Analysis of calcium channel properties in cultured leech retzius cells by internal perfusion, voltage-clamp and single-channel recording

Several types of calcium channel with distinctive properties have been identified and characterized in vertebrate and invertebrate neurones (Reuter, 1985; Byerly and Hagiwara, 1988; Tsien et al. 1988). These channels differ with respect to their ion selectivity, kinetics and sensitivity to toxins (Carbone and Lux, 1984b; Fedulova et al. 1985; Fox et al. 1987a,b; Swandulla and Armstrong, 1988). Relatively few studies have been made on the calcium channels that play a part in physiological processes such as transmitter release (Llinas et al. 1981; Augustine et al. 1987).

The aim of the present experiments was to describe divalent cation currents in cultured Retzius cells of the leech, which extend processes to establish chemical synapses. Single Retzius cells survive for days or weeks after isolation from the central nervous system (Ready and Nicholls, 1979). In culture, they maintain their membrane properties. When co-cultured with an appropriate partner they can form chemical synapses that show marked dependence on external Ca²⁺ (Fuchs et al. 1981, 1982; Henderson, 1983; Henderson et al. 1983; Dietzel et al. 1986; Nicholls, 1987). Moreover, optical recordings made with the Ca²⁺ indicator Arsenazo III show that Ca²⁺ entry following action potentials is influenced by the molecular composition of the substratum and is unevenly distributed over the surface of a growing Retzius cell (Ross et al. 1987, 1988).

Although the characteristics of Na⁺ and K⁺ channels have been measured in two-electrode voltage-clamp and with loose patch studies (Stewart et al. 1989b; Bookman et al. 1987; Nicholls and Garcia, 1989; Garcia et al. 1990), surprisingly little information is available about Ca²⁺ currents in leech neurones (see Stewart et al. 1989a,b). Certain technical difficulties arise with conventional microelectrodes. First, the high impedances of the microelectrodes required for penetration limit the speed with which the membrane potential can be displaced and clamped (approx. 1–2 ms). The capacity current artefacts obscure the initial phases of activation and deactivation, precluding analysis of rapid kinetics. A second difficulty is that the isolation of Ca²⁺ currents is difficult in the presence of unblocked, residual K⁺ currents (Stewart et al. 1989a,b). These problems have been largely eliminated in the present experiments by replacing the currentpassing microelectrode with a low-resistance ‘patch’-type pipette that passes large currents to charge the membrane capacitance rapidly (≈100 μs) and that permits internal perfusion. In addition, a tight-seal cell-attached patch-clamp method has been used to examine single Ca²⁺ channel activity. With these modifications it became possible to explore the following questions concerning the voltagedependent calcium channels present in Retzius cells in culture. (1) What are the kinetics of activation, deactivation and inactivation of Ca²⁺ channels? (2) To what extent do Ca²⁺, Ba²⁺ and Sr²⁺ permeate through the divalent cation channels? (3) Can more than one type of calcium channel be identified? (4) What are the properties of individual calcium channels? Preliminary reports of this work have been presented elsewhere (Bookman and Liu, 1986, 1987).

The techniques for the identification, isolation and culture of Retzius cells from the CNS have been described in detail elsewhere (Fuchs et al. 1981; Dietzel et al. 1986). In brief, individual Retzius cells were removed after enzyme treatment of ganglia with collagenase-dispase (2 mg ml⁻¹), and plated in microwells coated with poly-L-lysine in L-15 culture medium containing gentamycin (100 μg ml⁻¹) and foetal calf serum (2%). Measurements were made at room temperature (20–22°C) in appropriate solutions (see below). The perfusion system permitted changes in the bathing fluid to be effected within 5–60 s. Cells were examined 1–15 days after removal from the ganglion. For the formation of high-resistance seals between either the current-passing electrode or the patch pipette, it was essential that the surface of the neurone should be clean of debris.

Current-passing electrodes were double-pulled from haematocrit tubing (Clay Adams 1021), fire-polished and insulated with Sylgard (Hamill et al. 1981). For these electrodes, tip openings ranged from 2 to 8 μm and had resistances of 0.5–1.5 MQ in our standard external solution. Voltage microelectrodes were pulled from filament glass to have resistances of 10–30 MΩ when filled with 3 mol l⁻¹ KC1. Miniature Ag/AgCl pellets (E-255, In Vivo Metrics, Healdsburg, California) were used for the patch electrodes, the microelectrodes and for the reference electrode, which was connected to the bath by an agar bridge.

Frequency response was maximized and capacitative artefacts were reduced by using low-impedance patch electrodes for passing current. The intracellular fluid was perfused in experiments to isolate Ca²⁺ current from Na⁺ and K⁺ currents (Kostyuk and Krishtal, 1977; Lee et al. 1978; Hagiwara and Byerly, 1981; Byerly and Hagiwara, 1982). The current-passing pipette was first sealed onto the cell surface with a seal resistance of 1GΩ or more. As suction ruptured the membrane, there was a concomitant decrease of the tip resistance to about 50 MΩ, representing the parallel combination of the seal and membrane resistances. The resting potential was stabilized with hyperpolarizing current and the cell was then impaled with the high-resistance voltage-recording microelectrode.

The performance of this clamp is illustrated in Fig. 1. The membrane potential reached more than 90% of its new value in about 100 μs. Capacitative currents decayed with a time constant of approximately 45 μs. These speeds were achieved by using a conventional two-electrode voltage-clamp amplifier (Almost Perfect Electronics, Basel, Switzerland). The I–V converter headstage of a patch-clamp amplifier did not charge the whole-cell capacitance sufficiently fast for rapid activation or deactivation kinetics to be measured.

Electrodes were made from hard Pyrex glass H15–10 (Jencons Scientific, England) and had resistances of 2–6 MΩ. Channel activity was recorded in cell-attached patches with records filtered at 0.6–2 kHz and corrected for leak and capacity currents by adding scale-averaged records from 16 or 32 hyperpolarizing pulses with no channel activity.

The on-line data acquisition system was designed by one of us (RJB) in collaboration with C. M. Armstrong and D. R. Matteson. High-speed transfer of parallel data between the experimental interface and the computer was accomplished with a direct memory access parallel interface card (DRV11-WA, Digital Equipment Corp.) capable of transferring one 16-bit word every 2 μs. The experimental interface consisted of an optically isolated 100 kHz, 14-bit A/D converter, a 14-bit D/A converter to make voltage command pulses, D/A converters to display sweeps on an oscilloscope and associated logic to handle data transfers.

The signal proportional to membrane current was filtered at a corner frequency appropriate for the sampling rate to prevent aliasing (LPF 902, eight-pole Bessel low-pass filter, Frequency Devices, MA, USA). The sampling rates ranged from a to 100 kHz. Linear leakage currents and capacitative currents were subtracted by the computer using the P/n method of Bezanilla and Armstrong (1977). Here, the currents recorded during four hyperpolarizing pulses scaled to one-quarter of the amplitude of the test pulse were added to the record of current from the depolarization. These hyperpolarizing ‘subtraction’ pulses were applied from the same holding potential as the test pulse. In this way, records of membrane current were leak- and capacity-corrected and signal-averaged (2–10 sweeps). Data acquired on-line were stored on disc.

For current-voltage relationships, peak measurements represent the average of five points around the peak while isochronal measurements were from a single point. Time constants and single exponential fitting of current records were obtained by first performing a logarithmic transformation of the data and then calculating a weighted linear regression. Double exponential fits were verified with an implementation of the Levenberg–Marquardt method of least-squares fitting (Wavemetrics, Lake Oswego, Oregon, USA).

The standard internal solution contained (in mmol l⁻¹): caesium glutamate, 100; IV-methylglucamine fluoride, 25; TEA-CI, 25; EGTA, 2; Hepes-TEA-OH, 20. The external solution was (in mmol l⁻¹): NaCl, 140; CaCl₂,5; MgCl₂, 5; KC1, 2; Hepes-NaOH, 20. ‘75BaCl₂’ solution contained (in mmol l⁻¹): BaCl₂, 75; TEACI, 50; MgCl₂, 2; 4-aminopyridine, 5; Hepes-TEA-OH, 20. Permeation of different divalent cations was tested with 25 mmol l⁻¹ concentrations of Ca²⁺, Sr²⁺ or Ba²⁺ in 50 mmol l⁻¹ TEA-Cl, 20 mmol l⁻¹ Hepes-TEA-OH and 75 mmol l⁻¹ N-methylglucamine chloride. All solutions were adjusted to a final pH of 7.4 and an osmolarity of 350 mosmol l⁻¹.

A serious limitation in the measurement of macroscopic calcium channel currents is contamination by currents flowing through other channels or leak pathways (Hagiwara and Byerly, 1981; Byerly and Hagiwara, 1982). With internal perfusion through the current electrode, Retzius cell bodies could be perfused with the standard internal solution containing Cs⁺ and TEA⁺. The exchange of intracellular fluids was complete in about 2 min. Fig. IB shows the total current of a Retzius cell recorded during a 50 ms pulse to −10 mV while the cell was perfused with a solution containing 120 mmol l⁻¹ KC1 and with an external solution containing 140 mmol l⁻¹ NaCl without blockers. The early inward current carried by Na⁺ was followed by an outward K⁺ current (compare Stewart et al. 1989b).

Fig. 1C shows recording from a different cell perfused with intracellular solution containing 100 mmol l⁻¹ caesium glutamate and 25 mmol l⁻¹ TEA-CI. By the time that the upper trace in Fig. 1C was recorded (about 1.5 min after breaking in) most of the K⁺ current had already been blocked. By 2min, net outward current had been eliminated.

Single calcium channel currents were recorded from cell-attached patches in a solution containing BaCl₂ and TEA⁺ (75 BaCl₂ solution). Current recorded from a patch on the initial segment or ‘stump’ of a Retzius cell in culture for 5 days is shown in Fig. 2. Depolarization by 30 mV caused the appearance of brief events with a duration of 2–20 ms and about IpA in amplitude. During 50 mV depolarizations, the currents increased in frequency and decreased in amplitude to about 0.8 pA. Channels permeable to Ba²⁺ with similar characteristics were observed in 12 patches. Detailed analyses of Ba²⁺ single-channel currents were not possible for three reasons. First, it was difficult to establish the seal resistances of 10-100 GQ which were required for recording small single Ca²⁺ channel currents. Second, even when tight seals were established, most of the patches showed no channel activity in the presence of Ba²⁺. The 12 successful patches were the result of literally hundreds of trials. Third, it was not possible to test decisively whether currents such as those shown in Fig. 2 were due to more than one channel in the patch (see Discussion). Accordingly, the properties of calcium channels were studied by measuring macroscopic currents recorded with the hybrid voltageclamp as described in the following sections.

Fig. 3A shows a slowly activating inward current and a large tail current at the end of a depolarizing pulse to +10 mV from a holding potential of −60 mV, recorded in external solution containing 75 mmol l⁻¹ Ba²⁺ and 0 mmol l⁻¹ Na⁺. Both inward and tail currents were blocked by adding 200 μmol l⁻¹ CdCl₂ to the bathing fluid (Fig. 3B), indicating that the flow of ions was through calcium channels. A small initial transient outward current and a minor component of the tail current persisted in the presence of Cd²⁺ (see below). Dihydropyridines such as (−)202-791 or (+)202-791 (Sandoz, Switzerland; Kongsamut et al. 1985) at concentrations of 20 μmol l⁻¹ did not block or activate Ba²⁺ currents.

The voltage dependence of Ba²⁺ currents is plotted in Fig. 4. Inward Ba²⁺ currents became apparent with depolarizations to −25 mV from a holding potential of −60 mV. With larger depolarizations up to about +15 mV the inward currents became larger. Depolarizations beyond +15 mV led to smaller isochronal currents measured at 10ms. As the driving force diminished, the extrapolated current reversed at about +50 mV (with 75 mmol l⁻¹ Ba²⁺ outside and 100 mmol l⁻¹ Cs⁺ inside).

The selectivity of Retzius cell calcium channels was measured with 25 mmol l⁻¹ Sr²⁺ or Ba²⁺ in the bathing fluid. The current-voltage relationships of Fig. 4, all recorded from a single Retzius cell, show that Sr²⁺ currents were larger than those for Ca²⁺ or Ba²⁺. In most experiments to be described Ba²⁺ was used as the charge carrier because it blocked K⁺ channels (thereby improving the isolation of the current) and showed less inactivation (see below) (Standen and Stanfield, 1978; Armstrong and Taylor, 1980).

The conductance-voltage (G–V) relationship for Ca²⁺ channels (Fig. 5) was calculated from the magnitude of the instantaneous current change at the end of 20ms pulses upon return to the holding potential. This relationship is therefore a measure of those channels that were open after a depolarization of 20 ms duration. Ba²⁺ entry into the cell, estimated by integrating pulse and tail currents, showed that charge entered the cell mainly during the depolarization, with far less entry during the tail. The peak of the charge entry occurred at a pulse potential of + 10 mV.

The early portion of Ba²⁺ current activation was obscured by a fast transient outward current. Nevertheless, the activation kinetics appeared sigmoidal. The entire time course could not be fitted by a single exponential since the rise developed only after an initial delay (Fig. 6A). The activation kinetics was voltage-dependent; the rise times were fastest at +25 mV with a r value of 2.4 ms. Activation kinetics measured with Sr²⁺ or Ca²⁺ as the charge carrier was similar to that shown for Ba²⁺ in Fig. 6B.

The kinetics of channel closing or deactivation was estimated by measuring tail currents at the end of a depolarizing pulse. Deactivation, like activation, was voltage-dependent (Fig. 7). The decfine of Ba²⁺ tail currents was best fitted with two exponentials (Fig. 7A). The later portion of the tail was fitted first. Subtracting this slow component (τ_SLOW) from the total tail current left the fast component (τ_fast). The sum of these two functions described the tail currents. For a closing voltage of − 60 mV (the holding potential for most experiments), τ_fast was about 200 μs and τ_slow about 1.9 ms. In some cells, deactivation of Sr²⁺ currents was approximately 20% faster than for Ba²⁺ or Ca²⁺ currents.

The Ba²⁺ current through Ca²⁺ channels showed very little decline during 50 ms depolarizations. The current at the end of a 50 ms pulse was usually within 5–10 % of the peak level (Fig. 8). Similarly, changes in holding potential from −80 mV to −40 mV did not alter the current-voltage relationship for Ba²⁺ current or the activation kinetics. However, the decay of the current amplitude was greater with either Sr²⁺ or Ca²⁺ as the charge carrier. This was most readily observed during a 100–300 ms pulse. The amplitude decayed to as little as 10 % of the peak value for a 300 ms pulse to +10 mV. The kinetics of this decay was well described by the sum of two exponentials. Unfortunately, a detailed analysis of the inactivation properties was precluded by partial obscuration of the inactivation by a contaminating outward current in many cases. This was made more prominent by the absence of Ba²⁺ and by the longer pulses required to observe the slow time course of the decay. Thus, it was difficult to determine reliably that portion of the current decay which represented true inactivation of calcium channels.

A fast, transient outward current obscured the early phases of Ba²⁺ current activation. Since specific blockers of sodium channels, such as saxitoxin or tetrodotoxin, do not affect Na⁺ currents in Hirudo medicinalis, eliminating the external Na⁺ only reduces inward I_Na. Residual intracellular Na⁺ can still produce an outward I_Na. Under appropriate conditions (low external [Na⁺]), depolarizing pulses to the sodium reversal potential still revealed a second component. This component was never inward for small depolarizing pulses and was an Off-current at the end of a depolarizing pulse. Ion substitution experiments failed to identify a reversal potential and the rising phase was too fast to be an outward I_Na. It seems likely that this represents a non-linear capacity current.

One aim of the present experiments has been to describe the properties of calcium channels in leech neurones in culture; Retzius cells were of special interest because they form chemical synapses rapidly and reliably (Fuchs et al. 1982; Liu and Nicholls, 1989). In previous studies, calcium entry has been measured in these cells with conventional two-electrode voltage-clamp and with optical recordings using the Ca²⁺ indicator Arsenazo III (Stewart et al. 1989a; Ross et al. 1987). Although these techniques have provided information for correlating Ca²⁺ entry with transmitter release and growth, they are inadequate for measuring the kinetics of Ca²⁺ currents or for analysing how many types of calcium channel might be present.

The hybrid voltage-clamp method has offered certain advantages for approaching these problems. For example, the speed of the clamp and the small capacitative artefacts made it possible to measure more accurately the kinetics of activation and deactivation; also, isolation of the currents was improved since the technique permitted internal perfusion with Cs⁺ and TEA⁺ to block residual K⁺ currents unaffected by external 4-AP and TEA⁺. Nevertheless isolation of the Ca²⁺ current was never perfect, including the initial transient that masked early phases of Ca²⁺ current. One possibility is that this early transient represents gating currents for Ca²⁺ or Na⁺ channels (Adams and Gage, 1976; Kostyuk et al. 1981). This interpretation is problematic due to the large current amplitude in some cells (Armstrong, 1981). Another possibility was that the transient was caused by outward monovalent cation movement through sodium channels (Hille, 1984).

The results obtained by the hybrid voltage-clamp method suggested the presence of a single type of calcium channel, rather than a multiplicity of channel types. The kinetics of activation and the two phases of deactivation can most simply be explained by one type of calcium channel with more than one closed state (Llinas et al. 1981; Tsien, 1983; Brown et al. 1983; Byerly et al. 1984; Llano and Bookman, 1986). In its properties the calcium channel of Retzius cells described here resembled others that activate at high voltages and inactivate slowly, such as the HVA, L or FD types of calcium channel seen in a variety of cells (Carbone and Lux, 1984a; Nowycky et al. 1985; Armstrong and Matteson, 1985; Matteson and Armstrong, 1986). Retzius cell calcium channels were also similar to those of squid giant synapses and some other invertebrate neurones in their insensitivity to dihydropyridine agonists or antagonists (Augustine et al. 1987). With their slow and incomplete inactivation the calcium channels in Retzius cells behaved like those in nerve terminals or synaptosomes (Augustine et al. 1987; Lemos and Nowycky, 1989). This ensemble of properties, including high Sr²⁺ conductance and voltage-dependent activation and deactivation kinetics, set Retzius cell calcium channels apart from those found in other cells (Nachsen and Blaustein, 1982; Augustine and Eckert, 1984; Hille, 1984; Hess and Tsien, 1984; Hess et al. 1986; Tsien et al. 1987). In summary, Retzius cell calcium channels did not fall neatly into any of the conventional categories.

Unfortunately, patch recordings were too few to characterize Retzius calcium channels further at the single-channel level. The dearth of calcium channels in patch recordings was unexpected and cannot be simply explained. With peak macroscopic currents of approximately 20000pA and with cells having a membrane capacitance of approximately InF (corresponding to 100 000 μm²) one can estimate about one active channel per μm² (assuming that the single-channel current would be about 0.2 pA for such a peak depolarization). Single-channel pipettes covered approximately 1 μm² of the cell surface, from which one might expect most of the patches to have contained a channel. The small number actually observed could be explained by complicated infoldings of the cell membrane with calcium channels preferentially located in structures inaccessible to a patch pipette or by a non-homogeneous distribution of channels over the cell surface.

It was not possible by our techniques to assess whether different types of calcium channels exist in different regions of the Retzius cell, its soma, initial segment, axons and growth cones. A small population of such channels highly concentrated in one part of the cell would not necessarily give rise to distinctive currents measured under the present conditions. Optical recordings and loose patch-clamp experiments suggest that the voltage-sensitive calcium channels are not uniformly distributed but are more concentrated in the initial segment (Ross et al. 1987; Bookman et al. 1987; Nicholls and Garcia, 1989; Garcia et al. 1990).

In spite of such uncertainties regarding the numbers of calcium channel types and their distribution, these observations set the stage for further analyses of calcium channel function in relation to synapse formation and transmitter release by Retzius cells. The currents seen with whole-cell clamp can be compared with those observed by loose-patch clamp experiments and correlated with concentration changes measured by high-resolution optical imaging of calcium transients at newly formed synapses.

We particularly thank Dr J. G. Nicholls for his support, advice, criticism and help during all phases of this work. We thank Dr W. B. Adams for his help with voltage clamp and Dr R. R. Stewart for helpful discussions. We are indebted to Mr P. Battig, Ms H. Niederer and P. Muller for their superb technical assistance and also to Ms J. Wittker for her skilled typing. This work was supported by research grants from the Swiss National Fond (no. 3.556-0.86) to J. G. Nicholls. RJB was the recipient of a NATO Fellowship and an NRSA award from the US Public Health Service.