Responses of nerve fibres of the rat saphenous nerve neuroma to mechanical and chemical stimulation: an in vitro study

Abstract

1. The response of neuroma nerve endings to different stimuli was studied in a saphenous nerve neuroma preparation in vitro.

2. Electrical activity was recorded from 141 single fibres dissected of saphenous nerve. One-third (27 %) displayed spontaneous activity. Based on their response to mechanical and chemical stimuli, neuroma nerve fibres were classified as mechanosensory fibres (47.5 %), mechanically insensitive chemosensory fibres (17.0 %), polymodal nociceptor fibres (28.4 %) and unresponsive fibres (7.1 %).

3. Mechanosensory and polymodal neuroma endings responded to von Frey hair stimulation either with a few impulses (phasic units) or a sustained discharge (tonic units). Polymodal units were additionally activated by at least one of the following stimuli: acidic solutions; a combination of bradykinin, prostaglandin E₂, serotonin, substance P and histamine (all at 1 μM) plus 7 mm KCl (inflammatory soup); 600 mm NaCl and capsaicin.

4. Low pH solutions increased the firing discharge of polymodal endings proportionally to the proton concentration. The ‘inflammatory soup’ evoked a firing response characterized by the absence of tachyphylaxis, which appeared when its components were applied separately. Both stimuli sensitized polymodal fibres to mechanical stimulation. Hypertonic NaCl (600 mm) and capsaicin (3.3 mm) induced a prolonged discharge that outlasted the stimulus duration.

5. Mechanically insensitive chemosensory neuroma fibres exhibited responses to chemical stimuli analogous to polymodal fibres. They became mechanically sensitive after chemical stimulation.

6. These findings show that neuroma nerve endings in the rat saphenous nerve neuroma in vitro are functionally heterogeneous and exhibit properties reminiscent of those in intact mechanosensory, polymodal and ‘silent’ nociceptor sensory afferents, including their sensitization by algesic chemicals.

When peripheral sensory nerves are damaged, the axon distal to the section degenerates whereas the proximal portion sprouts and if regeneration towards the innervation target is impeded, forms a neuroma (Ramón y Cajal, 1913). During the regenerative process, injured afferents may develop spontaneous activity, depending on time after lesion, type of primary sensory neurone involved, and animal species (Govrin-Lippmann & Devor, 1978; Blumberg & Jänig, 1984; Meyer et al. 1985; Welk et al. 1990; Amir & Devor, 1993). Spontaneous activity appears to be generated at trigger zones either at the regenerating nerve endings (Govrin-Lippmann & Devor, 1978; Blumberg & Jänig, 1984; Meyer et al. 1985), at the site of nerve constriction (Xie et al. 1995; Tal & Eliav, 1996) or in the soma of dorsal root ganglion (DRG) neurons (Kajander et al. 1992; Babbedge et al. 1996).

Regenerating nerve endings in a neuroma are also activated by mechanical forces and/or chemical substances locally released by inflammatory cells or neighbouring autonomic efferents. Mechanical stimuli applied to the surface of neuromas of injured sciatic, saphenous or sural nerves of the rat excited both A and C fibres, the sensitivity of the response being related to the time after lesion (Blumberg & Jänig, 1984; Welk et al. 1990; Michaelis et al. 1995; Babbedge et al. 1996). Chemical sensitivity of neuroma endings to noradrenaline was reported by Wall & Gutnick (1974). Welk et al. (1990) obtained impulse responses to bradykinin, histamine and adrenaline in fibres of saphenous nerve neuromas. Similar excitatory discharges were evoked by capsaicin (Hartung et al. 1989) and by a combination of endogenous chemicals similar to those released during inflammation (Michaelis et al. 1997). Excitability changes in neuroma terminals appear to be due to the incorporation of ion channel proteins that are transported down to the regenerating terminal including voltage-sensitive Na⁺ channels (Matzner & Devor, 1994). In man, ectopic activity of neuromas has been associated with paraesthesias and pain (Torebjörk et al. 1979).

Most of the experiments studying responsiveness of regenerating sensory fibres in the neuroma to mechanical forces and chemical substances were carried out in anaesthetized animals, where the complexity of tissues surrounding the neuroma may interfere with the stimulus. The purpose of the present study was to explore the characteristics of the response to different stimuli of regenerating nerve endings in an in vitro saphenous nerve neuroma model. Some of these results have been published in abstract form (Rivera et al. 1993a,b).

Methods

Surgery

Adult rats were operated on under sodium pentobarbitone anaesthesia (Nembutal, Abbot, 50 mg kg⁻¹i.p.). Saphenous nerves were exposed at the level of the internal aspect of the mid-thigh, dissected free and tightly ligated with 8–0 silk. The nerve was cut just distal to the ligature and the cut end placed inside a 5 mm silicone tube (internal diameter 0.52 mm, external diameter 0.96 mm), to which it was fixed by means of an 8–0 silk ligature. The distal end of the tube was left open. Approximately 5 mm of the distal nerve stump was excised to prevent regeneration. The central end of the nerve with the silicone tube was replaced in its normal position and the incision closed in layers. Rats were allowed to recover postoperatively and then housed individually under standard conditions. They were treated with G-penicillin and inspected daily for infection as well as for abnormal behaviour. Some animals showed mild signs of autotomy such as gnawing nails during the observation time. Animals were treated according to the guidelines of the Universidad Miguel Hernandez Ethical Committee and the Ethical Guidelines for Investigations of Experimental Pain in Conscious Animals of the International Association for the Study of Pain.

After 1–40 weeks, rats were reanaesthetized (Nembutal, 50 mg kg⁻¹i.p.). The thoracic cavity was opened and the aorta cannulated through the left ventricle. The animal was perfused with cold (4°C) physiological salt solution (modified Tyrode solution). The composition was (mm): NaCl 140.3, KCl 5, MgCl₂ 1, CaCl₂ 2.9, glucose 5.5 and Hepes 1, adjusted to pH 7.4. The saphenous nerve was re-exposed at the level of the neuroma and dissected proximally to its point of disappearance in the inner aspect of the hip joint. Saphenous nerves were then transferred to a Petri dish containing cold salt solution bubbled with 100 % O₂. Under a dissecting microscope, the silicone tube was removed, the nerves cleaned of connective and fatty tissues and the perineurium desheathed for electrophysiological recording.

Superfusion chamber

The superfusion chamber (Fig. 1) consisted of a Lucite box (dimensions 32 mm × 13 mm × 6 mm) with two compartments separated by a thin Perspex wall and with the bottom covered with a layer of Sylgard. The neuroma end was placed in one of the compartments, carefully fixed with insect pins through the connective tissue and perfused at a flow of 3 ml min⁻¹ with physiological salt solution bubbled with 100 % O₂. Temperature was monitored with a thermocouple and maintained at 35 ± 1°C with an electronically controlled Peltier device. The proximal end of the nerve was passed to the second compartment of the Lucite chamber through a notch in the dividing wall that was sealed with grease. This compartment was filled with warm mineral oil.

Figure 1.

Schematic diagram of the recording chamber and experimental set-up.

Fine axon bundles were teased from the main nerve trunk using sharp dissecting forceps, and placed on the recording electrode. Every filament was maintained up to 3 min on the recording electrode to detect the presence of ongoing activity before proceeding to further dissection.

Monopolar recording of electrical activity was performed using a fine Ag-AgCl electrode mounted in a micromanipulator, with the reference electrode located under the neuroma. A bipolar Ag-AgCl stimulating electrode was placed on the main nerve trunk 3–4 mm proximal to the end of the neuroma (Fig. 1). Electrical activity was amplified, filtered, displayed on a storage oscilloscope and recorded on an FM tape. Recordings were analysed off-line with a window discriminator, an analog-to-digital converter (CED 1401–18 plus) and a computer.

Conduction velocity measurements

Conduction velocity was determined by stimulating the main nerve trunk with square wave pulses (0.1–0.5 ms, 0.5–3 mA). Conduction velocity was calculated from the delay of individual spikes, identified by their amplitude and shape, and the distance between the stimulating and the recording electrodes. Axons were considered to be Aδ fibres when the conduction velocity was over 2 m s⁻¹. Fibres with conduction velocity below 2 m s⁻¹ were classified as C fibres.

Mechanical stimulation

Mechanosensitivity of single units was explored by probing the neuroma surface with a small blunt glass rod (diameter 1.2 mm) and with calibrated von Frey hairs of variable force (5–40 mN).

Chemical stimulation

Chemical substances were applied either directly with a micropipette over the neuroma or by adding them to the perfusion fluid at a known concentration. The following substances were used: physiological saline solution (PSS) buffered with Mes instead of Hepes to pH 6.5, 6.0 and 5.5, applied in random order; 600 mm NaCl; 1.2 M sucrose; 3.3 mm capsaicin; ‘inflammatory soup’ composed of: bradykinin (BK), substance P (SP), prostaglandin E₂ (PGE₂), serotonin and histamine, all at 1 μM, and KCl at 7 mm (Handwerker & Reeh, 1991).

Experimental protocol

When a filament containing one or a few electrically evoked single units was isolated, the presence of spontaneous activity was ascertained. Thereafter, the neuroma surface was stimulated mechanically with a blunt glass rod and the mechanical threshold measured with von Frey hairs of increasing stiffness. The conduction velocity of the unit(s) was then measured. Chemical sensitivity was subsequently explored, applying the test substance with a washout time interval between tests of at least 10 min. Mechanical sensitivity was again explored with the von Frey hairs 5–10 min after application of the chemical agent. No more than four fibres were explored in each experiment.

Drugs

Bradykinin (Arg-Pro-Pro-Gly-Phe-Ser-Pro-Phe-Arg), capsaicin (8-methyl-N-vanillyl-6-nonenamide), Hepes, histamine, Mes, prostaglandin E₂ ([5z,11(13E,15S]-11,15-dihydroxy-9-oxoprosta-5,13-dienoic acid), substance P (Arg-Pro-Lys-Pro-Gln-Gln-Phe-Phe-Gly-Leu-Met-NH₂) and serotonin (3-(2-aminoethyl)-5-hydroxyindole) were all purchased from Sigma, Spain. NaCl, KCl, MgCl₂, CaCl₂, sucrose and glucose were obtained from Merck, Spain.

Analysis of data

Instantaneous and mean frequency of impulses of the spontaneous or evoked impulse activity were calculated. For the responses to chemical agents, the time delay between application of the test solution and the first impulse (latency of the first impulse) or the bin which had a maximum number of impulses (time to peak), maximum frequency and mean frequency (first 30 s after the 1st spike) were also measured.

Data are expressed as means ±s.e.m. (standard error of the mean). Differences between groups were determined with Student's t test and one-way ANOVA.

Results

General

Spontaneous or stimulus-evoked activity was recorded in 141 single units obtained from nerve filaments dissected from 42 neuromas. Thirty-eight units (27 %) displayed spontaneous activity at the beginning of the experiment. Sixty-seven units (47.5 %) responded exclusively to mechanical stimulation; 24 units (17.0 %) were activated only by chemicals and 40 units (28.4 %) responded to both mechanical and chemical stimuli. Ten units (7.1 %) were unresponsive to any of these stimuli.

Ongoing activity

The mean frequency of units that showed ongoing activity was 5.5 ± 0.2 impulses s⁻¹ (range 0.03–48.0). The pattern of the ongoing activity varied from an irregular discharge of single impulses (mean frequency 1.1 ± 0.05 impulses s⁻¹), present in about 70 %, to a burst-firing pattern of two or more impulses per burst, observed in the remaining spontaneously active fibres (mean frequency 18.6 ± 1.2 impulses s⁻¹). Figure 2A–C shows examples of various patterns of spontaneous nerve discharge.

Figure 2. Examples of the various patterns of spontaneous activity found in neuroma fibres.

A, continuous, irregular discharge. B, discharge in doublets. C, bursting discharge. Time scale bars: 5 s. In B and C a segment of the recording has been expanded 6 times.

Thirty-two per cent of the spontaneously active fibres were exclusively activated by mechanical forces; 42 % responded to both mechanical and chemical stimuli; 8 % were excited only by chemicals while 18 % were unresponsive to any form of stimulus.

Conduction velocity

Conduction velocity was measured in 45 units. Values ranged from 0.3 to 8.5 m s⁻¹ (Fig. 3A). Seventy per cent of the units conducted below 2 m s⁻¹ and were classified as C fibres (mean 0.8 ± 0.1 m s⁻¹, range 0.3–1.9 m s⁻¹, n = 31). The remaining units were classified as Aδ fibres (mean 4.1 ± 2.34 m s⁻¹, range 2.1–8.5 m s⁻¹, n = 14). No obvious association was found between conduction velocity and the responsiveness of the fibre to mechanical or chemical stimuli or the presence of spontaneous activity. No units conducting over 9 m s⁻¹ were found, presumably because of the short distance between the stimulating and the recording electrodes (Govrin-Lippmann & Devor, 1978; Devor & Jänig, 1981).

Figure 3. Conduction velocity and mechanical threshold of neuroma fibres.

A, histogram showing the distribution of conduction velocities of all functional types of fibres (n = 45). B, mechanical thresholds of mechanosensory (n = 64) and polymodal (n = 35) fibres.

Functional responses

Mechanosensory units

Fibres that responded exclusively to mechanical stimuli were categorized as pure mechanosensory units. Of these, 17 % showed spontaneous activity at the beginning of the experiment. Conduction velocity measured in 12 fibres was both in the C (mean 1.0 ± 0.4 m s⁻¹, range 0.3–1.7 m s⁻¹, n = 8) and the Aδ range (mean 4.9 ± 2.9 m s⁻¹, range 2.4–8.5 m s⁻¹, n = 4).

The mean force threshold was 16.3 ± 0.1 mN (range 3.9–29.4 mN, n = 64) (Fig. 3B). Three of the units were excited only by application of a blunt glass rod. Sixty per cent of the mechanosensory units n = 67) gave a single or a few nerve impulses immediately after application of a sustained indentation (phasic units, Fig. 4A, upper and middle traces). The rest (tonic units) produced a longer impulse discharge that often persisted throughout the stimulation time, sometimes presenting afterdischarge (Fig. 4A, lower trace). In tonic units, stimuli of higher intensities evoked a roughly proportional increase in the number of impulses per stimulus (Fig. 4A, lower trace). Repetition of the same stimulus at 30 s intervals often slightly increased the number of impulses per stimulus (Fig. 4B).

Figure 4. Response of mechanosensory fibres to mechanical stimulation.

A, sample recordings showing the response to sustained stimulation with von Frey hairs of a phasic fibre (conduction velocity (CV) = 1.2 m s⁻¹, upper trace), the simultaneous recruitment (middle trace) of a phasic unit (CV = 1.1 m s⁻¹) (larger amplitude unit) and a more tonic unit (CV = 0.4 m s⁻¹) (smaller amplitude unit) and the impulse discharge evoked by stimuli of increasing intensity (12 and 18 mN) in a tonic unit (CV = 1.25 m s⁻¹, lower trace). Time scale bars: 4 s, 1 s and 10 s in upper, middle and lower traces, respectively. B, effect of repeated stimulation on the firing response of mechanosensory units. Responses of mechanosensory units to a series of eight mechanical stimuli of the same intensity (18 mN) applied at 30 s (mean ±s.e.m., n = 5).

Polymodal units

Among the units exhibiting mechanosensitivity (n = 107), more than one-third (n = 40) also responded when challenged with at least one of the test chemicals (Table 1). Conduction velocity measurements performed in 16 units showed that about two-thirds were C fibres (mean 0.8 ± 0.3 m s⁻¹, range 0.4–1.1 m s⁻¹, n = 11) and the rest were Aδ fibres (mean 3.5 ± 0.9 m s⁻¹, range 2.6–4.4 m s⁻¹, n = 5). Forty percent of the polymodal units had spontaneous activity at the beginning of the experiment.

Table 1.

Recruitment by chemical stimuli of mechanically sensitive and mechanically insensitive neuroma fibres

	Mechanically sensitive	Mechanically sensitive
pH 6.5–5.5	16/22	8/14
IS	19/19	11/14
600 mm NaCl	12/12	7/7
3.3 mm capsaicin	11/12	3/5

Data are number of responding units/number of explored units. IS, inflammatory soup (see text).

Mechanical stimulation

The mean force threshold of polymodal units was 20.6 ± 0.1 mN (n = 35). This value was significantly higher than for pure mechanosensory units (range 5.9–39.2 mN, n = 35; P < 0.05,t test) (Fig. 3B). Five of the polymodal fibres (12.5 %) were recruited only by application of a blunt glass rod on the neuroma.

About half of the units gave a single or a few nerve impulses immediately after application of a sustained indentation and were thus considered phasic units. The remaining fibres were tonic and produced an impulse discharge that persisted throughout the stimulation time, often continuing at the end of the stimulus. The number of evoked impulses per stimulus increased with mechanical stimuli of growing intensity. This is illustrated in the example of Fig. 5.

Figure 5. Relationship between stimulus intensity and impulse response.

The figure shows an example of the mean firing frequency evoked by mechanical stimuli of increasing intensities in a polymodal Aδ unit (CV = 4.25 m s⁻¹ (r²= 0.99). Inset: sample recording of the neural discharge of the same unit evoked by the successive mechanical stimuli.

Acid

Sixteen out of twenty-two polymodal units in which sensitivity to acidic solutions (pH 6.5–5.5) was explored showed a positive response to protons. No difference was observed between Aδ and C fibres.

The response to acid consisted of a sustained discharge of nerve impulses with a late onset (30.2 ± 15.6 s) that outlasted the application time and persisted during the washout period generating long afterdischarges and poststimulus ongoing activity. The firing rate during the stimulus increased and the latency of the first evoked impulse discharge decreased significantly (P < 0.01, one-way ANOVA) with the lower pH values of the test solutions (Fig. 6).

Figure 6. Impulse response evoked by acidic solutions in polymodal neuroma fibres.

Peristimulus time histograms were obtained from sixteen polymodal units in response to 2 min superfusions (bars above histograms) with solutions of decreasing pH values (6.5, 6.0 and 5.5), applied with a 15 min interval at the times indicated by the bar. Bin width = 5 s. Data are means ±s.e.m. Insets: sample recordings of the neural discharge evoked in a polymodal C fibre (CV = 0.3 m s⁻¹) by the application of each acidic solution (bars above traces).

Inflammatory soup (IS)

The effect of IS was tested in 19 units, 11 of which conducted in the C fibre range. All units were excited by the IS with a response characterized by a rapid onset (8.2 ± 2.4 s), an afterdischarge and ongoing activity during the washout period. No difference was observed in the characteristics of the response after repetition of the stimulus (Fig. 7). Half of the IS-sensitive fibres were also activated by a pH 6.0 solution. Fibres that responded exclusively to IS were predominantly C fibres and gave a smaller frequency discharge after application of IS (0.9 ± 0.3 impulses (30 s)⁻¹) than those that were activated by both IS and low pH solution (2.5 ± 0.5 impulses (30 s)⁻¹), the difference in mean frequency of their impulse response being significant (P < 0.05,t test).

Figure 7. Impulse response evoked by ‘inflammatory soup’ (IS).

Peristimulus time histograms of nineteen polymodal units showing the results of three 2 min superfusions with inflammatory soup (bars above histograms), separated by 15 min intervals. Data are means ±s.e.m., bin width = 5 s. Insets: examples of the neural discharge evoked by IS (bars above traces) in a polymodal C unit (CV = 1.1 m s⁻¹) in response to IS. This unit responded to capsaicin and became mechanically sensitized after IS application.

Hypertonic NaCl

In all units tested, exposure of the neuroma to a saline hypertonic solution (600 mm NaCl) for 30 s (n = 12) produced an impulse discharge that appeared with a relatively short latency (8.8 ± 3.5 s) and reached comparatively high peak frequency values (30.1 ± 11.2 impulses s⁻¹) outlasting the stimulation time. In all units this was followed by a prolonged ongoing activity (Fig. 8A). None of these units was excited by hypertonic sucrose (1.2 m).

Figure 8. Response of polymodal fibres to hypertonic NaCl and capsaicin.

A, peristimulus time histogram of twelve polymodal units to 20 s superfusion with 600 mm NaCl as indicated by the bar. Bin width = 1 s. Inset: sample recording of the neural discharge evoked by application of 600 mm NaCl (shown by bar) in a polymodal C unit (CV = 0.85 m s⁻¹). B, peristimulus time histogram of response of 12 polymodal units to direct application of 3.3 mm capsaicin over the neuroma surface. Flow was interrupted after 20 s (bar). Data are means ±s.e.m. Bin width = 1 s. Inset: sample recording of the response of a polymodal C unit (CV = 1.1 m s⁻¹) to 3.3 mm capsaicin (applied during period shown by bar).

Capsaicin

Capsaicin (3.3 mm) was tested in 12 units that also showed sensitivity to other chemical stimuli. All but one of the units was excited by capsaicin. The response had a fast onset (2.9 ± 1.7 s) and consisted of an irregular discharge of impulses that outlasted the application time, generating a poststimulus ongoing activity (Fig. 8B). After capsaicin, the response to mechanical stimulation was suppressed.

Interaction between stimuli

A significant reduction of mechanical threshold was observed in 19 polymodal fibres following exposure of the neuroma to a pH 6.0 solution and to IS (Fig. 9A and B). This effect was not obtained after application of hypertonic NaCl.

Figure 9. Changes in sensitivity to mechanical stimulation of polymodal and mechanically insensitive units as a result of the application of chemical agents.

Mechanical threshold of polymodal fibres measured before and after the application of an acidic solution (A) or IS (B). Data are means ±s.e.m.; * P < 0.05,**P < 0.01; Student's t test). C, development of mechanical sensitivity in 9 units 15 min after application of either acidic solution or IS. N/R, no response.

Mechanically insensitive chemosensory units

Twenty-four out of 141 units (17 %) initially unresponsive to mechanical stimulation but sensitive to at least one of the test chemical substances were classified as pure chemosensory units. Sixteen per cent of these units had spontaneous activity at the beginning of the experiment. Their firing discharge in response to chemicals had a frequency and a temporal course similar to that observed in polymodal units. Similarly, nine of the chemosensory units developed sensitivity to mechanical stimulation after chemical activation (Fig. 9C). The response consisted of one to a few impulses that were evoked at the same mean threshold as in mechanosensory units.

All of the studied mechanically insensitive fibres were excited by hypertonic NaCl, two-thirds of them responded to IS while only half were activated by pH 6.0 solutions. Capsaicin, tested in five units excited three of the units and was ineffective in the remaining two (Table 1).

Unresponsive units

Ten out of 141 units (7.1 %) did not respond to any of the stimuli tested. Seven of them displayed spontaneous activity at the beginning of the experiment, and three developed ongoing activity following electrical stimulation.

Discussion

The present study shows that a significant proportion of the regenerating nerve endings trapped in a saphenous nerve neuroma, in vitro, display ongoing activity and respond to some of the stimuli that under normal conditions activate intact sensory nerve terminals.

The presence of spontaneous impulses in neuroma fibres, showing either an irregular pattern of discharge, or firing in doublets, triplets or bursts, has been reported in various models of regenerating sensory fibres in vivo. The proportion of spontaneously active units found in the neuroma in vitro was in general similar to that obtained by other authors in neuroma fibres recorded in situ (Govrin-Lippmann & Devor, 1978; Meyer et al. 1985; Leah et al. 1988; Kajander et al. 1992). Based on experiments in which the connection of the neuroma with the soma of DRG neurons was maintained, it has been proposed that the soma of DRG neurons may be the main source of spontaneous discharges of myelinated fibres in peripherally injured nerves (Kajander et al. 1992; Babbedge et al. 1996). The relatively large proportion of fibres displaying spontaneous activity (almost one-third of the Aδ and C fibres) in the present experiments suggests that the main source of ectopic ongoing activity is the endogenous depolarization of regenerating nerve endings in the neuroma, as was previously proposed in studies in vivo (Wall & Gutnick, 1974; Govrin-Lippmann & Devor, 1978; Blumberg & Jänig, 1984; Meyer et al. 1985). In the neuroma in vitro preparation it was not possible to distinguish the original receptor type of the recorded fibres. More than 30 % of those showing ongoing activity were exclusively mechanosensitive; they may include a significant proportion of slowly adapting proprioceptors, which have been shown to contribute more than low-threshold cutaneous mechanoreceptors to ectopic spontaneous activity in neuromas (Tal et al. 1999).

In the neuroma in vitro, inflammatory substances produced by surrounding tissues are presumably washed out. Therefore, these data confirm that hyperexcitability is directly attributable to changes in membrane electrical properties of the injured fibres, possibly associated with the expression of altered Na⁺ conductances (Matzner & Devor, 1994).

Mechanosensitivity was found in silent and spontaneously active units. The majority of neuroma fibres were exclusively mechanosensory, whereas one-third of them were polymodal units. Most mechanosensory units conducted in the C fibre range. This corresponds with the high number of unmyelinated polymodal nociceptors found in the normal nerve of the rat (Lynn & Carpenter, 1982). In neuromas in vivo, a comparatively higher proportion of mechanosensory Aδ fibres was found (Leah et al. 1988; Welk et al. 1990). This may be explained because in the present experiments all fibres with a conduction velocity below 2 m s⁻¹ were considered C fibres. Also, shorter distances for conduction velocity measurements were available in vitro, thus making more significant the contribution of the unmyelinated portion of the growing axon to the final conduction velocity value (Fried et al. 1991).

Impulse activity evoked by mechanical stimulation of Aδ and C fibres of the neuroma had thresholds that were in the range of those of intact sensory endings of the saphenous nerve of the rat (Lynn & Carpenter, 1982). Their firing frequency was related to the intensity of the stimulus, although they often exhibited afterdischarges and enhanced responsiveness upon repeated stimulation. The resemblance in the sensitivity to mechanical forces between neuroma endings and intact nerve terminals suggests that during the process of regeneration, the cellular machinery retains the capability to synthesize and to transport anterogradely to the injured area stretch-activated membrane channels and proteins (Garcia-Añoveros & Corey, 1997) necessary to transduce mechanical stimuli.

A proportion of the neuroma fibres was also activated by various chemical substances and were thus classified as polymodal units (Bessou & Perl, 1969). Some of the chemosensory units were devoid of mechanical sensitivity, which developed in a proportion of them after repeated application of algesic substances; thus they resemble ‘silent’ nociceptors (Meyer et al. 1991; Schaible & Schmidt, 1996). Hypertonic NaCl produced a strong activation of neuroma fibres that did not seem to be due to an osmotic action, because it was not mimicked by a sucrose solution of analogous osmolarity and is therefore attributable to a surface charge effect (Belmonte et al. 1991). Acid, another well known activator of nociceptive nerve endings (Belmonte et al. 1991; Steen et al. 1992) evoked in most Aδ and C neuroma fibres a sustained impulse discharge that was roughly proportional to the pH of the test solution as occurs in intact nociceptive endings (Belmonte et al. 1991). Confirming previous data (Hartung et al. 1989) an excitatory action was obtained by application of capsaicin, usually followed by inactivation of the fibre. This effect has been repeatedly observed in intact capsaicin-sensitive nerve terminals and is seemingly due to a large Ca²⁺ entry induced by capsaicin (Marsh et al. 1987; Gallar et al. 1993).

Several proton-gated ion channels have been described in primary sensory neurons (Krishtal & Pidoplichko, 1980; Waldmann et al. 1997a,b;Waldmann & Lazdunski, 1998). Also, a slowly inactivating, non-selective cation channel responsible for the capsaicin effect (the vanilloid receptor VR1) has been recently cloned. This channel is present in putative nociceptive neurons and is additionally activated by noxious heat and acid (Caterina et al. 1997; Tominaga et al. 1998). The sensitivity of neuroma endings to acid and capsaicin confirms that the membrane of regenerating nerve fibres contains proton-gated and VR1 ion channels similar to those present in the intact peripheral endings of sensory nerves.

A mixture of chemical agents, the ‘inflammatory soup’ (Handwerker & Reeh, 1991) evoked in the neuroma a discharge of impulses that outlasted the stimulus and induced afterdischarge and ongoing activity, preferentially in fibres that also exhibited sensitivity to acid. After IS and acidic solutions, an enhanced responsiveness to mechanical stimulation was also noticed. Responses of axotomized fibres to algesic substances have been described within a few hours of nerve transection (Michaelis et al. 1997). The present data indicate that sensitivity to endogenous chemicals persists in neuroma fibres for long periods after axotomy. The impulse response evoked by IS is similar to that obtained in intact nociceptive fibres of the rat skin-nerve preparation, including the absence of tachyphylaxis, defined as a rapidly decreasing response after repetition of a few doses of the same chemical stimulus (Kessler et al. 1992). Algesic agents present in the IS act either by activating protein kinase C, as does bradykinin (Cesare & McNaughton, 1996), or through an increase in intracellular cAMP levels, as is the case for PGE₂ or serotonin (Taiwo et al. 1989; England et al. 1996; Gold et al. 1996; Ingram & Williams, 1996). The effects of IS on neuroma fibres suggest that the intracellular pathways that mediate the excitatory and modulating effect of inflammatory substances on ion channels involved in nociceptor activation remain operative in neuroma nerve endings. The possibility that neuroma endings become sensitized by endogenous chemicals could explain the generation of pain sensations by innocuous mechanical stimulation of neuromas.

By studying in vitro the firing characteristics of saphenous neuroma fibres we demonstrate that regenerating terminals are functionally heterogeneous and exhibit properties reminiscent of those of intact mechanosensory, polymodal and ‘silent’ nociceptor sensory units, including their sensitization by algesic chemicals. Activity in the damaged end of thin myelinated and unmyelinated axons may contribute importantly to the arrival of a continuous sensory input to the spinal cord after peripheral lesion of sensory nerves. Most fibres of this size carry nociceptive information and are thus presumably the source of paraesthesias and neuropathic pain (Devor, 1991). The model of the isolated superfused neuroma in vitro may be useful for studying the physiological and pathophysiological responses of regenerating nociceptive nerve endings and their modulation by drugs under controlled experimental conditions.