). Preliminary studies showed that sucrose is more potent in releasing neurotransmitters from synaptosomes than KCl depolarization but that both secretagogues act on the same neurotransmitter pools. 2 Sucrose-triggered release similar to electrophysiological studies is partially inhibited by tetanus toxin. 2 Together these experiments suggest that the synaptosomes can be used to probe secretion of different types of neurotransmitters elicited by stimuli acting on distinct parts of the release machinery.
Effect of Blocking PIP and PIP 2 Synthesis on Neurotransmitter Release-We treated synaptosomes with 3 M PAO or control buffer and measured norepinephrine and glutamate release stimulated by KCl, ionomycin, and ␣-latrotoxin (Fig. 3A). After PAO treatment, norepinephrine secretion was blocked whereas glutamate release was unchanged. This was a surprising result because it suggests that glutamate and norepinephrine release may be mechanistically different, although both are Ca 2ϩ -dependent (Fig. 2). To exclude the possibility that the distinct effects of PAO were caused by differences between experimental conditions, we measured norepinephrine and glutamate release in the same preparation of PAO-treated and control synaptosomes (Fig. 3B). The PAO concentrations used were high enough to assure a virtually total inhibition of PI and 2 G. Lonart and T. C. Sü dhof, unpublished observation. PIP phosphorylation (Fig. 1). Again, norepinephrine release was inhibited whereas glutamate release was not. Thus PAO has opposite effects on the exocytosis of norepinephrine-and glutamate-containing vesicles.
The possibility that norepinephrine release but not glutamate release is inhibited under conditions that block PI and PIP phosphorylation is interesting because it implies that PI and PIP phosphorylation is not universally required for exocytosis. In addition, if the components of the release machinery differ between types of neurotransmitters, their exocytotic mechanisms must be distinct. To affirm these conclusions, we performed a large-scale study of the effects of PAO on synaptic vesicle exocytosis. Three transmitters, norepinephrine, glutamate, and GABA, were analyzed in parallel under PAO treatment conditions that block PI and PIP phosphorylation. Because PAO is also a tyrosine phosphatase inhibitor (11,12), we treated the synaptosomes with a strong tyrosine phosphatase inhibitor, vanadyl hydroperoxide (VOOH), as a further control (18). We conducted multiple independent experiments, normalized their results, and expressed the integrated release as percent of control (Figs. 4 and 5). Overall, the data confirm that PAO selectively inhibits exocytosis of norepinephrine containing synaptic vesicles independent of which secretory agent is used, without a consistent inhibitory effect on GABA or glutamate release.
When release was induced by KCl depolarization, PAO at low concentrations (3 M) severely depressed norepinephrine secretion but had no effect on glutamate release even at 10-fold higher levels (30 M PAO; Fig. 4). PAO also partially inhibited GABA release stimulated by KCl. Norepinephrine release triggered by ionomycin, the second Ca 2ϩ -dependent and probably most powerful secretagogue used here (Fig. 2), was also inhibited by PAO. Again, glutamate secretion was unaffected. With ionomycin, GABA release was also unchanged by PAO (Fig. 4). In addition, norepinephrine secretion triggered by ␣-latrotoxin or by sucrose was severely inhibited by 3 M PAO (Fig. 5). With both stimulation agents, we again observed no effect of PAO at concentrations of up to 30 M on glutamate secretion. Furthermore, PAO exerted no significant inhibition of GABA release stimulated by either hypertonic sucrose or ␣-latrotoxin.
In the design of our experiments, PAO treatments preceded the uptake of labeled neurotransmitters for the release measurements. To exclude the possibility that PAO changes the disposition of neurotransmitters after uptake into the synaptosomes, we performed release measurements in which the order of PAO treatment and neurotransmitter uptake was reversed. With this protocol, PAO still did not inhibit glutamate release but strongly suppressed norepinephrine secretion (data not shown). It is also unlikely that "glutamatergic" and "GABAergic" nerve terminals are selectively protected from the inhibitory effects of PAO. These two transmitters account for more than 90% of all synapses in brain cortex. Thus the changes in PIP and PIP 2 observed by TLC with PAO treatment must affect glutamatergic and GABAergic synaptosomes. DISCUSSION In the current experiments, we use synaptosomes to study the release of three neurotransmitters, norepinephrine, glutamate, and GABA. We applied PAO to inhibit PIP and PIP 2 synthesis and studied the effect of this inhibition on release. Our data show that norepinephrine secretion elicited with four stimulation protocols is severely inhibited by PAO. In contrast, PAO had no effect on glutamate secretion stimulated by all four stimulation protocols. PAO did not inhibit GABA release stimulated by ionomycin, sucrose, and ␣-latrotoxin, and moderately inhibited GABA release evoked by KCl. We used PAO concentrations that cause a nearly complete inhibition of PI and PIP phosphorylation. Thus glutamate and GABA secretion do not require newly synthesized PIP and PIP 2 . In contrast, similar to PC12 and chromaffin cells (9 -12), PI and PIP phosphorylation is essential for norepinephrine secretion. The fact that we employed different means of stimulating release ensures that the effects observed are not artifacts of a particular stimulation method.
These results have implications for our understanding of neurotransmitter release and membrane fusion. First, the data show that there are transmitter-specific differences in the mechanism of exocytosis. No such differences have been observed in the protein composition of synapses (e.g. see Refs. [5][6][7][8]. The selective requirement for phosphoinositides in norepinephrine but not glutamate and GABA release may reflect fundamental differences between exocytosis of dense-core synaptic vesicles containing catecholamines and clear synaptic vesicles containing glutamate and GABA. Second, exocytosis of norepinephrine-containing vesicles requires PIP phosphorylation at a step that precedes Ca 2ϩ action. We found that PAO equally inhibits Ca 2ϩ -dependent release induced by KCl and ionomycin and Ca 2ϩ -independent release evoked by hypertonic sucrose, which triggers exocytosis of all docked synaptic vesicles by an unknown mechanism prior to Ca 2ϩ entry (27).
Third, phosphorylation of PI and PIP is not universally involved in exocytosis. Although PAO could have multiple actions, our data document that it acts as an effective inhibitor of PI and PIP phosphorylation in synaptosomes. The undiminished capacity of synaptosomes to secrete glutamate and GABA in the absence of significant PI and PIP phosphorylation is striking. It thus is unlikely that PIP and PIP 2 function as general signals or lipidic mediators in membrane fusion.