Role of Peptidergic Nerve Terminals in the Skin: Reversal of Thermal Sensation by Calcitonin Gene-Related Peptide in TRPV1-Depleted Neuropathy

To investigate the contribution of peptidergic intraepidermal nerve fibers (IENFs) to nociceptive responses after depletion of the thermal-sensitive receptor, transient receptor potential vanilloid subtype 1 (TRPV1), we took advantage of a resiniferatoxin (RTX)-induced neuropathy which specifically affected small-diameter dorsal root ganglion (DRG) neurons and their corresponding nerve terminals in the skin. Thermal hypoalgesia (p<0.001) developed from RTX-treatment day 7 (RTXd7) and became normalized from RTXd56 to RTXd84. Substance P (SP)(+) and TRPV1(+) neurons were completely depleted (p = 0.0001 and p<0.0001, respectively), but RTX had a relatively minor effect on calcitonin gene-related peptide (CGRP)(+) neurons (p = 0.029). Accordingly, SP(+) (p<0.0001) and TRPV1(+) (p = 0.0008) IENFs were permanently depleted, but CGRP(+) IENFs (p = 0.012) were only transiently reduced and had recovered by RTXd84 (p = 0.83). The different effects of RTX on peptidergic neurons were attributed to the higher co-localization ratio of TRPV1/SP than of TRPV1/CGRP (p = 0.029). Thermal hypoalgesia (p = 0.0018) reappeared with an intraplantar injection of botulinum toxin type A (botox), and the temporal course of withdrawal latencies in the hot-plate test paralleled the innervation of CGRP(+) IENFs (p = 0.0003) and CGRP contents in skin (p = 0.01). In summary, this study demonstrated the preferential effects of RTX on depletion of SP(+) IENFs which caused thermal hypoalgesia. In contrast, the skin was reinnervated by CGRP(+) IENFs, which resulted in a normalization of nociceptive functions.


Introduction
Neuropeptides such as calcitonin gene-related peptide (CGRP) and substance P (SP) are presumably responsible for transmitting nociceptive stimuli from sensory nerve terminals of the skin. CGRP is implicated in several painful conditions, for example elevation of CGRP in migraines [1], and increased CGRP(+) nerve fibers in tumor-bearing bone tissues [2]. The differential expression of the CGRP gene was related to variable phenotypes of thermal responses to nociceptive stimuli [3]. The reduced expression of CGRP in the spinal dorsal horn was accompanied by reduced thermal sensations [4,5]. However, there are limited reports providing direct evidence of the role of CGRP in cutaneous nerve terminals for thermonociception after nerve degeneration.
In some patients with small-fiber sensory neuropathy, thermal hypoalgesia is associated with skin denervation [6,7]. Nerve terminals of CGRP and SP phenotypes in the skin belong to smalldiameter neurons, the soma of which are located in dorsal root ganglia (DRGs). These neurons also express transient receptor potential vanilloid subtype 1 (TRPV1) [8], which can be depleted by an ultrapotent capsaicin agonist, resiniferatoxin (RTX) [9,10]. To investigate pathophysiology of small-fiber sensory neuropathy, we and other groups previously established a system of RTXinduced neuropathy with thermal hypoalgesia due to cutaneous nerve terminal degeneration, particularly peptidergic nerve terminals [11][12][13][14]. In this model, there is a lack of systematic studies investigating the relationship between TRPV1 and neuropeptides and the influence of such changes on nociceptive functions after RTX-induced neuropathy, including the peptidergic profiles in DRG neurons and molecules compensating for the loss of TRPV1(+) neurons.
Botulinum toxin type A (botox) has emerged as an effective treatment for various pain disorders including migraine control by intramuscular administration [15] and bladder cystitis therapy by intravesical treatment [16]. A potential analgesic mechanism is related to interference with neurotransmission [17]. We hypothesized that CGRP might act as a candidate molecule and asked (1) whether botox caused cutaneous nerve degeneration or only changed the expression of neuropeptides and (2) how these changes paralleled thermal responses in RTX-induced neuropathy.
To address the above issues, we investigated (1) the peptidergic expression profiles of DRG neurons in relation to TRPV1, (2) the corresponding peptidergic innervation patterns of the skin, (3) nociceptive functions in relation to peptidergic innervation of the skin, and (4) the influence of botox on skin innervation in terms of peptidergic profiles of cutaneous nerve terminals.

Systemic RTX Treatment
RTX was used to create neuropathy in 8-week-old adult male ICR mice (35,40 g) [14]. Briefly, RTX (Sigma, St. Louis, MO) was dissolved in a vehicle (10% Tween 80 and 10% ethanol in normal saline). Animals received a single dose of RTX by an intraperitoneal injection (50 mg/kg, the RTX group). One group received an equal volume of vehicle as the control (the vehicle group). After an intraperitoneal injection, mice were housed in plastic cages on a 12-h light/12-h dark cycle and were given access to water and food ad libitum. All procedures were conducted in accordance with ethical guidelines for laboratory animals [18], and the protocol was approved by the Animal Committee of National Taiwan University College of Medicine (Permit Number: 20100021) and Kaohsiung Medical University (Permit Number: 100055). All experimental procedures were performed under 4% chloral hydrate (dose: 100 g in 1 ml) except for the behavioral evaluations which were performed on awake mice. All efforts were made to minimize suffering. Thermal withdrawal latencies were measured with a hot-plate  test. Briefly, mice were placed on a 52uC hot plate (IITC,  Woodland Hills, CA), enclosed by a Plexiglas cage. The withdrawal latencies of the hindpaw to noxious thermal stimulations were determined to an accuracy of 0.1 s. Each test session consisted of three trials, at 30-min intervals. The criteria of withdrawal included shaking, licking, or jumping on the hot plate. The cutoff limit was 25 s to avoid potential tissue damage. The mean latency was expressed as the threshold of individual mice to the noxious thermal stimulation. Mice were subjected to the hotplate test before RTX treatment (RTXd0), on RTX-treatment day 7 (RTXd7) and then weekly until the experimental endpoint of RTXd84.
For double-labeling studies, the tyramide signal amplification (TSA) technique was applied [20,21]. Briefly, sections were incubated with the first antiserum overnight at 4uC, a biotinylated-labeled secondary antibody for 1 h, and streptavidinhorseradish peroxidase (HRP) (1:200, PerkinElmer, Waltham, MA) for 30 min. Signals were amplified with the fluorescein tyramide reagent (1:100, PerkinElmer) for 3 min. After rinsing in Tris, sections were incubated with the second antiserum, followed by a Texas red-conjugated secondary antibody for 1 h (1:100, Jackson ImmunoResearch, West Grove, PA). Footpad sections were mounted on gelatin-coated slides for further analyses. Digital images of double-labeling sections were obtained at 400x magnification under Leica confocal microscope (Wetzlar, Germany).

Double-labeling Immunofluorescent Staining of DRG Neurons
For double-labeling of DRG neurons, DRG tissues were cryoprotected with 30% sucrose in PB overnight and cryosectioned with a cryostat (CM1850, Leica, Wetzlar, Germany) at 8 mm thickness. For adequate sampling, two ganglia (L4/L5) per mice and five,eight sections per DRG tissue (at 80-mm intervals) were immunostained. Briefly, sections were incubated with one of a mixture of primary antiserum:

Quantification of different Phenotypic DRG Neurons and IENFs
To quantify DRG neurons of different phenotypes, each DRG section was photographed at 200x under a fluorescence microscope (Axiophot microscope, Zeiss, Oberkochen, Germany) in a systematic fashion to produce a montage of the entire DRG section following established procedures [14,21]. To avoid a density bias, each section that only contained neuronal ganglia was measured with Image J vers. 1.44d software (National Institutes of Health (NIH), Bethesda, MD), and only neurons with a clear nuclear profile were counted. The colocalization ratios of different phenotypes DRG neurons were calculated respectively.
To quantify IENFs of different phenotypes, including PGP9.5(+), CGRP(+), SP(+), and TRPV1(+) IENFs were counted at 400x magnification (Axiophot microscope, Zeiss). The counting protocol followed established criteria in a coded fashion [22]. Fibers with branching points within the epidermis were counted as a single IENF. For fibers with branching points in the dermis, each fiber was counted as a single IENF. The length along the lower margin of the stratum corneum was defined as the epidermal length and determined with Image J software (NIH). IENF density was defined as the counted IENFs divided by the epidermal length (fibers/cm). In preliminary studies, PGP9.5(+) IENFs were 108.963.5 fibers/cm in the normal mice (n = 23). The values followed a Gaussian distribution (p = 0.27, Shapiro-Wilk normality test) and the variations among animals were minimal (3.21% of the mean).

Immunohistochemistry and Quantitation of CGRP(+) Nerve Terminals in the Spinal Cord
The lumbar cord sections of 50 mm in thickness were immunostained with anti-CGRP (1:2000) antisera as the described before. To ensure adequate sampling, every sixth section and a total of five sections for each animal were used. Sections were then mounted on slides for further quantitation [23]. Briefly, the areas of the superficial part of the dorsal horn (laminae I and II) were outlined under a dark-field microscope (Axiophot microscope, Zeiss). The areas of CGRP(+) nerves terminals on dorsal horn were quantified and normalized to the outlined areas of the superficial dorsal horn. The densities of CGRP(+) nerve terminals were expressed as mm 2 /mm 2 .

Intraplantar Administration of Botox after RTX-induced Neuropathy
This study investigated the role of CGRP in nociception using an intraplantar injection of botox (BOTOXH, Allergan, Irvine, CA) on RTXd56 when thermal hypoalgesia had been normalized. Briefly, botox was dissolved in sterilized saline and bilaterally administrated to the hindpaws (22.5 pg/paw in 10 ml) with a Hamilton microsyringe (Hamilton, Reno, NV). Thermal withdrawal latencies were measured by the hot-plate test as described above on post-botox day 1 (Bd1) through Bd14.

Evaluation of Motor Function with a Rotarod Retention Test
To evaluate whether botox caused motor dysfunction, the rotarod retention time was measured followed a previous protocol [24]. Briefly, mice were trained to balance on a rotarod treadmill (Med Associates, Georgia, VT) at 8 rpm for 120 s. The rotarod test was performed before and after botox at each time point following the hot-plate test up to Bd14. There were three trials of the rotarod test at 20-min intervals for each mouse, and the mean of the retention times for the three trials was used for further analysis.

Assessment of Neuromuscular Junction (NMJ) Innervation
Innervation of NMJ in the plantar muscle was evaluated with a combination of cholinesterase histochemistry and PGP9.5 immunostaining following our previously established protocols [25]. Briefly, the plantar muscles were carefully dissected and cryoprotected with 30% sucrose overnight. Series sections (30 mm) were cut on a cryostat (CM 1850, Leica). Every fifth section was stained with cholinesterase followed by PGP9.5-immunohistochemistry. Sections were incubated in a 5-bromoindoxyl acetate solution for 15 min to demonstrate the cholinesterase in NMJ followed by PGP9.5-immunohistochemistry. Coded sections from the botox and saline groups were examined, and 20 fields were randomly selected under microscopy at 200x magnification. The NMJ innervation ratios were calculated according to the number of innervated NMJ divided by the number of total NMJ on all sections.

Pharmacological Intervention with CGRP 8-37 by an Intraplantar Injection
Pharmacological experiments were performed with a single dose of the CGRP receptor antagonist, CGRP 8-37 (Sigma), at various concentrations of 0.13, 1.3, and 13 nmole/paw in a volume of 10 ml on RTXd56. Drugs were dissolved in normal saline and administrated to the bilateral hindpaws with a Hamilton micro-syringe (Hamilton). The withdrawal latency was measured according to the hot-plate test as described above. The other group of RTX mice received normal saline as the control for comparison. Changes in withdrawal latencies were assessed at 0.5, 1, 2, 4, 6, 8, and 24 h after CGRP  antagonism.

Evaluation of CGRP Contents in the Skin
To evaluate the contents of CGRP after botox, enzyme immunoassay (EIA) of CGRP was performed on the skin. Briefly, the plantar skin of the hindpaw was incubated in ethylenediaminetetraacetic acid (EDTA) solution at 37uC for 30 min. The epidermis was separated from the dermis. The epidermis was eluted in normal saline with ultrasonic bath for 30 min. The contents of CGRP were determined with the CGRP EIA kit (SPI-BIO, Massy Cedex, France) following the manufacturer's instructions. The protein contents of epidermis were determined with the protein assay kit (Bio-Rad laboratories, Hercules, CA). The CGRP content was normalized to the protein content (pg/mg).

Experimental Designs and Statistic Analysis
In this study, there were two experimental designs: RTXinduced neuropathy and a CGRP functional intervention, including botox and CGRP  antagonist treatments for RTXinduced neuropathy. In the neuropathy model, there were two groups: RTX and vehicle groups. For the CGRP functional studies, there were two experiment designs: (1) the effect of botox (the botox group) and (2) the effect of CGRP antagonism (the CGRP 8-37 group) on RTXd56. In both experimental designs, a separate group of mice received saline as a control (the saline group). Coding information was masked during the behavioral tests and quantification procedures. There were 120 mice in total with 5,7 mice in each group at different time points for the behavioral functional evaluation and morphology examinations. All data are expressed as the mean 6 standard derivation of the mean, and t-test was performed for data with a Gaussian distribution. For data which did not follow a Gaussian distribution, a nonparametric Mann-Whitney test was conducted. p,0.05 was considered statistically significant. For comparison of hot-plate latencies, two-way repeated measures ANOVAs were performed followed by Bonferroni's post-hoc test when p,0.05 was obtained.

Thermal Hypoalgesia with RTX-induced Neuropathy
Typical thermal hypoalgesia developed in the RTX group in comparison with the vehicle group (p,0.0001, two-way repeated measures ANOVA test). Before RTX treatment, the hot-plate latencies were similar between the RTX and vehicle groups  Thermal hypoalgesia existed from RTXd7 to RTXd49 (15.565.3 s, p,0.01) and hot-plate withdrawal latencies had become normalized on RTXd56 (12.363.0 s, p.0.05) through RTXd84, the end point of the study period (11.363.5 s, p.0.05) (Fig. 1).

Functional Effects of Botox on Thermal Sensations
For a functional intervention of CGRP(+) IENFs, we delivered botox by an intraplantar injection on RTXd56 of RTX-induced neuropathy after thermal sensations had become normalized. Botox reduced CGRP(+) IENFs, and thermal hypoalgesia reappeared. Under the dose of botox we used (22.5 pg/paw), there was no gross change in gait or extension response of the hind limbs compared to the saline group (Fig. 9A,C) Fig. 9D). Consistent with normal motor function, rotarod retention times were similar between the botox and saline groups at each time point (p = 1.00, Fig. 9E).

Effects of Botox on Cutaneous Innervation of CGRP
To investigate the influence of botox on skin innervation, we assessed PGP9.5(+) and CGRP(+) IENFs. The abundance of PGP9.5(+) IENFs did not change on Bd3 (Fig. 10A). In contrast, CGRP(+) IENFs were reduced on Bd3 compared to the saline group, and there was a return of CGRP(+) IENFs on Bd14 compared to Bd3 (Fig. 10B,D). Quantitatively, densities of CGRP(+) IENFs were markedly reduced on Bd3 (6.162.7 vs. 38.168.7 fibers/cm, p,0.0001), but they had recovered by Bd14 (18.1610.8 fibers/cm, p = 0.023); however, the value was still lower than that of the saline group (p = 0.003). Densities of PGP9.5(+) IENFs were similar on Bd3 (61.3617.3 fibers/cm, p = 0.19) and Bd14 (66.4621.8 fibers/cm, p = 0.86), which were comparable to the saline group (72.1612.4 fibers/cm) (Fig. 10E). In the botox experiment, densities of CGRP(+) IENFs paralleled hot-plate latencies (r = 20.66, p = 0.0003, Fig. 10F). We then measured the changes in CGRP contents of the skin which paralleled the CGRP expression in the skin nerve terminals. On RTXd7, CGRP contents in the RTX group were markedly reduced compared to the vehicle group (16.767.8 vs. 63.1618.7 pg/mg, p,0.01). Botox had similar effect on the reduction of CGRP content on Bd3 (16.367.8 pg/mg, p = 0.01) Figure 9. Effect of botulinum toxin type A (botox) on withdrawal latencies on the hot-plate test and motor performance in resiniferatoxin (RTX)-induced neuropathy. Botox was administered through intraplantar injection on RTX treatment day 56 (RTXd56). (A,C) Graphs show the extension ability of hindlimbs before botox treatment (A, Pre-botox), in the group of after botox treatment (B, botox) and in the saline-treated group (C, saline). There was no difference in gross appearance of hindlimb extension among the 3 groups. (D) The graph shows the effect of botox on the withdrawal latency on the hot-plate test at 52uC in the saline group (open circles, n = 5) and botox group (filled circles, n = 5). On RTXd56, the withdrawal latency was normalized (arrow). Thermal hypoalgesia reappeared on post-botox day 1 (Bd1, p,0.01) and persisted until Bd5 (p,0.05). Hot-plate latencies between botox and saline groups were analyzed with two-way repeated measures ANOVA followed by Bonferroni's post-hoc test. (E) There was no difference in retention times between the saline group (open circles, n = 5) and botox group (filled circles, n = 5) according to rotarod performance at a speed of 8 rpm (p.0.05). *p,0.05, **p,0.01, ***p,0.001. doi:10.1371/journal.pone.0050805.g009 and CGRP contents recovered on Bd14 (77.6623.4 pg/mg, p.0.05) (Fig. 10G).

Pharmacological Intervention with CGRP 8-37 Antagonism
To investigate the CGRP-mediated transmission of nociceptive stimuli, we produced a pharmacological blockade with the CGRP receptor antagonist, CGRP  , by intraplantar administration on RTXd56. RAMP-1 showed punctate profiles, and the immunor-eactivities were co-expressed with CGRP immunoreactivities in cutaneous nerves of the footpad skin (Fig. 12A, B).

Contribution of Co-localization with TRPV1 to different Patterns of Denervation and Reinnervation of SP(+) and CGRP(+) IENFs
The current study demonstrated that different degrees of the reinnervation of SP(+) and CGRP(+) IENFs were attributable to different co-localization ratios with TRPV1. The permanent depletion of SP(+) neurons and their cutaneous terminals resulted from a high co-localization ratio with TRPV1(+) neurons, which were depleted by RTX [9,10]. In addition, the parallel depletion of SP(+) and TRPV1(+) neurons and their cutaneous terminals in the late stage (RTXd84) implied the possibility of permanent loss of TRPV1(+) neurons after RTX treatment. In contrast, CGRP(+) neurons and IENFs were recovered at RTXd84. What is the possible mechanism of CGRP recovery? CGRP(+) neurons had a lower level of co-expression with TRPV1(+) neurons than SP(+) neurons. Moreover, there were limited SP(+)/CGRP(+) neurons and TRPV1(+)/CGRP(+) IENFs. Those observations suggests CGRP(+) neurons account for the relative resistance of CGRP(+) neurons to RTX treatment than SP(+) neurons [26] and provide the possibility of reinnervation by CGRP(+) IENFs after RTXinduced neuropathy [27], i.e., the reinnervation in the late phase may result from the remaining CGRP(+) neurons with their CGRP(+) IENFs. The ratios of TRPV1(+)/CGRP(+) and TRPV1(+)/SP(+) neurons corresponded to the differential depletion of peptidergic DRG neurons, and such differences accounted for the basis of different extents of depletion of SP(+) and CGRP(+) IENFs.

Distinct Functions of Peptidergic IENFs on Thermal Responses in Acute and Chronic Phases of RTX-induced Neuropathy
This study documented that CGRP(+) IENFs were sufficient for recovery of thermal sensation when TRPV1(+) and SP(+) IENFs were depleted. With RTX-induced neuropathy, thermal hypoalgesia occurred at RTXd7, corresponding to depletion of TRPV1(+) and SP(+) DRG neurons and their peripheral terminals in the skin. RTX-induced thermal hypoalgesia lasted for ,7 weeks and became normalized at RTXd56 through RTXd84. During this period of thermal hypoalgesia, SP(+) and TRPV1(+) IENFs were depleted. The current study indicates that the reinnervation CGRP(+) IENFs in the late phase provided a compensatory mechanism for the transmission of thermonociceptive stimuli when SP(+) and TRPV1(+) IENFs were depleted. The role of thermal transmission by CGRP(+) IENFs was further confirmed by (1) botox-induced thermal hypoalgesia with a parallel decrease in CGRP(+) IENFs and CGRP contents in the skin, (2) blocking of CGRP transmission by CGRP  antagonism.
TRPV1 and SP are known molecular mediators of thermal sensations [8,28,29] and are responsible for the induction of thermal hypoalgesia in RTX-induced neuropathy [11,14]. TRP channels of various classes mediate transduction of temperature stimuli [30][31][32]. In particular, TRPV1 and TRPV2 channels are responsible for nociceptive heat [33] and have different distributions [34,35]. In addition to TRPV1 and TRPV2, other types of TRP channels or polymodal C-nociceptors may be responsible for sensing heat when TRPV1 and TRPV2 are depleted [36,37]. In the current study, TRPV2 was mainly expressed in medium-sized DRG neurons and rarely colocalized with CGRP. Moreover, there was no change in the densities of TRPV2(+) neurons and the ratios of TRPV2/CGRP during RTX-induced neuropathy, indicating TRPV2 are not responsible for the thermal compensation in RTX-induced neuropathy. In contrast, this study demonstrated the recovery of thermal sensation in the absence of TRPV1(+) neurons and their cutaneous terminals and provides new evidence and mechanisms for the compensatory recovery of thermal sensation by CGRP(+) IENFs. This recovery was further reversed by an intraplantar injection of botox, which depleted CGRP(+) IENFs and again resulted in thermal hypoalgesia.
In cutaneous nerve terminals, CGRP was co-localized with RAMP1, a chaperone component required for a functional CGRP receptor [38]. RAMP1 is a small single transmembrane protein but essential for binding of CGRP pharmacologically [39] and CGRP effects can be blocked by RAMP1 antagonism [40]. In this report, the colocalization of RAMP1 and CGRP on cutaneous nerves provided structural evidence of CGRP(+) IENFs for thermosensation. The effect of CGRP transmission on thermal sensations was further confirmed by CGRP antagonism with CGRP  . The antagonistic effects of CGRP 8-37 were dosedependent and transiently evidenced by temporal patterns of withdrawal latencies.

Effects of Botox on CGRP Transmission in Cutaneous Nerve Terminals and Clinical Implications
This report documents the influences of botox on CGRPmediated thermonociceptive transmission, which may provide mechanisms underlying botox-mediated pain-relieving effects. Two issues merit discussion: (1) whether botox completely damages IENFs structurally and (2) whether motor functions are impaired by an injection of botox. The reduction in CGRP(+) IENFs could be attributed to degeneration of cutaneous nerve terminals [14] or a change in the phenotypes of IENFs. Botox induced the transient reduction of CGRP(+) IENFs and PGP9.5(+) IENFs remained the same. By our observations of colocalization of CGRP(+)/PGP9.5(+) IENFs, those resulted indicating that IENFs remained structurally intact after the use of botox, but reduction CGRP after botox. Thermal hypoalgesia induced by botox was due to reduced thermonociceptive transmission, because this study documented the intactness of the motor system after the botox injection: (1) a normal gait and reflex response of the hind limbs, (2) same retention time on the rotarod test as the saline group, and (3) intact neuromuscular innervation.
Taken together, this study suggests that botox functionally changed CGRP(+) IENFs in the skin. Both downregulation and reduced release could contribute to the reduced CGRP contents in the skin. These findings provide a foundation of new therapeutic strategy for pain control by modulating GGRP releases [41,42] and CGRP receptor at the nerve terminal of the skin. For example, transcutaneous botox administration is a easy and efficacy approach for modulating CGRP-mediated thermal nociception [43] and by developing specific antagonists of CGRP receptors without affecting cutaneous CGRP expression [40].
In summary, RTX, an ultrapotent ligand for TRPV1, permanently depleted TRPV1(+) and SP(+) neurons and their corresponding peripheral terminals in the skin, which resulted in thermal hypoalgesia. In this report, we further demonstrated that RTX had different degrees of influence on peptidergic epidermal nerves, mainly by depleting SP(+) IENFs. In contrast, CGRP(+) IENFs were more tolerant of the neurotoxicity of RTX and were responsible for normalization of the thermal responses in the late phase.