Distribution of Na⁺ and K⁺ Currents in Soma, Axons and Growth Cones of Leech Retzius Neurones in Culture

For a neurone to receive, integrate and transmit signals effectively, different populations of ion channels must be distributed over its surface membrane in a functionally appropriate pattern. For example, in many neurones the relative proportions of voltage-sensitive Na⁺, Ca²⁺ and K⁺ channels are different in different parts of the cell such as cell body, axon, dendrites, nodes of Ranvier, internodes and presynaptic terminals. Moreover, the pattern of distribution is characteristically different for each type of neurone (Waxman and Ritchie, 1985; Angelides et al. 1988; Thompson and Coombs, 1988; Byerly and Leung, 1988). Identified leech neurones such as Retzius and sensory cells display highly distinctive electrical properties and action potentials. By two-electrode voltage clamp it has been shown that these differences can be explained in terms of the distribution and properties of K⁺ channels (Stewart et al. 1989b).

This study describes experiments with two main aims: first, to compare the distribution of ion channel densities in various sites on the Retzius cell of the leech immediately after its removal from the CNS; second, to follow how the distribution changes as growth of new neurite processes occurs in culture. The Retzius cells are unipolar, serotonergic neurones which have a modulatory role in the animal and play a part in the swimming behaviour. In culture the cell sprouts rapidly on appropriate substrata and reliably forms specific chemical and electrical connections with an appropriate target cell (Nicholls, 1987) such as the pressure sensitive P cell. By a novel technique we can now remove the cell body together with the proximal axonal segment and the primary and secondary axons. Detailed information is available from voltage-clamp measurements about the properties of voltage-sensitive ionic currents in cultured Retzius cells. Thus, Na⁺ currents, as expected, activate and inactivate rapidly with depolarization, whereas Ca²⁺ currents activate and inactivate more slowly. At least four distinct types of K⁺ currents have been identified: (a) rapidly activating and inactivating (I_A) channels blocked by 4-AP; (b) more slowly activating and inactivating K⁺ channels blocked by TEA⁺ (I_Ki); (C) Ca²⁺-activated K⁺ channels (I_c), and (d) slow, delayed K⁺ channels (I_K2) (Stewart et al. 1989b).

There are various ways to explore the distribution of ion channels over the surface of an isolated cell. A successful approach to the measurement of Ca²⁺ channel distribution has been the use of the Ca²⁺-sensitive indicator dye Arsenazo HI to measure Ca²⁺ entry into different regions of the cell as it grows on various substrata (Ross et al. 1987, 1988). In spite of low spatial resolution, it has been possible to show that Ca²⁺ entry evoked by impulses is highly non-uniform, being largest in the proximal axonal segment. A completely different approach, the loose-patch method introduced by Almers et al. (1983), offers certain advantages over the dye method. A small area of membrane is rapidly and effectively voltage clamped through a pipette that can be moved many times from one place on the neurone surface to another without damaging the cell. With appropriate fluids in the pipette one can selectively isolate currents carried through Na⁺, Ca²⁺ and specific types of K⁺ channels.

The following questions have been studied with this method. (1) How are voltage-sensitive Na⁺ channels distributed over the surface of a Retzius cell immediately after removal; are there specific regions of high channel density? (2) Are the voltage sensitivity and kinetics of Na⁺ currents similar in different areas of the neurone? (3) Are Na⁺ channels redistributed as new neurites extend? (4) Are the various types of K⁺ channels distributed in the same manner as Na⁺ channels, and are they distributed separately or together? (5) Can Ca²⁺ channel distribution be measured by this method? Brief accounts of some of these experiments have been reported elsewhere (Bookman et al. 1987; Garcia et al. 1989; Nicholls and Garcia, 1989).

Ganglia were dissected out of the leech, pinned in a Petri dish and the capsules opened as described previously (Dietzel et al. 1986). They were then gently shaken during mild enzyme treatment with collagenase-dispase (2 mg ml⁻¹) for 30 min. Individual Retzius cells were identified in leech ganglia by direct observation of their location. A new procedure has been devised to isolate the cell body together with long processes. First, all the small cells surrounding the Retzius cells in the central packet were sucked out one by one; the ganglia were re-incubated on a shaker with enzyme for a further 30min. At this time the Retzius cell could be gently sucked out together with a length of axon attached to the initial segment or ‘stump’. The axon often continued beyond the first bifurcation point, providing a cell with primary and secondary processes (Fig. 1). Cells were immediately plated in normal L-15 medium on a substratum consisting of Concanavalin A, as previously described (Chiquet and Acklin, 1986). Video recordings during the first 24 h showed that the initial segment or stump contracted and became thicker.

Normal L-15 medium consisted of Leibovitz L-15 medium (GIBCO) supplemented with 2 mmol l⁻¹ glutamine (GIBCO), 6mgml⁻¹ glucose, 0.1mgml⁻¹ gentamicin sulphate (Garamycin; Schering, Kenilworth, NJ) and 2% foetal calf serum (Biological Industries, Kibbutz Beth Haemek, Israel).

Solution I, for recording Na⁺ currents, contained (mmol 1⁻¹): TEACI, 25; CdCl₂, 2; 4-AP, 5, in normal medium.

Solution II, for recording Na⁺ currents, contained (m mol1⁻¹): TEACI, 25; CdCl₂, 2; 4-AP, 5; NaCl, 125.

Solution III, for recording transient K⁺ currents, contained (mmol 1⁻¹): TEACI, 25; KC1, 4; CdCl₂, 2.

Solution IV, for recording delayed K⁺ currents, contained (mmol 1⁻¹): 4-AP, 5; CdCl₂, 2; KC1, 4.

Solution V, for recording Ca²⁺ currents, contained (mmol l⁻): TEACI, 25; 4-AP, 5; CaCl₂, 10; KC1, 4.

Solution VI, for recording Ba²⁺ currents, contained (mmol l⁻¹): BaCl₂, 75; TEACI, 50; Hepes, 20. This solution was used in the pipette and the bath.

Solutions II-V were buffered to pH 7.4 with TrisCl. The osmolarity was adjusted to 350—370 mosmol l^-1 with N-methyl-d-glucamine chloride.

Loose-patch electrodes with tip diameters of approximately 3–10 μm were pulled on a vertical puller and fire polished. Sylgard coating up to the tip considerably reduced capacity artefacts. Ag-AgCl wires were used in the pipette and in the agar bridge to the bath. The fluid in the patch pipette consisted of normal L-15 medium for the measurement of total currents. For isolation of individual ion currents, appropriate blocking agents were introduced into the pipette. To isolate Ca²⁺ currents TEACI was injected into the cells through an intracellular microelectrode (see below).

Suitable pipettes had resistances of 0.4–0.8MΩ. The same pipette was used for a whole series of experiments. To form the seal, the pipette was brought close to the membrane with the micromanipulator. Simultaneously, gentle suction was applied while some of the pressure of the pipette against the membrane was released. For tight and stable seals it was necessary to wait for 3 min before recording. With adequate seals, the seal resistance (R_s) was approximately 5–10 MΩ or 10 times the pipette resistance (R_p). In all experiments described below pipette and seal resistances remained constant in any one series of recordings and from patch to patch on the same cell. For results to be compared, R_s/(R_s+R_p) had to be constant. Pipettes were replaced several times at the same place with a precision of 10 μm. Results obtained at any one site were repeatable after a series of experiments at other locations. The equivalent circuit (Stühmer et al. 1983) and photographs of the pipette and cells are shown in Fig. 1. The electronics were constructed for us by Dr W. B. Adams. The amplifier was an EPC-7 (List Electronic, Darmstadt, FRG) and the head stage, which was specially designed, contained a single LF 356. Controls for electrode and seal resistance compensation were contained in a separate box.

The technical limitations of the loose patch-clamp method have been analysed and described by Stühmer et al. (1983). They arise mainly from low electrical seal resistances between the pipette rim and the cell membrane. Voltage control may be poor underneath the rim, and the noise generated by the low seal resistance precludes accurate recordings of small currents such as Ca²⁺ currents. To assess voltage control, in some experiments we placed an intracellular microelectrode as close as possible to the extracellular patch electrode. In an extreme case we found that during a voltage pulse of 80 mV generating a 4 nA outward current, the voltage error at the end of the 20 ms pulse was 5 mV. There was no noticeable voltage escape at the beginning of the pulse during the flow of a 2 nA inward current. Since most currents recorded in our study were only about half these sizes, inadequate voltage control should not have been a major source of error. Nevertheless, current densities varied greatly from one cell to another. Therefore, for statistical analyses, relative rather than absolute comparisons have been made. A satisfactory aspect of these experiments was the reproducibility of currents measured at the same spot time after time, after the pipette had been removed and the pipette fluids changed.

Holding potentials and pulses were applied directly by the pipette. Most records show displacements from an unknown resting potential, which, when measured by an intracellular microelectrode, was between −30 and −55 mV. A membrane potential of −40 mV was assumed for all plots of current-voltage relationships. One problem with loose-patch recording is that large currents flow through the clamped area of membrane and can alter the resting potential. As shown previously (Bookman et al. 1987) this was not a serious problem provided that pulses lasting less than 50ms were used (see Discussion). With longer pulses, outward K⁺ currents and leak currents became sufficiently large to alter the membrane potential of the cell as a whole, as measured by the intracellular microelectrode. Such intracellular measurements were made in the soma and primary axons but were not feasible in growth cones.

Pipette solutions were changed by ejecting them with positive pressure applied to the rear end of the pipette and sucking them up by applying negative pressure. In experiments using solutions I-V in the pipette, the bath solution was normal L-15 medium. All experiments were carried out at room temperature.

In all experiments the depolarizing pulse was followed by three pulses in the opposite direction scaled to one-third amplitude for subtraction of leakage and capacitative currents. The currents were digitalized and stored on a laboratory computer (HNC/DECLAB 23).

Ca²⁺ currents were recorded after injecting TEACI into the cell or into the stump to block K⁺ channels more efficiently. Constant depolarizing current (1–2nA) was passed through the TEA⁺ microelectrode for 15min. The loose patch pipette was filled with solution V containing a high Ca²⁺ concentration. The effect of intracellular TEA⁺ on K⁺ currents could only be observed near the site of injection. To record K⁺ currents, Na⁺ currents were reduced by using Na⁺-free solutions in the pipettes. Tetrodotoxin and saxitoxin (alas!) do not block leech sodium channels.

Characteristic inward and outward membrane currents were evoked by applying depolarizing pulses to the surface of a Retzius cell minutes after it had been removed from the CNS of the leech. Fig. 2A shows records of total currents at the tip of the secondary process obtained with normal L-15 solution in the bathing fluid and in the pipette. Separation of inward and outward currents was achieved by altering the pipette solution.

To reveal Na⁺ currents in isolation, pipettes were filled with solutions containing TEA⁺, 4-AP and CdCl₂ (solution I or II see Materials and methods), and depolarizing pulses were applied (Fig. 2B). Inward currents were maximal at approximately 0 mV (see Fig. 5) and became changed to outward ones when fluid containing 0 mmol 1⁻¹ Na⁺ was present in the pipette.

A steady 20 mV depolarization of the patch membrane by current applied through the pipette inactivated the inward current observed during a depolarizing step. Together, these results revealed the presence of voltage-sensitive, rapidly activating and inactivating Na⁺ channels in the patch. The properties of these Na⁺ channels resemble those observed by two-electrode voltage-clamp, with the advantage that the rising and falling phases could be observed more accurately owing to the faster speed of the clamp and the smaller amplitude of the capacitative artefacts. The results, in addition, show that the fluid in the pipette rather than that in the bathing solution effectively determines the pattern of ionic currents flowing through the membrane patch. This is shown by the blockage of individual current components by pharmacological agents and by alterations of ionic composition.

Four types of K⁺ currents have been characterized in R cells: I_A, I_K1, I_c and I_K2 (Stewart et al. 1989b). Two types of K⁺ currents were recorded with the loose patch in the isolated cells (Fig. 2C, D). The transient K⁺ current of Fig. 2C resembled an A current, was rapidly activating and inactivating, and was insensitive to TEA⁺. The external fluid and the loose-patch pipettes contained 25 mmol 1⁻¹ TEACI (solution III). In Fig. 2C depolarizing pulses evoked a rapidly rising outward current that decreased during the pulse. By contrast, the K⁺ current of Fig. 2D activated more slowly, did not inactivate, and was insensitive to 4-AP. In these records the loose-patch pipette contained 5 mmol 1⁻¹ 4-AP (solution IV) instead of TEA⁺. This current resembled the I_K1 current. No further attempt was made to separate the other K⁺ current components, I_c and I_K2-The reliability of the method is apparent from comparison of the two traces in Fig. 2C, which were taken before and after the recording shown in Fig. 2D using the same pipette on the same spot but with different blocking fluids. The current records superimpose.

Ca²⁺ currents were considerably harder to isolate than Na⁺ or K⁺ currents. The difficulty arose in part from the small amplitudes of the Ca²⁺ currents and in part from the presence of a masking outward current that had activation kinetics similar to that of the Ca²⁺ current. With pipettes containing no Na⁺ to eliminate Na⁺ currents, as well as TEA⁺ and 4-AP to block K⁺ currents, only very small currents were recorded during the pulse (Fig. 3A). At the end of the pulse, however, a large inward tail current was evident. Such tail currents increased in amplitude with increasing pulse durations and potentials (Fig. 3D). With Ba²⁺ in the pipette the amplitude of the tail currents increased. Ca²⁺ tail currents were most easily observed when TEA⁺ was injected intracellularly to block residual K⁺ currents. Even after TEA⁺ injection, Ca²⁺ tail currents were obvious mainly at or close to the site of injection. For example, after TEA⁺ had been injected into the proximal axon of a Retzius cell, loose-patch recordings revealed a tail current in that region but not in the soma. After injection of TEA⁺ into the soma, a Ca²⁺ tail current appeared first in the soma and then in the proximal axon after a delay of a few minutes. When Cd²⁺ was added to the pipette fluid, to block currents through Ca²⁺ channels, the inward tail was abolished and an outward current was revealed (Fig. 3B). This current had an activation half-time of the order of 1–2 ms, and very rapidly deactivated at the end of the pulse. The charge carrier for this outward current was not identified but it was found to persist in the presence of high concentrations of traditional channel blockers. Subtraction of the traces in Fig. 3B from those in Fig. 3A revealed an inward current that had been blocked by Cd²⁺ (Fig. 3C), presumably carried through Ca²⁺ channels. This ‘Ca²⁺ current’ had a half-time for the rising phase of 1–2 ms. The steady-state amplitude (Fig. 3D) increased with larger depolarizations, as expected. Owing to their similar amplitudes and activation kinetics, the Ca²⁺ current and the outward current effectively cancelled one another during the pulse. At the end of the pulse, however, in the absence of Cd²⁺, a Ca²⁺ tail current was revealed that reflected the extent of Ca²⁺ current activation during the pulse. (For technical reasons relating to the subtraction procedures, the fast tail current is not visible in Fig. 3C.)

Together the results show that Na⁺, Ca²⁺, fast K⁺ and slow K⁺ currents can be isolated. Their properties resemble those observed by conventional voltage clamp (Stewart et al. 1989b). Moreover, the currents could be recorded reliably again and again from the same spot with different bathing and pipette fluids.

Sodium channels were distributed over the surface of freshly dissected Retzius cells in a characteristic, non-uniform manner, as shown in Fig. 4A. Within a few minutes of removal from the CNS, large inward Na⁺ currents were recorded with step depolarizations applied to the axon. Invariably, the largest currents occurred at the broken tip of the axon from which neurite outgrowth would begin. No detectable Na⁺ current could be recorded from the soma or the initial axon segment (stump) of such a cell. The occurrence of the highest channel density at the tip was correlated with damage. Retzius cells were isolated with different break-points: at the secondary axon, at the primary axon, at the end of the stump or in the middle of the stump. In each type of preparation the largest current occurred at the sealed tip. Possible reasons include membrane infolding and channel redistribution (see Discussion). The kinetics and voltage dependence of sodium channels were similar at all sites recorded. Fig. 5 shows the currentvoltage relationship for records taken at the primary and at the secondary axon in a Retzius cell after 2 h in culture. At both sites the peak inward current occurred at about +2 mV. The inset of Fig. 5 shows similar activation and inactivation kinetics for the sets of channels at the two sites. The two traces recorded from the axon tip in Fig. 5 were taken before and after recording from the soma with the same pipette.

Na⁺ current was recorded from newly grown processes, and from growth cones, a few hours after removal. There was no obvious delay between the appearance of the membrane and the appearance of Na⁺ channel activity. After 2 days in culture, Na⁺ current was more uniformly distributed over the entire cell than at the time of isolation. Na⁺ current could then be recorded in the soma as well as on the axons and in the growth cones (see Fig. 8).

With a pipette containing TEA⁺ solution III to block slow K⁺ currents, the remaining outward transient K⁺ current activated and inactivated more rapidly (Fig. 6A). As expected, it was reduced or abolished by steady depolarization of the holding potential by 20 mV. With a pipette containing 4-AP, K⁺ currents were slower and did not inactivate during the 50 ms pulse (Fig. 6B).

The distribution of either current was more uniform than that of the Na⁺ currents. Although variations in K⁺ current density were observed in individual cells (Fig. 6), no variation was apparent when results from a number of cells were pooled (see Fig. 8). The different characteristics of fast and slow K⁺ currents were confirmed by studying current-voltage relationships (Fig. 7A). Fast and slow currents differed in their sensitivity to depolarizing steps: fast currents began to activate at depolarizing steps to −35 mV, whereas slow channels required a larger depolarization to at least −25 mV. Measurements made from the soma, initial segment, distal axons or growth cones showed no differences in the voltage sensitivity or activation characteristics: the different regions were similar in the properties of their fast or slow K⁺ currents. Current-voltage relationships in the soma and the axon for delayed and transient K⁺ currents are shown in Fig. 7B,C; the kinetics of K⁺ currents at different sites are given in the insets of Fig. 7.

K⁺ channels, as well as Na⁺ channels, appeared in growing processes and growth cones (Fig. 8). Both transient and delayed K⁺ currents were recorded as soon as new processes grew out (Fig. 8). In the present experiments the transient and delayed K⁺ currents were studied without any further separation of the various additional K⁺ current components identified by Stewart et al. 1989b. It was, however, evident that, in addition to the 4-AP-sensitive transient K⁺ currents and the TEA⁺-sensitive delayed K⁺ currents, other K⁺ currents with distinctive properties appeared in growth cones and newly grown processes after 2 days in culture. Such currents were more resistant to TEA⁺ and 4-AP. They were not analysed in detail.

For technical reasons it was difficult to compare quantitatively the calcium currents in different regions of the cell. The currents were small and required highly standardized seal resistances (see above). We did not investigate changes during development in culture.

Ion channels of the Retzius neurone of the leech, studied using the loose-patch technique, showed an uneven distribution over the surface of the cell that could change during development in culture. The observed properties of the channels agreed well with those found earlier using other methods (Stewart et al. 1989a,b; Ross et al. 1987, 1988).

In cells that had just been removed from the CNS, sodium current could be recorded only in the broken extremity of the axon. After 2–3 days in culture, this current was found in the soma and proximal region of the axon also. The two types of potassium current that were studied, fast I_A and slow I_K1, showed variation in distribution that was not consistent between cells, yielding a uniform distribution on average.

After 2 days in culture, when neurite outgrowth was extensive, Na⁺ and K⁺ currents could be recorded in growth cones. Current densities in these newly formed structures were similar to those recorded in the tip of the broken axon from which neurite outgrowth occurred. This indicates that the channels may first move towards the site of damage, and later diffuse into the newly formed neurites. Another possibility is insertion of new channels into these sites via axonal transport of the channel proteins. In rat embryonic cortical neurones and spinal cord cells Angelides et al. (1988) found different lateral mobility of Na⁺ channels within the membrane plane in different parts of the neurone. This would be in agreement with the first hypothesis mentioned above. In contrast, unequal distribution of Na⁺ channels has also been shown in skeletal muscle fibres (Beam et al. 1985). Na⁺ channel density is highest near the site of the neuromuscular junction of these cells. Na⁺ current densities vary up to threefold over short distances and the lateral mobility of the channels is undetectably small in frog skeletal muscle (Aimers et al. 1983). Further experiments with Retzius cells will be needed to decide whether lateral diffusion or axonal transport accounts for the highly localized distribution of these channels in the present study. Other techniques will also be needed to obtain a better resolution of the distribution of Ca²⁺ channels in these cells.

What are the functional implications of the special localization of Na⁺ channels in growing Retzius cells? The highest Na⁺ current density could always be recorded at the tip of the cut axons after isolation of the cells. Neurite outgrowth also occurred most actively in this region. The access resistance of the narrow neurites is presumably large. For spread of excitation into the neurites to occur, a high density of Na⁺ inward current is needed to overcome the access resistance. In addition, as discussed above, Na⁺ channels may diffuse from the sites of high density into the newly formed neurite membranes. Both functional implications have still to be verified. It will also be interesting to see whether and how channel distribution is affected by synapse formation.

It is our great pleasure to thank Drs J. G. Nicholls and W. B. Adams for their enthusiastic encouragement, their participation in some of the experiments and their help with the manuscript. Without their support this study would not have been completed. We also gratefully acknowledge the financial support by the Swiss National Science Foundation (grants 3.556-0.86 to J. G. Nicholls and 3.078–1.87 to HR). Ms S. Grumbacher-Reinert is the recipient of an award from Boehringer Ingelheim Stiftung.