Nerve growth factor sensitizes nociceptors to C‐fibre selective supra‐threshold electrical stimuli in human skin

Intradermal injection of 1 µg nerve growth factor (NGF) causes sustained nociceptor sensitization. Slowly depolarizing electrical current preferentially activates C‐nociceptors.


| INTRODUCTION
Nerve growth factor (NGF) plays a central role in chronic inflammatory pain. Antibodies targeting NGF relieved pain in osteoarthritis patients (Lane et al., 2010;Tiseo et al., 2014) and cutaneous injections of NGF evoked local mechanical and heat hyperalgesia in healthy human subjects (Dyck et al., 1997;Rukwied et al., 2010). Electrical high-frequency stimulation also caused enhanced pain in humans after NGF treatment Rukwied et al., 2014) suggesting axonal sensitization processes of primary sensory afferents (Obreja et al., 2018). In these earlier studies, rectangular supra-threshold electrical pulses were used for nociceptor activation, but rectangular stimuli activate both A-delta and C-nociceptors. In this study, we aimed to activate C-fibres preferentially in order to study whether NGF sensitizes primarily C-nociceptors. To evaluate this, we used an A-fibre compression block. An additional study rationale was to specify axonal sensitization after NGF in C-nociceptor subserving different sensory classes. To achieve this, we applied slowly depolarizing electrical stimuli to NGF-treated human skin and assessed psycho-physically the electrical excitability of mechano-sensitive and mechano-insensitive ('silent') C-nociceptors. We recently developed slowly depolarizing transcutaneous electrical stimulation profiles for C-fibre activation. One profile (500 ms half sine) preferentially activates mechano-sensitive C-nociceptors with increasing discharge frequencies at higher current intensities (Rukwied et al., 2020). The other (4 Hz sinusoidal) stimulates mechano-sensitive and mechano-insensitive ('silent') C-fibres in synchrony with the sinusoidal cycle (Jonas et al., 2018;Rukwied et al., 2020) and leads to pronounced accommodation of pain upon sustained application (Jonas et al., 2018). Notably, this accommodation in healthy subjects is absent in the symptomatic skin of neuropathic pain patients (Jonas et al., 2018). We hypothesized that employing sinusoidal current stimulation profiles in NGF-sensitized human skin would allow assessment of the C-nociceptor contribution to NGF-induced cutaneous sensitization.

| METHODS
The study protocol was approved by the Ethics Committee II of the University of Heidelberg, Germany. Investigations were performed in 14 human subjects (8 female, 6 male, average age 29 ± 2 years) according to the Helsinki Declaration of 1975, as revised in 1983, and after informed written consent was obtained from all volunteers. Medical history of all subjects was recorded and people with diseases associated with chronic pain, hereditary coagulation dysfunction, allergy, skin eczema, or analgesic medication were excluded. Accordingly, one subject with Raynaud syndrome was excluded from the A-fibre conduction block protocol.

| NGF administration
To familiarize each subject with the test-procedures, we performed a 'training' session involving the transcutaneous electrical stimulation protocol and mechanical impact stimuli. We repeated the test paradigms until the subjects' ratings on a numerical rating scale (NRS, 0-10) correctly reflected the applied stimulus intensities. Thereafter, we injected intradermally a volume of 100 µl containing 1 µg NGF (Miltenyi Biotec, Germany) dissolved in 0.9% NaCl (Braun Melsungen, Germany) into the volar forearm. The dose of 1 µg NGF proved to be sufficient to evoke a substantial hyperalgesia (Deising et al., 2012;Obreja et al., 2018;Rukwied et al., 2010;Weinkauf et al., 2012). About 3-4 cm lateral to the NGF site we injected 100 µl NaCl 0.9% (Braun Melsungen, Germany) as vehicle control. Hence, NGF and saline were applied during the same experimental session at day 0 and time course of hyperalgesia was assessed comparing responses at the NGF site to responses from the saline injection site. Subjects were blinded to the treatment condition (NGF/ NaCl) and kept their eyes closed throughout all test procedures to limit potential bias. Each injection spot was labelled and copied on a transparent foil for later recognition as described previously (Rukwied et al., 2010). Saline was injected into the NGF-treated arm to eliminate issues arising from handedness or from differential sensation between limbs and to minimize bias that may originate from the NGF sensitization when performing the sensory tests, particularly during the A-fibre conduction block. Skin sensitization is strictly localized at the NGF injection site and does not spread to adjacent areas (Rukwied et al., 2010). Therefore, the location of the saline control injection should not be affected at a distance of 3-4 cm from the NGF injection site. Investigators were not blinded to the treatment condition. Also, panellists were no longer blinded as soon as hyperalgesia had developed at the NGF site, which both are limitations of the experimental approach.

| Test protocol
The experimental protocol comprised functional nociceptor assessment upon electrical, mechanical and heat stimulation at days 1-14 after NGF or vehicle (NaCl) treatment assessment of skin C-nociceptors and their putative excitability changes under pathologic conditions. | 387 SCHNAKENBERG Et Al.
( Figure 1a). In addition, an A-fibre compression block with subsequent mechanical impact and electrical stimulation was performed at day 3 and 7.
All tests were performed at each pre-treated skin site twice and in randomized order taking into account adaptation or learning. Pain responses evoked by the stimuli were recorded on a numerical rating scale (NRS) with the endpoints 0 ('no pain') and 10 ('worst pain imaginable'). Subjects were allowed to respond in fractional numbers.

| Transcutaneous electrical stimulation and laser Doppler imaging
A pair of blunted bipolar platinum electrodes (diameter 0.4 mm, distance 2 mm, Nørresundby, Denmark) placed on an applicator printed with a 3D-printer was attached to the volunteers' forearm skin (Figure 1b). Sine wave and half sine wave pulses were generated by a constant current stimulator (Digitimer DS5, Welwyn Garden City, UK) and Digital-Analogue Converter (DAQ) NI USB-6221 (National Instruments, Texas, US) controlled by Dapsys 8 software (www.dapsys.net). Sine wave pulses of 4 Hz were delivered at intensities of 0.05-0.1-0.2-0.4 mA for 2.5 s each. Single half sine wave pulses of 500 ms duration were delivered with 0.2-0.6-1 mA. The intensities of the sine wave, respectively, half sine wave pulses were given in randomized order. For continuous (1 min) sine wave stimulation, the delivered current amplitude was adjusted to a rating of NRS 2-3 recorded upon 2.5-s sine wave stimulation (see below 3.1.3.).
Rectangular pulses (0.5 ms width) were generated by a Digitimer DS7A constant current stimulator and pulse generator PG1 (Rimkus Medizintechnik). At a frequency of 2 Hz, current intensity was increased from 0 in steps of 0.1 mA/s and the volunteers were requested to indicate when they perceived the pulses, when the perception was painful, and when pain reached a level of NRS 3. Thereafter, at intensity levels of NRS 3, rectangular pulses of 4-20-100 Hz were delivered in randomized order for 1 s and the corresponding pain intensity was recorded.
Axon reflex-mediated flare responses caused by the sine wave stimuli delivered for 2.5 s at intensities of 0.05-0.4 mA were recorded by laser Doppler imaging (Moor LDI). The scanning device was mounted perpendicularly to the skin surface at a distance of 50 cm using a plump line and the position ensured across sessions by focusing the laser beam at rest to the injection sites. Following a baseline image of the skin blood flow the scanning was paused, transcutaneous electrical sine-wave stimuli were delivered and the laser Doppler imaging sequence proceeded for 2 min. Each scan required 30 s and covered an area of 12 cm 2 (4,000 Pixels). Image analysis was performed off-line by dedicated software (MoorLDI, V5.0). Mean baseline skin blood flow ('flux') and its standard deviation were determined. The threshold for pixels indicating a significant blood flow increase was set at twofold standard deviation of the mean flux as reported previously (Chizh et al., 2007;Geber et al., 2007) and all pixels exceeding the threshold value were identified. The corresponding area of these pixels was calculated and indicated as the area of axon reflex erythema. F I G U R E 1 Experimental procedure and electrode configuration. (a) After informed consent and a training session at day 0, different functional sensory tests were assessed after NGF injection at days 1-14. A-fibre compression block was performed at day 3 and day 7 (grey columns) and a shortened protocol of sine wave, half sine wave and mechanical impact stimuli investigated. (b) Configuration of the electrode used for transcutaneous electrical stimulation. Note the ruler for the identification of the electrode dimension 388 | SCHNAKENBERG Et Al.

| Mechanical impact stimulation
Dynamic mechanical pain was assessed by a cylindrical plastic projectile (diameter 6 mm) accelerated pneumatically to velocities of 4 and 8 m/s in an 8 cm guiding barrel towards the skin surface (Kohlloffel et al., 1991). For each velocity, stimuli were repeated three times in intervals of 20 s and corresponding pain responses were recorded.

| CO 2 laser heat stimulation
A feedback-controlled CO 2 laser class 4 stimulation device (wavelength 10.6 µm, power 27 W, laser beam diameter 6 mm) was used to administer heat pulses of 60°C (SIFEC s.a.,). In principle, the temperature sensor and target presence detector are located in the aperture head which receives in both directions via optical fibres the CO2 laser beam controlled by a computer in real time for all data required to transmit the specified power for definite temperature.
We recorded a dose-response function for the heat stimulus and the corresponding pain magnitude (NRS). The CO 2 laser temperature was set at 60°C and heat pulses were delivered at increasing steps of 20 ms up to 100 ms pulse duration. Thereby, the heat pulse gradually reached deeper epidermal layers upon prolonged stimulation. This stimulation paradigm was used for a broader spatial distribution of the heat stimulus to different skin layers. Heat pulses of increasing duration were administered in 10-s intervals; sufficient time to allow the heat to move away from the administration site. CO2 laser-evoked pain was rated (NRS, 0-10) for each pulse by the volunteer and each skin site (NGF/NaCl) was measured twice in randomized order.

| A-fibre conduction block
A-fibre conduction block was induced by ischaemic compression of the upper arm as described previously (Casale et al., 1992;Koppert et al., 1998) on day 3 in n = 13 subjects (one subject with Morbus Raynaud was excluded from this part of the experimental protocol) and at day 7 in n = 5 subjects. In principle, a double-cuff tourniquet was placed around the bolstered upper arm (VBM Medizintechnik, Sulz, Germany) of the NGF/NaCl-treated forearm. Subjects lifted their arm and an Esmarch bandage was applied from the wrist up to the tourniquet to exsanguinate the arm. Thereafter, the double-cuff was inflated to 250 mmHg and the Esmarch bandage unwound. The arm was placed comfortably on a vacuum cushion (HEK medical GmbH, Ascheberg, Germany). The pressure of the cuff was kept at 250 mmHg until an A-fibre conduction block was achieved. In order to determine a conduction block, sensory tests for primary afferent units were applied in the vicinity of the NGF/NaCl-treated skin sites in 5-min intervals during ischaemic compression. Tests were performed in the following order: brushing the skin (SenseLab Brush-05, Somedic, Sweden), pinprick (256 mN, MRC Systems GmbH, Heidelberg, Germany), warm and cold stimuli applied by metal rolls set at 40°C and 18°C (Therroll, SBmedic, Sweden) for 3 s, transcutaneous electrical stimulation (rectangular pulses of 0.5 ms duration delivered for 1 s with 20 Hz, current intensity adjusted to individually assessed NRS 3 at 2 Hz as described above). A-fibre compression block was achieved when the sensation to brush was absent, the sensation to cold was either absent or perceived as burning hot and when the sensation to warmth was still present. Moreover, upon A-fibre block the sharp stinging pain of the pinprick stimulation was abolished and the pulsing character of pain induced by 20 Hz bursts of electrical stimuli was no longer perceived.

Sensory tests during A-fibre conduction block
During A-fibre conduction block we assessed pain upon slowly depolarizing transcutaneous electrical stimuli (as described in 2.2.1.) with current intensities of 0.4 mA (sinusoidal stimulation) and 1 mA (half sine stimulation) as well as mechanical impact pain (described in 2.2.2.). Tests were performed at day 3 (n = 13) and day 7 (n = 5) after NGF injection, that is at maximum CO 2 laser heat and mechanical impact pain. NGF-and NaCl-treated skin sites were investigated twice and in randomized order.

| Statistical analysis
Statistical analyses were performed with STATISTICA 7.1 (StatSoft Inc., Tulsa, US) using analysis of variance (ANOVA) and Bonferroni post hoc tests to identify significant differences (p < .05) between the factorial groups 'NGF versus NaCl treatment'-'stimulation intensity'-'day after NGF/NaCl injection' for pain and axon reflex erythema recordings. The levels for each factor, F numbers and the degrees of freedom for each ANOVA are given in T A B L E 1 Analyses of variance (ANOVA) for the factorial groups 'day after NGF/NaCl injection' (Day)/'NGF versus NaCl treatment' (NGF)/'stimulation intensity' (e.g. current intensity) for pain and axon reflex erythema and their interactions before (a) and during nerve fibre compression block (b)

(a) Factorial group/Interaction
Factor level  Figure 2a). Axon reflex flare responses recorded after the entire sine wave stimulation of 0.05-0.4 mA were significantly different between the factorial groups 'NGF versus NaCl treatment'-'day of investigation'-'time of the LDI recording' (F(16,192 = 1.96; p < .02, ANOVA), revealing a maximum flare of 3.1 ± 0.5 cm 2 on average at days 7-14 after NGF-treatment, compared to a maximum response of about 2.5 ± 0.5 cm 2 at the NaCl-treated site (Figure 2b).

| Pain during continuous sine wave stimulation
In order to compare pain development between NaCl-and NGF-treated skin sites during the continuous 1-min lasting sine wave stimulation, we adjusted the delivered current amplitudes to a value causing pain of NRS 2-3 at both NaCl and NGF sites that had been recorded before upon the 2.5 s lasting sine wave stimuli (reported above in 3.1.1). Hence, sine wave current amplitudes were set to 0.4 mA for NaCl skin site stimulation, but to only 0.2 mA for stimulating NGFtreated skin to evoke a similar pain intensity. We are aware that different current amplitudes for the NaCl control group and the NGF-sensitized group introduced another variable in this investigative protocol. If current intensities were set to the same levels for the NaCl and NGF sites, differences of pain recordings between these sites would be driven by the reduced electrical excitation thresholds caused by the NGF challenge. This effect of reduced excitation thresholds was monitored during the dose-response sine wave stimulation protocol reported in Section 3.1.1. For the 1-min continuous sine wave stimulation, we aimed to monitor alterations in the temporal pain profile between the NaCl and NGF sites. It was of particular interest in this protocol to monitor the previously observed accommodation of nociceptors to ongoing stimuli (Jonas et al., 2018) in order to identify putative alterations of this process by NGF. To achieve this, we circumvented the additional variable of altered excitation thresholds on pain and adjusted the delivered current intensities to record time course changes of pain at the NaCl and NGF sites, respectively. In general, during ongoing sine wave stimulation, pain was strongest at about 10 s of stimulation and thereafter decreased throughout the 1-min period (F(6,78) = 13.94, p < .00001, ANOVA). Pain recordings revealed a significant interaction between the factorial groups 'NGF versus NaCl treatment'-'time of stimulation' (F(6,78) = 2.23, p < .05, ANOVA). In particular, subjects reported significantly more pain during 20 s of stimulation at the NGF site (0.2 mA, NRS 2.8 ± 0.2) when compared with NaCl-treated skin (0.4 mA, NRS 2.1 ± 0.3, p < .02, Bonferroni test) on day 7 although stimulus intensity was twice as high at the NaCl control site. On days 10 and 14, however, no significant differences were recorded between the sites (Figure 2d).

| Pain upon rectangular electrical pulses in NGF-sensitized skin
Transcutaneous electrical stimuli of 0.5 ms width square pulses were delivered with a frequency of 4, 20 and 100 Hz to the NGF-and NaCl-treated skin sites for durations of 1 s. The applied current intensity was set to amplitudes causing pain of NRS 3 determined at 2 Hz (see Methods above) and were on average 3 ± 0.2 mA at the NGF and 3.6 ± 0.2 mA at NaCl-treated skin sites. Pain increased significantly with main effects for the factorial group 'stimulus frequency' (F(2,26) = 102.97; p < .00001, ANOVA) and 'NGF versus (c) Pain ratings (NRS) after a single half sine wave pulse (500 ms, 0.2-1 mA) recorded from NGF-(solid square) and NaCl-injected skin (open circles) at day 1-14 after treatment. Sensation was perceived as significantly stronger for 0.6 and 1 mA half sine wave pulses after NGF-treatment (p < .005, ANOVA, and p < .002 Bonferroni post hoc test marked by asterisks). (d) Pain during 4 Hz sinusoidal stimulation delivered continuously for 1 min and recorded from the NGF site (solid squares) and the NaCl site (open circles). Note that current amplitudes for sine wave stimulation were adjusted to a pain NRS-level of 2-3 obtained during 10 sine wave pulse stimulation. Correspondingly, current intensities of 0.2 mA were delivered to the NGF site, but 0.4 mA were applied to the NaCl sites. For comparison, pain ratings recorded upon 0.2 mA from untreated skin are depicted at panel day 1 (open squares) and were recorded during the training session. Pain intensity was significantly different between treatment sites (NGF vs. NaCl) over the 60-s stimulation period (p < .05, ANOVA), particularly during 10-15 s of stimulation at day 7 after treatment (p < .02, Bonferroni post hoc test, marked by asterisks)

| Evoked pain in NGF-sensitized skin during A-fibre conduction block
A-fibre conduction block was achieved after 29.5 ± 4 min of ischaemia. At that time, skin surface temperature lowered from 30.5 ± 1.5°C before block to on average 19 ± 2°C during the block [assessed by a noncontact thermometer (STProPlus, Raytek)]. Warmth perception was still present in all subjects and cold sensation was abolished. Pain upon supra-threshold electrical rectangular pulses of 20 Hz was NRS 4 ± 0.4 at ischaemia onset and was almost completely blocked (NRS 1 ± 0.3) after 30 min. It had entirely lost its pulsing character, indicating that the rectangular pulses evoked pulsating pain not via C-fibres. Thus, the test is ideally suited to screen for a differential nerve fibre block.

| DISCUSSION
Inhibition of nerve growth factor (NGF) signalling is analgesic in chronic inflammatory pain (Lane et al., 2010;Tiseo et al., 2014). In addition to traditional threshold measures, assessment of C-nociceptor electrical excitability may broaden the characterization of hypersensitivity in chronic pain patients. Here, we investigated sensitization of C-nociceptor sub-classes to NGF using electrical sinusoidal stimulation paradigms (Jonas et al., 2018;Rukwied et al., 2020). We found enhanced axonal excitability of mechano-sensitive C-nociceptors to 1 Hz (500 ms half sine) stimulation. Both mechano-sensitive and mechano-insensitive C-nociceptors showed an increased ability to maintain firing during sustained 4 Hz stimulation. Larger axon-reflex-mediated flare responses indicated augmented recruitment of mechano-insensitive C-nociceptors and a steeper stimulus-pain rating curve for CO2 laser heat-facilitated action potential generation in heatsensitive C-nociceptors with higher discharge frequencies.

| Facilitation of action potential generation in C-nociceptors by NGF
Intradermal injection of NGF is known to facilitate action potential generation and conduction Obreja et al., 2018;Rukwied et al., 2014). In order to induce action potential discharge preferentially in nociceptors we designed an electrode (Landmann et al., 2016) and a stimulus profile (Rukwied et al., 2020) to specifically depolarize mechanosensitive C-nociceptors. A half sine wave constant current stimulus provokes a burst of action potentials (Rukwied et al., 2020) and we found that NGF increased the pain ratings to the most intense amplitude of half sine wave stimulation. In F I G U R E 4 Slowly depolarizing electrical and mechanical impact pain during A-fibre block. (a) Sinusoidal currents (4 Hz, 0.4 mA, 2.5 s) were delivered at day 3 (left panel, n = 13) and day 7 (right panel, n = 5) prior to (control condition) and during A-fibre conduction block to NGF-treated skin (black and grey symbols) and NaCl injection sites (open symbols). Pain was reduced by about 50% during the conduction block at day 3 and day 7 (day 3 p < .0001 and day 7 p < .04, ANOVA, marked by hashtag) but sensation was significantly enhanced at the NGF site when compared to NaCl skin (day 3 p < .003 and day 7 p < .004, ANOVA, marked by asterisks). (b) Pain upon half sine wave stimulation (1 pulse 500 ms, 1 mA) recorded prior to (control condition) and during A-fibre conduction block (significant differences marked by hashtag, day 3 p < .0001 and day 7 p < .05, ANOVA). Stimuli delivered to the NGF sites (black and grey symbols) were felt as more painful compared to the NaCl injection site (open symbols) at day 3 (left panel, n = 13) and day 7 (right panel, n = 5) during A-fibre conduction block (day 3 p < .01 and day 7 p < .04, ANOVA, marked by asterisks). (c) Mechanical impact stimuli of 4 and 8 m/s were delivered at day 3 (left panel, n = 13) and day 7 (right panel, n = 5) prior to (control condition, ctr) and during A-fibre conduction block to NGF-treated skin (solid symbols) and NaCl injection sites (open symbols). Pain was significantly elevated at the NGF sites at both days (day 3 p < .002 and day 7 p < .05, ANOVA, marked by asterisks), particularly upon stimuli of 8 m/s delivered during the A-fibre block (day 3 p < .0001, Bonferroni post hoc test, marked by asterisks) | 395 SCHNAKENBERG Et Al.
DRG neuronal somata, maximum evoked discharge frequency to depolarizing voltage steps vary considerably between different classes of primary afferent neurons (Zheng et al., 2019). For instance, a 500 ms depolarization induced firing frequencies of about 10 Hz in non-peptidergic nociceptors, but some 25 Hz in low threshold mechano-sensitive C-fibres (Zheng et al., 2019). The authors identified a distinct constellation of voltage-gated ion-channels in various populations of sensory afferents that could underlie the unique discharge properties recorded upon depolarization (Zheng et al., 2019). Notably, low-intensity half sine wave electrical stimuli also activated low-threshold mechano-sensitive (LTM) C-fibres (Rukwied et al., 2020) but did not evoke any sensation. We therefore conclude that LTMC-fibres are less likely to contribute to enhanced pain responses following NGF. We further provide indirect evidence that NGF sensitizes mechano-sensitive C-nociceptors and facilitates higher frequency discharge.
The second protocol assessed action potential discharge during continuous 4 Hz sine wave stimulation for 1 min (Jonas et al., 2018) to evaluate the ability of mechano-sensitive and mechano-insensitive ('silent') nociceptors to maintain firing for prolonged periods. Sine wave stimulation at 4 Hz for 1 min was perceived as more painful at the NGF sites. With the stimulus intensity adjusted the same initial NRS-levels at NGF-and NaCl-treated skin sites 50% lower current intensities were required to evoke the same pain intensity at NGF-sensitized skin. One significant feature of 1 min, 4 Hz sinusoidal stimulation is the distinct accommodation of pain in healthy volunteers whereas neuropathic pain patients report an increase in pain sensation during the stimulus in their symptomatic skin (Jonas et al., 2018). Surprisingly, even though we recorded significantly enhanced pain during 1-min sine wave stimulation at the NGF sites, the accommodation during the stimulus appeared similar to that for normal skin and this observation is in contrast to the time profile recorded from neuropathic pain patients (Jonas et al., 2018). Our present results substantiate intradermal NGF injections as an experimental human model for chronic inflammatory pain rather than painful neuropathy. Accordingly, clinical studies with anti-NGF antibodies showed no efficacy in neuropathic (Wang et al., 2017) but profound reductions in primary outcomes for chronic inflammatory pain (Lane et al., 2010;Tiseo et al., 2014).
Sine wave stimulation at 4 Hz activates silent nociceptors (Jonas et al., 2018) and evokes larger erythema areas at the NGF sites. This result might have been expected as erythema development depends on activation of 'silent' C-nociceptors activation (Schmelz et al., 2000) and facilitated axonal excitation of these fibres was recorded in pig and human skin after NGF (Hirth et al., 2013;Obreja et al., 2018). Interestingly, even when using supra-threshold rectangular electrical pulses, we recorded unchanged axon reflex erythema in our previous NGF studies (Rukwied et al., 2010(Rukwied et al., , 2014. Here, we saw an increase in the erythema response after NGF which was apparently attributed to an activation of 'silent' C-nociceptor activation. Thus, the slow depolarizing stimulus delivered at 4 Hz appears to be more suited to identifying NGF-induced axonal excitability changes in 'silent' C-nociceptors.

| Sensitized signal transduction
Very superficial heat stimuli were achieved with a feedback-controlled CO 2 laser that applied 60°C for durations of 20-100 ms. Longer pulses allowed deeper penetration of heat into the skin (Marchandise et al., 2014). Both A-delta and C-nociceptors respond to laser heat stimuli (TREEDE et al., 1995;Wooten et al., 2014) with activation thresholds of about 47°C versus 40°C, respectively [determined by heat pulses of 50 ms duration with an initial 10 ms ramp (Churyukanov et al., 2012)]. Hence, our stimulus may have activated both nociceptor classes, but certainly included heat-sensitive C-fibres. Experiments in monkeys have shown that rapid heating by laser provokes bursts of about 70 Hz in C-nociceptors (Wooten et al., 2014). In NGF-treated skin, we observed a significantly stronger stimulus duration-response effect as compared to NaCl over 14 days. Acute heat sensitization by NGF is caused by enhanced expression and phosphorylation of the ion-channel TRPV1 (Bhave et al., 2003;Bonnington & McNaughton, 2003;Zhang et al., 2005), which is essential for heat hyperalgesia (Davis et al., 2000). In addition to increased heat transduction mechanisms, the increase in pain perception upon sinusoidal electrical stimulation suggests a lower axonal spike generation threshold following NGF. Indeed, the increase in pain ratings upon stronger, that is longer lasting, heat stimuli (Figure 2c), was significantly steeper at the NGF site. A possible explanation for the steeper slope of pain increase could be facilitated action potential generation leading to higher discharge frequencies of heat-sensitive C-nociceptors, as already discussed above for the response to slow depolarizing 500 ms half sine wave stimuli.

| Conduction block of A-fibres
A-fibre conduction block was used to investigate the contribution of C-nociceptors sensitized by NGF. The absence of A-delta nociceptor conduction following pressure block was confirmed by the loss of cold sensation and absence of sensation to transcutaneous supra-threshold electrical rectangular pulses delivered at 20 Hz. This latter approach proved to be a simple test for A-fibre conduction block when compared with commonly used pain reaction time measures (Henrich et al., 2015;Ziegler et al., 1999). We recommend this confirmatory test procedure for future studies involving nerve fibre block using pressure. Certainly, conduction in some mechanically sensitive C-fibres may also be impaired following pressure block and C-nociceptors can conduct frequencies of 20 Hz for short time periods (Weidner et al., 2002;Zheng et al., 2019) but only A-delta nociceptors can transmit the 20 Hz pulses in a synchronized fashion to the spinal cord and thereby evoke pulsating pain. Loss of conduction in A-delta nociceptors will mute this pain. Our data confirm earlier studies using differential nerve block of the superficial radial nerve in volunteers showing that mechanical impact pain is primarily mediated by C-fibres (Kilo et al., 1994). Here, we demonstrated that 4 Hz and 1 Hz sinusoidal current stimuli were more painful at the NGF-treated site. Compared to control skin, we recorded a significant pain reduction at both NGF and NaCl skin sites during the A-fibre block. It is possible that nociceptive signalling is affected by the ischaemia and temperature changes accompanying the double-cuff tourniquet. However, physiological increases of cutaneous sympathetic nerve activity that might be expected during ischaemia have previously been shown to not alter the firing properties of unmyelinated (polymodal) skin nociceptors in human subjects (Elam et al., 1999). Interestingly, mechanical impact pain was not reduced during A-fibre block, which might be linked to the endings activated by this strong mechanical stimulus being at a deeper location and thus somewhat warmer than the skin. It will be of interest to investigate systematically the influence of skin surface temperature on pain responses caused by sinusoidal currents, particularly in light of recent findings demonstrating that lower temperatures increase the excitability of unmyelinated nerve fibres ex vivo (Hugosdottir et al., 2019).