OUP user menu

Encoding of burning pain from capsaicin-treated human skin in two categories of unmyelinated nerve fibres

Martin Schmelz, Roland Schmid, Hermann O. Handwerker, H. Erik Torebjörk
DOI: http://dx.doi.org/10.1093/brain/123.3.560 560-571 First published online: 1 March 2000


Burning pain was induced in healthy human subjects by intracutaneous injections of capsaicin (20 μl, 0.1%) in the innervation territory of the cutaneous branch of the peroneal nerve and the pain responses were compared with the activation patterns of afferent C-fibres recorded by microneurography. Responsiveness of single units to mechanical or heat stimuli or to sympathetic reflex provocation tests was determined by transient slowing of conduction velocity following activation (marking technique). Capsaicin activated each of 12 mechano-responsive and 17 of 20 mechano-insensitive C-units. However, the duration of the responses to capsaicin was significantly longer in mechano-insensitive C-units (median 170 s; quartiles 80–390) compared with mechano-responsive C-units (8 s; 4–10). The activation times of mechano-insensitive C-units closely matched the duration of capsaicin-induced pain responses, whereas activation of mechano-responsive C-units was too short to account for the duration of the burning pain. The latter generally were desensitized to mechanical stimulation at the injection site, whereas 8 of 17 of the originally mechano-insensitive C-units became responsive to mechanical probing at the injection site after capsaicin. Responses typically started several seconds after the onset of the mechanical stimulus in parallel with pain sensations. We did not observe sensitization to brushing or to punctate stimuli in uninjured parts of the innervation territory. Differential capsaicin sensitivity adds to the cumulating evidence for the existence of two categories of functionally different nociceptors in human skin, with a special role for mechano-insensitive fibres in sensitization and hyperalgesia. Possible structural differences between these two categories are discussed, including the role of tetrodotoxin-resistant sodium channels.

  • nociception
  • hyperalgesia
  • capsaicin
  • receptive field
  • allodynia
  • CMH = C-units which are mechano- and heat-responsive
  • CM = C-units which are responsive to mechanical stimulation
  • CH = C-units which are responsive to heating
  • CMiHi = C-units which are mechano- and heat-insensitive
  • TTX = tetrodotoxin


Capsaicin injection into human skin leads to intense pain and characteristic forms of hyperalgesia in the surrounding skin area (Simone et al., 1989; LaMotte et al., 1991). Though the capsaicin injection model has been in use for decades and is well established for studying various forms of hyperalgesia under experimental conditions, the underlying neuronal mechanisms are only partly explored. Ongoing pain and hyperalgesia to heating and punctate mechanical stimuli at a capsaicin-treated skin site (primary hyperalgesia) are mediated by unmyelinated nociceptors, as revealed by experiments in which conduction in myelinated fibres was blocked differentially (Culp et al., 1989; Koltzenburg et al., 1992; Torebjörk et al., 1992). On the other hand, from results obtained by microstimulation techniques it was concluded that secondary hyperalgesia to touch is mediated by altered central processing of input from myelinated mechano-sensitive units (Torebjörk et al., 1992), although this altered central processing of mechanoreceptor input also requires ongoing activity in afferent nociceptive C-fibres (Koltzenburg et al., 1992).

However, apart from this indirect evidence, the role of C-nociceptors in capsaicin-induced pain and hyperalgesia is still unclear. To our knowledge, there is only one published microneurography study on 14 CMH units (mechano- and heat-responsive C-fibres, formerly polymodal nociceptors) demonstrating excitation, but also desensitization after capsaicin injection (LaMotte et al., 1992). In a study on a larger sample of nociceptors recorded from monkey, responses of polymodal myelinated and unmyelinated nociceptors were too transient and generally too weak to account for the pain after capsaicin injection in human psychophysical experiments (Baumann et al., 1991). Although this study concentrated on mechano-receptive nociceptors, the authors, by serendipity, observed a few mechano-insensitive `heat-specific' and `chemoceptive' C-fibres and discussed the possibility that such types of units might be responsible for the strong pain lasting many minutes after capsaicin injection (LaMotte et al., 1992). However, the significance of these observations remained unclear.

With improved techniques for reliably recording from single fibres in microneurography, we have proved the existence in human skin of other categories of nociceptive C-units beside mechano- and heat-receptive, `polymodal' nociceptors (Schmidt et al., 1995). Here we looked for possible activation patterns in one or several sub-population(s) of human C-nociceptors, which could match pain and hyperalgesia following intradermal capsaicin injection.

Some of the results have been published in abstract form (Schmelz et al., 1997).



Recordings were obtained from 34 subjects (23 male, 11 female, age 20–33 years) in the microneurography laboratories at Uppsala and Erlangen. None of the subjects showed signs of neurological or dermatological disease. The subjects were financially compensated for the time spent in the experiment. They were instructed that they could withdraw from the experiment at any time and this would not affect the financial compensation. All subjects gave their informed consent according to the Declaration of Helsinki and the study was approved by the local ethics committees in Uppsala, Sweden, and Erlangen, Germany.


Methods of microneurography employed in this study have been described in detail elsewhere (Schmelz et al., 1994; Schmidt et al., 1995). Microelectrodes were inserted at the level of the fibular head into the superficial branch of the peroneal nerve. When a stable recording position in a nerve fascicle was obtained, the skin field innervated by this fascicle was identified by stroking the skin and listening to the characteristic signal from an audio-amplifier displaying multifibre discharges of low threshold mechano-responsive A-fibres. The ensuing search for innervation territories of single C-units was by transcutaneous electrical stimuli to avoid a bias towards mechanically receptive units. To this purpose, a steel electrode with a pointed tip of 1 mm2 was gently pressed onto various points within the target area and electrical pulses of 0.2 ms duration were delivered from an insulated constant current stimulator (Digitimer DS7). Electrode gel was used to reduce the impedance. Stimulus strength was adjusted to a level that was slightly unpleasant, but still tolerable for the subjects (30–50 mA). When C-fibre responses were encountered, a pair of needle electrodes, 0.2 mm in diameter, was inserted for intracutaneous electrical stimulation of that site, tips 5 mm apart. If necessary, these needle electrodes were readjusted until C-fibre responses were obtained to iterative constant voltage stimulation (0.2 ms, 10–100 V, 4 s interstimulus interval) delivered by the stimulus insulation unit of a Grass S88 stimulator.

`Marking' of activated C-units

When responses of one or several C-fibres to the intracutaneous stimulation were recorded, the `marking' technique was used to characterize the unit(s). This technique is based on the slowing of conduction velocity in a C-fibre when it is activated by an additional stimulus (Torebjörk, 1974; Torebjörk and Hallin, 1974). Pronounced slowing after repetitive firing is characteristic for nociceptive C-fibres and probably due to prolonged changes of membrane properties after excitation (Thalhammer et al., 1994; Gee et al., 1996; Serra et al., 1999). It has been shown that even a single additional spike induced in a C-fibre by a conditioning stimulus produced an increased delay of the subsequent electrically induced spike by ~1 ms (intracutaneous electrical stimulation at 4 s intervals). The length of time of the delay is strongly correlated with the number of additional spikes (Schmelz et al., 1995).

The C-units were functionally tested in the following order.

Manoeuvres eliciting sympathetic reflexes

At first the `marking' technique was employed to test if the C-units were possibly sympathetic efferents by manoeuvres known to increase the skin sympathetic sudomotor and vasoconstrictor outflow greatly in conscious man (Torebjörk and Hallin, 1970; Delius et al., 1972; Hagbarth et al., 1972; Hallin and Torebjörk, 1974b). For this purpose, sympathetic reflexes were provoked by loud unexpected noises, by inciting the subject to laugh or to perform a deep inspiration. The efficiency of these manoeuvres was controlled by recording background activity of sympathetic burst discharges. C-units which showed latency increases related to sympathetic reflexes were classified as sympathetic fibres (Hallin and Torebjörk, 1974a; Schmelz et al., 1998).

Natural stimulation of the skin

A 750 mN v.Frey nylon filament was used (Stoelting Co., Chicago, Ill., USA) for determining whether units were responsive to mechanical stimulation and, if that was the case, also for mapping the extent of their receptive fields (Schmelz et al., 1994; Schmidt et al., 1997). In addition, thinner calibrated v.Frey filaments were used for determining mechanical thresholds. Heat stimuli were delivered by radiation from a halogen bulb focused to the target area and feedback controlled from a thermocouple lightly attached to the skin (Beck et al., 1974). Mechano-responsive units were tested inside their mechano-receptive field and mechano-insensitive units inside their electro-receptive field (Schmelz et al., 1994). For testing responsiveness to heating, the skin temperature was slowly increased by 0.25°C/s, from an adapting temperature of 30–32°C. Heating was stopped by the subject before the pain tolerance limit was reached (48–52°C). Cut-off temperature was 52°C. We did not map the heat-responsive innervation territory in this study.

Depending on the occurrence or absence of latency shifts related to these mechanical and thermal stimuli, afferent C-units were classified as in our previous studies into categories of units responding to mechanical and heat stimuli (CMH), only to mechanical stimuli (CM), only to heat stimuli (CH) or to neither mechanical probing nor heat, i.e. mechano- and heat-insensitive (CMiHi) (Schmidt et al., 1995).

The innervation territory of mechano-responsive units was mapped with 750 mN v.Frey filaments. The heat responsiveness was tested only at one spot of the innervation territory, at the site selected for the injection.

Transcutaneous electrical stimulation

For searching and identifying mechano-insensitive units (CH and CMiHi), transcutaneous electrical stimuli (0.2 ms, 30–50 mA) were delivered from a pointed surface electrode gently pressed onto the skin. This surface electrode was also used for assessment of the innervation territory (electro-receptive field) of mechano-insensitive units (Schmelz et al., 1994).

Electrical thresholds

A Digitimer DS7 constant current stimulator was used for measuring the electrical threshold of the units. Single pulses, 0.2 ms in duration, were delivered through a round compressed cotton disc, 5 mm in diameter, soaked in saline, and held by hand in a specially designed applicator (Magerl et al., 1990). A large (5 × 10 cm) metal plate attached to the skin on the lower leg served as reference electrode.

Capsaicin injection

Up to 20 μl of saline containing 0.1% capsaicin in Tween-80 (LaMotte et al., 1991) (i.e. a maximal dose of 20 μg) was injected intracutaneously at a distance of 5–10 mm from the stimulating needle electrodes, but within the innervation territory of the C-unit (i.e. the mechano-receptive field of mechano-responsive or the electro-receptive field of mechano-insensitive units). The injection was performed slowly (10 s) to allow the subject to rate continuously the increase of pain magnitude on a 10 point scale (0 = `no pain', 10 = `unbearable pain'). Injection was stopped immediately when the subject announced a pain intensity of `5' in order to prevent losing the unit due to unintentional pain-provoked movements of the subject. Capsaicin was delivered from a 100 μl Hamilton syringe (Hamilton Bonaduz AG, Basel, Switzerland) with 1 μl partitions through a sterile nylon microfilter with 0.22 μm pores (Micron Separations Inc., Westborough, Mass., USA), using a 25 gauge needle.

Post-injection testing

About 2–4 min after the end of injection, when the most intense pain from capsaicin had substantially decreased, mechanical responsiveness was tested in the following order: (i) responsiveness of the units to light stroking was tested with a cotton swab in the uninjured area around the injection site and at the injection site; (ii) the borders of the zone of secondary hyperalgesia to punctate stimuli (LaMotte et al., 1991) were determined with a 750 mN v.Frey filament; (iii) the same v.Frey filament was used to test mechanical responsiveness of the units outside their innervation territory but within the zone of secondary hyperalgesia to punctate stimuli; (iv) the non-treated parts of their innervation territories were then tested, excluding the injection site; (v) finally, v.Frey stimulation was applied at the site of capsaicin injection.

Responsiveness to heat stimuli was only tested at the injection site.

Data acquisition and analysis

C-unit responses to intracutaneous electrical stimulation were recorded on line by a PC computer via an interface card (DAP, Microstar, Wash., USA) using the SPIKE/SPIDI software package (Forster and Handwerker, 1990). A suitable time segment of the recording following each electrical stimulus pulse was displayed and subsequent traces were written from top to bottom on the computer screen for on-line assessment of latency shifts of the activated C-units. Digital matched filtering was implemented to facilitate the tracking of the latency shifts (Hansson et al., 1998). In addition, the recordings were stored on hard disk for off-line analysis.

Activation of C-units was documented by parameters derived from the `marking' response: (i) the maximal shift in latency (ms) induced by the capsaicin injection; (ii) the delay of maximum shift (s), i.e. the period from the end of

injection to the occurrence of maximal latency shift; (iii) the number of traces in which the conduction delay of the respective unit was increased abruptly (activation periods); and (iv) the time (s) from the first to the last activation period (duration).

The maximal latency shift is a semiquantitative measure of the peak activation of the unit. It increases with the number of capsaicin-induced impulses (Schmelz et al., 1995). However, in many recordings, this parameter could not be determined during the most intense bursting activity, but only during the beginning of the recovery phase after the burst (see Fig. 2, activation at the end of capsaicin injection). Thus, the maximal latency shifts reported here represent conservative measures. Since the ongoing electrical stimulation had a low frequency of 0.25 Hz, there is a possible error of the order of 4 s in the delay and duration measurements.

Unless stated otherwise, the Mann–Whitney U test was used for statistical analysis. P < 0.05 was considered statistically significant.

Data in Results and in the figures are given as median and quartiles (25–75%) or mean ± SEM as appropriate.


Sample of C-units

Forty C-units were tested with capsaicin injections. Eight of these were sympathetic, efferent C-fibres. Ten C-units were mechano- and heat-responsive (CMH) and two were responsive to mechanical stimulation, but not to heating (CM). Twenty units were insensitive to mechanical stimuli. Eleven of these responded to heating the skin (CH). The remaining nine units were mechano- and heat-insensitive (CMiHi).

There is cumulating evidence from our recent work that all mechano-insensitive units show similar characteristics independent of their responsiveness to heating, e.g. slow conduction velocities compared with other categories of C-units, pronounced slowing following activation and large innervation territories in the skin. Therefore, we combined the fibre classes `mechano- and heat insensitive' (CMiHi) and `heat-, but not mechano-responsive' (CH) in one category of mechano-insensitive C-units for the purpose of this study (n = 20). Likewise, the rare mechano-responsive units which did not respond to heating (CM) have many features in common with the much more common CMHs (polymodal nociceptors), e.g. similar mechanical thresholds, similar receptive field sizes and similar conduction velocities (Torebjörk et al., 1996). They were grouped together in this report in the category mechano-responsive C-fibres (n = 12).

The innervation territories of the units were located on the lower leg (n = 23) or on the dorsum of the foot (n = 17). Mean conduction velocities were 0.79 ± 0.02 m/s for sympathetic fibres, 0.97 ± 0.05 m/s for mechano-responsive and 0.77 ± 0.03 m/s for mechano-insensitive units. Electrical thresholds assessed with the electrolyte probe described in the Methods section were significantly lower (P < 0.01) for mechano-responsive (mean 2.1 ± 0.6 mA) than for mechano-insensitive units. In one unit of this category the threshold was 40 mA, but it exceeded the tolerance level of 50 mA in another 10 cases.

Eleven additional afferent C-units (five mechano-responsive and six mechano-insensitive) were lost due to unintentional movements of the subject during the capsaicin injection. These units are not included in the present material.

Sensory effects of capsaicin injections

The minimum volume of injected capsaicin solution was 8 μl, and each injection resulted in a small bleb at the injection site. Pain ratings generally increased during the first 30 s after the end of injection, and subsequently the pain decreased, waxing and waning over the next 5–10 min to vanish thereafter. There was hyperalgesia to gently stroking the skin or poking with the 750 mN v.Frey filament in the area of secondary hyperalgesia around the injection site during that time, confirming previous observations (Simone et al., 1989; LaMotte et al., 1991, 1992; Koltzenburg et al., 1992).

Responsiveness to capsaicin injection

The following peak pain ratings were obtained in experiments in which the respective unit types were recorded: 5 (4–6) [median (quartiles)] for sympathetic units, 5 (4–6) for mechano-responsive units and 6 (5–7) for mechano-insensitive units. Pain ratings did not differ significantly between the unit types (Mann–Whitney U test) suggesting similar stimulation intensity.

All 12 mechano-responsive units and 17 of 20 mechano-insensitive units were activated by capsaicin. Six of eight sympathetic C-units were also activated briefly in irregular reflex responses, caused either by the nociceptive input itself or by the arousal induced by the pain ratings.

A typical response of a mechano-receptive unit to heat and mechanical stimulation obtained before capsaicin injection is shown in Fig. 1. This unit had a mechanical threshold of 30 mN, and it started to respond to heating at 39°C.

Fig. 1

Response latencies of a mechano- and heat-responsive C-nociceptor (CMH) to electrical stimulation at 4 s intervals inside the receptive field are shown in subsequent traces from top to bottom. Mechanical stimulation with a v.Frey filament of 30 mN (arrow) activated the unit, leading to a sudden increase in response latency (`marking'). Increasing the skin temperature in the receptive field from 32 to 43°C also activated the unit at a threshold of 39°C.

The response of the same unit to capsaicin injection is shown in Fig. 2. There was an immediate response when the needle was inserted into the skin (Fig. 2, filled horizontal arrowhead), confirming its mechano-responsiveness. When the capsaicin injection began, there was again an immediate pronounced but short-lasting marking response, suggesting an intense, brief burst of activity lasting for a few seconds only. After the injection, the unit did not respond to poking with filaments from 30 to 750 mN at the injection site, suggesting desensitization (Fig. 2, open horizontal arrows), and it was also unresponsive to heating up to 48°C at this site. The desensitization was confined to the treated area and the unit was still responsive to mechanical stimuli in the untreated part of the receptive field with unchanged threshold (Fig. 2, filled horizontal arrow). It did not respond, however, to repeated poking with the stiff 750 mN filament just outside the innervation territory or to punctate stimuli within the zone of secondary hyperalgesia (Fig. 2, open vertical bar).

Fig. 2

Response of a mechano- and heat-responsive C-nociceptors (CMH) to intracutaneous injection of capsaicin (20 μl, 0.1%) in the receptive field. Subsequent traces are displayed from top to bottom. Note that traces are highly condensed such that single spikes can no longer be recognized. Insertion of the needle into the skin (filled horizontal arrowhead `in') activated the unit, as did the capsaicin injection (filled vertical bar). The open horizontal arrow (`out') denotes retraction of the needle. Note that the capsaicin-induced burst of activity lasted for a few seconds only as can be judged by the abrupt and marked increase in response latency. After the injection, the unit did not respond to repeated poking with the stiff 750 mN filament just outside the innervation territory but within the zone of secondary hyperalgesia to punctate stimuli (open vertical bar), nor did it respond to poking with filaments from 30 to 750 mN within the receptive field at the injection site, suggesting desensitization (open horizontal arrows). The unit was also unresponsive to heating up to 48°C at the injection site. The desensitization was confined to the treated area and the unit was still responsive to mechanical stimuli in the untreated part of the receptive field with unchanged threshold (30 mN) (filled horizontal arrow). Note that the second unit, responding with a latency of ~340 ms, had its innervation territory remote from the capsaicin injection site.

Different response patterns were found for mechano-insensitive units, as exemplified in Figs 3 and 4. This CMiHi unit was not activated by skin heating in its electro-receptive field before capsaicin injection (Fig. 3A), but thereafter it became responsive to heating above 40°C in the capsaicin-treated area (Fig. 3B). Insertion of the needle into the electro-receptive field did not activate the unit, confirming its insensitivity to mechanical stimuli. It began to respond at the end of the capsaicin injection and was then intensively active for a period of ~70 s. Before capsaicin, the unit was unresponsive to probing with the 750 mN v.Frey filament inside the electro-receptive field and it remained so thereafter in the non-treated part of the electro-receptive field. It also remained unresponsive to punctate stimuli just outside the innervation territory in the area of secondary hyperalgesia. However, mechanical probing with the 750 mN filament just at the border of the injection site evoked a moderate activation, and probing the centre of the injection site induced an intense response. Even a thinner v.Frey filament of 260 mN evoked a response at the injection site. The unit responded to laterally stretching the skin with the 750 mN filament just outside the injection zone, but was unresponsive to the same filament when applied perpendicularly to the same spot.

Fig. 3

Response of a mechano- and heat-insensitive C-nociceptor (CMiHi) to heating before (A) and after (B) capsaicin injection. The unit was not activated by heating to 48°C before the injection (upper panel) but only showed a slight decrease in response latency due to higher conduction velocity at higher skin temperatures. After capsaicin injection, the unit was sensitized to heat at the injection site and responded vigorously to heating to 44°C (lower panel).

Fig. 4

Response of a mechano- and heat-insensitive C-nociceptor (CMiHi) to capsaicin injection is shown on the left. Location of the electro-receptive field (hatched area), the stimulation needles (stim.), the zone of secondary hyperalgesia to pinprick (dotted line pp), the capsaicin injection site (x) and mechanical test spots (c and v) are given on the right. Note that letters on the left panel indicate the location of the stimulus as shown on the right. Insertion of the injection needle into the electro-receptive field did not activate the unit (left panel, filled arrowhead `in'), confirming its insensitivity to mechanical stimuli. Capsaicin injection (filled bar) activated the unit for a period of ~70 s after the end of injection (needle withdrawn at open arrowhead `out'). Probing with a 750 mN v.Frey filament in the area of secondary hyperalgesia to punctate stimuli, just outside the innervation territory (open bar pp) and in the non-treated part of the electro-receptive field (open bar pp'), did not activate the unit. However, mechanical probing with the 750 mN filament just at the border (c) of the injection site evoked a moderate activation (filled arrow c), and probing the centre of the injection site (x) induced an intense response (filled arrow x). Even a thinner v.Frey filament of 260 mN evoked a response at the injection site (filled arrow x). The unit responded to laterally stretching the skin with the 750 mN filament just outside the injection zone (hatched arrow v), but was unresponsive to the same filament when applied perpendicularly to the same spot (open arrow v').

These response features were typical for mechano-insensitive C-units of both types, CMiHi and CH units, whereas the two CM units responded similarly to the major group of mechano-responsive units, the CMHs.

The response of a CH unit (heat threshold 47°C) to capsaicin injection is shown in Fig. 5. It did not respond to insertion of the needle into its innervation territory, confirming the mechano-insensitivity of the unit. It was activated during the injection of capsaicin, and continued to discharge in vigorous bursts for 3 min thereafter.

Fig. 5

Responses of a mechano-insensitive, heat-responsive C-nociceptor (CH) to capsaicin injection (20 μl, 0.1%; filled bar). Note that insertion of the cannula (filled arrowhead `in') did not activate the unit, proving its mechanical insensitivity. Injection of capsaicin led to pronounced increases of response latency for several minutes.

Interestingly, the newly acquired responses to mechanical stimuli in the capsaicin-treated areas of previously mechano-insensitive units were often delayed, appearing several seconds after the v.Frey filament was removed from the skin. In parallel, the burning pain sensations elicited by these stimuli were also delayed for several seconds.

Differential activation patterns in different categories of C-units

The median of the maximum latency increases was 50 ms (quartiles 27.5 and 52.5) in mechano-responsive and 45 (quartiles 20 and 90) in mechano-insensitive C-units. The differences between the two groups of afferent units are not statistically significant. Responses in sympathetic units were less intense, with a median of maximum latency increase of 5 ms (quartiles 4 and 6).

In contrast, the delay from the end of capsaicin injection to the beginning of recovery from the maximal latency increase was different between the two unit groups, median 2 s for mechano-responsive units (quartiles 0–4) compared with 18 s (quartiles 4–40) for mechano-insensitive units (P < 0.01). Like mechano-responsive units, sympathetic units were also activated only briefly and thus recovery from the maximum shift increase was seen after a median of 2 s (quartiles 0–4) (Fig. 6).

Fig. 6

Summarizing statistics on the capsaicin response of mechano-responsive (mech+), mechano-insensitive (mech–) and sympathetic units (Symp.). The maximum shift of response latency (median and quartiles) following the injection, which reflects a semiquantitative measure of the degree of activation, is depicted on the ordinate. On the abscissa, the delay of the maximum shift after the injection is shown.

Another parameter derived from the marking response is the mean duration of capsaicin-induced responses, defined as the time interval from the first to the last abrupt increase in latency of the activated units. It was 8 s (4–10) for mechano-responsive and 170 s (80–390) for the mechano-insensitive units (P < 0.01). The longest duration of the capsaicin response in a mechano-responsive unit was 60 s with a total of five activation periods. It should be noted that some of the mechano-insensitive C-fibres were active for periods exceeding the usual observation time, of the order of 15–20 min after capsaicin injection. The difference between the two categories of mechano-insensitive units was not significant (Mann–Whitney U test, NS). The number of activation periods or `markings', indicating the intensity of the response, was significantly smaller for mechano-responsive units, 2 (1–3), as compared with mechano-insensitive units, 14 (9–40) (P < 0.01) (Fig. 7).

Fig. 7

Duration of activation in minutes (left panel) and number of activation periods (`markings', right panel) following capsaicin injection (20 μl, 0.1%) in different classes of C-fibres are shown (median and quartiles). The two measures reveal that mechano-insensitive units (CMiHi and CH, mech–) exhibited more intense and sustained activation than the mechano-sensitive [CM(H), Symp. and mech+] units.

Changes in mechanical and heat responsiveness at the site of capsaicin injection and in unaffected parts of the innervation territory

Ten of 12 mechano-responsive units became unresponsive to mechanical stimuli at the injection site after capsaicin. The result was ambiguous for the two other units of this category. In contrast, among the mechano-insensitive units, 5 of 10 CH units newly acquired responsiveness to mechanical probing at the injection site after capsaicin, and so did 3 of 7 CMiHi units. The latter were—by definition—unresponsive to heating before capsaicin and 2 of 4 of them became responsive afterwards in the injection area. The heat threshold of these units was lowered from non-responsiveness to 44 and 47°C. In the case of CH units that had responded to heating before capsaicin, 3 of 4 were sensitized by the injection in the treated area, resulting in a drop of heat thresholds from 47 to 42°C in one unit, 44 to 35°C in the second, and 47 to 43°C in the third unit.

The stiff 750 v.Frey filament was used during the period of pronounced pain and hyperalgesia to test whether the C-units under study could be activated by mechanical probing either just outside their innervation territories, but in the secondary zone for punctate hyperalgesia, or within their innervation territories, but clearly outside the site of capsaicin injection. In none of the units was sensitization of untreated parts of the receptive territory or extension of this territory observed. Stroking the skin with a cotton swab did not activate any of the C-units before or after capsaicin, regardless of in which part of their innervation territory the stimulus was applied.


Differential properties in sub-populations of human C-nociceptors

The aim of this study was to resolve the apparent mismatch between the high intensity and long duration of pain after capsaicin injections and the brief activity induced in polymodal nociceptors (Baumann et al., 1991). It was found that different classes of nociceptors in human skin are activated differentially by capsaicin injections. Mechano-responsive CM and CMH nociceptors were activated only briefly for a few seconds, whereas mechano-insensitive CH and CMiHi nociceptors exhibited prolonged bursting discharges for minutes. In this context, it is interesting to note that mechanically insensitive A- as well as C-nociceptors (MIAs) in monkey skin (Meyer et al., 1991; Davis et al., 1993) are also particularly responsive after capsaicin injection, showing intense prolonged discharges (Ringkamp et al., 1997).

Another striking difference between the two categories of C-nociceptors is the changed excitability after capsaicin injections. Mechano-responsive C-units were usually desensitized to mechanical stimulation, while mechano-insensitive C-units often became responsive to mechanical stimuli. Some of the mechano-insensitive units also became sensitized to heating. These changes in responsiveness were restricted to the injected area, i.e. to the axon terminals reached by the capsaicin.

It is known that the effects of capsaicin on human C-nociceptors are dependent on concentration. Topical application of 1% capsaicin on intact skin typically produces sensitization to heat in CMH nociceptors and no desensitization to mechanical stimuli (LaMotte et al., 1992), whereas intracutaneous injection often causes unresponsiveness of CMHs to mechanical and heat stimuli at the injection site (Baumann et al., 1991). Here we describe a different response pattern of mechano-insensitive units. These new findings may explain some important characteristics of the subjective pain responses to capsaicin (see below).

The absence of desensitization and the pronounced sensitization of mechano-insensitive units following capsaicin injection may have one of several possible causes. One may argue that a more extensive branching of mechano-insensitive as opposed to the mechano-responsive endings would allow greater options for different concentrations of capsaicin to act on different receptive sites of the axon terminal, perhaps causing desensitization in some parts and allowing enhanced responsiveness in other parts. An indication of this is the larger receptive territories of mechano-insensitive nociceptors (Schmidt et al., 1997, 1998). One may assume further that the typically delayed responses to probing in the injected area of mechano-insensitive units and the stretch-induced responses might be due to capsaicin squeezed out from the original injection site to stimulate adjacent, chemo-responsive terminals. According to another hypothesis, the sensitizing effect of capsaicin on mechano-insensitive units could be due to a deeper location of their receptive endings. Indeed, these units had much higher electrical activation thresholds than mechano-responsive units, suggesting that they either were situated deeper in the skin or had thinner terminal branches, or both. If a deeper location were the reason for the differential sensitivity, the concentration of injected capsaicin might perhaps be lower near their endings, causing sensitization rather than desensitization.

All these hypotheses based on morphological differences can hardly fully explain the opposing effects, because receptive fields of mechano-responsive units are also large enough (diameter up to 4 cm) to allow a sufficient concentration gradient of capsaicin. Thus, the differences between the two categories of unmyelinated nociceptors may also be due to molecular mechanisms. The density of channels operated by capsaicin (VR1) receptors might be lower in the membrane of mechano-insensitive than in mechano-responsive C-fibre terminals, which would allow for a lower influx of calcium ions for a given capsaicin concentration, leading to sensitization rather than neurotoxic effects. Alternatively, or in addition, the prevalence of tetrodotoxin (TTX)-resistant (Tate et al., 1998) sodium channels could account for the relative lack of desensitization in mechano-insensitive nociceptors. Recently, both a slow and a fast type of TTX-resistant Na+ channels have been described (Scholz et al., 1998a) in dorsal root ganglion cells of the rat. While in the fast subtype 50% of the channels are inactivated at a membrane potential of –55 mV, which is similar to the respective inactivation threshold of TTX-sensitive Na+ channels (–66 mV), 50% of the slow subtype channels remain active at a more depolarized membrane potential of –32 mV (Scholz et al., 1998a). Thus, a neuron containing a higher percentage of the slow subtype of TTX-resistant Na+ channels would retain responsiveness much longer when depolarized by capsaicin. It is known that TTX-resistant Na+ channels are crucial for capsaicin responses, as shown in C-nociceptors of the guinea pig cornea (Brock et al., 1998), and that Na+ channel blockers have an antihyperalgesic action (Trezise et al., 1998). There is some indirect evidence linking mechano-insensitive C-nociceptors with TTX-resistant Na+ channels: mechano-insensitive C-fibres exhibit a pronounced conduction velocity decrease upon electrical stimulation at frequencies as low as 0.25 Hz, in contrast to mechano-responsive nociceptive units (Weidner et al., 1998), a phenomenon that is not simply explained by differences in conduction velocity as also shown for the differentiation between nociceptive and cold-sensitive C-fibres in human (Serra et al., 1999). Interestingly, a comparable response pattern has also been described for TTX-resistant Na+ channels: electrical stimulation at a low frequency (0.4 Hz) induces a reduction in amplitude of sodium currents only in the slow subtype of TTX-resistant Na+ channels (Scholz et al., 1998b).

The new findings on differential capsaicin sensitivity add to the cumulating evidence for the existence of two categories of functionally different nociceptors in human skin. The most important clue is apparently the presence or absence of responsiveness to mechanical stimuli. Recently, differential response patterns of these two categories of nociceptor units have also been observed during prolonged tonic pressure in human skin; mechano-responsive units initially were activated followed by adaptation within a few seconds, whereas mechano-insensitive units initially were silent and later activated (Schmelz et al., 1997). Further characteristic differences were the significantly different electrical thresholds to controlled transcutaneous stimulation described in this report. This pattern fits strikingly with changes observed in null mutant mice for the α-subunit of the sensory neuron-specific TTX-resistant Na+ channel in which normal sensitivity to v.Frey filament stimulation co-existed with insensitivity to withdrawal responses to tonic tail pressure and lower transcutaneous electrical thresholds of C-fibres (Akopian et al., 1999). Other distinguishing features are the lower conduction velocities and the larger innervation territories of mechano-insensitive compared with mechano-responsive fibres (Torebjörk et al., 1996). Indeed, innervation territories of the mechano-insensitive units fit with their presumed role in the axon reflex flare reaction (Lynn et al., 1996).

Sensory correlates of activation of CH and CMiHi units

Both categories of C-units were activated during capsaicin injection, and could contribute to the rather intense early pain. However, only mechano-insensitive but not mechano-responsive C-units were active for several minutes (≥10) following injection. It seems reasonable to assume that their bursting discharges may induce the waxing and waning pain which is characteristic for this period. Their enhanced responsiveness to heat correlates with the subjective reports that warming aggravates the pain (Culp et al., 1989; LaMotte et al., 1992), and their delayed responses to punctate stimuli within the capsaicin injection territory fit with the delayed pain responses to such stimuli. It has also been found that after differential block of conduction in A-fibres, primary hyperalgesia to punctate stimuli persists (Culp et al., 1989; Koltzenburg et al., 1992). Further experiments with compression nerve blocks of impulse conduction in myelinated fibres would be needed to assess a possible contribution of mechano-sensitive and mechano-insensitive Aδ fibres (Ringkamp et al., 1997).

Secondary hyperalgesia to touch has been shown to be mediated by sensitive myelinated mechanoreceptor units (Torebjörk et al., 1992). However, its maintenance seems to require excitation of nociceptors (LaMotte et al., 1991; Koltzenburg et al., 1992), since the magnitude, duration and area of the touch-evoked secondary hyperalgesia are closely related to the magnitude and duration of ongoing pain after capsaicin injection (Simone et al., 1989). According to our present results, mechano-insensitive C-units are particularly important for the maintenance of this form of central sensitization. In contrast, secondary hyperalgesia to punctate stimuli can persist long after peripheral input in nociceptors has subsided or is blocked by lidocaine injection (LaMotte et al., 1991). Central processing of input from mechano-responsive nociceptors is held responsible for this form of hyperalgesia (Cervero et al., 1994), and recent psychophysical investigations provided evidence for a crucial role of input from capsaicin-insensitive Aδ fibres in punctate hyperalgesia (Meyer et al., 1998). However, a peripheral mechanism of punctate hyperalgesia was also postulated on the basis of similar development of secondary hyperalgesia and axon reflex vasodilation measured with thermography (Serraet al., 1998). It was assumed that mechano-insensitive C-nociceptors might be involved, since remote sensitization of mechano-insensitive `silent' C-nociceptors following capsaicin injection has been found by the same group (Serra et al., 1995).

Our data show no evidence for sensitization of remote parts of the electro-receptive fields of mechano-insensitive C-fibres outside the capsaicin injection site. None of the mechano-responsive or mechano-insensitive C-units were activated by punctate stimuli or touch in the zone of secondary hyperalgesia. These findings are consistent with previous reports on the absence of spread of sensitization to mechanical or heat stimuli by an `axon reflex' mechanism in regions of the receptive field clearly separated from the stimulated terminals (Baumann et al., 1991; LaMotte et al., 1992; Schmelz et al., 1996). Thus, our findings are consistent with the traditional interpretation that secondary hyperalgesia to touch and punctate stimuli is due to central rather than peripheral sensitization.

Possible sensitization of remote mechano-insensitive C-nociceptors by capsaicin injection was not tested in this study.

Clinical relevance

Our results lend further support to the crucial role of mechano-insensitive nociceptors in mechanical hyperalgesia (Reeh et al., 1987; Neugebauer et al., 1989; Kress et al., 1992; Schmidt et al., 1995; Schmelz et al., 1997), whereas `conventional' mechano- and heat-responsive nociceptors may control acute pain thresholds (Robinson et al., 1983). Mechanical hyperalgesia is often the most prominent symptom in chronic pain patients. The similar pattern of hyperalgesia in these patients (Ochoa and Yarnitsky, 1993; Koltzenburg et al., 1994) and in capsaicin injection-induced hyperalgesia provides the basis for the use of capsaicin as a model of neuropathic pain. Among the therapeutic strategies in neuropathic pain, systemically applied local anaesthetics have been shown to be effective (Rowbotham et al., 1991; Ferrante et al., 1996). Recent investigations on the antihyperalgesic effect of systemically applied lidocaine revealed a striking pattern: low concentrations of lidocaine inhibited mechanical hyperalgesia to tonic pressure, histamine-induced itch, capsaicin-induced hyperalgesia to punctate stimuli, touch-evoked hyperalgesia and axon reflex flare reaction, whereas acute heat pain thresholds and mechanical pain sensation were unaffected (Koppert et al., 1998, 2000). Assuming a prevalence of slow TTX-resistant Na+ channels in the terminal of mechano-insensitive C-units, these findings could be explained by the known high susceptibility of these channels to local anaesthetics. Consequently, the differential sensitivity of mechano- insensitive nociceptors could be a basis for the development of new specific antihyperalgesic drugs.


This work was supported by the Swedish Medical Research Council, Project no. 5206, the Deutsche Forschungsgemeinschaft (SFB 353), a Max Planck Research Award for International Collaboration to H.E.T, and a grant to R.S. from the Swedish Foundation for Brain Research.


View Abstract