Neural Immune Communication in the Control of Host-Bacterial Pathogen Interactions in the Gastrointestinal Tract

SENSORY NOCICEPTIVE NEURONS

Detection of noxious stimuli ranging from extremes in heat or cold to the presence of discrete chemical substances (2) has been long appreciated to occur through highly specialized “sensory nociceptive” neurons. This ability is conferred through the expression of a family of transient receptor potential (TRP) cation channels that open upon stimulation, such as ligand binding, resulting in neuronal activation (2). While there are a multitude of TRP channel family members, we will limit our scope to those expressed by sensory nociceptive neurons that are responsive to bacterial pathogens, their components, or the host response to infection. Interested readers are directed to the comprehensive reviews on TRP channels and their expression on various cell types (14).

Perhaps the most well-known of these TRP channels is the transient receptor potential cation channel subfamily V member 1 (TRPV1) that is activated by temperatures in excess of 43°C, low pH, and compounds such as ATP and capsaicin (14–17). Activation of TRPV1 on nociceptive neurons consequently induces signaling in the DRG of the spinal cord, initiating efferent signaling in a classical reflex arc (18). This signaling culminates in the release of neuropeptides such as substance P (SP), calcitonin-gene related protein (CGRP), and neurokinin A (NKA) into the organ. Activation can also cause localized release of these neuropeptides at the initial site or organ system through antidromic activation (2, 19). Although sensory nociceptors can be found throughout the body, the density, distribution, and sensitivity of these neurons can vary dramatically depending on the organ system. As a specific example, the GI tract is densely innervated by sensory nociceptors; however, unlike other organs such as the skin, the majority of these nociceptive neurons in healthy individuals are quiescent (20). It is important to note that these TRPV1-expressing sensory afferents in visceral organs can also function to support normal physiology, with these neurons in the small intestine and colon functioning in mechanosensation of organ distension (21). From a neuroanatomical perspective, sensory afferent neurons in the colon are derived from lumbar splanchnic and pelvic nerve of the thoracolumbar and lumbosacral regions of the spinal cord, respectively (22). Although historically these sensory neurons have been identified based on morphometry or electrical properties, single-cell sequencing of colonic projecting neurons from the thoracolumbar and lumbosacral DRG identified TRPV1 expression in three of seven discrete types of colonic sensory afferents (22). These data highlight that TRPV1⁺ sensory neurons are heterogenous in composition and respond to unique stimuli, and could result in different functional outcomes.

ROLES OF SENSORY AFFERENTS IN BACTERIAL INFECTION

Despite these differences between types of sensory innervation within different organs, the release of nociceptive neuropeptides has a profound effect on the molecular underpinnings of inflammation and host responses to pathogens. For example, SP acts directly as a chemotactic molecule (23), while increasing chemokine expression, endothelial permeability (24, 25), and lymphatic drainage (26). Complicating this generalized proinflammatory role of nociceptive neuropeptides (26), CGRP can increase vasodilation (27) and act synergistically with cytokines to increase neutrophil recruitment (28) in vivo, yet reduces in vitro T-cell proliferation (29). Thus, the overall effect of sensory nociceptive activation on host immune responses is likely to be determined by several variables, including the nature of the stimuli, the organ site, and the immune cells involved.

It is becoming increasingly clear that pathogens, or the resulting infection-induced inflammation, can activate sensory nociceptors. While these neurons express a range of Toll-like receptors (TLRs) to detect components of pathogens, typically TLR activation by pathogen-associated molecular patterns (PAMPS), such as lipopolysaccharide (LPS), reduces the threshold of neuronal activation (30). It is an important caveat that exquisite care and appropriate culture techniques must be used to exclude nonneuronal cells and the potential contribution of nonneural cells, such as satellite glial cells, in the assay of these transcripts. Activation of TLRs not only can reduce the threshold of activation but can further induce expression of proinflammatory cytokines, and stimulate prostaglandin synthesis. This further complicates our interpretation of bacterial activation of sensory nociceptors, since these neurons are activated by TNF-α, IL-1β, IL-6, IL-17A, IL-31, histamine, and prostaglandins (31–33). Additionally, bacterial infection can increase TRPV1 signaling through TLR4-dependent increases in TRPV1 expression, altering sensitivity to exogenous or endogenous ligands (18). As a polymodal nociceptive channel, TRPV1 can also function as a direct detector of bacterial components, becoming activated by LPS binding at a discrete site from the capsaicin-binding site (34). Other host-produced compounds besides cytokines, such as the lipid-derived compounds N-arachidonyldopamine (35), lysophosphatidic acid (36), and hydroxyeicosatetraenoic acid (20-HETE) (37), have been identified as TRPV1 ligands. However, at this time it is not certain what role if any these substances have during host immune responses to enteric bacterial pathogens. Further complicating analysis, activation of TRPV1 has also been shown to “gate” or reduce the activity of other ion channels, including the mechanically activated Piezo cation channels (38). This would suggest that activation of TRPV1 can induce changes in other signaling pathways, thus further complicating analysis of “TRPV1-dependent” responses. With this in mind, it would seem that infection and the resulting immune response would be expected to activate or facilitate a reduced threshold for activation of sensory nociceptive neurons through a variety of discrete mechanisms. While the focus has been on neuronal expression of this channel, care is required in the interpretation of results from constitutive knockout (KO) mice or neuronal ablation studies due to expression on nonneuronal cells such as epithelial cells and T cells (39, 40). In practice, these concerns can be mitigated by use of bone marrow chimeras, conditional knockout mice, or in ablation studies allowing sufficient time for replenishment of nonneuronal cells.

Adding to this complexity of the role of the nervous system during enteric infection is the expression of a diverse array of other TRP channels, which are specialized sensors of unique stimuli. The TRPA1 cation channel is activated by noxious cold (<17°C) and chemical irritants such as mustard oils (41), endocannabinoids, lipid peroxidation products generated during oxidative stress and inflammation (42, 43), and components of bacteria such as LPS (44). This ability suggests that sensory nociceptors expressing TRPA1 could become activated not only by LPS, but also by products of the host immune response during infection. Although the role of TRPA1 during enteric bacterial infection is unknown, in the bladder, LPS instillation induced nociceptive hypersensitivity, together with bladder voiding in wild-type (WT) but not TRPA1 KO mice, with no change in overt inflammation. These findings indicate that TRPA1⁺ nociceptors are responsive to LPS instillation; however, it is uncertain if this is a direct effect of LPS on this channel or if these changes are a consequence of the local induced inflammation (45). Similar to TRPV1, TRPA1 can also be activated by cytokines such as IL-4 and IL-13, suggesting that expression of this TRP channel confers neurons with the ability to monitor immune responses (32). This role of TRPA1 in monitoring of inflammation appears to be widespread, with TRPA1 activation in the lung increasing immune cell recruitment and airway hyperreactivity during ovalbumin (OVA)-induced airway hypersensitivity. Reduced inflammation and changes in lung function were correlated with reduced SP, CGRP, and NKA release in sensitized TRPA1 KO mice challenged with OVA compared to WT mice. Although these data demonstrate that endogenously produced immune reactive products activate TRPA1 to enhance lung inflammation, the location of TRPA1 was not evaluated (46). As a whole, these findings suggest that TRPA1-dependent nociceptor activation can amplify local innate immune responses to promote host protection.

Although TRPA1 and TRPV1 are established to influence a variety of immune responses, it is important to note that not all TRP channels expressed by sensory nociceptors appear to exert an influence over immune outcomes. While the cold- and menthol-sensing channel TRPM8 is expressed by sensory nociceptive neurons, immunological effects have been attributed to TRPM8 expression by macrophages (47, 48). These data suggest that careful dissection of the roles of sensory afferent neuron activation and TRP channels in immune responses is required.

Despite the well-established ability of SP and CGRP to drive neurogenic inflammation, activation of sensory nociceptors has paradoxically been found to drive maladaptive host immune responses by select pathogens. Although the majority of studies on nociceptive neurons during infection have been conducted in the skin or lung, there appear to be some commonalities between these sites and responses evoked in the GI tract. Infection of wounded skin with Staphylococcus aureus activates local sensory nociceptors in the immediate vicinity of the site that is not proportional to the wound size, indicating that the infection triggered activation of sensory nociceptive neurons. Using cultured DRG neurons, Ca⁺² signaling indicative of activation occurred in vitro following addition of live and heat-killed S. aureus and S. pneumoniae. This nociceptor activation, however, is detrimental to the host, as in vivo this nociceptor activation leads to a significantly inhibited host response. Confirming the role of sensory neurons, prior neural ablation with the excitotoxic TRPV1 agonist resiniferatoxin (RTX) (49) increased host protective immune responses. Using this approach, maladaptive immune responses were identified to be due to the release of CGRP and inhibition of neutrophil recruitment to the wound site. This ability to subvert host immune responses through sensory nociceptors was not restricted to a single bacterial pathogen, as Streptococcus pyogenes induced similar maladaptive responses due to TRPV1⁺-dependent neuronal activation and release of CGRP. Highlighting that these effects of neuronally induced immune subversion are not limited to the skin, CGRP released by sensory neurons in the lung reduced host survival during methicillin-resistant Staphylococcus aureus pneumonia (5). It is interesting to note that at this mucosal site, prior ablation of TRPV1⁺ neurons increased neutrophil recruitment, and γδ T-cell recruitment due to increased expression of the neutrophil chemotactic protein CXCL1 and TNF-α (5). While these studies suggest that bacterial activation of sensory nociceptors is a general mechanism of immune subversion, it is critical to note that different heat-killed bacterial pathogens elicited various degrees of nociceptor activation in vitro (3). These findings suggest that the response of sensory afferent neurons to bacterial pathogens is likely to be highly complex. In support of this, silencing of sensory afferents using the voltage-gated sodium channel blocker QX-314 reduced infection-induced neuronal excitation and hyperresponsiveness to noxious stimuli, with no reduction in S. aureus burden (50). These data could be indicative of different types of sensory nerves in the dorsal skin compared to the hind paw of the mouse, or that neural ablation versus silencing could target different types of sensory nerves. Either of these possibilities suggests that nociceptors in the GI tract or other organ systems could respond uniquely to bacterial pathogens.

This ability of sensory nociceptors to dictate host immune responses is not limited to the lung and skin. In the GI tract, neuropeptides released by sensory nociceptors can exert host-protective and proinflammatory effects. At this mucosal site, SP has long been regarded as proinflammatory, with increased release occurring in animal models of nonbacterial-induced colitis, and with selective SP receptor antagonists reducing the severity of inflammation (51, 52). During enteric bacterial infection, treatment of WT mice with SP receptor antagonists, or knockout mice for the gene encoding the SP receptor, Tacr1, infected with the invasive enteric pathogen Salmonella enterica serovar Dublin, have significantly reduced IL-12 and IFN-γ production, and consequently reduced host survival (53). The source of host-protective SP during enteric infection was not established, and is complicated by the ability of monocytes (54), macrophages (54, 55), and lymphocytes (56) to serve as nonneuronal sources. Similar to the skin, sensory nociceptors express not only SP but also CGRP and have been shown to exert a protective role during Salmonella enterica serovar Typhimurium infection. Sensory nociceptive neuron-derived CGRP was shown to support maintenance of segmented filamentous bacterial in the ileum, and reduce the differentiation of microfold “M” cells (57), a site of S. Typhimurium invasion (Fig. 1) (58). Although the extrinsic sensory innervation of the GI tract originates from either the DRG or nodose ganglion, and can express TRP receptors, including TRPV1, and SP (59), there are few CGRP⁺ vagal afferent neurons that innervate the intestinal tract (60). In addition, while these data suggest that CGRP is released by sensory neurons from the DRG during infection, there are two isoforms of CGRP, with CGRPα predominantly expressed by extrinsic neurons and CGRPβ by intrinsic innervation (61). The source of this host-protective CGRP has not been formally proven; however, given that CGRPβ-expressing intrinsic neurons do not express TRPV1 and are insensitive to capsaicin (61), these cells would be unaffected by sensory nociceptor ablation methods that are dependent on TRPV1 expression. Together, this would suggest that host protection is afforded by the sensory nociceptors arising in the DRG.

• Open in new tab
• Download powerpoint

FIG 1

Extrinsic sensory innervation of the colon drives maladaptive responses to Salmonella Typhimurium infection. Sensory nociceptive neurons that originate in the dorsal root ganglion (DRG) release CGRP in the small intestinal mucosa, thereby supporting luminal segmented filamentous bacteria (SFB) that consequently reduce susceptibility to S. Typhimurium infection. Release of CGRP also inhibits the differentiation of microfold M cells in Peyer’s patches, reducing bacterial burden and disease.

Coordination of protective host immune responses to enteric bacterial pathogens by sensory nociceptors appears to be a generalized phenomenon, regardless of the nature of the bacterial pathogen. Using the model noninvasive enteric pathogen Citrobacter rodentium, we found that ablation of nociceptive neurons with RTX significantly increased the severity of disease and delayed bacterial clearance (62). As TRPV1-deficient mice also exhibited increased bacterial burden and delayed clearance, this host-protective function of sensory nociceptive neurons appears to be TRPV1 dependent (Fig. 2) (62). Increased fecal and colonic adherent pathogenic bacteria were found to be due to reduced IL-22 production, a cytokine vital to the host response against C. rodentium. This reduction in host-protective immune function 10 days postinfection was due to delayed recruitment of CD3⁺ T cells into the infected colonic lamina propria. Ablation of sensory neurons appeared to reduce T-cell recruitment due to reduced expression of the mucosal addressin molecule MAdCAM-1 (62). In determining the potential role for CGRP in mediating the host-protective effects of the sensory neurons, mice were administered a highly selective and potent small molecule CGRP receptor antagonist (63, 64). Blockade of CGRP receptors did not increase the fecal or colonic adherent C. rodentium, or significantly reduce T-cell recruitment (62). In keeping with prior studies where C. rodentium infection did not increase colonic CGRP expression (65), our data demonstrate that nociceptor-mediated protection is independent of CGRP. These data highlight that neuroimmune interactions during enteric bacterial infection are complex and dependent on the region and the pathogen involved.

• Open in new tab
• Download powerpoint

FIG 2

Extrinsic sensory neurons in the colon coordinate protection during enteric infection. In mice with intact sensory innervation, infection with C. rodentium drives increased recruitment of IL-22-producing T cells into the colonic lamina propria. Ablation of TRPV1⁺ sensory nociceptors with RTX reduces expression of MadCAM-1 and consequently T cell recruitment during C. rodentium infection.

In these studies of the role of nociceptive neurons during infection of skin, lung, and intestine, the site of infection has received focus without considering the effects on developing immune responses in secondary lymphoid tissues, such as the spleen and lymph nodes. In support of nociceptive neuropeptides modulating responses in these immune tissues, SP increases lymphatic flow by causing the contraction of lymphatic muscle cells (66–69). Control over afferent lymphatics by nociceptive neurotransmitters appears to be a feature of the intertwining of SP-expressing neurons with lymphatic vessels (70). In contrast to SP-induced increased flow rates of afferent lymph, CGRP induces smooth muscle relaxation, consequently reducing lymph flow (71). Anatomically, this control is provided by the peri- and paravasulcar innervation of mesenteric, axillary, inguinal lymph nodes and spleen that coexpress SP and CGRP (72, 73). Nonvascular associated fibers have been found throughout the lymph node (74), although recent studies using a genetic fate map approach with tdTomato expressed in Nav1.8⁺ sensory neurons revealed only perivascular sensory nociceptor innervation of the lymph nodes (75). This disparity suggests discrete patterns of innervation in different strains of mice, or that SP and CGRP are expressed by other Nav1.8⁻ neuronal cell types. Beyond lymphatic endothelial cells, these neuropeptides activate other stromal cells, or immune cells in the lymphatic tissues (76). At this point, it has not been experimentally determined whether nociceptor-induced host-protective responses require signaling in these immune tissues, or what cell types are involved. These studies would require use of conditional knockout mice for the receptors of these neuropeptides, and for techniques to perform organ-selective neuronal ablation to be developed.

VAGAL AFFERENTS AND THEIR ROLE IN BACTERIAL INFECTION

Maintenance of a wide range of homeostatic functions in the body is provided through the sensory information detected by vagal afferents. As the longest cranial nerve, the vagus innervates numerous visceral organs, including the heart, lungs, spleen, gastrointestinal tract, liver, and pancreas, placing vagal sensory afferents at important interfaces between the host and the environment.

The cell bodies of afferent vagal sensory neurons reside in the nodose and jugular ganglia, with projections to the brainstem where most synapse to neurons in the nucleus tractus solitarius (NTS). Information is coordinated in the NTS complex with resulting outflow, i.e., efferent vagal nerve signaling, occurring through the activation of neuronal cell bodies of the dorsal motor nucleus (DMN). Given the positioning of vagal afferent sensory neurons at the interface between the body and the external environment, it is not surprising that the vagus is implicated as a critical sensor of pathogens and inflammation. As evidence of this role, infection of mice with the enteric bacterial pathogen Campylobacter jejuni activates vagal afferent neurons, evidenced by expression of the immediate early gene c-fos in neurons of the nodose and NTS (77). These data, in concert with the absence of detectable serum cytokines, suggested that pathogens could be directly detected by vagal afferent neurons (77). It is, however, worth noting that vagal afferent activation could occur indirectly due to pathogen-induced changes to host physiology or activation of intestinal enteroendocrine cells (78).

VAGAL EFFERENT NEURONS AND THEIR ROLE IN IMMUNE REGULATION AND BACTERIAL INFECTION

Stimulated or induced efferent vagal nerve fiber activity can impact host immune processes by directly communicating with immune cells in organs such as lung and intestine, or through intermediaries in secondary lymphoid tissues (79). For example, it is well appreciated that inflammation due to activation of tissue resident macrophages in the small intestine following surgery can be reduced by stimulation of vagal nerve efferents. Release of acetylcholine from these nerve terminals in the intestinal muscularis binds to nicotinic acetylcholine receptors on these macrophages to prevent NF-κB signaling and consequently reduce proinflammatory gene expression (80). Vagal innervation of secondary lymphoid organs, such as the spleen, is indirect, with efferent vagal nerve fibers synapsing onto sympathetic neuronal cell bodies in the superior mesenteric ganglion/celiac ganglion complex (81). These sympathetic fibers project into the spleen and mesenteric lymph nodes and, when activated, release norepinephrine (NE) into these tissues. This indirect neuroimmune circuit requires specialized T cells to release acetylcholine (ACh) that can inhibit NF-κB-dependent signaling transduction in macrophages (82). Although the consequences of vagal efferent signaling directly or indirectly to macrophages during enteric bacterial infection are unknown, we have shown that Ach-producing T cells exert a host-protective effect during infection with an enteric bacterial pathogen. While ChAT T cells are rare in the GI tract of naive mice, an inability of T cells to produce ACh caused significant changes in the commensal microbiota (83). Based on these observations during health, we assessed the contributions to host-pathogen interactions. Increased numbers of ChAT⁺ T cells are found in C. rodentium-infected colon, but not during chemically induced colitis, and conditional T-cell knockout mice have significantly increased bacterial burdens and delayed pathogen clearance (Fig. 3) (84). Despite the known role for sympathetic innervation to induce ACh release from these cells in the spleen and MLN (81, 82), the role of sympathetic nerves in communicating to ChAT⁺ T cells in the colon during infection is unknown.

• Open in new tab
• Download powerpoint

FIG 3

ChAT⁺ T cells induce host-protective responses during C. rodentium infection. T cells expressing ChAT and producing acetylcholine home to the colonic lamina propria during infection with C. rodentium. Production of ACh enhances expression of inducible nitric oxide synthase (iNOS), a known host-protective molecule.

Vagal innervation and stimulation can also enhance innate immunity and host protection in the peritoneal cavity by controlling production of enzymatically produced factors during the resolution of inflammation, termed specialized pro-resolving mediators (SPM) (85). This term applies to a broad class of mediators, including resolvins, lipoxins, and protectins, and protein conjugates of these lipids that display unique biological activity (86). Vagotomy in mice significantly altered the production of a wide variety of lipid mediators, including the peptide-lipid conjugate termed protection conjugates in tissue regeneration 1 (PCTR1) (87). While it is unknown if PCTR1 is produced in the GI tract, this protectin conjugate reduces inflammation following peritoneal infection with Escherichia coli by enhancing chemotaxis of monocytes and macrophages, and increasing bacterial phagocytosis (88). During peritoneal E. coli infection, vagotomy significantly delayed bacterial clearance, reduced macrophage phagocytotic capacity, and decreased the number of type 3 innate lymphoid cells (ILC3s) (87). Vagotomy-induced changes were due to reduced levels of ACh in the peritoneal cavity, as ACh induced PCTR1 production by ILC3s, and administration of either ILC3s or PTCR1 to vagotomized mice improved the resolution of inflammation (87). As ILC3s are a critical component of the host defense against C. rodentium (89), the effect of a vagal-protectin conjugate axis on ILC3 in the GI tract could be an important mechanism. Although Ach-producing cells were found in close proximity to ILC in the omentum, these cells were not T cells, supporting our observation of a large ChAT⁺ B cell population in the peritoneum that could regulate neutrophil migration (90). As a whole, these studies suggest that a multitude of vagus nerve-mediated neuroimmune circuits could exist to aid host protection or resolution following infection of the GI tract.

SYMPATHETIC INNERVATION

Neuroimmune circuitry that utilizes the sympathetic nervous system consists of the cholinergic anti-inflammatory pathway (CAIP) and pathways descending the spinal cord to prevertebral ganglia that provide sympathetic innervation to secondary lymph nodes, and the GI tract, including the liver (91). In the spleen, a sympathetic neuroimmune circuit has been characterized that is discrete from the CAIP, with stimulation of the splenic nerve sufficient to inhibit LPS-induced TNF-α production independent of α7R expression (92). This sympathetic inhibition of immune cell activity is also present in the GI tract, with stroke increasing sympathetic tone in the liver that reduced invariant natural killer T-cell crawling, IFN-γ, and IL-12 production, with increased IL-10 and IL-5 production (93). Confirming the role of sympathetic innervation, ablation of these neurons prevented immune suppression following stroke, while administration of NE alone recapitulated these effects (93). Thus, sympathetic innervation of the liver is capable of dictating immune cell activity and could have a significant impact on responses to pathogens. Highlighting the complexity of neuroimmune circuits in the GI tract, sympathetic neuronal activation is not always deleterious to the host production protective factors. Infection with Salmonella Typhimurium in the small intestine induced activation of extrinsic sympathetic neurons and NE-dependent reprogramming of muscularis macrophages from proinflammatory to a tissue-reparative phenotype (94). This signaling to muscularis macrophages was further found to protect against S. Typhimurium-delayed gastrointestinal transit caused by infection-induced loss of enteric neurons (Fig. 4) (95).

• Open in new tab
• Download powerpoint

FIG 4

The enteric nervous system participates in the immune response to S. Typhimurium and can be altered by infection. Enteric infection can result in loss of enteric neurons that precipitates reduced or altered intestinal motility. Infection also induces activation of extrinsic sympathetic neurons that cause the release of norepinephrine to activate muscularis macrophages. These macrophages appear to have a unique role in reducing the depletion of enteric neurons (A). Production of IL-18 by the enteric nervous system can also induce antimicrobial peptide expression by goblet cells, directly aiding the killing of enteric bacterial pathogens (B).

Despite these elegant experiments, it is uncertain what effect loss of this neuroimmune circuit would have on the host defense or overall immune response. In support of this host-protective role during systemic bacteremia, increased activation of sympathetic outflow reduced bacterial burden in the liver due to increased peritoneal macrophage and dendritic cell phagocytosis of E. coli, and increased monocyte and macrophage bactericidal activity, respectively (96). These studies reveal the complicated nature of neuroimmune interactions while highlighting the importance of neural input in resolving certain infections. Perhaps it is not surprising that certain pathogens have been reported to detect the host through the release of these neurotransmitters. It has been reported that NE can act as a ligand for bacterial quorum-sensing proteins (97), inducing bacterial proliferation and expression of virulence genes (98, 99). Moreover, mice that lack the ability to produce NE were reported to have reduced C. rodentium colonization (98), highlighting that specific enteric bacterial pathogens could utilize host-derived signals to the benefit of the pathogen. This direct effect of NE on enteric bacterial pathogens demonstrates the need for carefully controlled experiments in which sympathetic neuronal ablation is performed to rule out the potential of enhanced virulence gene expression as opposed to reduced host defense capacity.

ENTERIC NERVOUS SYSTEM

The ENS is comprised of an estimated 200 to 600 million neurons that reside in the intestinal wall from the oral to the anal cavity. These neurons are found within two layers: the myenteric plexus (MP) that is between the outer longitudinal muscle and the inner circular muscle and the submucosal plexus that is found between the circular muscle and the mucosa. Neurons in the myenteric plexus can be found in discrete ganglia, modulate GI motility, and are closely associated with muscularis macrophages, while submucosal neurons regulate ion secretion, epithelial permeability, and blood flow. Although early experiments demonstrated that the ENS can control GI physiology in the complete absence of extrinsic innervation, both the myenteric and submucosal plexus receive input from the CNS (1, 100). As it is beyond the scope of this review to detail the nearly 20 neuronal cell types identified, reflexes, and all the mechanisms that serve to control GI function, the interested reader is directed to the excellent comprehensive reviews on this subject in references 1 and 101. In addition to the host-protective functions of increased water secretion and contractility to expel pathogens or their toxins, the ENS can drive localized antibacterial responses. Recently, conditional deletion of IL-18 in ENS neurons significantly increased bacterial burden and morbidity during S. Typhimurium infection (102). This ENS-derived IL-18 was proposed to increase goblet cell antimicrobial peptide expression in the proximal colon during infection but not at steady state (Fig. 4) (102). It is uncertain if this effect of ENS-derived IL-18 has similar effects on Paneth or goblet cells in the small intestine, another site of S. Typhimurium entry. Finally, while IL-18 expression by ChAT⁻ neuronal nitric oxide synthase⁺ (nNOS) and ChAT⁻ nNOS⁻ myenteric plexus neurons was observed, the mucosal IL-18⁺ innervation was revealed by the general neuronal marker βIII tubulin (102). This makes it difficult to ascertain if these neurons are part of the intrinsic or extrinsic innervation, or if they are part of the ENS and originate in either the MP or submucosal plexus. The types of neurons in the ENS that project from the MP to the mucosa are limited to intrinsic primary afferent neurons, and are typically cholinergic, expressing ChAT and not nNOS (103). Whether this IL-18 expression in the ENS represents a novel neuronal type, or this ability to induce AMP is a conserved feature of neurons in both the myenteric and submucosal plexus is unknown. In addition, although conditional ablation studies were performed using Cre recombinase expression under the heart and neural crest derivatives expressed 2 (Hand2) promoter, to provide ENS-selective ablation, the specificity of this promoter is uncertain. Single-cell sequencing data identify Hand2 expression in the ENS and in sympathetic neurons, meaning that extrinsic sympathetic innervation could equally have been affected using this ablation strategy (104). Despite these concerns, these exciting data demonstrate the complexity and numerous layers of neuronal control of host defenses to enteric bacterial pathogens.

This multitude of neuroimmune circuits in the GI tract is further illustrated by recent reports demonstrating that enteric neurons producing vasoactive intestinal peptide (VIP) can affect select aspects of host immune function. These VIP⁺ neurons in the mucosa have been found in close association with tertiary lymphoid tissue in the intestine that contain ILC3 (105, 106). Although these ILC3 were found to express high levels of the VIP receptor 2 (VIPR2), the functional ramifications are contentious, with activation of VIPR2 in vitro reducing (105) and increasing IL-22 expression (106). To decipher the role of VIP in vivo during an infection, a chemogenetic approach was used (107, 108), whereby controlled selective expression of an activating designer receptor exclusively activated by a designer drugs (DREADD) receptor was achieved in VIP⁺ cells using Cre recombinase under the control of the VIP promoter (105). Mice infected with C. rodentium and administered the DREADD ligand clozapine-N-oxide (CNO) to induce VIP⁺ cell activation had fewer IL-22-producing ILC3s. This was accompanied by significantly increased C. rodentium translocation to the liver and spleen, and consequent mortality. These findings were confirmed using expression of an inhibitory DREADD whereby blockade of VIP⁺ cell activity reduced bacterial translocation to the spleen and liver. These intriguing studies leave much to be determined in understanding the contribution of this pathway to the regulation of mucosal immunity. In particular, there are at least four discrete types of neurons in the ENS that produce VIP, each with a unique functionality (1). While microscopy suggests that these VIP-expressing neurons are secretor motor neurons adjacent to tertiary lymphoid organs, it is not clear that DREADD-based activation or inhibition occurred in the GI tract. Neurons expressing VIP can be found in tissues ranging from the pancreas, where VIP exerts control over exocrine functions (109), to the suprachiasmatic nucleus in the brain that functions to entrain circadian rhythm (110, 111). Despite these concerns, these data illustrate that a number of neuroimmune circuits exist that exert control over host defenses, and demonstrate the need for further development of techniques to provide targeted stimulation of discrete neurons in the periphery. This need for precise identification and control of cell types while dissecting a presumed neuroimmune circuit is critically important, as neuron-associated cells such as glial cells have also been shown to be part of an immune regulatory network in the colon. Glial-derived neurotrophic factors, which induce development of the ENS, also induce IL-22 expression in ILC3s (112). These studies demonstrate the complex integrative physiology of neuroimmune circuits in the GI tract and will require unique and highly selective tools to identify the contributions of each component.

Conclusions. The crosstalk between the immune system and the nervous system is incredibly complex, and we have only begun to illuminate how the nervous system detects bacterial pathogens and modulates immune cell function. Neural immune communication has thus far been revealed to induce both host-protective and immune-inhibitory pathways that are exploited by pathogens. The sheer number and types of neurons in the ENS, and connections to extrinsic innervation, suggest that there are likely a number of neuroimmune circuits that have yet to be discovered. These pathways will have significant implications and present opportunities in the development of new treatment modalities for infections.