Unraveling the Plastic Peripheral Neuroimmune Interactome

The nervous and the immune systems both protect and warn individuals of local and systemic threats by virtue of many shared functions, including sensing the presence of pathogens as well as injured and dying cells. Both share the capability to memorize experiences, and such memory dictates functional outcome upon rechallenge. As a result of their coevolution and overlapping roles, neurons and immune cells respond not only to environmental input but also to signals from each other (1). We argue that this reciprocal interaction is fundamental to their roles and is capable of amplifying and diminishing the function of the two systems in different circumstances. Nociceptors express many receptors for soluble mediators produced by the immune system, including ions, amines, lipids, peptides, cytokines, and chemokines. Conversely, immune cells respond directly to neuropeptides and other neuromodulators released from the peripheral terminals of activated nociceptors. There exists, therefore, the opportunity for extensive cross-talk between these seemingly independent systems, one that is increasingly recognized to contribute both to homeostasis and disease.

Immune interactions can profoundly affect nociceptor activation and sensitivity, altering the level of input to the CNS and changing pain or itch perception. Immune signals can directly drive nociceptor activity by acting on transducer receptors like TRPA1, a detector of multiple endogenous chemical irritants, or induce posttranslational modifications in ion channels like TRPV1, which then renders the neurons sensitive to nondamaging stimuli (2, 3). Likewise, nociceptors can have drastic effects on immune responses. Neuropeptides act, for example, on tissue-resident myeloid cells, increasing their activation state and cytokine production, and can also cause immune suppression, which could be either beneficial or detrimental to the host (4). This nociceptor-driven neuroimmune interaction is local, peripheral, and independent of the CNS. A different form of neuroimmune cross-talk is the systemic “inflammatory reflex loop” in which inflammatory products like TNF-α made by immune cells activate sensory afferents in the vagus nerve, which in turn, via a reflex output in vagal efferents originating in the brain stem, attenuates inflammation through a cholinergic anti-inflammatory pathway to the spleen (5).

Currently, our knowledge of reciprocal neuroimmune interactions in health and disease is incomplete. Although individual studies have revealed several specific neuroimmune pathways, such reductionist approaches do not faithfully represent the complex nature of neuroimmune interactions within a tissue in its entirety. There is substantial heterogeneity of nociceptor subsets as well as of immune cell types, including the receptors and the ligands they express. In addition, the cell composition and ligand–receptor profiles vary between tissue types and change with disease. As inflammatory processes progress, both neurons and immune cells begin to express receptors or ligands that they do not during the healthy state, and an infiltration of new leukocytes into inflamed tissues changes the immune landscape. Together, these variables add to changes in ligand–receptor pairing operational in different conditions and times. Therefore, neuroimmune cellular interactions are highly dynamic and cannot be revealed in a single snapshot of time and space (Fig. 1).

We propose the need for constructing a comprehensive map of the bidirectional ligand–receptor “interactome” between sensory neurons and immune cells. We discuss approaches as well as challenges toward obtaining such an interactome. We argue that identifying the full nature and consequences of reciprocal interactions between nociceptors and proximal immune cells and revealing its dynamic changes in disease states will greatly assist in identifying promising novel targets for therapeutic intervention for pain and itch control as well as for alleviating many immune disorders. First, we set the stage by briefly highlighting some of the known peripheral neuroimmune interactions.

Originally, the role of nociceptors in inflammation (neurogenic inflammation) was thought to be limited exclusively to vasodilation and increased capillary permeability. However, once it was recognized that innate and adaptive immune cells express receptors for many of the neuropeptides and neuromodulators produced by nociceptors, it became evident that these sensory neurons may directly act on immune cells (1).

A productive immune response consists of a complex cascade of events that involves the activation of the resident innate immune system, the generation of immune effector functions, the priming of the adaptive immune system, and finally, the memory generation for long-term protection. Interaction of nociceptors with immune cells can potentially regulate any of these elements of an immune response, something that needs to be explored.

Although inflammation has always been associated with pain and is indeed one of its cardinal features, it is only recently that researchers have begun to ask if the somatosensory system plays a driving role in inflammatory disease. One motivation for this was the observation made several decades back that denervation leads to an alleviation of rheumatoid arthritis and psoriasis, suggesting that sensory neurons may somehow promote inflammation (6, 7). Recent work has now revealed that during inflammation, neuropeptides released from activated nociceptors modulate many aspects of innate immune activation, including regulating leukocyte infiltration, myeloid cell activation, and innate cytokine production (8). Substance P for example promotes the adhesion of immune cells to the vasculature, the phagocytosis of microbes, and TNF-α and IL-1β production by macrophages (9–11), whereas vasoactive intestinal peptide (VIP) stimulates cytokine production by CD4 T cells and innate lymphoid cells (12). In contrast, the neuropeptide, calcitonin gene-related peptide (CGRP), exerts anti-inflammatory effects by augmenting the LPS-induced release of IL-10 by macrophages as well as suppressing IL-1β and TNF-α production by dendritic cells (DCs) via the repression of the transcription factor inducible cAMP early repressor (13, 14).

Because nociceptors secrete neuropeptides rapidly following their activation (within seconds), innate immune cells are likely much more susceptible to nociceptor-mediated signals than adaptive immune cells. However, the quality and nature of the innate immune response directly affects adaptive immunity (15). Therefore, T cell– and B cell–driven adaptive immune response may also be modulated by nociceptors indirectly via innate immune regulation or by directly responding to neuropeptides. CGRP skews T cell response toward Th17 (16), whereas the absence of its RAMP1 receptor results in impaired IL-4 production by Th2 cells (17). VIP-treated macrophages are able to suppress Th1 type cytokine secretion and induce Th2 type cytokine secretion, and VIP-treated mice have increased the numbers of IL-4–secreting cells and reduced the numbers of IFN-γ–producing cells (18). Nociceptors promote IL-17 production by γδ T cells in the case of psoriasis and Candida albicans infection by enhancing dermal DC production of IL-23 (19, 20). Nociceptor-immune cross-talk enhances type 2 allergic airway inflammation as well as antiparasitic immunity. VIP promotes IL-5 transcription by Th2 cells (12). Similarly, neuromedin U drives the production of type 2 cytokines and tissue repair factors by ILC2 and Th2 cells (21–23).

Although the role of neuroimmune interactions during the effector phase of immune responses in peripheral tissues has been extensively studied, relatively little is known about whether the neuroimmune axis plays a role in priming the adaptive immune responses in secondary lymphoid organs. Evidence suggests that DC function could be altered via the adrenergic receptors they express, enabling them to detect neurotransmitter release from the sympathetic nervous system (24–26). In the context of allergic airway inflammation, DCs in proximity to nociceptors express CGRP receptors, and CGRP reduces DC maturation and allergen-specific T cells (27, 28). It will be interesting to see if and how nociceptor interactions with DCs regulate their Ag presentation, maturation, or migration to lymph nodes. A role for nociceptors in the generation of central or tissue resident memory is currently quite unexplored.

Most of what we know about the regulation of immune responses by nociceptors is in the context of ongoing inflammation, which they can either suppress or amplify. We need to explore if nociceptor activation is sufficient to initiate inflammation. In other words, does the nervous system only modulate inflammation, or can it initiate an inflammatory response? Transdermal activation of cutaneous neurons via optogenetic stimulation is sufficient to lead to an induction of type 17 inflammation–associated cytokines (29).

Uncovering the full nociceptor to immune cell interactome at a highly granular level will reveal which sensory neuronal signals regulate which immune cells and when and how. This will help tease out specific molecular interactions in diverse inflammatory conditions as well as identify targets for innovative therapies.

In inflammatory microenvironments, the presence of diverse immune signals can both initiate and enhance the perception of pain and itch. The initiation of pain and itch occurs by ligand-dependent activation of receptors that trigger action potential firing, whereas enhancement is the consequence of increasing the sensitivity of these receptors (30, 31). Direct activation drives spontaneous pain and itch, whereas enhancement is responsible for pain hypersensitivity and itch amplification at the site of tissue injury/inflammation. One example of a direct neuronal activator is histamine. Mast cells are the major producers of histamine upon degranulation (32). Sensory neurons express the H1R and H4R histamine receptors, driving histamine-evoked itch sensations (33, 34). Blocking of this immune cell to sensory neuron interaction using histamine receptor antagonists is a widely used therapy for itch. However, because many itch-causing diseases occur via histamine-independent mediation (35), efficacy is limited.

Most immune signaling to nociceptors appears to be pain or itch enhancing. Induction of Cox-2 levels, for example, in inflammatory macrophages is critical for PGE₂ production. PGE₂ acts on EP1 GPCR receptors on nociceptors to elevate cAMP levels and activate protein kinase A, leading to posttranslational changes in transducer channels (36). This increases the sensitivity of the nociceptors, an effect referred to as peripheral sensitization (37–39). Cox-2 inhibitors are, in consequence, successful analgesics for low level inflammatory pain, but their efficacy is limited to pathological conditions associated with Cox-2 induction in immune cells and in which PGE₂ is a prime driver of peripheral sensitization. Immune complexes of IgG made by B cells in rheumatoid arthritis can trigger mechanical hypersensitivity, an effect mediated by the expression of FcγR by nociceptors (40). Nociceptors also sense the inflammatory milieu via cytokine receptors. IL-1β and TNF-α are rapidly secreted by immune cells and act directly on nociceptors via IL-1R and TNFR signaling (41, 42). These signaling pathways activate MAPK via p38, leading to posttranslational modifications in sodium channels that increase membrane excitability (43, 44). Once the inflammation resolves, inflammatory mediators disappear from the milieu, and the sensitivity of the neurons is then normalized. The simultaneous presence of many pain-sensitizing ligands of immune origin that act on nociceptors implies that targeting more than one for treatment of inflammatory pain may have a higher efficacy than current monotarget therapies.

In addition to causing acute changes in the membrane excitability, immune mediators can also induce retrograde signaling in nociceptors. Nerve growth factor (NGF) is made by immune cells and has a direct local sensitizing property on nociceptor peripheral terminals via TrkA receptor, contributing to peripheral pain hypersensitivity (45–47). However, the NGF/TrkA receptor complex is transported from the peripheral terminal to the nociceptor cell body, where it initiates transcriptional changes, including for transducers like TRPV1 and neuropeptides (48). NGF is also critical for neuronal growth, and its presence may induce de novo sprouting of neurons into areas like wounds (49). Immune cells produce several additional neurotrophins like BDNF and NT-4/5 that may contribute to maintaining innervation (50). In contrast, immune activation is also associated with sensory neuronal degeneration and may be involved in die back neuropathies (51).

Neurons express chemokine receptors that were traditionally thought to be limited to the immune cell compartment and crucial for immune cell migration (52). Upon injury, stromal cells secrete chemokines, and immune cells follow the gradient of chemokine to reach the appropriate tissue. Chemokines have been suggested to contribute to pain hypersensitivity as well as maintain inflammatory pain in nociceptors (52–54). However, the exact mechanisms and extent of chemokine receptor signaling in nociceptors remains unclear. Do they only impact sensory function, or can they play a role in growth, directionality, or integrity?

It is now clear that ligand–receptor interactions between immune cells and nociceptors can fundamentally change multiple aspects of neuronal structure, expression, and function. Constructing a neuroimmune interactome will help with mapping relevant ligand–receptor interactions modulating neuronal behavior and develop immune-based therapies to treat diverse diseases of the peripheral nervous system.

We have discussed the diverse nature of the bidirectional neuroimmune “conversation” and its role in inflammatory conditions. The standard way of studying this is “listening” to one dialogue from one player to another at one time, using reductionist and observation-based experimental designs. However, with these approaches, discovery of novel interactions is often a chance occurrence. The complexity of the immune–nociceptor cross-talk and its reciprocal nature calls for a more global approach. That a comprehensive map of neuroimmune cross-talk may have more value than dealing with single interactions has been recognized for brain-integrated reflex-like relationships between the nervous and the immune systems, a “neuroimmune communicatome” (55). However, this important concept still relies on cataloging individual observations into a large repertoire of interactions present in specific contexts rather than on, as we propose in this study, an unbiased strategy to unravel the full neuroimmune interactome. Existing data on neuroimmune interactions are covered in these reviews (1, 56–58).

Local intercellular interactions are made possible when a ligand expressed or secreted by one cell type engages with its cognate receptor present on another cell type in close proximity. This can be mediated by a direct physical association of the two cell types, membrane to membrane, or by soluble factors (Fig. 1A). One way to begin to study peripheral neuroimmune interactions is to find out all the ligands as well as all the receptors expressed by nociceptors and immune cells in a given tissue, inflammatory state, and time. Because of the heterogeneity of both neuronal and immune subtypes, it is critical to know the cellular identity of receptor and ligand expression patterns. Single cell transcriptomic analysis can provide cellular identity information as well as the receptor and ligand expression profile. Expression maps obtained by such an approach will, with bioinformatic analysis of all known and potential ligand–receptor interactions, constitute a putative neuroimmune interactome.

The ligands in such an interactome need not be limited to classical-soluble or membrane-bound ligands that exert effects by directly activating receptors and their signaling pathways. A ligand can be a mediator of an interaction made by one cell type that sets into motion a chain of molecular events that eventually changes the functionality of another cell type. For instance, Cox-2 is not a secreted protein but one that can synthesize a ligand for a receptor expressed on nociceptors. Development of powerful bioinformatics approaches will be critical to extract indications of any possible direct and indirect cellular communication from such complex interactomes, and additional experiments will be required to validate the function of a particular interaction using reductionist approaches. Furthermore, a full analysis of transcription-dependent changes in the neuroimmune interactome will need to be captured to reveal the dynamic aspect of the interactome in disease conditions.

A prerequisite to understand a diseased state is to define the healthy or homeostatic conditions.

Under healthy conditions, the tissue-resident immune cells and innervating neurons are seemingly quiescent. The transcriptional expression of inflammatory proteins in immune cells is minimal to prevent unwanted inflammation. Likewise, because nerve terminals are physically located far away from their cell bodies, neuropeptides are synthesized and stored at the terminals in advance, requiring minimal dependence on the ongoing transcription of neuropeptides. However, steady-state neuroimmune interactions can still take place via ligand–receptor pairs constitutively expressed on neurons and immune cells. For instance, TNFR superfamily and its ligands are almost ubiquitously expressed at steady-state and drive survival, growth, and inflammation (59). Single cell transcriptional analysis of immune cells in healthy conditions will reveal all ligands as well as receptor expression on immune cells type. Sensory neurons cannot be subjected to standard single cell RNA isolation methods because enzymatic dissociation of neurons leads to damage as well as modifies the transcriptome of the cells. To study transcriptional changes in neurons, one can isolate nuclei from the soma located in the dorsal root ganglia. Single cell profiling is a reliable method to capture the cellular transcriptional landscape of peripheral neurons (60). Because transcriptional expression may not accurately represent levels present on the cell surface, single cell RNA sequencing can be coupled with CITE-seq technology. CITE-seq offers multiplexed protein marker detection using oligonucleotide-tagged Abs to cell surface proteins (61).

A map of all possible ligand–receptor interactions that can occur between immune cells and nociceptors in the steady-state will define a healthy interactome signature, which can then serve as reference for the efforts to restore homeostasis in disease conditions. Bioinformatic tools can be used to calculate the “distance” between the interactome in disease and healthy conditions, which might be an important parameter for assessing disease pathology and prognosis as it is possible that beyond a certain distance, pathological neuroimmune interactions become irreversible.

To successfully develop meaningful peripheral neuroimmune interactomes, we need to define several key features, some of which cannot be revealed by single cell transcriptomic profiling.

The nociceptor neuropeptome: what are all the neuropeptides made and released by nociceptors that can act on immune cells? Transducer channels in nociceptor terminals when activated by stimuli initiate a depolarizing generator potential, which both activates voltage-gated sodium channels to cause action potentials and voltage-dependent calcium channels, leading to a calcium influx that initiates the vesicular release of neuropeptides from the terminal. Nociceptors can secrete a wide spectrum of presynthesized neuropeptides, including substance P, CGRP, VIP, SST, and neuropeptide Y, which constitute the neuropeptome (62). These are the primary mediators of interactions with proximal immune cells (Fig. 1A). Different neuron subtypes make and release distinctive combinations of neuropeptides. Changes in peptide expression in disease conditions will alter the neuropeptome.

The immune cell secretome: what presynthesized immune ligands can act on nociceptors? There are instances in which immune mediators are presynthesized and are ready to be secreted (Fig. 1A). CD8 T cells and NK cells store granzyme B and perforin in vesicles, which are released upon activation. Specific populations of macrophages have premade TNF-α and pro–IL-1, which undergo signal-dependent translation and posttranslational processing (63, 64). Similarly, complement proteins are also preformed mediators that can act on nociceptors (65, 66). These mechanisms allow for a rapid immune action on nociceptors independent of de novo transcription.

Tissue proteomic analysis is valuable for identifying the repertoire of peptides/proteins released acutely by nociceptors and immune cells following stimulation. However, the specific cellular source of the ligands will remain unknown. Development of single cell proteomics will greatly aid defining the transcription-independent component of the neuroimmune interactome at high spatial resolution. Furthermore, imaging techniques can be used to gain insights into the physical proximity of specific immune cells to nerve endings, which may be predictive of direct neuroimmune interactions.

Inflammation is accompanied by transcriptional changes in nociceptors. Likewise, the immune landscape undergoes transformation due to transcriptional modifications in tissue-resident immune cells as well as leukocyte infiltration. These changes allow for interactions that do not exist at steady-state reshaping the interactome over the course of the inflammation, rendering it to be highly plastic (Fig. 1B).

The immune population in barrier tissues consists of diverse cell types, such as macrophages, innate lymphoid cells, DCs, and T cells. When the host senses microbial insult or tissue injury, stromal cells release chemokines to recruit additional immune cells to the damaged site. This process significantly alters the cellular composition of the tissue. An inflammatory microenvironment also causes an upregulation of chemokines, cytokines, and receptors for neuronal signals in immune cells. The conversion of inflammatory milieu to homeostasis was originally thought to occur simply via dissipation. However, we now know that the resolution of inflammation is a bioactive process involving negative feedback loops for inflammatory cytokines and chemokines as well as the synthesis of specialized proresolving lipid mediators by immune cells, which modulate neuronal function as well (67, 68). The immune population is therefore constantly changing through the initiation, progression, and resolution of inflammation. These immune changes together with transcriptional changes in neurons innervating inflammatory tissue create a ligand–receptor landscape that is dynamic and plastic (Fig. 1C).

RNA profiling of individual immune cells and nociceptors at various stages of disease progression will identify the dynamic changes in cellular composition and of ligands and receptors. It is conceivable that different inflammatory states will be associated with distinct ligand–receptor expression profiles constituting unique disease interactome molecular fingerprints.

Neuroimmune interactions are strongly contextual. Even at steady-state, every tissue will have its unique interactome. This is due to the distinct set of resident immune cells in different anatomical locations. Langerhans cells are uniquely present in the skin, whereas alveolar macrophages are only found in the lung. Functionally distinct nociceptor subtypes are found in in different tissues as well as within a given tissue (69–71). These individual neuronal subsets are likely to respond uniquely to a given insult, further adding to the complexity of the interactome. It is highly likely that the nature, degree, and timing of the plasticity of the interactome in disease conditions will also vary from tissue to tissue, something to be explored.

There are several ligands and receptors coexpressed by both immune cells and neurons. Neuropeptide Y is expressed by neurons, macrophages, DCs, and lymphocytes (72). It can act in an autocrine and paracrine manner as its Y1, Y2, Y4, and Y5 receptors are also present on both neurons and immune cells (72). Because the outcome of autocrine signaling might be different from paracrine effects, it is important to try to extract which of these modes of signaling is operational under what circumstances. One can profile neurons and immune cells in isolation and use bioinformatic tools to identify “corrected” gene expression profiles that eliminate the calculated effects of autocrine signaling (73).

Finally, in the context of an infection, another variable to take into account is the nature of the pathogen itself. Different classes of pathogens will likely result in different interactome signatures due to their unique immune activation mechanisms, site of infection, and direct actions on nociceptors (4). Pathogen gene expression data during the course of an infection may together with that from immune cells and nociceptors give insight into pathogen-specific neuroimmune signatures.

The complexities of intercellular signaling suggests that constructing disease- and tissue-specific neuroimmune interactomes is merely a first, but critical, step toward fully understanding the bidirectional conversation between nociceptors and immune cells.

Nociceptors and immune cells interact bidirectionally, a conversation made possible by the wide spectrum of cells involved and the receptors and ligands they express. Several direct interactions between the two cell types, and their consequences in health and disease, have already been revealed. However, from our growing understanding of the complexity, dynamic nature, and contextual features of the reciprocal interactome, it is imperative that we define all possible interactions that occur between nociceptors and immune cells and how they collectively modify tissue properties. Single cell RNA sequencing and bioinformatic tools will enable us to begin to study the peripheral neuroimmune interactome and define its plasticity and temporal changes. We anticipate that many inflammatory diseases will soon be defined by a distinct interactome signature, which may well contribute to the development of targeted treatment options.

The authors have no financial conflicts of interest.