INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Many neurons respond to injury of a peripheral axon with persistent alterations of excitability as well as regenerative growth (Titmus and Faber 1990; Walters 1994). In the case of somatic sensory neurons in mammals (Abdulla and Smith 2001b; Gallego et al. 1987; Gurtu and Smith 1988; Kim et al. 1998; Liu et al. 2000; Stebbing et al. 1999; Zhang et al. 1997) and in the mollusk, Aplysia (Ambron et al. 1996; Bedi et al. 1998; Clatworthy and Walters 1994; Gunstream et al. 1995; Walters et al. 1991), axon injury often produces persistent hyperexcitability of the soma as well as parts of the neuron close to the site of injury (Billy and Walters 1989; Dulin et al. 1995). Nociceptive sensory neuron hyperexcitability in mammals is thought to contribute to hyperalgesia and neuropathic pain (Abdulla and Smith 2001b; Babbedge et al. 1996; Kajander et al. 1992; Na et al. 2000; Study and Kral 1996; Wall and Devor 1983). In Aplysia, similar hyperexcitability of sensory neurons is produced by axotomy and by manipulations associated with learning and memory. This similarity suggested that mechanisms of hyperexcitability that can be induced by nerve injury might also be utilized in other long-term alterations, such as memory (Walters 1994; Walters et al. 1991).

The simple form of learning that has been linked to long-term hyperexcitability of Aplysia sensory neurons is behavioral sensitization produced by noxious stimulation (Cleary et al. 1998; Walters 1987). Noxious stimulation of intact Aplysia causes a long-lasting (1 day) depression in sensory neurons of outward currents that resemble the S-type ("serotonin-sensitive") K⁺ current, I_K,S (Cleary et al. 1998; Scholz and Byrne 1987). Short-term (and perhaps long-term) sensitization is thought to involve release of serotonin (5-HT; reviewed by Byrne and Kandel 1996), and brief application of 5-HT to these sensory neurons transiently reduces I_K,S (Baxter and Byrne 1989; Klein et al. 1982; Pollock and Camardo 1987; Pollock et al. 1985; Shuster and Siegelbaum 1987; Shuster et al. 1991; Siegelbaum et al. 1982). I_K,S is characterized by its partial activity at resting membrane potential (RMP), weak voltage dependence, slow activation with depolarization, lack of inactivation during prolonged depolarization, relative insensitivity to Ca²⁺ and to various K⁺ channel blockers, including tetraethylammonium (TEA) and 4-aminopyridine (4-AP), activation by arachidonic acid, and reduction by cAMP. In mammals a two-pore-domain K⁺ channel, called TREK-1 (Fink et al. 1996; Patel et al. 1998) or KCNK2 (Goldstein et al. 2001), has recently been identified and shown to have these same properties, suggesting that I_K,S may be conducted by related channels.

In principle, axotomy-induced hyperexcitability of Aplysia sensory neurons might be accounted for by a reduction of I_K,S. Short-term reduction of I_K,S produced by 5-HT increases excitability by reducing spike frequency adaptation (accommodation) during trains of action potentials (Baxter and Byrne 1990; Klein et al. 1986) and by lowering the threshold for spike initiation (Klein et al. 1982). Reduction of I_K,S should have particularly important effects on responses that occur within the voltage range in which I_K,S makes a large relative contribution to membrane potential (V_m), which includes the range between RMP (approximately 50 mV) and the threshold for spike initiation (approximately 35 mV). Because at least one component of I_K,S has very slow kinetics of activation, a reduction of I_K,S would be expected to have prominent effects during prolonged depolarizations, such as the slow afterdepolarization that often follows a burst of spikes evoked by noxious peripheral stimulation (Clatworthy and Walters 1993b; Walters 1987; Walters and Byrne 1983). Large afterdepolarizations can lead to afterdischarge, which is enhanced following axotomy (Walters et al. 1991). In addition, because I_K,S is partially active at RMP, a reduction of I_K,S would be expected to alter passive membrane properties, including input resistance (R_in) and the membrane time constant ().

In the present study, we show that peripheral axotomy of nociceptive sensory neurons in Aplysia does, in fact, produce a long-lasting reduction of I_K,S and alteration of passive membrane properties. In addition to passive alterations attributable to reduced I_K,S, we also found that the apparent membrane capacitance of axotomized sensory neurons was increased, which is consistent with morphological evidence that axotomy causes sprouting of the sensory neuron within the pleural ganglion (Steffensen et al. 1995). A concomitant decrease in I_K,S and increase in input capacitance (C_in) suggest that the reduction in I_K,S functions, in part, to compensate for the decrease in excitability that would otherwise occur during regenerative growth. Some of these results have been described in abstract form (Gasull et al. 2001).

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

General procedures

Aplysia californica (70-250 g) were supplied by the National Institutes of Health-Aplysia Resource Facility (Miami, FL) and Alacrity Marine Biological Services (Redondo Beach, CA). Animals were housed in aquaria containing artificial seawater (ASW; Instant Ocean, Burlington, NC) at 15-17°C for 1-10 days before surgery. Body weight was maintained on a diet of Gracilaria seaweed or dried seaweed laver. Axotomy was produced in vivo after anesthetizing the animal with an injection of ~30% body volume of isotonic MgCl₂ into the neck. An incision (~1 cm) was made just above the pedal ganglia, and a pair of blunt forceps was used to crush all pedal nerves on one side of the animal 5-10 mm from the pedal ganglion (~2 cm from the sensory neuron somata in the pleural ganglion). The side crushed was alternated across animals. The incision was closed with stitches, and the animal was placed back into its home tank. Five to 10 days following nerve crush, animals were injected with isotonic MgCl₂ (~50% body vol) and dissected. Pedal-pleural ganglia from both sides of the animal were excised and pinned in silicone elastomer (Sylgard)-covered dishes. The pleural ganglia were surgically desheathed in a 1:1 solution of MgCl₂ and ASW, which was then exchanged for buffered ASW composed of (in mM) 460 NaCl, 11 CaCl₂, 10 KCl, 55 MgCl₂, and 10 Tris buffer (pH 7.6).

Two-electrode current clamp

After 1 h in buffered ASW, sensory neurons from corresponding locations in the axotomized and control VC clusters (Walters et al. 1983) were tested at 15-18°C. Each sensory neuron soma was impaled with two glass microelectrodes (10-20 M) filled with 3 M K-acetate, with one electrode used for passing current and the other for measuring V_m. Note that the site of axon injury is probably too distant (~2 cm away) to directly influence electrical measurements made in the soma. For example, even very large changes in holding potential of the soma in the pleural ganglion have no detectable influence on transmitter release onto motor neurons 2-3 mm away in the pedal ganglion, even though hyperpolarization of sensory neuron axons or somata near release sites in the same ganglion significantly reduces release (Hammer et al. 1989; X. Liao and E. T. Walters, unpublished observations). In the present study, cells that did not have action potential amplitude >70 mV, RMP < 40 mV, and input resistance R_in >10 M were excluded. In most experiments RMP was held at 50 mV. R_in and  were determined with 2-s hyperpolarizing pulses at either 0.2 or 0.5 nA or with the increasingly negative series of pulses depicted in Fig. 1B. Soma spike threshold was measured with a standard series of 20-ms depolarizing pulses (Walters et al. 1991) or, for determination of rheobase, during an ascending series of 2-s pulses delivered at 15-s intervals. Repetitive firing was quantified by counting the number of spikes evoked by either a 1-s intracellular depolarizing pulse at 2.5 times the threshold current determined with the 20-ms pulses or with the constant currents indicated in the text.

View larger version (11K): [in this window] [in a new window]   Fig. 1. Axotomy alters responses of sensory neurons to prolonged (2 s) hyperpolarizing pulses under 2-electrode current clamp. A: representative examples of responses in control and axotomized cells to the smallest test current used, 0.2 nA (for 2 s), from a holding potential (Vh) of 50 mV. The relaxation of membrane potential, Vm, between the early peak hyperpolarization and the end hyperpolarization is termed "sag" (see Table 1). B: for control cells and axotomized cells the V-I curves (i.e., the end hyperpolarization in each case as a function of injected current) were significantly different (see text). Note the divergence of the curves with smaller injected currents, and convergence of the curves with larger currents. Error bars in this and all other figures indicate SE.

Two-electrode voltage clamp

Voltage-clamp studies were conducted at 15-18°C using methods similar to those used by Baxter and Byrne (1989) on the same sensory neurons. The pleural sensory neurons arborize extensively in the pedal ganglion (2-5 mm from their somata in the pleural VC cluster) and in the pleural ganglion underneath their somata, and thus present major obstacles to an effective space clamp (Nazif et al. 1991; Steffensen et al. 1995). To achieve an adequate space clamp, following desheathing these compartments of the sensory neurons were at least partially removed by either excising the cluster of somata from the pleural ganglion using iridectomy scissors or by severing the pedal-pleural connective near the VC cluster. Each cluster was pinned down in a 200-µl chamber. The current-passing and recording electrodes were filled with 3 M K-acetate and beveled to a final resistance of 5-7 M. After impaling a sensory neuron, the two electrodes were shielded from each other by a foil strip lowered as close to the bath as possible. Each cell was voltage clamped using an Axoclamp-2A amplifier (Axon Instruments, Foster City, CA) and conventional two-electrode voltage-clamp techniques. Currents were sampled either at 3 or 10 kHz, digitized on-line with an ITC-16 computer interface (Instrutech), and stored and analyzed on a Power Macintosh G3 computer using Axograph 4.6 software (Axon Instruments). Apparent input capacitance (C_in) was computed using the same software by integrating under a single exponential fit to the first 2 ms of the response to a 10-ms, 10-mV voltage step. No correction was made for linear leak current during integration of the capacity transient. In all cells, most of the capacity transient was fit well by single exponentials, suggesting that most of the membrane contributing to the transient was effectively isopotential. However, in some cells, the late phase of the transient deviated from the single exponential, suggesting that incompletely clamped compartments may have contributed to some of the measurements. We did not attempt quantitative comparisons of the goodness of fit between groups. To minimize cumulative inactivation of membrane conductances, voltage-clamp pulses were separated by 90 s. Instantaneous currents were measured immediately after the capacity transient, 3-5 ms after the beginning of the pulse. Only recordings from cells with the following characteristics were accepted: holding currents < ±2 nA, RMP < 40 mV, R_in > 10 M. In addition, cells with slowly rising or distorted voltage responses were excluded. Tetrodotoxin (TTX; Sigma), 4-aminopyridine (4-AP; Sigma), and tetraethylammonium chloride (TEA; Sigma) were added directly to the experimental chamber in small (2-10 µl) ASW aliquots to final concentrations of (respectively) 150 µM, 1.5 mM, and 50 mM. Each cell was held at 50 mV and given steps (0.1, 0.4, or 2 s) to the indicated potentials.

Statistical analysis

Most data are represented as means ± SE, and were analyzed using t-tests or two-way ANOVA for repeated measures followed by Bonferroni post hoc tests, using Prism software (Graphpad, CA). Except where indicated, two-tailed tests were used, with statistical significance set at P < 0.05. One-tailed t-tests were used when planned, a priori comparisons were made on the basis of statistically significant findings from pilot studies. This permitted the use of smaller n's in the final design of some of the experiments. Frequencies were compared with Fisher's exact test.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Axotomy decreases RMP, increases R_in, and decreases sag during hyperpolarizing pulses

Because I_K,S is partially active at the normal RMP of Aplysia sensory neurons, a reduction of this current by axotomy would be expected to alter resting membrane properties. A comparison of axotomized sensory neurons to contralateral controls 5-10 days after unilateral pedal nerve crush revealed a small but significant decrease in RMP (Table 1). These RMP values were obtained ~1 min after impalement, when V_m had reached a relatively stable level. However, RMP sometimes showed very slow additional hyperpolarization (over tens of minutes), which was not analyzed.

Axotomy also increased R_in, as determined with 1-s, 0.5-nA hyperpolarizing pulses delivered from RMP (Table 1). Tests using 2-s pulses delivered from a manually clamped holding potential (V_h) of 50 mV showed that the axotomy-induced increase in R_in occurred at some but not all hyperpolarizing pulse amplitudes. Examples of membrane responses to 0.2-nA hyperpolarizing pulses are illustrated in Fig. 1A. In this case, the animal that these cells were taken from had received unilateral pedal nerve crush 7 days earlier. A two-way ANOVA with repeated measures on the group data (Fig. 1B) revealed significant effects of axotomy and test current level (P < 0.002 in each case) but not of their interaction. Although multiple comparisons with Bonferroni post hoc tests did not reveal significant differences at any test current level, planned comparisons using paired t-tests showed that axotomized cells had larger R_in when tested with the smallest current, 0.2 nA (P = 0.003), and with a current close to our standard 0.5-nA test pulse, 0.6 nA (P = 0.02). No significant difference was seen with the largest current (1.6 nA), and it can be seen that the relationships between voltage and injected current (V-I curves) for axotomized and control cells began to converge with more negative current pulses (up to 1 nA), which caused hyperpolarizations to V_m < approximately 80 mV (Fig. 1B). These results, obtained with two electrodes in each cell, extend previous findings, using single electrodes, that suggested that axotomy might alter RMP and R_in but that usually failed to reach statistical significance (Clatworthy and Walters 1994; Gunstream et al. 1995; Walters et al. 1991; unpublished observations).

Another effect of axotomy is illustrated in Fig. 1A, where more "sag" is seen in the hyperpolarizing response of the control cell than the axotomized cell. Sag is a notable relaxation of V_m from an early peak hyperpolarization during prolonged pulses. Defining sag arbitrarily as a ratio of V_m (end hyperpolarization)/V_m (peak hyperpolarization) < 0.95, we found that 28 of 30 control cells displayed sag, whereas only 19 of 30 axotomized cells did during responses to 0.2-nA hyperpolarizing pulses (Table 1, P = 0.010 by Fisher's exact test).

Axotomy increases  and C_in

In general,  is determined by the product of R_in and C_in. Because C_in should be either unchanged or increased following axotomy (see following text) and R_in was found to be increased, we predicted that  would be greater in axotomized than control sensory neurons. Thus we measured time constants during the falling phase of hyperpolarizations produced by 2-s, 0.6-nA pulses, which were more than long enough for complete charging of membrane capacitance. Only cells showing little or no sag (end hyperpolarization/peak hyperpolarization > 0.90; see Fig. 1A) were included in this analysis.  in axotomized cells was significantly longer than that in control cells (Table 1, P = 0.04, paired t-test). This difference in  did not depend on differences between axotomized and control cells in the magnitude of the hyperpolarization, V_m, during the 0.6-nA pulse (see following text).

The difference between axotomized and control sensory neurons in  appeared somewhat greater than the difference in R_in (axotomized  = days 217% of control, whereas axotomized R_in = 152% of control in the same cells), suggesting that C_in also may have been increased by axotomy. Indeed, when cells were selected on the basis of the same range of R_in (20-50 M, determined with 0.6-nA pulses), axotomized cells still had significantly longer  values (36.5 vs. 20.6 ms, P = 0.04 by paired t-test, R_in = 30.9 and 31.4 M, respectively, for axotomized and control neurons, n = 18 and 21 cells, respectively, in 7 animals). C_in was estimated by using brief voltage-clamp pulses and integrating single exponentials fit to the initial capacitive transient. Apparent C_in was significantly greater in axotomized than control sensory neurons (Table 1). Although differences between axotomized cells and control cells in the contribution of incompletely clamped membrane to this estimate of membrane capacitance may account for some of this difference (see DISCUSSION), an increase in apparent C_in is consistent with the increase in total membrane area that is likely to occur during axotomy-induced growth of new processes near the soma (Steffensen et al. 1995).

Altered passive properties account for prolonged responses to brief depolarizing pulses

Soon after beginning our investigations of axotomized sensory neurons, we noticed that an excellent predictor of hyperexcitability was a dramatically slowed recovery of V_m following brief, subthreshold depolarizations (Fig. 2A, top) applied during ascending series of pulses used to determine spike threshold. This prolongation of responses to brief depolarizing pulses was documented by comparing the rate that V_m relaxes in axotomized and control cells following 20-ms pulses. Our index for comparison was the percent recovery of V_m measured 20 ms after the end of the 20-ms test pulse (Fig. 2A, "20-ms recovery"). Axotomized cells showed significantly lower 20-ms recovery (slower relaxation) at all depolarizing (and hyperpolarizing) current levels (Fig. 2B, 2-way ANOVA with repeated measures, P < 0.0001 for effect of axotomy, with P < 0.05-0.001 at different levels by Bonferroni post tests).

View larger version (25K): [in this window] [in a new window]   Fig. 2. Axotomy-induced slowing of the relaxation of Vm following brief depolarizing or hyperpolarizing pulses is accounted for by alterations in passive membrane properties. A: examples of depolarizing and hyperpolarizing responses to ±0.6-nA pulses in a control cell and an axotomized cell. Twenty-millisecond recovery is defined as the percentage of the recovery from the peak Vm measured 20 ms after the end of the pulse. The 20-ms recovery in axotomized cells was significantly lower than in control cells (B), whereas the peak Vm (or, equivalently, Vm because all responses were evoked from the same Vh) showed no difference (C, see text).

In principle, two underlying alterations could contribute to this slower relaxation of V_m in axotomized cells: the observed axotomy-induced changes in passive properties (specifically, increases in R_in and C_in) and direct or indirect (via reduction of opposing outward current) enhancement of a rapidly activating, very low-threshold, voltage-gated depolarizing current. Several observations indicated that enhancement of voltage-gated depolarizing currents was unlikely to be important for prolonging brief, subthreshold depolarizations in axotomized sensory neurons. First, as shown in Fig. 2, A and B, following hyperpolarizing current pulses both axotomized and control cells displayed relaxations of V_m that were roughly symmetrical with those that followed depolarizing pulses of the same absolute valueespecially for the smallest current steps. Second, a voltage-gated depolarizing current that could significantly influence the 20-ms recovery measurement would have to be activated very quickly and thus would be likely to increase the peak amplitude of V_m during subthreshold depolarizing pulses. However, the relationship between injected current and V_m was constant for both depolarizing and hyperpolarizing pulses (Fig. 2C). Third, differences in 20-ms recovery between axotomized and control cells persisted when large reductions were made in extracellular Na⁺, Ca²⁺, and Cl concentrations and the cells were tested with 20-ms pulses of ~0.6 nA (in this study a series of successively greater depolarizing pulses were applied to each cell and those producing depolarizations of 10 ± 1 mV were compared). When extracellular Na⁺ was greatly reduced (<1% normal) by substituting Tris-Cl for NaCl and extracellular CaCl₂ was omitted from the bathing solution, the 20-ms recovery of the axotomized cells was 65 ± 4 versus 82 ± 4% for control cells (P = 0.006, n = 23 and 20 cells, respectively). When extracellular Cl was reduced (~14% normal) by substituting Na-acetate for NaCl and extracellular CaCl₂ was omitted from the bathing solution, the 20-ms recovery of the axotomized cells was 79 ± 2 versus 97 ± 4% for control cells (P = 0.006, n = 14 and 17 cells, respectively). Although the control values for 20-ms recovery shifted in these solutions (see DISCUSSION), the fact that 20-ms recovery in axotomized cells was still significantly lower than in control cells suggests that the more prolonged relaxations of V_m following brief, subthreshold depolarizing pulses in axotomized neurons do not depend on activation or inactivation of Na⁺, Ca²⁺, or Cl conductances.

These data indicate that an axotomy-induced increase in  accounts for the prolongation of responses to brief depolarizing pulses. In general,  should be a function of the product of R_in and C_in. Therefore we were curious to see how the ratio of  in axotomized and control cells compared with the ratio of the products of R_in and C_in. The mean ratio of axotomized/control was nearly identical for the product of R_in and C_in and for directly measured  (2.15 and 2.17, respectively). For both control and axotomized cells, the product of R_in and C_in was ~60% of the empirically measured values of  in the same groups. This difference is not surprising given that  and R_in were measured in intact pleural-pedal ganglia and C_in was measured in partially excised clusters as well as the fact that finding a simple equality of  with the product of R_in and C_in seems unlikely in cells with such complex geometry. The close similarity of the axotomized/control ratios for  and for R_in × C_in even though the measurements of , R_in, and C_in involved different experiments and different preparations supports the conclusion that increases in R_in and C_in are sufficient to account for the prolongation of responses to brief depolarizing pulses (and the slowing of other responses) following axotomy.

Altered responses to long depolarizing pulses are consistent with decreased I_K,S

A notable feature of I_K,S is that, in addition to a resting leakage component, there is a prominent voltage-dependent component that slowly activates and fails to inactivate during prolonged depolarizing pulses (see following text). The time course of this voltage-dependent component is similar to the time course of spike accommodation during prolonged depolarizations under current clamp. Previous studies have shown that axotomy produces hyperexcitability of Aplysia sensory neurons that is marked by a large decrease in spike accommodation (Ambron et al. 1995; Clatworthy and Walters 1994; Gunstream et al. 1995; Liao et al. 1999; Walters et al. 1991). However, no studies have described responses of axotomized sensory neurons to subthreshold depolarizations. Because I_K,S shows significant activation between RMP and the threshold for spike initiation (approximately 35 mV), we used two-electrode current-clamp methods to compare the responses of axotomized sensory neurons and contralateral control neurons to prolonged depolarizing pulses in this subthreshold voltage range. We also extended previous studies of suprathreshold responses, which had only used single-electrode methods.

Figure 3 illustrates typical responses of control and axotomized sensory neurons to depolarizing pulses given from a V_h of 50 mV. During subthreshold pulses, both the control and axotomized sensory neurons showed an initial peak depolarization followed by a partial relaxation (Fig. 3A). We compared V_m at the end of the 2-s depolarizing test pulses ("end depolarization"). Two-way ANOVA with repeated measures showed significant effects of axotomy, test current level, and their interaction (P < 0.0001 in each case). Multiple comparisons with Bonferroni post tests showed that axotomized cells were more depolarized at all test currents from 0.4 to 1.8 nA (P < 0.05 to P < 0.001). A planned comparison using a paired t-test at 0.2 nA revealed a significant difference in end depolarization at even the lowest stimulation current used (P = 0.002). The peak depolarizations were also significantly different. Because the axotomized cells, unlike control cells, often fired action potentials during stimulation with low currents (see following text), we could only compare subthreshold responses at 0.2 nA (even with this low current, 6 of 27 axotomized cells spiked, whereas 0 of 25 control cells did). Comparison of the depolarizing responses in axotomized and contralateral control clusters revealed significantly greater peak depolarizations during 0.2-nA pulses in the axotomized sensory neurons than in control neurons (7.2 ± 0.7 vs. 5.1 ± 0.4 mV, n = 8 animals, P = 0.027, paired t-test). The peak depolarizations occurred early in the pulse (within 50-100 ms) but showed somewhat longer latencies in axotomized cells than control cells (Fig. 4A). The longer latency to peak may be a consequence of the increase in  in axotomized neurons.

View larger version (17K): [in this window] [in a new window]   Fig. 3. Axotomy alters responses of sensory neurons to prolonged depolarizing pulses under 2-electrode current clamp. A, left: traces show representative examples of depolarizing responses to 2-s subthreshold pulses. V-I curves of the end depolarizations were significantly different for control and axotomized cells (right, see text). B, left: traces show representative examples of action potentials (spikes) evoked by an intermediate test current (0.8 nA). The relationships between the number of spikes evoked by each pulse and the injected current were significantly different (see text).

View larger version (14K): [in this window] [in a new window]   Fig. 4. Axotomy decreases rheobase and increases the latency to spike initiation at rheobase. A: examples of spikes initiated at rheobase in control and axotomized cells. Vh was 50 mV and the pulse was 2 s long. Rheobase data were obtained from the study displayed in Fig. 2. Significant differences were found between axotomized and control cells in stimulus current at rheobase (B) and in latency to the action potential at this threshold current (C, see text).

As observed previously (Clatworthy and Walters 1994; Gunstream et al. 1995; Walters et al. 1991), suprathreshold pulses evoked more action potentials in axotomized sensory neurons than controls (Fig. 3B). The response of the axotomized sensory neuron illustrated in Fig. 3B to a 0.8-nA pulse shows the typical high frequency of firing at the beginning of the pulse and then the decrease in frequency that occurs as the cell undergoes spike frequency adaptation. A two-way ANOVA with repeated measures revealed significant effects of axotomy, test current level, and their interaction (P < 0.0001 in each case). Multiple comparisons showed that axotomized cells fired more spikes at all test currents from 0.6 to 1.8 nA (P < 0.001 in each case). Furthermore, a planned comparison using a paired t-test at 0.2 nA suggested a difference at the lowest stimulation current used (P = 0.04). We also used the same excitability tests as used in previous studies (Clatworthy and Walters 1994; Gunstream et al. 1995; Walters et al. 1991) and found that axotomized cells fired significantly more spikes when tested either with a 1-s pulse set at 2.5 times the threshold current determined with 20-ms pulses or determined with a 2-nA pulse (Table 1).

In addition, we measured rheobase, which is defined as the lowest current (in a pulse of any duration) that can elicit an action potential (Jack et al. 1975). Because the pleural sensory neurons always fire an action potential within 200 ms after the beginning of a suprathreshold pulse (usually within 50-100 ms), the 2-s pulses used in the study shown in Fig. 3 were more than long enough to reveal rheobase. We estimated rheobase as the current halfway between the level that first elicited one or more action potentials and the preceding level during the ascending series of test currents used in this study. Axotomized sensory neurons displayed a significantly lower rheobase than control neurons (Fig. 4, P = 0.0006, paired t-test). In contrast, in this same study, the current required to reach spike threshold during brief, 20-ms pulses did not significantly differ between axotomized and control cells (Table 1). The reduction in rheobase was associated with a significant increase in the latency for spike initiation at the rheobasic threshold (Figs. 4, A and C, P = 0.003, paired t-test). This increase in latency is likely to be a consequence of the increase in  in axotomized neurons.

Axotomy reduces outward current evoked by subthreshold depolarization

To investigate the mechanisms underlying differences between axotomized and control sensory neurons in their responses to subthreshold depolarizing currents (Fig. 3A), we used two-electrode voltage-clamp methods to examine currents evoked in ASW by prolonged (400 ms) steps from 50 to 40 mV. Figure 5A shows averaged currents from 8 axotomized and 10 control sensory neurons. Subtraction of the averaged currents reveals general features of the current reduced by axotomy. There is an apparently instantaneous outward current, which then increases slowly throughout the 400-ms pulse and is followed by a slowly decaying outward tail current after termination of the pulse. Because both the increase in apparent C_in (Table 1) and previous morphological observations (Steffensen et al. 1995) suggest that the membrane area of axotomized sensory neurons increased in the pleural ganglion, we normalized the currents to C_in so that current densities rather than raw currents could be compared. Current densities measured 3 ms after initiation of the pulse ("early"), at the end of the pulse ("steady-state"), and 4 ms after termination of the pulse ("tail"), revealed that each was significantly lower in axotomized than control sensory neurons (Fig. 5B, P = 0.001, P = 0.003, and P = 0.015, respectively, paired t-tests). Similar differences between axotomized and control sensory neurons were found when absolute current magnitudes rather than current densities were compared (early current: 0.2 ± 0.1 vs. 1.3 ± 0.3 nA, P = 0.004; steady-state current: 0.4 ± 0.1 vs. 1.8 ± 0.4 nA, P = 0.008; tail current: 0.06 ± 0.06 vs. 0.5 ± 0.2 nA, P = 0.037, respectively, paired t-tests). As discussed in the following text, these properties are consistent with an axotomy-induced reduction of I_K,S.

View larger version (23K): [in this window] [in a new window]   Fig. 5. Axotomy decreases early, steady-state, and tail currents evoked by 10-mV depolarization from Vh = 50 mV. A: averages of currents measured in control and axotomized cells with the difference of the averages on the right. Early current was measured 3 ms after initiation of the step, steady-state current was measured at the end of the 400-ms step, and tail current was measured 4 ms after the end of the pulse. B: normalization of currents to apparent Cin (yielding current density). Each component was significantly reduced in axotomized cells, whether expressed as absolute currents (not shown; see text) or current densities. The average holding current (Ih) in these and other voltage-clamp studies was not significantly changed by axotomy, although there was less dispersion of Ih values in axotomized cells (C, see text).

We also examined the holding currents (I_h) necessary to maintain soma V_m at 50 mV. No significant difference between axotomized and control cells was found in mean I_h. However, there was less dispersion of I_h values in these cells (Fig. 5C). Indeed, axotomized cells required significantly less current to maintain the soma at 50 mV when comparisons were made of the absolute values of the holding currents (P = 0.04, paired t-test). This observation is consistent with the significant difference in R_in and relatively small difference in RMP between axotomized and control cells (Table 1, and see DISCUSSION), and suggests that the V_h of 50 mV is relatively close to the "natural" RMP in both the axotomized and control cells under these conditions.

Outward current reduced by axotomy has the voltage dependence and pharmacological properties of I_K,S

The axotomy-sensitive outward current that was evoked by a small depolarization in ASW resembled I_K,S (Fig. 5). To investigate this outward current further, we used a combination of drugs that has been used previously with these cells to partially isolate I_K,S: 150 µM TTX to block fast Na⁺ current (I_Na), 1.5 mM 4-AP to block delayed, voltage-dependent K⁺ current (I_K,V), and 50 mM TEA to block calcium-activated K⁺ current (I_K,Ca) as well as I_K,V (Baxter and Byrne 1989; Klein et al. 1982; Walsh and Byrne 1989). In addition, in both this study and the study shown in Fig. 6, the fast, transient K⁺ current (I_K,A) was inactivated by holding the cell at 50 mV (Baxter and Byrne 1989; Klein et al. 1982). Other subtypes of I_K,A in Aplysia (although not the fastest subtype, which appears to be the prominent subtype in the sensory neurons) would, if present, be blocked at least partially by the 1.5 mM 4-AP (Furukawa et al. 1992). Therefore under these conditions outward current in these cells should represent primarily I_K,S (see DISCUSSION). Figure 6 shows averaged currents from six axotomized and eight control sensory neurons in response to a range of 2-s voltage steps from 50 mV. Again, subtraction of the averaged currents reveals general features of the current reduced by axotomy. With depolarizing steps there is an apparently instantaneous early outward current, which then increases slowly over several hundred milliseconds and is maintained at a nearly constant level for the duration of the 2-s pulse. In addition, a large inward tail current observed in control cells during steps to +20 mV was eliminated following axotomy (Fig. 6A). This is a presumptive Ca²⁺ current because it was measured under conditions that permit Ca²⁺ influx, whereas voltage-gated Na⁺ channels should have been blocked by TTX. Its properties will be described elsewhere.

View larger version (21K): [in this window] [in a new window]   Fig. 6. Axotomy alters an outward current with the properties of S-type current (IK,S). Currents in response to 2-s pulses from Vh = 50 mV were measured in the presence of 150 µM TTX, 1.5 mM 4-aminopyridine (4-AP), and 50 mM TEA, which block Na+ current and major voltage-dependent K+ currents in Aplysia sensory neurons. Each trace shows the averaged currents (top) and averaged voltage responses (bottom) at the indicated clamp potentials for control cells and for axotomized cells at lower gain (A) and higher gain (B). During each pulse, early current was measured after 3 ms and steady-state current after 1.95 s. The difference currents (right) display the slow activation and lack of inactivation characteristic of IK,S. In addition, an inward tail current following steps to +20 mV was dramatically reduced by axotomy.

Figure 7 plots the absolute current magnitudes and current densities in axotomized and control cells as functions of V_m for the steady-state and early outward currents. For both currents and current densities, two-way ANOVA with repeated measures on the steady-state values showed significant effects of axotomy, test potential, and their interaction (P < 0.0001 in each case). Multiple comparisons with Bonferroni post tests showed that axotomized cells exhibited lower steady-state currents and current densities at all test potentials from 0 to 30 mV (P < 0.001 in each case). In addition, a planned comparison using a paired t-test at 40 mV revealed a significant difference in current density at that membrane potential as well (P = 0.045, 1-tailed). Differences were also found in the early outward current (Fig. 7B): 2-way ANOVA with repeated measures showed significant effects of axotomy, test potential, and their interaction for both absolute current magnitudes and current densities (P < 0.01 in each case). Multiple comparisons with Bonferroni post tests showed that axotomized cells exhibited lower late currents at 20 mV (P < 0.01). In addition, planned comparisons using paired t-tests at 80 and 20 mV revealed significant differences at these membrane potentials as well (P = 0.025 and 0.025, respectively, 1-tailed). A prior study using 100-ms pulses rather than 2-s pulses, but with the same drugs and same test potentials (n = 8 axotomized cells and 12 control cells), yielded an identical pattern of statistically significant results and was used as the basis for the planned comparisons in the study shown with 2-s pulses (Fig. 7). These results indicate that axotomy reduces both a leakage component of I_K,S and a voltage-sensitive component of I_K,S. In addition, the finding of reduced I_K,S in TTX suggests that ongoing, activity-dependent release of neuromodulators such as 5-HT is not necessary for expression of axotomy-induced LTH.

View larger version (30K): [in this window] [in a new window]   Fig. 7. Axotomy causes a significant decrease in both a voltage-sensitive, steady-state component (top) and an early leakage component (bottom) of IKS. I-V curves are plotted for the cells whose averaged responses are shown in Fig. 7. During each pulse early current was measured after 3 ms and steady-state current after 1.95 s. Responses are shown as both absolute current magnitudes (A) and current densities (B), which have been normalized to apparent Cin because of the likelihood that total membrane area near the soma of axotomized cells is greater than that of control cells (see text).

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

The present study has shown that axotomy causes long-lasting alterations of passive properties of Aplysia sensory neurons, including RMP, R_in, , and, perhaps, membrane capacitance. Most importantly, it has demonstrated that a major mechanism contributing to long-term hyperexcitability and to the changes in RMP, R_in, and  is a persistent reduction in I_K,S.

Long-term, axotomy-induced changes in excitability

The present study has confirmed previous studies (Ambron et al. 1996; Bedi et al. 1998; Clatworthy and Walters 1994; Gunstream et al. 1995; Walters et al. 1991) in showing that, after axotomy, Aplysia sensory neurons fire more action potentials in response to standardized depolarizing test stimuli applied to the soma than do control neurons (Table 1). Extending earlier studies, we have shown that axotomy causes a profound reduction in rheobase, which has been the most commonly observed electrophysiological consequence of axotomy in mammalian DRG neurons (Abdulla and Smith 2001b; Gallego et al. 1987; Gurtu and Smith 1988; Kim et al. 1998; Liu et al. 2000; Stebbing et al. 1999; Zhang et al. 1997). Between RMP (approximately 50 mV) and spike threshold (approximately 35 mV) I_K,S may be particularly important for opposing depolarizing inputs because the other major outward currents characterized in these cells are inactivated, outside the voltage range for effective activation, or relatively inactive in the absence of Ca²⁺ influx that occurs during action potentials (see following text). During measurements of rheobase (Fig. 4) and repetitive firing (Fig. 3) using prolonged depolarizing pulses, we found that action potentials are not initiated until ~40 ms after initiation of the pulse (~80 ms in axotomized cells), which is long enough for activation of slower, voltage-gated outward currents. A decrease in a current, such as I_K,S, that has a prominent voltage-dependent component as well as a leakage component can explain the large decrease in rheobase in axotomized cells (Table 1). Moreover, because the voltage-dependent component is slowly activating (Figs. 4 and 5), a decrease in I_K,S also can explain the much weaker effect of axotomy on spike threshold measured with 20-ms pulses than with 200-ms pulses because the leakage component is smaller than the voltage-dependent component near threshold. A practical implication of this finding is that long pulses (>100 ms) should be used to test possible alterations of spike threshold in cells expressing I_K,S.

Long-term, axotomy-induced changes in passive properties

Reduction of an outward current that is partially active at rest, such as I_K,S, would, in the absence of any other changes, be expected to depolarize RMP. Although we found a statistically significant reduction in RMP when we pooled data from two studies (Table 1), this effect appears to be relatively weak. Previous published (Ambron et al. 1996; Clatworthy and Walters 1994; Gunstream et al. 1995; Walters et al. 1991) and unpublished studies from this lab have generally failed to find statistically significant effects of nerve injury on RMP. Nevertheless, in every study known to us the mean RMP of axotomized Aplysia sensory neurons has been slightly depolarized compared with control sensory neurons. A weak trend toward depolarization of RMP in axotomized sensory neurons is also implicit in classes of mammalian dorsal root ganglion (DRG) cells, in which similar small differences in RMP were observed by Abdulla et al. (2001b). A weak effect of axotomy on RMP would be expected if axotomy-sensitive leakage currents constitute a minor part of the currents generating the RMP or if changes in some leakage currents are compensated for by opposing changes in other resting currents.

Consistent with the weakness of axotomy's effect on RMP, we found no significant difference in the mean holding current required to keep sensory neurons voltage-clamped at 50 mV. On the other hand, the absolute value of the holding currents was reduced significantly in the axotomized cells (Fig. 5C). This indicates that axotomized and control sensory neurons show little difference from each other in the proportions of cells with RMPs negative to and positive to this holding potential but that less current is necessary to bring V_m of axotomized cells to the holding potential of 50 mV. This is consistent with the significant increase in R_in caused by axotomy (Fig. 1, Table 1). All of these effects, including the slight resting depolarization, could, in principle, be explained by a decrease in a leakage conductance to K⁺. Interestingly, axotomy has recently been shown to reduce leakage current in mammalian DRG cells (Abdulla and Smith 2001a). However, axotomized Aplysia sensory neurons as well as axotomized DRG cells have revealed inconsistent effects on R_in. In the present study, a clear increase in R_in was found, but increases in R_in observed in previous studies have not always been statistically significant (Clatworthy and Walters 1994; Gunstream et al. 1995). Similarly, in axotomized DRG cells, a significant increase in R_in was found in one study (Kim et al. 1998) but not in others (Czeh et al. 1977; Gallego et al. 1987; Gurtu and Smith 1988; Zhang et al. 1997). Three factors are likely to contribute to variation in measured R_in. One is inadvertent leakage resulting from impalement with sharp microelectrodes (see Abdulla and Smith 2001b). In the present study, this was minimized by setting relatively conservative lower limits for acceptable values of R_in, RMP, and spike amplitude. A second factor is the error introduced by using a single electrode to both inject current and record voltageeven with careful bridge balance, R_in measurements through a single electrode can be imprecise. The present study is the first in Aplysia or mammals to test R_in in axotomized sensory neurons with two independent intracellular electrodes. The third factor is the voltage range in which R_in is measured. In the present study, the largest differences in R_in were found with the smallest hyperpolarizing pulses (0.02 nA). Previously reported R_in measurements in axotomized Aplysia sensory neurons were made with only a single test pulse current of either 0.5 or 1.0 nA (Clatworthy and Walters 1994; Gunstream et al. 1995).

The axotomy-induced increase in R_in appears to involve a persistent decrease in a leakage conductance at RMP. Measurements in ASW (Fig. 5) and in solutions containing ion channel blockers (Figs. 6 and 7) revealed an apparently instantaneous component of current that was reduced significantly by axotomy. Because this early current was measured at a time (3 ms after initiation of the pulse) and from a holding potential (50 mV) at which voltage-dependent K⁺ or Cl currents in Aplysia neurons are unlikely to show significant activity (see following text), the early current probably represents instantaneous current flow through open leakage channels in response to the step in potential.

A previously unreported effect of axotomy is an increase in . This effect is readily apparent during the slowed relaxations of V_m that follow brief subthreshold pulses (Fig. 2). The increase in  probably accounts for the doubling of spike latency at rheobase in axotomized cells (Fig. 4). Because  depends on the product of R_in and C_in, the increase in R_in after axotomy accounts for at least part of the increase in . We also found evidence, however, that apparent C_in increases following axotomy. An increase in membrane capacitance would be consistent with the likelihood that the total amount of membrane that can be charged by current injection into the soma increases because of injury-induced growth of regions of the cell electrotonically close to the soma. Steffensen et al. (1995) and Dulin et al. (1995) provided morphological and electrophysiological evidence for extensive sprouting of axotomized sensory neurons within the pleural ganglion. Some of the observed growth was in the neuropil directly underneath the sensory neuron somata. Such growth would be likely to contribute to measurements of C_in even if substantial amounts of neuropil are cut away for voltage-clamp experiments. A limitation, however, of our C_in measurements is that they assumed the capacity transient is fit by a single exponential. We were unable to characterize quantitatively the goodness of fit during these measurements and noticed small deviations from single exponentials at the end of some of the transients, suggesting that incompletely clamped compartments (presumably in fine branches remaining in the neuropil) (see Nazif et al. 1991) may have introduced some error. The axotomy-induced increase in membrane resistance might have increased the space constant and allowed more of the sensory neuron to contribute to C_in measurements. If so, some of the apparent increase in C_in would not represent an actual change in membrane capacitance. Despite these reservations, the increase in this estimate of membrane capacitance after axotomy is important to note because, in combination with the morphological observations of Steffensen et al. (1995), it suggests that axotomy-induced alterations in passive properties are probably not produced solely by a decrease in I_K,S.

Axotomy causes a long-term reduction in I_K,S

The most important conclusion of this report is that many of the long-term effects of axotomy on the passive properties and excitability of sensory neurons can be accounted for, at least in part, by long-lasting reduction of I_K,S. Such alterations include a depolarized RMP, increased R_in, increased , enhanced depolarization during subthreshold pulses, reduced rheobase, and reduced spike accommodation. This conclusion follows from voltage-clamp experiments demonstrating, first, that prolonged depolarizing steps produced an outward current that had an instantaneous (leakage) component as well as a slowly activating, noninactivating component, as has been described for I_K,S (Baxter and Byrne 1989; Klein et al. 1982; Pollock and Camardo 1987; Pollock et al. 1985; Shuster and Siegelbaum 1987; Shuster et al. 1991; Siegelbaum et al. 1982). Second, a separate study has shown that this axotomy-sensitive outward current is carried largely by K⁺ ions because the current has a reversal potential of approximately 75 mV and is blocked by intracellular Cs⁺ ions (Gasull and Walters, unpublished observations). Third, both components of the axotomy-sensitive outward current were expressed under conditions used previously to isolate I_K,S from other K⁺ currents in these cells: holding the cells at 50 mV to block I_K,A, and bathing the cells in 1.5 mM 4-AP, which blocks I_K,V, and 50 mM TEA, which blocks I_K,Ca as well as I_K,V (Baxter and Byrne 1989; Klein et al. 1982; Walsh and Byrne 1989). The currents we measured probably underestimate the magnitude of I_K,S because 50 mM TEA partially blocks I_K,S (K_D for external TEA = 90 mM) (Shuster and Siegelbaum 1987). K⁺ currents other than I_K,S that have been described in Aplysia sensory neurons show little or no activation at potentials near the holding potential of 50 mV, are not activated by hyperpolarizing pulses, and show relatively slow activation (compared with leakage currents) during depolarizing pulses (Baxter and Byrne 1989; Klein et al. 1982; Walsh and Byrne 1989). The one fast outward current that might contribute in this voltage range, the transient K⁺ current, I_K,A, is inactivated in these cells at the holding potential of 50 mV (Baxter and Byrne 1989; Furukawa et al. 1992).

As with any voltage-clamp study on neurons with complex geometries, one must consider whether artifacts related to inadequate space clamp could account for the results. We attempted to minimize possible differences in space clamp by severing the sensory neuron axons near their somata either directly underneath the cluster or at the base of the pedal-pleural connective. Given the number of fine processes near the soma (Nazif et al. 1991) and some deviations of the late phase of the capacity transients from single exponentials, we cannot rule out small but potentially significant contributions of incompletely clamped regions to our measurements of current. However, the alterations in currents we observed after axotomy cannot be accounted for by an improvement in space clamp (which would be produced by the increase in R_in). An improved space clamp should increase voltage-gated outward currents during large depolarizing steps, whereas the observed effect of axotomy was to decrease outward currents. This decrease was clearly robust, whether expressed as absolute current magnitudes or normalized to C_in, our estimate of membrane capacitance. Although there are possible errors associated with C_in, it is important to note the differences in estimated current densities because the two groups are likely to differ in total membrane area near the soma (Steffensen et al. 1995).

A distinctive property of I_K,S (and current carried by potentially homologous TREK-1/KCNK2 channelssee following text) is its depression by cAMP (Baxter and Byrne 1989, 1990; Goldsmith and Abrams 1992; Klein et al. 1980; Pollock et al. 1985; Shuster et al. 1985; Siegelbaum et al. 1982). The outward current we measure in Aplysia sensory neurons is also depressed by cAMP under control conditions and after axotomy (Gasull and Walters, unpublished observations), providing further evidence that it is, in fact, I_K,S. Interestingly, another potential background current exists in these cells that is turned off by cAMP (Buttner 1992): a hyperpolarization-activated, inwardly rectifying Cl current that was first described by Chesnoy-Marchais (1982, 1983) in Aplysia cerebral neurons. Properties of this Cl current suggest, however, that it does not normally play a major role in shaping passive properties of the sensory neuron or in regulating excitability. First, its activation depends on elevation of the intracellular Cl concentration; Buttner did not detect the current in the sensory neurons except when intracellular Cl levels were elevated by dialysis during whole cell voltage clamp. Second, even with Cl loading, this voltage-dependent current is most active at potentials considerably more hyperpolarized than normal RMP in the sensory neurons. Furthermore, we have found that large changes in extracellular Cl concentration do not reduce the differences observed between electrophysiological properties of axotomized and control sensory neurons; indeed, axotomy appears to increase rather than decrease this Cl current (Gasull and Walters, unpublished observations). Therefore alteration of this or other Cl currents is unlikely to contribute to the axotomy-induced reduction of outward current described in the present report.

Together, these data indicate that axotomy causes a persistent decrease in current through S-type K⁺ channels (Pollock and Camardo 1987; Shuster and Siegelbaum 1987; Shuster et al. 1991; Siegelbaum et al. 1982). A decrease in leakage through open S-type K⁺ channels would account for the decrease in apparently instantaneous early current following axotomy (Figs. 5A and 6). In addition, because open S-type K⁺ channels show outward (Goldman-Hodgkin-Katz) rectification, a decrease in leakage through these channels would be consistent with the decrease in early current seen in the nonlinear portion of the I-V curve shown in Fig. 7B. Some voltage-gated S-type K⁺ channels may also be open at RMP. Although closure of these channels by hyperpolarization might account for a very small part of the sag shown in Fig. 1, we found little evidence for time-dependent current on hyperpolarization from 50 mV (Fig. 6). Other currents, perhaps including hyperpolarization-activated Cl current (Chesnoy-Marchais 1982, 1983; Gasull and Walters, unpublished observations) and hyperpolarization-activated cation currents such as I_H (Abdulla and Smith 2001a), are probably more important than I_K,S for sag during hyperpolarizing steps from normal RMP.

Recently, a two-pore-domain K⁺ channel, called TREK-1 (Fink et al. 1996; Patel et al. 1998) or KCNK2 (Goldstein et al. 2001), has been cloned from mammals and shown to be an open rectifier with properties that are remarkably similar to those of S-type K⁺ channels (Patel et al. 1998). The widespread distribution of TREK-1 in mammalian CNS and nerves (Bearzatto et al. 2000; Fink et al. 1996; Maingret et al. 2000; Medhurst et al. 2001), combined with the present results, raises the possibility that axotomy-induced changes in the expression or regulation of this channel may have broad significance. Given the magnitude of the change in I_K,S after axotomy of Aplysia sensory neurons, the similarity between properties of S-type K⁺ channels and mammalian TREK-1 channels (Patel et al. 1998), and the presence of TREK-1 channels in rat sciatic nerve (Bearzatto et al. 2000), an important question is whether axotomy-induced regulation of channels homologous with those carrying I_K,S in Aplysia contributes to the decrease in rheobase (and decrease in leakage current) observed in mammalian neurons. Conversely, the similarity of TREK-1 channels to S-type K⁺ channels raises interesting questions about mechanisms underlying the persistent reduction of I_K,S in Aplysia. For example, like S-type K⁺ channels in Aplysia (Shuster et al. 1991), TREK-1 channels are often open at rest but also show voltage-dependent opening (Patel et al. 1998). An important recent discovery was that cAMP-dependent phosphorylation can reversibly convert TREK-1 channels from a voltage-independent leak channel to a voltage-gated channel that only opens in response to depolarization (Bockenhauer et al. 2001). This suggests an interesting extension of the ability of cAMP-dependent protein kinase (PKA) to close S-type channels in Aplysia sensory neurons (see also Braha et al. 1990; Goldsmith and Abrams 1992; Scholz and Byrne 1988; Shuster et al. 1985; Siegelbaum et al. 1982). Axotomy might persistently alter the dynamics of conversion between different forms of this K⁺ channel. Nerve injury also induces a long-lasting increase in PKA activity (Liao et al. 1999), so PKA might act continuously to close S-channels and to maintain them in a voltage-gated state. In addition, recent observations on the effects of cAMP and other signals on I_K,S suggest that axotomy might cause a long-term reduction in the number of available S channels. Interestingly, decreased expression of some K⁺ channels after axotomy has been reported in mammalian DRG neurons (Ishikawa et al. 1999).

Multiple currents may be altered by axotomy

Several currents are reported to be reduced in mammalian DRG neurons following axotomy, including Ca²⁺ current and voltage-gated K⁺ currents similar to I_K,V and I_K,A (Abdulla and Smith 2001a; Everill and Kocsis 1999). Preliminary evidence indicates that similar effects occur in axotomized sensory neurons of Aplysia. A large inward tail current during steps to +20 mV was eliminated by axotomy (Fig. 6A). This is likely to be a Ca²⁺ current because it was measured under conditions that permit Ca²⁺ influx but block voltage-gated Na⁺ channels. Effects of axotomy on other electrophysiological properties of these cells, such as action potential duration (Walters et al. 1991; Gasull and Walters, unpublished observations), suggest that additional K⁺ currents may be altered in these cells as well. Finally, recent evidence suggests that hyperpolarization-activated Cl current is increased by axotomy (Gasull and Walters, unpublished observations).

Functional implications of decreased I_K,S

The decrease in I_K,S and consequent increase in excitability following axotomy may have adaptive significance (see Walters 1994). First, depression of I_K,S will increase the safety factor for conduction. Failure of action potential conduction can occur both centrally and peripherally in these cells (Clatworthy and Walters 1993a). Without compensating changes in membrane conductance, injury-induced growth within the CNS (Steffensen et al. 1995) and in the periphery (Billy and Walters 1989) might greatly increase the probability of conduction block (by decreasing R_in and increasing C_in). The chances of conduction block should be particularly high in thin regenerating axons in the periphery (Steffensen et al. 1995). However, the observation that the threshold for spike initiation decreases in regenerating receptive fields (Billy and Walters 1989; Dulin et al. 1995) suggests that the decrease in I_K,S we have observed in the soma may also occur near peripheral terminals of these mechanosensory neurons. This location suggests a second function that should be important under natural conditions, where partial axotomy within a sensory neuron's receptive field can easily be produced by trauma to the animal's soft body. In these cases, a decrease in I_K,S would decrease spike threshold and increase the number of spikes generated peripherally by mechanical stimuli, thereby contributing to nociceptive sensitization in the injured area. The same effects within central regions of the sensory neuron, including the soma, could lead to centrally generated afterdischarge (Clatworthy and Walters 1993b) during responses to peripheral stimuli, further enhancing nociceptive sensitization. Finally, decreased I_K,S may contribute to the presynaptic facilitation (Byrne and Kandel 1996; Klein et al. 1982) that is thought to persist for long periods following noxious peripheral stimulation (Cleary et al. 1998; Frost et al. 1985; Walters 1987) or nerve injury (Clatworthy and Walters 1994; Walters et al. 1991).