Neuronal Regulation of Cutaneous Immunity

The skin serves as a physical and immunological interface between the host and environment. Emerging work has shifted our view of the skin as an inert mechanical barrier to a dynamic immunological organ capable of integrating microbial, mechanical, thermal, and chemical cues into uniquely tailored immune responses. Barrier disruption or dysregulation of the cutaneous immune system leads to skin pathologies such as infection, allergy, inflammation, or cancer. Therefore, understanding the immunological mechanisms at the skin interface and how cellular coordination is both attained and lost will enhance our understanding of various cutaneous diseases and lead to the rational design of novel therapies.

Recent studies have demonstrated that the nervous system and immune system have a synergistic relationship. They both communicate via synaptic structures and share many molecules involved with intra- and intercellular communication (1, 2). The immune system communicates with neurons via soluble factors (i.e., cytokines) that activate neurons to elicit sensations such as itch and pain and corresponding behavioral responses (3–6). Alternatively, the nervous system, in addition to conveying sensory information centrally, can secrete neuropeptides in innervated peripheral tissues (i.e., skin) to coordinate local immune responses (7). Whereas the immune system operates on a time scale of hours to days, the nervous system is capable of transmitting information on a millisecond time scale. Our understanding of the synchronization between neurons and immune cells has begun to provide clarity to the mechanisms that regulate barrier tissue immunity. In this review, we will highlight recent advances in our understanding of cutaneous neuroimmunology. We will focus on murine studies addressing the role of sensory neurons in cutaneous inflammation and host defense, with particular emphasis on the ability of neurons to modulate the function of immune cells within the skin.

Skin is densely innervated by functionally and anatomically distinct neurons of the peripheral nervous system (8). Within the peripheral nervous system are somatosensory afferents that interact with the immune system during inflammation and host defense. In the skin, sensory afferents mediate sensations such as pain, itch, touch, and proprioception. Of particular interest are the specialized C-fiber afferents such as nociceptors or pruritoceptors, which sense noxious or itch-inducing stimuli, respectively (5, 9, 10). Most work on neuroimmune interactions in the skin have focused on this subset of cutaneous neurons based on their ability to communicate with immune cells through secreted factors such as neuropeptides. These fibers innervate both the dermis and epidermis and are therefore well located to “sense” environmental stimuli (11).

C-fibers can be broadly classified as either peptidergic—implying potential for neuropeptide release—or nonpeptidergic. Traditionally, distinguishing between neuronal subsets has been done by comparing cell physiology with physical characteristics and immunofluorescent analysis of hallmark neuronal cell surface markers. Using these techniques, it has been shown that markers enriched in the peptidergic population include TRPV1, CGRP, GFRa3, and substance P (12–14). TRPV1 is a well characterized receptor for capsaicin as well as other noxious agents and is commonly used to identify and manipulate peptidergic neurons (15). TRPV1-expressing neurons communicate noxious and heat sensations that are interpreted as painful (15, 16). Markers for nonpeptidergic neurons include MrgprD, GFRa2, and IB4, and their activation results in the sensation of both itch and pain. There is, however, considerable phenotypic and functional overlap in these two broad and somewhat ill-defined categories (17–26). Identifying neuronal subsets using a small number of surface markers is complicated by a significant overlap of markers across neuronal subtypes including expression of TRPV1 by some nonpeptidergic neurons (27–29). This is especially important to consider when using Cre lines that drive expression using a subset-defining marker, which is further complicated by the potential for transient expression of Cre during ontogeny (18). Recently, work using single-cell RNA sequencing of dorsal root ganglion neuron cell bodies has begun to integrate functional and phenotypic information to more effectively segregate sensory neuron subtypes (28, 29). In these studies, neurons identified as peptidergic can be further subdivided into additional subsets. Additional subsets associated can be defined based on expression of markers such as Nppb, MrgprA3, and MrgprD. Notably, the Nppb-expressing subset that is associated with itch also expresses TRPV1, further demonstrating the functional and phenotypic overlap between sensory neuron subsets and potentially explaining why TRPV1 is important for both pain and itch (21, 28). Despite the diversity of cutaneous innervation, there are few publications demonstrating that TRPV1-negative neurons modulate skin immunity. To date, the field has primarily focused either on C-fibers as a whole or on TRPV1-expressing neurons.

A variety of mouse models, reagents, and technologies are available for studying neuronal regulation of immunity in mouse skin in vivo. The Nav1.8-cre mouse line is a common approach used to determine the requirement of neurons in loss-of-function models (Table I). Nav1.8 is a sodium ion channel subtype primarily expressed by sensory neuron C-fibers of the dorsal root ganglion (30). In Nav1.8-cre DTA mice, in which the diphtheria toxin fragment A is constitutively active in Nav1.8-expressing cells, the immune responses can be evaluated in the constitutive absence of these fibers (3). This approach has been quite useful but is limited because of the constitutive ablation of neurons and the possibility of compensatory effects and the lack of discrimination between different subsets of cutaneous afferents. An alternative approach uses TRPV1-cre to drive expression of the primate diphtheria toxin receptor (DTR). TRPV1-DTR mice allow for the inducible ablation of TRPV1-expressing neurons upon administration of diphtheria toxin (31, 32). Although less broadly expressed by peripheral sensory neurons than Nav1.8, transient expression of TRPV1 during neuronal development results in depletion of some TRPV1-negative neuron subsets (18). TRPV1-expressing neurons can also be pharmacologically ablated by administering RTX—a potent functional capsaicin analogue (33). Unlike TRPV1-DTR, RTX directly targets the TRPV1 receptor, which allows for specific depletion of TRPV1-expressing neurons (34). Notably, there are several reports that nonneuronal cell types such as keratinocytes and T cell express TRPV1 (35, 36). For reasons that are unclear, TRPV1-cre drives expression of Cre exclusively in neurons; expression in these other cell types is not observed (31, 37) Until recently, gain-of-function approaches were less commonly used to study cutaneous neuroimmunology. As a mainstay technique in the field of neuroscience, optogenetics enables in vivo neuronal activation with high temporal and spatial precision (38). Neurons are genetically engineered to express light-sensitive proteins known as channelrhodopsins (Chr2) within their cell membrane. Upon illumination with pulsed laser light of a specified wavelength, rhodopsin channels undergo conformational change leading to transient ion flux and neuronal activation. This approach allows for the selective activation of specific types of neurons at a controllable rate firing and allows for location specific neuron activation in vivo without physical manipulation or damage to tissue.

Cutaneous nerve endings innervate the dermis and epidermis of the skin and thus are ideally positioned to “sense” microbes. Recognition of microbial pathogen-associated molecular patterns and their secreted products by nerve endings can initiate an action potential that travels centrally toward the DRG and spinal cord (39–42). Cutaneous neurons express membrane bound PRRs including TLR3, TLR4, TLR7, and TLR9 allowing them to respond to various microbial ligands (40–44). For example, activation of neuronal TLR4 by bacterial LPS is sufficient to elicit persistent pain (45). In an elegant study using a Staphylococcus aureus infection model, it was demonstrated that intraplantar administration of S. aureus activates Nav 1.8-lineage neurons via N-formylated peptides and α-hemolysin pore-forming toxins. This results in painful stimuli often associated with S. aureus infection (3). Ablation of sensory afferent neurons using Nav-1.8 DTA mice abrogated pain-associated behavior in this model. Similarly, Streptococcus pyogenes secretes streptolysin S, which directly activates TRPV1+ neurons, resulting in infection-associated pain (32). Ablation of TRPV1+ neurons using RTX administration or TRPV1-cre DTR mice diminished bacterial detection and significantly abrogated infection-associated pain responses. In other studies, it was demonstrated in vitro that TRPV1+ neurons directly recognize the fungal pathogen Candida albicans (46). Using mice deficient in TRPV1, TRPA1, or P2RX3, it was later shown that β glucan activation of sensory neurons is mediated via the Dectin-1–ATP–P2RX3/P2RX2/3 and Dectin-1–phospholipase C–TRPV1/TRPA1 axis (47). In contrast to the aforementioned pathogens, Mycobacterium ulcerans induces cutaneous hypoalgesia via the secretion of the toxin mycolactone that impairs nociceptor function and triggers neuronal death (48). Together these studies demonstrate that neurons use distinct mechanisms to recognize pathogens, some of which are shared with the innate immune system and some which are unique.

The relationship between cutaneous sensory afferents and the development of in vivo immune responses has been explored using murine skin infections in TRPV1 neuron loss-of-function models. Ablation of cutaneous TRPV1+ neurons using RTX results in a defective innate immune response to epicutaneous C. albicans infection, as evidenced by fewer recruited neutrophils and higher CFU at sites of infection (46) (Fig. 1). Local responses to C. albicans are driven by IL-23 from dermal cDC2 and IL-17 from dermal γδ T cells. Notably, neuron cell bodies isolated from dorsal root ganglia release the neuropeptide CGRP when exposed to C. albicans in vitro. In vivo administration of a CGRP inhibitor abrogates release of IL-23 from dermal cDC2, resulting in a reduced IL-17–driven infiltrate and higher CFU (46). Thus, release of CGRP from TRPV1-expressing neurons is required for development of a local type 17 immune response in the setting of C. albicans infection.

The sufficiency of TRPV1 neuron activation in vivo can be examined using TRPV1-Chr2 optogenetic mice (17, 38). Local cutaneous photostimulation of TRPV1-Chr2 mice resulted in the development of type 17 inflammation with an associated expansion of neutrophils and IL-17–expressing T cells in the skin at the site of photostimulation (37). Photostimulated TRPV1-Chr2 mice showed augmented host defense to epicutaneous C. albicans and S. aureus infection. Thus, TRPV1 neuron activation in isolation is sufficient to initiate type 17 inflammation and augment host defense. Taken together, the ability of TRPV1+ pain-inducing neurons to both sense the presence of pathogens and trigger type 17 inflammation argues that painful stimuli represent immunological danger.

The beneficial effect of TRPV1-neuron–induced type 17 inflammation is not universal. Like S. aureus and C. albicans, S. pyogenes is sufficient for activation of TRPV1-expressing neurons and for induction of CGRP release in vitro (32). Dermal infection with S. pyogenes (a murine model of necrotizing fasciitis) resulted in elevated CGRP at the site of infection. Unlike infection with C. albicans, however, recruitment of neutrophils was inhibited, resulting in reduced host defense that could be reversed by inhibiting either CGRP or its release. The mechanism underlying this dichotomy is unclear. S. pyogenes, unlike C. albicans, is a professional pathogen and may have evolved mechanisms to avoid CGRP-mediated immune clearance. In addition, epidermal and dermal innervation is distinct with a significant proportion of TRPV1+ neurons residing in deeper skin layers (14, 18). The composition of dendritic cells also differs between the epidermis and dermis resulting in the same pathogen driving different immune responses when administered to the epidermis or dermis (51). Thus, the difference between C. albicans and S. pyogenes may derive from epidermal versus dermal routes of infection. In addition, the biology of CGRP in skin is complex, with reports of both proinflammatory and anti-inflammatory effects (52–54). Thus, the capacity of TRPV1-expressing neurons to modulate cutaneous host defense is clear, but important variables such as the pathogen, which neurons are activated, the precise location of infection, and possibly the amounts of CGRP released during infection may affect the outcome.

Psoriasis is a chronic type 17 inflammatory skin disease. The cause of psoriasis is understood to involve inflammatory cytokine loops dominated by IL-23, TNF-α, and IL-17 (55, 56). Targeted therapies against these cytokines have proven highly successful in the treatment of psoriasis. Interestingly, there is mounting support that cutaneous TRPV1+ neurons participate in driving psoriasis pathogenesis.

Evidence in support of neurocutaneous mechanisms in psoriasis pathogenesis comes from observations in patients experiencing resolution of psoriatic lesions following peripheral nerve damage (57). These observations are consistent with findings that levels of neurotrophins and neuropeptides including nerve growth factor, substance P, and CGRP are elevated in the skin of psoriasis patients (58, 59). In murine models of psoriasis, surgical denervation reverses the disease in a neuropeptide-dependent manner (50). Denervation resulted in decreased levels of CGRP and substance P as well as fewer infiltrating CD11c⁺ dendritic cells and CD4⁺ T cells. The effects of denervation could be reversed by administration of exogeneous substance P and CGRP. RTX-mediated ablation of TRPV1-expressing neurons inhibits release of IL-23 from dermal cDC2 and downstream IL-17 responses in the imiquimod dermatitis model of psoriasis (49). Moreover, administering intradermal botulinum neurotoxin A, which prevents vesicle fusion and release of neuropeptides from neurons, reduced levels of CGRP and substance P, resulting in a strong reduction in skin pathology in mouse models (60). Notably, in psoriasis patients, injection of botulinum neurotoxin A directly into chronic psoriatic plaques has been shown to have a beneficial effect (61, 62). Thus, cutaneous afferent neurons participate in the maintenance of established psoriasis lesions and represent a potentially important therapeutic target for the treatment of psoriasis.

Allergic contact dermatitis (ACD) is a type IV hypersensitivity reaction that manifests as a T cell–mediated allergic reaction caused by contact with a foreign sensitizer. The ensuing cell-mediated response leads to redness, thickening, and scaling of the skin. Recent work has highlighted a role of cutaneous sensory neurons in modulating the immune response in ACD. In a squaric acid dibutylester (SABDE)–induced model of ACD, ablation of TRPV1+ neurons or genetic deficiency of the TRPV1 channel exacerbated cutaneous inflammation (63). An opposite role for TRPA1—a neuron ion channel that gives rises to somatosensory modalities such as pain, itch, and cold in response to environmental irritants—was observed in hapten-induced cutaneous inflammation, as TRPA1-deficient mice displayed reduced production of proinflammatory cytokines following hapten challenge (63, 64). These data provide evidence that the TRPV1 and TRPA1 channels participate in inflammatory processes. How neurons participate in disease onset is unknown. It is possible that in addition to detecting pathogens, neurons may also respond to haptens and allergens or possibly to endogenous factors released by these agents.

Atopic dermatitis (AD) is a type 2 inflammatory skin disease driven by the production of epithelial-derived cytokines such as IL-25, TSLP, and IL-33, as well as Th2 and innate lymphoid cell type 2 (ILC2)-derived cytokines including IL-4, IL-5, and IL-13 (65, 66). Pruritis is a hallmark of AD and is a central component of disease pathogenesis. AD lesions are hyperinnervated with increased penetration of nerve endings into the epidermis (67, 68). Although the precise cutaneous neuron subset responsible for itch in AD remains to be fully elucidated, it is likely these are Nppb- and MRGPRA3-expressing neurons.

Cutaneous neurons express receptors for the type 2 cytokines IL-31, IL-4, and IL-13, which are sufficient to elicit itch behavior (4, 69). In the setting of chronic type 2 inflammation, cutaneous neurons are activated by the cytokines IL-4 and IL-13. Mice with the IL-4R ablated in sensory neurons or mice treated with a JAK1 inhibitor, which suppresses IL-4R–mediated signaling, show reduced itch in the MC903 model of AD (4). AD patients treated with anti–IL-4R–neutralizing mAbs report greatly reduced pruritis and improvement in dermatitis (70).

An early event in AD pathogenesis is the production of epithelial-derived TSLP and IL-33. Similar to IL-4 and IL-13, these cytokines can activate neurons and cause pruritis (6, 71). Unlike TRPV1-expressing neurons, in which activation has been shown to be sufficient for triggering type 17 immune responses, it remains unclear whether activation of pruritis-associated neuron subsets can trigger type 2 responses. In the lung and gut, the neuropeptide neuromedin U (NMU) stimulates ILC2 and the elaboration of type-2 inflammation, whereas CGRP, in concert with other type 2 cytokines, constrains ILC2 activation (72–75). It seems likely that similar type 2 neuroinflammatory circuits operate in the skin in the setting of AD, although this remains to be demonstrated.

A unique feature of the nervous system is its capacity to transmit directional information on a time scale of milliseconds. Neuronal activation in the periphery travels both orthodromically (toward the CNS) and also antidromically (toward the periphery). The ability to transmit neuronal signals multidromically facilitates the dissemination of action potentials throughout the skin via a neuronal reflex arc (76, 77). For example, in an axon reflex, depolarization of nerve endings initiates an action potential that travels orthodromically (toward the DRG). Centrally traveling impulses can then be diverted antidromically (away from the DRG) along collateral branches of the same axon, which in turn activates nerve terminals in the skin surrounding the original neuronal stimulus (78, 79). Activation of cutaneous TRPV1+ neurons either optogenetically or by C. albicans is sufficient to elicit type 17 immunity both at the original site of neuronal stimulation and in neighboring, unstimulated tissue via the dissemination of action potentials to surrounding skin (Fig. 1). As a result, adjacent tissue rapidly receives information of potential infection leading to type 17 “anticipatory immunity” and host defense in surrounding uninfected tissue (37).

In this review we have discussed how specific subsets of peripheral cutaneous afferent neurons interact with skin-resident immune cells to augment host defense and inflammation. TRPV1-expressing neurons are necessary and sufficient to initiate type 17 responses at least in part through the elaboration of the neuropeptide CGRP. The result is augmented host defense to some extracellular pathogens, although there is at least one example in which the effect is detrimental. TRPV1-expressing neurons also participate in maintaining inflammation loops in psoriasis. They may also participate in the initial pathogenesis of psoriasis and ACD lesions. Itch-sensing neurons respond to type 2 cytokines, which results in pruritis, a key component of AD. Approaches that interrupt neuronal function have been beneficial in the treatment of both psoriasis and AD.

Despite recent progress in the field of cutaneous neuroimmunology, the field remains nascent. Many major questions remain. It is unclear why activation of the same neuron subset by different pathogens induces distinct responses. This may represent activation of what are, in fact, similar but distinct subsets of neurons or possibly other unappreciated contextual cues. The precise function of CGRP and other neuropeptides on immune cells in the skin has yet to be fully elucidated. Given the ability of neurons to modulate T cell responses, it is plausible that they may alter Ag presentation in the skin, although the relationship between cutaneous neurons and adaptive immunity has not been studied. Most studies on cutaneous neuroimmune interactions have focused on the setting of inflammation or infection. Cutaneous neuroimmune cell interactions in the skin under homeostatic conditions warrants further exploration. Finally, the immune contribution of nonpeptidergic neuron subsets in the skin as well as neurons of the sympathetic system have all yet to be explored in detail. Clearly, much remains to be learned, but it does appear clear that neurons play a key role in modulating cutaneous immune responses and this interaction appears to be a ripe target for therapeutic manipulation for the treatment of many skin disorders.

The authors have no financial conflicts of interest.