Macrophages in the Smooth Muscle Layers of the Gastrointestinal Tract

Abstract

Muscularis macrophages are a newly discovered population of immune cells populating the smooth muscle layers of the gastrointestinal tract. Beyond their well-established role in modulating innate immunity, these cells are emerging for their ability to communicate with cells required for gastrointestinal motility. This chapter will describe the factors contributing to muscularis macrophages’ phenotype and the functional connections these cells established with different cell types.

1. Introduction

The gut is home to the body’s largest population of immune cells [1, 2]. Beyond the frontline defenses against unparalleled exposure to foreign antigens, gut macrophages (MΦ) also constantly communicate with an intricate network of cell types orchestrating gastrointestinal (GI) functions [3, 4]. By now, the plasticity of MΦ in different organs of the body (inter-diversity) and within tissue layers (intra-diversity) has been established. While the M1/M2 dogma has served to advance the field of Immunology and understand MΦ phenotype and function [5, 6], immunologists now generally agree that there is a spectrum of phenotypes between the two classifications [7, 8, 9]. This chapter will outline the phenotype and function of a population of MΦ in the GI tract, called muscularis macrophages (MMΦ). As the name states, MMΦ resides in the muscularis layers of the GI tract, called muscularis propria. MMΦ fulfills multiple functions across development, adulthood, and under disease conditions. This chapter will report the factors contributing to MMΦ’ phenotype heterogeneity and describe the functional interaction MMΦ establish with cells populating the same environment.

2. Identification of MMΦ: a new population of macrophages

Mikkelsen and colleagues first identified “macrophage-like” cells in the muscularis propria of the small intestine using combination of electron microscopy and immunohistochemistry [10]. Subsequent studies by the same group confirmed these cells to be MMΦ after endocytosis of FITC-dextran and F4/80 labeling co-labeled. In addition to the identification of MMΦ by immunohistochemistry, electron microscopy analysis revealed distinct morphological features based on location within the muscularis propria. These cells were noted to contain a nucleus, Golgi bodies, smooth and rough endoplasmic reticulum, and enveloped by the processes of interstitial cells of Cajal (ICC). The GI muscularis propria (Figure 1) is separated from the external environment by the mucosa, which is in constant contact with ingested food along with the gut microbes and other pathogens.

Figure 1.

Different regions of the gastrointestinal tract along with the varying morphologies of MΦ.

The primary function of the mucosa is to absorb the dietary nutrients and protect against the different external stimuli [11, 12, 13]. On the other hand, the muscularis propria is mainly responsible for coordinating contractions for proper food movement through the GI tract. The cellular anatomy of the muscularis propria is complex and characterized by different regions. The two muscular layers are called longitudinal and circular, respectively, based on their orientation. The myenteric plexus, also known as the Auerbach plexus, contains a significant number of enteric neurons (ENs) and pace-making ICC, which regulate peristalsis between the two muscle regions [14, 15, 16, 17] (Figure 1). Advanced technologies further demonstrated MMΦ’s heterogeneity within the muscularis propria, as recent studies described differences in MMΦ distribution and phenotype within different regions of the GI muscularis propria. Current understanding revolves around MMΦ acquiring distinct phenotypes upon exposure to intrinsic GI cues. However, emerging evidence suggests that MMΦ display differences in their functions and phenotypes that are not exclusively driven by the GI milieu but also by their ontogeny.

3. MMΦ heterogeneity: ontogeny vs. environmental cues

For the longest time, the origin of MΦ was attributed entirely to circulating blood monocytes, which after engrafting the tissue, acquired tissue-resident MΦ resemblance [18, 19, 20]. However, studies in the late 2010s challenged this paradigm as scientists theorized that populations of resident MΦ in homeostatic tissues also derived from embryonic progenitors of the yolk sac and fetal liver [21, 22, 23]. After this finding, there is now a consensus on the double origin of tissue-resident MΦ in multiple organs, where monocyte-derived- and embryonic MΦ coexist [24]. Embryonic MΦ are established before birth and their number is maintained by cell division, independently of circulating monocytes’ recruitment. On the other hand, a population of MΦ is continuously replenished by a monocyte waterfall wherein circulating adult monocytes ingress into tissues and differentiate to tissue-resident MΦ [24].

While MMΦ ontogeny in homeostatic condition has been recently studied [25], most of the research in the last couple of decades has primarily been focused on disease models. Like microglia in the central nervous system (CNS), MMΦ contains populations of different origins, as cells of both embryonic and circulating monocyte origin constitute the entire pool of tissue-resident MMΦ. MMΦ as microglia in the brain express the high level of CX3CR1 as a canonical marker of tissue-resident cells.

Using a lineage tracing mouse model, CX3CR1 MMΦ were followed from the embryonic stage to adulthood. This population represents the totality of tissue-resident MMΦ at the embryonic stage and declines with time. This decline starts between 4 and 20 weeks after birth and stops after this timepoint. Consequently, the population that remains seeded has come to be known as long-lived MMΦ. Importantly, this population of embryonic MMΦ has a different transcriptional profile compared with monocyte-derived MMΦ. In fact, a population of tissue-resident MMΦ expresses high levels of CX3CR1, also known as CX3CR1^hi. CX3CR1^low MMΦ, on the contrary—as the name suggests—express low levels of CX3CR1 and highly express C-C chemokine receptor 2 (CCR2). CCR2 is involved in monocyte homing in response to inflammation in the local tissue environment [26, 27]. Resident MMΦ exhibits a general anti-inflammatory phenotype at steady-state conditions compared to the more inflammatory phenotype of mucosal MΦ. This is underscored by the expression of their wound healing and tissue-protective genes.

Long-lived MMΦ express genes responsible for cell-to-cell adhesion, cytoskeletal anchoring, and neuron development, suggesting their anatomical association with ENs. In addition, 26% of the genes enriched in a subpopulation of long-lived MMΦ is unique compared to the data set from tissue-resident MΦ populating other tissues. However, most of this population express gene previously associated with microglia in the brain, such as Fc receptor-like scavenger (FCRLS), cystatin C (CST3), platelet factor 4 (PF4), apolipoprotein E (ApoE), and disabled-2 (Dab2). Although a subtype of MMΦ is maintained by cell division, most tissue-resident MMΦ are continuously replenished by circulating monocytes. In fact, bone marrow transplanted CX3CR1 ^EGFP cells engrafted into the muscle layers, and within 4 weeks of transplantation, they re-established the number of tissue-resident MMΦ [28].

A study aimed at differentiating between long-lived MΦ and monocyte-derived MΦ used Tim4 and CD4 as surface markers to separate the two populations [29]. Long-lived MΦ were Tim-4⁺/CD4⁺ as evidenced by the slower rate of turnover compared to Tim-4^{− /}CD4⁻ cells as expressed by chimerism of cells post-irradiation. Although the total pool of small bowel MΦ was considered, this segregation is largely driven by mucosal tissue-resident MΦ in commensal-rich areas as noted in the study due to the higher rate of monocyte-macrophage turnover. Interestingly, Tim-4 is as an important apoptotic cell uptake receptor indicating that these long-lived MΦ might be playing a crucial role in efferocytosis, a process which has been shown to resolve inflammation in the brain by microglia [30], and other tissue types [31].

A new population of embryonic MMΦ called perivascular (PVMs) was recently identified in anatomical association with blood vessels within the muscularis propria of the ileum and small intestine [32]. The authors identified a gene—musculoaponeurotic fibrosarcoma (Maf)—which is required for the development of this subpopulation in white adipose tissue (WAT). This population, named vasculature-associated macrophages (VAMs), existed in all organs in proximity to blood vessels. This commonality led them to understand the effect of Maf regulation in the muscularis propria of the ileum and small intestine since they harbor MMΦ expressing similar markers to VAMs, specifically, VAM2. Furthermore, when the Maf gene was deleted, there was a total loss of CD206+ MMΦ in the small and large intestines, implicating its critical role in their phenotype and function.

All these data suggest the bivalent origin of tissue-resident MMΦ with the coexistence of monocyte- and embryonic-derived MMΦ. This level of heterogeneity has been recently described in the single-cell transcriptomic analysis of colonic MMΦ. Colonic MMΦ can be divided into three different populations, including “transient” monocyte-like MMΦ expressing high levels of calprotectin (heterodimer of S100A8 and S100A9) and long-lived calprotectin-negative MΦ expressing a TRM phenotype [33, 34].

Further investigation is needed to outline the key similarities and differences between murine and human MMΦ distribution, morphology, and composition. Although the murine monocyte waterfall in the intestine depicts a detailed transition of Ly6C^hi CX3CR1^int monocytes to CX3CR1^hi MHC-II^hi CD64+ MMΦ, information on human MMΦ is sparse, generally relied upon immunohistochemistry and morphological studies.

Another study by Bernardo and colleagues tracked the transition phenotype of human monocytes to tissue-resident MΦ [35]. It was found that CD14⁺ monocytes differentiated into inflammatory monocyte-like cells upon entering healthy and inflamed colonic mucosa. These cells, identified by CD11c^hi CCR2^hi CX3CR1⁺ expression, then transitioned through an intermediate phenotype of CD11c^dim CCR2^low CX3CR1^low before finally becoming tissue-resident—or tolerogenic—MΦ with a CD11c⁻ CCR2⁻ CX3CR1⁻ signature. As is true in the mouse model, CCR2 remains critical in recruiting monocytes in humans. Furthermore, the authors found that homing was abrogated when it was blocked on monocytes before migration. Whether this is true in MMΦ is yet to be elucidated. Changes to monocyte-derived or long-lived embryonic MMΦ can alter the total MMΦ number in diseases where homeostasis is challenged. For example, in diabetic and diabetic gastroparetic mice [36], there is an increase in MMΦ which is linked to the recruitment of inflammatory circulating monocytes.

MMΦ phenotype depends on regional distribution across the muscularis propria and the interaction MMΦ established with other cell types populating the same environment. MMΦ have a different morphology [37] in the different regions of the muscularis propria. MMΦ located in the myenteric plexus and serosal regions is multipolar, with many branches originating from the main body (Figure 1).

On the other hand, MMΦ distributed within the muscular layers have a bipolar shape that follows muscle cells’ orientation. Further data are needed to understand if morphological differences between these diverse MMΦ populations translate into functional changes. Such differences have been easier to study in the brain since the various CNS regions are accessible and can be separated. Whereas this has enabled the study of microglial phenotype residing in each area, it is more complicated to get the same type of information from the muscularis layers of the GI tract due to the technical difficulties in separating the different regions.

Long-lived MMΦ occupy a specific anatomical niche within the GI muscularis propria. De Schepper and colleagues [25] identified this MMΦ population as essential for maintaining ENs, located within the myenteric plexus region where they interact with ENs. The critical role of the environmental cue in shaping MMΦ phenotype is evident as embryonic MΦ are also present in the mucosa, but they have an overall distinct phenotype compared to long-lived MMΦ. Interestingly, another population of embryonic MMΦ does not express CX3CR1, but in this case, it appeared to be anatomically coupled with blood vessels [32].

Another essential feature contributing to MMΦ heterogeneity is the location of MMΦ in different regions of the muscularis propria. Whereas in the small intestine MMΦ are mostly concentrated in the myenteric plexus, gastric MMΦ are evenly distributed between the myenteric plexus and smooth muscle layers [38]. Further studies are needed to understand if this difference in MMΦ distribution is also responsible for functional changes. Phenotypically, at resting, gastric MMΦ do not express high levels of CD206 as MMΦ from the small intestine does. MMΦ density also differs between the small intestine and colon in young mice, with a reduction of MMΦ density observed in the colon [39]. However, this type of difference was no longer observed in adult mice. It also appeared that MMΦ in the different gut regions responds to external stimuli differently, suggesting a possible intrinsic phenotypic difference. For example, in diabetes, gastric MMΦ change their phenotype leading to gastric dysfunction, whereas, in the context of the same disease, MMΦ in the small intestine are unchanged [36].

Although we have a clearer picture of MMΦ distribution in different smooth muscle layers at steady-state conditions, we have only partial information about their distribution in states of altered homeostasis and disease. In aging, clusters of CD163-IR immune cells are visualized in proximity to sympathetic hyperinnervation in the jejunum of rats [40]. In a mouse model of diabetic gastroparesis, an increased number of MMΦ has been described with the onset of diabetes, but no changes in the MMΦ distribution have been reported [41].

More studies are needed to understand the differences between the populations of MMΦ residing in the different gut regions looking at (1) phenotypic changes, (2) changes in response to inflammation/stimuli, and (3) origin.

4. MMΦ: new players of gastrointestinal motility

As anticipated in the previous section of the chapter, the muscularis propria contains numerous ENs that work in concert with the CNS to control digestive function. The enteric nervous system (ENS) has 200–600 million ENs distributed in thousands of small ganglia [15, 16]. Importantly, the ENS can function independently from the CNS to control digestive function. The GI tract is innervated by intrinsic ENs and the CNS axons of extrinsic sympathetic and parasympathetic neurons.

Since some MMΦ are closely associated with ENs, this raises the following questions: Do these MMΦ functionally interact with enteric nerves, and what does such communication entail?

4.1 Functional interaction between MMΦ and intrinsic innervation

MMΦ-ENs functional interaction has been studied extensively in the homeostatic and diseased gut. Muller and colleagues showed for the first time an active interaction between MMΦ and ENs in 2014 [42]. In this study, the investigators showed that MMΦ expresses a high level of bone morphogenetic protein 2 (BMP2) compared to mucosal MΦ. Notably, certain ENs express the receptor (BMP2r) that, upon interaction with BMP2, respond by a pSMAD1/5/8 related mechanism. This type of functional interaction is regulated by microbiota, as microbiota-free mice have reduced BMP2 expression. Depletion of MMΦ results in poorly coordinated colonic contractions in an ex vivo model and abnormal colonic transit time in vivo. Adding exogenous BMP2 to the colonic rings from MMΦ -deficient mice decreases stretch-induced contractions. Enteric neuron number results from a dynamic balance between the ENs dying by apoptosis and the continuous production of new ENs by neurogenesis. As microglia in the CNS, MMΦ played an essential role in clearing cellular debris resulting from neuronal death. In vitro models have shown the bidirectionality of this interaction. Oxytocin (OT) is traditionally considered a nonapeptide hormone synthesized in the hypothalamus that is released from the posterior pituitary into circulation and is involved in milk let-down and uterine contraction. Polarized pro-inflammatory MMΦ regulate the expression of OT and its receptor, OTR in cultured enteric neurons via STAT3 or NF-κB pathway [43].

On the other hand, TGF-β released by anti-inflammatory MMΦ induces the upregulation of OT/OTR [44]. Interestingly higher levels of pro-inflammatory cytokines correlated with a lower level of OT/OTR in DSS-colitis. In the colon, ENS is reduced by pro-inflammatory MMΦ via the GK1-FOXO3 pathway.

Most studies described a close association between MMΦ and nerve fibers. However, recently a paper for the first time also describes a rare population of MMΦ distributed within the ganglia [45, 46], which house the bodies of the ENs. In their study, the authors demonstrated that this population of MMΦ, called intra ganglionic macrophages (IGMs), has processes in this region of mouse colon, suggesting phagocytic capability. Colitis-induced mouse models are characterized by an increased level of pro-inflammatory MMΦ and associated with a reduction of IGMs. Notably, the loss of IGMs in colitis is associated with enteric neuroinflammation, characterized by neuronal hypertrophy.

A series of studies questioned the role of MMΦ in regulating the total number and the genetic coding of ENs. CX3CR1 MMΦ [25] of embryonic origin persisted with aging and remained primarily associated with ENs in the myenteric plexus region. The conditional removal of this population of MMΦ during development results in the overall reduction of ENs, leading to GI dysfunction. Csf1^op/op mice, which do not have MMΦ from birth, have an abnormal myenteric plexus and more ENs than control mice [41, 42]. Although the number of nitrergic ENs is increased in Csf1^op/op mice [47], the number of cholinergic ENs is not altered, suggesting that MMΦ may regulate different subtypes of ENs (Figure 2).

Figure 2.

Bidirectional interactions between EN and MMΦ comparing homeostasis and disease.

In the same animal model, Cipriani and colleagues also showed that the absence of MMΦ from birth is associated with more ENs sharing cholinergic and nitrergic phenotypes, indicative of a more undifferentiated population of ENs. A reduced number of anti-inflammatory MMΦ in aged mice is linked to ENs loss [48, 49].

To further understand this intimate relationship between MMΦ and the ENS, papers understanding their dynamics during development have shed some light on this. MMΦ colonize the gut independently of ENs at E9.5. Although they engraft into the muscularis propria, tissue-resident MMΦ are not close to ENs but are closer to other cells populating the same environment during this period [50]. The investigators identified a plausible explanation for this observation as the absence of CSF1 release by ENs during this period, contrary to adulthood where ENs represent the primary source of CSF1. An increasing number of MMΦ engraft into the muscularis propria during development, where they occupy a distinct niche compared to mucosal MΦ and establish an intimate connection with neuronal processes [51]. Conditional depletion of irf8, a gene enriched in this population of MMΦ, leads to impaired intestinal GI motility. More studies are needed to dissect further the possible role of MMΦ in orchestrating ENs differentiation and distribution.

4.2 Functional interaction between MMΦ and extrinsic innervation

The amount of data describing the functional interaction between tissue MMΦ, and peripheral nerves is limited, mainly because the number of MMΦ is minimal compared to the total number of tissue MΦ. For example, MMΦ expressing CX3CR1 are closely associated with sympathetic nerve fibers of adipose tissue [52]. Precise extrinsic afferent (visceral sensory) and efferent (sympathetic and parasympathetic) innervation of the gut is fundamental for gut-brain cross talk. While the extrinsic nerves do not directly modulate gut motility, they can affect it by regulating other cell types within the ENS [53]. Interactions between MMΦ and extrinsic innervation and the effect of sympathetic and catecholaminergic signaling in the immune cells’ modulation of multiple organs have been extensively studied in the past [54]. However, their possible involvement in modulating MMΦ polarization phenotype in the gut has been described only recently. In their study, Gabanyi and colleagues [37] suggest the regulation of MMΦ activation by the β-adrenergic receptor β-2AR receptor (β2AR). Β2AR⁺ MMΦ resides near neuronal cell bodies or processes of the myenteric ganglia. Because of this interaction, MMΦ expresses higher levels of β2AR, a neuropeptide receptor, than lamina propria MΦ. Notably, adrenergic signaling through this receptor reduces ENs loss following infection [55].

Acetylcholine (Ach) represents the primary parasympathetic neurotransmitter released by preganglionic nerve fibers and the vagus nerve. Ach has been studied for its anti-inflammatory effects in the periphery, as its stimulation is sufficient to suppress systemic inflammation in response to endotoxin. Cholinergic neuronal release during vagal nerve stimulation (VNS) induces an anti-inflammatory MMΦ phenotype activation via the alpha7 nAChR (α7nAChR), ameliorating muscular inflammation [56]. In addition, vagal manipulation leads to an increased number of gastric MHCII+ MMΦ, resulting in delayed gastric emptying [57].

The vagal nerve represents the longest nerve in the body and the main component of parasympathetic innervation. It is well established that the vagal nerve innervating the GI tract originates from 2 central regions of the CNS: the ambiguous nucleus and the dorsal motor nucleus of the vagus. This indeed represents one of the most studied roots to access the ENS from the CNS [58]. VNS mediates MMΦ anti-inflammatory phenotype activation in a model of inflammation induced by mechanical stimulation of the mucosa. This pathway is independent of vagal stimulation from the spleen as vagus denervation from the spleen did not prevent MMΦ activation. In this mouse model, the activation of anti-inflammatory MMΦ through VNS reduced the overall level of inflammation in the tissue. It appeared that this type of pathway is effective through the α7nAChR since MMΦ from α7nAChR knockout mice did not respond to VNS [56]. Extrinsic vagal innervation is involved in regulating contractions generated by the stomach. Preclinical studies underlined this pathway’s involvement and its possible therapeutic role in preventing muscularis propria inflammation in inflammatory bowel disease (IBD). Mice in which the vagus is resected develop severe colitis associated with increased pro-inflammatory cytokine levels such as IL-1β, IL-6, and TNF-α [59].

Interestingly, patients with depression and psychological stress are typically associated with adverse and worst forms of ulcerative colitis. This important association also translates into animal models of depression, which are more prone to develop forms of colitis [60]. Notably, the beneficial effect of antidepressant drug application is abolished after vagotomy. Although this mechanism is not entirely understood, the transfer of MΦ from animal models with depression made the recipient mice more susceptible to developing forms of colitis [61].

Gastroparesis is a disease that affects the stomach and is associated with impaired motility and increased pro-inflammatory MMΦ. VNS stimulation in patients with gastroparesis induced anti-inflammatory MMΦ activation and an incremental improvement in symptom scores [62]. In addition, VNS prevents gastroparesis by inducing anti-inflammatory MMΦ through STAT3-JAK2 mediated mechanism. Abdominal surgery is often associated with pro-inflammatory MMΦ activation leading to an overall inflammation of the muscularis propria and affected gastric motility, which VNS3 abolishes [63].

It appears that this interaction is bidirectional, as ENs can also affect MMΦ phenotype, differentiation, and maintenance. In an animal model in which pharmaceutical and genetic sympathetic innervation is deprived, there is an increase of circulating monocytes that ingress into the muscularis propria compared to controls [64]. In addition, isolated MMΦ from sympathetic-deprived mice have anti-inflammatory phenotype and concomitant increment of pro-inflammatory phenotype that led to accelerated GI transit time. Intestinal manipulation in postoperative ileus (POI) promoted an increased level of anti-inflammatory ED1 MMΦ [65] in the colon blocked by an anti-α7nAChR antibody. The authors proposed a mechanism in which ENs released acetylcholine upon intestinal manipulation that binds its receptor (α7nAChR) on the surface [56]. This represents a mechanism that could be targeted to prevent POI. A mouse model of post-infectious irritable bowel syndrome (IBS) is associated with subtype-specific neuronal loss via NRLP6 and caspase 11 mechanism and dysmotility. Notably, in this model, β2-AR signaling depletion in MMΦ resulted in increased loss of ENs in the post-infectious IBS model. These results indicate that, while short-term depletion of MMΦ does not impact intrinsic enteric-associated neurons (iEANs) survival in the unperturbed state, MMΦ may play an iEAN-protective role during enteric infection [56].

5. MMΦ interactions with non-neuronal cells

5.1 MMΦ-ICC

The GI tract represents a highly heterogeneous system where multiple cell types coexist and contribute to GI contractility. Similar to the pacemaker cells of the heart, the gut contains cells called ICC that set up the GI contractility pattern. ICC was described more than 100 years ago by Ramon y Cajal. For many years these cells were characterized only by non-specific histological stains and later, more reliably, by electron microscopy. The ultrastructural features and the ICC’s anatomical distributions suggested their critical physiological roles: (1) they are pacemaker cells and propagate slow waves [66, 67], (2) they mediate both inhibitory and excitatory neurotransmission [68, 69, 70], (3) they work as mechanosensory cells.

Limited data describe MMΦ-ICC interaction in homeostatic conditions in the ENS. Electron microscopy and immunofluorescence analysis showed that MMΦ populations are closely associated with ICC [10, 14], suggesting a potential functional role for this type of interaction. Ji and colleagues [38] recently demonstrated that despite their close association, MMΦ rarely touches ICC, but they are separated by a space of 300 microns. In Csf1^op/op mice, that do not have MΦ from birth, ICC appears to have a normal distribution, and the level of expression of kit, a specific ICC marker, is not different from controls (Figure 3).

Figure 3.

Bidirectional interactions between ICC and MMΦ in homeostasis and disease.

The information relative to MMΦ-ICC functional interaction in GI disease is more extensive. Blockade of IL-17A signal reduced ICC loss in sepsis by affecting the overall number of pro-inflammatory MMΦ [71]. Numerous MMΦ are observed closely associated with ICC in the antiinflammation model compared to controls at ultrastructural level [72]. In the same model with the progression of inflammation, MMΦ has large phagosomes and lysosomes in the proximity of injured ICC, suggesting a possible ongoing phagocytic activity. During development, ICC releases CSF1 in the small intestine, thus contributing to MMΦ migration and survival during this period. IL-6 released by MMΦ during GI surgery promotes upregulation of miR-19a responsible for ICC depletion [73]. An increase of pro-inflammatory cytokines produced by MMΦ is associated with decreased c-kit positive ICC in the dilated colon of Hirschsprung disease (HSCR) associated with enterocolitis [74].

Most of the information acquired in recent years describing a functional interaction between MMΦ and ICC has been produced in the context of diabetic gastroparesis, a functional disease affecting the stomach. One of the main cellular changes observed in both mice and patients with diabetic gastroparesis is ICC depletion and changes to MMΦ composition. Conditioned media from pro-inflammatory activated MMΦ reduces kit positive ICC in-vitro [75]. The ineffectiveness of the same conditional media in the presence of a TNF-alpha neutralizing antibody suggests this cytokine’s implication in regulating kit expression. Patients with diabetic gastroparesis have fewer CD206⁺, anti-inflammatory MΦ. This reduction correlates with ICC loss, suggesting a protective role of anti-inflammatory MMΦ on ICC, that is impaired in diabetic gastroparesis [76]. Anti-inflammatory MMΦ also secrete interleukin 10 (IL-10) to induce heme oxygenase (HO1) expression [77], an enzyme that has a protective effect on ICC. Mice with delayed gastric emptying treated with exogenous IL-10 return to regular gastric emptying with higher levels of HO1 and better connected, more organized, and evenly distributed ICC networks [78]. Thus, treatment to promote MMΦ polarization to an anti-inflammatory phenotype may be a viable treatment for diabetic gastroparesis. Progression of diabetes and development of delayed gastric emptying is associated with increased levels of pro-inflammatory MMΦ and reduced anti-inflammatory MMΦ.

A small number of data suggest the contribution of MMΦ-ICC functional interaction in the development of other GI diseases. For example, Crohn’s disease is associated with ICC injury and changes to MMΦ morphology [79]. In Achalasia, ICC closely related to immune cells are preserved [80]. Cytokines released by MMΦ have been considered responsible for ICC network disruption in endothelin-B receptor null rat, a model of Hirschsprung’s disease [81]. Other functional diseases, such as slow transit constipation, are associated with ICC loss. In this disease, ICC loss depends on the release of exosome miR-34c-5p from MMΦ [82].

5.2 MMΦ/smooth muscle cells and fibroblast-like cells

Although MMΦ exists in the muscle layers of the gut, the limited research surrounding their spatiotemporal dynamics with smooth muscle cells (SMCs) has hampered their full understanding. Only recently has their spatial interactions with SMCs been characterized in a paper. In it, MMΦ establishes membrane-to-membrane contacts with SMCs forming structures akin to peg-and-socket joints observed using Transmission Electron Microscopy [38] (TEM) in both humans and mice. Due to this tight interaction, it is speculated that chemokines/cytokines that are released by SMCs and vice versa maybe be pertinent for the maintenance of homeostasis in the local environment and ensure proper gut motility (Figure 4).

Figure 4.

Bidirectional interactions between SMC and MMΦ comparing homeostasis and disease.

In fact, a study revealed a transient receptor potential vanilloid 4 receptor (TRPV4) mediated interaction between MMΦ and SMCs in the colon of mice [83]. TRPV4 is a biosensor that can detect mechanical, thermal, and chemical cues and has been implicated in various GI disorders. However, because its effect on gut motility was not established, the authors sought to determine its role, specifically in colonic motility. Using TRPV4−/− and TRPV4+/+ as comparisons, they found various indications of impaired colonic motility. For example, they found that TRPV4−/− mice had a significantly increased number of pellets retained in the colons as opposed to their WT controls. This led to the identification of a subtype of MMΦ expressing the TRPV4 channel responsible for colonic contractions independent of ENS input. This independence was confirmed by optogenetic stimulation of the same MMΦ while applying tetrodotoxin (TTX) which resulted in prostaglandin E2 release leading to spontaneous contractions.

Postoperative ileus (POI) is a common abdominal complication almost after every intra-abdominal surgery characterized by a prolonged absence of bowel movement. This leads to symptoms such as a distended abdomen, nausea, vomiting, and other complications that prolong patients’ stay in the hospital, furthering the epidemiological burden for all parties involved. Shortly after intestinal manipulation in mice, the MMΦ network in the muscularis propria evokes a localized inflammatory response that recruits neutrophils and mast cells. The extravasation of leukocytes in tandem with MMΦ [84] results in the synthesis of prostaglandins and nitric oxide which directly impair SMCs contractility leading to POI. In the inflammatory cascade that ensues initially, there is an upregulation of MIP-1a, IL-1B, IL-6, ICAM-1, and MCP-1 compared to WT controls, indicating that MMΦ recruits monocytes, which then become resident-MΦ, and subsequently attract more monocytes. The cytokines and chemokine released are injurious to the SMCs resulting in impaired contractility depletion of this MMΦ network has been shown to decrease overall inflammation and prevent POI completely [85].

Crohn’s disease is an inflammatory bowel disease (IBD) characterized by persistent inflammation that typically affects both terminal ileum and colon. Due to simplicity, reproducibility, and commonalities associated with humans, TNBS-colitis in the murine model is used to study Crohn’s disease [86]. In this model, MCP-1 RNA and protein levels are upregulated in the muscularis layer of the gut [46]. Similarly, in humans, inflammation resulting from Crohn’s is associated with an increased number of pro-inflammatory MMΦ, neutrophils, and other immune cells recruited in the smooth muscle layers via a CCL2 and MCP-1 dependent mechanism [87]. Consequently, prolonged inflammation and hypertrophy of the surrounding smooth muscle layers have been observed [88]. Further studies are required to properly understand how MMΦ contribute to the pathogenesis of Crohn’s disease.

In the process of characterizing ICC at the ultrastructural level, cells with morphological similarities to ICC were discovered [89]. These fibroblast-like cells were shown to contain gap junctions with SMCs in mice, rats, and guinea pigs indicating their potential involvement in GI motility. Years later, they came to be known as PDGFRα positive cells/telocytes (TC) since they were positive for the platelet-derived growth factor receptor alpha (PDGFRα) and negative for c-kit thus differentiating them from ICC. These cells are located all throughout the GI tract but specifically in the muscularis propria, as they can be found encircled around muscle bundles and ganglia. Their long processes form an intricate mesh-like network interacting with ICC networks, once again highlighting their importance in motility and function—specifically, purinergic motor transmission (Figure 5) [90].

Figure 5.

Bidirectional interactions between PDGFRα-positive/TC and MMΦ comparing homeostasis and disease.

Recently, it has been shown that MMΦ establishes cell-to-cell contacts with PDGFRα-positive cells using TEM. Due to the close association of PDGFRα-positive cells with SMCs and the connections made with MMΦ, it has further bolstered the notion that MMΦ contributes to homeostasis and GI motility in some fashion that has yet to be elucidated in greater detail.

5.3 MMΦ: enteric glial cells

Enteric glial cells (EGCs) are found in the muscularis propria surrounding the ENs, sharing similar features with the brain macroglia, represented by astrocytes and oligodendrocytes. As oligodendrocytes in the brain, EGCs contribute significantly to ENs maintenance, survival, and function [91, 92, 93, 94]. As previously described, a subpopulation of MMΦ is specifically situated in this space, making the two cell types sharing the same anatomical space and a functional interaction possible.

EGCs as astrocytes in the brain can shift the phenotype in disrupted homeostasis conditions, such as inflammation. They actively mediate acute and chronic inflammation in the gut by regulating circulating monocyte recruitment during inflammation and the expression of pro-inflammatory cytokines. In vivo and in vitro studies demonstrate that EGCs secrete several cytokines and chemokines, including interferon-γ (IFN-γ), chemokine ligand 20, TNF-α, and prostaglandin D2 [95, 96].

The increased number of CD68 MMΦ during inflammation, partially due to circulating monocyte extravasation, is reduced after conditional depletion of connexin 43 (Cx43)—a gene which encodes for gap junctions—from EGCs. IL-1Β level, which is increased in inflammation, binds to its receptor on EGCs and regulates CSF1 expression by a Cx43 dependent mechanism [97].

A similar mechanism involving EGCs-MMΦ crosstalk was also observed in POI, which is associated with an increased level of IL-1β. Intraperitoneal injection of IL-1β promoted the expression of pro-inflammatory genes, and deficient mice for IL-1R1, a receptor for IL-1β, are protected from POI. Interestingly, immunohistochemistry study showed that IL-1R1 in POI co-labeled with EGC labeled with GFAP. Culture enteric glial stimulated with 10 ng/mL of IL1-β for 24 h expressed a high level of MCP-1, suggesting the possible involvement of these cells on circulating monocyte recruitment [98].

Another study showed that EGCs, after inflammation, express CCL2, which promotes monocyte recruitment upon binding with its receptor CCR2 [99]. Another study on POI described the activation of ATP, which through the P2x receptor triggered IL6 production, which is selectively blocked by the P2x2 antagonist ambroxol [100].

In colitis induced by Heligmosomoides polygyrus infection, transcriptome analysis of isolated MMΦ revealed the enrichment of IFNγ in the colon. This data is consistent with an enrichment of IFNγ from EGC from patients with undergoing inflammation. Interestingly IFNγ drives a feedback effect on EGC by eliciting chemokine interferon-γ inducible protein 10 kDa (CXCL10) and guanylate-binding protein 10 (GBP10) expression through STAT1, leading to reduced proliferation of EGC. CXCL10 and GBP10 are involved in host defense and mediate immune responses with regards to anti-bacterial immunity and cancer, respectively [101, 102]. In addition to directly impacting ECs, the IFNγ-EGC-Cxcl10 signaling axis regulates tissue repair after helminthes infection through MMΦ via CCL8, CCL7, CXCL2, and CCL2 activation [102].

α-Synuclein (α-Syn) aggregates are found in the brain of patients with Alzheimer’s disease. Most patients with this disease suffered from GI functional disorders, however, the mechanism is not understood. Application of α-synuclein (α-Syn) aggregates into the muscularis propria promotes the expansion of EGCs and overall tissue inflammation. Although it is not directly tested, it is possible that in this context, EGCs orchestrate the overall α-Syn mediated inflammation by talking with MMΦ, as multiple genes expressed following EGCs expansion are associated with MMΦ (Figure 6) [103].

Figure 6.

Bidirectional interactions between EGC and MMΦ comparing homeostasis and disease.

Pro-inflammatory MMΦ products, such as IL-1, IL-4, and TNF-α, promote EGCs activation like reactive gliosis in the CNS. It is also evident that MMΦ products can affect EGCs phenotype during inflammation. Esposito and colleagues showed that upon LPS treatment, EGCs acquire an activated phenotype that coordinates the inflammatory response in the ENS. Inhibiting the NF-κB pathway on EGCs ameliorated the overall inflammatory response in colitis and in-vitro models. On the other hand, treatment of EGCs in vitro with LPS promoted the expression of genes associated with an anti-inflammatory response [104].

As in the CNS, EGCs play a role in maintaining gut homeostasis by interacting with ENs. By interacting with EGC, MMΦ can potentially contribute to ENs maintenance. In fact, the application of TNF-α and IL-1β, 2 cytokines produced by pro-inflammatory MMΦ, can induce the expression of nerve growth factor (NGF), which is implicated in neuronal outgrowth.

6. Conclusion and future directions for MMΦ in the GI tract

MMΦ are specialized phagocytic cells that fulfill an important role in regulating GI homeostasis and disease. They contain different subpopulations whose phenotype depends on their GI tract location and origin. These unique MMΦ, compared to mucosal MΦ, share an anti-inflammatory phenotype. It is evident now that MMΦ has a dual origin. The MMΦ pool is maintained by both monocytes derived- and embryonic-derived MMΦ. Although there is evidence suggesting the involvement of embryonic-derived MMΦ in regulating ENs, no information supports the possible contribution of circulating monocytes to tissue homeostasis. Depending on their location, MMΦ can interact functionally with cells that are important for GI physiology. Further studies are needed to elucidate the underlying mechanisms regulating this type of interactions.