Peripheral and Central Mechanisms of Itch

Abstract

Itch is a unique sensory experience that is encoded by genetically distinguishable neurons both in the peripheral nervous system (PNS) and central nervous system (CNS) to elicit a characteristic behavioral response (scratching). Itch interacts with the other sensory modalities at multiple locations, from its initiation in a particular dermatome to its transmission to the brain where it’s finally perceived. In this review, we summarize the current understanding of the molecular and neural mechanisms of itch by starting in the periphery, where itch is initiated and perceived, and discussing the circuits involved in itch processing in the CNS.

INTRODUCTION

Itch (also known as pruritus) is an uncomfortable everyday experience that evokes a desire to scratch. Exciting advances in the field of somatosensory physiology over the past decade have led us to view itch as a distinct sensory modality within the somatosensory system (along with the other senses of pain, temperature, touch and proprioception). Itch is a unique sensory experience, is encoded by genetically distinguishable neurons both in the peripheral nervous system (PNS) (Han et al., 2012; Usoskin et al., 2015) and central nervous system (CNS) (Mishra and Hoon, 2013; Mu et al., 2017; Sun and Chen, 2007b; Sun et al., 2009b) and elicits a characteristic behavioral response (scratching) in both humans and laboratory animals (Shimada and LaMotte, 2008). Yet itch also interacts with the other sensory modalities at multiple locations, from its initiation in a particular dermatome to its transmission to the brain where it’s finally perceived.

In this review, we summarize our current understanding of the molecular and neural mechanisms of itch. We will start in the periphery, at the site of the skin where itch is initiated and perceived, and ask the question from a neuron-centric standpoint: what molecules in the skin activate the itch sensory neurons and how is this signal converted into action potentials?

I. PERIPHERAL MECHANISMS OF ITCH

Causes of acute and chronic itch

When felt acutely, itch evokes an innate scratching response that assists in the removal of irritants (such as insects and poisonous plants) and calls attention to the affected skin areas. While this response may be evolutionarily useful, numerous skin conditions and systemic diseases can generate pathological chronic itch that lasts more than 6 weeks causing immense suffering (Ständer et al., 2007). Pathological itch conditions are generally classified into one of four categories: 1) Dermatological conditions such as atopic dermatitis and psoriasis; 2) Systemic diseases that arise from organs other than the skin, including liver and renal failure and drug-induced pruritus; 3) Neurological diseases that cause neuropathic itch through damage or malfunction of the peripheral or central nervous systems and 4) Psychiatric disorders (Bernhard, 2005; Ständer et al., 2007; Yosipovitch and Bernhard 2013). We will briefly review the causes of itch in the context of the affected dermatome and go over the major categories of known itch mediators, their cellular sources, their receptors on the neuronal membrane, and the signal transduction pathways that lead to the firing of an action potential by which the sensory information is transmitted to the CNS.

Figure 1 illustrates a schematic dermatome where itch is to be felt. The skin is a stratified organ that covers the whole body and serves as a barrier against environmental irritants. The outermost layer in direct contact with the environment is the epidermis, consisting primarily of keratinocytes (Joost et al., 2016; Merad et al., 2008; Pasparakis et al., 2014). Below the epidermis, separated by a basement membrane, is the dermis where skin appendages, blood and lymphatic vessels, and a variety of immune cells reside. Among the immune cells are tissue resident mast cells, a type of granulocytes filled with cytosolic granules containing histamine and other inflammatory mediators (Pasparakis et al., 2014). Further below the dermis lies the hypodermal adipose tissue. The sensory endings of itch neurons terminate in the epidermis forming branched free nerve endings containing a myriad of membrane receptors for various mediators (Han et al., 2012; Imamachi et al., 2009a; Liu et al., 2009; Zylka et al., 2005).

Figure 1. Peripheral mediators of itch.

The skin is a laminated structure that serves as a protection barrier. When a skin area is irritated or infected, keratinocytes and local immune cells release various mediators to mount an immune response and eradicate the insult. Itch sensory neurons that innervates the dermatome extend free nerve ending in the epidermis and express receptors that, once bound by the ligands, can trigger neuronal firing. Itch mediators with identified neuronal receptors include TSLP and IL-33 most secreted by keratinocytes, mast cell mediators including histamine, serotonin and various proteases, as well as several cytokines associated with T_H2 type T helper cells including IL-4, IL-13 and IL-31.

During an episode of acute itch (such as one triggered by a mosquito bite), the epidermis often suffers a breach or chemical insult (Green and Dong, 2016). The keratinocytes and local immune cells detect the damage or pathogen related molecular patterns and release a number of chemical mediators in response (Pasparakis et al., 2014). Mast cells undergo degranulation and release the content of their cytosolic granules. These granules contain mediators including histamine, serotonin, leukotriene, proteases, cytokines and chemokines (Rao and Brown, 2008). Histamine and other signals trigger local vasodilation and attract circulating immune cells including neutrophils and leukocytes to the region to clear the potential pathogen and repair the lesion (Pasparakis et al., 2014). Together, the changes in local chemical environment can be detected by the itch sensory nerve endings causing a temporary itch sensation. The response usually resolves within a day or two.

However, in various forms of dermatitis (inflammation of the skin), repetitive or persistent immune activation causes dramatic elevation of histamine and other mediators in the afflicted skin area, resulting in thickening and morphological changes of the epidermis, massive immune cell infiltration, and chronic itch sensation. Infiltrating CD4+ T cells usually exhibit an inflammatory T helper 2 (T_H2) phenotype, likely induced by local cytokine constituents (Brandt and Sivaprasad, 2011). Common dermatological diseases including atopic dermatitis, psoriasis, irritant or allergic contact dermatitis, and xerosis (dry skin) all bear complex genetic and environmental triggers and are characterized by distinct alterations in skin presentation, histology and chemical content (Nattkemper et al., 2018). On the other hand, incoming blood circulation may carry systemic itch mediators that arise in other diseased organs or prurigenic drugs. For example, patients with kidney failure or liver diseases often suffer from long term pruritus without identifiable primary skin lesions. The molecules that activate itch neurons in these diseases remain largely unknown (Huesmann et al., 2013; Kittaka et al., 2017; Kremer et al., 2010; Usoskin et al., 2015).

Since the focus of this review is the neuronal mechanism of itch, we will restrict our discussion on molecular mediators with well-established neuronal receptors that directly activate itch neurons or enhance their activation. Molecules that cause itch indirectly (for example, through activating mast cells) will not be included here. The involvement of currently known peripheral mediators in various itchy conditions is summarized in Table 1.

Table 1.

Peripheral mediators and neuronal receptors of itch

Molecular mediator	Main cellular source	Neuronal receptor	Ion channel	DRG neuron subtypes	Cause of itch
Histamine	Mast cells	H1R, H4R	TRPV1, TRPV4	NP2, NP3	Insect bites, dermatitis
Serotonin (5-HT)	Mast cells, keratinocytes	HTR7 HTR2	TRPA1 TRPV1, TRPV4	NP3	Atopic dermatitis
Proteases	Mast cells, plants	PAR2 MrgprC11	TRPA1, TRPV1	NP2	Cowhage
TSLP	Keratinocytes	TSLP receptor (IL-7Rα + TSLPR)	TRPA1		Atopic dermatitis
IL-31	Th2 T helper cells	IL-31 receptor (IL-31Rα+ OSMR)	TRPA1, TRPV1	NP3	Atopic dermatitis, T cell Lymphoma,
IL-33	Keratinocytes	IL-33 receptor (IL-1RAcP + ST2)	TRPA1, TRPV1	NP2	Allergic contact dermatitis
IL-4 and IL-13	Th2 cells, ILC2s, basophils	Il-4Rα, IL-13Rα1	TRPA1, TRPV1	NP1, NP2, NP3	Atopic dermatitis, chronic idiopathic pruritus
Poly I:C Imiquimod	Pathogens Drug	TLR3 TLR7		NP2?	Psoriasis Xerosis (Dry skin)
BAM8-22 peptide	Keratinocytes	MrgprC11	TRPA1, TRPV1	NP2	Xerosis (Dry skin)
Chloroquine	Medicine in circuilation	MrgprA3	TRPA1, CNO1	NP2	Drug-induced itch
β-alanine	Medicine in circulation	MrgprD		NP1	Drug-induced itch
Bile acid	Bile	TRA5	TRPA1	NP1	Cholestatic itch
lysophosphatidic acid (LPA)	Bile	LPAR5	TRPA1, TRPV1	NP1	Cholestatic itch

Histamine and histamine receptors

Histamine has long been the “gold standard” itch mediator and has been studied for more than 100 years (Dale and Laidlaw, 1910). When applied to human skin, histamine causes local vasodilation and gives rise to a characteristic “redness, flare and swelling” response (Lewis, 1926) accompanied by an intense itch sensation (Simone et al., 1987). Mast cells contain large amounts of histamine in their granules and are the main sources of histamine in the body, yet histamine can be synthesized by several other cell types including basophils, neurons and keratinocytes. There are currently four known histamine receptors: H1R, H2R, H3R and H4R. Among these receptors, H1R and H4R are expressed in itch-sensing dorsal root ganglia (DRG) neurons and mediate histamine-induced itch (Kashiba et al., 1999; Simons and Simons, 2011; Strakhova et al., 2009; Usoskin et al., 2015). Both H1R and H4R knockout mice scratch less in response to histamine while simultaneous antagonism of H1R and H4R effectively blocked histaminergic itch (Dunford et al., 2007; Roβbach et al., 2009; Sugimoto et al., 2003). H3R expression in DRG neurons has also been detected, but it seems to be an inhibitory receptor and its interaction with H1R and H4R remains inconclusive due to the lack of genetic analyses (Cannon et al., 2007; Rossbach et al., 2011).

All four histamine receptors are G-protein coupled receptors (GPCRs). H1R is coupled to Gαq/11 G protein and when activated, mediates neuronal calcium elevation through phospholipase A2, phospholipase C-β3 (PLCβ3), and protein kinase C δ (PCKδ) (Han et al., 2006; Imamachi et al., 2009a; Kim et al., 2004b; Shim et al., 2007). Neuronal activation requires the transient receptor potential cation channel subfamily V1 (TRPV1) (Imamachi et al., 2009a; Kim et al., 2004a; Shim et al., 2007; Valtcheva et al., 2015a). TRPV1 is a non-selective cation channel whose activation leads to a rise in membrane potential and collective activation of the entire sensory terminal can cause the firing of an action potential. (Caterina, 1997; Imamachi et al., 2009a). Another TRP channel TRPV4 also contributes to histamine-mediated neuronal activation (Kim et al., 2016). Anti-histamine drugs have been widely used to alleviate allergy and itch symptoms but are shown to be ineffective in many itch conditions. Itch that cannot be blocked by anti-histamines are usually referred to as non-histaminergic itch.

Serotonin and serotonin receptors

Serotonin (5-hydroxytryptamine, 5-HT) is an important neurotransmitter in the CNS but can also be released by mast cells independent of histamine (Moon et al., 2014; Theoharides et al., 1982). When applied to human or mouse skin at high doses (>1mM), 5-HT elicited both pain and itch sensation (Akiyama et al., 2010a; Akiyama et al., 2010b; Weisshaar et al., 1997), whereas lower doses of 5-HT (100µM) only causes itch (Morita et al., 2015). There are 14 known 5-HT receptors, divided into 7 classes (Nestler et al., 2008). Several 5-HT receptors have been detected in DRG neurons (Morita et al., 2015; Ohta et al., 2006; Usoskin et al., 2015). Antagonists to HTR 1 and 2 (Nojima and Carstens, 2003; Yamaguchi et al., 1999), and HTR3 (Ostadhadi et al., 2015; Weisshaar et al., 1997) have been shown to reduce 5-HT mediated itch. Genetic knockout of HTR7, however, completely abolished scratching behavior when 5-HT was applied at a low dose and also partially reduced scratching in a mouse model of atopic dermatitis (Morita et al., 2015). HTR3 is itself a ligand-gated ion channel while all other 5-HT receptors are GPCRs (Reeves and Lummis, 2002). HTR7 is Gas coupled and, through adenylyl cyclase (AC) and cyclic AMP (cAMP), activate the TRPA1 cation channel to achieve neuronal activation (Morita et al., 2015). It has also been shown that PLCβ3 and another TRP channel TRPV4 are required for 5-HT mediated itch, likely downstream of another 5-HT receptor (Akiyama et al., 2016; Imamachi et al., 2009b).

Proteases and Mas-related G protein-coupled receptors (Mrgprs)

In addition to histamine and serotonin, mast cells also release a number of proteases including tryptases, chymases and cathepsins (Moon et al., 2014; Wernersson and Pejler, 2014). These endogenous proteases, as well as exogenous proteases such as mucunain from cowhage spicules or papain, can cause strong non-histaminergic itch (Reddy et al., 2008; Reddy and Lerner, 2010). Originally, proteases were proposed to cause itch via protease-activated receptor 2 (PAR2). The PARs are a family of GPCRs with N-terminal domains that, once cleaved by proteases, can auto-activate the GPCR itself (Akiyama et al., 2015a). However, recent genetic experiments have alluded to members of the MRGPR family of GPCRs as the transducers of protease-induced itch (Liu et al., 2011; Reddy et al., 2015). One member of the family, MrgprC11 is activated by cleavage at its N-terminus. Cysteine proteases, cathepsin S and papain can both activate this receptor. In Mrgpr cluster knockout animals in which MrgprC11 is deleted, DRG neuronal activation as well as scratching response triggered by the cysteine proteases were abolished, whereas PAR2 mutant animals were indistinguishable from wild type (Reddy et al., 2015). The PAR2 N-terminal peptide SLIGRL which was previously used as a PAR2 specific agonist, was also found to induce itch behavior by activating MrgprC11 (Liu et al., 2011). It was shown that a shorter peptide SLIGR, which was specific to PAR2, elicited pain but not itch (Liu et al., 2011). Taken together, these observations point to MrgprC11 as the neuronal receptor for protease-induced itch.

The Mrgprs are a large family of GPCRs with 27 members in mice and 8 in humans excluding pseudogenes (Bader et al., 2014; Dong et al., 2001; Han et al., 2002; Meixiong and Dong, 2017; Zylka et al., 2003). Besides cysteine proteases, MrgprC11 is also activated by a proteolytic product of proenkaphalin A, bovine adrenal medulla 8–22 peptide (BAM8-22) (Liu et al., 2009). Although the amount of BAM8-22 in the skin has yet to be quantified, the precursor proenkaphalin A is expressed in keratinocytes and is up-regulated in atopic dermatitis patients (Slominski et al., 2011). Another member of the family, MrgprA3, responds to an anti-malaria drug chloroquine, causing strong itch (Liu et al., 2009). MrgprD of the same family mediates itch induced by β-alanine (Liu et al., 2012a). Activation of Mrgprs in both mice and humans causes strong itchy responses (Sikand et al., 2011a). Mouse models of dry skin and allergic contact dermatitis pruritus show significant overexpression of MrgprA3 and MrgprC11, while Mrgpr cluster knockout animals lacking both receptors exhibit less scratching in these chronic itch conditions (Zhu et al., 2017). Consistently, human MRGPRs were shown to be up-regulated in atopic dermatitis and psoriasis patients (Nattkemper et al., 2018). MrgprA3 and MrgprD are specifically and highly expressed in mutually exclusive DRG neuronal populations and are in fact key defining molecular features of 2 types of non-peptidergic itch sensing neurons (Dong et al., 2001; Liu et al., 2009; Usoskin et al., 2015; Zylka et al., 2005). MrgprC11 and MrgprA3 are expressed in almost completely-overlapping populations (Liu et al., 2009; Usoskin et al., 2015). It was shown that MrgprA3 signals through Gβγ to TRPA1 (Wilson et al., 2011) and possibly TRPC3, TRPV1 and TRPV4 (Kim et al., 2016; Than et al., 2013), while MrgprC11 is coupled to PLCβ and signals via TRPA1 leading to neuronal activation (Wilson et al., 2011). However, this conclusion was recently disputed by electrophysiological recordings showing that MrgprA3-mediated neuronal responses to chloroquine did not require TRP channels, but is instead partially transmitted by a calcium activated chloride channel, ANO1(Ru et al., 2017).

Cytokines and cytokine receptors

Cytokines are a broad category of signaling molecules utilized by immune cells for communication. Multiple forms of dermatitis are strongly associated with T_H2 cytokines. These cytokines induce and maintain CD4+ T helper cells in the inflammatory T_H2 type, triggering further chronic inflammation. It has been recognized in recent years that itch sensory neurons also express cytokine receptors and utilize similar intracellular signaling pathways as immune cells, providing a likely mechanism to coordinate the nervous and immune systems to warn the body about potential damage and infections.

TSLP

Thymic stromal lymphopoietin (TSLP) is an epithelial-derived cytokine and a master initiator of T_H2 type inflammation (Indra, 2013). High levels of TSLP are a signature of atopic dermatitis (Han et al., 2017; Indra, 2013; Li et al., 2006; Soumelis et al., 2002). In the skin, keratinocytes release TSLP in response to a range of stimuli including protease activation of PAR2. Both subunits of the TSLP receptor, IL7Rα and TSLPR, are detected in a small subset of nociceptive neurons that do not overlap with either histamine or chloroquine-responsive neurons. TSLP can trigger itch through direct neuronal activation via the TSLP receptor and requires TRPA1 for its downstream signaling (Wilson et al., 2013).

IL-31

Interleukin-31(IL-31) has also been associated with atopic dermatitis and itch caused by cutaneous T cell lymphoma and is predominantly produced by T_H2 type T cells (Cevikbas et al., 2013; Nattkemper et al., 2016; Nattkemper et al., 2018). IL-31 belongs to the IL-6 family of cytokines and the IL-31 receptor consists of 2 subunits, IL-31Rα and oncostatin M receptor (OSMR). Both subunits are expressed in a group of small diameter itch-sensing neurons that also express receptors for 5-HT (Usoskin et al., 2015). Subcutaneous injection of IL-31 in mice triggered scratching behavior (Cevikbas et al., 2013; Usoskin et al., 2015), whereas human subjects receiving IL-31 administration did not report immediate itch sensation (Indra, 2013). IL-31R signals through mitogen-activated protein kinase (MAPK) pathway and results in phosphorylation of extracellular signal-regulated kinase (ERK). Both TRPV1 and TRPA1 were shown to be critical components of the signaling pathway(Cevikbas et al., 2013).

IL-33

IL-33 is a member of the IL-1 cytokine family that drives multiple immune cells to produce T_H2 cytokines (Brandt and Sivaprasad, 2011; Han et al., 2017). Recently it has been shown that keratinocyte-secreted IL-33 is a major itch mediator in allergic contact dermatitis induced by poison ivy allergens urushiol and oxazolone (Liu et al., 2016). IL-33 receptors consist of a membrane bound IL-33-specific ST2 chain and an accessory IL-1 receptor-like 1 (IL-1RAcP) chain and both are expressed in a population of chloroquine and histamine-sensitive neurons. Direct IL-33 application triggers a mild increase in neuronal Ca²⁺ and does not by itself cause itch in naïve animals. Yet IL-33 and ST2 receptor are necessary for and potently promote chronic itch in urushiol-induced pruritus. Both TRPV1 and TRPA1 are required for IL-33 signaling in neurons (Liu et al., 2016).

IL-4 and IL-13

IL-4 and IL13 are closely related both in their sequences and functions. These cytokines are master initiators of T_H2 immunity as they drive the differentiation of naïve CD4+ helper T cells into T_H2 cells (Brandt and Sivaprasad, 2011). T_H2 cells themselves then produce large amounts of IL-4 and IL-13 to maintain the T_H2 phenotype. IL-4 also promotes B cell class-switch to produce IgE antibody, leading to IgE-dependent allergy. IL-4 and IL-13 receptors were found in multiple types of itch sensing neurons that overlap with histamine, chloroquine, TSLP and IL-31 responsive neurons (Chiu et al., 2014; Oetjen et al., 2017; Usoskin et al., 2015). When activated, these receptors signal through the Janus kinase (JAK) pathway as well as both TRPV1 and TRPA1, causing elevation in neuronal Ca²⁺. IL-4 and IL-13 injection in the skin do not trigger acute scratching behavior. However the IL-4/13 –> IL-4R/13R –> JAK signaling axis is required for the development of chronic itch in both mouse models of atopic dermatitis and in human patients with chronic idiopathic itch, likely through potentiating neuronal response to other itch mediators (Oetjen et al., 2017).

Toll-like receptors

Another family of classical immune receptors involved in itch are the toll-like receptors (TLRs). Originally cloned in Drosophila, these large leucine-rich repeat containing proteins were later recognized as key pattern recognition receptors of the innate immune system (Janeway and Medzhitov, 2002; Medzhitov, 2001). Epithelial and immune cells utilize these receptors to recognize pathogen-associated molecular patterns (PAMPs) such as components of the bacteria cell walls. Of the 14 TLRs in mice, TLR1/2, TLR2/6, TLR4, TLR5 and TLR11 are known to localize on the plasma membrane for detection of extracellular PAMPs and TLR3, 7, 8 and 9 reside on the membranes of intracellular vesicles for detection of intracellular PAMPs. Surprisingly, TLR3 and TLR7, both of the latter group, have been shown to be expressed in itch-sensing DRG neurons. Tlr3−/− and tlr7−/− mutant animals exhibited severely compromised response to itchy stimuli and showed reduced scratching in a dry skin chronic itch model (Liu et al., 2012b; Liu et al., 2010a). Application of TLR3 ligand poly I:C (PIC) and TLR7 ligand imiquimod evoked action potentials in the neurons. It was thus hypothesized that unlike in other cell types, TLR3 and TLR7 might function on the plasma membrane to directly regulate neuronal excitability. Another group, however, reported that imiquimod could elicit itch responses in both wild type and tlr7−/− animals, suggesting that TLR7 is not directly involved in itch sensation but the drug mediates itch via other TLR7-independent mechanisms (Kim et al., 2011). The precise mechanisms by which TLRs mediate itch remain unclear. Whether itch is evoked by these receptors in the primary sensory neuron, or by peripheral non-neuronal cells or central itch processing circuits will need to be dissected carefully using cell-type specific genetic approaches.

In summary, itch is initiated in the periphery when itch-sensing neurons expressing specific membrane receptors detect their corresponding ligands in the dermatome they innervate. Many itch receptors are GPCRs and signal through diverse G protein signaling pathways. The TRP family of ion channels, most importantly TRPV1 and TRPA1 serve as key signal transduction components downstream of the GPCRs, though the specific coupling of each GPCR and TRP channels remain under debate (Geppetti et al., 2015; Sun and Dong, 2016). Cytokines traditionally thought to be components of the immune system also play important roles in itch sensory signaling (Figure 2). We will start the next section on itch sensing neural circuits by first reviewing the expression of neuronal receptors for itch mediators in distinct populations of primary sensory neurons.

Figure 2. Neuronal itch receptor activation and downstream signaling.

Several known itch mediators bind to G protein coupled receptors (GPCRs) on the neuronal membrane. Each of the GPCRs signal through their coupled G protein pathways but all require the TRP family cation channels. Cytokine receptors, on the other hand, signal through classical Janus kinase and MAP kinase pathways but also require the TRP ion channels to trigger itch neuronal firing.

II. NEURAL CIRCUITS OF ITCH

Primary sensory neurons of itch

Together with neurons that mediate other somatosensory modalities, the cell bodies of itch sensory neurons are located in the DRGs and trigeminal ganglia (Figure 3). The cell body sends out one pseudounipolar axon which, upon exiting the DRG, bifurcates and extends towards both the peripheral dermatome that it innervates and the spinal cord for central transmission. DRG neurons can be categorized by the size of their cell body, degree of myelination, electrophysiological properties, peripheral innervation of sensory organs, central projection in the spinal cord, and gene expression profiles (Abraira and Ginty, 2013; Basbaum et al., 2009; Lallemend and Ernfors, 2012; Zimmerman et al., 2014). There are four widely accepted categories: large diameter, thickly myelinated proprioceptive neurons; large diameter, myelinated Aβ low threshold mechanoreceptors that mediate touch; medium-sized, lightly myelinated Aδ nociceptive neurons and small diameter, unmyelinated C nociceptive neurons that detect noxious stimuli including pain, itch and temperature (Basbaum et al., 2009). Electrophysiological recordings revealed that most itch neurons belong to C-type neurons and a small population are Aδ neurons (LaMotte et al., 2014a; Ringkamp et al., 2011).

Figure 3. Hypothetical schematic of peripheral and central itch circuits based on current knowledge.

Primary somatosensory neurons in the DRG can be categorized by their gene expression profiles. Single cell expression profiling revealed 3 subtypes of itch neurons termed NP1, NP2 and NP3. NP1 neurons are characterized by their expression of MrgprD, NP2 by MrgprA3 and NP3 by serotonin receptors. The MrgprA3+ NP2 neurons have been shown to be itch specific neurons since activating these neurons elicit only itch behaviors. Itch sensory neurons make synaptic connections with GRP+ interneurons in the spinal cord, which are in turn connected with GRPR+ neurons. Ablation of GRPR signaling or GRPR+ neurons will completely abolish response to itchy stimuli, indicating vital importance of this pathway in itch transmission. Information about itch stimuli in the periphery is eventually conveyed to higher brain centers by projection neurons. The parabrachial nucleus (PBN) in the brainstem serves as an important relay center in the process.

Itch interacts with other sensory modalities in the spinal cord. GRP+ neurons receive both itch and pain stimuli but will inhibit their own responses to high intensity pain sensation through enkaphalin signaling. Meanwhile, a painful stimulus will activate both itch and pain sensory neurons. But only pain will be perceived because itch is inhibited by pain via inhibitory interneurons in the spinal cord. One group of these interneurons is marked by the transcription factor BHLHB5. These neurons are activated by a variety of anti-itch stimuli and likely inhibit itch via the release of dynorphin and glycine. Similarly, a group of inhibitory interneurons labeled by the NPY neuropeptide were shown to mediate the inhibition of mechanical itch by light touch, though the sensory neurons and further transduction circuits of mechanical itch remain to be identified.

The spinal cord dorsal horn also receives descending neuromodulation from higher brain centers. The raphe magnus (NRM) sends serotoninergic projections to the spinal cord and can stimulate GRPR+ neurons and potentiate itch by co-activating HTR1A and GRPR receptors. The dorsal horn also receives tonic noradrenergic inhibition, likely from the locus coeruleus (LC). GRPR+ neurons in the suprachiasmatic nucleus (SCN) of the hypothalamus have been suggested to play a role in contagious itch.

In recent years, several groups have conducted transcriptome profiling of DRG neurons on both single-cell and population levels using next generation RNA sequencing (RNAseq) (Chiu et al., 2014; Goswami et al., 2014; Li et al., 2017; Reynders and Moqrich, 2015; Thakur et al., 2014; Usoskin et al., 2015). Despite minor discrepancies on the expression and specificity of certain genes, the RNAseq analyses are generally in agreement and the assignment of neuronal subtypes is consistent with traditional criteria based on cell size and electrophysiological properties. For the purpose of this review, we have chosen to adopt the categories and nomenclature of an unbiased single cell RNAseq study (Usoskin et al., 2015). In this study, principal component analysis of single cell transcriptomes determined that mouse DRG neurons were best categorized by 11 subgroups: NF1-3 correspond to low threshold Aβ mechanoreceptors; NF4 and 5 are proprioceptors; TH correspond to C type low threshold mechanoreceptor that mediate mechanical pain and pleasant touch; PEP1 (peptidergic) are thermosensitive nociceptors; PEP2 are lightly myelinated Aδ nociceptors. Three types of non-peptidergic nociceptors NP1, 2 and 3 are likely itch sensory neurons. NP1 correspond to MrgprD+ neurons. NP2 and 3 are partially overlapping populations. Both groups express histamine receptors and IL-33 receptor, with NP2 being characterized by the expression of MrgprA3 and MrgprC11, while serotonin receptors, IL-31 receptor and leukotriene receptors are specific to NP3. IL-4 and IL-13 receptors are detected in all three types of neurons. TRPA1 and TRPV1 channels required for downstream signaling of the receptors are consistently present in the neurons. The gene expression profiles of DRG neurons are extremely useful in assigning physiological functions of neuron types: Receptors expression is a good predictor of ligand specificity while ion channels and transmitter synthesis/transport machineries indicate the mode of conductance and synaptic transmission in the spinal cord.

Coding of itch and pain in the DRG

The neural mechanisms that distinguish itch from pain have long been the focus of itch research. Both itch and pain are detected by small diameter nociceptive neurons in the DRG and it was once hypothesized that itch is simply a low intensity form of pain and is not encoded by distinguishable neuronal populations (Bishop, 1948; Lewis and Zotterman, 1927; von Frey, 1922). In fact, all itch neurons described in the previous section express TRPV1 and can be activated by the TRPV1 ligand capsaicin, which is a classical pain stimulus (Caterina et al., 1997). Capsaicin and histamine, the golden standard pain and itch agents, elicited very similar sensations in human subjects when applied at a low dose in a punctate manner (Sikand et al., 2011b). However, genetic and functional analyses in the recent years support the existence of itch specific neurons in the DRGs. When applied on the cheek of mice, itch mediators trigger a characteristic scratching behavior in which the animals use the hind limb on the same side to scratch, whereas pain agents cause wiping with forelimbs. The two sensations can thus be clearly distinguished in a behavioral readout. Inhibiting histaminergic and non-histaminergic itch neurons using an activity-dependent sodium channel blocker QX-314 did not affect the animals’ wiping responses to capsaicin, suggesting that the two senses are coded by distinct neuronal populations (Roberson et al., 2013). MrgprA3+ neurons, which probably include all NP2 and a small population of NP3 neurons, were shown to be itch-specific sensory neurons. Genetic ablation of MrgprA3+ neurons drastically reduced scratching behavior triggered by various itch mediators as well as dry skin or allergic contact dermatitis pruritus, while capsaicin-induced wiping was not affected in these animals. More importantly, when MrgprA3+ neurons were selectively activated, only scratching but not wiping behaviors were triggered. This result strongly suggest that MrgprA3+ neurons carries information about itch but not pain (Han et al., 2012). Thus in the PNS, itch is at least partially represented by a “labeled line” that can be distinguished from pain. MrgprD+ NP1 neurons, on the other hand, are likely polymodal and respond to itch, pain, mechanical and thermo stimuli (Wang and Zylka, 2009; Zylka et al., 2005). Interactions between pain and itch in the CNS will be further discussed in the following sections.

Itch processing circuits in the spinal cord

Transmission of itch signal from DRG to spinal neurons

Nociceptive neurons, including itch sensitive neurons, are excitatory glutamatergic neurons (Brumovsky et al., 2007; Liu et al., 2010b) that express vesicular glutamate transporter type 2 (VGLUT2). Interestingly, genetic knockout of VGLUT2 from all nociceptive neurons abolished pain behaviors while enhancing itch (Liu et al., 2010b). Thus itch neurons do no solely rely on glutamate for synaptic transmission but have instead been proposed to communicate via neuropeptides. Genetic knockout of natriuretic polypeptide b (NPPB, also called B-type natriuretic peptide, BNP) strongly reduced scratching behavior induced by histamine, chloroquine, 5-HT and the SLIGRL peptide, suggesting that it is required for the transmission of NP2 and NP3 signals (Mishra and Hoon, 2013). Recent RNAseq analyses, however, suggest that the expression of NPPB is restricted to NP3 (Li et al., 2017; Usoskin et al., 2015), so NP2 might utilize other transmission mechanisms. In a simplified model, the central projection of MrgprA3+ neurons terminate in lamina II of the spinal cord dorsal horn and make mono-synaptic connections with a group of interneurons that express another neuropeptide gastrin-releasing peptide (GRP). Most GRP+ neurons express NPPB receptor NPRA, so NP3 neurons likely signal to GRP+ neurons via NPPB-NPRA and NP2 to GRP+ neurons via another transmitter (Mishra and Hoon, 2013; Sun et al., 2017). GRP+ neurons then relay the signal to another group of interneurons in lamina I that express GRP receptor GRPR (Mishra and Hoon, 2013). It has also been proposed that primary itch neuron may signal to GRPR+ neurons directly via GRP release (Nattkemper et al., 2013; Sun and Chen, 2007a; Sun et al., 2009a). The debate centers on the expression of GRP in DRG neurons or spinal cord interneurons. At the transcription level, GRP mRNA in the DRG could not be detected in RNAseq experiments or GRP-cre animals (Li et al., 2017; Mishra and Hoon, 2013; Solorzano et al., 2015; Sun et al., 2017; Usoskin et al., 2015), yet the detection of GRP protein using antibodies yielded conflicting results (Devin et al., 2016; Liu et al., 2014; Mishra and Hoon, 2013; Nattkemper et al., 2013; Solorzano et al., 2015). Either activated directly or secondarily via other interneurons, the GRPR+ neurons appear to serve as a cardinal hub for itch signaling since the genetic deletion of GRPR and ablation of GRPR+ neurons completely abolished scratching behavior mediated by histamine, 5-HT, Chloroquine and SLIGRL without affecting pain, thermo or mechano-sensations (Sun and Chen, 2007a; Sun et al., 2009a). These results demonstrate the necessity of the GRPR+ neurons for the transmission of itch signals in the spinal cord but do not conclusively show that this circuit is dedicated to itch. The behavior consequences of selective activation of NPRA+ and GRPR+ neurons using genetic approaches remains to be elucidated.

Interactions between itch and pain in the spinal cord

Although itch-specific primary sensory neurons have been recognized to specifically convey itch stimuli, these neurons also express channels typically associated with pain. For example, the TRPV1 ligand capsaicin can activate all itch neurons and also a large population of non-itch-sensing pain neurons. Yet capsaicin exclusively elicits pain and not itch. It has long been recognized that simultaneous perception of pain will inhibit itch. Scratching itself is thought to temporarily alleviate itch through activation of pain neurons (Braz et al., 2014). It was also observed that conditional knockout of VGLUT2 in all nociceptors will abolish pain but significantly elevate itch responses (Aresh et al., 2017; Liu et al., 2010b). This is likely because pain transmission requires glutamate, and the elimination of pain cause excessive itch by releasing its inhibition by pain. Consistent with this gate control theory, it was shown BHLHB5+ inhibitory interneurons can be activated by a variety of counter-itch stimuli including pain and menthol, and release kappa opioid dynorphin to inhibit GRPR+ or downstream neurons of the itch pathway (Kardon et al., 2014; Ross et al., 2010). Loss of BHLHB5+ neurons causes excessive scratching phenotype similar to that of nociceptive VGLUT2 knockout (Ross et al., 2010). It was recently shown that activating inhibitory glycinergic interneurons (GLYT2+ neurons) also cause a reduction in itch behavior (Foster et al., 2015). These neurons receive direct synaptic input predominantly from myelinated low threshold mechanoreceptors in the DRG and thus represent inhibition of itch by touch. Given that BHLHB5+ neurons are not a homogenous population, it is possible that a subpopulation of GABAergic BHLHB5+ neurons inhibit itch through kappa opioid signaling while the others exert inhibition through glycine (Foster et al., 2015; Kardon et al., 2014).

Interestingly, the GRP+ interneurons that relay input from itch specific MrgprA3+ neurons to GRPR+ neurons also receive synaptic input from DRG neurons that sense pain (Sun et al., 2017). While GRP+ neuron activation by itch mediators is linearly dose-dependent, response to painful stimuli at high concentration triggers the release of delta opioid enkaphalin from local enkaphalin neurons which in turn inhibit the GRP+ neurons themselves. This differential negative feedback provides a “leaky gate” mechanism as an amendment to the interneuron mediated gate control model and explains a human psychophysical observation that itch sensation is often accompanied by a slight nociceptive perception (LaMotte et al., 2014b; Sun et al., 2017). Overall, an itchy stimuli will activate itch-specific sensory neurons, whereas a painful stimuli will activate both pain and itch primary sensory neurons but the itch circuit will be inhibited in the spinal cord, leaving only the sensation of pain.

Mechanical itch circuit in the spinal cord

In addition to chemical itch triggered by the binding of various mediators on the terminals of itch sensory neurons, there is also mechanical itch which is elicited by light tactile stimuli (Akiyama et al., 2012a; Bourane et al., 2015; Fukuoka et al., 2013). A population of neuropeptide Y (NPY) positive inhibitory interneurons in the spinal cord was shown to inhibit mechanical itch. When NPY neurons were ablated or silenced, mice exhibit excessive scratching behavior in response to normally non-itch-mediating low force stimuli. This pathway appeared to be independent of known chemical itch processing pathways since deletion of neither GRPR+ neurons nor Bhlhb5 affected mechanical itch. NPY ablation itself did not modify chemical itch behavior either (Kardon et al., 2014). Thus the two pathways, mechanical itch and chemical itch, likely act in parallel. It was shown that NPY+ neurons received synaptic input from low threshold mechanoreceptors that innervates hairy skin. The primary sensory and spinal neurons that positively mediate mechanical itch remain unknown (Bourane et al., 2015). The NPY neurons are also glycinergic. Thus inhibitory glycinergic interneurons receiving input from touch receptor neurons inhibit both chemical and mechanical itch through genetically distinct neuronal sub-populations.

Projection neurons

The itch-specific GRPR+ neurons in lamina I of spinal cord are excitatory interneurons that require a sodium channel VGLUT2 for neurotransmission (Aresh et al., 2017; Lee et al., 2014). Classical neuroanatomy outlines that nociceptive information in the superficial spinal cord laminae is conveyed to the brain by projection neurons whose cell bodies are located in lamina I. The axons of projection neurons are thought to cross the midline to the contralateral side, and join the spinothalamic tract (STT) that project to the thalamus and the spinoparabrachial tract that project to the brainstem (Basbaum et al., 2009; Cameron et al., 2015; Ikoma et al., 2011). Recording of STT neurons in cat (Andrew and Craig, 2001), primates (Davidson et al., 2009; Davidson et al., 2007), rat (Carstens et al., 2010; Moser and Giesler, 2014) and mice (Akiyama et al., 2015b) revealed populations of projection neurons that respond to itch stimuli including histamine, 5-HT, BAM8-22, chloroquine and cowhage.

In rats and mice, many of these projection neurons express the substance P receptor NK1R (Akiyama et al., 2015b; Al-Khater and Todd, 2009; Carstens et al., 2010). Ablation of NK1R+ neurons reduced chloroquine-evoked acute itch as well as spontaneous scratching and mechanically evoked scratching in a mouse model of atopic dermatitis. Retrograde tracer injected in the thalamus and parabrachial nucleus (PBN) in the brainstem overlapped with NK1R+ neurons, showing that some of these neurons do indeed project to the brain (Akiyama et al., 2015b). However, itch-sensitive projection neurons appeared to be polymodal and responded to pain and thermo stimuli as well (Davidson et al., 2012; Davidson et al., 2009; Hachisuka et al., 2016). It is unclear how itch information is conveyed by the projection neurons or how it is represented in the brain. It is possible that the projection neurons are not a homogeneous population and requires further genetic and functional dissection. Or it could be that, similar to retinal ganglion cells which perform analogous central projection function in the visual system, somatosensory projection neurons carry processed and integrated information (“features” instead of primary sensory information) that will be decoded in higher brain centers.

Itch circuits in the brain

Itch processing in the PBN

Itch circuits in the brain remains largely unexplored. Spinal projection neurons project to both the thalamus and the PBN in the brainstem (Akiyama et al., 2015b; Cameron et al., 2015; Davidson et al., 2012; Hachisuka et al., 2016). It was recently discovered that PBN serves as a key itch processing nucleus. A group of PBN neurons were shown to be bi-synaptically connected with the GRPR+ neurons, likely through projection neurons. These PBN neurons show a strong response to histamine and chloroquine, and conditional knockout of the VGLUT2 glutamate transporter significantly reduced both acute scratching response to histamine and chronic itch triggered by allergic contact dermatitis, without affecting the responses to mechanical and thermo stimuli (Mu et al., 2017). Whether itch-sensing PBN neurons can be genetically distinguished from other PBN neurons, the identities of spinal neurons that directly synapse onto these neurons, and the projection target into higher brain regions remain to be elucidated.

Itch representation in the thalamus, somatosensory cortex and other brain regions

The thalamus is traditionally viewed as a relay station for all the senses except olfaction. Vision, taste, audition and somatosensation are projected onto topologically organized thalamic nuclei and subsequently relayed to corresponding sensory cortices. Itch sensing STT neurons project contralaterally to the ventrobasal and posterior thalamus (Akiyama et al., 2015b; Akiyama et al., 2012b; Davidson et al., 2012; Davidson et al., 2009; Davidson et al., 2007; Moser and Giesler, 2014). Functional magnetic resonance imaging in human subjects experiencing itchy stimuli revealed activation of the primary somatosensory cortical regions (S1) topographically corresponding to the stimuli, indicating that S1 may represent the location of itch. In addition, itch stimuli elicit a combination of characteristic activities in brain regions including the secondary somatosensory cortex (S2); areas involved in emotional processing and reward (cingulate and prefrontal cortex, amygdala and the limbic systems); areas for evaluation and self-awareness (insular cortex); premotor, motor and supplementary motor areas (likely involved in the initiation of scratching behavior) and the pattern is significantly altered in patients with chronic itch disorders (Ishiuji et al., 2009; Schneider et al., 2008).. The periaqueductal gray (PAG) is activated during scratching, likely correlated with descending itch suppression by the opioid enkaphalin (Herde et al., 2007; Kwan et al., 2000; Mochizuki et al., 2009; Papoiu et al., 2013; Vierow et al., 2009; Yosipovitch et al., 2007; Yosipovitch et al., 2008). How itch is represented in the brain in relation to pain, touch and other senses, and how these circuits are established during development and altered in diseases remain big questions to be understood.

Descending modulations of itch

Electrical responses of STT neurons to itchy stimuli can be suppressed by scratching and other counter-itch stimuli via GABAergic and Glycinerigc mechanisms (Akiyama et al., 2011). Itch processing neurons in the spinal cord also receive descending neuromodulations from higher brain centers. Serotoninergic neurons in the brainstem nucleus raphe magnus (NRM) sends descending projections to the spinal cord and can stimulate GRPR+ neurons and potentiate itch by co-activating HTR1A and GRPR receptors. Inhibition of 5-HT signaling in the spinal cord effectively reduced dry skin-mediated chronic itch (Zhao et al., 2014). Since scratching and painful stimuli were shown to increase spinal 5-HT from the CNS, this mechanism provides an explanation to why chronic itch is often exacerbated by scratching despite being temporarily relieved due to direct inhibition by pain. In addition, it has been shown that the spinal cord dorsal horn also receives tonic noradrenergic inhibition, likely from the locus coeruleus (Gotoh et al., 2011a; Gotoh et al., 2011b). mRNAs of α-adrenergic receptors have been detected in the dorsal horn and antagonizing these receptors dramatically enhanced behavior responses to 5-HT or mosquito-elicited itch, likely through reducing the activities of inhibitory neurons (Figure 3).

Contagious itch

Contagious itch is a curious phenomena in which viewing others scratch causes itch sensation in the viewers themselves (Papoiu et al., 2011). Human fMRI studies have shown that observing scratching behaviors can activate several itch-related brain centers including the thalamus, primary somatosensory cortex, premotor cortex and insula, representing a case of psychogenic itch in the absence of real cutaneous stimuli (Holle et al., 2012). A surprising finding was recently published stating that mice also exhibit contagious itch behavior and that the scratching behavior in response to viewing is mediated by the suprachiasmatic nucleus of the hypothalamus (Yu et al., 2017). It was shown that SCN neurons were robustly activated during contagious itch and mediate scratching behavior through local GRP-GRPR signaling. Selective ablation of GRP or GRPR activity in the SCN eliminated contagious itch behavior, while optogenetic activation of SCN GRP+ neurons triggered scratching (Yu et al., 2017). The robustness of the behavior paradigm is under debate (Barry et al., 2017; Liljencrantz et al., 2017) and it remains to be explained how the SCN conveys visual or social information about scratching. The SCN was known to control circadian rhythm through direct input from photo-sensitive retinal ganglion cells instead of receiving information from visual cortices (Berson et al., 2002). Given the importance of GRP and GRPR in globally mediating itch and scratching behavior, it is important to confirm that the effect is indeed restricted to the SCN but does not require other GRPR neurons including those that mediate itch sensation in the spinal cord.

Conclusion and remarks

In summary, itch perception and processing relies on dedicated sensory neural circuits. In the periphery, MrgprA3+ itch-specific DRG neurons convey sensory information to lamina II of the spinal cord, where the information is relayed by GRP+ and GRPR+ interneurons and integrated with other sensory modalities, pain and touch. GRPR+ neurons then synapse onto projection neurons whose axons connect with itch-sensing neurons in the PBN. Neuronal mechanisms mediating descending modulation of itch from the CNS and circuits responsible for mechanical itch and contagious itch have also been partially elucidated (Figure 3). Despite exciting progress, several key questions still remain to be answered.

First, the genetic and environmental causes of multiple chronic itch conditions in humans remain elusive. Mouse genetics has been enormously useful in dissecting both the molecular and circuit mechanisms of itch. Yet mice and humans bear many functional distinctions in immunological, dermatological and neurological perspectives. For example, γδ T cells are pivotal in mouse epidermis but are rare in normal human skin (Laggner et al., 2011; Pasparakis et al., 2014). Great caution shall be taken when drawing conclusions solely based on Cre-dependent markers as many Cre lines give rise to wider and less specific expression than the genes and cell types being studied. The cellular segregation of histamine and β-alanine mediated itch (which correspond to NP2,3 vs NP1 neurons in mice) in primate DRGs suggests that the neuronal sub-populations identified in mice may hold true (Wooten et al., 2014). Besides changes in immunological factors and concentrations of itch mediators, morphological and gene expression alteration in the itch neurons themselves may be of central importance. Mouse models of dry skin and allergic contact dermatitis revealed higher itch fiber density in the epidermis (Valtcheva et al., 2015b) and enhanced expression of itch receptors in DRGs and spinal cord, likely contributing to the diseased animals’ significantly increased response to itchy stimuli (Akiyama et al., 2010a; Zhao et al., 2013). Recent application of genomic, transcriptomic and proteomic technologies on lesioned, pruritic and normal skins from human patients are starting to reveal some correlative and causative factors (Nattkemper et al., 2018).

Second, it is unclear how itch information is conveyed by the spinal projection neurons and how itch is decoded from other nociceptive senses in the brain. The identity of projection neurons connecting GRPR+ interneurons and itch-sensing neurons in the PBN is likely key to the question. The cellular and molecular nature of peripheral and spinal neurons that mediate mechanical itch are also unknown. And it is curious whether these mechanical and chemical itch are represented differently in the brain since in everyday experience both senses are perceived and conceptualized as “itch” and both elicit scratching response. How itch and all other somatosensory features are represented in the somatosensory cortex and other higher brain centers are also open questions. In theory, itch sensation should be topographically represented in the thalamus and primary sensory cortex, reflecting specific dermatomes where the stimuli were initiated.

Last but not least, the knowledge accumulated from skin immunology and neurophysiology can likely be extended to the mucosal system, which is the other barrier surface heavily involved in host defense and chemical sensation. It has recently been shown that MrgprC11, in addition to mediating itch in the skin, is also detected in a small population of jugular sensory neurons of the vagal ganglia. These neurons innervate the airway and, when stimulated by the MrgprC11 ligand BAM8-22, mediate cholinergic bronchoconstriction and airway hyper-responsiveness (Han et al., 2018). It is tempting to speculate that similar perceptions on mucosal surfaces (ocular pruritus that drives eye-rubbing, and airway itch induced sniffing and coughing), are mediated by similar molecular and cellular mechanisms and also come in both histaminergic and non-histaminergic forms (LaVinka and Dong, 2013). Such knowledge may lead to effective therapies for allergic and inflammatory diseases that inflict the mucosal surface, which are sometimes even more debilitating than cutaneous itch.