TRPV1 is a sodium/calcium ion channel expressed in a subpopulation of DRG neurons that respond to noxious heat, endogenous algesic compounds, and the vanilloid agonist capsaicin 123. Capsaicin responses are detected in subpopulations of unmyelinated C-fiber neurons and myelinated Aδ-fibers 45. Electrophysiological studies with radiant heat have shown that thermal sensing C-fibers mediate responses to stimuli that heat the skin at low rates (≤ 0.9°C/sec) whereas Aδ-fibers mediate responses to stimuli that heat the skin at high rates (≥ 6.5°C/sec) 6. TRPV1 in C-fibers is responsible for burning pain sensations plus the integration of inflammatory chemical signals in many pathological pain states, and multiple drug development efforts have been directed at antagonizing TRPV1 for pain control 789. TRPV1 agonists, such as the ultrapotent capsaicin analogue resiniferatoxin (RTX), have also been proposed as therapeutic agents for treating acute and chronic pain 1011. The binding of RTX leads to a sustained influx of sodium and calcium through TRPV1 channels 12 leading to channel desensitization and/or the loss of TRPV1-expressing DRG neurons and/or their fibers and terminals via calcium-induced cytotoxicity 1314. Thus, although the mechanisms diverge, ultimately either agonists or antagonists can be used as analgesic agents.

While antagonists can be administered orally, for agonists local administration is required and the route or site of administration is a critical factor. For example, RTX injection into the intrathecal space results in a loss of centrally projecting TRPV1-expressing fibers in the dorsal roots and at higher concentrations a loss of TRPV1-expressing DRG neuronal perikarya; both cases produce permanent regional analgesia 141516. In contrast, injections of low concentrations of RTX into peripheral sites (e.g., subcutaneous injections) spare the neuronal perikarya while ablating or temporarily inactivating TRPV1-expressing peripheral terminals and fibers 1718. This approach therefore results in temporary analgesia at focal sites until the fibers reactivate or regenerate. Systemic injections of RTX have also been used to induce analgesia; however, higher concentrations of RTX are needed and the analgesic effect is widespread rather than regional or focal 19.

Although TRPV1 is expressed in C- and Aδ-fibers, most animal studies that have ablated TRPV1 fibers with capsaicin or RTX focus on behavioral responses associated with C-fibers 14172021. This can be attributed, primarily, to the prevalence of thermal pain assays that examine C-fibers rather than Aδ-fibers in awake rodents. Indeed, in the rat, where the axons are relatively short (20 cm or less), it may be difficult to distinguish behavioral responses mediated solely by fast-conducting Aδ-fibers. For example, the conduction velocity of the relatively slower C-fibers is sufficient enough to produce laser evoked potentials (LEP) as early as 225 ms after heating the paw with a brief CO₂ laser pulse 22.

One goal of this study was to evaluate the usage of an infrared diode laser to examine behavioral responses that are discretely associated with Aδ activation in the rat. Infrared diode lasers, in contrast to CO₂ lasers, are capable of directly heating the skin at depths where cutaneous nerve fibers terminate, thereby providing rapid (capable of > 200°C/sec), efficient, non-damaging thermal stimulation of Aδ-fibers 232425, which can also initiate rapid behavioral responses. We combined RTX-induced ablation of TRPV1 fibers in plantar hind paw 17 with infrared laser stimulation and provide an excellent model to examine the contribution of Aδ-fibers to nociceptive behaviors in rats. Our findings demonstrate that RTX-induced thermal hypoalgesia is sustained longer than previously thought, and RTX greatly attenuates behavioral responses characteristic of Aδ stimulation. Importantly, this methodology suggests infrared laser stimuli can be used to analyze behavior associated with fast-conducting Aδ-fibers, which has been relatively difficult to accomplish in awake, unrestrained rodents.

We sought to specifically stimulate afferent nerve fiber subtypes in the glabrous skin of the rat hind paw using either short-pulse, high-rate (Aδ) or long-pulse, low-rate (C) skin heating with an infrared diode laser. Studies in rats have suggested that withdrawal activity following short-duration, high-intensity stimulation of the hind paw with a CO₂ laser is mediated by C-rather than Aδ-fibers, in part from recordings of electromyographic (EMG) activity in the stimulated limb 26 and/or of laser-evoked cortical potentials (LEPs) 222728, that show latencies consistent with a conduction velocity in the C-fiber range. The infrared diode laser used here has the ability to specifically activate Aδ fibers, which has been determined from direct extracellular recordings of trigeminal ganglion neurons in rats 25. In order to further establish a relationship between Aδ conduction and nociceptive response a high-speed video camera (500 fps) was used to record stimulation-induced movements in awake, freely-moving rats during the brief peristimulus time period.

With the high-speed camera we observed that the actual withdrawal of the stimulated limb was often preceded by a panoply of dedicated motor programs. For example, video frames of a rat before, and in the first 300 msec after stimulation (6.08 W/mm2) are shown in Figure 1. In this example, visible head movement towards the stimulated hind paw could be detected by 110 msec. The right eye was also seen to squeeze shut by 110 msec, however, this behavior was not a prominent feature in many of the trials. Figure 1C illustrates that by 210 ms, the stimulated toe has been retracted off the glass platform, and by 310 ms withdrawal of the hind limb is fully apparent (Figure 1D). The latency from the noxious phase (the period after the first 30 ms from when the laser stimulation began, see Methods) of the stimulus to the first observable laser-evoked movement (fLEM) in this trial was 74 ms which occurred in a forelimb rather than the stimulated limb (see also Additional file 1 Movie S1). In contrast, the latency to fLEM in the stimulated limb was 112 msec. Nerve length from the toe to the dorsal root entry zone was 16.8 cm, based on dissection (350 g rat). Hence, this example suggests a minimum conduction velocity of 2.27 m/s (0.168 m/0.074 s), which is well within the range for Aδ-fibers [≥ 1.3 m/s] 2930 and beyond the range for the TRPV1 population of C-fibers [~0.4 m/s] 31.

Calculated estimates of conduction velocity based on latency to fLEM for a series of normal rats are listed in Table 1. We observed that response latency appeared highly dependent on the rat's posture at the moment of stimulation. From 16 video recorded trials, 6 of the 8 shortest latencies (n = 4 rats) to fLEM occurred when body weight was not supported by all four limbs. In these cases, one or both forepaws were not on the floor and hence the rat was in the "reared" position prior to stimulation. For these 16 trials, the shortest latency to fLEM was 56 ms and the average was 159.5 ± 36.7 msec. In contrast, 6 of the 8 longest latencies to fLEM occurred when weight was distributed to all four paws; from these trials, the shortest latency was 74 ms and the average was 250.0 ± 28.9 msec. It should be emphasized that our conduction velocity estimates are conservative. They do not account for central delay (possibly as high as 80 ms in the rat) 32, motoneuron conduction velocity (33-85 m/s) 33 or the time required for muscle contraction/relaxation (34-116 ms) 33, which would necessarily yield faster conduction velocity estimates. Furthermore, the cascade of bilateral extensor/flexor muscle activity required for abrupt postural adjustment is complex, and the possibility is strong that the earliest muscle twitches in some trials may have been missed entirely by the viewing angle of the camera. We cannot predict, at present, where movement will occur first. Despite this, we could capture in 31% of trials movements associated with nerve conduction velocities higher than 1.7 m/s, clearly within the range for Aδ-fibers, following a short, intense noxious laser stimulus. We conclude that Aδ-fibers must be recruited in some, if not all, responses under this stimulus paradigm.

Rats were treated with intraplantar injection of 5 or 50 ng RTX and examined behaviorally over several weeks. Two stimulus intensities were used to examine the effect of RTX on Aδ-fiber activity. At baseline, the probability of withdrawal to the lower power stimulus (5.12 W/mm²) was 0.69 ± 0.08 and the severity of withdrawal (on the behavioral rating scale, see Methods) was 2.61 ± 0.24 (Figure 2A and 2B). After RTX injection, the probability of withdrawal and the severity of withdrawal were eliminated or barely detectable during the first two weeks. The 50 ng RTX-treated paws remained significantly suppressed for up to 5 weeks. By week 7, there was no longer a significant difference in behavioral responses to the Aδ stimulus between the RTX and vehicle treated paws. A statistically significant difference was observed between the 5 and 50 ng doses over the course of the study (p < 0.01, probability of withdrawal, Figure 2A; p < 0.001, intensity of withdrawal, Figure 2B).

When stimulating with higher power (6.08 W/mm²), the probability of withdrawal and intensity of withdrawal were close to maximum at baseline (0.92 ± 0.05 and 3.33 ± 0.19, Figure 2C and 2D, respectively). The stronger behavioral responses with the 6.08 W/mm² stimulus indicate more efficient heating of the skin as compared to the 5.12 W/mm² stimulus. After 50 ng RTX, rats were almost completely unresponsive to the 6.08 W/mm² Aδ stimulus for the first two weeks. By three weeks there was a substantial return towards baseline, nonetheless, a small but significant difference could be detected up to week 5 suggesting a small deficit in fiber function remained. By week 7, there was no difference between vehicle and RTX-treated paws, similar to the 5.12 W/mm² setting. Responses from the 5 ng RTX treated paws returned to baseline levels by week 3. As with the lower power setting, a statistical difference between the 5 and 50 ng groups was observed when stimulating with the 6.08 W/mm² laser pulse (p < 0.01, probability of withdrawal, Figure 2C; p < 0.05, intensity of withdrawal, Figure 2D).

Our previous work demonstrated that 62.5 ng of RTX attenuates C-fiber responses for about 2 weeks 17. In the present study, normal rats withdrew from the laser generated C-fiber stimulus at about 8 s (Figure 2E). In rats lightly anesthetized with 0.5% isoflurane, a withdrawal response is retained, and the skin temperature (stimulation site) at the time of withdrawal was 48.5 ± 0.5°C when measured using a thermal camera [data not shown]. Both doses of RTX produced significant increases in withdrawal latency compared to the vehicle injected (contralateral) paw. Withdrawal latencies at the mid-plantar injected site reached cutoff (16 sec) during week 1, the latencies then decreased by week 2 but remained significantly different from contralateral paws until they finally normalized by week 3 (Figure 2E). Analgesia occurred with both doses of RTX and the time course of recovery was similar but the effect, however, was greater in the 50 ng group (p < 0.05). Figure 2F and 2G directly compare behavioral responses to the Aδ and C stimuli on day 18; after 50 ng RTX, rats were generally unresponsive to the 5.12 W/mm² Aδ stimulus (as compared to vehicle (p < 0.05). In contrast, responses to the C stimulus were not significantly different between the RTX and vehicle groups by day 18.

The mid-plantar hind paw is innervated by the tibial nerve. Since the toes are also innervated by the tibial nerve, we separately examined the effect of intraplantar RTX on recovery kinetics in the toes. The probability of withdrawal for the toes using the 5.12 W/mm² Aδ stimulus was 0.72 ± 0.08 prior to RTX or vehicle treatment, similar to that obtained with mid-plantar stimulation. By 4 weeks, the probability of withdrawal and the intensity rating were near maximal for the vehicle-injected paw (0.94 ± 0.04 and 3.47 ± 0.16, respectively) (Figure 3A and 3B). In contrast, on the RTX injected side, the toe response was significantly reduced for at least 7 weeks for the 5 ng group and 10 weeks for the 50 ng group. Stimulation of the toes at higher intensity (6.08 W/mm²) resulted in higher baseline responses (probability of withdrawal = 0.89 ± 0.05 and response rating = 3.33 ± 0.21). Despite the stronger intensity, significant reductions in behavioral responses mediated by this Aδ stimulus could be seen for up to 7 weeks after intraplantar RTX (Figure 3C and 3D). These data show that behavioral sensitivity to Aδ stimuli is regained more slowly in the toes compared to the mid-plantar injection site. However, as with the mid-plantar site, analgesia in the toe was significantly greater in the 50 ng as compared to the 5 ng group (p < 0.001, Figure 3A and 3B; p < 0.05, Figure 3C).

Responses to C stimuli at week 5 revealed a small but significant difference between RTX (50 ng) and vehicle-treated toes (Figure 3E). By 10 weeks, there was no difference. There was no measurable statistical difference between the 5 ng RTX and vehicle-treated paws at week 5 or week 10.

Application of RTX initially evokes nociceptive activity by binding to and activating TRPV1 14. This is followed by a dying-back (axonopathy) of unmyelinated and myelinated TRPV1-expressing fibers and potentially a recruitment of macrophages that can release cytokines, which in turn could activate the remaining proximal end of the nerve fiber. Using RT-PCR, we examined the DRG and spinal cord for expression of genes associated with nociceptive activity and nerve regeneration after RTX-treatment. As shown in Figure 4 the gene encoding the nociceptive chemokine MCP-1 3435 was increased in the DRG at 24 h, likely due to an initial burst of nociceptive activity following TRPV1 activation 36, but expression normalized by day 10 (Figure 4A-C). In contrast, the nerve regeneration marker ATF3 37 was induced at 24 h but the increase was maintained through day 10 (Figure 4A, B and 4D). The regeneration marker galanin was also increased at day 5 and 10 (Additional file 2). Upregulated c-Fos in the spinal cord is also an early marker of nociception 3839. Consistent with the elevation in ganglionic MCP-1 transcripts, c-Fos protein levels increased early (6 h) but reverted by 24 h (Figure 4E-I), indicating that regeneration after RTX-induced axonopathy is devoid of ongoing spontaneous nociceptive activity sufficient to induce these markers.

In the present study, we examined (a) behavioral responses associated with thermal laser stimulation of Aδ- and C-fibers in awake rats, (b) such behavioral responses after peripheral ablation of the two fiber types following intraplantar administration of the vanilloid agonist RTX, (c) the time-course of recovery/regeneration, and (d) the expression of markers for nociception and nerve regeneration. The results demonstrate that the infrared diode laser can be tuned to evoke behavioral responses temporally consistent with activation of Aδ-fibers in the rat hind paw. The results also demonstrate that mid-plantar injection of RTX removes both Aδ- and C-fiber mediated thermal sensations in the footpad and toes. This is followed by a progressive recovery of behavioral function from proximal (mid-plantar) to distal (toes) over several weeks.

One of the main aims of the present report was to evaluate infrared laser evoked Aδ- and C-fiber nocifensive behaviors. Two approaches were used: analysis of behavioral reactions in the immediate 300 msec of the post-stimulus period and modulation of behaviors following intraplantar injection of RTX. We first established, with high-speed videography, that nocifensive behaviors following short duration (100 msec), high-intensity laser pulse stimulation (6.08 W/mm²) are mediated by Aδ-fibers, consistent with reported electrophysiological 2425 and laser-evoked potential 22264041 observations. Importantly, the latency to withdrawal of the stimulated limb was not always useful for estimating the conduction velocities of activated fibers. Instead, by capturing the behavior of the entire animal with a high-speed camera, we often observed that a rapid set of movements occurs before withdrawal of the stimulated limb. In most cases the latencies of these early movements indicated a faster afferent conduction velocity than if the latency to withdrawal of the stimulated limb was used as the endpoint. The progression of motion after the first observable laser-evoked movement (fLEM) was often consistent with postural adjustments to ensure stability that preceded a later, yet brisk, withdrawal of the stimulated limb. The shortest latency to fLEM was 56 ms, yielding a conduction velocity estimate for the primary afferent volley of 3 m/s, which is in the range for Aδ fibers and well above the range for C-fibers. In a previous study using a CO₂ laser, EMG recordings taken from the gluteus maximus and biceps femoris of the stimulated limb yielded evoked electrical activity at ~280 ms 26. However, these studies were performed on anesthetized animals whereas the present observations, obtained from awake unrestrained rats, suggest that complex postural changes and even reflexive Aδ-type behaviors are suppressed when the animal is under anesthesia. Our high speed imaging of the earliest, reflex-like movements is indicative of an Aδ-mediated nociceptive response to 'first pain' 42 and an activation of the Aδ "alarm system" 43.

When rats supported their body weight on 2 or 3 paws, fLEM often occurred at < 160 ms and frequently the movement occurred on the contralateral (i.e., unstimulated) side. This suggests that a readily observable short latency Aδ-evoked movement is more likely to be revealed when rats are in a "reared" position, or any posture that opportunistically requires rapid mediolateral balance adjustment prior to paw withdrawal. In contrast, when rats are standing with body weight distributed to all four legs, and therefore more neutrally balanced, latency to fLEM increased to ~250 msec. When all four paws are on the ground, early postural adjustments are difficult to detect if they primarily involve the axial musculature rather than repositioning of the forepaws or head and neck, all of which can occur in the first 100 msec following the end of the laser pulse. An additional consideration is that in the rat, central delay to acute laser stimulation has been reported to last for up to 80 ms 32. Whether this makes a significant contribution to "preparatory" adjustments 44 or long latency movements is being further investigated but it is clear that controlled balance at the moment of retraction of the stimulated paw is a priority for spinal sensory-motor circuits (for an example in humans, see 45).

The actions of RTX are manifested in both Aδ- and C-fiber types. We observe that peripherally administered RTX temporarily eliminates behavioral responses to both Aδ- and C-fiber mediated thermal nociception. Our data corroborate previous human studies in which, after prolonged capsaicin treatment, cortical LEPs corresponding to activated Aδ- and C-fibers are eliminated, as well as the behavioral response 4647. The capacity to independently test the two fiber types using our current laser settings suggest differential recovery rate for Aδ compared to C-fibers. After administering the higher dose of RTX (50 ng), responses to the C-fiber stimulus recovered first, but responses to the Aδ stimulus lagged by several weeks. However, the C-fiber stimuli also covered a greater area of skin, increasing the likelihood of activating a functional nerve ending. In both cases, responses to the C-fiber and Aδ stimuli, which were initially eliminated, returned gradually rather than abruptly. The gradual return is consistent with the idea that, within a single fiber type, the rate of regeneration is heterogeneous. It is also possible that, during recovery, C-fibers played a more pronounced role in mediating behavioral responses as they were no longer preempted by the faster Aδ input 48. The lag in recovery of the full, archetypal Aδ behavioral responses after RTX may indicate that the Aδ-fibers are greatly compromised and/or undergo a more extensive axonopathy and/or that return of function depends on complete re-myelination. Because the diameter of the Aδ laser spot was small we could stimulate multiple discrete areas of skin and spatially map the thermal sensitivity of the paw. Spatial mapping of the paw demonstrated a slower return of distal sensitivity (toes) compared to return of sensitivity proximal to the injection site (mid-plantar). Therapeutically, prolonged behavioral recovery at sites distal to an RTX injection indicates that regional analgesia can be sustained longer by injecting RTX more proximal to the DRG and has potential therapeutic ramifications for treatment of peripheral neuropathic pain problems subsequent to nerve injury or possibly chemotherapy.

During the first 3 to 5 weeks of repeated testing, we observed that the rats became progressively more sensitive to the Aδ stimulus (Figure 2 and 3). This may be due to one or more concurrent processes, including (a) cued or non-cued learning or anticipation, which may be greatly influenced by the aversive nature of the stimulus, and (b) habituation to the novelty of the testing environment. Interestingly, the rats did not become more sensitive to C-fiber stimuli over the course of the study. This likely reflects a specific difference between the C and Aδ stimuli. Our C-fiber test involves slow-rate heating over several seconds (average latency ~8 sec) and the sensation of warming occurs well before the stimulus becomes noxious. Even if distracted by environmental cues, the warming phase can act as a thermal cue for the subsequent noxious threshold. Nonetheless, the rats maintained consistent withdrawal latency to C-fiber stimuli and did not exhibit a trend towards shorter latencies after repeated testing. This suggests that neither learning or anticipatory cues nor habituation critically affect responses in longitudinal C-fiber testing. In contrast, the sharp, pricking sensation of an Aδ stimulus (as reported by human subjects, 49) may make it substantially more aversive than a C-fiber stimulus and therefore couple it more tightly to anticipatory or learning circuits and suggest that further exploration of these aspects is warranted.

The initial RTX-induced nociceptive phase, subsequent analgesia, and nerve regeneration were each discernible at the molecular level. We assessed the temporal regulation of several molecular markers of nociceptive and regenerative processes in DRG or spinal cord following intraplantar injection of RTX. MCP-1 is widely thought to be a marker of pain since transcript and protein levels are upregulated in the DRG by peripheral inflammation 36 and in neuropathic pain models 505152. With intraplantar RTX, we observed that ganglionic MCP-1 was induced at 24 h, consistent with our previous findings 36. The increase in MCP-1 was not sustained: no elevation was detected at day 5 (data not shown) or at day 10. In contrast, ATF3, a marker of nerve regeneration 3753, exhibited an increase at 24 h that was sustained at both day 5 and 10, indicative of a long-lasting regenerative process after intraplantar injection of RTX. Strong upregulation of galanin, another marker of nerve regeneration, was also detected on day 5 and 10. Cumulatively these data suggest that processes involved in peripheral nerve regeneration do not contribute to the maintenance of MCP-1 upregulation. Further investigation may provide clues to the mechanisms of MCP-1 upregulation in neuropathic and inflammatory models of pain and the heterogeneity of neuronal expression in the DRG. At least in the case of RTX, action potentials or calcium signaling 12 may be involved, since RTX induces acute, prolonged nociceptive behaviors at the 50 ng dose 36 and a massive influx of calcium 13 precedes desensitization. It is also possible, however, that MCP-1 is induced by the initial dying back of the TRPV1-expressing fibers (e.g., due to a loss of trophic factor signaling) rather than the initial burst of nociceptive activity. Consistent with the transient increase in MCP-1 following peripheral RTX, we observed that expression of the c-fos gene, measured by c-Fos protein immunostaining, was transiently increased in the spinal cord immediately after intraplantar RTX injection, consistent with our previous findings 17. Induction of c-Fos in the spinal cord is widely accepted as a marker of pain signaling 3839. A drop in c-fos gene expression towards baseline levels by 24 h further indicates that RTX induces an immediate inactivation of nerve fibers since sustained nociceptive input can produce a long-lasting change in Fos proteins in dorsal spinal cord 38. The lack of sustained elevation in protein occurs concurrently with a pronounced behavioral analgesia.

Together our data demonstrate that analgesia following intraplantar administered RTX lasts much longer than previously assumed. The previous underestimate is due, in part, to the undetected extended time required for complete behavioral recovery to an Aδ stimulus. Highspeed videography suggests that Aδ nociception engages a complex process of rapid sensory-motor integration involving coordinated postural and balance adjustments prior to limb retraction. Infrared laser stimulation in awake, unrestrained animals can be used to assess the behavioral role of Aδ-fibers in several pain models and to differentiate mechanisms related to myelinated and non-myelinated fiber populations.

MIN is the CEO of Lasmed, LLC., which provided the laser and initial stimulus protocol for this work.

BDB, JN, ML carried out the behavioral studies. JN, ML, NM and LS participated in the immunohistochemical and gene analysis. KM, BDB, JMK and MJI participated in the design of the study and performed data analysis. KM, JMK, MIN and MJI coordinated the study and drafted the manuscript. GH conceived of and conducted the high-speed videography recordings. All authors read and approved the final manuscript.