Neuron,

Vol. 8, 1069-1077,

June, 1992, Copyright

0 1992 by Cell Press

Transmitter Release Increases Intracellular Calcium in Perisynaptic Schwann Cells In Situ Babak S. and Milton National Department Medical University Toronto, Canada,

Jahromi, Richard Robitaille, P. Charlton Centers of Excellence of Physiology Sciences Building of Toronto Ontario M5S IA8

Summary Glial cells isolated from the nervous system are sensitive to neurotransmitters and may therefore be involved in synaptic transmission. The sensitivity of individual perisynaptic Schwann cells to activity of a single synapse was investigated, in situ, at the frog neuromuscular junction by monitoring changes in intracellular Ca*+ in the Schwann cells. Motor nerve stimulation induced an increase in intracellular Ca*+ in these Schwann cells; this increase was greatly reduced when transmitter release was blocked. Furthermore, local application of the cotransmitters acetylcholine and ATP evoked Ca*+ responses even in the absence of extracellular Ca*+. Successive trains of nerve stimuli or applications of transmitters resulted in progressively smaller Ca*+ reponses. We conclude that transmitter released during synaptic activity can evoke release of intracellular Ca*+ in perisynaptic Schwann cells. This Ca*+ signal may play a role in the maintenance or modulation of a synapse. These data show that synaptic transmission involves three cellular components with both postsynaptic and glial components responding to transmitter secretion. introduction In the brain, glial cells embrace synapses and are thus well positioned to have a regulatory role in synaptic transmission. For instance, glial cells can respond to many transmitter chemicals in vitro (Orkand, 1982; Pearce et al., 1985; Hamprecht, 1986; Lauder and McCarthy, 1986; Enkvist et al., 1989; Cornell-Bell et al., 1990; Ahmed et al., 1990; Cornell-Bell and Finkbeiner, 1991; Dave et al., 1991) and can be depolarized by K+ accumulation near active neurons in situ (Orkand et al., 1966; Villegas, 1981; Marrero et al., 1989). However, owing to complicated connections and morphology, the physiological relationships between individual glial cells and nerve terminals are difficult to study in the brain. The counterparts of glial cells in the peripheral nervous system are Schwann cells. Skeletal neuromuscular junctions (NMJs) are covered by a few large, nonmyelinating Schwann cells (NMSC; Figure 1; Heuser and Reese, 1977) and thus provide a simple system in which to investigate signaling between single nerve terminals and individual glial cells. At the frog NMJ,

the NMSCs cover the entire thread-like nerve terminal and send fine processes around it (Dreyer and Peper, 1974; Heuser and Reese, 1977). The Schwann processes are usually located between the active zones at irregular intervals (I-4 pm), within 0.5 pm of an active zone and often as close as a few tens of nanometers (Dreyer and Peper, 1974; Heuser and Reese, 1977; Figure IB). The NMSCs are therefore well positioned to detect transmitter release and are strategically located to have a modulatory role in nerve terminal activity. This close association between synapses and glial cells is a typical feature present at vertebrate NMJs and at synapses in the central nervous system (Heuser and Reese, 1977; Orkand, 1982). In the present report we tested whether perisynaptic NMSCs at the frog NMJ can detect synaptic activity generated by a single nerve terminal. Changes in intracellular Ca2+ were monitored as an indicator of NMSC activity during synaptic transmission. We report here that the releaseof neurotransmittersevoked by stimulation of the motor nerve triggers an increase in intracellular Ca2+ and that this increase is mimicked by local application of the cotransmitters acetylcholine (ACh) and ATP. Results Intracellular Ca2+ in NMSCs at the frog NMJ was monitored using confocal microscopy with the permeant fluorescent Ca2+ indicator fluo 3-AM (Kao et al., 1989; Tsien, 1989), and synaptic transmission was induced by electrical stimulation of the motor nerve (40 Hz, 30 s). Muscle contractions and all electrophysiological signs of postsynaptic responses were eliminated by blocking muscle ACh receptors with a-bungarotoxin (a-BuTx; data not shown). We have previously reported Ca*+ signals in the nerveterminal underlyingtheseSchwanncells.These recordings of nerve terminal Ca*+ were obtained under conditions in which nerve terminals retained flue-3, but the Schwann cells did not (see Figure 4 in Robitaille and Charlton, 1992). We have noticed that the nerve terminal Cal+ signals begin immediately upon nerve stimulation and remain elevated as long as stimuli occur and that successive trains of stimuli elicit similar Ca2+ responses. These characteristics are unlike those of the Ca*+ signals in Schwann cells. In the present study, both Schwann cells and nerve terminals retained flue-3 and could both produce a Ca*+ response. To reduce contribution of the nerve terminal Ca*+ signal to measurement of the Schwann cell signal, we measured Cati responses only at the ellipsoidal (7.0 f 0.3 x 21.4 f 0.8 pm) NMSC soma, which is much wider than the cylindrical nerve terminal (l-2 pm diameter; Figure IA). In addition, Ca2+ responses in Schwann cells lying directly over the nerve terminal were similar to responses in cells located

NeUrOll 1070

Figure 1. Perisynaptic the Frog NMJ

Schwann

Cells

at

(A) Nonmyelinating Schwann cells covering branching nerve terminals at frog NMJs in the cutaneus pectoris muscle. The bright ellipsoids are the cell bodies of the NMSCs (arrows). These ceils have been loaded with flue 3-AM. Bar, 50 pm. (B) A Schwann cell surrounding the nerve terminal at the frog NMj. This electron micrograph is a cross section of a frog NMJ from thecutaneous pectoris muscle, showing the nerve terminal (nt) with synaptic vesicles (sv) clustered around active zones (az), where neurotransmitter secretion occurs. The muscle fiber (mf) has a postjunctional fold, wherecholinergic receptors are clustered opposite each active zone. The NMSCs cover the nerve terminal and send processes (scf) between the nerve terminal and the muscle frber. Note the presence of the large nucleus (n) in the cell body of the Schwann cell. The cell body is, on average, 7 pm wide and 21 pm long. Bar, 1 urn.

lateral to the nerve terminal, where the nerve terminal fluorescence could not contribute to measurements of Schwann cell fluorescence (see below). Therefore, owing to their temporal and spatial separation, Schwann cell Ca*+ signals may be detected in isolation from nerve terminal signals. Nerve-Evoked Ca2+ Responses in Perisynaptic Schwann Cells Nerve stimulation (40 Hz, 30 s) triggered an increase in flue3 fluorescence (128% f 17.5%; 16 cells in 8 muscles) in NMSCs, indicating an increase in intracellular Caz+ (Figure 2A). The Ca2+ response began, on average, 1.7 k 0.6 s after the beginning of the stimulation and rose to a peak in 14.1 k 2.4 s. The duration of elevated Ca2+ during motor nerve stimulation was 15.5 + 2.0 s, and the Ca2+ response decayed to baseline

after the nerve stimulation in 96.3 f 15.0 s. Nerve stimulation at lower frequencies (IO or 20 Hz) resulted in smaller Ca*- responses than stimulation at 40 Hz (data not shown). In cell bodies arranged lateral to the nerve terminal (see Figure IA), Ca2’ signals were occasionally observed to originate at the terminal side and propagate as a wave into the rest of the cell body. The stimulus-induced Ca*’ signal was also seen in the elongated parts of the NMSC covering the nerve terminal. In contrast to the NMSC, no Ca2+ responses were detected in the myelinating Schwann cells present at the last myelinated segment of the motor axon during stimulation of the motor nerve. The nerve-evoked Ca*’ response in the perisynaptic Schwann cells was also characterized by a progressive rundown during successive stimulations (40 Hz; Figure 3A). The Ca* response evoked by a second train

Nerve-Evoked 1071

CaZ+ Signals

in Schwann

Cells

NORMAL

A7

cd - CgTX

Ca+ +

3

B4 c

c 210 iz a 140R i brl .w m + + 6

2; 180 c 5 120 -

70-

3G 60 -

01 40~ I 0

60

I

I

120 180 Time (s)

I 240

‘2 + + 6

2

O-

0

I

I

I

1

20

40

60

80

Time (s)

Figure 2. Nerve-Evoked Cal+ Responses in Schwann Cells flue-3. Pixel intensities arecoded so that higher fluorescence False color images of a frog NM] loaded with the fluorescent Ca >+ indicator are red and lower intensities are blue. (A) Ca*+ response in a Schwann cell before (Al), intensities, and hence higher Ca2+ activities, during (A2-A3), and after (A4-A5) sustained transmitter release induced by repetitive stimulation of the motor nerve (40 Hz, 30 s). (A6) Schematic diagram of the NM) shows the studied area delimited by a square. (A7) Changes in fluorescence (AFIF) before, during (bar), and after the nerve stimulation for the Schwann cell body presented in Figures Al-A5. The corresponding time at which each image was taken is illustrated on the graph with the corresponding number of the figures above. Bar (Al-A5), 10 urn. (6) Ca*+ response in 2 Schwann cells after blockade of neurotransmitter release by oCgTx. The images show the NM] before (61) and at peak intensity of (B2) the Cam response during nerve stimulation. (83) Schematic diagram of the NMJ with the studied area delimited by a square. (B4) Changes in fluorescence due to nerve stimulation (bar) in the same Schwann cells presented in (Bl) and (B2). The time at which images were taken is illustrated on the graph with the corresponding number of the figures above. The increase in fluorescence intensity was reduced when transmitter release was blocked by o-CgTx, but local application of ACh still induced a Ca*+ response (see Figure 4). Different NMJ from that in (A). Bar, 20 urn.

Neuron 1072

Tc-s u 60 j

\

-2 40 .c! -g 20 s? 0I

‘+\ I

I

E?

1

2

3

Train Number

B-G 5 100 s +

80

+ 3 60 .N 23 40 g 20 s 0 ACh Applications

Figure

3. Rundown

of W’

Responses

(A) Relative increase in Ca2+ fluorescence expressed as a percentevoked by the first train of stimuli to age of the Ca2+ response the motor nerve (6 ceils in 5 muscles). Note that the relative increase in fluorescence induced by a second and third consecutive train of stimulation is much reduced. (6) Relative increase of fluorescence expressed as a percentage of the first Cap response evoked by local application of ACh (5 cells in 3 muscles). A rundown similar to that in (A) was observed. Different experiments from that in (A). Error bar is SEM of the percent decrease.

of stimuli was only 32.4% f 7.4% the size of the first one, and a third train of stimuli evoked an average response of only 9.7% + 4.5% the size of the initial one (6 ceils in 5 muscles). Often no Ca*+ response could be evoked by a fourth train. Only the data obtained from the first train of stimuli are used in the present study. The rundown of the Ca*+ response is unlikely to be due to a reduction in the amount of transmitter secreted, since the amount of transmitter release varies little with successive trains of stimuli (Robitaille and Charlton, 1992). Does the Ca*+ Response Require Transmitter Release? If the Ca2+ response in NMSCs is due to molecules secreted by the nerve terminal during neurotransmission, then the response should be reduced when transmitter release is blocked. This hypothesis was tested by blocking transmitter release with o-conotoxin GVIA (w-CgTx), which binds irreversibly to certain types of Ca2+ channels (Gray et al., 1988), blocks evoked transmitter release at the frog NMJ (Kerr and Yoshikami, 1984), and blocks the entry of

Ca*+ in the nerve terminal as determined by blockade of the fluo-3 Ca*+ signal (Robitaille and Charlton, 1992). After blockade of transmitter release by o-CgTx (IO pg/ml), the nerve-evoked Ca*’ signal in NMSCs was, on average, 89.8% smaller than that in the absence of o-CgTx (13% rt 2%; 10 cells in 5 muscles; Figure 26). Thus, substances released by the nerve terminal during evoked activity are necessary to induce the response in the Schwann cells. The remaining nerve-evoked Ca2’ signal, which is resistant to o-CgTx, could be caused by K+ accumulation near the active terminal (Takeuchi and Takeuchi, 1961; Erulkar and Weight, 1977). Thiswould depolarize the Schwann cell membrane and allow Ca2+ entry through voltage-sensitive Ca2+ channels (MacVicar, 1984; Newman, 1985,1986; Berwald-Netter et al., 1986; Barres et al., 1990). Indeed, local application of KCI (25 mM in a micropipette) in the presence of o-CgTx elicited a Schwann cell Ca2+ response (69% 5 17%; 11 cells in 5 preparations). Furthermore, the w-CgTxresistant nerve-evoked signal was abolished in Cai-free, 5 mM Mg2+ saline, suggesting that Ca*+ entry into Schwann cellsthrough channels insensitivetow-CgTx was required for this small response. The persistence of a depolarization-dependent Ca*+ response after w-CgTx application is consistent with the failure to detect o-CgTx binding on these Schwann cells (Robitaille et al., 1990; Cohen et al., 1991). Can Transmitters Trigger CaZ’ Responses? If substances such as the cotransmitters ACh and ATP (Silinsky, 1975; Smith, 1991) secreted by the active nerve terminal cause Schwann cell Ca*’ signals, then application of these substances alone should mimic the effects induced by nerve stimulation. ACh was applied to Schwann cells by pressure via a micropipette. Neostigmine (3 Kg/ml), a blocker of synaptic acetylcholinesterase, was used to avoid the production of choline and acetate. Application of ACh (20 PM to 1 mM in the micropipette) induced a CaZ- response in NMSCs (107% c 18%; 23 cells in 7 muscles; Figure 4A). The sensitivity to ACh was not affected byo-CgTx because Caz+ responses to ACh were still obtained after application of the toxin (124% + 49%; 6 cells in 3 muscles). Therefore, the reduction of the nerveinduced response by o-CgTx (Figure 2B4) is not due to blockade of the ACh effect. Application of the two degradation products of ACh, choline (choline chloride) and acetate (potassium acetate), for up to 10 s (both at 1 mM in the micropipette), in the presence of neostigmine, failed to induce any increase of CaZ’ in the NMSCs. However, when applied after choline or acetate, ACh still caused a typical Ca2+ response. Therefore ACh, but not its breakdown products, is a signal recognized bythe Schwann cells. Local application of ATP (0.1-5 mM ATP in the micropipette) also induced a Ca2’ response (213% i- 28%; 7 cells in 3 muscles) in NMSCs (Figure 5A). Sincethe nerve-evoked Ca2+response in the NMSCs decreases during consecutive trains of stimuli (Figure

Nerve-Evoked 1073

3A), duced would ond only to a elicit

Ca2+ Signals

in Schwann

Cells

we

wondered whether the Ca*+ responses inby local applications of the neurotransmitters also be susceptible to rundown. Indeed, a secapplication of ACh evoked responses that were 10.1% k 5.3% of the first response. The response third application of ACh was more difficult to (i.e., more ACh had to be applied) and was often

completely

absent

(Figure

3B).

Release of Ca2+ from Internal Stores To determine whether the source of the Ca*+ signal triggered by ACh or ATP is influx of extracellular Ca2+ or Ca*+ released from intracellular stores, we removed extracellular Ca*+ and locally applied ACh or ATP. Local applications of ACh induced a fluorescence increase of 169% k 20% (15 cells in 3 muscles) in the absence of extracellular Ca2+ (Figure4B). ACh induced Ca2+ responses (125% & 34%; 7cells in 2 muscles) even when the Ca*+chelator EGTA (1 mM) was added to the Ca*+-free, 5 mM Mg*+ saline to reduce extracellular Ca2+ further. Local application of ATP in the absence of Ca2+ caused an increase in fluorescence intensity of 232% f 11% (11 cells in 2 muscles; Figure 58). The occurrence of responses to ACh and to ATP in the absence of extracellular Ca*+ indicates that these signalsareduemainlytothereleaseofCa2+fromintracellular stores. Discussion We present here direct evidence that, under normal physiological conditions, perisynaptic Schwann cells in situ respond to secretion of transmitter substances by an elevation of their intracellular Ca2+ level and are thus active participants during synaptic transmission. Most of the Ca*+ response was abolished when evoked transmitter release was blocked. The cotransmitters ACh and ATP can independently produce Ca2+ responses in the absence of extracellular Ca2+. Thus, evoked release of these transmitters may trigger Ca*+ release from internal stores in perisynaptic Schwann celIs.Wecannot, however,excludethepossibilitythat other cotransmitters, such as peptides (Matteoli et al., 1988, 1990), also participate in the Schwann cell Ca2+ response to nerve stimulation. Recently, Ca2+ responses were observed in astrocytes in the CA3 area of rat hippocampal organotypic cultures subsequent to electrical stimulation of the dentate gyrus (Dani et al., 1992). Nerve stimulation caused propagation of Ca*+ waves within individual cells that spread to neighboring astrocytes, thus revealing an extended network between glial cells. This indicates that the Ca2+ responses in these astrocytes may depend on multiple signals such as nerve-evoked transmitter release and propagating Ca2+ signals from neighboring astrocytes. In contrast, the data presented here demonstrate that synaptic activity at a single nerve terminal is sufficient to trigger a Ca*+ response in a single perisynaptic glia-like cell. Therefore, these responses are not an emergent property

dependent on a complicated multicellular network. Owing to its unusual accessibility and simplicity, the frog NMJ provides a preparation in which the physiological relationships between single glia-like cells and single identified nerve terminals can be studied in situ with both electrophysiological and optophysiological techniques. Indeed, we and others (McMahan et al., 1972; Dennis and Miledi, 1974) have penetrated these Schwann cells with microelectrodes, measured resting potentials, and injected dyes into the cytoplasm. By permitting independent manipulation of the Schwann cell Ca*+ signal, these techniques will allow us to study the regulatory roles of the Schwann cells in synaptic transmission. TheNMSCCa2+responsesarecharacterized byslow onset, rise, and decay times and a reduction in their amplitude after repetitive nerve stimulations or local applications of ACh. These characteristics are also found in other types of glial cells isolated from brain preparations. Glial cells isolated in vitro respond to application of several other transmitters in addition to cholinergic agonists and ATP (Orkand, 1982; Hamprecht, 1986; Lauder and McCarthy, 1986; Cornell-Bell et al., 1990; Neary et al., 1988; Pearce et al., 1985; Dave et al., 1991; Kimelberg and Norenberg, 1989). Local applications of neurotransmitters and ligands, such as glutamate, y-aminobutyric acid, and carbachol, induce the release of Ca*+ from internal stores, and these responses are sometimes reduced after repetitive applications (for a review, see Cornell-Bell and Finkbeiner, 1991). Similar to responses in Schwann cells, nerve-evoked Ca2+ responses in hippocampal astrocytes also follow nerve stimulation by a few seconds (Dani et al., 1992). Although we have not identified the receptors involved, the general characteristics of the Ca2+ responses suggest that the ACh receptor might be of the muscarinic type. Similar to our results, muscarinic receptors have been shown to cause release of Ca*+ from internal stores, with rundown following repetitive application of agonists (Pearce et al., 1985; Dave et al., 1991). In addition, our preliminary results at the frog NMJ indicate that nicotinic cholinergic antagonists such as a-BuTx or d-tubocurarine chloride do not block the Ca2+ response to applied ACh, whereas local application of the muscarinic agonist muscarine induces a Ca*+ response (data not shown). The Schwann cell Ca*+ signals persist after application of sufficient a-BuTx to block postsynaptic electrophysiological responses to evoked ACh release and to applied ACh. Because the Schwann cell receptors are different from the nicotinic postsynaptic receptors, they can be differentially blocked. This allows us to deduce that the nicotinic response of postsynaptic cells to ACh is not required to trigger the Schwann response. Additional experiments at the frog NMJ are in progress to determine the nature of the receptors and possible second messengers involved in the NMSC Ca2+ response. Schwann cells can be separated into two groups,

NWKKl 1074

4A

NORMAL

5A

Cat’

NORMAL

Ca+ +

REST

ACh

4Al

g

5Al

120 1

0

4B

f?

30

60 90 Time (s)

0-Ca++

b+‘ 300 L a 225 s ; 150 c .ean 75 m 0 + t 0

120

5B

REST

REST

ACh

ATP

45

90 135 180 Time (s)

Nerve-Evoked 1075

Ca2+ Signals

in Schwann

Cells

For imaging Ca2+ in perisynaptic Schwann cells, muscles were incubated with fluo 3-AM, the membrane-permeant form of the fluorescent Ca2+ indicator fluo-3 (Molecular Probes) (Kao et al., 1989; Tsien, 1989). which increases fluorescence with Ca’+. Solutions of fluo 3-AM (IO PM) were prepared in normal frog Ringer’s solution containing a final concentration of 1% (v/v) dimethyl sulfoxide and 0.02% pluronic acid (Molecular Probes), a detergent that facilitates indicator loading. Muscles were incubated for 90-120 min at room temperature (18”C-2OOC). Heavy metals that bind to flue-3 and limit its response to Ca2+ were partially chelated by tetrakis @pyridylmethyl) ethylenediamine (TPEN, IO-20 PM; Molecular Probes) (Arslan et al., 1985) added to the saline. Similar increases in fluorescence were obtained without TPEN, but results were more uniform when it was used.

myelinating and nonmyelinating, each bearing different molecular markers (lessen and Mirsky, 1991). The data presented in this study reveal other differences. A Ca2+ signal was evoked in NMSCs by brief nerve stimulation or by local application of the transmitters, whereas the myelinating Schwann cells at the last myelinated segment of the axon did not produce any Ca2+ signal. This indicates that the myelinating Schwann cells may lack appropriate receptors coupled to internal Ca2+ stores. The roles of the NMSCs at the frog NMJ and the role of the Ca*+ response induced by nerve-evoked synaptic activity remain unknown. The Ca*+ signal may trigger the secretion of nerve terminal modulators or neurotransmitter substances (Bevan et al., 1973; Dennis and Miledi, 1974; Philibert et al., 1988) that could affect the operation of the synapse (for a review, see Martin, 1992). This signal might also increase glucose metabolism by Schwann cells and triggerthe reuptakeof metabolites released during nerve terminal activity (Ransom and Carlini, 1986; Massarelli et al., 1986; Derouiche and Frotscher, 1991). Since these Schwann cells undergo profound changes upon denervation (Bevan et al., 1973; Dennis and Miledi, 1974), the Ca2+ signal might also function to maintain the cells in a physiological state compatible with the normal activity of the nerve terminal. Experimental

Fluo-3 Fluorescence Measurement Changes in fluorescence intensity were observed with a Bio-Rad 600 confocal laser scanning microscope (Shotton, 1989) using a 40x water immersion objective (Nikon; 0.55 NA). The 488 nm excitation line of the laser was attenuated to 1% of the maximum power, and emission was detected through a low pass emission filter with cutoff at 515 nm. The endplates were located using standard phase microscopy. Changes in fluorescence in Schwann cells were monitored during transmitter release evoked by nerve stimulation (40 Hz, 30 s), and images (128 x 192 pixels) were collected every 500 ms. The fluorescence intensity (F) was averaged over the cell body area of the Schwann cells using custom software, and fluorescence changes induced by the stimulation were expressed as AF/F All values

as mean

f

SEM.

Local Applications of Drugs For application of transmitters, NMJs were prepared as described above. ACh or ATP, dissolved in the same saline as that in the bath, was ejected by pressure pulse (