CGRPα-Expressing Sensory Neurons Respond to Stimuli that Evoke Sensations of Pain and Itch

Calcitonin gene-related peptide (CGRPα, encoded by Calca) is a classic marker of nociceptive dorsal root ganglia (DRG) neurons. Despite years of research, it is unclear what stimuli these neurons detect in vitro or in vivo. To facilitate functional studies of these neurons, we genetically targeted an axonal tracer (farnesylated enhanced green fluorescent protein; GFP) and a LoxP-stopped cell ablation construct (human diphtheria toxin receptor; DTR) to the Calca locus. In culture, 10–50% (depending on ligand) of all CGRPα-GFP-positive (+) neurons responded to capsaicin, mustard oil, menthol, acidic pH, ATP, and pruritogens (histamine and chloroquine), suggesting a role for peptidergic neurons in detecting noxious stimuli and itch. In contrast, few (2.2±1.3%) CGRPα-GFP+ neurons responded to the TRPM8-selective cooling agent icilin. In adult mice, CGRPα-GFP+ cell bodies were located in the DRG, spinal cord (motor neurons and dorsal horn neurons), brain and thyroid—reproducibly marking all cell types known to express Calca. Half of all CGRPα-GFP+ DRG neurons expressed TRPV1, ∼25% expressed neurofilament-200, <10% contained nonpeptidergic markers (IB4 and Prostatic acid phosphatase) and almost none (<1%) expressed TRPM8. CGRPα-GFP+ neurons innervated the dorsal spinal cord and innervated cutaneous and visceral tissues. This included nerve endings in the epidermis and on guard hairs. Our study provides direct evidence that CGRPα+ DRG neurons respond to agonists that evoke pain and itch and constitute a sensory circuit that is largely distinct from nonpeptidergic circuits and TRPM8+/cool temperature circuits. In future studies, it should be possible to conditionally ablate CGRPα-expressing neurons to evaluate sensory and non-sensory functions for these neurons.


Introduction
Small-to-medium-diameter neurons in the dorsal root ganglia (DRG) have classically been divided into peptidergic and nonpeptidergic subsets [1,2]. Many of these neurons respond to noxious thermal, mechanical and chemical stimuli, making them nociceptive, whereas others respond to innocuous stimuli, such as warming and cooling. The most widely recognized markers of peptidergic neurons are CGRP and substance P, while IB4binding and fluoride-resistant acid phosphatase (FRAP; also known as Prostatic acid phosphatase, PAP) classically mark nonpeptidergic neurons [3,4].
The sensory functions of these circuits were recently examined through the use of sophisticated genetic and physiological techniques. Nonpeptidergic, Mrgprd-expressing neurons are unmyelinated and contribute to mechanosensation but not thermosensation or cold sensation [5,6]. Peptidergic CGRP + neurons are myelinated (A-fibers) or unmyelinated (C-fibers) and, depending on fiber type, respond to nociceptive stimuli or guard hair displacement [7,8]. TRPV1 + neurons, a subset of which are peptidergic [9], detect noxious thermal stimuli and some pruritogens [5,10,11,12,13]. However, the extent to which the broader class of peptidergic CGRP + neurons is required for innocuous and noxious stimulus detection in mammals is currently unknown.
CGRP is not a single peptide but two separate peptides (CGRPa and CGRPb) encoded by separate genes (Calca and Calcb). Calca is alternatively spliced, giving rise to CGRPa in neurons and calcitonin in thyroid C cells [14]. And, CGRPa and CGRPb are nearly identical at the amino acid level. As a result, antibodies typically cannot distinguish CGRPa from CGRPb, necessitating use of the term ''CGRP-immunoreactivity'' (CGRP-IR). CGRP-IR cells and fibers are present in multiple tissues, including the brain, stomach, intestine, skin and bladder [15,16,17,18]. In studies where expression of each gene was resolved, both CGRPa and CGRPb were expressed in the DRG although CGRPa was expressed at two-fold higher levels [16,17].
When released peripherally from neurons, CGRPa causes vasodilatation, relaxes smooth muscle cells and contributes to migraine pathogenesis [19]. CGRPa is also released in the dorsal spinal cord and potentiates excitation caused by noxious stimuli and pronociceptive chemicals [20,21]. CGRPa levels also regulate sensitivity to noxious heat [22]. Notably, CGRPa knockout mice have reduced behavioral responses to capsaicin and impaired heat hyperalgesia although acute heat responsiveness is not affected [23,24,25].
To directly study the projections and sensory functions of CGRPa neurons, we generated a knock-in mouse that expresses an axonal tracer and a conditional cell ablation construct from the Calca/Cgrpa locus. We used these mice to prospectively identify peptidergic DRG neurons in culture and show that they respond to agonists that evoke sensations of pain and itch.

Results
CGRPa-GFP + neurons respond to agonists that evoke sensations of pain and itch At the time we began this study, there were no ways to prospectively identify CGRP + sensory neurons for physiological studies. To permit direct visualization of CGRP + sensory neurons and axons, we knocked-in a floxed (LoxP flanked) membranetethered axonal tracer (farnesylated enhanced GFP) to the Calca locus ( Fig. 1A) [26]. This floxed GFP also conditionally blocks expression of downstream DTR (Cre recombinase-dependent expression of DTR will be described in a subsequent study). Heterozygous (CGRPa-GFP +/2 ) mice, which contain one functional Calca allele and one GFP allele, were used throughout this study. The mice were viable and showed no obvious phenotypic or behavioral abnormalities.
We next loaded cultured DRG neurons from CGRPa-GFP +/2 mice with the calcium indicator Fura2-AM. CGRPa-GFP + neurons were readily identifiable based on intrinsic GFP fluorescence and accounted for 39.9% of all Fura2-loaded neurons (Fig. 1B, arrowheads; n = 1292 Fura2 + neurons analyzed). No CGRPa-GFP + cells were present from wild-type littermate controls. A majority (54.8%) of these CGRPa-GFP + neurons were 17-30 mm in diameter, with the remainder being either smaller or larger.
CGRPa-GFP genetically marks a circuit that is largely distinct from nonpeptidergic and TRPM8 + sensory circuits To determine if CGRPa-GFP was expressed in peptidergic sensory neurons, we next immunostained sections of lumbar DRG with antibodies to GFP and various neuronal markers. We found that the vast majority (88.960.5%) of all CGRP-IR neurons were CGRPa-GFP + ( Fig. 2A-C, Table 2). Conversely, 67.860.8% of all CGRPa-GFP + neurons were CGRP-IR. This lack of complete overlap was likely due to the greater sensitivity of GFP immunostaining-GFP filled cells in their entirety and was easier to detect than CGRP-IR, especially in cells with low levels of CGRP-IR. Interestingly, ,10% of the CGRP-IR neurons did not colocalize with CGRPa-GFP. Because the CGRP antibody we used recognizes CGRPa and CGRPb, these CGRP-IR-only cells could represent DRG neurons that express CGRPb alone [16,17]. In addition, approximately 50% of the CGRPa-GFP + neurons expressed TRPV1 (Fig. 2D-F, Table 2), consistent with our functional studies above.
In the spinal cord, CGRPa-GFP (Fig. 3A) and CGRP-IR ( Fig. 3B) were colocalized in lamina I and lamina II outer, with fibers extending into lamina V and towards lamina X. There was little overlap between CGRPa-GFP + and IB4-binding terminals in lamina II (Fig. 3C, D), revealing segregation between peptidergic and nonpeptidergic spinal circuitry. When taken together, our data indicate that CGRPa-GFP genetically marks a distinct subset of small-to-medium-and large-diameter DRG neurons in adult mice and constitutes a circuit that is largely distinct from nonpeptidergic circuits and TRPM8 + /cool temperature-sensing circuits.
CGRPa-GFP is expressed in motor neurons and a small population of neurons intrinsic to the dorsal spinal cord CGRP-IR in the dorsal horn is typically attributed to primary afferent axons and their terminals; however CGRP-IR is also present in a subset of dorsal horn neurons in rats and mice [35,36]. To detect these cells immunohistochemically, these groups performed dorsal rhizotomies or treated animals with colchicine (colchicine arrests axonal transport, allowing CGRP to accumulate). Scattered Cgrpa/Calca-expressing cells were also detected in the dorsal horn by in situ hybridization, in Allan Brain Atlas adult spinal cord images [37]. The high sensitivity of the membranetethered GFP axonal tracer allowed us to detect these intrinsic CGRPa + neurons without manipulating mice surgically or chemically. When examined at higher magnification, these spinal neurons were located between axon terminals of CGRPa-GFP + and IB4 + sensory neurons, with CGRPa-GFP + neurons being predominantly located in lamina II inner and lamina III ( Fig. 4A-F, arrowheads). Very few of these intrinsic CGRPa-GFP + dorsal horn neurons contained PKCc (Fig. 4C,F), a marker of some lamina II and III neurons [38,39]. In the ventral horn, CGRPa-GFP labeled many CGRP-IR motor neurons ( Fig. 4G-I) along with their axons, which terminate at motor endplates in skeletal muscle ( Fig. 5A-C). There were also a number of CGRP-IR motor neurons that lacked CGRPa-GFP, likely reflecting a subset of motor neurons that only express CGRPb [40].
Most of these CGRPa-GFP + endings had a straight and stubby morphology that was distinct from the meandering ''zig-zag'' shape of PGP9.5 + /CGRPa-GFP 2 (presumably nonpeptidergic)  fibers. We previously observed this same morphological distinction between peptidergic and nonpeptidergic fibers when targeting farnesylated GFP to Mrgprd + /nonpeptidergic neurons [26]. Interestingly, we also noticed that some of the epidermal CGRPa-GFP + fibers had small spheres at their tips (see arrowheads, Fig. 5Ginset). These spheres may simply result from membrane budding or intriguingly might constitute a novel transduction unit at the tips of some peptidergic afferents. CGRPa-GFP + afferents were also present within sweat glands of glabrous skin (Fig. 5G-I). These afferents, which were also PGP9.5 + , are likely sensory in origin because CGRPa is not expressed in sympathetic ganglia of mice [17]. In hairy skin, CGRPa-GFP + fibers progressed through the dermis and terminated in the epidermis and on guard hair follicles ( Fig. 5J-L).
In addition, CGRPa-GFP + fibers were present in the submucosal/smooth muscle layers of the small intestine ( Fig. 7A-C), consistent with previous studies [44,45]. There were also numerous green fluorescent cells in intestinal villi; however, these cells were not CGRPa-GFP + because: a) they did not co-stain for CGRP-IR (Fig. 7B) and more importantly, b) they were detectable in wild-type mice (i.e., mice lacking GFP; Fig. 7C-inset). These cells are likely a population of autofluorescent stromal cells [46]. There were also a large number of CGRP-IR cells in the intestinal villi that were not CGRPa-GFP + (Fig. 7B). These CGRP-IR-only cells likely express CGRPb, particularly given that CGRPb/Calcb is the primary CGRP gene expressed in the gut [16,17]. Lastly, we observed CGRPa-GFP + afferents in the bladder (Fig. 7D-I), a visceral tissue that is innervated by sensory afferents. When taken together, our data indicate that CGRPa-GFP + neurons innervate diverse cutaneous and visceral structures.
CGRPa-GFP labels other cell types that express Calca, including thyroid cells and neurons in the brain Since GFP was targeted to exon 2 of Calca, an exon that is common to CGRPa and calcitonin [14], CGRPa-GFP should be   present in all tissues where Calca is expressed. Indeed, we found that CGRPa-GFP was co-localized with CGRP-IR in parafollicular cells of the thyroid (Fig. 7J-L). We next thoroughly mapped CGRPa-GFP expression in the brain. To do this, we immunostained adult mouse brain sections and noted all locations where CGRPa-GFP + cell bodies were found (Table 3). With the exception of the abducens nucleus, Purkinje cells, cuneiform nucleus and the dorsomedial thalamic nucleus, we detected CGRPa-GFP + cell bodies in all regions previously known to express CGRPa [47,48,49,50]. Representative regions where cellular and/or fiber staining were observed include the spinal trigeminal nucleus caudalis (Fig. 8A), the parabrachial nucleus (Fig. 8B), the peripeduncular and posterior intralaminar thalamic nuclei (Fig. 8C), the subparafascicular nucleus of the thalamus (Fig. 8D), the nucleus accumbens (Fig. 8E), the subiculum (Fig. 8F) and weakly in the visual cortex (Fig. 8F, inset). Calca-GFP BAC transgenic mice produced by the GENSAT project show a similar distribution of cellular and axonal labeling in the brain [51]. Taken together, our data indicate that CGRPa-GFP knock-in mice reproducibly mark all cells and tissues that are known to express Calca.

Discussion
We generated the first knock-in reporter mouse to directly visualize and functionally study CGRPa-containing sensory neurons. While characterizing these mice, we found that CGRPa-GFP faithfully marked the peptidergic subset of DRG neurons, as well as other cell types throughout the body that express Calca. In contrast, cells that express Calcb/CGRPb, including intramural neurons of the intestine [17], were devoid of CGRPa-GFP immunoreactivity. Our reporter mice can thus be used to discriminate Calca-expressing cells from cells that express Calcb. The membrane-tethered GFP reporter allowed us to prospectively identify live CGRPa-expressing neurons in culture for functional studies. Remarkably, half (,50%) of all CGRPa-GFP + DRG neurons expressed TRPV1 and half of all CGRPa-GFP + DRG neurons responded to the TRPV1 agonist capsaicin, suggesting that CGRPa + neurons may play a significant role in capsaicin and noxious thermal sensitivity in vivo. In addition, .50% of all histamine-and chloroquine-responsive neurons were CGRPa-GFP + , suggesting a major role of CGRPa-expressing neurons in histamine-dependent and histamine-independent itch. Likewise, there is a large degree (,90%) of overlap between TRPV1/capsaicin-responsive neurons and histamine-responsive neurons [31,52], suggesting thermal pain and histamine-dependent itch are encoded by the same class of sensory neurons.
In contrast, Takashima et al. found that TRPM8-GFP and CGRP-IR overlap by ,20% when using a BAC transgene to mark Trpm8-expressing neurons [54]. BAC reporters often drive higher levels of gene expression when compared to knock-in reporters, but can suffer from position effects that compromise expression specificity [51]. Thus, higher detection sensitivity and/ or position effects could explain why there was a greater degree of overlap between CGRP-IR and BAC reporter driven Trpm8 expression than we and others observed when examining endogenous Trpm8 expression.
We also found that 14.365.0% of all CGRPa-GFP + cells were menthol-responsive. Contrary to what is commonly stated in the literature, menthol is not a TRPM8 specific agonist. Menthol activates TRPA1 at sub-to low-micromolar concentrations and inhibits TRPA1 at higher concentrations [28,29]. This bimodal modulation provides one of many explanations for why a smaller percentage of CGRPa-GFP + neurons responded to menthol in culture than to the TRPA1 agonist mustard oil (Table 1).
With regard to position effects, it will be interesting to determine if the Calca-GFP BAC transgenic mouse line made by the GENSAT project reproduces CGRPa expression in DRG, brain and peripheral tissues to the same extent as our knock-in mouse [51]. In addition, it will be interesting to determine if this BAC transgenic line distinguishes Calca-expressing cells from Calcbexpressing cells. Calca and Calcb are located ,80 kb apart in the mouse genome. This genomic proximity could contribute to their similar but not identical expression patterns. Baillie and colleagues recently used Calca-GFP BAC transgenic mice and optical imaging techniques to visualize an axon reflex in an individual CGRPa + sensory afferent [55].
In what is perhaps the most comprehensive physiological study of CGRP + sensory neurons to date, Lawson and colleagues found that CGRP-IR neurons can be classified as C-fiber and Ad-fiber nociceptive units (responsive to noxious thermal and high  threshold mechanical stimuli), unresponsive C-fibers or Aa/b guard hair afferents. None of the CGRP-IR neurons had Ccooling/cold or C-low threshold mechanoreceptive (C-LTMR) properties. These findings, combined with TRPV1 cell inactivation studies (described above) and our current work, consistently point to a role for CGRP + neurons in sensing noxious heat.
CGRPa-GFP might also mark the CGRP-IR + Aa/b guard hair units that were identified by Lawson and colleagues [7], particularly since CGRPa-GFP + fibers terminated on guard hairs in hairy skin and ,25% of all CGRPa-GFP + neurons expressed NF200, a marker of myelinated afferents. Guard hairs add sheen to the coat of furry mammals, are often water repellent, and drive activity in sensory afferents when deflected [7,56,57]. Whether activation of guard hair afferents has sensory and/or non-sensory functions in mammals is currently unknown. Ultimately, it should be possible to directly evaluate the in vivo functions of CGRPa + sensory neurons by taking advantage of the LoxP-stopped DTR that we knocked-in immediately behind GFP (Fig. 1A). DTR, when combined with injections of diphtheria toxin, can be used to conditionally ablate cells and neurons in adult mice [5,58]. Importantly, DTR expression was completely blocked in DRG (Table 2). We engineered DTR so that its ATG start codon will precisely substitute for the start codon of GFP upon CRE recombinase-mediated excision. DTR should thus be expressed in all cell types that jointly express CGRPa and CRE recombinase (including cells that expressed CRE at any time during development). When crossed with sensory neuron selective lines, such as Nav1.8-Cre or Advillin-Cre [59,60,61,62], this could permit selective expression of DTR in DRG neurons while maintaining GFP expression in all other Calca-expressing cell types. Given that Calca is expressed in many other cell types, this strategy could be broadly employed to genetically label, ablate and study the function of diverse peptidergic CGRPa-containing cell types throughout the brain and body.

Materials and Methods
All procedures and behavioral techniques involving vertebrate animals were approved by the Animal Care and Use Committee at the University of North Carolina at Chapel Hill.

Molecular Biology
Recombineering was used to generate Calca targeting arms from a C57BL/6-derived bacterial artificial chromosome (BAC; RP24-136021). The start codon located in exon 2 is common to CGRPa and calcitonin and was replaced with an AscI site to facilitate cloning of an axonal tracer and a conditional cell ablation construct: AscI-LoxP-EGFPf-3x pA-LoxP-DTR-pA-Frt-PGK-NeoR-Frt-AscI. EGFPf = farnesylated enhanced GFP [26]. DTR = human diphtheria toxin receptor [58]. NeoR = neomycin resistance. The LoxP sites were oriented so that the first ATG encountered was in GFP or, after Cre recombinase-mediated excision, DTR. Correct targeting was confirmed in 5.8% of all embryonic stem cell clones by Southern blotting using flanking 59 and 39 probes and a NeoR internal probe. High percentage chimeras were crossed to C57BL/6 females to establish germline transmission and then crossed to ACTFLPe mice (B6.Cg-Tg(ACTFLPe)9205Dym/J, Jackson Laboratory) to remove the Frt-flanked selection cassette (removal confirmed by PCR). Next, mice were backcrossed to C57BL/6 to remove the ACTFLPe allele (removal confirmed by PCR) and then backcrossed to C57BL/6 mice for 8 generations to establish the CGRPa-GFP knock-in line. As a technical note, we were only able to detect GFP expression in DRG neurons after removal of the PGK-NeoR selection cassette.

Calcium Imaging
Adult (4-6 week old) male CGRPa-GFP +/2 mice were decapitated, DRG were dissected then neurons were dissociated using collagenase (1 mg/mL; Worthington, CLS1) and dispase (5 mg/mL; Gibco, 17105-041) in DH10 media (1:1 Ham's DMEM/F12, 10% FBS and 1% penicillin/streptomycin) [63,64]. Medium was supplemented with 25 ng/mL of glialderived neurotrophic factor (GDNF; Upstate, GF030). The neurons were plated onto coverslips coated with 0.1 mg/mL poly-D-lysine and 5 mg/mL laminin. After 24 h, neurons were washed 26 with Hank's balanced salt solution (HBSS) and incubated for 1 h with 2 mM Fura2-AM in the dark at room temperature. Next, the cells were washed 36 with HBSS and maintained at room temperature for 30 min prior to imaging. After a 60 s baseline, agonists (1 mM capsaicin, 100 mM mustard oil, 200 mM menthol, 4 mM icilin, 100 mM ATP, 100 mM histamine, 1 mM chloroquine or acidic pH 5-6 HBSS) were perfused onto the neurons. Following activation, cells were perfused with HBSS to remove the agonist, which was followed by addition of 100 mM KCl to determine the total number of neurons present. Images were acquired on a Nikon Eclipse Ti

Histology
Mice were sacrificed by overdosing with pentobarbital. The thyroid, brain, bladder, hindpaw skin, lumbar DRG, lumbar spinal cord and small intestine were dissected and immersion-fixed in 4% paraformaldehyde (5 h, 24 h, 5 h, 3 h, 4 h, 8 h and 2 h, respectively) and were cryopreserved in 30% sucrose at 4uC. Tissue was embedded in TissueTek and cryosectioned (20 mm for small intestine and DRG; 40 mm for thyroid, bladder and spinal cord; 50 mm for brain and skin). Sections were either immunostained free-floating or thaw mounted onto SuperFrost Plus slides and stored at 220uC until needed.
For diaminobenzidine (DAB) staining, brain sections were processed as described above with chicken anti-GFP (1:5,000). On day 2, the sections were washed in TBST and then blocked for 30 min. Sections were incubated for 2 h in biotinylated donkey anti-chicken IgG (1:500), which was followed by washes in TBST. Sections were incubated with the Vectastain ABC complex in TBST for 2 h and washed. Sections were treated with a DAB solution (0.02% DAB, 0.01% H 2 O 2 and 0.005% nickel ammonium sulfate in TBST) for 15 min. Following TBST washes and a PBS rinse, the sections were immersed in 0.2% gelatin in water, mounted onto Superfrost Plus slides and then air-dried for 4 days. Lastly, the sections were dehydrated with graded ethanols, cleared with xylene and coverslipped with DPX.

Supporting Information
Table S1 Percentage of CGRPa-GFP+/2 DRG neurons of a given size class (small, medium, large diameter) that respond to the indicated agonists.