fJeuron,

Vol. 6, 517-524,

April,

1991, Copyright

0 1991 by Cell Press

Differential Release of Amino Acids, Neuropeptides, and Catecholamines from Isolated Nerve Terminals Matthijs Verhage,*+ Harvey T. McMahon,* Wim E. J. M. Ghijsen,* Frans. Boomsma,§ Greet Scholten,* Victor M. WiegantJ and David G. Nicholls* “Department of Experimental Zoology iJniversity of Amsterdam Amsterdam !*letherlands ‘:Department of Biochemistry Jniversity of Dundee Dundee Scotland +Department of Internal Medicine ‘irasmus University Rotterdam ‘Netherlands Rudolf Magnus Institute for Pharmacology Jniversity of Utrecht iJtrecht Yetherlands

Summary We have investigated transmitter release from small and large dense-core vesicles in nerve terminals isolated from guinea pig hippocampus. Small vesicles are found in clusters near the active zone, and large dense-core vesicles are located at ectopic sites. The abilities of Ca2+ channel activation and uniform elevation of Ca2+ concentration (with ionophores) to evoke secretion of representative amino acids, catecholamines, and neuropeptides were compared. For a given increase in Ca2+ concentration, ionophore was less effective than Ca2+ channel activation in releasing amino acids, but not in releasing cholecystokinin-8. Titration of the average Ca2+ concentration showed that the Ca2+ affinity for cholecystokinin-8 secretion was higher than that for amino acids. Catecholamine release showed intermediate behavior. It is concluded that neuropeptide release is triggered by small elevations in the Caz+ concentration in the bulk cytoplasm, whereas secretion of amino acids requires higher elevations, as produced in the vicinity of Ca2+ channels. tntroduction it has been established that the release of fast-acting neurotransmitters occurs by the exocytosis, probably within 1 ms, of small synaptic vesicles clustered in the proximity of active zones, whereas neuropeptides are released by the fusion of large dense-core vesicles at ectopic sites outside these active zones (Zhu et al., 7986; Buma, 1989). At the squid giant synapse it has ’ Present address: Rudolf Magnus lar Biology, University of Utrecht, tdetherlands.

Institute/Institute Padualaan 8,3584

for MolecuCH Utrecht,

been shown that high, localized concentrations of free Ca2+ may be generated in close proximityto active zone Ca2+ channels (see Zucker and Stockbridge, 1983; Simon and Llinas, 1985; Zucker and Fogelson, 1986; Smith and Augustine, 1988), and it has been proposed that these may be required to trigger fast-acting transmitter exocytosis. From these experiments, a picture has emerged of a localized coupling between Ca2+entrythrough channels located in theactivezone and fast-acting transmitter release. This situation is in contrast with exocytosis of neuropeptides via a more delocalized, bulk-phase event. The small (> [Ca2+]laVe.In contrast, since mobilecarrier ionophores insert randomly into the membrane and typically transport about 1 ion per ms (Pressman, 1976), the Ca*+ ionophore ionomycin will induce a uniform elevation in [Ca*+] within the terminal, such that [Ca2+]local = [Ca2+lave = [Ca2+lo/t0. lonomycin transports Ca2+ across a variety of lipid bilayers, without the requirement of specific constituents of biological membranes (Blau and Weismann, 1988). It

of the active

zone.

The apparent

exclusion

of large dense-core

vesicles

can thus be expected that its action is uniform in the heterogeneous population of synaptosomes. It follows that, for a given fura 2 signal, a neurotransmitter released by [Ca2+]local in the micro-environment of a Ca2+ channel will be much more effectively released by Ca*+ entry through those channels (evoked by KCI or Qaminopyridine) than by ionomycin. In contrast, a neurotransmitter released by [Ca2+lnit0 would demonstrate no such preference.

[Caz+].Ve and Amino Acids,

the Evoked Release of Neurotransmitter Noradrenaline, and CCK-8

Synaptosomes were exposed to 30 mM KCI, 2 mM Qaminopyridine (which induces spontaneous repetitive firing of synaptosomes; Tibbs et al., 1989), or 1 uM ionomycin, in order to produce comparable increases in Ka*+L, as indicated by fura 2 fluorescence (Figure 3). In addition, a higher [Ca2+laVewas attained by using 5 PM ionomycin. In parallel with the fura 2 measurements, the Ca*‘-dependent release of glutamate, CABA, noradrenaline, and CCK-8 was determined (Figure 4). A 3 min exposure to either 2 mM Caminopyridine or 30 mM KCI, known to evoke optimal release of each of the transmitters (Woodward et al., 1988;

D fferential 519

Release

of Neurotransmitters

2

3

4

RANK

2. The Occurrenceof Small and Large Dense-Core Synapticvesiclesand the Length of the Active Zone in Successive Ultra Thin Sections of Synaptosomes Figure

The section exhibiting the most small clear-core vesicles was asscribed the rank order zero, and adjacent sections were numbered consecutively. Serial sections of synaptosomeswere taken into account only when 4 or more successive sections could be assessed and when both types of vesicles were found in at least 1 section (n = 24). Columns represent the average number of vesicles per section or length of the active zone. Statistical analys s of differences is described in the text.

McMahon and Nicholls, 1991; Verhage et al., 1989, 1991a, 1991b), was equally effective in releasing each transmitter(Figure4, left-hand panels).Thetwoagents were again equally effective when the release during the first 15 s was determined (which encompasses the most rapid phase of amino acid or catecholamine release; Figure 4, right-hand panels). In contrast, no Ca2+-dependent release of GABA or glutamate was evoked by 1 uM ionomycin, even though this produced the same fura 2 signal as 30 mM KCI or 2 mM 4-aminopyridine (Figure 3). lonomycin (1 PM) did not

significantly deplete the synaptosomal ATP/ADP ratio below the control value of 5.1 rt 0.5 (n = 3). To induce a release of glutamate comparable in extent to that seen with 2 mM 4aminopyridine or 30 mM KCI, 5 uM ionomycin was required. This concentration of ionomycin caused a saturation of the fura 2 signal, indicating that [Ca2+lave rapidly rises above 5 uM (Figure 3). However, it also caused some plasma membrane depolarization (data not shown), thus firing of voltage-activated Ca2+ channels cannot be eliminated. Additionally, a component of the release with 5 PM ionomycin could have a cytoplasmic origin as a result of a slow depolarization-induced reversal of the electrogenic acidic amino acid carrier in the plasma membrane (Nicholls, 1989;Verhage et al., 1989). This possibility cannot be allowed for by a Cati-independent control (see Experimental Procedures), since the highionomycin-induced depolarization is itself Ca2+ dependent. From these provisos, it is likely that the exocytotic release of glutamate evoked by the uniform entry of Ca2+ catalyzed by 5 PM ionomycin is substantially less than the total release, further increasing the discrepancy between ionomycin and plasma membrane depolarization. In contrast to the amino acids, the neuropeptide CCK-8 was effectively released by a uniform elevation in [Ca2i],yt0 to 400 nM. Indeed, the effect of 1 PM ionomycin was even greater than that of 30 mM KCI or 2 mM Caminopyridine, whereas 5 uM ionophore induced only a limited additional effect (Figure 4). Noradrenaline release showed an intermediate preference for depolarization-versus ionophore-evoked Ca2+ entry (Figure 4). Dopamine release resembled that of noradrenaline, but was considerably slower (data not shown). Dopamine release was very small in guinea pig hippocampus (maximal Ca2+-dependent release was 224 f 69 fmol per mg of protein). Several reports suggest that catecholamines occur with a dual distribution in both small and large dense-core vesicles (Klein et al., 1982; Zhu et al., 1986; Lundberg et al., 1990) and in synapses both with and without active

5bM ION + 1.5 Ca j

1.5uMION+1.5Ca 2 4.AP + 1.5 Ca

Figure

3. [CaZ+],“.

in Fura

2-Loaded

Synaptosomes

following

Depolarization

or lonomycin

Treatment

Synaptosomes were incubated as described in Experimental Procedures. Additions, as indicated on the traces, were made at the arrows: K, KCI; ION, ionomycin;4-AP,4-aminopyridine; Ca, CaCb. Concentrationsgiven are in millimolarunless indicatedotherwise(ionomycin). Fura 2 traces are the means of 4 preparations. SEM values do not exceed 10% of the mean.

Neuron 520

5-

GLU

-A E

10 mM KCI

F 2mM 4-AP

t

1

Figure5. Average Membrane Potential Population Estimated with Cyanine Dye

e-

7

PI ihR

>NM

[CaliM3

>%M

in the

Synaptosomal

Synaptosomes were preincubated in incubation medium with the addition of 0.1 uM DiS-G(5). The arrow indicates additions of KCI, 4aminopyridine, or ionomycin. The right vertical axis gives an approximate estimation of the membrane potential according to the reduced Nernst potential, E,(mV) = 58log([K+]JO.l M), as calculated by K+ titration. The trace after Caminopyridine addition is corrected for an absorption blank. Traces are means of 3 preparations; SEM values do not exceed 15% of the mean.

NA

5irzi? [Cal m

>$M2do

PaI ihR

>9M

Release of Glutamate, GABA, Figure4. Net Ca 2+-Dependent CCK-8, and Noradrenaline As a Function of [Ca2’],,, following Depolarization or lonomycin Treatment Release was evoked under conditions parallel to those shown in Figure 3 and is expressed per mg of synaptosomal protein. The first point of each graph in each panel represents the [Ca?‘],,, of synaptosomes incubated in the absence of added Caz+ and the absence of releasing agent (the Ca2+-dependent transmitter release by definition being zero). The second point of each graph represents synaptosomes in the presence of 1.5 mM Ca*+ but no releasing agent; note that some release of CCK-8 occurs under these conditions (8). (A)-(D) show total Ca*+-dependent release in the 180 s following addition of Ca2+ with or without releasing agent. (E)-(H) show the early phase of Ca*+-dependent release in the first 15 s after additions. Data points represent means f SEM of 4-6 independent experiments.

zones (Smith and Augustine, data may reflect such a dual

1988). release

Therefore mechanism.

these

The Role of the Membrane Potential In Exocytosis As 1 PM ionomycin caused little depolarization (Figure 5), it is possible that the ineffectiveness of the ionophore in releasing amino acids is due to an inherent role of plasma membrane depolarization in exocy-tosis in addition to its ability to activate Ca2+ channels (Parnas and Parnas, 1988; Hochner et al., 1989). Since activation of the putative exocytosis-coupled

Ca*+ channel itself requires plasma membrane depolarization, this option is less easy to eliminate. In any case, an inherent role of the membrane potential could apply only to the amino acid transmitters, since both CCK-8 and noradrenaline are released in the absence of significant depolarization (with 1 PM ionomycin; Figures 4 and 5). The ineffectiveness of ionomycin in releasing glutamate is retained in synaptosomes pre-depolarized by 30 mM KCI in the presence of 20 mM Mg2’ to inhibit Ca*+ entry through voltage-activated Caz+ channels. At 1.4 uM, ionomycin (giving the same fura 2 signal as 1 uM ionophore in normal Mg2’ medium) released less than 10% of the total releasable glutamate pool from the pre-depolarized synaptosomes in 30 s, compared with25% releasefrom polarized synaptosomes in normal Mg*+ medium (difference significant at p < 0.001). Discussion Is the Synaptosome a Valid Model for Investigating Ca*+ Secretion Coupling of Different Transmitter Classes? Isolated nerve terminals (synaptosomes) retain a high degree of integrity, as judged by a low [Ca2+lave, a high membrane potential, tight bioenergetic coupling, and the ability to release amino acid neurotransmitters (DeBellerocheand Bradford, 1977; Nicholls, 1989;Verhage et al., 1989), catecholamines (Drapeau and Blaustein, 1983; Woodward et al., 1988), and neuropeptides (Floor, 1983; Terrian et al., 1988; Verhage et al., 1991b) by Ca*+dependent exocytosis. Glutamate is the dominant neurotransmitter released by this prepara-

IXfferential ! ,21

Release

of Neurotransmitters

tion, and compartmental analysis reveals its vesicular origin (Wilkinson and Nicholls, 1989), whereas most, if not all, CCK-8 coexists with amino acid transmitters, particularly GABA, in the hippocampus (Henry et al., -1984; Gall, 1984; Kosaka et al., 1985; Sloviter and Nilaver, 1987; Lopes da Silva et al., 1990). As shown in Figures 1 and 2, the present preparation of synaptosomes also retains the morphological characteristics ofthe intactterminal, includingtheclusteringof small but not large dense-core vesicles in the vicinity of ,rctive zones. Although the preparative procedure guarantees maximal purity of synaptosomes (Verhage et al., 1988), rhe fraction is still contaminated to a limited extent with non-synaptosomal structures. If a proportion of t:he fura 2 signal were to originate from these contaminating organelles, and if these were capable of main:aining a low [Ca2+]cyt0 but lacked voltage-activated r *+ channels, ,,a ionomycin, unlike depolarization, $:ould increase the Ca*+ saturation of the probe in i:hese organelles and artifactually enhance the change n fura 2 signal originating from the synaptosomes themselves. This in turn would result in a greater indicated [Ca2+lave being associated with a given release of !:ransmitter. However, this explanation would simi-

RESTING TERMINAL

larly bias the neuropeptide release, especially since this may arise from the same terminals as the amino acid release (see above). The time resolution with which [Ca2+lave and neurotransmitter release can be monitored is of necessity slow in relation to the phasic response during physiological stimulation. However, detailed kinetics analyses of the release of glutamate (McMahon and Nicholls, 1991; Verhage et al., 1991a), GABA (Turner and Goldin, 1989), and dopamine (Drapeau and Blaustein, 1983) have shown that a substantial proportion of the total transmitter is released in the first 2 s of KCI depolarization. Our 15 s time point, chosen as the minimum interval at which CCK-8 release could be accurately determined (Verhage et al., 1991b), largely reflects this initial release. Furthermore, at least in the case of the amino acids (Turner and Goldin, 1989; McMahon and Nicholls, 1991), neurotransmitters are released by channels that do not undergo voltage inactivation on to the plateau

prolonged of elevated

Do the Results Support

depolarization [Ca2+lave shown

the Working

Three alternative explanations tivenessof depolarization-and

and give rise in Figure 3.

Hypothesis?

for the differing ionophore-evoked

effecele-

STIMULATED TEFUvllNALS

KCI

Figure

6. Proposed

Mechanism

for the

Differential

Release

of Glutamate,

CCM,

and

Noradrenaline

Drawings represent nerve terminals containing two types of synaptic vesicles. Under resting conditions ([Ca2’],,, = 140 nM), no release occurs (upper left scheme). When the terminal is depolarized by high [K’] (upper right scheme), voltage-activated Ca*+ channels in the active zone open. [CaZ+] in the immediate proximity of the inner aspect of the channel is high before Cal+ is able to diffuse radially into the bulk cytoplasm and produce a [CaZ’],, = 400 nM. The Ca*+-sensitive trigger for amino acid exocytosis experiences the high [Ca2+] in the micro-environment of the mouth of the open channel, which is necessary for amino acid transmitter release (see graph). Stimulation with low ionomycin (bottom right scheme) produces a similar [Ca*+laM (400 nM), but through random penetration of nerve terminal membrane. As in the case of K+ depolarization this elevation of [CaZ’],, is sufficient to activate the neuropeptide release mechanism, which is sensitive to relatively small Ca*+ elevations (see graph), but does not produce the high elevations in [Ca*+]I,,r necessary to evoke release of amino acid transmitters. The graph (bottom left) summarizes ionomycin data from Figure 4, showing the differences in Caz+ sensitivity of the release mechanisms for the different transmitter classes.

NellrOn

522

vations in [Ca2+lave can be eliminated. Membrane potential does not play an inherent role in amino acid release, contaminating non-synaptosomal organelles cannot account for the difference between the amino acid and CCK-8 release, and 1 PM ionomycin does not deplete synaptosomal ATP/ADP ratios and terminate the ATP production required for exocytosis (Kauppinen et al., 1988). Furthermore, the heterogeneity of the synaptosomal preparation cannot explain the present results for two reasons: First, the representative peptide (CCK-8) does not occur in a distinct population of terminals in the preparation, but originates (largely) from the same terminals as the amino acid transmitters studied (see above). Second, the effect of ionomycin is presumably uniform among the different types of terminals in the preparation (see Hypothesis and Experimental Approach) and still evokes the differential release of transmitters (see Figure 6). Therefore we conclude that the present results are consistent with the hypothesis that amino acids are released by [Ca2+]rocal in the micro-environment of voltage-activated Ca2+ channels, whereas neuropeptides are released by bulk [Ca2+]cV0. A nerve terminal with a diameter of