Abstract
The hallmark of inflammatory bowel diseases (IBD) is chronic intestinal inflammation with typical onset in adolescents and young adults. An abundance of neutrophils is seen in the inflammatory lesions, but adaptive immunity is also an important player in the chronicity of the disease. There is an unmet need for new treatment options since modern medicines such as biological therapy with anti-cytokine antibodies still leave a substantial number of patients with persisting disease activity. The role of the central nervous system and its interaction with the gut in the pathophysiology of IBD have been brought to attention both in animal models and in humans after the discovery of the inflammatory reflex. The suggested control of gut immunity by the brain–gut axis represents a novel therapeutic target suitable for bioelectronic intervention. In this review, we discuss the role of the inflammatory reflex in gut inflammation and the recent advances in the treatment of IBD by intervening with the brain–gut axis through bioelectronic devices.
Introduction
Inflammatory conditions in the gut are of rising concern and the prevalence of inflammatory bowel diseases (IBD) is estimated at 1.6 million and up to 3 million patients in the USA and Europe, respectively (1–3). The pathophysiology behind chronic gut inflammation is not fully known, but intestinal dysbiosis as well as immune system dysfunction are likely of key importance. There is a growing body of evidence for a role of the central nervous system (CNS) and its interaction with the gut in the pathophysiology of IBD as has been discussed both in relation to animal models and in human clinical trials (4–7). The brain–gut axis involves complex two-way communication of metabolic, hormonal and neural signals (8). There is compelling evidence for the involvement of the intestinal microflora as well as gut immune dysfunction in this axis of regulation (4, 8, 9).
The role of the intestinal microflora includes microbial production of short chain fatty acids (SCFAs), which have impact on the CNS effects on energy homeostasis through free fatty acid receptors (FFARs) and intestinal gluconeogenesis (9–11). The activation of FFARs by SCFAs appears to be of key importance in the connection between SCFAs and the CNS by promoting the production of local cholecystokinin (CCK), peptide tyrosine tyrosine (PYY) and glucagon-like peptide-1 (GLP-1), which can influence the hypothalamus to modulate appetite and energy homeostasis (12–16). SCFAs and butyrate, in particular, also stimulate the hypothalamic–pituitary–adrenal (HPA) axis directly (17). Moreover, butyrate also activates gene expression involved in intestinal gluconeogenesis in enterocytes, a mechanism suggested to control food intake by inducing satiety (9, 18).
This connection between intestinal SCFAs and the CNS is thought to inform hypothalamic nuclei to control energy intake by detection of increased upstream portal glucose levels by specific neurons adjacent to the portal veins (10, 19–21). Furthermore, metabolism of digested fibers by luminal bacteria is suggested to result in intraluminal biosynthesis of neurotransmitters such as γ-aminobutyric acid, serotonin, dopamine and noradrenaline. Local accumulation of these neurotransmitters has the potential to affect synaptic activity in afferent enteric neurons and thereby influence signals to CNS, even though this remains to be proven (13, 22–26).
A putative connection between the brain–gut axis and intestinal immunity has been suggested through the discovery of the inflammatory reflex. The gut immune system has traditionally been described as circulating hematopoietic cells recruited to the intestine and activated through cell–cell interaction and cytokine/chemokine signaling. However, the discovery that functions of both innate and adaptive immune cells can be modulated by the CNS via complex bidirectional innervation, a mechanism described as the inflammatory reflex, draws the attention to the brain–gut axis as a potential regulator of gut immunity (27–34). The bowel represents one of the largest sites for immune interactions in the body, and potential targets for treating inflammatory conditions in the gut may be found in the complex innervation of the intestine (35). Further, neural control of physiological processes can be particularly suitable for therapeutic intervention by bioelectronic devices, which has been demonstrated in the treatment of epilepsy and pain by electric stimulation of nerves.
In this review, we provide an overview of the role of the inflammatory reflex in gut inflammation and discuss the recent advances in the treatment of IBD by intervening with the brain–gut axis through bioelectronic devices.
Inflammatory bowel diseases
IBD include the two disease entities Crohn’s disease (CD; segmental deep gut wall inflammation throughout the gastrointestinal tract) and ulcerative colitis (UC; continuous superficial colonic inflammation). The conditions particularly affect young persons with a peak incidence between 15 and 30 years of age and the hallmark of IBD is chronic inflammation with flares of disease involving innate as well as adaptive immunity with an abundance of both neutrophils and T cells in the inflammatory lesions (5, 36–38).
Genome-wide association studies have identified a number of IBD-associated genetic loci involved in intestinal barrier integrity, innate and adaptive immunity, cell migration and autophagy (39). Nucleotide-binding oligomerization domain 2 (NOD2) has the most studied genetic polymorphisms that have been associated with CD. NOD2 belongs to the cytosolic Nod-like receptor family and plays an important role in antigen-presenting cells as well as epithelial cells by recognizing muramyl dipeptide of the wall of gram+ and gram– bacteria (40). Through nuclear factor κB (NF-κB) and mitogen-activated protein kinase (MAPK) signaling pathways, immune cells are activated by NOD2 after exposure to intracellular bacteria (39, 41, 42). NOD2 polymorphisms associated with CD hamper this activation of NF-κB, i.e. loss-of-function defects in the local immune response may trigger the continuous inflammation (40).
An additional intracellular mechanism of host defense is autophagy, a mechanism dependent on the gene ATG16L1, which encodes a key molecule in the clearance of intracellularly invading microbes (40, 43–45). Autophagy upholds cellular function by trapping pathogens and other detrimental cytoplasmic elements in autophagosomes that merge with lysosomes for degradation (40). This process is also important for the ability of antigen-presenting cells to activate antigen-specific CD4+ T cells through major histocompatibility complex class II (MHCII), and polymorphism of ATG16L1 is associated with CD (46–48). In addition, a direct link has been identified between a NOD2 genetic polymorphism and dysfunctional autophagy, which underlines the importance of NOD2/ATG16L1 for the balance between luminal antigens and gut immunity (40).
Another suggested mechanism associated with IBD pathogenesis is single-nucleotide polymorphisms in the IL-23-receptor, which implicates Th17 cells since IL-23 is required for Th17 cell differentiation (49). However, even though many European CD patients carry at least one risk locus, only a small proportion of the general population with the described genetic variations actually develop IBD, which further illustrates the multifactorial mechanisms that underlie this inflammatory condition (50).
The luminal microbiota are clearly a major trigger for IBD. Chronic intestinal inflammation can only be induced in animal models in the presence of gut intestinal microflora and diversion of the luminal content through a loop ileostomy is sometimes used to treat refractory Crohn’s colitis (51–54). The symbiotic existence with the commensal gut flora is fundamental to health by maintaining immune homeostasis. This balance is reflected by gut immune tolerance against nutritional antigens while the immune defense can eradicate luminal pathogens threatening the integrity of the body.
No single specific microbial species has yet been identified to be a major instigator of the inflammation in IBD. Rather, an imbalance in the composition of commensal flora with a lack of diversity, dysbiosis, seems to be associated with the chronic intestinal inflammation (55). This lack of homeostasis is suggested to induce immune reactions against microflora that the gut would normally tolerate. For example, the identification of antibodies reactive to flagellin on commensal flora in CD patients has been suggested to reflect this status of abrogated tolerance (56, 57). Moreover, commensals producing anti-inflammatory butyrate (Faecalibacterium prausnitzii and Roseburia spp.) have been less abundant, whereas producers of mucin degraders (adherent-invasive Escherichia coli and Ruminococcus gnavus) have been more numerous compared with healthy controls in some IBD studies (58–63).
Tumor necrosis factor α (TNF-α), together with IL-23, have been identified as major drivers of IBD. Monoclonal antibodies against TNF-α (infliximab, adalimumab) or IL-23 (ustekinumab) have proven efficacious in the treatment of CD and UC (5). Other cytokines have also been identified in the chronic inflammation such as IL-1β, IFN-γ and IL-6, but the inhibition of these by antibodies (canakinumab, fontalizumab and tocilizumab) has not been successful in treating IBD. Interestingly, inhibition of IL-17A in CD even worsened the disease in some patients in a phase II clinical trial with secukinumab (64). The long-term efficacy of biologics, especially anti-TNF-α antibodies, is dampened by both primary non-response of approximately 30% and gradual loss of response of another 20–30% within 1 year, resulting in 12-month remission rates of around 30% in patients with severe IBD and the need of an alternative, more efficacious anti-inflammatory therapy is evident (65–67).
The brain–gut axis and the inflammatory reflex
An important conduit of communication between the brain and the gut is the autonomic nervous system, which includes the sympathetic nerves in the splanchnic branch and the parasympathetic nervous system (the vagal and pelvic nerves). Afferents from the parasympathetic vagal nerves converge into the nucleus tractus solitarius (NTS) of the medulla in the CNS. The NTS is functionally connected with the dorsal motor nucleus of the vagal nerve, which relays signals to the efferent vagal branches (4, 68, 69).
The NTS also communicates with the rostral ventrolateral medulla, which transmits signals through the sympathetic efferent nerves, and this entire network—labeled the autonomic brainstem loop—forms a regulating circuit for gut motility, acid secretion and satiety (68, 70, 71). The loop also communicates with the prefrontal cortex through the amygdala, hippocampus, hypothalamus and locus coeruleus, which may enable synchronization of signals from the gut with cognitive and behavioral functions, which may account for a reciprocal impact of emotional states on gut function (68).
The vagus nerve establishes an information highway between brain and gut with 80% of the neurons in afferent and 20% in efferent directions (72). The exact anatomy of the vagal nerve in the bowel has not been exhaustively mapped. There are reports that vagal nerve endings are found throughout the colon, whereas others suggest a vagal distribution limited to the right and transverse colon (73). The vagal afferents originate from several layers of the gut wall and may be activated to fire action potentials by cytokines as well as peptides such as IL-1β, CCK, GLP-1 and serotonin (74–77). Additional information relayed to the CNS through the vagus nerve is luminal osmolarity, intestinal distention and carbohydrate content (4).
The immunoregulatory properties of the vagus nerve are mediated through two major pathways. The classical anti-inflammatory gut–brain loop is mediated by the HPA axis consisting of afferent vagal nerves projecting into the NTS that relay the signal to corticotropin-releasing hormone (CRH) neurons in the hypothalamic nuclei. CRH induces the pituitary to release adrenocorticotropic hormone, which results in the production of adrenal glucocorticoids that has systemic anti-inflammatory properties (78–80).
Another anti-inflammatory pathway situated in the vagus nerve is called the inflammatory reflex and was first described in rodents by Borovikova et al. in 2000 (28). The authors showed that in mice subjected to intra-peritoneal injection of bacterial endotoxin [lipopolysaccharide (LPS)], serum and liver TNF-α levels were significantly higher in vagotomized mice. Reciprocally, electrical stimulation of the cervical vagus nerve reduced TNF-α and attenuated LPS-induced septic shock development in a rat model. This anti-inflammatory mechanism was mediated by acetylcholine (ACh) released in response to vagus nerve stimulation and subsequent studies have demonstrated reduced release of TNF-α, IL-1β and IL-6 from LPS-activated macrophages that express nicotinic ACh receptors (28, 78, 81).
Vagal nerves end in the myenteric plexus of the intestinal wall but the mechanism of interaction between immune cells residing in the gut wall and the vagus nerve is not fully understood (Fig. 1) (5, 82, 83). In addition, the immunoregulatory effects of vagus nerve activation may potentially also be mediated through a route parallel to the direct brain–vagus–gut connection (34). Vagus nerve branches reach the celiac ganglion, and signals are functionally propagated to the noradrenergic splenic nerve (Fig. 1). The splenic nerve, as well as choline acetyltransferase-expressing T cells (T-ChAT), and the alpha7 nicotinic ACh receptor subunit on immune cells are required to relay neural signals from the vagus nerve to cytokine-releasing immune cells with subsequent attenuation of inflammation (34, 81, 84–87). Whether the vagus nerve and the inflammatory reflex regulate ACh-producing T cells in the gut wall remains to be elucidated.
The inflammatory reflex. The vagus nerve functionally projects into the celiac ganglion, which relays signals to the sympathetic splenic nerve. ACh is released from choline acetyltransferase+ T cells (T-ChAT), which attenuates release of pro-inflammatory mediators from splenic macrophages. Vagus nerve endings also project into the gut wall and interface with the enteric nervous system through the myenteric plexus, and the function of T-ChAT cells in the gut is yet uncertain. NE, norepinephrine/noradrenaline. Some figure elements are sourced from Servier Medical Art under a Creative Commons Attribution 3.0 Unported License.
Although inflammation is an important response to tissue injury, failure to properly resolve inflammation may promote inflammatory disease (88, 89). Studies imply that homeostatic neural reflexes regulate not only the onset of inflammation, but also its active resolution phase (90, 91). For example, experimental disruption of vagus nerve signaling increased the intensity and duration of peritonitis and reduced the levels of specialized pro-resolving mediators, a class of potent bioactive lipid mediators that actively regulate inflammation resolution (88). Activation of the inflammatory reflex also reduces extracellular levels of another key mediator of inflammation, the high mobility group box 1 (HMGB1) (92). HMGB1 has been implicated in regulation of inflammation resolution, development of chronic inflammatory diseases and slow recovery after critical care and is associated with aspects of IBD (93–97). Moreover, a recent paper suggested that a novel neural liver–brain–gut reflex is important for the induction and maintenance of peripheral T regulatory cells in the intestine (98).
Taken together, available data indicate that homeostatic neural reflexes such as the inflammatory reflex regulate important processes through onset to resolution of inflammation.
Human studies
In parallel with several animal studies of different models of inflammation, observational evidence supports a role for the brain–gut axis in the pathophysiology of human IBD. Rubio et al. reported that decreased parasympathetic tone as measured by heart rate variability (HRV) was observed in patients with CD as compared with healthy individuals (99). Furthermore, in CD patients, an inverse relationship was seen between vagal tone and circulating TNF-α. Pellissier et al. showed that patients with high resting parasympathetic tone [HRV within the high frequency (HF) spectrum, which reflects the parasympathetic activity] had lower plasma TNF-α levels (100). In another study, in UC patients in remission after a flare of the disease, the sympathovagal balance expressed by the ratio between low frequency (LF) and HF spectrum of HRV differed from healthy controls. The authors reported that the sympathovagal balance was associated with disease activity in terms of higher levels of C-reactive protein and serum TNF-α levels, measured at onset of disease. Likewise, the LF/HF ratio was associated with mucosal expression of IL-8 and IFN-γ RNA. Further, the study concluded that UC patients with higher parasympathetic activity after a disease flare demonstrated reduced systemic inflammation during the following 3 years (101).
In a recent epidemiological study including 15 637 vagotomized patients from 1964 to 2010, the incidence of IBD during an average of 21.6-year follow-up was 0.38 per 1000 person-years compared with 0.25 for non-vagotomized individuals (hazard ratio: 1.50; 95% confidence interval: 1.25–1.80). The significant association was limited to CD and not found in UC, and one reason for this difference might be that UC mainly affects the distal colon and rectum, which are not innervated by the vagal nerve. Nevertheless, the observation highlights the potential role of the gut–brain axis especially in CD (102).
To date, two small clinical trials intervening with the brain–gut axis have been carried out in CD patients. Both studies utilized implantable vagus nerve stimulators (VNS) (Cyberonics®), which is an established medical device used in the treatment of drug-refractory epilepsy (7, 103, 104). Sinniger et al. reported on a 12-month follow-up study including nine moderately active, biologic therapy naive CD subjects treated daily with the implanted device delivering 30 s of electrical stimulation to the left vagal nerve every 5 min, the same parameters used to treat epilepsy. The intensity of the pulse was gradually increased according to individual tolerability. Seven patients completed 1 year of VNS therapy and they demonstrated a restored vagal tone and reported reduced clinical scores with regards to pain. Their cytokine profile shifted toward a profile of healthy individuals with strongest trends in the levels of IL-6, IL-12/IL-23, transforming growth factor β1 (TGF-β1) and TNF-α. The treatment was well tolerated by all and no major safety issues were raised (104). The authors pointed out that this pilot study is encouraging, but larger randomized controlled studies are needed.
In the other clinical trial, 16 subjects with moderate-to-severe disease refractory to monoclonal antibody treatment were included (103). These treatment-refractory subjects had failed or were intolerant to up to six prior biologics. The first eight subjects had biologics washed out prior to device implantation and were subsequently treated with VNS as monotherapy while the second eight had the option to remain on biologics during the trial. The Cyberonics® VNS stimulator was implanted in this study as well but reprogrammed to deliver study-specific stimulations. Contrary to the continuously repeating stimulation used in the Sinniger trial (104), these patients were only stimulated for one to four times daily limited in sessions lasting 1–5 min. The primary study ended after 16 weeks and the authors report improved clinical scores, endoscopic scores and calprotectin levels in a preliminary report in abstract form (103). There were no treatment-related serious adverse events reported.
Bioelectronic devices
A key concept in bioelectronic medicine is to harness the function of well-characterized homeostatic reflexes and, by electrical stimulation of specific neural circuits, replace or supplement drugs with electrons. The long-term vision includes development of closed-loop systems capable of autonomous treatment-adjustment in response to disease-relevant physiological measurements. The devices available for human clinical trials today are, however, primarily manually programmable electrical nerve stimulators.
Electrical vagus nerve stimulation is an established therapeutic option for treatment of drug-refractory epilepsy and, according to manufacturers’ data, stimulators have been implanted in over 100 000 patients. The implanted components of this device consist of a hermetically sealed implantable pulse generator (IPG) with a primary cell battery that is placed subcutaneously in the chest, and a 430-mm lead wire that terminates in two helical electrodes and one anchor helix that are coiled around the vagus nerve (105). Since these neurological indications are ‘high duty cycle’ (they require the stimulator to deliver current often during the day), the power requirement of the implanted battery is high, so it needs to be relatively large and heavy. This ‘can and lead’ approach to device design may be acceptable for epilepsy, but the implantation procedure requires two separate incisions, the tunneling of a long, breakable lead and the application of electrodes to the nerve that are not readily amenable to revision or removal (106).
Activation of the inflammatory reflex by electrical vagus nerve stimulation does not require the high duty cycle stimulation used in epilepsy treatment but can be achieved by electrical stimulation lasting seconds to minutes per day (92, 107). Hence, it has become possible to design new devices for electrical stimulation of the vagus nerve intended for treatment of excessive inflammation that better facilitate a simple implantation procedure and have a lower weight and volume.
One example of this type of device intended for stimulation of the vagus nerve in treatment of CD and rheumatoid arthritis is currently in clinical trials (108, 109). It consists of one functional unit that comprises the rechargeable battery housed within a small IPG with integrated electrodes for the nerve interface (Fig. 2). It is sufficiently compact and light to be implanted directly on the vagus nerve through a single incision without the need for tunneling to a distant pocket. As there are no helical electrodes that typically become attached to the nerve by development of local fibrosis, the device is likely to be more amenable to simple surgical extraction and/or replacement if required. Without doubt, improved device technology that simplifies surgery and minimizes the risk of short- and long-term complications is of key importance for facilitating the larger, well-designed clinical trials that are greatly needed for the next clinical steps in bioelectronic medicine.
Vagus nerve stimulation. An example of a small bioelectronic device intended for activation of the inflammatory reflex to treat CD and rheumatoid arthritis. Clinical studies using bioelectronic devices have thus far utilized cervical stimulation locations (solid line) but other positions (the subdiaphragmatic vagus nerve and splenic nerve; dotted lines) have been proposed as distal targets for future study. Some figure elements are sourced from Servier Medical Art under a Creative Commons Attribution 3.0 Unported License.
With increasing mechanistic understanding and improved anatomical mapping of the neural circuits that regulate immune responses, it will become valuable to target specific branches of the peripheral nervous system to achieve more localized and targeted treatment effects. It is possible that a more distal device placement may produce a more localized effect on cytokine release and inflammation, and some experimental data support this notion. Implementation of a strategy for distal, near-target organ stimulator placement will be facilitated by high device surface biocompatibility and miniaturization. Of note, the requirement for a surgical procedure before start of treatment may, in some cases, such as for acute conditions, reduce the usability of fully implanted VNS for treatment of excessive inflammation.
Efforts are ongoing to develop other modalities than direct nerve stimulation by electrical current from a surgically implanted device to activate specific nerves, for example focused ultrasound, transcutaneous electrical stimulators and other innovative methods (110–112). There is a continuum between more invasive and less invasive techniques, which likely will include a tradeoff between better targeting, greater specificity and more energy efficiency. Thus, there are several potential advantages and drawbacks with the available methods and new techniques under development, and we are only at the beginning of what will likely be a lengthy process to determine which methods may be sufficiently efficacious and reliable over the long term for treatment of specific diseases in which excessive inflammation plays an important role.
Conclusions
The evolving mechanistic understanding of neuro-immune interactions in the peripheral nervous system and the immune system is expanding the field of inflammation neuroscience. Discoveries on the interplay between intestinal innervation, immune cells and microbiota have elevated the interest in the brain–gut axis and opportunities for new treatment options in IBD are expanding with the increasing body of high-quality data about inflammation neuroscience. Continual progress in the functional mapping of the neural reflex regulation of immunity together with advances in nerve–machine interface technology are important for delineating new potential therapeutic targets for attenuation and resolution of excessive inflammation. In the future, the combination of miniaturized nerve simulator technology with sensors for inflammation biomarkers promises to provide a significant improvement of personalized medicine. Better monitoring and frequent treatment adjustments in response to data provide hope.
Conflicts of interest statement: M.E. has received honoraria for lectures and consultancy from AbbVie, Merck (MSD), Takeda, Ferring, Orion Pharma, Otsuka, Tillotts, Novartis, Pfizer, and Janssen, received research funding from AbbVie and MSD, and is a former shareholder of Emune AB. Y.A.L. is employed by SetPoint Medical. L.T. has no disclosures. P.S.O. has received lecture honoraria from Ferring and Janssen and he is shareholder of Emune AB. M.E. was supported by Stockholm County Council ALF [project number 20180565].
