Communication between the nervous and immune systems can significantly alter immune cell function and response to inflammatory stimuli (1–5). Anatomically, neuroimmune interfaces have been discovered and studied within the spleen (6, 7), intestines (8, 9), the adrenal gland [HPA/cortisol (10) and dopamine reflex (11)], lymph nodes (12, 13), bone marrow (14), spinal column (15), pancreas (16), and heart/cardiovascular system (17). Both vagal and sympathetic afferent nerves have been shown essential in sensing and communicating peripheral immune status to the brain, which then modulates outflow to effector/immune cells within these interfaces to maintain homeostasis (18). Of these, the splenic neuroimmune interface has been intensively studied (1, 6, 7, 18, 19). Implant-based or pharmacological stimulation of this reflex, named the splenic cholinergic anti-inflammatory pathway (CAP) (3, 6, 7, 20, 21), has been shown to mediate control of specific cytokine production [including tumor necrosis factor (TNF)]. Mechanistically, it has been shown that vagus nerve stimulation results in norepinephrine (NE) release within the spleen (19). Splenic T cells are then modulated by the NE, and release acetylcholine (ACH) (1, 6, 7). This increase in splenic ACH results in inhibition of macrophage TNF production through alpha7 nicotinic acetylcholine receptor signaling (7). More recently, this anti-inflammatory pathway has been shown to be controlled by a cluster of cholinergic neurons within the dorsal motor nucleus (DMN), which projects to the celiac ganglion, and when activated results in the increased splenic nerve activity associated with cytokine/TNF inhibition (18). On the afferent side of the pathway, sensory neurons within the nodose ganglion (expressing specific cytokine receptors) have been shown to communicate cytokine specific nerve signals to the central nervous system (22, 23) and are hypothesized to modulate the activity (and thus level of cytokine suppression) of CAP. In addition, pain is likely sensed by nociceptors in infected tissue (such as lungs) and may also contribute to sensory modulation of neuroimmune pathways (24–26). Thus, significant evidence shows that these pathways contain the neuro-immune components to both sense inflammatory status (through afferent neurons) and respond to modulate cytokine production. Despite the extensive mechanistic investigation of the components of the splenic CAP, models used to study the effect of CAP activation have typically utilized systemic [i.e., severe endotoxemia (19)] or pathological [i.e., models of disease, such as collagen-induced arthritis (27)] inflammatory stressors. Therefore, the status of CAP activity during, and the contribution of the neuroimmune pathway response to, local/acute inflammatory stressors and infection remain unknown.

Several practical challenges have impeded the study of CAP stimulation during the progression of, and response to, local/acute infection. First, studies using implanted nerve stimulators [such as VNS (6, 18, 19)] are invasive and remain practically challenging, requiring surgical implantation of the device and long healing times to ensure that the inflammatory effects of the surgery do not impede the study. Second, use of electrical implants or pharmaceutical neuroimmune stimulators provide imprecise methods of neuroimmune modulation. For instance, cervical VNS implantation places the stimulator on a major vagus nerve trunk, potentially causing activation of many underlying neuroimmune (and other) reflexes during stimulation (3, 28, 29). Smaller electrical stimulators may enable near-organ implantation, and more precise activation of neurons entering specific organs (30, 31). However, low-invasive methods for laparoscopic implantation of these miniature stimulators are still under-development, and not available to the broader scientific community. In addition to implants, several pharmacological agonists of molecular components of neuroimmune pathways have also been discovered (32, 33). However, typical intravenous (i.v.) methods of administration often results in non-specific modulation of neuroimmune interfaces across the body (e.g., inhibition of MAPK signaling in both splenic and intestinal CAP locations using i.v. administration of semapimod), and/or require invasive methods of administration to modulate a specific neuroimmune interface (such as intracerebroventricular (i.c.v.) administration of semapimod).

Recently, our group (12, 28, 29, 34–38) and others (39–42) have shown the capability of pulsed ultrasound stimulation [targeted to either near organ (36, 43), whole organ (39, 40), or sub-organ anatomy (28)] to precisely modulate underlying nerve reflexes. In the spleen, the Okusa Lab (40–42) has shown that whole spleen ultrasound stimulation prevents ischemia-reperfusion injury (IRI) via the splenic CAP. Pathway specificity for splenic CAP (versus intestinal CAP or other neuroimmune pathways) was shown, as both physical splenectomy and chemical sympathectomy (with 6-OHDA) prior to ultrasound eliminated the protective effect. Furthermore, adoptive transfer of splenocytes from ultrasound-treated (but not sham) mice to naive mice was sufficient to protect recipients from IRI. In our group (28), splenic ultrasound reduced LPS-induced cytokine (e.g., TNF) release, and this protection was coincident with ultrasound-induced local but not systemic/plasma increases in CAP-related neurotransmitters. In addition, the ultrasound effect was not apparent in nude (lacking functional T cells), CD4 ChAT knock-out, or α7nAChR knock-out mice (key components of the CAP pathway). Splenic ultrasound was shown to have an equivalent effect on LPS-induced cytokine reduction but lack several of the associated off-target side-effects of cervical VNS, including effects on heart rate and metabolic function. Based on these previous findings, ultrasound-based CAP activation provides a new tool with which to study the effects of CAP stimulation over extended periods of time (i.e., days-weeks) non-invasively.

Many of the immunological steps associated with the progression of a local infection to a systemic and coordinated immune response are well-known. An initial innate immune response to infection includes upregulation of cytokines at the local site of infection, due to pathogen/inflammagen interactions or intracellular signaling with Toll-like and NOD-like receptors in immune cells, and activation of NF-κB-mediated (and other) intracellular signaling pathways (44). This local cytokine signaling then results in further leukocyte recruitment to the infected area. Upon elimination of the invading pathogen, these acute inflammatory responses become self-limiting, due to the release of resolvins and other anti-inflammatory molecules. However, failure to resolve the local infection or injury, dysregulation in the inflammatory response, and progression to unresolved and chronic inflammatory states can result in severe and lethal outcomes (including sepsis). It has previously been hypothesized that a physiological role of splenic CAP is to provide a protective mechanism that limits the potential negative effects of severe/systemic or chronic inflammation (1). This hypothesis has been tested by activating splenic or intestinal CAP prophylactically or immediately following severe infection or injury to provide protection prior to the challenge (1, 3, 6, 7, 18, 19, 40). These investigations have included models of endotoxemia (18, 21), sepsis (21), hemorrhagic shock (45), postoperative ileus (46), and kidney ischemia reperfusion injury (40). This hypothesis has also been tested in models of inflammatory disease [such as collagen- or serum-induced arthritis (27, 39) and chemical-induced colitis (35)], in which the CAP pathway is activated to provide an anti-inflammatory effect in the presence of chronic inflammatory disease. In each of these models, electrical (2), ultrasonic (28), or pharmaceutical (32) activation of CAP resulted in reduced cytokine production and disease-alleviating therapeutic effects.

However, there have been no studies investigating the effects of CAP activation during the progression and resolution of a localized and acute infection. Interestingly, it was recently shown that LPS challenge did not induce a TNF response in models of post-sepsis survival (i.e., animals surviving a septic episode that exhibit persistent immune impairment and immune fatigue, or compensatory anti-inflammatory response syndrome) (47). In these experiments, the TNF response to LPS could be rescued by first pharmacologically blocking the CAP pathway, demonstrating that suppression of the LPS-triggered TNF response in sepsis survivors was likely due to constitutive vagus nerve activation. We therefore hypothesized that the LPS response may also be suppressed during other physiological periods of natural CAP activation, such as the response to an acute infection prior to its resolution.

Herein, we applied a model of acute lung inflammation by first titrating the inoculum of S. pneumoniae during intra-tracheal instillation in rats (48), and then identifying the inoculum that resulted in positive bacterial lung cultures 16 hours following infection but did not result in systemic bacteremia (enabling symptom reduction (e.g., lower pain score) within 48 hours). We then investigated CAP status using the standard whole blood TNF response assay (7, 18, 20, 28) at multiple timepoints (i.e., 4-, 16-, and 48-hours) following infection. We measured a reduced cytokine response during infection, including complete suppression of the cytokine response at 16 hours (corresponding with the time of most severe symptoms and innate immune response markers). We next activated CAP using non-invasive ultrasound-based neuromodulation (28, 29, 34, 37, 39–42) and demonstrate that further stimulated CAP suppression of the TNF response was also time- or immune status-dependent during infection. That is, CAP stimulation did not result in further suppression of cytokine response during the intermediate phase (i.e., 16 hours following inoculation) of infection, but was observed immediately following instillation/challenge (i.e., hours) and returned within 48 hours. We also performed plasma cytokine and blood cell transcriptomic profiling and demonstrate that the 16-hour timepoint (in which CAP-mediated cytokine suppression was not observed) was associated with a peak in plasma CXCL1/KO response (neutrophil chemokine) and a transcriptional-shift in circulating immune cells.

It has been hypothesized that the physiological role of the cholinergic anti-inflammatory pathway (CAP) is to limit excessive systemic activation of the immune system, during response to infection (1, 3, 53). Evidence suggests that the nervous system is capable of sensing peripheral inflammation (through afferent neurons) and providing an integrated response to dampening the immune system (through efferent neurons) during periods of excessive inflammation (1, 5, 19, 22, 23). This protective effect has been shown by activating CAP prophylactically or immediately following injury in multiple models of severe inflammation or trauma, including sepsis, haemorrhagic shock, postoperative ileus, and kidney ischemia reperfusion injury (1, 6, 19, 21, 40, 45, 46). The protective effect has also been shown in models of chronic and pathological inflammation, including collagen-induced and serum transfer models of arthritis and DSS-induced colitis (20, 27, 35, 39). However, the activity of CAP, and the effect of stimulating the CAP pathway during progression of a local or acute infection has not previously been measured.

Herein, we utilized the whole blood cytokine response assay to monitor CAP activity [i.e., the cytokine response (20, 28)] at various timepoints following infection, and measured the effect of additional CAP stimulation (i.e., splenic ultrasound-based activation) at those same post-infection timepoints. The bacterial challenge was first demonstrated to result in positive lung culture (i.e., infection) 16 hours following infection, but did not result in systemic bacteremia, and enabled reduction in symptom measures (e.g., pain) within 48 hours. Weight loss was observed in the model, despite lack of bacteremia. However, weight loss during lung infection has been previously shown to correlate with cytokine concentrations in bronchial lavage (not systemic blood measures) (57). In addition, several lung specific mechanisms responsible for weight loss during lung infection are under-investigation including the effects of IL-22 upregulation in infected lungs (58), alterations in gut microbiota post-infection (59), and decreased food intake (57). Our results showed complete cytokine suppression in the whole blood response assay (i.e, maximum CAP activation) at 16 hours post-infection without any additional ultrasound-based CAP stimulation, which recovered to ~50% cytokine suppression (compared to non-infected controls) by the 48-hour timepoint. In non-infected animals, cytokine response was suppressed to the same level using the ultrasound stimulation procedure [previously shown to provide maximum CAP activation or full cytokine suppression in an LPS-induced inflammation model (28)]. Therefore, no additional cytokine suppression was observed after ultrasound-induced CAP activation 16-hours post infection (i.e., in animals already responding to the bacteria challenge and experiencing endogenous bacteria-induced CAP cytokine suppression).

The absence of cytokine suppression following exogenous CAP stimulation/activation has been reported in only one other previous report (47), that is, in animals surviving cecal ligation and puncture (CLP) induced polymicrobial sepsis. This previous report demonstrated that additional cytokine suppression following exogenous CAP stimulation/activation was not observed in the sepsis survivors due to constitutive endogenous activation of the CAP. That is, CAP was found to be pathologically active in the sepsis survivors (i.e., a potential mechanism associated with the prolonged post-sepsis state of immunosuppression and compensatory anti-inflammatory response syndrome). However, whether or not CAP is activated during a normal response to infection (i.e., an infection that resolves and does not result in systemic bacteremia or sepsis) has not been previously investigated. Herein, we demonstrate that cytokine response was suppressed during the progression of immune response to local/lung infection from initial challenge toward resolution. Furthermore, the level of suppression was time dependent (following the initial challenge) and reached a maximum level of suppression at the 16-hour post-challenge timepoint, which coincided with maximum bacteria-induced chemokine signaling and neutrophil mobilization.

Both blood cell counts ( Figure 3G ) and bulk RNA sequence profiling ( Supplementary Figures 5 – 8 ) data confirmed a strong immune response to the infection at the 16-hour timepoint, including an increase in the circulating neutrophil/leukocyte ratio and upregulation of genes associated with inflammatory response to bacteria and its regulation. In addition, there was an observed effect of ultrasound stimulation on several monocyte polarization- and plasmacytoid/dendritic cell maturation-related gene signatures (60–67) ( Supplementary Figures S7 – 9 ) in saline controls, which were all also observed in bacteria-instilled animals. This further validates the cytokine response data, which showed a maximal cytokine suppression (i.e., CAP activity) in bacteria challenged animals at 16 hours, and no additional effect of ultrasound stimulation at that time point. Future studies assessing gene expression at multiple time-points and measuring single cell/blood cell type specific gene expression during infection will be necessary to further examine the effect of CAP activity (both natural and stimulated) during the course of infection.

The whole blood cytokine response used herein, has become the standard assay for assessing CAP activation in both pre-clinical and clinical trials (19, 20, 28). However, we investigated CAP response in an acute infection model, and at multiple timepoints post-challenge, which have not previously been reported. These new experimental parameters led to the observation that CAP is activated during the normal course of immune response to a local/acute bacterial challenge, the level of activation is time dependent during the course of the immune response, and the maximal CAP activation coincides with the time period of maximal chemokine signaling and innate immune system mobilization. These observations are consistent with the long-standing hypothesis that the physiological role of CAP is to limit excessive and systemic activation of the innate immune system during an immune response (2, 53) but have not been previously measured. Despite this important initial observation in an acute infection model, several questions remain unresolved, and must be further addressed in future studies.

First, investigation of CAP activity across a larger range of challenge doses will be further informative. Based on the currently reported CAP hypothesis (i.e., that the efferent arm of CAP is modulated based on input signaling from cytokine and pathogen responsive afferents), the level and post-challenge timing of cytokine suppression/CAP activation should be expected to vary across different challenge doses. Furthermore, pharmacological blockade of CAP (28, 47) at pre- or post-challenge timepoints should modulate the pathogen effect on whole blood cytokine response. However, blocking of CAP may also affect the natural course of immune response and progression, and these experiments will require careful design (with respect to blockade versus challenge timing) to decouple those effects. The study herein focused on lung infection. However, afferent neurons project to many organs in the body, and the extent and time-course of CAP activation may be dependent on infection site location (22, 23). As an example, additional efferent arms of the anti-inflammatory pathway has recently been mapped within the intestinal tract (i.e., intestinal nerve pathways that modulate local macrophage activity and cytokine secretion independent of the splenic pathway) (8, 46). It is currently unknown if the site of local infection (and subsequent activation of different afferent and sensory neurons) results in differential activation of the splenic, intestinal, or other [i.e., the hypothalamic pituitary axis (HPA) or adrenal/dopamine] anti-inflammatory pathways. Simple observational studies, which measure the extent of activation of each of these different anti-inflammatory pathways by measuring their specific immunomodulatory effector [i.e., cortisol (HPA), dopamine (adrenal), TNF (intestinal vs. splenic/circulating)] under different inflammatory and immune challenges are warranted.

CAP activation [i.e., pharmacological (32), implant/electrical-based (20), or ultrasound-based (28, 39, 40)] has typically been studied in the context of testing a bioelectronic or neuromodulation-based therapy. For instance, implant/electrical-based CAP activation has been used to modulate cytokine levels in pre-clinical models of rheumatoid arthritis or irritable bowel disease (27, 68), and now in several human feasibility studies (20, 68, 69). In these studies, the underlying hypothesis is that CAP signaling is impaired in the disease and contributes to the elevated levels of cytokines (such as TNF) in the pathological state. In this respect, the mechanism associated with the CAP-based bioelectronic medicine (e.g., electrical/implant-based CAP activation, and cytokine/TNF reduction) is being investigated to replace a specific drug-based mechanism (i.e., cytokine TNF reduction) of pharmaceutical medications that are currently applied to the disease (i.e., anti-TNF based biologics). Based on the data we present herein, we speculate that the status of the CAP pathway (i.e., level of CAP activation) may also have further diagnostic relevance. We demonstrate that CAP is activated at different levels throughout the progression of the immune response to an infection, and that the whole blood cytokine response assay can be used to assess periods of CAP activity during this progression. Thus, the ability of the CAP stimulus to produce further cytokine suppression is indicative of the current immune state, and inflammatory response to the bacterial challenge. To further confirm the diagnostic ability of this test, we also calculated sensitivity and specificity, and created receiver operating curves (ROC). Indeed, the whole blood cytokine assay (pre-US sampling) has a capability to detect local infection with high sensitivity and specificity ( Figure 4A ). However, the diagnostic capability of the whole blood cytokine assay increased to near-ideal in the ROC of the change in whole blood cytokines (pre US – post US) ( Figure 4C ). This further indicates that ultrasound of the spleen adds additional value to the whole blood cytokine assay and can be utilized in the diagnosis of infections.

The whole blood assay has been the standard method of assessing CAP activity in past reports (1, 2, 7, 19, 20, 28). However, Supplementary Figure 10 shows additional observations we have made by replacing whole blood with PBMCs during the cytokine response assay. Macrophages have been reported as the main immune cell components of CAP. However, both macrophages and neutrophils (which are removed during processing of whole blood into PBMCs) are known to be endogenous sources of TNF, and activated neutrophils are known to modulate macrophages toward an “M1” or proinflammatory phenotype (70). In the PBMC response assay (i.e., lacking granulocytes, such as neutrophils), the total TNF response ( Supplementary 10A ) was greatly reduced compared to the whole blood TNF response in the no bacteria controls (Supplementary 2A). In addition, there was no additional suppression in TNF response after splenic ultrasound stimulation in the PBMC samples. This suggests that neutrophils are required for eliciting a maximal cytokine response from macrophages within the response assays, and that observation of the canonical CAP effect requires this neutrophil-macrophage interaction. Furthermore, at the 16-hour timepoint ultrasound stimulation/CAP activation resulted in an increased TNF response in the PBMC samples. This again demonstrates a rapid differential effect of CAP pathway stimulation in challenged/infected versus non-infected cohorts, and further supports the need to study these neuroimmune pathways within the context of different immune and inflammatory states (including the initial hours or pre-symptomatic stages of infection).

In summary, we present the first data examining CAP activity (as measured using the standard whole blood cytokine response assay) during progression and resolution of a local/acute bacterial infection. Our data shows a time-dependent level of CAP activation during infection, including low-level activation at infection onset (i.e., hours after challenge), maximum activation at the 16-hour timepoint post-challenge, and a decrease in activation during progression toward infection/inflammation resolution. We demonstrate that the response to additional ultrasound-based CAP activation is also dependent on the time-frame post-infection that the animal is stimulated. We further speculate that this non-invasive ultrasound tool for CAP activation can be utilized to further investigate the diagnostic utility of the status of CAP activity during infection and infectious disease progression.

The original contributions presented in the study are publicly available. This data can be found here: https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE197466.

The animal study was reviewed and approved by Institutional Animal Care and Use Committee (IACUC) at the Feinstein Institutes for Medical Research.

UA conceived and designed experiments, performed experiments, analyzed and interpreted experimental results, and wrote the manuscript. JG analyzed and interpreted experimental results. AD, OY, CM, and VC performed experiments, analyzed and interpreted experimental results. IM critically reviewed the manuscript. CD conceived and designed experiments, and reviewed the manuscript. SZ supervision, conceived and designed experiments, and analyzed and interpreted experimental results. CP supervision, conceived and designed experiments, analyzed and interpreted experimental results, and wrote the manuscript. All authors reviewed the manuscript. All authors contributed to the article and approved the submitted version.

Funding for this work was provided by a BARDA contract GE 75A50119C00056. This project has been funded in whole or in part with Federal funds from the Department of Health and Human Services; Office of the Assistant Secretary for Preparedness and Response; Biomedical Advanced Research and Development Authority, DRIVe, under this contract. The funder was not involved in the study design, collection, analysis, interpretation of data, the writing of this article or the decision to submit it for publication.

Author CP, JG, VC, and CM is/was employed by General Electric. SZ and UA have received previous financial support from General Electric.

The remaining authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.