From all of the pruritogens identified to date, we chose to test the antimalarial drug chloroquine by injection into our Mrgpr-cluster Δ ^−/− mice (Liu et al. 2009); we selected this pruritogen first because of its medical relevance and wide clinical usage. Most black Africans have a severe itch reaction to chloroquine, and this reaction can be so severe that many patients even refuse to take chloroquine to treat their malaria (Olatunde 1977; Abila et al. 1989; Mnyika and Kihamia 1991; Ademowo et al. 1998); interestingly, other races do not have such an itch response to chloroquine (Sams and Epstein 1965; Spencer et al. 1982). Because patients experience itch when they first ingest chloroquine and because antihistamines do not effectively prevent this side effect, researchers in Africa concluded that chloroquine-induced itch is not an allergic response, nor is it mediated by histamine receptors (Abila et al. 1994). Interestingly, injecting chloroquine into the skin of mice evokes a robust scratching response in the mice, and various inbred strains of mice exhibit different degrees of scratching responses (Green et al. 2006). Based on this response, we tested whether our Mrgpr-cluster Δ ^−/− mice respond differently from wild-type following chloroquine injection. The results were immediately obvious—the Mrgpr-cluster Δ ^−/− mice scratched significantly less than the wild-type mice (Liu et al. 2009). We were quite encouraged by this robust phenotype, as we had been searching for a function for Mrgprs in more than 7 years since they were first identified. This important finding opened a new research path in our laboratory that ultimately led to the functional characterization of Mrgprs and Mrgpr-expressing neurons in mediating itch.

To determine the specific itch-inducing effect of chloroquine, we also tested a synthetic precursor of chloroquine (called quinoline), in which one of the benzene rings lacks a specific side chain. Injecting quinoline did not evoke scratch in either wild-type or Mrgpr-cluster Δ ^−/− mice (Liu et al. 2009), suggesting that the unique chemical structure of chloroquine is required for inducing itch. Because the majority of Mrgpr genes that are deleted in the cluster knockout mice are expressed specifically in DRG neurons, we hypothesized that the chloroquine itch-resistant phenotype of Mrgpr-cluster Δ ^−/− mice was likely due to a functional deficit in the DRG neurons in these mice. Using Ca²⁺ imaging in cultured wild-type DRG neurons, we found that a small subpopulation of DRG neurons (representing approximately 5 % of the total population) is indeed activated by chloroquine (Liu et al. 2009). Although this small subpopulation of chloroquine-responding neurons could be easily overlooked or considered to reflect a nonspecific response, cultured DRG neurons from Mrgpr-cluster Δ ^−/− mice did not respond at all to chloroquine, indicating that the activation of wild-type DRG neurons by chloroquine is both specific and Mrgpr dependent. Similar results were obtained by performing electrophysiological recordings of DRG neurons to monitor their action potentials in response to chloroquine. Taken together, the behavioral and cellular data pointed to the exciting likelihood that Mrgprs are receptors for chloroquine. To test this hypothesis directly, we cloned the cDNAs of each of the 12 deleted Mrgprs and expressed each receptor individually in HEK293 cells (which do not express endogenous Mrgprs). We found that only the MrgprA3-expressing HEK293 cells responded robustly to chloroquine; the remaining 11 lines of Mrgpr-expressing HEK293 cells were completely insensitive to chloroquine (with the exception of MrgprA1-expressing cells, which responded weakly) (Liu et al. 2009). We were quite surprised by this high specificity of activation, particularly given that MrgprA3 is highly homologous to the other MrgprAs that we tested.

Our group and others reported that the preproenkephalin peptide BAM8-22 is a specific ligand for both mouse MrgprC11 and human MrgprX1 receptors (Han et al. 2002; Lembo et al. 2002). The precise physiological function of BAM8-22 is currently unclear. However, because both the MrgprA3 and MrgprC11 receptors are expressed in a single subset of small-diameter DRG neurons, we tested whether BAM8-22 is a pruritogen. Indeed, intradermal injection of BAM8-22 evoked a scratch response in wild-type mice (Liu et al. 2009). In collaboration with Robert LaMotte’s group at Yale University, we used cowhage spicules to deliver BAM8-22 to the skin of healthy human volunteers (Sikand et al. 2011). The participants reported an itch response that was similar in intensity to histamine-induced itch. Importantly, the BAM8-22-loaded spicules did not trigger skin flares or wheals (both of which are hallmarks of histamine-mediated itch). Moreover, pretreating the participants with antihistamine cream completely blocked histamine-evoked itch but had no effect on BAM8-22-induced itch. Finally, BAM8-18, a truncated form of BAM8-22 that lacks the Mrgpr-interacting motif, failed to elicit any sensation in the participants. Taken together, these data support the results of our mouse study and suggest that BAM8-22 is likely an endogenous mediator of histamine-independent itch.

SLIGRL is a 6-amino acid peptide cleaved from the N-terminus of activated PAR2 following protease cleavage. SLIGRL and its human ortholog SLIGKV evoke itch in mice and humans, respectively (Steinhoff et al. 2003; Shimada et al. 2006). Although it is generally accepted that this type of itch is mediated by PAR2, no study has directly addressed this issue (e.g., using PAR2-knockout mice). Because SLIGRL is used widely as a pruritogen for studying histamine-independent itch, we tested the ability of this peptide to induce itch in Mrgpr-cluster Δ ^−/− mice (Liu et al. 2011). To our surprise, the Mrgpr-cluster Δ ^−/− mice scratched significantly less in response to SLIGRL injection compared with their wild-type littermates. Perhaps even more surprisingly, the SLIGRL-induced itch response in PAR2-knockout mice was similar to the response induced in wild-type mice, suggesting that this response is not mediated by PAR2. In vitro, SLIGRL activated a small subset of both wild-type and PAR2-knockout DRG neurons (approximately 3 % of the total neuron population), whereas Mrgpr-cluster Δ ^−/− neurons did not respond at all. Interestingly, most of the SLIGRL-sensitive DRG neurons also responded to BAM8-22, suggesting functional overlap.

Based on these unexpected yet compelling results, we next tested whether SLIGRL can activate Mrgprs directly by expressing each of the 12 Mrgprs in heterologous cells and monitoring activation with Ca²⁺ imaging. Because HEK293 respond to SLIGRL (presumably because they express endogenous PAR2), we used SLIGRL-insensitive Chinese hamster ovary (CHO) cells for these experiments. Strikingly, among the 12 genes that were tested, only MrgprC11-expressing CHO cells were activated by SLIGRL. In parallel experiments, we also found that SLIGKV (the human version) specifically activated human MrgprX2 but not MrgprX1, MrgprX3, or MrgprX4. Based on a dose-response study, we determined that the EC50 of SLIGRL-induced MrgprC11 activation is on par with the EC50 of SLIGRL-induced PAR2 activation, suggesting that the peptide can activate these two receptors equally well. When we examined the SLIGRL sequence in detail, we hypothesized that the C-terminus of the peptide is required for the activation of MrgprC11, as this region contains a sequence motif that is similar to other MrgprC11-activating peptides such as RYamide and RFamide (Dong et al. 2001; Han et al. 2002). We next generated a truncated SLIGR peptide by removing the C-terminal leucine and found that SLIGR failed to activate MrgprC11 in heterologous cells. On the other hand, the truncated peptide still activated PAR2, as the PAR2-activating motif resides at the N-terminus of SLIGRL (which is intact in the truncated peptide). Therefore, SLIGR provided us with a useful tool to separate MrgprC11 and PAR2 signaling. Unlike SLIGRL, SLIGR cannot directly activate wild-type DRG neurons, nor does it evoke a scratching response in wild-type mice. Taken together, these data strongly suggest that SLIGRL-evoked itch is mediated by MrgprC11 but not PAR2.

Using the truncated peptide SLIGR, we also found that PAR2 plays a role in inflammatory pain by sensitizing TRPV1 channels in the DRG. Moreover, we cannot exclude the possibility that PAR2 mediates itch in other cell types such as keratinocytes. It is also highly conceivable that the function of PAR2 in the DRG switches from pain to itch under chronic itch conditions. For example, human psychophysical studies demonstrated that bradykinin and acetylcholine—two stimuli that are normally painful—evoke an itch response when applied to atopic dermatitis patients (Rukwied and Heyer 1999; Hosogi et al. 2006). Various proteases—including plant-derived proteases such as mucunain and papain and mammalian cathepsin S—evoke an itch response that is likely mediated by PAR2 (or related PARs), as PAR2 can be cleaved and activated by these proteases (Hollenberg and Compton 2002; Ossovskaya and Bunnett 2004). This hypothesis can be tested using PAR2-knockout mice. Our results also raise the interesting possibility that Mrgprs are targets for proteases.

Both MrgprA3 and MrgprC11 are deleted in our Mrgpr-cluster Δ ^−/− mice. Thus, is it possible that other Mrgprs that are not within the cluster are also itch receptors? A promising candidate is MrgprD, which is located far from the Mrgpr cluster near the telomere of mouse chromosome 7 and which has a clear human ortholog. Genetic labeling experiments revealed that MrgprD-positive axons only innervate the skin and single-gene MrgprD-knockout (MrgprD ^−/− ) mice have been generated (Zylka et al. 2005). Although MrgprD is expressed in 70 % of IB4-positive DRG neurons (which represent approximately 30 % of all DRG neurons), MrgprD ^−/− mice do not exhibit any sensory (e.g., thermal or mechanical) deficits. Previous studies revealed that β-alanine, a naturally occurring β-form of alanine, directly binds and activates both the human and mouse MrgprD receptors (Shinohara et al. 2004), although a literature search did not yield any published study demonstrating β-alanine-evoked itch. Nevertheless, we noticed that several weblogs by body builders contained many complaints of severe itch experienced immediately after taking oral β-alanine. Interestingly, β-alanine is a popular supplement used to build muscle; β-alanine is a major component of the dipeptide carnosine, a buffer that prevents muscle cells from becoming acidic, thereby decreasing fatigue during exercise (Crush 1970; Derave et al. 2007; Hill et al. 2007). In order to mimic the effect observed in humans, we gave wild-type and MrgprD ^−/− mice sweetened water containing β-alanine or sweetened water without β-alanine (Liu et al. 2012). The sweetened water alone (without β-alanine) did not elicit scratching in either wild-type or MrgprD ^−/− mice. However, several minutes after drinking the β-alanine-containing water, the wild-type mice began scratching (primarily at the back of the neck), and this behavior lasted for at least 1 h. In contrast, the MrgprD ^−/− mice scratched significantly less after drinking the β-alanine-containing water; the level of scratching in the MrgprD ^−/− mice was comparable to the scratching response in the mice that were given sweetened water without β-alanine. These data suggest that as in humans, β-alanine induces itch in mice and this effect is solely MrgprD dependent.

We were also excited by the ability to induce itch by delivering a compound orally, as we typically induce itch by injecting a pruritogen into the skin. However, many oral drugs have itch as a side effect. Thus, we can employ this oral delivery model to study the mechanisms of itchy side effects in mice. In addition to the oral delivery assay, we also injected β-alanine into the cheek (Liu et al. 2012). Consistent with oral administration, wild-type mice that were injected with β-alanine exhibited robust scratching with apparent pain response; in contrast, the MrgprD ^−/− mice did not exhibit an itch response. l-alanine itself (which does not activate MrgprD receptors) did not evoke a significant scratch response in wild-type animals, suggesting that only the β-form of the amino acid (i.e., with the amino group located at the β-position from the carboxylate group) can induce MrgprD-mediated itch. At the cellular level, β-alanine directly activated a subset of wild-type DRG neurons, but this response was absent in MrgprD ^−/− neurons (see below) (Liu et al. 2012). To quantify and analyze the site-specific effect of β-alanine-induced itch in humans, Dr. LaMotte’s group injected β-alanine into the skin of healthy volunteers (Liu et al. 2012). Upon injection of β-alanine, each participant experienced modest itch that was accompanied by tingling and a mild stinging sensation. Importantly, injecting β-alanine did not generate a typical histamine-mediated itchy skin reaction such as flares or wheals. Finally, as reported in mice, l-alanine did not evoke an itch sensation in humans.

Structurally, β-alanine is similar to the primary inhibitory neurotransmitters GABA and glycine, and β-alanine activates both GABA and glycine receptors (albeit with less potency) (Wu et al. 1993; Rajendra et al. 1997). Furthermore, β-alanine binds to glycine’s co-agonist site on NMDA receptors and suppresses glutamatergic signaling (Ogita et al. 1989; Pullan and Powel 1992). Therefore, in the central nervous system, β-alanine acts as an inhibitory neurotransmitter or modulator. Based on these findings, β-alanine-induced itch has been proposed to involve a central mechanism. However, our analysis of knockout mice strongly suggests that β-alanine-induced itch depends solely on MrgprD receptors. Our skin injection experiments in mice and humans combined with the finding that MrgprD receptors are expressed in DRG neurons—but not in CNS neurons—suggest that β-alanine-induced itch is mediated by a peripheral, cutaneous mechanism.

The chloroquine receptor MrgprA3 is expressed in 5–8 % of all DRG neurons (the reported percentage differs depending on the detection method used and the spinal level of the ganglion that is examined) (Liu et al. 2008, 2009; Han et al. 2013). These neurons are positive for both IB4 labeling and CGRP expression and negative for substance P (Han et al. 2013). However, the level of CGRP expression in MrgprA3-positive neurons is lower than in canonical peptidergic neurons. Therefore, the pattern of MrgprA3 expression is not random, but is specific to a subset of DRG neurons. Because MrgprA3-positive neurons can be activated by several pruritogens (including chloroquine, BAM8-22, and histamine) (Liu et al. 2009), they are a promising candidate as the itch-specific neurons. To test this hypothesis, we used the BAC transgenic approach to generate a new mouse line in which a GFP-Cre fusion protein is expressed selectively in MrgprA3-positive neurons (Han et al. 2013). Thanks to recent advances in mouse genetics, we were able to cross this MrgprA3-Cre line with several Rosa26 reporter lines, thus enabling specific expression of various reporter genes in MrgprA3-positive neurons without the need to generate additional mouse lines. With these tools in hand, we first examined the morphology of these DRG neurons using a reporter mouse expressing the red fluorescent protein tdTomato. Strikingly, MrgprA3-positive axons (i.e., red fluorescent fibers) innervate the skin but not any other tissue examined; this finding provides a cellular explanation for why we only feel itch in the skin but not in other, deeper tissues such as muscles and visceral organs (Han et al. 2013). Dr. LaMotte’s group then used their powerful in vivo DRG neuron recording technique to study the cellular properties of MrgprA3-positive neurons (identified clearly by their intrinsic green fluorescence). They recorded soma activation of lumbar DRG neurons in response to various stimuli applied to the receptive fields in the hind limbs and hind paws of live mice. Strikingly, they found that the electrophysiological properties of the MrgprA3-positive neuron population are remarkably homogenous. These neurons are typical polymodal nociceptive neurons, as they have slow conduction velocity (0.5 m/s) and respond to noxious heat, noxious mechanical force, and capsaicin (Han et al. 2013). In addition, MrgprA3-positive neurons can be activated by a variety of pruritogens, including histamine, BAM8-22, chloroquine, and cowhage. On the other hand, MrgprA3-positive neurons do not respond to the MrgprD agonist β-alanine (Han et al. 2013). Interestingly, ablation of MrgprA3-positive neurons specifically affects itch but not pain. Finally, we used a genetic approach to generate mice in which TRPV1 is exclusively expressed in MrgprA3-positive neurons. Strikingly, capsaicin, a normally painful stimulus, caused itch and not a pain response in the MrgprA3-TRPV1 mice. Together, these data strongly suggest that MrgprA3-positive neurons are itch-specific neurons and most fit the labeled-line theory.