Characterization of action potential-evoked calcium transients in mouse postganglionic sympathetic axon bundles

Abstract

1. Action potential-evoked Ca²⁺ transients in postganglionic sympathetic axon bundles in mouse vas deferens have been characterized using confocal microscopy and Ca²⁺ imaging.

2. Axonal Ca²⁺ transients were tetrodotoxin sensitive. The amplitude depended on both the frequency of stimulation and the number of stimuli in a train.

3. Removal of extracellular Ca²⁺ abolished the Ca²⁺ transient. Cd²⁺ (100 μm) inhibited the Ca²⁺ transient by 78 ± 10 %. The N-type Ca²⁺ channel blocker ω-conotoxin GVIA (0.1 μm) reduced the amplitude by −35 ± 4 %, whereas nifedipine (10 μm; L-type) and ω-conotoxin MVIIC (0.1 μm; P/Q type) were ineffective.

4. Caffeine (10 mm), ryanodine (10 μm), cyclopiazonic acid (30 μm) or CCCP (10 μm) had no detectable effects.

5. Blockade of large and small conductance Ca²⁺-dependent K⁺ channels with iberiotoxin (0.1 μm) and apamin (1 μm), respectively, or Ca²⁺-dependent Cl⁻ channels by niflumic acid (100 μm) did not alter Ca²⁺ transients.

6. In contrast, the non-specific K⁺ channel blockers tetraethylammonium (10 mm) and 4-aminopyridine (10 mm) markedly increased the amplitude of the Ca²⁺ transient. Blockade of delayed rectifiers and A-like K⁺ channels, by tityustoxin-K (α) (0.1 μm) and pandinustoxin-K (α) (10 nm), respectively, also increased the Ca²⁺ transient amplitude.

7. Thus, Ca²⁺ transients are evoked by Na⁺-dependent action potentials in axons. These transients originate mainly from Ca²⁺ entry through voltage-dependent Ca²⁺ channels (80 % Cd²⁺ sensitive of which 40 % was attributable to N-type). Twenty per cent of the Ca²⁺ transient was not due to Ca²⁺ entry through voltage-gated Ca²⁺ channels. Intracellular stores and mitochondria were not involved in the generation of the transient. Ca²⁺ transients are modulated by A-like K⁺ channels and delayed rectifiers (possibly K_V1.2) but not by Ca²⁺-activated ion channels.

A sympathetic neurone consists of a well-defined cell body which gives rise to an axon which, after a variable length (1 cm to 1 m), repeatedly branches into a complex network of varicosities separated by thin intervaricose regions. Recently, a technique has been developed to measure action potential-evoked changes in Ca²⁺ levels in sympathetic nerve terminals (Brain & Bennett, 1997). Using confocal microscopy and Ca²⁺ imaging, the authors characterized action potential-evoked Ca²⁺ transients in individual varicosities and in the intervaricose region of postganglionic sympathetic nerve terminals. Ca²⁺ transients have also been detected in cell bodies of sympathetic neurones (Hua et al. 1993). What is less well known is that Ca²⁺ transients are elicited by action potentials in the main axonal trunks of neurones. Axonal studies of Ca²⁺ dynamics have mainly been carried out in the central nervous system i.e. rat cerebellar interneurones (Callewaert et al. 1996; Forti et al. 2000), cortical neurone cultures (Mackenzie et al. 1996), neonatal (Kriegler & Chiu, 1993; Lev-Ram & Grinvald, 1987; Sun & Chiu, 1999) and adult rat optic nerve (Brown et al. 2001). Axonal Ca²⁺ transients have also been detected in adult rat preganglionic vagus nerve trunks in the peripheral nervous system (Wächtler et al. 1998). However, to date, no action potential-evoked Ca²⁺ transients have been reported in the main axon trunks of unmyelinated autonomic neurones in regions far removed from the secretory terminals.

The paucity of data on Ca²⁺ dynamics in axons is due mainly to their small diameter (0.1–1 μm; Cottee et al. 1996) and their inaccessibility to conventional electrophysiological techniques such as the patch clamp. What little is known about axonal ion channel activity in postganglionic sympathetic neurones innervating rodent vas deferens is based on experiments where the action potential was monitored using extracellular recording techniques (Cunnane & Stjärne, 1984; Brock & Cunnane,1987,1988; Stjärne & Stjärne, 1989). It is noteworthy that removal of extracellular Ca²⁺ significantly reduced the amplitude of the nerve action potential in axons of sympathetic neurones innervating rodent vas deferens (Cunnane & Stjärne, 1984). In addition, tetraethylammonium chloride (TEA), a non-specific blocker of Ca²⁺-activated K⁺ channels, alters the configuration of the nerve action potential (Cunnane & Stjärne, 1984). One possible role for Ca²⁺ influx is to regulate the frequency of action potential firing in these nerves perhaps by activating Ca²⁺-dependent K⁺ channels.

The aims of the present study were: (1) to characterize physiologically, action potential-evoked Ca²⁺ transients in sympathetic axons using confocal microscopy and Ca²⁺ imaging techniques; (2) to establish the origin of the Ca²⁺ transient; (3) to determine whether axonal Ca²⁺ transients are modified by blockers of Ca²⁺-activated K⁺ and Cl⁻ channels; and (4) to determine whether axonal Ca²⁺ transients are modified by other voltage-dependent K⁺ channels.

METHODS

A schematic illustration of the cell body, main axon and varicose secretory terminals of a postganglionic sympathetic axon is shown in Fig. 1.

Figure 1. Diagram of a postganglionic sympathetic neurone.

A, schematic diagram of a postganglionic sympathetic neurone, showing the difference in size between cell body, axon and varicosity. B, image of a set of axon bundles loaded with the Ca²⁺ indicator Oregon Green 488 BAPTA-1 dextran. Note that only a single axon is represented in the corresponding box in A. C, image of part of a sympathetic terminal loaded with the same Ca²⁺ indicator.

Ca²⁺ indicator loading

Aganglionic vasa deferentia were removed from 8- to 12-week-old Balb/C mice, which had been humanely killed by cervical fracture. The bathing Krebs solution contained (mm): 118.8 NaCl, 25 NaHCO₃, 1.13 NaH₂PO₄, 4.7 KCl, 1.8 CaCl₂, 1.2 MgCl₂, 11.1 glucose and was constantly gassed with 95 % O₂-5 % CO₂ to pH 7.4. The cut prostatic end of each vas deferens was gently sucked into a glass pipette containing a saturated solution of Oregon Green 488 BAPTA-1, dextran linked with a molecular mass of 10 kDa (Molecular Probes Inc., OR, USA), in 2.5 % Triton X-100. Vasa deferentia were loaded for 3 h in the dark at room temperature to allow time for the indicator to travel along axons, and then washed for a further 2 h to remove any extracellular dye.

Stimuli and recording

The prostatic end of a vas deferens was mounted in a 2 ml organ bath and placed on the stage of a Leica TCS NT laser-scanning confocal microscope. The preparation was secured with a pair of parallel platinum electrodes and stimulated using an optically isolated stimulator (Digitimer DS2). The pulse width was set between 0.06 and 0.4 ms and the applied voltage was adjusted to give a reliable change in fluorescence following a single stimulus. The stimulus voltage was then increased by about 20 % to ensure that the stimulus was suprathreshold for that axon bundle thus preventing variation in the Ca²⁺ transient due to the recruitment of unstimulated axons in the bundle. The stimuli were electronically synchronized with the confocal microscope scans. The 488 nm wavelength of an Argon ion laser was used for exciting fluorescence. A 515 nm long pass emission filter was used. When detecting Ca²⁺ transients, sets of images were captured for 56 s every 3 min. This protocol prevented excessive photobleaching and phototoxicity. All experiments were carried out at 33 °C in the presence of nifedipine (10 μm; an L-type Ca²⁺ channel blocker), prazosin (1 μm; a competitive α₁-adrenoceptor antagonist) and α,β-methylene ATP (1 μm; a P_2X-receptor desensitizing agent) to reduce or abolish contractions elicited by high frequency stimulation.

Stock solutions of tetrodotoxin (TTX), α,β-methylene ATP, ω-conotoxin GVIA, ω-conotoxin MVIIC, tityustoxin-K (α), pandinustoxin-K (α), bepridil and iberiotoxin were dissolved in distilled water. Prazosin, ryanodine, carbonyl cyanide m-chlorophenyl hydrazone (CCCP) and cyclopiazonic acid (CPA) were dissolved in dimethyl sulfoxide, and apamin in 5 % acetic acid. Solutions were prepared and aliquoted before storing at −20 °C; this ensures that the drugs only passed through one freeze-thaw cycle. Stock nifedipine (in ethanol), niflumic acid (in methanol) and cadmium chloride (in distilled water) solutions were diluted as required on the day of the experiment. Caffeine, TEA and 4-aminopyridine (4-AP) were all made up fresh each day and diluted in Krebs solution to the required final bath concentration. The pH of 4-AP was titrated to pH 7.4 with 1 m HCl. CPA was obtained from Calbiochem (Nottingham, UK) and all other compounds were obtained from Sigma (Dorset, UK). Vehicle controls had a final concentration < 0.01 % and had no detectable effect on control axonal Ca²⁺ transients.

Analysis

Images were analysed using either Scion or NIH Image (available from URL http://www.rsb.info.nih.gov/nih-image) and custom-written macros. As Oregon Green 488 BAPTA-1 is a non-ratiometric dye, changes in intracellular Ca²⁺ concentration are reported as the change in fluorescence divided by the basal fluorescence (ΔF/F). An example of a typical region analysed is highlighted in yellow in Fig. 2A. Data are presented as means ± standard error of the mean (s.e.m.); n is the number of axon bundles analysed, all of which were obtained from > 3 vas deferens from > 3 animals. Statistical significance was determined using Student's paired t tests. In the figures, *P < 0.05, **P < 0.01 and ***P < 0.005.

Figure 2. Effect of TTX on action potential-evoked axonal Ca²⁺ transients.

Six example scans are shown; a prestimulus control frame (0.0 s) and frames 0.4, 0.8, 1.2, 1.6 and 2.0 s after the beginning of stimulation (10 stimuli applied at 10 Hz, 0.06 ms, 15 V). The green bars indicate the frames during which the stimuli were applied. The width of each frame is 40 μm. To the right of each panel is a graph showing the change in fluorescence over 50 s, including the period of electrical stimulation, in a single bundle (indicated by the yellow area in the first frame). B as A, 20 min after the bath application of TTX (0.3 μm). C shows a recording 30 min after wash out of TTX. Small differences between bundles may be due to small changes in the focal plane. D, histogram shows the mean data ( ± s.e.m.) from 15 axon bundles.

RESULTS

Characterization of Ca²⁺ transients

Effects of TTX

Small nerve trunks, which typically contain 2–10 unmyelinated axons (Cottee et al. 1996), were loaded with Oregon Green 488 BAPTA-1 dextran. The Ca²⁺ transients evoked by a train of 10 stimuli at 10 Hz, in a set of axon bundles, are shown in Fig. 2A. The change in fluorescence was abolished by TTX (0.3 μm; Fig. 2B) and returned after wash out of TTX (Fig. 2C). The mean inhibition of the amplitude of Ca²⁺ transients evoked by 10 stimuli at 10 Hz in the presence of TTX is shown in Fig. 2D (-106 ± 4 %, n = 15 axon bundles; P < 0.001). Thus the Ca²⁺ transients detected in axonal bundles are elicited by Na⁺-dependent, nerve action potentials.

Effects of increasing the number of stimuli in a train

Ca²⁺ transients could be detected in sympathetic axon bundles in response to even a single action potential (Fig. 3A). The amplitude of the Ca²⁺ transients could vary markedly from axon bundle to axon bundle, presumably due to either variable dye loading or physiological differences. Increasing the number of stimuli from 1 to 5, 10, 20, 30 or 50 at 10 Hz induced a progressive increase in the magnitude of the evoked Ca²⁺ transient (Fig. 3A; 1 stimulus, 0.1 ± 0.03; 3 stimuli, 0.2 ± 0.1; 5 stimuli, 0.3 ± 0.1; 10 stimuli, 0.6 ± 0.2; 20 stimuli, 0.7 ± 0.1; 30 stimuli, 0.8 ± 0.2; and 50 stimuli, 0.7 ± 0.2; n = 16 axon bundles). The evoked Ca²⁺ transient was less than that observed in varicosities (1 stimulus, 0.8 ± 0.2; 3 stimuli, 1.4 ± 0.2; 5 stimuli, 1.5 ± 0.2; 10 stimuli, 1.9 ± 0.3; 20 stimuli, 1.9 ± 0.2; 30 stimuli, 1.9 ± 0.1; and 50 stimuli, 2.1 ± 0.3, n = 19 varicosities). The mean fluorescence change observed with increasing number of action potentials at 10 Hz (n = 16 axon bundles) is shown in Fig. 4A. Experiments were limited to 50 stimuli and/or 50 Hz because of movement artifacts. It was evident that the maximum change in fluorescence was evoked by about 30 stimuli at 10 Hz. When the relative change in fluorescence was plotted against the number of stimuli in a train, a non-linear relationship resulted (Fig. 4A).

Figure 3. Effects of stimulus frequency and train length.

A, Ca²⁺ transients evoked at a constant frequency of 10 Hz but with a variable number of stimuli (1, 5, 10, 20, 30 and 50). A change in fluorescence was observed even in response to a single stimulus. Note that significant broadening of the Ca²⁺ transient occurred after 20–50 stimuli. Inspection of the Ca²⁺ transient evoked by 50 stimuli shows that the signal rose rapidly for about 1200 ms before reaching a plateau that was maintained throughout the period of stimulation. The bars indicate the periods of nerve stimulation. B, Ca²⁺ transients evoked by trains of 10 stimuli applied at frequencies of 1, 2, 5, 10, 20 and 50 Hz.

Figure 4. Plots of relative changes in fluorescence against stimulus frequency, train length and the effects of ionomycin and 4-AP.

A, plot of relative changes in fluorescence against number of stimuli at a constant frequency of 10 Hz (n = 16). Error bars represent the s.e.m. A non-linear relationship was revealed with a maximum change after 30 stimuli before a plateau was reached. B, plot of relative changes in fluorescence against stimulation frequency, n = 22 (except 2 Hz, n = 15). Again, a non-linear relationship was revealed with a maximum change observed at about 20 Hz before the plateau was reached. C, plot of relative changes in fluorescence after a control period of 10 stimuli at 50 Hz and 90 min after the application of ionomycin (10 μm; top trace). The dashed line represents the maximum output of the photomultiplier tube (i.e. 256, the maximum fluorescence output of an 8-bit detection system). Clearly, the magnitude of the evoked Ca²⁺ transient is not limited by dye saturation. D, effects of 4-AP on Ca²⁺ transients evoked by trains of 10 stimuli at 10 Hz (n = 15). 4-AP (10 mm; ▪) produced a threefold increase in the amplitude of the Ca²⁺ transient compared to control (⋄). Hence, K⁺ channels regulate the Ca²⁺ axonal transient and the non-linearities observed in Fig. 3A and B are not a consequence of dye saturation. E, lack of effect of 4-AP on the kinetics of recovery of the Ca²⁺ transient. Same data as in D, but the error bars have been removed and the control response scaled to match the 4-AP trace and superimposed.

Effects of increasing the stimulation frequency

The effects of varying the frequency of nerve stimulation but keeping the number of stimuli in a train constant were also investigated. Increasing the frequency of stimulation from 1 to 2, 5, 10, 20 and 50 Hz (10 stimuli), increased the fluorescence signal in all axon bundles studied (Fig. 3B; 10 stimuli at 1 Hz, 0.19 ± 0.04; 2 Hz, 0.30 ± 0.06; 5 Hz, 0.48 ± 0.05; 10 Hz, 0.52 ± 0.07; 20 Hz, 0.63 ± 0.07; 50 Hz, 0.61 ± 0.11, n = 15 axon bundles). The evoked Ca²⁺ transient was smaller than that observed in varicose regions (10 stimuli at 1 Hz, 1.5 ± 0.7; 2 Hz, 1.2 ± 0.1; 5 Hz, 1.8 ± 0.2; 10 Hz, 2.0 ± 0.2; 20 Hz, 2.5 ± 0.2; 50 Hz, 3 ± 0.2, n = 19 varicosities). A plot of the relative change in fluorescence against frequency (n > 15 axon bundles) is shown in Fig. 4B. A plot of ΔF/F versus the frequency of stimulation revealed a non-linear relationship, the Ca²⁺ transient reaching a maximum value after a train of 10 stimuli at 20 Hz.

Is the axonal Ca²⁺ transient amplitude limited by dye saturation?

To investigate whether the peak response was limited by dye saturation, the Ca²⁺ ionophore ionomycin (10 μm) was used. Ionomycin produced a peak fluorescence that was greater than that produced by nerve stimulation (Fig. 4C; control 10 stimuli at 50 Hz, 129 ± 16; ionomycin, 225 ± 6, n = 3 axon bundles). Furthermore, the application of the K⁺ channel blocker 4-AP (10 mm) also increased the action potential-evoked Ca²⁺ transient about threefold (Fig. 4D; 10 stimuli at 10 Hz, n = 15 axon bundles; P < 0.001) with no change in the time course of the signal (Fig. 4E). Taken together, these data strongly suggest that the non-linearity of the Ca²⁺ transient, observed during repetitive stimulation (Fig. 4), is unlikely to be due to dye saturation, and indicate that K⁺ channel activation might limit the magnitude of axonal Ca²⁺ transients.

Origin of Ca²⁺ transient

Evidence for Ca²⁺ influx

The potential sources of Ca²⁺ involved in the genesis of the axonal Ca²⁺ transients were investigated. When tissues were perfused with Krebs solution containing no added Ca²⁺ and with the Ca²⁺ chelator EGTA (1 mm) present, the evoked axonal Ca²⁺ transient was abolished. Thus, the presence of extracellular Ca²⁺ is obligatory for the generation of the Ca²⁺ transient. A histogram summarizing the effects of Ca²⁺-free solution on axonal Ca²⁺ transients evoked by trains of 10 stimuli at 10 Hz (-98 ± 4 %, n = 9 axon bundles; P < 0.001) is shown in Fig. 5A. It is noteworthy that in addition to abolishing the evoked axonal Ca²⁺ transient, removal of extracellular Ca²⁺ also reduced the basal fluorescence (-28 ± 5 %, n = 10 axon bundles; P < 0.005).

Figure 5. Effects of Ca²⁺-free Krebs solution and of Cd²⁺ on action potential-evoked axonal Ca²⁺ transients.

A, effects of Ca²⁺-free Krebs solution (EGTA 1 mm) on the Ca²⁺ transient evoked by 10 stimuli at 10 Hz in a single axon bundle. The histogram shows the mean data ( ± s.e.m.) from 9 axon bundles in different preparations. B, effects of the non-specific Ca²⁺ channel blocker Cd²⁺ (100 μm) on the Ca²⁺ transient evoked by 10 stimuli at 10 Hz in a single axon bundle. The histogram shows the mean inhibition of the Ca²⁺ transient produced by Cd²⁺ in 8 axon bundles ( ± s.e.m.).

Are voltage-gated calcium channels the route of Ca²⁺ entry into the axon?

The most likely route of Ca²⁺ entry into axons is through voltage-gated Ca²⁺ channels (VGCCs). This potential route of Ca²⁺ entry was investigated using the non-specific VGCC blocker Cd²⁺ (100 μm). A sample trace showing the inhibitory effect of Cd²⁺ on the Ca²⁺ transient evoked in a single axonal bundle (10 stimuli at 10 Hz) is shown in Fig. 5B. The inhibitory effect of Cd²⁺ was not readily reversible upon wash. A histogram summarizing the mean effect of Cd²⁺ on action potential-evoked Ca²⁺ transients (10 stimuli at 10 Hz, −78 ± 10 %, n = 8 axon bundles; P < 0.01) is shown in Fig. 5B. Interestingly, about 20 % of the fluorescence signal remained 1 h after the application of Cd²⁺. It should be pointed out that the Na⁺–Ca²⁺ exchanger may be blocked by Cd²⁺ (Trosper & Philipson, 1983; Hobai et al. 1997) so these results by themselves cannot distinguish between Ca²⁺ influx through axonal VGCCs and reverse mode Na⁺–Ca²⁺ exchange. However, the Na⁺–Ca²⁺ exchange blocker bepridil (100 μm) had no detectable effect on the axonal Ca²⁺ transient indicating that Ca²⁺ entry into the axons was mainly through VGCCs (10 stimuli at 10 Hz, 2 ± 14 %, n = 5 axon bundles; P > 0.05).

Pharmacological characterization of VGCCs

The L-type Ca²⁺ entry blocker nifedipine (10 μm) had no detectable effects on axonal Ca²⁺ transients evoked by a single stimulus (20 ± 18 %, n = 5 axon bundles; P > 0.05). The relatively high concentration of nifedipine employed here may also inhibit low-threshold T-type VGCCs (Richard et al. 1991; Takahashi & Akaike, 1991). All subsequent experiments were carried out with nifedipine present in the Krebs solution, to minimize smooth muscle contraction during high frequency stimulation. The most common VGCCs expressed in sympathetic nerve terminals, and shown to be involved in transmitter release, are the N-, P- and Q-types (Smith & Cunnane, 1997; Brain & Bennett 1997). The N-type Ca²⁺ channel blocker ω-conotoxin GVIA (0.1 μm) inhibited, but did not abolish, the axonal Ca²⁺ transient (Fig. 6A). Alternative routes of Ca²⁺ entry were investigated by using relatively high concentrations of ω-conotoxin MVIIC (0.1 μm), which blocks P/Q-type VGCCs. ω-Conotoxin MVIIC had no detectable effect on the axonal Ca²⁺ transient even when applied for 1 h or more (Fig. 6B; 10 stimuli at 10 Hz, 10 ± 12 %, n = 14 axon bundles; P > 0.05). To establish whether different types/populations of VGCCs were recruited when the frequency of stimulation was varied, the effects of N-, P- and Q-type VGCC blockers were investigated over a wide range of stimulation frequencies (1, 10 and 50 Hz). ω-Conotoxin GVIA produced the same degree of inhibition at each stimulation frequency employed (Fig. 6C; 1 Hz, −43 ± 14 %; 10 Hz, −35 ± 4 %; 50 Hz, −42 ± 6 %; n = 10 axon bundles; P < 0.005). Similarly, ω-conotoxin MVIIC was ineffective at all stimulation frequencies employed (Fig. 6C; 1 Hz, −2 ± 21 %; 10 Hz, 26 ± 14 %; 50 Hz, 17 ± 13 %; n = 9 axon bundles; P > 0.05).

Figure 6. Effects of ω-conotoxin GVIA and ω-conotoxin MVIIC on axonal Ca²⁺ transients.

A, effects of the selective N-type VGCC blocker ω-conotoxin GVIA (0.1 μm) on the Ca²⁺ transient evoked by 10 stimuli at 10 Hz in a single axon bundle. The toxin inhibited the transient by ∼40 %, even after 1 h of exposure. B, effects of the P/Q-type VGCC blocker ω-conotoxin MVIIC (0.1 μm) on the Ca²⁺ transient evoked by 10 stimuli at 10 Hz in a single axon bundle. The toxin had no significant effect even after 1 h of exposure. C, histogram showing the effects of VGCC blockers on Ca²⁺ transients evoked at different stimulation frequencies. The inhibitory effects of ω-conotoxin GVIA (0.1 μm) were independent of stimulation frequency (1 Hz, −43 ± 14 %; 10 Hz, −35 ± 4 %; 50 Hz, −42 ± 6 %; n = 10 axon bundles). ω-Conotoxin MVIIC was ineffective at all stimulation frequencies employed (1 Hz, −2 ± 21 %; 10 Hz, 26 ± 14 %; 50 Hz, 17 ± 13 %; n = 9 axon bundles).

Possible role of intracellular Ca²⁺ stores

Although removal of extracellular Ca²⁺ abolished the axonal Ca²⁺ transient, these experiments did not rule out the potential involvement of Ca²⁺-induced Ca²⁺ release (CICR). Ryanodine acts in a use- and time-dependent manner at the ryanodine receptor (Nagasaki & Fleischer, 1988) to either lock the channel in a sub-conductance state or to block it (Rousseau et al. 1987; McPherson et al. 1991). Nerves were therefore stimulated continually for 60–90 min in the presence of ryanodine to ensure that any potential ryanodine receptors present were activated by Ca²⁺ entry and therefore blockable by ryanodine (Smith & Cunnane, 1996). Under these experimental conditions, ryanodine (10 μm) did not alter the axonal Ca²⁺ transient even when long trains of stimuli were applied. The mean effects of ryanodine on the amplitude of the axonal Ca²⁺ transient elicited by several stimulation protocols were: 1 stimulus, −8 ± 25 %, n = 5 axon bundles; 10 stimuli at 10 Hz, 9 ± 47 %, n = 5 axon bundles; 50 stimuli at 10 Hz, −1 ± 37 %, n = 6 axon bundles, P > 0.05. Caffeine was also employed to deplete putative axonal Ca²⁺ stores. Caffeine (10 mm) had no significant effects on evoked axonal Ca²⁺ transients (1 stimulus, −16 ± 13 %; 10 stimuli at 10 Hz, 3 ± 16 %; 50 stimuli at 10 Hz, 2 ± 12 %; n = 9 axon bundles; P > 0.05). Ryanodine and caffeine had no detectable effect on basal fluorescence (8 ± 16 %, n = 9 axon bundles and 4 ± 15 %, n = 6 axon bundles, respectively; P > 0.05). Similarly, high concentrations of the Ca²⁺-ATPase inhibitor CPA (30 μm) had no detectable effect on the axonal Ca²⁺ transient (-7 ± 9 %, n = 8 axon bundles; P > 0.05) or on the basal fluorescence (10 ± 10.5 %, n = 9 axon bundles; P > 0.05).

To investigate whether mitochondria play a role in modulating the axonal Ca²⁺ transient, the effects of CCCP were investigated. CCCP (10 μm) had no effect on the axonal Ca²⁺ transient even after 30 min exposure (10 stimuli at 1 Hz, −12 ± 12 %; 10 Hz, −10 ± 13 %; 50 Hz, −10 ± 17 %; n = 11 axon bundles; P > 0.05), nor on the basal fluorescence (12 ± 29 %, n = 11 axon bundles; P > 0.05).

Do Ca²⁺-activated K⁺ and/or Cl⁻ channel blockers modulate axonal Ca²⁺ transients?

To determine whether Ca²⁺-activated K⁺ channel blockers could affect the Ca²⁺ transient, the non-selective K⁺ channel blocker TEA was used. TEA (10 mm) produced a large potentiation of the action potential-evoked axonal Ca²⁺ transient as shown in Fig. 7. As it has been suggested that activation of K⁺ currents in axons might depend on the stimulation frequency employed (see Hille, 1992), the effects of TEA (10 mm) were tested over a wide range of stimulation frequencies (10 stimuli at 1, 10 and 50 Hz). TEA increased the amplitude of the evoked Ca²⁺ transient at all frequencies of stimulation studied (Fig. 7; 10 stimuli at 1 Hz, 116 ± 32 %, n = 16 axon bundles; 10 Hz, 89 ± 38 %, n = 18 axon bundles; 50 Hz, 39 ± 17 %, n = 10 axon bundles; P < 0.005). TEA had no effect on the basal fluorescence (-1 ± 7 %, n = 13 axon bundles; P > 0.05).

Figure 7. Effect of TEA on axonal Ca²⁺ transients.

Effects of the K⁺ channel blocker TEA (10 mm) on the Ca²⁺ transient evoked by 10 stimuli at 10 Hz in a single axon bundle. Histogram showing the mean ( ± s.e.m.) increase in amplitude of the Ca²⁺ transient produced by TEA following trains of 10 stimuli applied at 1, 10 and 50 Hz from n = 16, 18 and 10 bundles, respectively.

As TEA is believed to block large conductance Ca²⁺-activated K⁺ channels, the specific blocker iberiotoxin was used to characterize further the nature of the K⁺ channel involved. Iberiotoxin (0.1 μm) had no effect on the action potential-evoked Ca²⁺ transients, at any stimulation frequency employed (Fig. 8; 10 stimuli at 1 Hz, 7 ± 9 %; 10 Hz, −2 ± 6 %; 50 Hz, 9 ± 10 %; n = 11 axon bundles; P > 0.05). Iberiotoxin had no detectable effect on the basal fluorescence (2 ± 7 %, n = 11 axon bundles; P > 0.05).

Figure 8. Effect of iberiotoxin on axonal Ca²⁺ transients.

A, graphs showing the change in fluorescence in a single axon bundle when a stimulus was applied (trains of 10 stimuli) at 1, 10 and 50 Hz (from left to right in the figure). The period of electrical stimulation is shown by the bar. The response in the absence (▴) and presence (□) of iberiotoxin (0.1 μm) is shown. B, histogram shows the mean ( ± s.e.m.) for trains of 10 stimuli applied at 1, 10 and 50 Hz from 11 axon bundles. Iberiotoxin had no significant effect.

To test whether the action potential-evoked Ca²⁺ transient was affected by small conductance Ca²⁺-activated K⁺ channels, the specific blocker apamin was applied. High concentrations of apamin (1 μm) failed to affect the evoked Ca²⁺ transient (10 stimuli at 1 Hz, 25 ± 22 %, n = 6 axon bundles; 10 Hz, −3 ± 10 %, n = 6 axon bundles; 50 Hz, −5 ± 7 %, n = 4 axon bundles; P > 0.05). Apamin had no effect on the basal fluorescence (7 ± 7 %, n = 9 axon bundles; P > 0.05).

As specific Ca²⁺-activated K⁺ channel blockers did not modify the action potential-evoked axonal Ca²⁺ transients, the possible involvement of Ca²⁺-activated Cl⁻ channels was investigated using niflumic acid. Niflumic acid (100 μm) had no detectable effect on the Ca²⁺ transient evoked by 10 stimuli at 10 Hz (-9 ± 13 %, n = 6 axon bundles; P > 0.05). Niflumic acid did not detectably alter the basal fluorescence (6 ± 8 %, n = 6 axon bundles; P > 0.05).

Pharmacological characterization of the K⁺ channels modulating axonal Ca²⁺ transients

The data obtained with iberiotoxin suggest that TEA does not potentiate the axonal Ca²⁺ transient by blocking large conductance Ca²⁺-activated K⁺ channels. One possibility is that TEA exerts its effects on K⁺ channels that are also sensitive to 4-AP. It is interesting to note that 4-AP (10 mm) also potentiated Ca²⁺ transients evoked by 10 stimuli at 10 Hz (Fig. 9). Like TEA, the effects of 4-AP were observed at all stimulation frequencies studied (Fig. 9; 1 stimulus, 662 ± 126 %, n = 13 axon bundles, P < 0.005; 10 stimuli at 1 Hz, 430 ± 217 %, n = 8 axon bundles, P < 0.01; 10 Hz, 238 ± 45 %, n = 18 axon bundles, P < 0.005; 50 Hz, 106 ± 50 %, n = 8 axon bundles, P < 0.01). 4-AP had no effect on the basal fluorescence (4 ± 6 %, n = 18 axon bundles; P > 0.05).

Figure 9. Effects of 4-AP on action potential-evoked Ca²⁺ transients.

Effects of the K⁺ channel blocker 4-AP (10 mm) on the Ca²⁺ transient evoked by 10 stimuli at 10 Hz in a single axon bundle. Histogram showing the mean ( ± s.e.m.) increase in amplitude of the Ca²⁺ transient produced by 4-AP following a single stimulus, or trains of 10 stimuli applied at 1, 10 and 50 Hz from n = 8–18 bundles.

4-AP is known to block both the transient A-like K⁺ current (I_A) and delayed rectifiers (I_V). Therefore, the selective antagonist for I_A, pandinustoxin-K (α), and the selective antagonist for the slow inactivating I_V, tityustoxin-K (α), were studied (Juhng et al. 1999). Pandinustoxin-K (α) (10 nm) potentiated the action potential-evoked axonal Ca²⁺ transients (Fig. 10). The mean effect of pandinustoxin-K (α) (10 stimuli at 1 Hz, 54 ± 13 %, n = 7 axon bundles, P < 0.005; 10 Hz, 48 ± 5 %, n = 6 axon bundles, P < 0.005; 50 Hz, 58 ± 23 %, n = 6 axon bundles, P < 0.01) is shown in Fig. 10B. Pandinustoxin-K (α) had no effect on the basal fluorescence (2 ± 2 %, n = 6 axon bundles; P > 0.05).

Figure 10. Effect of pandinustoxin-K (α) on action potential-evoked Ca²⁺ transients.

A, change in fluorescence in a single axon bundle when a stimulus was applied (trains of 10 stimuli) at 1, 10 and 50 Hz (from left to right in the figure). The period of electrical stimulation is shown by the bar. The response in the absence (▴) and presence (□) of pandinustoxin-K (α) (10 nm) is shown. B, histogram shows the mean ( ± s.e.m.) for trains of 10 stimuli applied at 1, 10 and 50 Hz from 7, 6 and 6 axon bundles, respectively. Pandinustoxin-K (α) significantly increased the amplitude of the evoked response.

Like pandinustoxin-K (α), tityustoxin-K (α) (100 nm) also potentiated the action potential-evoked axonal Ca²⁺ transients evoked by 10 stimuli at 1, 10 and 50 Hz (Fig. 11). Inspection of the histogram in Fig. 11B shows that the effect of tityustoxin-K (α) was independent of stimulation frequency (10 stimuli at 1 Hz, 75 ± 43 %, n = 11 axon bundles, P < 0.01; 10 Hz, 72 ± 31 %, n = 16 axon bundles, P < 0.005; 50 Hz, 42 ± 12 %, n = 13 axon bundles, P < 0.005). Tityustoxin-K (α) had no effect on the basal fluorescence (1 ± 3 %, n = 11 axon bundles; P > 0.05).

Figure 11. Effect of tityustoxin-K (α) on action potential-evoked Ca²⁺ transients.

A, change in fluorescence in a single axon bundle in response to trains of 10 stimuli at 1, 10 and 50 Hz (from left to right in the figure). The period of electrical stimulation is shown by the bar. The response in the absence (▴) and presence (□) of tityustoxin-K (α) (100 nm) is shown. B, histogram shows the mean data ( ± s.e.m.) for trains of 10 stimuli applied at 1, 10 and 50 Hz from 11, 16 and 13 axon bundles, respectively. Tityustoxin-K (α) significantly increased the amplitude of the evoked response.

DISCUSSION

The present study shows that Ca²⁺ transients are evoked in the main axon of postganglionic sympathetic nerve fibres (see Fig. 1) by single or brief trains of action potentials. Axonal Ca²⁺ transients may well be a general phenomenon of mammalian axons, having been detected in axons in the central nervous system (Lev-Ram & Grinvald, 1987; Kriegler & Chiu, 1993; Callewaert et al. 1996; Mackenzie et al. 1996; Sun & Chiu, 1999; Forti et al. 2000; Brown et al. 2001) and in preganglionic parasympathetic axons (Wächtler et al. 1998).

The first question to address is whether action potential-evoked Ca²⁺ transients occur in axon bundles or Schwann cells. Unlike previous studies with Ca²⁺ indicators in axons (for example, Lev-Ram & Grinvald, 1987; Kriegler & Chiu, 1993; Sun & Chiu, 1999), the present experiments involved the use of a dextran-conjugated dye. This membrane-impermeable dye cannot penetrate a series of sequential Schwann cells and therefore the evoked fluorescence changes report Ca²⁺ transients within axons rather than in Schwann cells.

Characterization of axonal Ca²⁺ transients

When the number of stimuli in a train was increased, the axonal Ca²⁺ transient reached a stable plateau after about 30 stimuli. Our findings differ from those obtained in single axons of rat cerebellar Purkinje neurones (Callewaert et al. 1996), where an increase in the number of evoked-action potentials produced an increase in the amplitude of the axonal Ca²⁺ transient that was directly proportional to the number of stimuli in a train (although only the first 20 stimuli were tested). A hyperbolic relationship was found in cell bodies of rabbit vagal afferent neurones by Cohen et al. (1997) who concluded that the plateau phase was due to CICR. However, this is unlikely to be the case in the present experiments as ryanodine, caffeine and CPA failed to affect the axonal Ca²⁺ transient.

The plateau phase does not appear to result from dye saturation as the Ca²⁺ ionophore ionomycin produced about a fivefold greater increase in fluorescence. This is more than twice that evoked by maximal stimulation. Furthermore, the K⁺ channel blocker 4-AP markedly increased the amplitude of the Ca²⁺ transient without altering its kinetics of recovery. Taken together, these findings suggest that the plateau phase of the Ca²⁺ transient is determined by biological mechanisms. It is possible that this plateau is where Ca²⁺ influx is in equilibrium with Ca²⁺ extrusion mechanisms.

What is the origin of the action potential-evoked axonal Ca²⁺ transient?

Removal of extracellular Ca²⁺ has previously been shown to reduce the amplitude of the extracellularly recorded nerve action potential in axon bundles innervating rodent vas deferens (Cunnane & Stjärne, 1984). In the present study, removal of extracellular Ca²⁺ abolished the axonal Ca²⁺ transient, showing that the genesis of the Ca²⁺ transient requires extracellular Ca²⁺ and is wholly dependent on Ca²⁺ influx. It must also be noted that the axonal basal [Ca²⁺]_i was reduced in Ca²⁺-free conditions, as there is now a change in the electrochemical gradient for Ca²⁺.

How does Ca²⁺ enter the neurone?

There are three main potential routes of Ca²⁺ entry into neurones: Ca²⁺ entry through VGCCs, by reverse mode Na⁺–Ca²⁺ exchange, or by Ca²⁺ entry through voltage-dependent Na⁺ channels.

Classification of VGCCs involved in action potential-evoked axonal Ca²⁺ transients

The axonal Ca²⁺ transient was reduced by about 80 % by the non-selective VGCC blocker Cd²⁺ making it likely that most of the Ca²⁺ enters through VGCCs. Ca²⁺ entry into axons of rat cerebellar Purkinje neurones (Callewaert et al. 1996) is abolished by ω-agatoxin IVA indicating Ca²⁺ entry through P-type VGCCs. In our experiments, Ca²⁺ transients were insensitive to P/Q-type Ca²⁺ channel blockers. In neonatal rat optic nerve axons, 58 % of the Ca²⁺ influx was through N-type Ca²⁺ channels (Sun & Chiu, 1999) but in adult rat optic nerve axons, only L-type Ca²⁺ channels were found to be present (Brown et al. 2001). Our studies suggest that 40 % of the Ca²⁺ enters through N-type VGCCs with no entry through L-type channels, similar to that in neonatal rat optic nerve axons. These findings suggest that the types of VGCCs expressed in neurones can be very different in both central and peripheral axons and may well be neurone specific and dependent on its stage of development. It seems likely that the type and relative distribution of VGCCs will have important functional consequences.

The relative contributions of N- and P/Q-type VGCCs to transmitter release differ according to the stimulation frequency used to activate the sympathetic nerves innervating the mouse vas deferens (Waterman, 1997). In the present study the effects of N- and P/Q-type VGCC blockers on the action potential-evoked axonal Ca²⁺ transient were independent of stimulation frequency.

Cd²⁺-resistant Ca²⁺ entry

It is possible that the residual 20 % Cd²⁺-resistant component of the Ca²⁺ transient results from Ca²⁺ entry through ‘conventional’ Na⁺ channels. Baker et al. (1971) previously reported that depolarization of the squid axon led to a leakage of Ca²⁺ into the axon; a small increase (∼1 %) was detected and it was concluded that Ca²⁺ entered through voltage-dependent Na⁺ channels. It is also possible that Ca²⁺ enters neurones through Na⁺ channels operating in a ‘slip-mode’ conductance state (Santana et al. 1998). An alternative, plausible explanation is that Ca²⁺ enters through non-selective cation channels, which are difficult to investigate pharmacologically.

Ca²⁺ influx in rat CNS white matter appears to proceed via reverse mode Na⁺–Ca²⁺ exchange during anoxia (Stys et al. 1992). However, in our experiments bepridil, which blocks Na⁺–Ca²⁺ exchange, had no effect on the Ca²⁺ transients. In addition, the lack of effect of bepridil on the action potential-evoked Ca²⁺ transient indicates that the Na⁺–Ca²⁺ exchanger is not responsible for Ca²⁺ extrusion from postganglionic sympathetic axons. An alternative likely candidate is the plasma membrane Ca²⁺-ATPase.

Intracellular stores

Ca²⁺ entry into some neurones can be amplified by CICR from ryanodine-sensitive intraneuronal stores. This mechanism has been shown to operate in sympathetic ganglia (Lipscombe et al. 1988; Hua et al. 1993) and in sympathetic nerve terminals (Smith & Cunnane, 1996). However, our data suggest that CICR does not occur in sympathetic axons. Thus, there seem to be clear regional differences in the expression of CICR mechanisms in neurones. Interestingly, Sun & Chiu (1999) also reported that pharmacological manipulations of internal Ca²⁺ stores with ryanodine and thapsigargin had no significant effect on the axonal Ca²⁺ transients in the rat optic nerve.

Similarly, on the basis of the experiments with CCCP, we could find no evidence for the role of mitochondria in mediating the axonal Ca²⁺ transients even with high frequency stimulation.

Which K⁺ channel(s) is involved in membrane repolarization during an action potential?

It has previously been shown, using focal extracellular recording in rodent vas deferens, that TEA prolongs the nerve terminal impulse, an effect believed to result from the blockade of Ca²⁺-dependent K⁺ channels (Stjärne et al. 1991; Brock & Cunnane, 1995). Similar conclusions regarding the action of Ca²⁺-dependent K⁺ channels have been reported using optical measurements from rat optic nerve axons (Lev-Ram & Grinvald, 1987). In the present study, TEA increased the amplitude of the axonal Ca²⁺ transient. However, it would appear that TEA was not acting on large conductance Ca²⁺-activated K⁺ channels, as the specific blocker iberiotoxin had no effect on the axonal Ca²⁺ transient. Apamin, which blocks small conductance Ca²⁺-activated K⁺ channels, was also ineffective. It is possible that TEA acts on 4-AP-sensitive, delayed rectifier channels (see Hille, 1992). Like TEA, 4-AP increased the amplitude of the Ca²⁺ transient at all stimulation frequencies employed. It follows that either delayed rectifier (K_V) or A-like (K_A) K⁺ channels (or both) are involved in regulating the configuration of the nerve action potential and hence the magnitude of the axonal Ca²⁺ transient.

K_V and K_A channels are normally isolated and studied electrophysiologically. However, in the mouse vas deferens, the axon bundles are too small to permit patch-clamp studies. As a result, the contribution of K_V and K_A channels to repolarization following an action potential was investigated pharmacologically by using the purified toxins pandinustoxin-K (α) and tityustoxin-K (α). Pandinustoxin-K (α) preferentially blocks the rapidly inactivating K_A channels (Rogowski et al. 1996) while tityustoxin-K (α) is selective for slowly inactivating delayed rectifier type channels (Werkman et al. 1993). The current study shows that K_A and K_V channels contribute to axonal membrane repolarization as both toxins increased the action potential-evoked axonal Ca²⁺ transient, an action that was independent of stimulation frequency. However, it must be noted that tityustoxin-K (α) has an IC₅₀ of 210 pm for blockade of K_V 1.2 K⁺ channels (Werkman et al. 1993), and similarly pandinustoxin-K (α) has been reported to block K_V 1.2 K⁺ channels, but with an IC₅₀ of 32 pm (Rogowski et al. 1996) raising the possibility that they may both inhibit K_V 1.2 in sympathetic axons.

Possible functions for action potential-evoked axonal Ca²⁺ transients

It is possible that the Ca²⁺ channels are simply ‘in transit’ from the cell body to the terminal. However, there is no evidence in postganglionic neurones that Ca²⁺ channels migrate in this manner. It is likely that Ca²⁺ channels are packaged in the cell body and transported to different regions of the nerve where they are functionally expressed. Based on our data, a number of points can be addressed. First, if the types and numbers of Ca²⁺ channels differ significantly between the terminal varicosities and main axons, that would indicate that the axonal channels are not simply ‘in transit’. In axons and varicosities about ∼50 % of the action potential-evoked Ca²⁺ transient is due to N-type Ca²⁺ channels, i.e. sensitive to ω-conotoxin GVIA and insensitive to blockers of P/Q-type Ca²⁺ channels. However, it is interesting to note that there are Cd²⁺-resistant Ca²⁺ channels in axons but not in varicosities (Brain & Bennett, 1997; Jackson, 2000). Second, the axonal Ca²⁺ channels are tightly regulated by K⁺ channels that are sensitive to 4-AP indicating that they may play an important physiological role.

In axons, the physiological role of Ca²⁺ influx following an action potential is unclear. A possible role would be to activate K⁺ channels, which in turn would contribute to repolarization of the action potential. TEA, which is often used as a blocker of large conductance Ca²⁺-activated K⁺ channels, greatly potentiated the action potential-evoked Ca²⁺ transient. In contrast, the specific blocker of large conductance Ca²⁺-activated K⁺ channels iberiotoxin had no effect. Inhibition of the small conductance Ca²⁺-activated K⁺ channels with apamin also failed to affect the Ca²⁺ transient. It is noteworthy that blockade of Ca²⁺-activated Cl⁻ channels with niflumic acid had no effect. The results indicate that the influx of Ca²⁺ following an action potential is not modulated by Ca²⁺-dependent ion channel activity in the axon membrane.

It has been suggested that intracellular Ca²⁺ acts to regulate the safety factor for action potential propagation in cultured rat dorsal root ganglion neurones (Lüscher et al. 1996). However, action potential propagation in postganglionic sympathetic nerves innervating rodent vas deferens does not fail during repetitive stimulation (Brock & Cunnane,1987, 1992).

The axonal Ca²⁺ transient may drive axonal transport as it has been demonstrated that high concentrations of Ca²⁺ (Chan et al. 1980) increase axonal active transport. Llinás et al. (1989) and Breuer et al. (1992) report evidence that fast axonal transport may be strictly Ca²⁺ dependent. If similar mechanisms operate in sympathetic axons, then Ca²⁺ entry into axons following action potentials may be the source of Ca²⁺ utilized to drive axonal transport.

It is possible that there is non-exocytotic Ca²⁺-dependent release of neurotransmitter from axons. For example, action potentials release NO in a Ca²⁺-dependent fashion from all regions (including axons) of some neurones (Wiklund et al. 1997). Therefore, it is possible that substances released from axons (Weidmann, 1994) might activate receptors on Schwann cells. ATP is a likely candidate as it produces intracellular Ca²⁺ transients in Schwann cells by activating purinergic (P_2Y) receptors (Jahromi et al. 1992; Lyons et al. 1994; Robitaille, 1995; Mayer et al. 1997). Whether this occurs in sympathetic axons is purely speculative.

The reported findings may have greater, or at least more direct, implications for nerve cell biology than for sympathetic system physiology. In addition to the issues raised above, the presence of different types/subtypes of calcium channels in different types of axons has implications for calcium channel sorting and trafficking mechanisms in main axons and varicosities.

Conclusions

The present experiments show that action potentials evoke Ca²⁺ transients in postganglionic sympathetic nerve axons. Axonal Ca²⁺ transients are mainly generated by Ca²⁺ entry through both N-type VGCCs, pharmacologically uncharacterized VGCCs and an unidentified Cd²⁺-resistant mechanism. Intracellular Ca²⁺ stores were not involved. The amplitude of the axonal Ca²⁺ transient is strongly regulated by K⁺ channels and pharmacological evidence suggests that K_A and K_V (possibly K_V 1.2) channels are involved. No convincing evidence was found to suggest that Ca²⁺-dependent K⁺ or Ca²⁺-dependent Cl⁻ channel activity occurs in main axons.