Anatomical And Clinical Implications Of Vagal Modulation Of The Spleen.

Abstract

The vagus nerve coordinates most physiologic functions including the cardiovascular and immune systems. This mechanism has significant clinical implications because electrical stimulation of the vagus nerve can control inflammation and organ injury in infectious and inflammatory disorders. The complex mechanisms that mediate vagal modulation of systemic inflammation are mainly regulated via the spleen. More specifically, vagal stimulation prevents organ injury and systemic inflammation by inhibiting the production of inflammatory cytokines in the spleen. However, the neuronal regulation of the spleen is controversial suggesting that it can be mediated by either monosynaptic innervation of the splenic parenchyma or secondary neurons from the celiac ganglion depending on the experimental conditions. Recent physiologic and anatomic studies suggest that inflammation in infectious and inflammatory disorders is regulated by neuro-immune multi-synaptic interactions between the vagus and the splanchnic nerves to modulate the spleen. Here, we review the current knowledge on these interactions, and discuss their experimental and clinical implications in infectious and inflammatory disorders.

1. INTRODUCTION

During the 20^th century, the application of immunology has become integral to modern medicine. In the past, the immune system was studied mainly at the cellular level in isolated cultures responding to specific infectious and inflammatory stimuli. Recent studies have reported that psychologic stress affects immunity and susceptibility to infections, suggesting a connection between the nervous and immune systems. Many connections are humoral signals based on the production of diffusible factors such as catecholamines released from the adrenal gland into the bloodstream. However, recent studies have revealed that innervation of specific immune organs such as the spleen plays a critical role in regulating inflammation and organ function disrupted during trauma, infectious, or inflammatory disorders. Understanding how the nervous system regulates the immune system is critical to designing novel therapeutic strategies to control inflammation and organ injury in multiple disorders.

Innervation of immune organs has significant clinical implications in infectious and inflammatory disorders. For example, electrical stimulation of the vagus nerve reduces systemic inflammation in experimental sepsis by inhibiting the production of inflammatory cytokines in the spleen (Huston et al., 2006). Indeed, vagal stimulation attenuates inflammation in multiple disorders including experimental ischemia and reperfusion (Bernik et al., 2002), hemorrhage and resuscitation (Cai et al., 2009), pancreatitis (Van Westerloo et al., 2006), postoperative ileus (Stakenborg et al., 2017), colitis (Meregnani et al., 2011), arthritis (Bassi et al., 2017; Koopman et al., 2016), asthma (Miner et al., 2012; Yin et al., 2018), and sepsis (Huston et al., 2006; Vida et al., 2011b). We reported that the spleen is the major source of inflammatory cytokines in the blood, and that acetylcholine, the principal neurotransmitter released by the vagus nerve, inhibits the production of inflammatory cytokines in splenic macrophages, the main cellular source of inflammatory cytokines (Huston et al., 2006). These findings suggest a vagal anti-inflammatory network that can control systemic inflammation and prevent organ injury by regulating immune cells in the spleen. However, the neuronal regulation of the spleen is debated, as the results appear to depend on experimental conditions. Here, we present a critical analysis of neuroanatomic and neurophysiologic studies on neuro-immune modulation of the spleen, from the early macroanatomic to recent advanced neurotracing, and discuss their physiologic and clinical implications.

2. NEUROANATOMIC STUDIES

One of the earliest studies on the innervation of the spleen comes from Swan in 1834, who described a posterior vagal trunk ending in the celiac plexus in human dissections. Perman (1916) reviewed these studies and reported no vagal gastric plexus and vagal innervation into the spleen. With the advance of Golgi’s Method using silver staining technique to visualize neurons, a vagal-celiac ganglion connection was observed in human embryos (Hammar, 1935). However, this technique was not fully detailed, and the findings were not confirmed (Mosimann, 1954). Colouma (1937) reported a vagal-celiac ganglion connection in his anatomic studies of 131 non-human mammals, from marsupials and cetaceans to artiodactylians and primates, and 48 human dissections. On the other hand, Kuntz (1938) reported that ablation of the vagus nerve in cats (via a Wallerian degeneration mechanism) did not alter the number of terminal branches or synaptic connections in the celiac ganglion. These findings were later explained in part by the limitation of the technique and the limited number of vagal fibers that innervate the plexus as previously reported by Mosimann (1954). In addition, the vagus nerve is not isomorphic among different species, and vagal innervations differ even in subjects of the same species, including humans (McCrea, 1924; Ruckley, 1964). For example, anatomic studies in rats and rabbits showed an uncommon accessory celiac branch of the vagus nerve that arises from the anterior vagal trunk and follows the left gastric artery in parallel with the celiac branch. These results suggest two separated vagal branches can innervate the celiac ganglion via different pathways (McCrea, 1924; Prechtl and Powley, 1985). Abdominal paraganglia is another unusual anatomic finding associated with the celiac branches (Deane et al., 1975; Prechtl and Powley, 1985). Whether paraganglia is part of the vagal system or simply local ganglia associated with gastrointestinal tissue must be determined by future studies. Finally, some axons from pre-vertebral ganglia can transverse other ganglia without establishing local synapses before reaching their targets (Dalsgaard and Elfvin, 1982; Parr et al., 1993; Trudrung, 1994). This transganglionic pathway has been observed in domestic birds where branches of the vagus nerve transverse the celiac ganglion to reach the abdominal organs (Hiramatsu and Watanabe, 1993). In general, most studies support the vagal-celiac ganglion connection , although other studies did not confirm this connection due to the limitations of the experimental techniques (Please see references and summary in Table 1). In contrast to the role of the parasympathetic vagus nerve, innervation of the spleen by the sympathetic splenic nerve is well documented by a number of neuroanatomic studies in different species including rodents (Guyot et al., 2019; Reilly et al., 1976), cats (Fillenz, 1970; Gillespie and Kirpekar, 1966), dogs (Zetterström, 1973), and humans (Kudoh, 1979; Verlinden et al., 2019).

Table 1.

Chronologic recording of macroanatomic studies on vagal innervation to the spleen and celiac ganglion.

Author	Subject	Technique	Vagus-celiac ganglion connection
Swan, 1834	Human	Dissection	Yes
Bourgery, 1844	Human	Dissection	Yes
Kollman, 1860	Human, Rabbit, Cat, Dog	Dissection	Yes
Luschka, 1863	Human	Dissection	Yes
Mivart, 1881	Cat	Dissection	Yes
Krause, 1884	Dog	Dissection	Yes
Jonnesco, 1895	Human	Dissection	Yes
Sobotta, 1907	Human	Dissection	Yes
Spalteholz, 1911	Human	Dissection	Yes
Perman, 1916	Human	Dissection	No
Latarjet & Wertheimer, 1921	Dog, Human	Dissection	Yes
McCrea, 1924	Human, Dog, Rabbit, Cat	Dissection	Yes
Glaser, 1929	Human, Pig, Calf, Rabbit, Guinea-pig	Rongalit – Methylene blue Silver impregnation	Yes
Hammar, 1935	Human fetus	Silver impregnation	Yes
Colouma, 1937	Mammals	Dissection	Yes
Kuntz, 1938	Human, Cat	Subdiaphragmatic vagus nerve cut - Wallerian degeneration	No
Mitchell, 1938	Human	Dissection	Yes
Clara, 1942	Human		Yes
Utterback, 1944	Cat	Subdiaphragmatic vagus nerve cut - Wallerian degeneration	No
Jackson, 1949	Human	Dissection	No
Niederhausern, 1953	Human	Dissection	No
Mosimann, 1954	Rat	Silver impregnation	No
Mizeres, 1955	Dog	Dissection	Yes
Greene, 1963	Rat	Dissection	Yes
Baljet & Drukker, 1979	Neonatal and Adult Rats	Acetylcholinesterase method	Yes
Hammar & Santer, 1981	Rat	Dissection	No
Boekelaar, 1985	Human and Rat fetuses	Acetylcholinesterase method	yes

3. NEURO-TRACT TRACER STUDIES

Recently developed neuronal labeling and tracing methods are revolutionizing neurobiology. Neurotracers are dye- or probe-based compounds that are transported along a specific neuronal axis. These compounds can trace the axon because of their ability to move away from the cell body (anterograde property), toward the cell body (retrograde property), or along a bidirectional course (Köbbert et al., 2000).

3.1. TRACER INJECTION INTO THE SPLEEN

Horseradish peroxidase is one of the first neurotracers used in modern tracing techniques (Kristensson et al., 1971). It is engulfed by the neuron and fused into lysosomes, where the peroxidase is retrogradely transported towards the cell body and proximal dendrites (Harper et al., 1980). David Kuo and George Krauthamer (1981) performed one of the first neurotracing studies to investigate the organization of cell bodies in the paravertebral ganglia. They exposed the proximal ends of the splenic nerve with peroxidase, and traced the signal to the paravertebral and dorsal root ganglia, as well as to the superior mesenteric and celiac ganglia of the solar plexus. Detailed analyses of the plexus revealed an unusual neuronal clustering that is seen primarily when fibers are close to the target organ, but not when they leave the ganglia. These findings may have resulted from injecting the tracer in close proximity to the plexus, thereby labeling the splenic and nearby unrelated nerves such as those from the renal plexus. Meckler and Weaver (1984) repeated these experiments, but injected peroxidase a few millimeters away from the spleen instead of a few millimeters from the solar plexus. As a result, the left and right celiac poles contained > 90% of all labeled cells (the other 10% were found in the lumbar chain ganglia), evidencing direct innervation of the spleen from the celiac ganglion (Meckler and Weaver, 1984). Baron and Jänig (1988) also reported that about 98% of splenic nerve fibers originate in the solar plexus, while the other 2% were identified in the paravertebral sympathetic trunk. Other studies also supported mostly sympathetic innervation into the spleen (Bellinger et al., 1989; Chevendra and Weaver, 1991; Felten et al., 1987; Quinson et al., 2001).

The discovery of neuropeptides, such as neuropeptide Y and the vasoactive intestinal peptide, inside the spleen parenchyma suggested additional sources of afferent innervation other than the sympathetic system (Lundberg et al., 1985). Nance and Burns (1989) used four different tracers to determine the sources of splenic afferent innervation. Injections of regular horseradish peroxidase, peroxidase conjugated to agglunitins, Fluoro-Gold, or Fast-Blue in different parts of the body of the spleen stained neurons in the celiac-mesenteric plexus and the sympathetic ganglia (efferent supply), but no staining was found in the cervical dorsal root or the nodose ganglia (supposedly the afferent supply). These findings suggested exclusive sympathetic, but no parasympathetic, vagal innervation in the spleen (Nance and Burns, 1989). However, Chen and colleagues (1996) found additional splenic afferent fibers, after they injected horseradish peroxidase conjugated to wheat-germ agglutinin into the splenic hilus of cats. They reported tracer residues in the celiac ganglion and in the dorsal motor and solitary tract nuclei, two additional parasympathetic brain structures that convey afferent information from abdominal viscera via the vagus nerve. Although Nance and Burns (1989) also observed labeling of additional afferent components (in their case, the dorsal root ganglia) after injecting FluoroGold in the splenic hilus, these findings could be due to tracer extravasation into the vasculature.

Unlike dyes, virus-based neuronal tracers only infect neurons and thus, are ineffective when leaked into the bloodstream (Ugolini, 2011). Cano and colleagues (2001) were among the first group using viruses to study splenic innervation. They injected Bartha-pseudorabies virus at six different sites along the hilar axis of the spleen, and reported different degrees of neural infection at different time points. Infection was first observed within the thoracic intermediolateral column (via celiac-mesenteric ganglia) at 60 hours, in the brainstem at 72 hours, and in the forebrain at 77 hours post inoculation. The nodose ganglion showed no trace of the virus until 110 hours post inoculation, after extensive infection of the spinal cord and many brain areas.

3.2. TRACER INJECTION INTO THE VAGUS NERVE

Potential vagal innervation of the prevertebral chain led Berthoud and Powley (1993) to selectively label vagal preganglionic innervations by injecting DiL (a newly available lipophilic tracer) into the dorsal motor nucleus of the vagus (brain nucleus) in combination with a new FluoroGold technique that counterstains neurons of the autonomic nervous system (Berthoud et al., 1990). DiL-labeled fibers were found in both the celiac and superior mesenteric arteries. Some of these fibers appeared to enter the celiac ganglion, while many others bypassed it and entered the mesenteric artery plexus. Since neural fibers from and to the dorsal motor nucleus are located in the nodose (inferior) ganglion, DiL was also injected into the vagal sheath below this ganglion as an additional experimental control. This procedure stained nerve terminals throughout the gastrointestinal tract, although only a small quantity of neurons was labeled in the brain nucleus, and no signal was found in the solar plexus. As none of the DiL-labeled axons showed simultaneous FluoroGold fluorescence, which was injected intraperitoneally, Berthoud and Powley (1993) suggested that DiL incorporation and transport in the nodose ganglion occurred only in neurons damaged by the injection, and concluded that the celiac ganglion receives vagal terminals. This notion was reinforced when Berthoud and Powley injected DiL or DiA into the dorsal motor nucleus, and confirmed labeled terminals in the celiac ganglion (Berthoud and Powley, 1996). These findings are quite unusual because DiA and DiL are lipophilic neurotracers that diffuse laterally through the cell membrane rather than depending on active axonal transport (Honig, 1989), and because the neuronal staining intensity observed in the celiac ganglion should mirror that of the gastrointestinal tract. Thus, Bratton and colleagues (2012) performed similar experiments by counterstaining DiL with Fast-Blue. Both dyes were found inside the celiac ganglion after injecting DiL into the dorsal motor nucleus and Fast-Blue into the splenic parenchyma. However, confocal microscopic analyses failed to show direct synapses between DiL- and Fast-Blue-labeled terminals (Bratton et al., 2012), and in vivo electrophysiologic experiments also failed to confirm this direct interaction (Please see Splenic Nerve Activity Studies below). These results contradicted earlier confocal microscopic analyses by Berthoud and Powley (1996), which showed highly varicose DiL- or DiA-positive nerve endings originating in the dorsal motor nucleus and tightly surrounding individual celiac ganglion cells. Other neural tracers produced staining patterns similar to those described by Berthoud and Powley in 1993 and 1996. For example, Ting et al. (2017) injected Dextran-Texas red, a lipophilic tracer, into the dorsal motor nucleus of mice, and reported red fluorescent nerve terminals in the celiac ganglion in apposition to sympathetic cell bodies. However, labeled vagal terminals in the celiac ganglion expressed only a faint red fluorescence, which is in contrast to the high red intensity shown by the myenteric plexus. These findings suggest that Dextran-Texas red is incorporated only into damaged neurons, as previously suggested by Berthoud and Powley (1993). Finally, Domoto et al (1995) performed neuronal tracing from the solar plexus to vagal-related structures. After injecting FluoroGold into the superior mesenteric ganglion of rats, they observed labeled cells in the thoracic dorsal root ganglia, intestines, and nodose ganglion (Please see summary in Table 2). Taken together, all these studies strongly suggest that few neural fibers of vagal origin innervate the celiac ganglion; however, functional interactions between these structures must be confirmed.

Table 2.

Chronologic recording of neuroanatomic studies using neurotracers.

Author	Subject	Technique	Local of injection	Local of visualization	Vagus-celiac ganglion connection
Nance & Burns, 1989	Rat	FGo Fast-blue	Spleen: body	Mesenteric ganglia Nodose ganglion	No
Berthoud & Powley, 1993	Rat	Dil FGo	Nodose ganglion DMN	Pre-vertebral ganglia	Yes
Domoto et al., 1995	Rat	FGo	Celiac ganglion	thoracic dorsal root ganglia nodose ganglion	Yes
Chen et al., 1996	Cat	WGA-HRP	Spleen: splenic artery	DMN	Yes, but see commentaries
Cano et al., 2001	Rat	PRV	Spleen: hilus	Nodose ganglion	No
Buijs et al., 2008	Rat	PRV	Spleen: hilus	IML DMN	Yes
Cailotto et al., 2012	Mouse	CTB	Spleen: poles SI	NTS DMN	Yes
Bratton et al., 2012	Rat	Fast-blue Dil	Spleen: body DMN	Celiac and suprarenal ganglia	No
Ting et al., 2017	Mouse	Dextran-Texas Red	DMN	Celiac ganglion	Yes, but see commentaries

FGo: FluoroGold; WGA-HRP: wheat germ agglutinin-horseradish peroxidase; PRV: pseudorabies virus; CTB: cholera toxin B; DMN: dorsal motor nucleus of the vagus nerve; DDMO: dorsal division of medulla oblongata; NTS: nucleus tractus solitarius; SI: small intestine; IML: intermedio-lateral column of the spinal cord.

4. CHOLINERGIC MARKERS IN THE SPLEEN

Cholinergic markers are found in pre-ganglionic terminals, ganglionic sympathetic pseudomotor, and all ganglionic parasympathetic fibers. These markers include the vesicular acetylcholine transporter (VAChT), choline acetyltransferase (ChAT), which catalyzes the transfer of an acetyl group from acetyl-CoA to choline to form acetylcholine, and acetylcholinesterase (AChE), the enzyme responsible for the hydrolytic metabolism of acetylcholine yielding choline and acetate.

Several investigators reported that spleen extracts from mammals express significant levels of both ChAT and AChE (Bulloch et al., 1994; Consolo et al., 1972; Dale and Dudley, 1929; Stephens-Newsham et al., 1979); however, these studies did not specify the cellular source of these enzymes. Gomori (1948) performed the first immunohistochemical studies that allowed direct visualization of cholinesterase inside the cells through enzymatic reactions using fatty acid esters of choline as substrates, but this method was very ineffective. Koelle and Friedenwald (1949) improved the method by introducing acetylthiocholine (AChI) as a substrate. Tissue cholinesterases hydrolyze AChI, and the free thiocholine is then captured, precipitated, and visualized. D’Agostini and Rossatti (1959) used this new method to visualize AChE in the spleen of cats, rabbits, and rats, and reported cholinesterase activity in the capillaries that supply the germinal centers. Likewise, Ballantyne (1968) showed AChI hydrolyzation in cortical capillaries and marginal sinuses supplying the splenic follicles. Fillenz (1970) then used several methods of neuronal labeling, i.e., silver impregnation, AChI, and immunofluorescence, in combination with electron microscopy to describe cholinergic nerve terminals in the spleen of cats. In this study, AChE-positive neurons were observed only in the splenic hilum. Acetylcholinesterase activity was lost when the nerve terminals enter the spleen parenchyma, and only non-neural splenic structures expressed AChE. Likewise, analyses of spleen samples from other mammals, such as ungulates, guinea pigs, and mice, also showed AChE in very small quantities only in nerve fibers restrained to blood vessels and in non-neural structures, such as megakaryocytes, erythrocytes, platelets, and occasionally in nonspecific deposits in the white pulp (Reilly et al., 1976; Schmidtova et al., 2004; Stephens-Newsham et al., 1979).

Surgically selective denervations revealed the source of splenic AChE-positive neurons in rats (Bellinger et al., 1993). In this study, splenic AChE overlapped with norepinephrine in sympathetic terminals found in splenic blood vessels and reticular tissue, as well as in non-neural structures and in the cellular bodies of the celiac ganglion. Moreover, chemical (5-hydroxydopamine administration) or surgical (celiac ganglion removal) sympathectomy eliminated both neural norepinephrine and AChE activity in the spleen. On the other hand, bilateral subdiaphragmatic vagotomy did not affect splenic AChE or norepinephrine staining. These results suggested that the spleen lacks cholinergic innervation and that the neural AChE found in the spleen mostly derived from sympathetic nerves.

Studies in human spleens reported similar findings. Rakhawy et al (1976) analyzed 22 surgically excised or post-mortem healthy spleens, and found thiocholine deposits in blood vessels and very small quantities in white and red pulp. Kudoh and colleagues (1979) evaluated 52 human spleens from patients with advanced gastrointestinal cancer, and found AChE activity only in nerves close to arteries of the splenic trabecular system. Recent studies in 2017 on 8 fresh spleen samples from non-surviving septic patients, reported abundant nerve fibers containing tyrosine hydroxylase (a neuronal sympathetic marker) in arteries and arterioles, while no terminals contained vesicular acetylcholine transporter (Hoover et al., 2017).

Choline acetyltransferase (ChAT), the enzyme responsible for the synthesis of acetylcholine, is extensively used to label cholinergic neurons (Oda, 1999). In early experiments, ChAT activity was measured by analyzing the formation of radioactive acetylcholine after combining natural tissue choline with synthetic acetyl-[¹⁴C]coenzyme A. Consolo et al (1972) used this method to calculate global splenic acetylcholine levels and reported very little, if any, ChAT activity in the spleen of cats. Stephens-Newsham et al (1979) found that spleen samples from various mammals expressed different levels of ChAT depending on the species. For example, spleens from horses and donkeys exhibited high levels of acetylcholine, while those from ox, pigs, sheep, and wallabies showed very low or negligible levels. Studies in rats yielded similar results, and no evidence for cholinergic innervation was found in murine spleens (Bellinger et al., 1993). These studies appear to suggest a lack of ChAT in the spleen; however, the indirect method used, i.e., formation of radioactive acetylcholine, is affected by ChAT inhibitors and other contaminants during sample processing. Thus, Gautron et al (2013) used ChAT-Cre-tdTomato mice, a genetically modified mouse line that emits a reddish fluorescence in all cells expressing ChAT. In these animals, only lymphocyte-containing areas of white pulp and arterioles of the spleen exhibited fluorescent fibers, albeit in small quantities. Although none of the red ChAT neurons contained tyrosine hydroxylase, they were often found in close apposition to these terminals, especially around arterioles. Although the authors did not identify the source of these neurons, they deduced their origin in ganglionic sympathetic neurons in the para- and/or pre-vertebral chains. In a very recent studies in 2019, using ChATBAC-eGFP mice, a transgenic mouse line emitting green fluorescence on cholinergic neurons, Kaestner and colleagues (2019) reported a dense network of cholinergic terminals surrounding sympathetic cell bodies (tyrosine hydroxylase positive) inside the celiac ganglion (Please see summary in Table 3). Overall, the splenic parenchyma has feeble levels of neuronal AChE and ChAT, and those cholinergic fibers detected inside the spleen are apparently derived from the sympathetic system, as ganglionic sympathetic neurons can also exhibit a cholinergic phenotype (Burn and Rand, 1960; Elfvin et al., 1993; Holmstedt and Sjoqvist, 1959; Schafer et al., 1997). In fact, Lever (1970) showed that about 0.5% of the 2055 unmyelinated axons studied in the splenic nerve of a cat exhibited unusual cholinergic phenotype, and this percentage increased to 2% after nerve constriction, e.g., accumulation of AChE in the proximal stump. Decades later, Belliger et al (1993) showed cholinergic cell bodies inside the celiac ganglion, but could not assert that the axons from these neurons innervate the spleen.

Table 3.

Chronologic recording of neuroanatomic studies with cholinergic markers in the spleen.

Author	Subject	Cholinergic marker	Method	Positive Location
D’Agostini & Rossatti, 1959	Cat, Rabbit, Rat	AChE	Thiocholine precipitation	Capillaries and germinal centers Lymphoid cells surroundings
Ballanthyne, 1968	Rabbit	AChE	Thiocholine precipitation	Cortical capillaries Marginal sinuses
Lever et al., 1969	Cat	AChE	Thiocholine precipitation	Splenic nerve fibers
Fillenz, 1970	Cat	AChE	Silver impregnation Thiocholine precipitation	Non-neural structures
Consolo et al., 1971	Cat	ChAT	Radiochemical	Non-neural structures
Reilly et al., 1976	Mouse	AChE	Thiocholine precipitation	Non-neural structures Occasional in the white pulp
Rakhawy et al., 1976	Human	AChE	Thiocholine precipitation	Blood vessels Feeble in white and red pulps
Kudoh et al., 1979	Human	AChE	Thiocholine precipitation	Trabeculae
Stephens-Newsham et al., 1979	Horse	AChE ChAT	AChE: Thiocholine precipitation ChAT: Radiochemical assay	AChE: blood vessels ChAT: Non-neural structures
Bellinger et al., 1993	Rat	AChE ChAT	AChE: Thiocholine precipitation ChAT: Radiochemical assay	AChE: white/red pulp, arterioles (see comments) ChAT: not detected
Schmidtova et al., 2004	Guinea-pig	AChE	Thiocholine precipitation	Blood vessels
Gautron et al., 2013	Mouse	ChAT	ChAT-Cre-tdTomato animals	Few ChAT+ fibres in white and red pulp
Hoover et al., 2017	Human	VAChT	Immunohistochemistry	No detection
Guyot et al., 2019	Mouse	ChAT	Immunohistochemistry	Apical splenic nerve branch
Verlinden et al., 2019	Human	ChAT	Immunohistochemistry	Non-neural structures
Kaestner et al., 2019	Mouse	ChAT	ChATBAC-eGFP mice	Celiac ganglion

ChAT: choline acetyltransferase; AChE: acetylcholinesterase; VAChT: vesicular acetylcholine transporter.

Further, cholinergic innervation of the celiac ganglion may not necessarily be vagal. Viscerofugal neurons from the gastrointestinal tract traveling cranially to synapse the celiac ganglion have been reported in the literature (Domoto et al., 1995; Furness et al., 2000; Messenger and Furness, 1992), as well as their cholinergic phenotype (Furness et al., 2000; Lomax et al., 2000; Mann et al., 1995). These studies support the notion that the celiac ganglion is part of the abdominal feedback that modulates reflex inputs from different parts of the digestive tract (Frantzides et al., 1987; King and Szurszewski, 1989; Semba, 1954). As the gastrointestinal tract is densely innervated by vagal afferents, vagal stimulation may activate these gut-celiac networks (Figure 1). In addition, vagal fibers have a double phenotype. Early studies reported that vagal stimulation can induce chronotropic and inotropic effects in the heart of sympathectomized animals, depending on the electrical intensity applied to the nerve (Benfey and Greeff, 1961; Chiang and Leaders, 1966; McEwen, 1956). These findings indicate that the vagus nerve is not completely parasympathetic. In fact, Muryobayashi et al (1968) and Nielsen et al (1969) showed monoamine-fluorescent axons in both the subdiaphragmatic and cervical vagus coming from the superior cervical ganglion (sympathetic component). Other anatomic studies confirmed these findings in cats (Ahlman et al., 1976), rats (Ohsumi, 1974), and guinea pigs (Tansy et al., 1971), while other investigators added the stellate, medial cervical, inferior, and/or intrathoracic ganglia as additional sources of sympathetic fibers (Ahlman et al., 1978; Ahlman et al., 1979; Kawagishi et al., 2008; Liedberg, 1973; Verlinden et al., 2016). Future studies will be required to determine whether these sympathetic vagal fibers participate in the anti-inflammatory potential of the vagus nerve.

Figure 1. The cholinergic anti-inflammatory pathway and alternative neuro-immune connections.

A: The cholinergic anti-inflammatory network is based on connections between the vagus nerve and the spleen via the celiac ganglion (CG) and the splenic nerve. B: Neuro-immune pathways involving the efferent vagus nerve. (1) The vagus nerve directly innervates the spleen, bypassing the celiac ganglion; (2) Vagal terminals release acetylcholine (ACh) into the core of the adrenal glands to induce dopamine (DA) production; (3) Vagal stimulation also activates intestinal-celiac reflexes. C: Neuro-immune networks involving vagal afferent pathways. (1) Vagal stimulation activates the splenic nerve through a central reflex mechanism dependent on the spinal cord, the splanchnic nerve, and the celiac ganglion; (2) Vagal stimulation also induces the systemic release of norepinephrine (NE) from sympathetic chain neurons. Figure templates extracted from the Servier Medical Art (https://smart.servier.com)

5. SPLENIC NERVE ACTIVITY STUDIES

Electrophysiologic nerve assessment offers more dynamic and immediate data analyses than neuro-tracing, nerve ablation, or enzymatic methods. Early electrophysiologic studies showed that the activity of the splenic nerve is affected by blood pressure as it exhibits a pulse-synchronized pattern. This suggests that several mechanisms are involved, and that baroreceptors and cardiac afferent fibers can modulate the splenic nerve (Green and Heffron, 1966; Ninomiya and Irisawa, 1975; Weaver et al., 1983). More recent studies support this hypothesis, showing that stimulation of aortic baroreceptors increases splenic nerve activity (Meckler and Weaver, 1985; Weaver et al., 1983), and that the splenic nerve is under reflex modulation by vagal afferents (Meckler and Weaver, 1988a; Meckler and Weaver, 1988b; Taylor and Schramm, 1987).

Interest in the splenic nerve reappeared when we reported that electrical stimulation of the vagus nerve prevents septic shock, organ damage, and systemic inflammation in experimental sepsis by inhibiting production of inflammatory cytokines in the spleen (Huston et al., 2006). Speculation arose regarding whether the vagus nerve modulates the splenic nerve via a peripheral (efferent) or a central (afferent) pathway (Figure 2). By sectioning the cervical vagus nerve and specifically stimulating its distal (peripheral) part, we showed that specific efferent vagal stimulation attenuated serum TNF levels in endotoxemic wild-type mice, but not in mice lacking alpha7 nicotinic acetylcholine receptors (α7nAChRs) (Vida et al., 2011a). These findings showed that efferent vagal stimulation reduced serum TNF levels via α7nAChRs, but also that electrical stimulation of the intact vagus nerve reduces serum TNF levels in septic α7nAChRs-KO mice. Acetylcholine inhibits the production of inflammatory factors from splenic macrophages by inhibiting STAT3 phosphorylation and the NF-kB pathway via α7nAChRs (Peña et al., 2010; Ulloa, 2013). Taken together, these observations provided evidence that electrical stimulation of the intact vagus nerve can activate both afferent (central) networks via the central nervous system and efferent (peripheral) pathways, which modulate serum TNF levels in endotoxemic mice via α7nAChRs in splenic macrophages.

Figure 2. Neuronal and cellular modulation of splenic macrophages.

Electrical stimulation of the vagus nerve triggers a peripheral (efferent) or a central (afferent) pathway. Vagotomy (VGX) of the vagus nerve and specific stimulation of its distal part induces the activation of the splenic nerve to release norepinephrine, which in turn activates beta2-adrenoceptors (β2AdrRs) in modulatory cholinergic lymphocytes (Chly) to produce acetylcholine (ACh). Acetylcholine inhibits the production of inflammatory factors in splenic macrophages via alpha7 nicotinic acetylcholine receptors (α7nAChRs). Figure templates extracted from the Servier Medical Art (https://smart.servier.com)

Our results also showed that electrical stimulation of the vagus nerve attenuates serum TNF levels in wild-type but not in lymphocyte-deficient nude mice (Peña et al., 2011; Vida et al., 2011b). Thus, electrical stimulation of the vagus nerve activates the splenic nerve to release norepinephrine, which activates cholinergic lymphocytes in the spleen to inhibit cytokine production in resident macrophages (Figure 2) (Peña et al., 2011; Vida et al., 2011b). These findings were later confirmed by other investigators showing that cholinergic lymphocytes produced acetylcholine to inhibit cytokine production in macrophages via α7nAChRs (Rosas-Ballina et al., 2011), and agreed with the previous studies showing ChAT expression in lymphocytes of ChAT-Cre-tdTomato mice (Gautron et al., 2013). In fact, earlier studies reported that ChAT is present in various immune cells including NK cells (Jiang et al., 2017), macrophages (Verlinden et al., 2019), and lymphocytes (Martelli et al., 2019; Reardon et al., 2013). Thus, stimulation of the parasympathetic vagus nerve activates the sympathetic splenic nerve, and the two autonomic systems work in synchrony to modulate systemic inflammation in critical conditions such as sepsis.

Recently, Bratton et al. (2012) used electrophysiologic techniques to monitor the splenic nerve during vagal stimulation to determine whether vagal afferent or efferent fibers are involved in the cholinergic anti-inflammatory pathway. Cervical vagal stimulation activated the splenic nerve, but nerve activity remained constant after removing the afferent pathway by crushing the vagus above the stimulation electrodes (Bratton et al., 2012). The splenic nerve responded only after stimulation of the splanchnic nerve, i.e., sympathetic preganglionic neuron. The investigators suggested that the vagus nerve affects the splenic nerve reflexively, but not directly. Although these results may depend on experimental conditions, they concur with previous findings that vagal stimulation may also control the spleen via an afferent pathway through the central nervous system (Abe et al., 2017; Vida et al., 2011a).

6. A DIRECT VAGAL SYNAPSE TO THE SPLEEN?

From an anatomic perspective, early macro- and microanatomic studies discarded the notion of direct vagal innervation of the spleen. However, more recent analyses indicate that parasympathetic nerve branches may innervate the spleen. Inoculation of pseudorabies virus, a retrograde neurotracer, into the splenic hilus infected pre-ganglionic neurons of the spinal cord and cell bodies of the dorsal motor nucleus of the vagus nerve in the brain (Buijs et al., 2008). After denervation of the splenic hilus, virus inoculation infected only neurons in the brain, suggesting additional sites of splenic innervation. In search of these unknown sites, Buijs et al. (2008) removed all connective tissue from the hilus up to both extremities of the spleen, leaving intact only the connective tissue along the splenic artery. Surprisingly, after “denervation” of the splenic apices, virus inoculation into the splenic hilus infected neural cells bodies in the spinal cord, but not in the brain, suggesting that vagal fibers enter the spleen somewhere between the hilus and the apices. Cailotto and colleagues (2012) reported similar results showing infected neurons in the dorsal motor nucleus after inoculation of cholera toxin B at both splenic extremities. After denervation of the splenic apices, they found no toxin in the dorsal motor nucleus. Although both studies discarded possible tracer leakage, their findings did not provide histological evidence to support innervation of the spleen. (For comments, please see Anderson et al (2015)).

Recent studies of Guyot and colleagues (2019) have just provided the first histologic evidence of an independent apical splenic innervation in mice. They identified three distinct nerve-like structures that appeared to innervate the spleen. Two branches entered the spleen via the splenic artery at the level of the hilum, and the other entered the spleen via both apices. Only the apical branches had cholinergic characteristics. However, these cholinergic fibers were identified only on the outer surface of the spleen and not deeper into the organ parenchyma. Still, stimulation of this apical nerve increased splenic acetylcholine levels in lymphocyte-deficient nude mice, indicating functional cholinergic nerve activity. Almost at the same time, Verlinden and colleagues (2019) have just reported that their 3D reconstruction of human spleen innervation confirms the presence of apical splenic branches. However, these new branches lack cholinergic markers (ChAT or AChE), and arise from the splenic nerve surrounding the splenic artery, e.g., the hilus region. Furthermore, anatomical analyses showed that these apical branches make a turn and enter the spleen via the splenic hilum, and intrasplenic analysis showed that these branches surround only the trabecular system and central arteries. Interestingly, macroanatomic studies did not find direct vagal innervation of the spleen or an extra splenic branch, even in large mammals such as horses and oxen. Such a discovery in large mammals would be remarkable, and perhaps could be traced back along mammalian evolution. However, we cannot rule out that the apical splenic nerve is an unusual anatomic variation of the splenic nerve or simply a rogue branch from elsewhere, and future research will be required to determine whether this apical splenic nerve is functional.

Other studies have indicated alternative pathways between the vagus nerve and the spleen. For example, we showed that vagal stimulation controls joint inflammation independent of the spleen in experimental arthritis (Bassi et al., 2017), and transdermal neuronal stimulation of the sciatic network with electroacupuncture improves sepsis survival by activating the adrenal to release dopamine (Torres-Rosas et al., 2014). In addition, Vida et al (2011b) reported that splenic cytokine production is inhibited by stimulating β2-adrenoceptores in cholinergic lymphocytes, which was later confirmed by other investigators (Rosas-Ballina et al., 2011) (Figure 2). Indeed, Martelli and colleagues (2014) suggested non-neural connections in which stimulation of splenic nerve terminals occurs via incoming ChAT+ T cells from vagal targets. In fact, as mentioned earlier, several studies have found ChAT in various immune cells including NK cells (Jiang et al., 2017), macrophages (Verlinden et al., 2019), and B-lymphocytes (Reardon et al., 2013).

7. CLINICAL IMPLICATIONS

Electrical stimulation of the vagus nerve was approved by the FDA for the treatment of refractory epilepsy in 1997, and for the treatment of chronic depression in 2005. During these years, electrical vagal stimulation has proven to be safe and without significant side effects. These treatments involve an electrical generator that is surgically implanted under the skin of the chest and connected to electrodes wrapped around the cervical left vagal trunk. The electrical polarity of the stimulation is directed toward the central nervous system to modulate brain activity in epilepsy and depression. Investigators are now changing the polarity to induce efferent signals toward the periphery to control inflammatory disorders such as rheumatoid arthritis (Koopman et al., 2016) and Crohn’s disease (Bonaz et al., 2016). Vagal stimulation reduced blood C-reactive protein, fecal calprotectin, and abdominal pain, and improved mood in 5 out of 7 patients with Crohn’s disease (Bonaz et al., 2016). Vagal stimulation also improved clinical symptoms of rheumatoid arthritis, and significantly lowered serum levels of inflammatory cytokines such as TNF and IL6 in 18 patients after 3 months of treatment (Koopman et al., 2016). Additional clinical trials are in progress to analyze the effects of vagal stimulation in other inflammatory conditions such as polyneuropathies ( NCT04053127), heart failure ( NCT03425422), and traumatic brain injury ( NCT02974959). (Please see summary in Table 4 and Bonaz et al. (2016)). Thus, electrical vagal stimulation is a promising approach for treating chronic inflammatory disorders such as Crohn’s disease and rheumatoid arthritis.

Table 4.

Clinical trials (and their identifiers) that are using vagal stimulation to treat inflammatory conditions.

Clinical Condition	Participants	Stimulation Device	Clinical Trials (status)	Country of Study	Inflammatory Outcome
Acute ischemic stroke	*150	Transcutaneous	NCT04050501 (not started)	Netherlands	Infarcted area
Acute stroke	*60	Transcutaneous	NCT03733431 (recruiting)	Turkey	Morbidity and mortality
Atrial fibrillation	53	Transcutaneous	NCT02548754 (active)	USA	Blood cytokines
Chron’s disease	15	Implantable	NCT02311660 (active)	USA	Crohn's Disease Activity Index
Demyelinating polyneuropathy	*10	Implantable	NCT04053127 (recruiting)	USA	Nerve conduction Blood cytokines
Gastroparesis	*45	Transcutaneous	NCT03120325 (recruiting)	USA	Intestinal motility
Heart failure	*80	Transcutaneous	NCT02898181 (recruiting)	USA	Blood cytokines and CRP levels
Heart failure	*50	Transcutaneous	NCT03945058 (not started)	USA	Blood cytokines
Heart failure	*72	Transcutaneous	NCT03327649 (recruiting)	USA	Blood cytokines
Heart failure	*800	Implantable	NCT03425422 (recruiting)	USA	Cardiovascular parameters Mortality
Inflammatory bowel disease	*30	Transcutaneous	NCT03863704 (recruiting)	USA	Blood cytokines
Juvenile Idiopathic Arthritis	*12	Transcutaneous	NCT01924780 (recruitment by invitation)	Sweden	In-vitro production of cytokines from whole blood by LPS
Lung cancer	*12	Transcutaneous	NCT03553485 (recruiting)	Belgium	Blood cytokines
Lupus Erythematosus	*18	Transcutaneous	NCT02822989 (recruitment by invitation)	USA	Musculoskeletal pain
Osteoarthritis	*20	Transcutaneous	NCT03919279 (recruiting)	France	Pain and motor function
Postoperative Atrial Fibrillation	42	Transcutaneous	NCT03533140 (active)	Austria	Blood cytokines
Postoperative ileus	*130	Intra-operatory	NCT02524626 (active)	Belgium	Gastrointestinal transit and tissue cytokines
Rheumatoid arthritis	18	Implantable	NCT01552538 (active)	USA	Disease Activity Score
Sepsis	*34	Transcutaneous	NCT03992378 (not started)	USA	Blood cytokines
Stroke	17	Implantable	NCT02243020 (active)	USA	Motor function
Stroke	*120	Implantable	NCT03131960 (recruiting)	USA	Motor function
Thoracic Surgery	*200	Transcutaneous	NCT02783157 (recruiting)	USA	Morbidity and mortality Blood cytokines
Traumatic Brain Injury	*30	Implantable	NCT02974959 (recruiting)	USA	Bradycardia Blood cytokines
Ulcerative Colitis	*46	Transcutaneous	NCT03908073 (recruiting)	UK	In-vitro production of cytokines from whole blood by LPS

Data obtained from the (NCT) ClinicalTrials.gov.

^*: expected number of participants.

Given that the right vagal cervical trunk has a more direct effect on the cardiovascular system, most clinical trials prefer to stimulate the left vagal cervical trunk to avoid potential side effects on the cardiovascular system (Groves and Brown, 2005). However, stimulation of the right vagus nerve could be useful for treating inflammatory conditions related to the heart, such as heart failure (Li et al., 2004). Although the right vagal branch has a larger surface area and twice the number of catecholaminergic fibers compared to the left branch (Verlinden et al., 2016), laterally unique anti-inflammatory mechanisms of the cervical vagus nerve are not likely. Indeed, experimental studies have shown no differences in anti-inflammatory effects after stimulating the right vs left cervical vagus nerve (Bernik et al., 2002; Olofsson et al., 2015). Likewise, subdiaphragmatic stimulation of the anterior vs the posterior trunk reduced inflammation in a similar fashion (Stakenborg et al., 2017). However, some researchers suggest that stimulation of different subdiaphragmatic nerve branches, e.g., hepatic, gastric, or celiac, may have unique effects on inflammation (Somann et al., 2019), and nerve implants to stimulate specific subdiaphragmatic branches are now in clinical trials for treating morbid obesity (VBLOC therapy) (de Lartigue, 2016).

Electrical vagal stimulation also provides significant advantages for treating acute infectious and inflammatory disorders. Indeed, early studies on vagal stimulation revealed its potential to attenuate systemic inflammation in experimental sepsis including endotoxemia and polymicrobial peritonitis (Bernik et al., 2002; Borovikova et al., 2000). Likewise, a recent study reported that cervical vagal stimulation increases E. coli internalization by hepatic Kupffer cells in septic mice (Fonseca et al., 2019). Although cellular TNF levels did not decrease, vagal stimulation increased ChAT levels in the Kupffer cells (Fonseca et al., 2019). This mechanism is similar to that by which vagal stimulation induces acetylcholine in cholinergic lymphocytes (Peña et al., 2011; Vida et al., 2011b). Ultrasound-guided vagus nerve stimulation also reduces systemic TNF levels after endotoxemia and rescued delirium-like behavior in mice (Huffman et al., 2019). However, these acute conditions represent a technological challenge to design noninvasive approaches such as transdermal vagal stimulation (Ulloa et al., 2017). We demonstrated that transdermal nerve stimulation with electroacupuncture decreases serum TNF levels, and rescues mice from polymicrobial peritonitis (Torres-Rosas et al., 2014). Although the precise mechanisms of nerve stimulation with acupuncture or electroacupuncture are not well known, they provide a non-invasive strategy for selective nerve stimulation with limited side effects (Grech et al., 2016; Ulloa et al., 2017). Neuronal stimulation on acupoint ST36 activates a sciatic-vagal network that can induce the production of dopamine in the adrenal medulla (Torres-Rosas et al., 2014). Recently, Cotero et al. (2019) have reported that noninvasive ultrasound stimulation of the spleen can activate neuromodulation and the cholinergic anti-inflammatory pathway to reduce cytokine response to endotoxin similar to that induced with implant-based vagal stimulation. Although the precise mechanisms of nerve stimulation with ultrasound is unknown, previous studies aimed to stimulate large peripheral nerves (outside the organ) produced conflicting results, recent studies reported successful activation of nerve terminals inside organs such as brain, kidney, or spleen (Cotero et al., 2019; Gigliotti et al., 2015). Non-invasive neural approaches, such as electroacupuncture (Torres-Rosas et al., 2014), transcutaneous electromagnetic fields (Rossi et al., 2009), focused ultrasound (Cotero et al., 2019; Gigliotti et al., 2015), or nerve fibers (Abe et al., 2017; Downs et al., 2018), may be useful to stimulate specific neurons to control organ function. These studies also revealed the relevance of vagal innervation to specific organs, such as the spleen or the intestines (Willemze et al., 2015), and its implications in healthy physiologic homeostasis and treatment of infectious and inflammatory disorders to reestablish physiologic homeostasis.

One fundamental clinical challenge is identifying the pathologic conditions that interfere with neuronal modulation, i.e., the potentially unresponsive cohort of patients. Indeed, most experimental studies on inflammatory models and neural stimulation are performed in healthy animals. Further, many experimental anti-inflammatory strategies that were successful in rodents, failed in clinical trials, in part, because normal healthy mice do not mimic patients in clinical settings (Feketeova et al., 2018; Joseph et al., 2019). For example, vagal stimulation attenuates serum TNF levels in experimental sepsis via a mechanism that is mediated by the spleen and thus, splenectomized patients are more susceptible to sepsis. Another example is that neuronal stimulation on the acupoint GB30 in healthy uninjured rats decreases thermal nociception by activating delta and mu opioid receptors at low-frequencies under 10Hz, but via kappa receptors at high-frequencies above 100Hz (Han, 2003). However, in injured rats treated with a plantar injection of complete Freund's adjuvant (CFA), stimulation with both low and high frequencies on GB30 decreases nociception via mu and delta opioid receptors, but not via kappa receptors (Kim et al., 2009; Zhang et al., 2004). Thus, kappa receptors can control thermal nociception in control rats but not in injured rats. Further, electrical stimulation on the acupoint ST36 can improve rodent survival in experimental sepsis by inducing production of dopamine in the adrenal medulla (Torres-Rosas et al., 2014). However, some septic patients have adrenal insufficiency and therefore, become unresponsive to this neuronal mechanism. Given that dopamine mediates the anti-inflammatory potential of electroacupuncture, these patients can then be treated with specific dopaminergic agonists that mimic the neuronal mechanism of modulation. Importantly, dopamine is used in critical care as the precursor of norepinephrine, which has stronger cardiac dromotropic and inotropic effects. However, dopamine does increase the risk for tachyarrhythmia (Povoa et al., 2009; Sakr et al., 2006), and can worsen survival in septic animals (Oberbeck et al., 2006). Thus, selective dopaminergic agonists, such as the selective D1 receptor agonist fenoldopam, can avoid these side effects of dopamine, and confer protection to renal function (Joseph et al., 2019; Morelli et al., 2005). It should be noted here that the efficacy of electroacupuncture is controversial because of the lack of proper mechanistic explanations for its effectiveness in some diseases and patients, but not in others with similar symptomatology. Nevertheless, taken together, these studies reveal that innervation of specific organs, such as the spleen and adrenal glands, is not only important in normal healthy physiology but also under pathologic conditions. Thus, these mechanistic studies will allow the design of novel noninvasive techniques for nerve stimulation as well as potential pharmacologic treatments for unresponsive patients.

8. SUMMARY AND CONCLUSIONS

Regulation of organ function and physiologic homeostasis is finely tuned by a combination of the nervous and immune systems. In this review, we focused on vagal innervation of the spleen and its role in modulating inflammation. Although functional vagal innervation to the celiac ganglion is debated, most studies showed absence of neuronal acetylcholine and related markers, such as ChAT and AChE, in the splenic parenchyma. Despite initial studies that suggested a vagus/spleen connection via the celiac ganglion, recent studies indicate alternative pathways with similar anti-inflammatory properties that can have clinical implications in infectious and inflammatory disorders. It is also important to acknowledge vagal innervation of other organs and tissues, such as the adrenal glands and the intestines, to understand its biological activity and clinical implications to control organ function and physiologic homeostasis.

Highlights.

• The vagus nerve can control inflammation and organ injury in infectious and inflammatory disorders by regulating the production of inflammatory factors in the spleen.

• Most studies showed absence of neurogenic cholinergic markers in the spleen.

• Recent studies suggest that vagal regulation of inflammation is mediated by multi-synaptic interactions between the vagus and the splanchnic nerves

• These mechanisms are allowing the design of novel therapeutic strategies for infectious and inflammatory disorders.