235

Journal of Physiology (1992), 447, pp. 235-256 With 12 figures Printed in Great Britain

CALCIUM INFLUX AND CALCIUM CURRENT IN SINGLE SYNAPTIC TERMINALS OF GOLDFISH RETINAL BIPOLAR NEURONS BY RUTH HEIDELBERGER AND GARY MATTHEWS From the Department of Neurobiology and Behavior, the State University of New York, Stony Brook, NY 11794-5230, USA

(Received 22 April 1991) SUMMARY

1. The calcium influx pathway in large synaptic terminals of acutely isolated bipolar neurons from goldfish retina was characterized using Fura-2 measurements of intracellular calcium and patch-clamp recordings of whole-cell calcium current. 2. Depolarization of bipolar cells with high [K+]0 resulted in a sustained, reversible increase in [Ca2+]i in both synaptic terminals and somata. Removal of external calcium abolished the response, as did the addition of 200 /tM-cadmium to the bathing solution, indicating that the rise in [Ca2+]i was due to entry of external calcium. Dihydropyridine blockers of voltage-gated Ca2+ channels also blocked the influx, and the Ca2+ channel agonist Bay K 8644 potentiated influx, implicating voltage-activated, dihydropyridine-sensitive channels in the influx pathway. 3. Under voltage clamp, depolarization from a holding potential of -60 mV evoked a slowly inactivating inward current that began to activate at -50 to -40 mV and reached a maximal amplitude between -20 and -15 mV. This current was identified as a calcium current because it decreased when the extracellular calcium concentration was lowered, increased when barium was the charge carrier, and was blocked by 200 JtM-external cadmium. The current was substantially blocked by 1 /uM-nitrendipine and potentiated by 0 1 JIm-Bay K 8644, as expected for L-type Ca2+ channels; it was unaffected by w)-conotoxin. No evidence for transient or rapidly inactivating Ca2+ current was found. 4. At a given level of potassium depolarization, both the amplitude and the speed of increase in [Ca2+]i were greater in synaptic terminals than in somata. For instance, depolarization by 32-6 mM-potassium caused an increase in intracellular calcium of 400 + 23 nm in terminals and 180 + 20 nm in somata (mean +s.E.M., n = 73 terminals, n = 30 somata), with maximal rates of change of 40 + 3 and 12 + 2 nM/s, respectively. 5. The contribution of terminal and somatic currents to the total whole-cell Ca2+ current was determined under voltage clamp by local application of calcium or of blocking agents. While there was no qualitative difference between currents in terminals and somata, synaptic terminals accounted for 64+3% (mean+S.E.M., n = 12) of the total whole-cell calcium current, and somata accounted for 39 + 2 %. Thus, the density of Ca2+ current was higher in the terminal, accounting for the greater magnitude and speed of Ca2+ influx observed in terminals in Fura-2 experiments. MS 9320

236

R. HEIDELBERGER AND G. MATTHEWS

6. The existence of a single type of calcium current in goldfish bipolar neurons and its prevalence in the synaptic terminal, as well as its physiologically relevant activation range, suggest that the calcium current described in this paper plays an important role in synaptic transmission. INTRODUCTION

The role of calcium in synaptic transmission is well established (Katz & Miledi, 1967; Llina's, Steinberg & Walton, 1981; Augustine, Charleton & Smith, 1985). However, the small size and inaccessibility of most synaptic endings hinder direct physiological study of calcium influx in vertebrate CNS nerve terminals. In this paper, we characterize Ca2" influx in the unusually large synaptic terminal of a CNS interneuron, the goldfish retinal bipolar neuron. In the retina of goldfish and carp, some types of bipolar neuron have synaptic terminals that measure 8-12 ,tm in diameter (Ishida, Stell & Lightfoot, 1980; Saito & Kujiraoka, 1982; see also Fig. 1). Intact bipolar neurons, including the synaptic terminal, can be isolated from goldfish retina (Kaneko & Tachibana, 1985), which has enabled us to make patch-clamp and Fura-2 investigations of calcium regulation in individual synaptic terminals of largeterminal bipolar cells. We found a single type of Ca21 current that underlies depolarization-induced Ca21 influx in these cells. This current is activated from a relatively depolarized holding potential, is sensitive to dihydropyridine Ca21 channel drugs, and inactivates slowly, all of which are features of L-type Ca2' channels (Fox, Nowycky & Tsien, 1987). Preliminary reports of some of the data have appeared (Heidelberger & Matthews, 1990; Matthews & Heidelberger, 1990). METHODS

Cell isolation Bipolar cells were freshly isolated each day from goldfish retina by the method of Tachibana (1983). Briefly, dark-adapted goldfish 8-12 cm in length were killed by decapitation, and the eyeballs were removed and placed into a cold, oxygenated low-calcium solution (0-2 mM-Ca2+ Ringer solution) containing (mM): NaCl, 120; KCl, 2-6; MgCl2, 1 0; CaCl2, 0-2; glucose, 10; HEPES, 3; pH 7-3. Retinas were dissected free and treated for 20 min with hyaluronidase (1100 units/ml; Sigma, type V). After rinsing, each retina was chopped into eight to sixteen pieces and placed for 15-90 min into digestion solution composed of (mM): NaCl, 115; KCl, 2-5; MgCl2, 1-0; CaC12, 05; glucose, 10; HEPES or PIPES, 10;cysteine, 2-7; pH 6-9, and papain (7-30 units/ml, depending on source and lot number). All enzyme treatments were performed at room temperature (18-21°C). After enzyme treatment, the retinal pieces were rinsed several times with 0 2 mM-Ca2' Ringer solution supplemented with 0 5 mg/ml bovine serum albumin (BSA) and stored at 12 'C. Pieces were triturated with fire-polished glass pipettes in the BSA-containing 0-2 mM-Ca2' Ringer solution and, for use in Fura-2 experiments, were plated out onto clean glass cover-slips in 0-2 mM-Ca2+ Ringer solution and stored in a moist oxygen atmosphere at 17 'C. Cells used for voltage-clamp studies were freshly triturated as needed from stored pieces. Experiments were done on bipolar cells with large synaptic terminals. On the basis of their flaskshaped cell body, thick main dendrites, and 8-12 ,m synaptic terminal, cells were probably the rod-dominant, on-type bipolar neurons termed typeB1 by Ishida et al. (1980). Figure1 shows representative cells, which illustrate the range of cell length and cell morphology encountered. The variation in length presumably reflects the dorsal-ventral location from which the cells were obtained in the retina, which in goldfish is thicker at the ventral pole.

Fura-2 measurements Microscope modifications, equipment and software used to make fluorescence measurements were similar to those described in detail by Neher (1989). A filter wheel rotating at 4 revolutions/s

CALCIUM IN RETINAL BIPOLAR CELLS

237

was placed in the epifluorescence excitation path. The wheel contained two filters selected to give alternating excitation near 360 and 380 nm. The intracellular free calcium concentration was calculated at the rate of two data points per second from the ratio of the emitted fluorescence at these two wavelengths by the method of Grynkiewicz, Poenie & Tsien (1985), using calibration A

L

25 pml'-,0--;, C

D ......I.*

E-_

Fig. 1. Examples of large-terminal bipolar neurons isolated from goldfish retina after papain digestion. The circle around the synaptic terminal of the cell in B shows the size of 30 /um diameter region from which light was collected for the Fura-2 measurements of [Ca21]i. The cells in C and D have patch pipettes placed on the terminal and soma, respectively. constants experimentally determined in cells as described by Neher (1988, 1989). In this method, cells (either bipolar neurons or rat pheochromocytoma (PC12) cells) were internally dialysed via a whole-cell patch pipette with solutions containing 100 /tm-Fura-2 and known concentrations of free Ca21 set by Ca-EGTA/K -EGTA buffer combinations. A pinhole aperture was inserted before the photomultiplier tube so that emitted light from the Fura-2 fluorescence was collected solely from a region of the total field measuring 30 4um in diameter. The size of the aperture is indicated by the circle around the synaptic terminal of the cell in Fig. IlB. The aperture excluded 98-3 ± 0-2% (mean ± SE.M. , n =10) of the ligaht originating from an object placed 2 Itm outside the rim of the aperture. The distance between terminal and soma in isolated large-terminal bipolar neurons ranged from 20 to 40 /um. Thus, by appropriate positioning of the cell, it was possible to selectively collect light from either the terminal or the soma. Cells were plated onto glass cover-slips and loaded in darkness for 15 20 min at 15-17 'C with 1 /Sm membrane-permeant Fura-2 acetoxymethyl ester (Fura-2 AM). After the loading period, the cells were washed with 15 ml of the extracellular bath solution to remove any excess dye not taken up by the cells. To determine the proportion of loaded Fura-2 AM that had been converted to the free acid and that was accessible to calcium entering from the external solution, cells loaded with Fura-2AM were exposed to 1-10 #sm-4Br-A23187 (Deber, Tom-Kun, Mack & Grinstein, 1985) and then to 2-5 mm-external Mn21. As with Quin-2 (Hesketh, Smith, Moore, Taylor & Metcalfe, 1983; Grynkiewicz et al. 198,5), Mn21 binds to de-esterified Fura-2 and quenches the fluorescence. Introduction of Mn2" after 4Br-A23187 quenched 94-2 + 0-7 % (mean +S5 E.M. , n = 3) of cellular fluorescence, demonstrating that the bipolar neurons had effectively converted Fura-2AM to Fura2 and that the dye was largely accessible to ionophore-mediated divalent cation entry. For fluorescence measurements, a cover-slip of cells was transferred to a glass-bottomed chamber on the cooled stage of the microscope. During experiments, the bath temperature ranged between

238

R. HEIDELBERGER

ANTD

G. MA 1THEWS

15 and 17 'C. The standard extracellular bath solution (choline Ringer solution) consisted of (mM:) NaCl, 60; choline chloride, 60; KCl, 2-6; CaCl2, 2-5; MgCl2, 1-0; glucose, 10; HEPES, 3; pH 7-3. Cells were depolarized with high-K+ solutions in which K+ replaced the appropriate amount of choline chloride, keeping external sodium concentration and osmolarity constant. In some of the later experiments, choline chloride was eliminated, and the total NaCl concentration was increased to 120 mm (2-5 mM-Ca2` Ringer solution). For these experiments, then, added K+ replaced Na' in high-K+ solutions; no difference in the results was observed compared to experiments in which K+ replaced choline at constant [Na+]O. All solution changes were made by bath exchange, using push-pull syringes so that fluid was removed from the chamber at the same rate it entered, keeping bath volume constant. The rate of exchange varied, depending on how firmly the cells were attached to the underlying cover-slip, but complete exchange of the 15 ml bath volume was commonly achieved in approximately 10 s. Drug effects were calculated as the ratio of the K+induced change in [Ca2+]i in the presence of the test solution divided by the average of the control responses to high K+ before and after the test solution. Voltage clamp Recordings were made using a List EPC-9 patch-clamp amplifier (EPC-7 in some early experiments). Whole-cell. electrodes were pulled from thick-walled Pyrex glass, with an outer diameter of 1-2 mm. After fire-polishing, typical electrode resistances were 10-30 MQ. In order to block potassium current and reveal whole-cell calcium current, the standard intracellular solution contained Cs+ and TEA, with no added K+; to more closely approximate the presumed low-Clinternal fluid of intact cells, the principal anion of the pipette solution was gluconate. Thus, the standard intracellular solution (caesium gluconate) used for recording whole-cell calcium currents consisted of (mM): caesium gluconate, 120; TEA-Cl, 10; MgCl2, 2; Na2ATP, 2; GTP, 0 3; leupeptin, 0-2; EGTA, 5; HEPES, 3; pH 7-2 with CsOH. In some experiments CsCl was substituted for caesium gluconate. In experiments in which it was not desired to block potassium currents, 130 mM-potassium gluconate replaced caesium gluconate and TEA. The extracellular solution was 2 5 mM-Ca2' Ringer solution. The sampling interval during voltage steps was 0 5 ms. Capacitative and leak currents were subtracted, except where noted, using a p/n method, with n 4 or 6. In this protocol, each test pulse is followed by n copies of the test pulse, each with amplitude 1/n of the test pulse amplitude. The responses to the p/n pulses were then summed and subtracted from the response to the test pulse to eliminate linear components. The p/n pulses were delivered at a potential chosen to prevent activation of voltage-sensitive conductances during the p/n pulse (-70 mV in the present experiments). Linear voltage ramps at 100 mV/s were sometimes used to obtain current-voltage relations, in which case sampling interval was5 ms. Control experiments showed that current-voltage relations measured with ramps were identical to those measured more conventionally with families of 70 ms rectangular voltage pulses. Only cells with leak current less than 20 pA and series resistance less than 50 MQ were chosen for study. Typically, about 50 % of series resistance was electronically compensated, using the automatic series-resistance compensation of the EPC-9 amplifier. To minimize calcium current run-down, repetition intervals greater than 6 s were used, and depolarizing pulses lasted 70-150ms. Recording electrodes were typically placed on the terminal, but there was no difference in the measured currents if the recording electrode was on the soma. All reported voltages are corrected where appropriate for the -6 mV junction potential between gluconate-containing pipette solution and Ringer solution. Unless otherwise noted, drugs were applied by local superfusion. Application pipettes (orifice diameter = 5-10,um) were used to deliver test solutions to the soma, terminal, or entire cell. Application pipettes were continuously and gently blowing, and were lowered or raised in the bath solution to apply or remove drugs. w-Conotoxin (Calbiochem; Research Biochemicals, Inc.) was made freshly each day as a1 mM stock solution. Potency ofwo-conotoxin was verified by blockade of neurotransmission at the frog neuromuscular junction. All other drugs were made up as stock solutions, aliquoted and stored in the dark at -20 'C, except Bay K 8644, which was stored at 4 'C. Fura-2AM (Molecular Probes) was dissolved in chloroform, aliquoted into 50 samples, dried and stored desiccated. Samples were reconstituted in dimethyl sulphoxide (DMSO) to give a stock concentration of1 mm. The dihydropyridines nitrendipine and Bay K 8644 were made up in 95 % ethanol as 5 or 10 mm stock solutions. =

CALCIUM IN RETINAL BIPOLAR CELLS

239

RESULTS

Fura-2 measurements Characterization of calcium influx The effects of membrane depolarization in intracellular calcium were studied in freshly isolated bipolar cells loaded with the membrane-permeant form of the calcium-sensitive fluorescent dye Fura-2. Cells were depolarized by elevating extracellular potassium. Figure 2A shows Fura-2 fluorescence measurements of [Ca2+]i from a synaptic terminal (top trace) and soma (bottom trace) of a bipolar neuron at rest (2-6 mm [K+]0) and during depolarization by 62-6 mM-potassium. In response to K+-induced depolarization, there was a sustained increase in [Ca2+]i in both the soma and synaptic terminal that was reversible upon wash-out of the highK+ solution. Reproducible increases in [Ca2+]i upon depolarization could be elicited for more than 20 min in a given neuron. Whole-cell patch-clamp recordings under current clamp showed that bipolar neurons depolarized from -50 + 3 to -17 + 3 mV (mean + S.E.M., n = 4) when [K+]o was increased from 2-6 to 62-6 mm. This placed the membrane potential within the activation range of Ca2+ current (e.g. Fig. 6). To test whether the increase in [Ca2+]i induced by high [K+]o is due to influx of external Ca2+, cells were depolarized in extracellular Ringer solution containing 5 mM-EGTA and no added Ca2+, and the response was compared with that in 2-5 mMCa2+ Ringer solution (Fig. 2A). In contrast to the results in 2-5 mM-Ca2+ Ringer solution, high K+ in the absence of extracellular Ca2+ caused no change in [Ca2+]i. Upon return to 2-5 mM-Ca2+ Ringer solution, K+ depolarization once again produced an increase in [Ca2+]i. Thus, influx of extracellular calcium is essential for an increase in intracellular calcium to occur upon depolarization. This was true of both synaptic terminals (n = 6) and somata (n = 7). To determine whether the calcium that enters the cell upon depolarization enters through voltage-gated Ca2+ channels, the effect of cadmium, a blocker of voltagegated Ca2+ channels, was examined on depolarization-induced Ca2+ influx. Addition of 200 ,uM-cadmium to the bathing solution reversibly blocked the depolarizationinduced increase in [Ca2+]i in both terminals (n = 3) and somata (n = 8). This is illustrated in Fig. 2B for a terminal (top trace) and soma (bottom trace). The combined results of the experiments shown in Fig. 2 demonstrate that upon membrane depolarization, Ca2+ enters bipolar cell somata and synaptic terminals through voltage-gated Ca2+ channels, elevating [Ca2+]i.

Pharmacology of calcium influx To establish the pharmacology of the influx pathway, the actions of dihydropyridine Ca2+ channel drugs were examined on depolarization-induced Ca2+ influx. At concentrations of 1 ,UM or below, dihydropyridines have been shown to act specifically on L-type Ca2+ channels, but not N- or T-type (Bean, 1984; Nowycky, Fox & Tsien, 1985a). The dihydropyridine Ca2+ channel blocker nitrendipine reduced depolarization-induced Ca2+ influx in both synaptic terminals and somata of bipolar neurons (Fig. 3A). In the traces shown in Fig. 3A, 100 nM-nitrendipine blocked 56% of the increase in [Ca2+]i in response to 62-6 mM-potassium in a bipolar cell soma, and

R. HEIDELBERGER AND G. MATTHEWS

240 A

Synaptic terminal 0

+'--

+

Soma

-

-

---

62.6 mM

62.6 mM + +2+ aK O

62.6 mM K

Cal+ LJ 50 s

1 -

+

~

~

O

K040 Ca2+1 -

-

-

-

62.6 mM

62.6 mM

62.6 mM

Ko

Ko + O Ca+

Ko

100 s

B

Synaptic terminal

200

uM-Cd2+

--

0

-

'

-

LJ---

62.6 mM 62.6 mM N Ko+200X&& + Cd+2+

62.6 mM

K+

100 s

Soma l200uM-Cd2+

0

-

-

-

-

-

-

62.6 mM

-

-

-

-

62.6 mM

82.6 mM

100 s

Ko KO+ Cd+ Ko Fig. 2. Effects of external Ca2+ and Cd2+ on the increase in [Ca2+]1 elicited by elevated [K+]0 in bipolar neurons. Cells were loaded with Fura-2 by incubation in Fura-2AM. In all instances, the bars indicating exposure to a particular solution show the time from the beginning of bath exchange applying the solution to the beginning of wash-out of the solution. The control solution at all times not indicated by the bars was choline Ringer (see Methods). A, effect of removing external Ca2+ on the increase in [Ca2+]1 elicited by depolarization with high [K+]. Top trace shows changes in [Ca2+]i recorded from a synaptic terminal, and the lower trace shows [Ca2+]i of the cell body of a different neuron. During the filled bars below the traces, cells were exposed to 62-6 mm [K+]0; periods of zero external calcium are indicated by the grey bars above the traces. Dashed line indicates zero level. 0 Ca2+ Ringer solution contained 5 mM-EGTA and no added Ca2+. B, 200 4uM-Cd2+ added to the bath solution blocked the K+-induced rise in [Ca2+]1. Terminal (top trace) and soma (bottom trace) are from different neurons than those in A. Filled bars below traces indicate periods of 62-6 mM-K+, and grey bars above the traces indicate periods of exposure to Cd2 . In these experiments, elevation of [K+]o was achieved by replacing choline chloride in the extracellular solution with K+, so that the osmolarity and external [Na+] remained constant.

CALCIUM IN RETINAL BIPOLAR CELLS

241

A 0 1 PM nitrendipine

Synaptic

terminal -~1

CU

c- n

-

-

-

-

-

-

-

-

-

-

-

-

-

62.6 mM

62.6 mM

K0

K0 + nitrendipine

L

--

-

-

100 s

-

-

-

-

62-6 mM Ko

01 pM nitrendipine

Som3 CD

0

-

-

-

-

-

62.6 mM

-

-

-

-

-

-

-

L

62.6 mM Ko+ nitrendipine

62.6 mM

Ko

---

Ko

62.6 mM

200 s

K0

B

Synaptic

0.1 pM-Bay K

terminal CU --

-

-

-

22.6 mM

22.6 mM

22.6 mM

Ko

K4+ Bay K

Ko

L-

100 s

0-1 pM-Bay K Soma CU -

-

-

-

-

-

32.6 mM

-

-

m

-

32.6 mM

32.6 mM

-

LJ

32.6 mM

100 s

KKo Ko+BayK Ko Fig. 3. Effect of dihydropyridine Ca2+ channel drugs on depolarization-induced increase in [Ca2+]i. A, 100 nM-nitrendipine, a Ca2+ channel antagonist, reversibly blocked 52 % of the peak increase in [Ca2+]1 in a terminal (top trace) and 56 % of the increase in a soma in response to depolarization by 6216 mM-K+. B, 100 nM-Bay K 8644, a Ca2+ channel agonist, enhanced the depolarization-induced increase in [Ca2+]i in both the synaptic terminal (top trace) and soma (lower trace). 2216 mM-K+ was used to depolarize the terminal, and 3216 mM-K+ was used to depolarize the soma.

52% of the peak increase in a terminal. The effect of nitrendipine reversed upon wash-out. The dose dependence of nitrendipine block is shown in Fig. 4. In general, the sensitivity of terminals and somata to nitrendipine was identical, and therefore results from terminals and somata were combined for each concentration of

R. HEIDELBERGER AND G. MA TTHEWS

242

nitrendipine. Figure 4 shows that 100 nM-nitrendipine blocked 55 + 4 % (mean + S.E.M., n = 22) of the peak increase in [Ca2+]i in terminals and somata in response to depolarization by 62-6 mM-Ky, 500 nm-nitrendipine blocked 64 + 7 % (n = 8) and 1 /IM-nitrendipine blocked 86 + 3 % (n = 5). At the concentrations of nitrendipine 100 0 0

80

C, 60

-

U

,, 40

~

C

C

_

20

0

0.2

0.4

0-6

0.8

1.0

Nitrendipine (pm) Fig. 4. Concentration dependence of the block of Ca`+ influx by nitrendipine. Points are pooled data from terminals and somata. Cells were depolarized with 62 6 mm [K+].. For 0 1 ,um-nitrendipine, n = 22; for 0 5 ,um-nitrendipine, n = 8; for I /um-nitrendipine, n = 5. Error bars indicate+S.E.m. Data points are peak change in [Ca2+]i induced by high [K+]. in the presence of drug divided by the average of peak K+-induced changes in [Ca2+]i before and after the drug.

used, the effect of the dihydropyridine was reversible, but complete blockade of Ca2+ influx was not routinely observed. However, the block achieved was comparable to that reported previously for L-type Ca2+ current (Bean, 1984) and to the block observed in our own patch-clamp studies of Ca 2+ current in bipolar cells (see below; Figs 8 and I10). Thus, it appears that a maj or part, if not all, -of the Ca2+ influx induced by high K+ occurs via dihydropyridine-sensitive, L-type Ca 2+ channels. If influx occurs via L-type Ca 2+ channels, then the dihydropyridine Ca2+ channel agonist, Bay K 8644, which enhances L-type Ca2+ currents by prolonging Ca 2+ channel open times (Hess, Lansman & Tsien, 1984; Nowycky, Fox & Tsien, 1985b), should enhance K+-induced Ca 21 influx in somata and terminals of bipolar neurons. As shown in Fig. 3B, this was indeed the case. For these experiments, an intermediate amount of K+ depolarization was chosen, which would be large enough to give a reliable rise in [Ca 2+ ]i but small enough to allow enhancement to be seen. In somata, the response to 326 mm-potassium was 5-4 +24 times larger (mean +S.E.M., n = 4) in the presence of 0 1 /,tm-Bay K 8644 than in high K+ alone. In terminals, the response to 22-6 mm-potassium was 1 7 + 0 1 times larger (n = 10) in the presence of 0-1 ,um-Bay K 8644. These results, again, suggest the involvement of dihydropyridine-sensitive Ca2+ channels in depolarization-induced Ca2+ influx in

bipolar neurons.

CALCIUM IN RETINAL BIPOLAR CELLS

243

Comparison of soma and terminal The increase in [Ca2+]i at a given level of [K+]. was greater in synaptic terminals than in somata, as illustrated in Fig. 5. When [K+]. was increased to 32-6 mm, [Ca2+]i in the terminal shown in Fig. 5A increased by 390 nM; in seventy-three terminals, the Synaptic terminal

A 0-5

+0>..

I

O

- - - -

-............. -

-

50-. C

32.6 mM-K+

B

Soma

0.51 ]

+

-

-

.... -

-

-

-

-

-

-

-

.-

-

-

-

.-

.

Coc

50-

~~~~32-6 mM-K+

-

C 0.5 .1 ._0-5

Soma without terminal

.I 50 C,,

I

32.6 mM-K+

50s

Fig. 5. Comparison of increase in [Ca2+]i elicited by 32-6 mm [K+]. in a bipolar cell synaptic terminal (A), in the soma of an intact bipolar neuron with axon and terminal (B), and in the soma of a bipolar neuron lacking axon and terminal (C). All three cells are from the same preparation. The top trace of each pair shows [Ca2+],, and the lower trace of each pair shows the rate of change of [Ca2+], (d[Ca2+],/dt). The time scales and vertical scales are the same in A, B and C. Timing of application of 32-6 mm [K+],, is indicated by the filled bars below each pair of traces. The bars indicate the time from beginning of rwash-in to beginning of wash-out of high-K+ solution.

244

R. HEIDELBERGER AND G. MATTHEWS

peak increase in [Ca2+]i elicited by 32-6 mm [Ki]w was 400 + 23 nm (mean + S.E.M). In the soma of a different bipolar neuron in the same preparation (Fig. 5B), the same level of [K+]. increased [Ca2+]i by 240 nM. In thirty somata, the increase in [Ca2+]i with 32-6 mm [K+]. averaged 180 + 20 nM. In addition, the rate at which [Ca2+]i increased, once the response had begun, was greater in terminals than in somata, as illustrated by the lower traces in Fig. 5A and B. In the seventy-three terminals, peak d[Ca2+]j/dt in response to 32-6 mm [K+]. averaged 40 + 3 nM/s, while in the thirty somata, the comparable value was 12+2 nm/s. It can also be seen in Fig. 5 that the delay between onset of the bath exchange to the high-K+ solution and the beginning of the rise in [Ca2+]i was greater in the soma than in the terminal. Bath exchange is not well-suited to the study of such temporal properties, because the time of onset of the actual depolarization in the cell is unknown and possibly variable, depending on flow dynamics and location of the cell within the recording chamber. Nevertheless, we can eliminate one possible explanation for the greater delay in somata, namely that all the Ca2+ influx actually occurs in the terminal, with the delay arising from internal diffusion of Ca2+ to the soma. Bipolar somata that lost their axons and synaptic terminals during dissociation show depolarization-induced increase in [Ca2+]i that is similar to that of somata with intact terminals. Figure 5C shows the response of one such bipolar soma without an axon, in the same preparation as the other two neurons of Fig. 5. In eighteen axonless bipolar somata, the increase in [Ca2+]i elicited by 32 mm [K+]o was 142 + 26 nm (mean +s.E.M.) and the peak d[Ca2+]j/dt was 9 + 3 nm/s. In addition, patch-clamp recordings from axonless bipolar somata confirmed that Ca2+ current is present in somata (seven out of seven cells examined), and experiments on the origin of whole-cell Ca21 current in cells with intact terminals also point to a somatic contribution of Ca2+ current (see Fig. 12). Thus, the delayed rise of [Ca2+]i in somata compared with terminals is not due to exclusive location of Ca2+ channels in the terminal; to study the mechanisms of such temporal properties of the rise and fall of [Ca2+]i, better temporal control of depolarization than can be achieved with bath exchange will be required. Whole-cell calcium current Characterization and pharmacology of calcium current The results of the Fura-2 experiments suggest that bipolar cell somata and synaptic terminals contain L-type Ca2+ channels. To examine the Ca2+ current directly, whole-cell Ca2+ currents were recorded under voltage clamp from freshly isolated bipolar cells, using pipette solution containing Cs+ and TEA (see Methods) to block outward currents. A slowly inactivating, inward current was elicited in cells held at -60 mV and stepped to potentials positive to -40 mV. Figure 6A shows the inward current evoked by a 150 ms voltage step to -10 mV in an isolated bipolar neuron. The inward current showed virtually no inactivation throughout the duration of the test pulse. The current-voltage relation of the inward current was bellshaped (Fig. 6B), reaching a maximum amplitude (commonly 100-150 pA with the standard [Ca2+]o of 2-5 mM) at around -20 mV. Even in the absence of K+ channel blockers, peak inward currents of 15-50 pA could be evoked. To prevent progressive disappearance of the inward current during an experiment, interpulse intervals of > 6 s were required. Therefore, in order to speed acquisition of current-voltage

245

CALCIUM IN RETINAL BIPOLAR CELLS

relations, the voltage dependence of current activation was measured in some experiments with linear voltage ramps (50-100 mV/s) rather than with voltage steps. In cells in which both methods were employed (n = 11), the two gave identical results, as illustrated in Fig. 6 C. A

+50

B

mV

100 pA 50 ms

-10 mv

-60m

c

-60

-40

-20

+40

mv '-

-

-80

Fig. 6. Bipolar cell Ca2+ current recorded under whole-cell voltage clamp. A, inward current in response to a 150 ms voltage step from -60 to -10 mV. Trace shows raw membrane current, without subtraction of capacitative or leak current. B, current-voltage relation of the total whole-cell current from the same neuron as in A. Data were obtained from 150 ms pulses to the indicated voltages from a holding potential of -60 mV, and are not leak subtracted. C, comparison of current-voltage relations of the total whole-cell current measured with a linear voltage ramp at 100 mV/s (line) or with a family of 70 ms voltage pulses (A) from a holding potential of -60 mV. Leak and capacitative currents were not subtracted. Different neuron from A and B. Pipette solution in A, B and C was caesium gluconate.

On the basis of its similarity to Ca2+ currents in other cells, we suspected that the inward current in Fig. 6 was a Ca 2+ current. To test this, we examined the effects on the inward current of removing external Ca2+, adding external Cd2+, and adding external Ba2+. When isolated bipolar neurons were superfused with solution containing no added Ca2 , the inward current was dramatically attenuated, as illustrated in Fig. 7A and B. We showed earlier that depolarization-induced Ca2+ influx in bipolar neurons was blocked by external Cd21 (Fig. 2B), and as shown in Fig. 7 C, the addition of 200 JLM-cadmium to the normal extracellular goldfish Ringer

246

R. HEIDELBERGER AND G. MATTHEWS

solution abolished the inward current. Finally, adding 20 mM-Ba2+ to the external solution increased the peak inward current 7-fold on average (n = 4; data not shown). All of these effects were rapidly reversible. Thus, the inward current has the characteristics expected of a Ca2+ current. A

oCa2+

_ _ __&AV.

2.5 mM-Ca2+

100 pA

50 ms

-10 mV

-60 mV

B -50 mV

C -80

-60

-40

pA -+50 +20 -20

1_

Cd

+40 mV

-50

Normal -

-100

-

-150

Fig. 7. Effects of external Ca2' and Cd2+ on voltage-activated inward current. A, in nominally zero external Ca2+ (i.e. no added Ca2+, no EGTA), there is little inward current (top trace). In contrast, in 2-5 mM-Ca2+ Ringer solution (bottom trace) there is more than 100 pA of inward current. Pulse protocol is shown at the bottom. Traces have capacitative and leak currents subtracted using a p/6 protocol. B, current-voltage relation of inward current in the absence of extracellular calcium (0), and in 2-5 mM-Ca2+ Ringer solution (@). Same cell as A. Data were obtained from 150 ms steps to the indicated voltages from -60 mV. C, current-voltage relation generated by linear voltage ramp at 100 mV/s in the presence and absence of 200 4aM-external Cd2+. 200 ,uM-Cd2+ blocked the inward current. Different neuron from A and B. Intracellular pipette solution in A and B: CsCl. Intracellular pipette solution in C: caesium gluconate.

CALCIUM IN RETIVAL BIPOLAR CELLS

247

Fura-2 measurements (Fig. 3) showed that depolarization-induced Ca21 influx was sensitive to dihydropyridine Ca21 channel drugs. Therefore, the actions of dihydropyridines were tested on whole-cell Ca2+ current. Figure 8A shows the Ca2+ current from a bipolar cell stepped from -60 to -20 mV before, during and after -20 mV

A -60 mV

20 pA| 50 ms B

-60

20 ,A +40 mV gr-1

Nitrendipine A Initial A Final Control O

Initial

*

Final

u ' -60 Effect 8. of Fig. nitrendipine on Ca2+ current. A, the effect of 1 /tM-nitrendipine on current elicited by 150 ms voltage step from -60 to -20 mV. Superimposed traces show membrane current before, during and after nitrendipine application. Capacitative and leak currents were subtracted using a p/6 protocol. Nitrendipine block was more complete near the end of the pulse than at the start. B, effect of nitrendipine on the Ca2+ current-voltage relation. Same neuron as in A. Circles are control data after wash-out of nitrendipine, and triangles are data in the presence of 1 /tm-nitrendipine. Open symbols represent measurements of current amplitude within the first 20 ms of the pulse, and filled symbols are measurements taken within the last 20 ms of the pulse. Nitrendipine was applied by local superfusion via application pipette. Intracellular pipette solution: CsCl.

application of 1 ltM-nitrendipine. Nitrendipine blocked the Ca2+ current in a reversible, time-dependent manner (Fig. 8A; see also Fig. lOB), which is characteristic of dihydropyridines (Bean, 1984; Rane, Holz & Dunlap, 1987). The

R. HEIDELBERGER AND G. MATTHEWS

248

magnitude and time dependence of the block can also be seen in the current-voltage relations of Fig. 8B, in which open symbols represent current measured near the start of the voltage pulse, and filled symbols represent current measured at the end of the 150 ms pulse. Nitrendipine reduced the Ca21 current across the entire voltage range -30 mV

A

+50 pA

B

-60 mV -50

-25

+25

Control +50 mV 0.1 uM-Bay K

50 pA 20 ms

-1 -50-

20 ms

Fig. 9. Effect of Bay K 8644 on Ca2+ current. A, Bay K 8644 (0 1 ,SM) reversibly increased the Ca2+ current generated by a voltage pulse to -30 mV, but had little effect on the Ca2+ current generated by a pulse to 0 mV in the same cell. The control traces in both cases are the Ca2+ current before application and after wash-out of Bay K 8644. B, current-voltage relation before (O), during (@) and after (A) application of Bay K 8644. In addition to increasing the peak inward current, Bay K 8644 shifted the peak of the Ca2+ current to a more negative potential. Intracellular pipette solution: caesium gluconate.

Of the Ca2+ current. Additionally, nitrendipine block was more pronounced near the end of the 150 ms pulse than at the start of the pulse. In twenty-one similar experiments, the Ca2+ current in the presence of 1 JtM-nitrendipine at the end of the pulse was 0*16 + 0*02 (mean ± S.E.M.) of the average control Ca2+ current. The average control Ca2+ current was calculated from values taken before addition of the drug, and after wash-out to take into account the possibility of Ca2+ current run-down. Note that the magnitude of the reduction in Ca22 current by r 1 sM-nitrendipine is similar to the reduction of Ca2minflux by 1 elM-nitrendipine in the Fura-2 studies (Fig. 4). As with Ca2+ influx elicited by high K+, 01otM-Bay K 8644 applied to synaptic terminals of bipolar neurons was found to potentiate the whole-cell Ca2± current. The potentiation was greatest for moderate depolarizations and became progressively less for larger depolarizations. In the example shown in Fig. 9A, 01 /iM-Bay K 8644 approximately doubled the Ca2A current at -30 mV, but had little effect on the current at 0 mV. The voltage dependence of the potentiation by Bay K 8644 can be seen in the current-voltage relation of Fig. 9B. Bay K 8644 increased the peak current by about 30%, and shifted the peak-current potential to a more negative value.

CALCIUM IN RETINAL BIPOLAR CELLS

249

Similar effects of Bay K 8644 were observed in six other cells. Bay K 8644 has previously been reported to shift the current-voltage relation for Ca2+ current to hyperpolarized potentials (Hess et al. 1984; Kostyuk, Shuba & Savehenko, 1988), in a manner like that shown here.

A

B

-20 mV

-20 mV -60 mV

-60 mV

Before conotoxin

Before drugs 10 pM-conotoxin

Nitrendipine l,04

Nitrendipine

+

conotoxin

After conotoxin

After drugs 100 pA

100 pA

50 ms

50 ms

Fig. 10. Effect of wo-conotoxin on Ca2+ current in bipolar neurons. A, 10 ,M-wO-conotoxin had no effect on the Ca2" current. Traces shown are before (upper), during (middle) and after (lower) local application of w-conotoxin. B, 10 ,tM-w-conotoxin was unable to reduce the nitrendipine-insensitive component of the Ca2+ current. Top trace is control Ca2+ current prior to drug application. The middle two traces are superimposed records showing the effect of 1 /uM-nitrendipine alone compared with nitrendipine in combination with 10 ymM-0-conotoxin. The lower trace shows current after wash-out of both drugs. Different neuron from A. Intracellular pipette solution: CsCl.

Based on the pharmacology of the Ca2+ influx and whole-cell Ca2` current, it can be concluded that synaptic terminals and somata of bipolar cells contain L-type Ca2` channels. However, because there was still a small amount of inward current (16 %) not blocked by nitrendipine, we checked to see whether w)-conotoxin could block this residual current. &o-Conotoxin has been reported to cause persistent block of N-type, dihydropyridine-insensitive Ca2+ currents in other neurons (Reynolds, Wagner, Thayer, Olivera & Miller, 1986; McClesky, Fox, Feldman, Cruz, Olivera & Tsien, 1987). Figure IOA shows that 10 tm-w-conotoxin had no effect on the Ca2` current in goldfish bipolar neurons; in ten cells, the inward current in response to a step from -60 to -20 mV in the presence of wo-conotoxin was 97+4% of the control (mean+s.E.M.). We also tested whether the residual current unblocked by 1,uM-

250

R. HEIDELBERGER AND G. MATTHEWS

nitrendipine was sensitive to w-conotoxin. In Fig. 10 B, the reduction in the wholecell Ca2+ current caused by 1 /tM-nitrendipine is compared to the reduction caused by 10 /uM-&o-conotoxin co-applied with 1 /tM-nitrendipine. Also shown is the control Ca2+ current before and after drug treatment. In five such experiments, w-conotoxin was -20 mV -60 mV -90 mv

100 pA 20 ms Fig. 11. Effect of holding potential on Ca2+ current. Superimposed traces show Ca2+ current elicited by pulses to -20 mV from holding potentials of -60 and -90 mV. Intracellular pipette solution: caesium gluconate.

unable to reduce the nitrendipine-insensitive component of the Ca2+ current. These data suggest that rather than representing a rapidly inactivating component of Ca2+ current attributable to an N-type Ca2+ current, the remaining Ca2+ current seen in the presence of 1 jtM-nitrendipine is due to incomplete block by nitrendipine of Ltype Ca2+ current, as has been reported by others (Bean, 1984; Fox et al. 1987, Yawo, 1990). The activation of the Ca2+ current from a holding potential of -60 mV, its slow inactivation, and its pharmacology are suggestive of a high-threshold, voltageactivated L-type Ca2+ current. However, the pulse protocol used in our experiments would not be expected to reveal a transient component of Ca2+ current due to the relatively depolarized holding potential (-60 mV). Indeed, a holding potential of -60 mV has been used to isolate L-type Ca2+ currents from N- and T-types (Fox et al. 1987). Therefore, to remove inactivation of any transient Ca2+ currents, either the holding potential as changed to -90 mV or hyperpolarizing pre-pulses to -90 mV for 200 ms were used. Shown in Fig. 11 are superimposed responses to pulses from holding potentials of -90 and -60 mV to -20 mV. There was no difference between the Ca2+ current evoked from -90 and -60 mV, indicating the lack of any transient component. Figure 11 also shows that there was apparently no inactivation of the Ltype current at -60 mV. In eight such experiments with a hyperpolarizing pre-pulse and three experiments in which holding potential was changed, no transient currents were revealed, suggesting that there is only a single type of slowly inactivating Ca2+ channel in bipolar neurons. The possibility that transient Ca2+ channels have been selectively eliminated by papain digestion cannot at present be eliminated; however,

CALCIUM IN RETINAL BIPOLAR CELLS 251 transient Ca21 channels in mammalian bipolar neurons are spared by papain (Kaneko, Pinto & Tachibana, 1989).

Site of origin of Ca2+ current Whole-cell Ca2+ currents measured from bipolar cells reflect contributions of Ca2+ current from both the synaptic terminal and soma. To estimate the contributions from the soma and synaptic terminal, bipolar neurons were bathed in Ringer solution containing 0-2 mM-Ca2+, and 2-5 mM-Ca2+ Ringer solution was locally superfused on either the soma or synaptic terminal with a small-bore pressure application pipette that was oriented perpendicular to the long axis of the cell, as shown in Fig. 12B (application pipette 1). To estimate the total Ca2+ current of the entire cell in 2-5 mM [Ca2+]0, 2-5 mM-Ca2+ Ringer solution was applied across the entire cell by a second, larger-bore application pipette oriented parallel to the long axis of the cell, so that its flow superfused the entire cell (application pipette 2, Fig. 12 B). In the example shown in Fig. 12A, application of Ca2+ locally to the terminal via pipette i added 44 pA of current; with pipette 1 pointing at the soma, Ca2+ current increased by 21 pA. When the entire cell was bathed in 2-5 mM-Ca2+ Ringer solution using the second, large-bore pipette (pipette 2), the Ca2+ current increased by 60 pA. From these numbers, we estimated that in this neuron, the terminal accounted for 73 % (44/60) of the total whole-cell current, while the soma/dendrites accounted for 34 % (21/60). If the local application via pipette 1 were perfectly restricted to the application site, the percentages from the soma and terminal should add up to 100 %. Their sum was 107 %, and the deviation from 100 % suggests a small degree of spread of Ca2+ from the site of application to other parts of the cell. In twelve such experiments, the Ca2+ current generated in the terminal contributed 64+3 % (mean + S.E.M.) of the total whole-cell current, while the soma contributed 40 + 2 % (mean +S.E.M.). The fact that the majority of whole-cell Ca2+ current originates in the synaptic terminal offers a simple explanation for the fact that changes in [Ca2+]i in the Fura-2 experiments were larger and faster in the terminal (Fig. 5). However, roles for other factors, such as differences in Ca2+ buffering or Ca2+ efflux between soma and terminal, cannot be ruled out. Another estimate for the contribution of Ca2+ current from the terminal was obtained by bathing cells in 2-5 mM-Ca2+ Ringer solution and determining the amount of current blocked by application of Cd2+ to the terminal via pipette 1, compared with the amount of current blocked by superfusing the entire cell via pipette 2. An example of one such experiment using Cd2+ is shown in Fig. 12 C. In this cell, 100 jtM-Cd2+ applied to the terminal via pipette 1 blocked more than half the Ca2+ current, while application via pipette 2 eliminated all the current. In five similar experiments, Cd2+ applied to the terminal blocked 74-2 + 7 % (mean + S.E.M.) of the whole-cell Ca2+ current; this percentage is similar to the 64 % estimated contribution from the terminal obtained by local application of 2-5 mm [Ca2+]O, above. In the experiments employing local application of 2-5 mM-Ca2 , it was notable that there were no obvious differences in the waveform of the Ca2+ current added by the local Ca2+ application, regardless of whether the patch pipette was placed on the soma or terminal and regardless of whether the application was directed at the recording site or at the opposite pole of the cell. In addition, we obtained the same

R. HEIDELBERGER AND G. MA TTHEWS

252

estimate for the percentage contribution of terminal Ca2+ current to the whole-cell current whether the recording site was the terminal (62 ± 4 %; n = 7) or the soma (67 + 5%; n = 5). These results suggest that the axon connecting the synaptic terminal and the soma presents no significant electrical barrier to the flow of current. -20 mV

A

B

,11-,1C

-60 mV

IApplication -A.

0.2 mM-Ca2+

pipette 1 Ca2+

Patch pipette

2.5 mM-Ca2+ on terminal

:EE~

Application pipette 2 Ca2+

~~ ~0.2

mm-Ca2+1 -20 mV

C 2 5 mM-Ca2+ on soma

--60

0.2 mM-Ca2+ 2.5 mM-Ca2+ on entire cell

50 pA 20 ms

mV -.& --

Cd2+ on entire cell Cd2+

on

terminal

Control

50 pA 20 ms

Fig. 12. Effect of local application of Ca2" and Cd2+ on whole-cell Ca2+ current. A, results from an experiment in which a cell was bathed in Ringer solution containing 0-2 mM-Ca2+, and 2-5 mM-Ca2+ Ringer solution was applied to the synaptic terminal (upper traces), to the soma (middle traces) or to all parts of the cell (lower traces). The traces labelled 0-2 mM-Ca2" show control currents in 0-2 mM-Ca2' Ringer solution before and after local application of 2 5 mM-Ca2+. B, the arrangement of the cell, recording pipette, and the two application pipettes for the experiment of A. For the upper and middle traces in A, Ca2+ was applied via application pipette 1, which was oriented perpendicular to the long axis of the cell and was positioned to point at the soma (upper position) or the synaptic terminal (lower position). For the lower trace in A, 2-5 mM-Ca2` was applied via largerbore application pipette 2, which was oriented to flood the area with 2-5 mM-Ca2+ Ringer solution. Although both application pipettes are shown in the diagram, only one pipette at a time was actually in position at the cell. Application pipettes were normally raised out of the bath and were lowered into position to apply Ca2 . In this experiment, the patch pipette was placed on the synaptic terminal, but in other experiments, it was sometimes placed on the soma. C, results of an experiment in which the cell was bathed in 2-5 mMCa2+ Ringer solution, and 100 /tM-Cd2+ (in 2-5 mM-Ca2' Ringer solution) was applied to the synaptic terminal (middle trace) or to the entire cell (upper trace). Control trace (lower traces) show responses before and after Cd2` application. The arrangement of application pipettes was as in B, with Cd2` applied to the terminal via smaller-bore application pipette 1 and to the entire cell via larger-bore application pipette 2. Different cell from A. Intracellular pipette solution: caesium gluconate.

CALCIUM IN RETINAL BIPOLAR CELLS 253 This is not surprising for a non-spiking neuron that normally depends on electrotonic spread for the transmission of signals from the dendrites to the synaptic terminal. DISCUSSION

In this series of experiments, we have characterized depolarization-induced increases in [Ca2+]i and whole-cell calcium currents in the synaptic terminal of a tonically active neuron from the vertebrate central nervous system. Our data show that large-terminal bipolar cells of the goldfish retina have a single type of Ca2+ current that is present in both the synaptic terminal and soma. This Ca2+ current is voltage gated, slowly inactivating and dihydropyridine sensitive. The pharmacology, kinetics, and activation from a holding potential of -60 mV are all characteristic of L-type Ca2+ channels (Nowycky et al. 1985a; Fox et al. 1987). Although the Ca2+ current originating in somata and terminals is qualitatively the same, the Ca2+ current density is greater in synaptic terminals. The identification of only one type of Ca2+ current in bipolar cells, its prevalence in the synaptic terminal, and the physiologically appropriate kinetics and activation range of this Ca2+ current, all point towards an important role for L-type Ca2+ channels in synaptic release in bipolar neurons. The Ca2+ current in the bipolar cell activates at a more negative potential than Ltype channels described in other neurons, such as dorsal root ganglion cells (Tsien, Lipscombe, Madison, Bley & Fox, 1988). This negative shift places the activation range of the Ca2+ current within the physiological operating range of on-type bipolar neurons of goldfish retina. In the dark, the resting potential of these cells in situ is -40 to -50 mV (Saito, Kondo & Toyoda, 1979; Kaneko, Famiglietti & Tachibana, 1979), which is near the level at which Ca2+ current began to activate in our voltage-clamp experiments. In response to light, large-terminal on-bipolar neurons depolarize by 10-20 mV (Saito, Kujiraoka & Yonaha, 1983), bringing the potential into the range in which the Ca2+ current increased rapidly with depolarization (e.g. Fig. 6). Thus, illumination would be expected to activate voltage-gated Ca2+ current, resulting in Ca2+ influx and elevation of intracellular calcium, particularly in synaptic terminals where the density of Ca2+ current is highest. In contrast to the sustained response to illumination common in bipolar neurons, transient light responses are typical in ganglion cells and most types of amacrine neurons. In mouse, retinal bipolar cells have been reported to have solely transient Ca2+ current (Kaneko et al. 1989), and in salamander, bipolar cells have a combination of transient and slowly inactivating Ca2+ currents (Maguire, Maple, Lukasiewicz & Werblin, 1989). In these species, then, the transient Ca2+ current may control transmitter release from bipolar cells and thus contribute to the conversion of sustained to transient light responses. In goldfish, however, our results show that sustained depolarization gives rise to sustained elevation of [Ca2+]i in the synaptic terminal and that no transient Ca2+ current could be detected. These results were obtained from cells isolated after enzymatic digestion with papain; but if the same is true in intact gold fish retina, then the temporal properties of the voltage-activated Ca21 channels in bipolar cell synaptic terminals do not contribute to the conversion of sustained to transient light responses. This suggests that processing of photic

254

R. HEIDELBERGER AND G. MATTHEWS

signals in goldfish retina to enhance responses to onset and offset of illumination will depend on other aspects of retinal circuitry, such as the extensive feedback synaptic input from amacrine cells (Marc, Stell, Bok & Lam, 1978; Yazulla, Studholme & Wu, 1987). In voltage-clamp studies of morphologically complex neurons, one important consideration is whether the cells are isopotential under voltage clamp. Because bipolar neurons rely on electrotonic transmission of signals, it would be expected a priori that this would be less of a problem in the present experiments and that the cells are specialized to be nearly isopotential. We have three lines of evidence that support this expectation. First, in the experiments using local application of Ca2" to estimate the contribution of the terminal to total whole-cell Ca2" current (Fig. 12), the estimated contribution from the terminal did not depend on the position of the recording electrode; thus, the same estimate of the terminal contribution was obtained whether the patch pipette was on the terminal or the soma. This suggests that the recording pipette effectively had equal electrical access to all parts of the cell. Second, the waveform of the Ca 2± current was the same whether local application of Ca2+ or of Ca2+ channel blockers was made at the recording site or at a distance from it. Third, the resting input conductance of bipolar neurons in our experiments was low (typically 300-500 pS), so that membrane resistance is high (20-40 kQ cm2). Assuming that the conductance is spread equally over the cell surface, the calculated length constant of a bipolar cell axon with a diameter of 1 ,tm would be approximately 700 ,tm (rm = 64 MQ cm; ri = 12-7 GQ cm-' for axoplasmic resistivity of 100 Q cm). The length of the axon (20-30 ,tm) is a small fraction of this length constant. If channels are preferentially excluded from the membrane of the axon, the effective length constant may be even longer. How do the Ca2+ channels of the large-terminal bipolar neuron from goldfish retina compare with those of other neuronal preparations that permit direct physiological characterization of Ca21 influx and Ca2+ currents in individual synaptic terminals? High-threshold, slowly inactivating presynaptic Ca2+ currents were found in the squid giant synapse (Smith & Augustine, 1988; Charleton & Augustine, 1990), chick ciliary ganglion (Yawo, 1990) and neurohypophysis (Lemos & Nowycky, 1989). The chick ciliary ganglion and neurohypophysial terminals additionally contained highthreshold, rapidly inactivating presynaptic Ca2+ currents (Lemos & Nowycky, 1989; Stanley & Atrakchi, 1990; Yawo, 1990). It is interesting to note that the Ca2+ current in the synaptic terminal of the bipolar cell, a tonically active CNS interneuron, is very similar to the presynaptic Ca21 current of squid giant synapse in the kinetics and voltage range of activation, and to the L-type presynaptic Ca2+ current described in chick ciliary ganglion and neurohypophysial terminals. The common theme of a slowly inactivating, high-threshold presynaptic Ca2+ current in a variety of terminals suggests that this current could be important in regulating synaptic release. Unique in this group of synaptic terminals, the terminal of the bipolar neuron receives well-characterized synaptic inputs; because of this and because it has only a single type of Ca2+ current, the bipolar neuron is well suited for studies of neurotransmitter modulation of presynaptic calcium current. This work was supported by NIH grant EY03821 (G.M.) and NIMH training grant MH18010

(R.H.).

CALCIUM IN RETINAL BIPOLAR CELLS

255

REFERENCES

AUGUSTINE, G. J., CHARLETON, M. P. & SMITH, S. J. (1985). Calcium entry and transmitter release at voltage-clamped nerve terminals of squid. Journal of Physiology 367, 163-181. BEAN, B. (1984). Nitrendipine block of cardiac calcium channels: high affinity binding to the inactivated state. Proceedings of the National Academy of Sciences of the USA 81, 6388-6392. CHARLETON, M. P. & AUGUSTINE, G. J. (1990). Classification of presynaptic calcium currents at the squid giant synapse: neither T-, L-, nor N-type, Brain Research 525, 133-139. DEBER, C. M., TOM-KUN, J., MACK, E. & GRINSTEIN, S. (1985). Bromo-A23187: a nonfluorescent calcium ionophore for use with fluorescent probes. Biochemistry 146, 349-352. Fox, A. P., NOWYCKY, M. C. & TSIEN, R. W. (1987). Kinetic and pharmacological properties distinguishing three types of calcium currents in chick sensory neurons. Journal of Physiology 394, 149-172. GRYNKIEWICZ, G., POENIE, M. & TSIEN, R. Y. (1985). A new generation of Ca2+ indicators with greatly improved fluorescence properties. Journal of Biological Chemistry 260, 3440-3450. HEIDELBERGER, R. & MATTHEWS, G. (1990). Depolarization-induced calcium influx in goldfish bipolar cells. Investigative Ophthalmology and Visual Science 31, 389 (abstract). HESKETH, T. R., SMITH, G. A., MOORE, J. P., TAYLOR, M. V. & METCALFE, J. C. (1983). Free cytoplasmic calcium concentration and the mitogenic stimulation of lymphocytes. Journal of Biological Chemistry 258, 4876-4882. HESS, P., LANSMAN, J. B. & TSIEN, R. W. (1984). Different modes of Ca channel gating favoured by dihyropyridine Ca agonists and antagonists. Nature 311, 538-544. ISHIDA, A. T., STELL, W. K. & LIGHTFOOT, D. A. (1980). Rod and cone inputs to bipolar cells in goldfish retina. Journal of Comparative Neurology 191, 315-335. KANEKO, A., FAMIGLIETTI, E. V. & TACHIBANA, M. (1979). Physiological and morphological identification of signal pathways in the carp retina. In Neurobiology of Chemical Transmission, ed. OTSUDA, M. & HALL, Z., pp. 235-251. Wiley, New York. KANEKO, A., PINTO, L. H. & TACHIBANA, M. (1989). Transient calcium current of retinal bipolar cells of the mouse. Journal of Physiology 410, 613-629. KANEKO, A. & TACHIBANA, M. (1985). A voltage-clamp analysis of membrane currents in solitary bipolar cells dissociated from Carassius auratus. Journal of Physiology 385, 131-152. KATZ, B., & MILEDI, R. (1967). A study of synaptic transmission in the absence of nerve impulses. Journal of Physiology 192, 407-436. KOSTYUK, P. G., SHUBA, Y. M. & SAVCHENKO, A. M. (1988). Three types of calcium channels in the membrane of mouse sensory neurons. Pflugers Archiv 411, 661-669. LEMOS, J. R. & NOWYCKY, M. C. (1989). Two types of calcium channels coexist in peptide-releasing vertebrate nerve terminals. Neuron 2, 1419-1426. LLINA'S, R., STEINBERG, I. Z. & WALTON, K. (1981). Presynaptic calcium currents in squid giant synapse. Biophysical Journal 33, 289-322. MCCLESKEY, E. W., Fox, A. P., FELDMAN, D. H., CRUZ, L. J., OLIVERA, B. M. & TSIEN, R. W. (1987). w-Conotoxin: direct and persistent blockade of specific types of calcium channels in neurons, but not muscle. Proceedings of the National Academy of Sciences of the USA 84, 4327-4333. MAGUIRE, G., MAPLE, B., LUKASIEWICZ, P. & WERBLIN, F. (1989). y-Aminobutyrate type B receptor modulation of L-type calcium channel current at bipolar terminals in the retina of the tiger salamander. Proceedings of the National Academy of Sciences of the USA 86, 10144-10147. MARC, R. E., STELL, W. K., BOK, D. & LAM, D. M. K. (1978). GABAergic pathways in the goldfish retina. Journal of Comparative Neurology 182, 221-246. MATTHEWS, G. & HEIDELBERGER, R. (1990). Calcium current of synaptic terminals of goldfish retinal bipolar cells. Society for Neuroscience Abstracts 16, 1273. NEHER, E. (1988). The influence of intracellular calcium concentration on degranulation of dialysed mast cells from rat peritoneum. Journal of Physiology 395, 193-214. NEHER, E. (1989). Combined fura-2 and patch clamp measurements in rat peritoneal mast cells. In Neuromuscular Junction, ed. SELLIN, L. C., LIBELIUS, R. & THESLEFF, S., pp. 65-76. Elsevier, Amsterdam. NOWYCKY, M. C., Fox, A. P. & TSIEN, R. W. (1985a). Three types of neuronal calcium channels with different calcium agonist sensitivity. Nature 316, 440-443.

256

R. HEIDELBERGER AND G. MATTHEWS

NOWYCKY, M. C., Fox, A. P. & TSIEN, R. W. (1985b). Long-opening mode of gating of neuronal calcium channels and its promotion by the dihydropyridine calcium agonist Bay K 8644. Proceedings of the National Academy of Sciences of the USA 82, 2178-2182. RANE, S. G., HOLZ, G. G. & DUNLAP, K. (1987). Dihydropyridine inhibition of neuronal calcium current and substance P release. Pfiilgers Archiv 409, 361-366. REYNOLDS, I. J., WAGNER, J. A., THAYER, S. A., OLIVERA, B. M. & MILLER, R. J. (1986). Brain voltage-sensitive calcium channel subtypes differentiated by w-conotoxin fraction GVIA. Proceedings of the National Academy of Sciences of the USA 83, 8804-8807. SAITO, T., KONDO, H. & TOYODA, J. (1979). Ionic mechanisms of two types of ON-center bipolar cells in the carp retina. Journal of General Physiology 73, 73-90. SAITO, T. & KUJIRAOKA, T. (1982). Physiological and morphological identification of two types of on-center bipolar cells in the carp retina. Journal of Comparative Neurology 205, 161-170. SAITO, T., KUJIRAOKA, T. & YONAHA, T. (1983). Connections between photoreceptors and horseradish peroxidase-injected bipolar cells in the carp retina. Vision Research 23, 353-362. SMITH, S. J. & AUGUSTINE, G. J. (1988). Calcium ions, active zones and synaptic transmitter release. Trends in Neurosciences 11, 458-464. STANLEY, E. F. & ATRAKCHI, A. H. (1990). Calcium currents recorded from a vertebrate presynaptic nerve terminal are resistant to the dihydropyridine nifedipine. Proceedings of the National Academy of Sciences of the USA 87, 9683-9687. TACHIBANA, M. (1983). Ionic current of solitary horizontal cells isolated from goldfish retina. Journal of Physiology 345, 329-351. TSIEN, R. W., LIPSCOMBE, D., MADISON, D. V., BLEY, K. R. & Fox, A. P. (1988). Multiple types of neuronal calcium channels and their selective modulation. Trends in Neurosciences 11, 431-438. YAWO, H. (1990). Voltage-activated calcium currents in presynaptic nerve terminals of the chicken ciliary ganglion. Journal of Physiology 428, 199-213. YAZULLA, S., STUDHOLME, K. M. & WU, J.-Y. (1987). GABAergic input to the synaptic terminals of Mb-i bipolar cells in the goldfish retina. Brain Research 411, 400-405.