The diversity of neuroimmune circuits controlling lung inflammation

Abstract

It is becoming increasingly appreciated that the nervous and immune systems communicate bidirectionally to regulate immunological outcomes in a variety of organs including the lung. Activation of neuronal signaling can be induced by inflammation, tissue damage, or pathogens to evoke or reduce immune cell activation in what has been termed a neuroimmune reflex. In the periphery, these reflexes include the cholinergic anti-inflammatory pathway, sympathetic reflex, and sensory nociceptor-immune cell pathways. Continual advances in neuroimmunology in peripheral organ systems have fueled small-scale clinical trials that have yielded encouraging results for a range of immunopathologies such as rheumatoid arthritis. Despite these successes, several limitations should give clinical investigators pause in the application of neural stimulation as a therapeutic for lung inflammation, especially if inflammation arises from a novel pathogen. In this review, the general mechanisms of each reflex, the evidence for these circuits in the control of lung inflammation, and the key knowledge gaps in our understanding of these neuroimmune circuits will be discussed. These limitations can be overcome not only through a better understanding of neuroanatomy but also through a systematic evaluation of stimulation parameters using immune activation in lung tissues as primary readouts. Our rapidly evolving understanding of the nervous and immune systems highlights the importance of communication between these cells in health and disease. This integrative approach has tremendous potential in the development of targeted therapeutics if specific challenges can be overcome.

INTRODUCTION

Bidirectional communication between the immune and nervous systems serves to initiate and refine host responses to immunological challenges. This communication between neurons and immune cells, or neuroimmune circuits, has been identified in primary and secondary lymphoid organs (1–6) and various organ systems. In the lung, neuroimmune circuits regulate the inflammatory response to bacterial and viral pathogens, as well as modulate the allergic responses (7–11). Although the function of these neuroimmune circuits has been predominantly characterized in the context of systemic inflammation (12–14), common mechanisms appear to regulate a diverse array of immune cells in different tissues (8, 15–20). This commonality is most likely due to the extensive innervation throughout the body by the peripheral nervous system, and the expression of neurotransmitter receptors by immune cells that facilitate the control of inflammation (21, 22). The highly specialized innervation regulates physiology in various organ systems as a function of neuroanatomy, with terminal axons releasing specific neurotransmitters in the target organ. These key features have been shown to regulate immune responses, highlighting the potential for targeted therapeutic intervention with neuronal stimulation by exploiting neuronal circuits that innervate specific tissues (22). Indicative of the therapeutic potential for this approach beyond systemic inflammation, neuronal stimulation reduces inflammation in preclinical models of postoperative ileitis (23), colitis (24), acute kidney injury (25, 26), ventilator-induced lung injury (VILI) (9), and acute respiratory distress syndrome (ARDS) in the lung (27). These promising results are based on the underlying assumption of a common neuronally evoked mechanism to limit inflammation; however, protection afforded by neuronal stimulation can be achieved through several neuronal pathways, differentially modulating immune cell types in the same target organ. As an additional complication, pathogens have also been identified to exploit these neuroimmune pathways to reduce host protective immunity (8), demonstrating the need for pathway-selective targeting. These concerns have not prevented limited clinical trials from demonstrating the efficacy of neurostimulation in the treatment of inflammatory diseases, including Crohn’s disease (28, 29) and rheumatoid arthritis (30). With these successes, and the emergence of SARS-CoV-2 as a novel respiratory virus that can induce significant morbidity and mortality as a consequence of inflammation, unfounded suggestions for the use of neurostimulator devices, such as vagus nerve stimulation (VNS), have been proposed as a possible therapeutic (31). This review seeks to introduce some of these pertinent issues and highlight the continued significant gaps in our understanding of neuroimmune circuits in the control of lung inflammation.

NEUROIMMUNE CIRCUITS

Interest in the impact of neuroimmune cross communication on host physiology has been reinvigorated in recent years. As there is tremendous diversity in the types of neurons in the peripheral nervous system, and in their innervation of organs, it should not be surprising that these neuroimmune circuits and the functional outcomes depend on the neuronal type, tissue innervated, and immune cell targets. It is well established that neurons express receptors for immune cell-derived molecules that can change host physiology and drive pathology. Sensory nociceptors can be directly activated by these proinflammatory mediators, in diverse organ systems ranging from the skin (32) to the lung (33, 34). The close proximity of immune cells to neurons has been well established in the lung to exert control over physiological processes during pathology. Neuronal activation and sensitization can occur in response to histamine, leukotrienes, and proteases released following the activation of mast cells, eosinophils, and neutrophils (35–38). In the lung, the consequences of immune cell-induced neuronal sensitization can include increased airway hypersensitivity during infection or allergic airway inflammation (7, 39). Physical contact between dendritic cells or T cells and neurons has been shown, highlighting the potential for communication between immune cells and the innervation at mucosal tissues such as the lung (40). Here we review the evidence for the regulation of lung inflammation by neuronal signaling.

Expression of Neurotransmitter Receptors by Immune Cells

Fundamental to the control of immune responses by neuronal signaling is the expression of neurotransmitter receptors by immune cells. Ionotropic receptors including nicotinic AChRs in addition to metabotropic G-protein-coupled receptors have recently been described to regulate key immune cells in both innate and adaptive populations. Complicating our understanding of neurotransmitter signaling that alters immune cell function is the structural homology of related receptors and G-protein-coupled receptors (GPCR), which complicates the production of specific antibodies (41). This homology means that studies assessing GPCR expression on immune cells, using antibodies that have not been validated using target gene knockout samples or other biological negatives to demonstrate specificity, should be viewed with some degree of skepticism. In the absence of this information, excellent resources such as RNAseq and single-cell SEQ databases [e.g., the Immunological Genome Project (42)], CellxGene (43), and antibody databases such as the Human Protein Atlas (44) can be used to help assess if specific genes are expressed by a cell type of interest. When using these references, low expression, sequencing depth, and the number of cells analyzed should be carefully considered as these critical factors could result in a “no-call” to be misinterpreted as a gene is not expressed. Here, we discuss select neurotransmitter receptors that are integral to neuroimmune circuits. Readers interested in comprehensive listings of neurotransmitter receptor expression by various immune cells are directed to recent in-depth reviews (45–48). In addition, we have provided a table highlighting critical ligands, receptors, and outcomes of activation on specific populations of immune cells for further clarification (Table 1).

Table 1.

Listing of neurotransmitter ligands and receptors present on immune cell populations

Class	Subtype	Cell Types	Inflammation
Muscarinic acetylcholine	M1 Gq/11	Dendritic cells (53) *mouse	ND
		Monocytes (54)	ND
		Alveolar macrophages (54)	No effect (55)
	M2 Gi/o	Dendritic cells (53) *mouse	ND
		Monocytes (54)	No effect (54)
		Alveolar macrophages (54)	No effect (54, 55)
		Neutrophils (56)	ND
	M3 Gq/11	Dendritic cells (53)	↓ TNFα (57)
		Monocytes (54)	No effect (54)
		Alveolar macrophages (54, 58)	↑LTB4, ↓NF-κB signaling (55)
	M4 Gi/o	Dendritic cells (53)	↓ TNFα (57)
		Monocytes (54)	ND
		Neutrophils (56)	ND
		Alveolar macrophages (54)	ND
	M5 Gq/11	Dendritic cells (53)	↓ TNFα (57)
		Monocytes (54)	ND
		Alveolar macrophages (54)	ND
		Neutrophils (56)	ND
Nicotinic acetylcholine	α2	Neutrophils (59)	ND
		Dendritic cells (53)	ND
		Monocytes (60)	ND
	α3	Neutrophils (59)	ND
	α4β2	Neutrophils (59)	ND
		Alveolar macrophages (61)	↓ IL-6, IL-12, TNFα (61)
	α3β2	Neutrophils (59)	↑ IL-8
	α4	Neutrophils (59)	↑ IL-8
	α5	Neutrophils (59)	ND
		Dendritic cells (53)	ND
		Monocytes (60)	ND
	α6	Neutrophils (59)	ND
		Dendritic cells (53)	ND
		Monocytes (60)	ND
	α7	Monocytes (62)	↓ TNFα, IL-1β, IL-12 (63)
		Alveolar macrophages	↓ TNFα, MIP2, HMGB1 (123)
		Neutrophils (59, 62)	↓ TNFα, MIP2, HMGB1 (123)
		Dendritic cells (53)	Inhibit antigen processing (64)
	α9	Neutrophils (59)	ND
		Monocytes (60)	↓TNFα, IL-1β, IL-12 (63)
	α10	Dendritic cells (53)	ND
		Monocytes (60)	ND
α Adrenergic	α1	Monocytes (65)	↑ IL-1β, ↑ p38 MAPK activation (66), ↑ complement (C2) synthesis (67), ↓ TNFα, IL-8, MIP1B (68)
		Neutrophils (69)	↑ proliferation (70)
		Dendritic cells (71, 72)	↓ mobility (71)
		NK cells (73)	↑ cytotoxicity (74)
	α2	Neutrophils (69)	NE (70)
		Dendritic cells (75)	↑ mobility (71)
		NK cells (73)	↑ cytotoxicity (74)
	α3	Neutrophils (69)	ND
β Adrenergic	β1	Monocytes (76, 77)	↑ cAMP, ↑ IL-1β (77)
		Neutrophils (69)	Inhibition of migration (78)
		Dendritic cells (75, 79)	ND
	β2	Alveolar macrophages (80)	↓ TNFα, ↓ JNK phosphorylation (81, 82), ↑ IL-10 (82, 83)
		Monocytes (76)	↓ TNFα, ↓ GM-CSF (84)
		Neutrophils (85, 86)	↓ IL-8 induced chemotaxis (86), ↓ netosis (87)
		Dendritic cells (72, 75)	↓ TNFα, ↑ IL-10 (83)
		NK cells (73)	↑ proliferation (88)
		Eosinophils (89)	↓ respiratory burst (90), LTC4 (91), No effect on degranulation (89)
		Basophils (92)	Inhibit histamine release (93)
		Mast cells (94)	↓ histamine (95–97) / leukotriene (98) release
	β3	Neutrophils (69)	ND

Ligand	Receptor	Cell Types	Inflammation
SP	NK1R	Neutrophils (99)	↑ MIP-1a, MIP-2, CCR1, CXCR2 (100)
		NK cells (101)	↓ cytotoxic activity, degranulation, ↓ phospho-ERK (102)
		Monocytes (103)	↑ NF-κB expression, NLRP3 (104)
		Mast cells (105)	↑ mast cell accumulation (106) Degranulation (105)
		Dendritic cells (107)	Modulate T-cell proliferation (107), ↑ DC survival (108)
		Cell type not specified	Pro-oncogenic (109)
	NK2R	Airway dendritic cells (110, 111)	↑ type I IFN expression (110)
		Airway eosinophils (112)	↑ survival (113)
CGRP	RAMP1/CLR	Alveolar macrophages (114)	↓ TNFα, FcγR-mediated phagocytosis, ↓ pro-IL-1β (115)
		Neutrophils (114)	Granule secretion (116)
		Dendritic cells (117)	↓CCR2 and CCR7 expression (118)
		Monocytes (114)	↓ MCP-1, TNFα, ↑ IL-6 (149)
		Mast cells (150)	NE on degranulation (105)
		Eosinophil precursors (151)	↓ recruitment (152)

ND, not detected.

Cholinergic Anti-Inflammatory Pathway

The cholinergic anti-inflammatory pathway (CAIP) is a well-established neuroimmune circuit that regulates inflammation in a variety of organ systems. Similarly, selective efferent VNS, achieved by transection of the vagus and application of electrical stimulation distal to the cut site, significantly inhibits macrophage activation and consequently reduces TNFα production in the spleen and select lymph nodes. Immunomodulatory effects evoked by VNS, however, are indirect, since there is no direct vagal innervation of the spleen or lymph nodes. Instead, vagal efferent fibers activate sympathetic neurons in the superior mesenteric/celiac ganglia complex that project into the spleen and MLN (45). Sympathetic activation results in localized norepinephrine (NE) release and induces a unique T-cell population that expresses the enzyme choline acetyltransferase (ChAT) to release acetylcholine (ACh) (2, 4, 119). This T-cell-derived ACh binds to the nicotinic α 7 receptor (α7R) on macrophages to suppress proinflammatory cytokine production by inhibiting NF-κB (14, 23, 120). Together, these cells constitute the CAIP (Fig. 1), and numerous studies have highlighted the utility of activating this circuitry in models of kidney ischemia-reperfusion injury (27) and intestinal inflammation (23, 24). Although VNS has demonstrated great potential in these preclinical studies and early clinical trials, great care should be taken in ascribing an anti-inflammatory effect of VNS to activation of the CAIP. Recent studies have revealed several neuroimmune circuits, each with unique components, which converge in the spleen and lymph nodes and are capable of regulating immune cells. It should be noted that the composition of the vagus nerve in most animal species is predominately afferent fibers. With respect to the lung, the bronchial vagal afferents are ∼60% afferent and 40% efferent (121), and although the effects of CAIP have been ascribed to activation of vagal efferent fibers, current vagal nerve stimulators activate both efferent and afferent signaling. Highlighting the importance of understanding the circuits activated by VNS, we have shown that blockade of LPS-induced inflammation by efferent VNS required ChAT⁺ T cells, whereas afferent selective VNS prevented inflammation independently of these specialized T cells (119). Afferent VNS activates a neuroimmune circuit that reduces immune regulation, dependent on β2 adrenergic receptors (β2AR) signaling but independent of the CAIP (119). In addition, afferent VNS has been shown to induce an efferent splanchnic anti-inflammatory mechanism, capable of suppressing LPS-induced proinflammatory cytokine production systemically (122) (Fig. 2). With this in mind, it is possible that prior reports of VNS-induced regulation of inflammation may occur independently of CAIP, through activation of other unknown neuroimmune circuits. We propose that VNS-mediated control of inflammation should not be ascribed to the CAIP unless there is sufficient evidence for components of this neuroimmune circuit such as ChAT⁺ T cells being required.

Figure 1.

A schematic illustration of the cholinergic anti-inflammatory pathway (CAIP) in the spleen. Vagal efferent fibers activate sympathetic neurons in the superior mesenteric ganglia (SMG) or celiac ganglia (CG) that project into the spleen. This leads to a local release of norepinephrine, which induces choline acetyltransferase (ChAT)-positive T cells to release acetylcholine. Acetylcholine in turn activates macrophages by binding to the nicotinic α 7 receptor (α7nChR), which inhibits the production of proinflammatory cytokine, such as TNFα and IL-6. Together, these cells constitute the CAIP, and selective efferent VNS modulates inflammation in the spleen via this pathway. It remains elusive if the same neuroimmune circuit is present in the lungs. [Image created with BioRender.com and published with permission.]

Figure 2.

The organization of sympathetic innervations in the spleen and the lungs. Cell bodies of sympathetic preganglionic neurons illustrated here (red) are in the thoracic spinal cord between level T1 and T12. Axons from preganglionic neurons (located in T1 to T5) synapse with postganglionic neurons (blue) at the paravertebral ganglia within the sympathetic chain. Axons from these postganglionic neurons travel to and innerve the lungs. The immunomodulatory effect of this pathway remains unknown. In contrast, axons from preganglionic neurons (located in T6 to T8) enter the greater splanchnic nerve, synapse with postganglionic neurons at the celiac ganglion, and innervate the spleen. Activation of this sympathetic splanchnic pathway is known to suppress inflammation in the spleen. [Image created with BioRender.com and published with permission.]

Evidence for the CAIP in the lung.

Although there is growing evidence for neuroimmune control of lung inflammation, a critical distinction should be made between the CAIP and other neuroimmune circuits. It is exciting to consider that prior literature describing the regulation of inflammation in the lung may have incorrectly ascribed these effects to CAIP, instead of other novel neuroimmune circuits activated by VNS. The vagus nerve is critical in mediating lung inflammation, with vagotomy significantly increasing VILI and inflammation in rats, as indicated by increased IL-6 production and neutrophil recruitment compared with surgical controls (9). Further supporting a lung anti-inflammatory role of the vagus, efferent selective VNS reduced IL-6 production and granulocyte recruitment (9). Although α7R expression and signaling in bronchial epithelial cells were investigated in vitro, these studies did not demonstrate that the anti-inflammatory effects of VNS required α7R to suppress cytokine production in vivo (9). It is unclear from these studies if the effects of VNS required α7R expressed on lung epithelial cells, macrophages, ChAT⁺ T cells, or the spleen/lymph nodes as components of the CAIP.

Further evidence for a vagal anti-inflammatory pathway in the lung is suggested by the observations that vagotomy increases inflammation during bacterial pneumonia (123). Although these effects were proposed to be dependent on the α7R, this conclusion is based on reduced inflammation after administration of selective α7R agonists, or increased inflammation induced by E. pneumonia in wild-type (WT) compared with α7R^−/− mice, without assessing the contribution of the vagus to either effect (123). Increased inflammation has also been observed in α7R^−/− mice or mice with right cervical vagotomy during influenza A infection (124). More recently famotidine, an H2 histamine receptor antagonist, reduced inflammation during systemic endotoxemia that was abrogated when the mice underwent a bilateral vagotomy (125). In each of these studies, it remains unclear if the vagal anti-inflammatory effects were due to CAIP, or another anti-inflammatory neuroimmune pathway. It should be noted that experiments using α7R agonists to reduce inflammation during bacterial pneumonia (123) or ventilator-induced injury and inflammation (126) are not direct evidence of the CAIP in the lung, but are instead the direct effects of these compounds, and are in agreement with previously identified anti-inflammatory roles of nicotinic receptors.

Regulation of Lung Inflammation by Catecholamines and Sympathetic Innervation

The release of catecholamines (epinephrine, norepinephrine, and dopamine) occurs following sympathetic neuronal activation, or from chromaffin cells in the adrenal medulla. Although there is abundant sympathetic innervation of the lung, historically this innervation is thought to be associated with the vasculature with few fibers identified beyond the lung hilus in mice and rats (127). Although this would suggest that these fibers are not in close physical proximity to the majority of immune cells, ablation of lung sympathetic innervation by surgical denervation or chemical sympathectomy significantly enhanced LPS-induced lung inflammation (128). It is important to note that, although chemical sympathectomy reduced tyrosine hydroxylase staining in the lung but not in the spleen, thymus, or bone marrow, the adrenal medulla was not assessed (128). These data suggest that chemical sympathectomy was restricted to the lung, but do not exclude potential off-target effects from the adrenal glands. Nevertheless, these exciting results demonstrate that alveolar macrophages (AMs) and innate lymphoid cells (ILC) express β2AR and that NE can block the production of proinflammatory cytokines induced by LPS or IL-33 (128).

As evidence of the complexity in the regulation of immune responses, proinflammatory roles for β2AR have also been described (129). Inhalation of particulate matter (PM) typical in air pollution increases the expression of proinflammatory cytokines, such as IL-6, by AMs and consequently induces thrombosis. This inflammatory state is dependent on PM-inducing catecholamine release and activation of β2AR on macrophages. Observations supporting the critical role of AMs include prevention of PM-induced inflammation on chlodronate liposome intratracheal administration (130), as well as significantly reduced IL-6 in LysM.Cre β2AR^f/f mice compared with controls (131). However, as neither approach selectively targeted AMs, it cannot be conclusively determined that this effect is solely restricted to AMs.

Although it is uncertain how β2AR signaling contributes to airway inflammation, noncanonical β arrestin-mediated signaling in AM appears to play a significant role in the pathogenesis of allergic airway inflammation. In β2-arrestin (ARRB2)^−/− mice, sensitization and subsequent exposure to ovalbumin (OVA) reduced T-cell chemotaxis, recruitment, and Th2 cytokine production in the lung (e.g., IL-4, IL-5 production) compared with WT mice. This reduction in immune-mediated pathology was not associated with reduced OVA-specific IgE, or with simply increased production of Th1 type cytokines. Moreover, acute inflammation following LPS challenge was identical in ARRB2^−/− and WT mice (132). These data suggest that GPCR activation leading to ARRB2 binding can be proinflammatory in specific models of inflammation and unique immune cells and that this aspect of neuronal control of inflammation will require additional careful study.

Control of Lung Inflammation by Sensory Neurons

Similar to many organs, the lung is densely innervated with sensory neurons that induce physiological changes in response to noxious stimuli such as mechanical injury and inflammation (133, 134). The innervation consists of unmyelinated C-fibers arising from the dorsal root ganglia (DRG) in the spinal cord and specialized vagal afferent nociceptive neurons originating from the nodose and intracranial jugular ganglia (34, 135). Anatomically these nociceptors are extensively branched, similar to somatosensory nociceptors, and can be found from the conducting airways through to alveoli (34, 135). In these regions of the lung, vagal nociceptive afferents can form close associations and protrude into clusters of pulmonary neuroendocrine cells (PNECs) in a neuroendocrine bundle (136). This bundle is in contrast to the sensory nociceptors from the DRG that form plexi adjacent to PNECs (136), suggesting that each type responds to divergent stimuli. Activation of these neurons through the nociceptive transient receptor potential cation channel subfamily V member 1 (TRPV1) triggers the release of peptide neurotransmitters such as substance P, calcitonin gene-related peptide, and neurokinin-A. This localized neurotransmitter release is the foundation of neurogenic inflammation, resulting in chemotaxis of immune cells (137), vasodilation (138), increased endothelial permeability (139), and adhesion molecule expression (140). In addition to these classically defined roles, sensory nociceptive neurons have recently emerged as playing a critical role in the coordination of host protective immunity during bacterial infection of the lung or skin, highlighting a previously unappreciated role of neuroimmune communication in maintaining host protection (141, 142).

Despite the well characterized immune cell recruitment and activation during neurogenic inflammation, the mechanistic role of sensory nociceptive neurons in coordinating mucosal immune responses during infection in the lung have only begun to be elucidated. Elegant experiments have demonstrated that Staphylococcus aureus activates TRPV1⁺ vagal nociceptors to suppress host protective responses by inhibiting neutrophil and γδ T-cell recruitment (8). This effect was attributed to CGRP release by TRPV1⁺ nociceptors, as a CGRP peptide antagonist improved survival following lethal S. aureus challenge (8). These results are in keeping with the activation of TRPV1⁺ somatosensory neurons by the bacterial pathogen Streptococcus pyogenes in the skin, which inhibits neutrophil recruitment in a CGRP-dependent manner (142). Although these results suggest TRPV1⁺ nociceptors could function generally as inhibitory neuroimmune circuits, TRPV1⁺ neurons are also required for host protection during infection with the enteric bacterial pathogens Citrobacter rodentium and Salmonella enterica serovar Typhimurium (143, 144). Together, these results suggest that the response of a neuroimmune circuit comprised of sensory nociceptors could depend on the type of pathogen.

Sensory nociceptors have also been described to contribute to the development of allergen-induced airway inflammation (35). The transient receptor potential channel TRPA1, which functions as a sensor of both mechanical stress and reactive compounds, such as cinnamaldehyde and hydrogen peroxides, is often coexpressed with TRPV1 by nociceptors (145). Highlighting the divergent roles of these channels in allergic lung inflammation, mice deficient for TRPA1 but not TRPV1 showed a significant reduction in immune cell recruitment into the lung, Th2 type cytokine production, goblet cell hyperplasia, and mucus production in the OVA-sensitization and challenge model (35). The significantly reduced antigen-induced airway inflammation in TRPA1^−/− mice was proposed to be due to reduced release of neuropeptides including CGRP and neurokinin-A in the lung and could be recapitulated using WT mice treated with a TRPA1 selective antagonist (35). Thus, although depletion of sensory neurons by targeting TRPV1⁺ neurons can result in functional changes, one should be careful in ascribing these changes to this particular receptor.

OUTLOOK/POTENTIAL FOR NEURAL STIMULATION IN THE TREATMENT OF LUNG INFLAMMATION

The preclinical advances in understanding of neuroimmune circuitry and, in particular, of VNS-induced suppression of inflammation have led to a push for clinical translation in a variety of inflammatory conditions including COVID-19 (31, 49, 146–148). Despite these successes, it is informative to look at each of the study outcomes, noting differences depending on the site of stimulation and the VNS parameters used. Clinical VNS is typically provided by surgical implantation of an electrode on the cervical vagus, transcutaneous electrical nerve stimulation (TENS) unit, or an electrode on the concha of the ear, which provides stimulation of the auricular branch of the vagus (50). As each paradigm is unique it should not be surprising that different parameters are required. For example, direct contact between the nerve and electrode requires significantly less current to evoke stimulation compared with TENS or auricular devices. These differences in VNS techniques can have significant effects on the ability to selectively induce vagal nerve activation and clinical outcome. Nonetheless, therapeutic effects of VNS are clear in diseases such as rheumatoid arthritis, with significant reductions in inflammatory cytokines, C-reactive protein (CRP), and disease activity scores achieved in patients implanted with a helical electrode to provide cervical VNS (30). Remarkable reductions in clinical disease (primary end point) and serum TNFα have also been observed in a proof-of-concept study using noninvasive auricular VNS (51). With the unique VNS site and parameters selected, it is, however, uncertain if this effect is due to the same mechanism of action. At present, it is unclear how much these factors contribute to VNS nonresponsive patients. A thorough understanding of the parameter space, time from disease onset, and when in the course of disease treatment VNS should be applied are desperately needed in the field.

Although there are several registered VNS clinical trials in the treatment of both acute COVID-19 (NCT04368156 NCT04379037, NCT04514627, NCT05058742) and long COVID-19 (NCT05225220), peer-reviewed results have not been published. At present, results from small open-label studies and preprint publications have suggested mixed benefits of VNS in acute COVID-19. Although limited cases reported improved patient symptoms with transcutaneous VNS (49), a randomized control trial of patients with COVID-19 (47 VNS, 50 control patients) found significantly reduced inflammation by serum CRP, and procalcitonin with VNS treatment compared with standard of care controls (52). Despite these promising findings, no benefit was observed in respiratory or clinical outcomes. These data suggest that VNS may have some therapeutic potential; however, the identification of individual parameters or time for treatment are critical factors. Confirmation of neuronal engagement, such as monitoring heart rate variability (HRV), during stimulation does not appear in many protocols, making it unclear if the lack of benefit was due to a lack of biological effect or inappropriate stimulation parameters. Although no adverse events were reported in these studies, the use of a neural stimulation device without understanding the basic mechanisms of action in the lung is concerning. From preclinical studies with bacterial pathogens where vagal sensory nociceptors in the lung prevented clearance of the pathogen in a host maladaptive process, it is unclear if VNS could engage these circuits as well, preventing desired host responses to viral infection. These uncertainties argue for continued research into the effects of VNS, not only in mice but in other animal species with similar neuroanatomy to humans, with a consideration of the parameters used to evoke neuronal action. Finally, careful control and reporting of these parameters, site of stimulation, and confirmation of neuronal engagement should be undertaken. We would further propose that a closed-loop system whereby HRV is monitored in real-time while adjusting VNS parameters could be used to elicit maximal engagement while minimizing undesired effects.

CONCLUSIONS

Communication between the nervous and immune systems evoked by inflammation or pathogens can significantly affect the host immune response. Although exogenous exploitation of these circuits (typically in the form of VNS) reduces inflammation in a multitude of immunopathologies including VILI-ARDS in the lung, continued research is required to identify new neuroimmune circuits and refine existing ones. These gaps in our knowledge are significant and attempting clinical translation of these approaches when critical parameters remain unknown risks complete failure and abandonment of the technique and its therapeutic potential. These challenges are not insurmountable but instead should instruct and inspire future research in neuroimmunology and neuroimmune circuitry in the treatment of lung inflammation.

GRANTS

This work was supported in part by National Institute of Allergy and Infectious Diseases Grants R01 AI150647 and R21 AI148188 (to C. Reardon) and National Institute of General Medical Sciences Grants T32 GM099608 and GM144303 (to S. Schreiber).

DISCLOSURES

No conflicts of interest, financial or otherwise, are declared by the authors.

AUTHOR CONTRIBUTIONS

E.X.Y.T. prepared figures; M.C., S.S., and C.R. drafted manuscript; M.C., S.S., K.M., and C.R. edited and revised manuscript; C.R. approved final version of manuscript.