Itch suppression in mice and dogs by modulation of spinal α2 and α3GABAA receptors

Chronic itch is a highly debilitating condition affecting about 10% of the general population. The relay of itch signals is under tight control by inhibitory circuits of the spinal dorsal horn, which may offer a hitherto unexploited therapeutic opportunity. Here, we found that specific pharmacological targeting of inhibitory α2 and α3GABAA receptors reduces acute histaminergic and non-histaminergic itch in mice. Systemic treatment with an α2/α3GABAA receptor selective modulator alleviates also chronic itch in a mouse model of atopic dermatitis and in dogs sensitized to house dust mites, without inducing sedation, motor dysfunction, or loss of antipruritic activity after prolonged treatment. Transsynaptic circuit tracing, immunofluorescence, and electrophysiological experiments identify spinal α2 and α3GABAA receptors as likely molecular targets underlying the antipruritic effect. Our results indicate that drugs targeting α2 and α3GABAA receptors are well-suited to alleviate itch, including non-histaminergic chronic itch for which currently no approved treatment exists. Chronic itch affects about 10% of the general population, however current treatments are largely ineffective. Here, the authors show that targeting of inhibitory α2 and α3GABAA receptors reduces itch in mice and in a canine model, suggesting this a potentially useful therapeutic approach.

C hronic itch affects between 4-17% of the general population 1,2 . Most drugs currently used to treat itch are histamine H1 and H4 receptor blockers that work well against acute itch. By contrast, chronic itch is mostly histamineindependent and largely irresponsive to these medications 3 . Frequent causes of histamine-independent itch include, besides atopic dermatitis, cholestatic liver disease, end stage kidney failure, and opioid-therapy 4 . Drugs used to treat itch in these conditions include immune suppressants and drugs acting at the CNS such as gabapentinoids, antidepressants, and opioid receptor antagonists. In the majority of cases, these treatments do not provide adequate relief or cause severe side effects 5 .
Pruritic (itch) stimuli are detected by sensory neurons (primary pruritoceptors) that innervate the skin and transform these stimuli into electrical signals, i.e., action potentials. These action potentials are then relayed via the peripheral and central axons of primary pruritoceptors to central neurons in the spinal or medullary dorsal horn 6 . Only recently have researchers begun to understand the signaling molecules, receptors, transmitters and neuronal pathways of itch. Several "new" G protein-coupled receptors expressed by primary sensory neurons have been identified that are activated by pruritogens. One such receptor is the mas-related G protein coupled receptor A3 (MrgprA3 in mouse, or MRGPRX1 in human), which is activated by the antimalarial drug chloroquine 7 . Sensory neurons expressing this receptor become excited by a wide variety of pruritogens involved in acute histaminergic and nonhistaminergic itch, as well as in chronic itch 8 . Other work addressed neuronal pathways involved in the spinal relay of itch. These studies identified excitatory interneurons expressing gastrin releasing peptide (GRP) 9 or the GRP receptor (GRPR) 10,11 as key elements of this process. These itch-relay pathways appear to be under tight control by dorsal horn inhibitory neurons. Lack of a certain subset of these neurons that depend on the transcription factor Bhlhb5 leads to severe chronic itch in mice 12 . Local ablation of inhibitory neurons of the deep dorsal horn induces abnormal grooming and biting behavior, and localized hair loss reminiscent of chronic itch syndromes 13 . Conversely, local activation of these neurons through DREADDs (designer receptors exclusively activated by designer drugs 14 ) suppressed histamine-dependent and histamineindependent itch, demonstrating that inhibitory dorsal horn neurons exert a profound control over spinal itch relay 13 .
Inhibitory neurons of the spinal dorsal horn release two fast amino acid transmitters, GABA and glycine, to reduce the excitability of their postsynaptic target neurons. In the present study, we focused on the GABAergic system and investigated whether itch, in particular chronic itch, can be suppressed through pharmacological modulation of specific subtypes of spinal GABA A receptors (GABA A Rs). GABA A Rs are pentameric anion channels built from a repertoire of 19 subunits 15 . Most GABA A Rs in the brain and spinal cord are composed of α, β, and γ subunits in a 2:2:1 stoichiometry. The mammalian genome harbors 12 genes encoding for these subunits (α1-6, β1-3, and γ1-3). Spinal GABA A Rs mainly contain α1, α2, α3, or α5 subunits together with β2/3 subunits and a γ subunit. α4 and α6 subunits are only sparsely expressed or completely lacking 16,17 . Differences in the physiological functions and pharmacological properties of these GABA A Rs are mainly determined by the α subunit 18 . In the present study, we first used genetically modified mice to identify α2/α3 containing GABA A Rs as key elements of spinal itch control. Building on this result we assessed potential antipruritic actions of an α2/α3GABA A R selective compound (TPA023B; ref. 19 ) and showed that it not only reduced acute histaminedependent and histamine-independent itch in mice but also chronic itch in mice and dogs without apparent side effects.

Results
Inhibitory input to peripheral and spinal pruritoceptors. We first verified that itch relaying GRP neurons receive input from local inhibitory interneurons. Retrograde mono/transsynaptic rabies virus-based tracing experiments 20 initiated from GRP::cre neurons identified numerous inhibitory and excitatory neurons presynaptic to GRP neurons (Fig. 1). About half of the inhibitory neurons were located in lamina II of the dorsal horn, where most inhibitory neurons are purely GABAergic 21 . The other half resided in deeper layers, where most inhibitory neurons co-express glycine and GABA 13 .
We next investigated the presence of α1, α2, α3, and α5GABA A R subunits on spinal axon terminals of primary MrgprA3 positive pruritoceptors and on spinal GRP neurons (Fig. 2). Both MrgprA3 fibers and GRP neurons are concentrated in lamina II 8,22 , which also harbors α2 and/or α3GABA A R subunits at high density 16,17,23 . To visualize MrgprA3 axons and terminals and GRP neurons, we used GRP::eGFP and MrgprA3:: cre-eGFP;ROSA26 lsl-tdTom transgenic mice. Immunostaining of spinal cord sections of these mice confirmed that the region of α2 and α3GABA A R subunit expression overlapped with that of MrgrpA3 terminals and GRP neurons (Fig. 2a,c). A similar GABA A R subunit expression pattern was found in the cervical spinal cord and the medullary dorsal horn (see Supplementary  Fig. 1). By contrast, α1 and α5GABA A R subunits were largely missing from lamina II. Confocal analysis at higher magnification further demonstrated that α2 and α3GABA A R subunits were located on MrgprA3 fibers and GRP neurons (Fig. 2b,d). To allow a quantification of α2 and α3GABA A R subunit expressing neurons, we performed fluorescent in situ hybridization in sections of lumbar dorsal root ganglia (DRGs) and lumbar spinal cords (Fig. 2e). In these experiments, we also included spinal cord sections from GRPR::eGFP transgenic mice. Only about 20% of MrgprA3 positive DRG neurons expressed α2 and α3GABA A R subunit transcripts. This low expression is consistent with previously published single cell RNAseq data 24 (Supplementary  Table 1). By contrast, virtually all GRP and GRPR neurons expressed α3GABA A R subunit transcripts, and more than 60% of these neurons also expressed α2GABA A R subunits. Co-expression of α2 and α3GABA A R subunit transcripts with GRP and GRPR transcripts was also observed in human spinal cord tissue ( Supplementary Fig. 2).
GABA A R subtypes with antipruritic efficacy. We then asked whether pharmacological targeting of GABA A Rs containing α2 or α3 subunits (α2/α3GABA A Rs) reduces itch. To this end, we administered diazepam, a classical benzodiazepine that nonselectively potentiates the activation of benzodiazepine-sensitive GABA A Rs. To restrict diazepam's action to a single GABA A R subtype (α1, α2, α3, or α5), we used triple GABA A R point mutated mice, which carry a histidine to arginine (H → R) point mutation in three of the four benzodiazepine sensitive GABA A R α subunits 23 (designated as HRRR, RHRR, RRHR, and RRRH mice, for mice in which only α1, α2, α3, or α5GABA A Rs remained diazepam sensitive). Because mice of this particular genetic background (129SvJ) have not yet been systematically analyzed in itch experiments, we first assessed their sensitivity to different pruritogens. We found that injection of α-methyl serotonin (αmethyl 5-HT), a metabolically more stable derivative of the pivotal itch messenger serotonin 25 , induced robust dosedependent scratching behavior ( Supplementary Fig. 3). We then tested the effect of systemic diazepam (10 mg kg −1 , p.o.) on scratching responses evoked by α-methyl 5-HT (20 µg). Selective targeting of α2 or α3GABA A Rs (in RHRR or RRHR mice) strongly reduced scratching bouts (Fig. 3a-d). No significant reduction was observed after activation of α5GABA A Rs (in RRRH mice). A possible contribution of α1GABA A Rs could not be addressed as their activation induces strong confounding sedation 23,26 .
To test whether also chronic itch would respond to α2/ α3GABA A R facilitation, we employed the oxazolone model of chronic atopic-like dermatitis 27 (Fig. 3e-h). Oxazolone was repeatedly applied to the shaved nape of the neck for 17 days. This procedure induced profound scratching responses in wildtype mice and the three lines of triple GABA A R point mutated mice. Systemic treatment with diazepam (10 mg kg −1 , i.p.) again strongly reduced scratching when only α2 or α3GABA A Rs were targeted (in RHRR and RRHR mice). It had no obvious effects on home cage behavior in these mice (see Supplementary Movies 1-4).
Antipruritic efficacy of an α2/α3 selective GABA A R modulator. We then tested whether the data obtained in genetically modified mice would translate into therapeutic efficacy of GABA A R subtype-selective compounds. To this end, we tested the antipruritic efficacy of the α1-sparing GABA A R modulator TPA023B (ref. 19 ). As a prerequisite, we verified the in vitro pharmacological profile of TPA023B in transiently transfected HEK293 cells. As reported previously 19 , TPA023B had partial agonistic activity at the benzodiazepine binding site of α2β3γ2 and α3β3γ2 GABA A Rs, but did not potentiate α1β2γ2 GABA A Rs and had very weak potentiating effects on α5β2γ2 GABA A Rs (Fig. 4a). It did not activate GABA A Rs in the absence of GABA. We then asked whether this partial agonistic activity would translate to a facilitation of GABAergic inhibition in MrgprA3 or GRP neurons (Fig. 4b, c). In dissociated MrgprA3::cre-eGFP positive DRG neurons, TPA023B (1 µM) led to a slight statistically insignificant increase in GABA A R current amplitudes (P = 0.068, two-sided paired t-test, n = 10) (Fig. 4b) Fig. 1 Inhibitory input onto spinal GRP positive itch relaying interneurons. a Neurons providing synaptic input onto GRP positive neurons were identified using transsynaptic rabies virus based tracing. GRP::cre;ROSA26 TVA mice (n = 3) were first injected with AAV.flex.mCherry-RabG into the left lumbar spinal cord, and 14 days later with an EnvA pseudotyped glycoprotein deficient rabies virus (EnvA.RabiesΔG-eGFP). b-g Transverse sections of the injected dorsal horn. Cre positive GRP neurons infected with AAV.flex.RabG (red) primary and secondary Rabies virus infected neurons (green). Co-staining with Lmx1b and Pax2 revealed excitatory and inhibitory neurons, respectively. b primary rabies virus infected neurons express eGFP and mCherry, and appear yellow. Secondary rabies virus infected neurons are green only. c Co-staining with Lmx1b (blue) reveals the excitatory population of secondary infected (transsynaptically labeled) neurons (magenta; eGFP and Lmx1b positive but mCherry negative) d Classification of transsynaptically labeled neurons as excitatory (39%, Lmx1b positive) or inhibitory (45%, Pax2 positive). 16% of transsynaptically infected neurons (eGFP positive but mCherry negative) were neither stained with antibodies against Lmx1b or Pax2, and remained unclassified. e Co-staining with PKCγ (red) indicates border between lamina II and III. mCherry was not visible in e-g because it required signal amplification by an mCherry antiserum. f Co-staining against Pax2 revealed transsynaptically labeled inhibitory neurons (magenta; Pax2 and eGFP positive). Scale bar, 100 µm. We did not find any mCherry/eGFP/Pax2 triple positive neurons (not shown) consistent with the known excitatory phenotype of spinal GRP neurons 22 . g Higher magnification of the area indicated in f. Arrow heads indicate rabies virus infected (eGFP positive) inhibitory (Pax-2 positive) neurons. Of the inhibitory transsynaptically labeled neurons, 48% were in laminae I/II and 52% in laminae III/IV. Quantifications were performed on 9 sections obtained from 3 mice vehicle (open circles) for the three genotypes. P values were obtained from unpaired two-sided t-tests corrected for three independent comparisons. n = 8 and 10 (α2, vehicle, diazepam); n = 8 and 8 (α3, vehicle, diazepam); n = 7 and 9 (α5, vehicle, diazepam). c Dose-dependence of the antipruritic effects of diazepam in RHRR mice (only α2GABA A Rs sensitive to diazepam). ANOVA followed by Dunnett's post hoc test F(6,36) = 6.02; *P < 0.05; **P < 0.01, n = 7, 7, 4, 6, 6, 7, 5, for vehicle, and 0.1, 0.3, 1.0, 3.0, 10, and 30 mg kg −1 . d Same as c but RRHR mice (only α3GABA A Rs sensitive to diazepam). ANOVA followed by Dunnett's post hoc test F(6,37) = 4.42; *P < 0.05; **P < 0.01, n = 7 (vehicle) and 6 for all other groups. e-h Suppression of chronic itch. Antipruritic effects of α2, α3, and α5GABA A Rs in the oxazolone model of atopic-like dermatitis. Mice were sensitized to oxazolone over 17 days and treated with diazepam (10 mg kg −1 , i.p.) or vehicle on day 18. Scratching bouts were counted for 6 h starting 15 min after drug or vehicle administration. e-g Number of scratching bouts plotted versus time after drug or vehicle administration. h P values obtained from unpaired two-sided t-tests for the three genotypes, corrected for three independent comparisons. Unpaired two-sided t-tests, n postsynaptic currents by 43 ± 10% (P < 0.01, paired two-sided ttest, n = 7) (Fig. 4c). Together with the in situ hybridization data, these results suggest that GABA A Rs inhibit the spinal relay of itch signals primarily at the level of dorsal horn interneurons rather than via inhibition of primary pruritoceptors.
Would the favorable in vitro profile of TPA023B translate into reduced propensity to side effects? Consistent with the lack of agonistic activity at α1GABA A Rs, TPA023B did not induce sedation at doses up to 3 mg kg −1 (p.o.), but instead increased locomotor activity at 1 and 3 mg kg −1 (Fig. 5a). TPA023B did not cause muscle relaxation (Fig. 5b) and did not impair motor coordination (Fig. 5c).
We then continued investigating the efficacy of systemic (p. o.) TPA023B against acute itch evoked by α-methyl 5-HT (20 µg), chloroquine (100 µg), or histamine (100 µg) (Fig. 6a). Chloroquine-induced scratching was reduced by TPA023B in wild-type mice at doses ≥ 0.03 mg kg −1 . Similar effects were obtained with α-methyl 5-HT and histamine. To verify that the antipruritic effect of TPA023B was due to its interaction with the benzodiazepine binding site of GABA A Rs, we used GABA A R triple point mutated mice (HRRR mice), in which all GABA A Rs susceptible to modulation by TPA023B (α2, α3, and α5 GABA A Rs) had been rendered benzodiazepineinsensitive. The antipruritic action of TPA023B was completely lost in these mice (Fig. 6b). We also confirmed that the H → R point mutation prevented potentiation and binding of GABA A Rs by TPA023B ( Supplementary Fig. 4). TPA023B (3 mg kg −1 , p.o.) had no significant effect on responses evoked by stimulation with a paint brush or von Frey filaments, and did not reduce responses to acute noxious heat ( Fig. 6c-e) excluding a non-specific block of sensory relay through the spinal cord.
Would TPA023B also alleviate chronic itch? We first employed again the oxazolone model of atopic-like dermatitis. Acute treatment with TPA023B (1 mg kg −1 , i.p.) caused a highly significant reduction in the number of scratching bouts (determined during the interval 15-210 min after drug injection) ( Fig. 8a; for potential effects of i.p. TPA023B on locomotor behavior and motor performance see Supplementary Fig. 5). A similar antipruritic action was observed in the dry skin dermatitis model 30 (Fig. 8b). Treatment of chronic itch conditions with TPA023B would require that no loss of therapeutic activity occurs during repeated applications. To test whether TPA023B would retain therapeutic activity during chronic treatment, we compared the antipruritic activity of TPA023B in drug naïve mice with that in mice treated with TPA023B (1 mg kg −1 i.p.) once daily for ten days (Fig. 8c). No significant loss of antipruritic activity was observed. Treatment with TPA023B over several days also alleviated skin lesions. After seven days of once daily treatment with TPA023B (1 mg kg −1 , i.p.), mice showed a progressive reduction in their dermatitis scores (Fig. 8d). Furthermore, chronic treatment with TPA023B also reduced the infiltration of the affected skin by macrophages (Supplementary Fig. 6). By contrast, topic treatment with TPA023B (100 µl 0.3 µM, once daily) over the same time period neither changed scratching behavior nor dermatitis scores (Fig. 8e).
Additional experiments verified a spinal site of action also for chronic itch. First experiments in global α2 R/R point mutated mice revealed that eliminating diazepam sensitivity from α2GABA A Rs was not sufficient to block the antipruritic action of TPA023B (Fig. 9a). Because selective targeting of α3GABA A Rs had an even stronger antipruritic effect than targeting α2GABA A Rs (compare Fig. 3h), we focused our efforts on the α3GABA A R subtype and investigated α3GABA A R subunit knockout mice, in which deletion of the α3GABA A R subunit gene was achieved through the insertion of a duplicated exon (5*) flanked by loxP sites 31 . This design allows a cre-dependent (tissuespecific) rescue of α3GABA A R subunit expression (Fig. 9b). The antipruritic effect of TPA023B was completely lost after global deletion of α3GABA A R subunits but was fully restored in spinal cord-specific hoxB8-GABA A Rα3 resc/resc mice (Fig. 9c).
Antipruritic efficacy of TPA023B in pruritic dogs. Encouraged by its efficacy in mice and the absence of apparent side effects, we tested the antipruritic efficacy of TPA023B in a second, hierarchically higher, species (Fig. 10). We chose dogs because models of atopic dermatitis that closely mimic natural disease are wellestablished in this species 32 and because the tolerability of TPA023B in dogs had already been established 33 . We performed a pseudo-randomized observer-blind placebo-controlled crossover trial and in twelve beagle dogs sensitized to house dust mites by repeated exposure of their lower abdomen to lyophilized extracts of Dermatophagoides farinae 34 . After a sensitization period of 8 weeks during which the dogs were exposed to the extracts once a week, three challenges on three consecutive days were made in week 12. Nine of the 12 dogs developed clinical signs of pruritus and were included in subsequent experiments. These nine dogs were again challenged in weeks 15 and 18 on three consecutive days. The challenge on day 2 was used to obtain baseline values. The challenge on day 3 was followed by TPA023B (20 mg in one tablet, equivalent to about 2 mg kg −1 , p.o.) or placebo administration. The same procedure was repeated in week 18 with a cross-over design. Both the time spent scratching and the numbers of scratching bouts were counted over 6 h after drug administration and normalized to the values on the day before drug exposure. Compared to placebo, a significant reduction was observed for both read-outs. Five of the 9 dogs (56%) responded with a reduction by more than 50% (for responses of the individual dogs see Supplementary Table 2). These results indicate that specific targeting of α2/α3GABA A Rs alleviates itch not only in mice but also in dogs supporting the potential for translation to more complex species, including possibly, also human patients.

Discussion
Local application of TPA023B to the spinal cord and specific genetic ablation of spinal α2GABA A Rs have shown that most if not all of its antipruritic action originates from the spinal (or medullary) dorsal horn. This is consistent with our finding that virtually all GRP and GRPR neurons in the mouse spinal cord express α3GABA A R transcripts and more than 60% of these neurons also express α2GABA A Rs. Using post mortem tissue samples, we confirmed a similar coexpression of α2 and α3GABA A R subunits on GRP and GRPR neurons for the human spinal cord, in line with the highly conserved expression pattern of spinal GABA A R subunits in rodents and humans 35,36 .
MrgprA3 is primarily activated by chloroquine 7 . However, the fibers carrying this receptor respond to many pruritogens and convey signals related to both histaminergic and nonhistaminergic itch 8 . A critical role of GRP 37 and GRPR 38 neurons in spinal itch circuits is meanwhile also well established. After local spinal ablation of GRP neurons, mice respond less to pruritic stimuli 37 . This is also supported by a study that employed BNP-conjugated saporin to ablate spinal GRP neurons 9 . Besides GRP and GRPR positive interneurons, neurokinin 1 (NK1) receptor expressing neurons of the dorsal horn are also relevant to spinal itch relay 39 . Previous work from our group has shown that these neurons also express α2 and α3GABA A Rs 40 .
α2/α3GABA A Rs do not only control itch but also pain 23,40 (for a review see ref. 41 ). The latter function is supported by the expression of α2/α3GABA A Rs receptors on sensory fibers and dorsal horn nociceptive neurons 16,40 and also by the efficacy of spinal GABAergic neuron transplants against pain and itch 42,43 . Unlike opioids that reduce pain but cause itch 44 , drugs targeting α2/α3GABA A Rs should alleviate both itch and pain. Activation of α3GABA A R had a stronger impact on scratching responses than α2GABA A Rs. This ratio is reverse in case of analgesia 23,40 in line with the view that sensory fibers and dorsal horn neurons processing itch or pain are not identical 3,45 . There were also differences in the contribution of α2 and α3GABA A Rs between models of acute and chronic itch. These may hint at potential neuroplastic changes occurring during the transition from acute to chronic itch.
In the present study, we have assessed unwanted effects that typically occur with classical non-selective benzodiazepine site ligands. TPA023B was devoid of sedative effects, did not impair motor coordination and did not lose antipruritic activity during prolonged treatment. The absence of a sedative effect is consistent with the pivotal role of α1GABA A Rs in benzodiazepine-induced sedation 26 . The increase in locomotor activity observed in the present study is possibly related to the anxiolytic action of α2GABA A Rs (see also ref. 23 ). Previous studies in triple GABA A R point mutated mice have attributed the impairment of motor coordination and tolerance development to α1 or α3GABA A Rs 23 . The lack of these side effects in TPA023B treated mice may either come from the absence of activity at α1GABA A Rs or from the only partial agonistic activity (relative to the full agonist diazepam) at α3GABA A Rs.
In human patients, the most challenging type of pruritus is chronic non-histaminergic pruritus which is often due to liver or kidney failure or treatment with opioids 46 . While antihistaminergics provide good itch relief in cases associated with urticaria, other forms respond less well to these drugs or not at all. At present, these conditions are mostly treated off-label with systemically applied anticonvulsants, such as gabapentin and pregabalin, antidepressants, opioid antagonists, or more recently with NK1 receptor blockers (aprepitant), or immunosuppressants 47 . None of these compounds have been approved for systemic treatment of chronic itch conditions. In our study, TPA023B was similarly effective against non-histaminergic and histamine-induced itch, and, in general, more effective in chronic than in acute itch. If the antipruritic efficacy combined with good tolerability observed in the present study translates to human patients, TPA023B or related compounds should be well-suited for the treatment of chronic itch in humans.
In a broader context, our study adds to a growing body of evidence indicating that subtype selective GABA A R modulators promise not only better tolerability but may also open new avenues to the treatment of disorders that have hitherto not been considered indications for benzodiazepines. Such potential new opportunities include, in addition to itch and pain, cognitive enhancement by inverse agonists at α5 GABA A Rs 48,49 , depression and autism spectrum disorders (for a review see ref. 50  (ROSA)26Sor tm14(CAG-tdTomato)Hze /J (ROSA26 lsl-tdTom , also known as Ai14; www.jax.org/strain/007914) were obtained from Jackson Laboratories. Sns-cre mice (Tg(Scn10a-cre)1Rkun) 54 were kindly provided by Dr. Rohini Kuner from University of Heidelberg. HoxB8-cre (Tg(Hoxb8-cre)1403Uze) 55 and GABA A Rα3 −/− mice (Gabra3 tm2Uru ) 56   Intraspinal virus injections. Animals were anesthetized with 2-5% isofluorane and lumbar vertebrae L4 and L5 were exposed. The animal was then placed in a motorized stereotaxic frame and the vertebral column was immobilized using a pair of spinal adaptors. The vertebral lamina and dorsal spinous process were removed to expose the L4 lumbar segment. The dura was perforated about 500 μm Image analysis. Fluorescent images were acquired on a Zeiss LSM710 Pascal confocal microscope using a 0.8 NA 20x Plan-apochromat objective or a 1.4 NA 63x EC Plan-Neofluar oil-immersion objective and the ZEN2012 software (Carl Zeiss). Whenever applicable, contrast, illumination, and false colors were adjusted in ImageJ or Adobe Photoshop (Adobe Systems, Dublin, Ireland). Cell numbers were quantified in a total of 9 sections prepared from 3 animals. In order to avoid double counting of cells in adjacent sections, all sections used for quantification were taken at a distance of at least 50 µm. The numbers of immune reactive cells were determined using the ImageJ Cell Counter plug-in.
Immunohistochemistry and image analysis of GABA A R subunits. Colocalization of GABA A R α subunits with MrgprA3 axons and GRP neurons was visualized on 40 µm thick transverse mouse lumbar spinal cord cryosections. Mice were deeply anaesthetized with pentobarbital (nembutal, 50 mg kg −1 , i.p.) and perfused with oxygenated aCSF. Spinal cords were rapidly collected by pressure ejection and placed in ice-cold 4% PFA for 90 min. The spinal cords were then cryoprotected overnight in a 30% sucrose/PBS solution, snap frozen with dry ice and cut in 40 μm thick coronal free-floating slices kept in antifreeze at −20°C until the day of staining 16 . Antibodies were home-made subunit-specific antisera 16,58 . Final dilutions were 1:20,000 (α1), 1:1,000 (α2), 1:10,000 (α3), and 1:3,000 (α5). The distribution of GABA A R α subunits in dorsal horn GRP neurons and MrgprA3 axons was analyzed by immunofluorescence staining on coronal sections prepared from 2-3 male GRP::eGFP transgenic mice as described above. For staining, the sections were incubated overnight at 4°C with a mixture of primary antibodies diluted in Tris buffer containing 2% normal goat serum. Sections were washed extensively and incubated for 1 h at room temperature with the corresponding secondary antibodies conjugated to Cy3 (1:500), Cy5 (1:200) (Jackson ImmunoResearch) or Alexa488 (1:1000, Molecular Probes, Eugene, OR). Sections were washed again and cover-slipped with fluorescence mounting medium (DAKO, Carpinteria, CA). Double-immunofluorescence signals were visualized by confocal microscopy (LSM 710; Zeiss AG, Jena, Germany) using a 63x Plan-Apochromat objective (N.A. 1.4). The pinhole was set to 1 Airy unit for each channel and separate color channels were acquired sequentially. The acquisition settings were adjusted to cover the entire dynamic range of the photomultipliers. Typically, stacks of confocal images (1024 × 1024 pixels) spaced by 0.3 μm were acquired at a magnification of 56-130 nm/pixel. For display, images were processed with the image analysis software Imaris (Bitplane; Zurich, Switzerland). Images from all channels were overlaid (maximal intensity projection) and background was subtracted, when necessary. A low-pass filter was used for images displaying α subunit staining. Analysis of the distribution of α subunit-IR in GRP::eGFP neurons and dendrites and MrgprA3::cre-eGFP axons was performed in single confocal sections acquired at a magnification of 78 nm / pixel in 8-bit gray scale images, using a threshold segmentation algorithm (minimal intensity, 90-130; area > 0.08 μm 2 ). Distribution of GABA A Rα3 protein was studied in brain and spinal cord sections obtained from adult hoxB8-α3 resc/resc , global α3 −/− , and wild-type mice. For immunoperoxidase staining, a polyclonal antibody directed against the N-terminal fifteen amino acids (pGluGESRRQEPGDFVKQ) of the rat GABA A R α3 protein was used as the primary antibody. Sections from GABA A R-mutated mice and from controls were treated in a strictly parallel fashion.
RNAscope fluorescent in situ hybridization. Multiplex fluorescent in situ hybridization was performed using RNAscope (Advanced Cell Diagnostics; ACD; ref. 59 ). In brief, dissected tissue was snap frozen in liquid nitrogen and later on embedded in NEC50. Thirty five µm sections were cut on a Hyrax C60 cryostat, mounted on superfrost + glass slides and stored at −80°C until use. The in situ hybridization was carried out according to the manufactures protocol. The following RNAscope probes were used: GABA A R α1 (catalog number: 435351); GABA A R α2 (435011-C2); GABA A R α3 (435021-C3); GABA A R α5 (319481); GRP (317861 and 317861-C2); GRPR (317871 and 317871-C2); EGFP (400281 and 400281-C3). The latter was used to detect eGFP expressed under the genetic control of the MrgprA3 gene in MrgprA3::cre-eGFP transgenic mice.
Human spinal cord tissue samples. Tissue has been extracted during routine autopsy from five patients. Two samples had been formalin fixed and paraffin embedded, and three were fresh frozen tissue samples. All samples were screened with control RNAscope fluorescent in situ hybridization probes provided by the manufacturer of the assay [human Polr2a (C1), PPIB (C2) and UBC (C3)]. Tissue samples of one patient were well enough preserved to yield reliable and specific in situ hybridization signals. These samples were from a 37 years old male patient with a post mortem time of 16 h who had died of septic shock and right heart failure. From this patient, small samples from the cervical, thoracic and lumbar spinal cord had been fixed in 4% formalin for 14 days, and were then paraffinembedded. Ten micrometer thick slices were mounted on superfrost glass slides. Multiplex fluorescent in situ hybridization on human spinal cord tissue was performed using RNAscope® Multiplex Fluorescent Reagent Kit Version 2 (323100, Advanced Cell Diagnostics; ACD; ref. 59 ). The in situ hybridization was carried out according to the manufactures protocol with the following modifications. Target retrieval was carried out at 94-98°C for 30 min in target retrieval buffer. Protease incubation was carried out for 30 min at 40°C with the protease plus reagent. The following RNAscope probes were used: α2GABA A R (Hs-GABRA2 448211-C2), α3GABA A R (Hs-GABRA3 320269-C3), GRP (Hs-GRP 465261-C1) and GRPR (Hs-GRPR 460411-C1) were used to detect co-expression of GRP / GRPR and α2GABA A R or α3GABA A R subunit transcripts.
Skin histology and immunofluorescence. Inflamed and healthy back skin was collected. Tissues were embedded in OCT compound (Sakura Finetek, Torrance, USA) and frozen on dry ice. Cryostat sections (7 μm) were placed on glass slides, air dried, fixed with acetone for 2 min at −20°C and subsequently rehydrated with 80% methanol for 5 min at 4°C. Specimens were incubated with 5% donkey serum, 0.1% Triton-X and 1% BSA in PBS for 1 h at room temperature, followed by overnight incubation with rat anti-mouse CD68 (  Behavioral experiments in mice. All behavioral experiments were performed in 7-12-week-old female and male mice. Care was taken to ensure equal numbers of female and male mice. All behavioral experiments were made by an experimenter, blinded either to the genotype of the mice or to their treatment with drug or vehicle. In experiments involving comparisons between diazepam or TPA023B and vehicle, mice were randomly assigned to the different groups. No formal sample size calculation was made. Group sizes were chosen based on previous experience with the respective behavioral test. Mechanical sensitivity was assessed with electronic von Frey filaments (no. 7; IITC, Woodland Hills, CA) and quantified as the change in the paw withdrawal thresholds measured in g. Heat hyperalgesia was evaluated in the Hargreaves test as the change in the latency of paw withdrawal to a defined heat stimulus. Responses to light mechanical stimulation of the hairy skin was tested as the change in the paw withdrawal responses upon gentle stimulation with a paint brush using the following score: 0 (no evoked movement), 1 (walking away or brief paw lifting of 1 s or less), 2 (sustained lifting of more than 2 s), 3 (strong lateral lifting above a 90°angle) or 4 (flinching/ licking of the affected paw). Six measurements were made for each animal for all three tests.
Acute itch. was assessed in mice that received intradermal microinjections of pruritogens or 0.9% saline into the right cheek, which had been shaved at least 1 day before the experiment. In two sets of experiments that addressed the contribution of GABA A Rs on primary and secondary pruritoceptors, pruritogens were injected into the skin of the left thigh (Fig. 7a-c). Before injection, mice were acclimatized to a 15 cm diameter cylindrical enclosure for more than 30 min with cage bedding on the floor. Background white noise generated by a radio at ambient volume was applied to prevent auditory distraction. A 30 gauge needle was inserted bevel-up and pushed 5 mm horizontally into the skin beyond the point of insertion, before injection of the pruritogen (in a total volume of 10 µl). No anesthesia was used. Correct injection was confirmed by the appearance of a slightly domed bulla. After injection, mice were placed back into the cylindrical enclosure and videotaped for 30 min. Videos were analyzed off-line. Scratching with the hind paw directed to the ipsilateral cheek was counted in bouts, with one bout defined as an instance when the mouse lifted its paw to scratch until it returned the paw to the cage floor. In case of experiments in which the pruritogen was injected into the skin of the thigh, the time spent biting the injected skin area was counted in s min −1 as a measure of itch.
Chronic itch. was investigated in the contact dermatitis model 27 and the dry skin model 30 . To induce contact dermatitis, mice were treated on day 0 with 10% oxazolone in acetone/olive oil (4:1 v v −1 ) on the shaved nape of the neck (100 µl). After a resting period of 7 days, mice were treated with 1% oxazolone in acetone/ olive oil (4:1 v v −1 ) on the nape of the neck (100 µl) every other day for 10 days. On the day of the experiment, mice were injected with drug or vehicle i.p. under short light isoflurane anesthesia. Scratching of the hind paw directed to the ipsilateral cheek was quantified as the number of scratching bouts. In the dry skin model, mice were treated with a mixture of acetone and diethylether (1:1) on the shaved nape of the neck for 15 s, followed by distilled water for 30 s, twice daily for 10 days. On the day of the experiment, vehicle or drug was administered i.p. under short light isoflurane anesthesia.
To quantify the severity of skin lesions a dermatitis score was determined 64 . Hemorrhage/erythema, dryness/scaring, and hyperplasia were scored as 0 (none), 1 (mild), 2 (moderate), or 3 (severe) once per day resulting in a score between 0 and 9. Photographs of atopic dermatitis-like skin were taken before treatment (day 11) and 9 days after treatment (day 20).
Locomotor activity, motor coordination and muscle relaxation. TPA023B (1 mg kg −1 , p.o. or i.p.) or vehicle was administered 60 min before the tests. Locomotor activity was measured in an open field arena (10 cm radius) equipped with four pairs of light beams and photosensors and analyzed for the time interval between 60 and 120 min after TPA023B administration. Motor coordination was assessed with a rotarod accelerating from 4 r.p.m. to 40 r.p.m. within 5 min. Fifteen measurements were taken per mouse. To assess muscle relaxation, mice were placed with their forepaws onto a metal horizontal wire placed 20 cm above ground. Successes and failures to grab the wire with at least one hindpaw were recorded between 60 and 120 min after TPA023B administration.
Pruritus study in dogs. Twelve 6 months-old beagle dogs (6 females, 6 males) weighing between 9.0 and 13.0 kg (see Supplementary Table 1) were included in the study. They were sensitized using lyophilized extracts of the house dust mite Dermatophagoides farinae in mineral oil. In order to expose the living epidermis, the skin of the abdomen was tape-stripped. D. farinae extract was gently applied on the tape-stripped skin once a week for eight weeks. At this time point, dogs were considered D. farinae-sensitized even though most of them (8 out of 12) did not exhibit pruritus or clinical signs of atopic dermatitis. Four weeks later, the dogs were challenged using D. farinae extracts after tape stripping of the abdominal skin. This challenge was made on three consecutive days (days 1 to 3). Pruritus and clinical signs were assessed on day 2, 3, and 4. All dogs exhibited pruritus and clinical signs compatible with atopic dermatitis on the site of challenge but also in remote areas. After the third challenge all dogs exhibited histological signs of atopic dermatitis and nine dogs showed positive reactions in intradermal allergen exposure test. These nine dogs were enrolled to assess the antipruritic effect of TPA023B in a randomized placebo-controlled observerblind cross-over study. TPA023B was administered at a dose of 20 mg per dog as a tablet, 30 min after the challenge. For the following 6 h, dogs were kept in groups of three in closed arenas of 2 × 4 m and continuously video recorded. Both the number of scratching bouts and the total time spent scratching per hour were determined. Recordings were made on the day of the second challenge (baseline) and on the following day after drug or placebo administration. The same procedure was done three weeks later with switched treatment (TPA023B and placebo). Number of scratching bout and total time spent scratching were normalized to the control values obtained the day before (second challenge). We did not correct our outcome parameters for baseline scratching present in the absence of the pruritogenic challenge. Blood samples were taken from all dogs at the end of the video recording and whole blood concentrations of TPA023B were measured by high performance liquid chromatography/high resolution electrospray time-of-flight mass spectrometry to verify drug exposure.
Statistics. For most experiments, results of individual mice or cells are displayed as individual symbols. Normal distribution of data was assumed when t-tests or ANOVAs were applied.
Data availability. Excel files including the data that support the findings of this study are available at G-Node.org with the identifier doi: 10.12751/g-node.fb5bd5 [https://doid.gin.g-node.org/fb5bd596a1ff2f6e8cd29f51be351ad3/]. Additional raw data of this study are from the corresponding author upon reasonable request.