Neuropods.

Enteroendocrine cells (EECs) are sensory cells of the gastrointestinal tract. Most EECs reside in the mucosal lining of the stomach or intestine and sense food in the gut lumen. Food signals stimulate the release of hormones into the paracellular space where they either act locally or are taken up into the blood and circulate to distant organs. It recently was recognized that many EECs possess basal processes known as neuropods that not only contain hormones but also connect to nerves. This review describes how neuropods contribute to EEC function beyond typical hormonal actions. For example, gastrointestinal hormones not only act on distant organs, but, through neuropods, some act locally to stimulate other mucosal cells such as intestinal stem cells, enterocytes, or other EECs. With the recent discovery that EECs communicate directly with enteric nerves, EECs not only have the ability to sense food and bacteria in the gastrointestinal tract, but can communicate these signals directly to the nervous system.

Enteroendocrine cells (EECs) are sensory cells of the gut that communicate by releasing hormones locally through paracrine action or into the blood stream. Recently it has been discovered that EECs possess cytoplasmic processes known as neuropods that extend to distant cells including nerves. Thus, EECs regulate ingestive behavior through local, hormonal and neural signaling.

Traditionally, enteroendocrine cells (EECs) have been viewed as spindle or flask-shaped cells that reside in the mucosa of the gastrointestinal tract.1, 2 Most EECs are open to the intestinal lumen where a small portion of their apical surface is exposed to intestinal contents. Similar to enterocytes, microvilli cover their luminal surface. In this manner, EECs can sample food or microbes in the intestine. A small number of EECs are of the closed type, and even though their cell bodies are contained in the mucosa, they do not come into contact with the lumen of the intestine.

Hormones are stored within vesicles that are concentrated in the basal region of the EEC and are released into the paracellular space when the cell is stimulated. Secreted hormones can act locally on adjacent cells or be taken up by the blood stream where they can bind to cell surface receptors on distant organs. In this manner, EECs can sense gut contents and communicate these signals throughout the body.

Most EECs have been identified through their expression of a predominant gastrointestinal peptide and were assigned a single letter designation. For example, in the stomach, gastrin-containing cells were called G cells, while somatostatin cells, whether they are found in the stomach or pancreatic islets, have been referred to as D cells. With the exception of L cells, which produce both peptide YY (PYY) and glucagon-like peptides (GLPs), it was believed that a single EEC produced only 1 peptide hormone. Recently, however, using mice with transgenic expression of fluorescent markers, it has been shown that EECs previously thought to express only a single hormone actually produce many gastrointestinal peptides. In fact, it appears that secretin is expressed in most, if not all, EECs of the intestine. Thus, not only is the terminology for describing specific EECs incomplete, the diversity of EECs is wider than ever expected. The variety of hormones expressed in EECs has broad implications for the function of these cells that is only beginning to be explored. For the purposes of this review, we refer to EECs by the hormone(s) they produce.

Anatomic Features

Within the gastrointestinal tract, the anatomy of EECs has been guided largely through microscopic analysis of tissues stained for chemical properties such as silver (eg, argentaffin), or immunohistochemical staining using general vesicle markers (eg, chromogranin) or specific hormone antibodies. These methods have shown that EECs comprise approximately 1% of the intestinal mucosal cells, are dispersed among enterocytes, and contain secretory vesicles along their basal pole. Electron microscopy has shown that microvilli cover the luminal surface of EECs and that both electron-dense core vesicles typical of hormone-containing granules and clear vesicles resembling neurotransmitter-containing granules comprise the vesicular compartment of EECs.

Because traditional microscopy generally is performed in 2 dimensions, it has been difficult to appreciate all of the anatomic features of EECs that deviate from the traditional spindle or flask-shape view. The exception has been somatostatin-containing cells of the stomach. Somatostatin, which has broad inhibitory effects, is found in gastric cells that have long cytoplasmic processes that extend to adjacent gastrin-producing EECs and hydrochloric acid–producing parietal cells. Thus, it appeared that somatostatin could be delivered to and released from cytoplasmic processes. This discovery provided compelling evidence for paracrine control of cell function in the gastrointestinal tract.

It was assumed that paracrine effects made manifest through cytoplasmic extensions were unique to somatostatin cells and were most apparent in the stomach despite several curious observations. Serotonin-containing enterochromaffin cells, a subtype of EECs, also are believed to exert paracrine effects within the gastrointestinal tract. By using immunohistochemical staining and careful cell dissection techniques, a physical connection was observed between EECs and other cells through an axon-like process. However, the presence of such processes was not limited to somatostatin cells.

In the intestine, mucosal cells arise from crypt stem cells and migrate toward the lumen. Within 4–5 days most mucosal cells reach the villus tip, undergo apoptosis, and are sloughed into the intestinal lumen. Although it was observed that an occasional EEC of the distal small intestine possessed a lagging segment of cytoplasm along the basal surface, it was assumed this occurred as the EEC was being pushed upward by younger cells arising from the crypt. Even though this explanation seems unlikely because lagging cytoplasm has not been observed for other migrating mucosal cells such as enterocytes or goblet cells, it largely has gone unchallenged.

With the development of transgenic mice expressing green fluorescent protein (GFP) downstream from the specific gastrointestinal hormone promoters such as cholecystokinin (CCK) and PYY, EECs could be analyzed using confocal fluorescence microscopy.12, 13 With the ability to visualize cells in 3 dimensions, it was possible to trace cellular processes that extended outside a simple plane. Remarkably, a large number of EECs from the proximal small intestine of CCK-GFP mice showed basal cytoplasmic processes. These processes generally extended from the basal portion of the cell but occasionally arose from the midportion of the cell body. They were usually less than 10 μm in length and radiated in all directions from the EEC. Their multidirectionality clearly differed from the unidirectional cytoplasmic processes reported previously and were not consistent with the notion that they were caused by EECs being pushed toward the villus tip.

However, where these cytoplasmic processes went and what they did was not known. In contrast to CCK cells in the proximal small intestine, PYY cells had only 1 process per cell, although they often were much longer, with some extending over 70 μm in length. Long cytoplasmic processes tunneled under enterocytes along the lamina propria (Figure 1). How to assess the ultrastructure of EECs that contained long cytoplasmic extensions posed a formidable problem. First, it was difficult to locate a rare cell type (an EEC) among a field of enterocytes and other mucosal cells using traditional electron microscopy. Second, by its very nature, electron microscopy is ideal for analyzing a small region of a cell, but does not lend itself to tracing a long and possibly tortuous cellular process that may not reside in a single plane. To circumvent these problems, confocal microscopy was used to locate a fluorescent EEC combined with a newly developed technique of 3-dimensional scanning electron microscopy. This approach enabled imaging an entire EEC with its cytoplasmic process from the distal small intestine. Several unique features of the cytoplasmic process were revealed. First, 70% of the EEC’s secretory vesicles were contained within the process. Second, both small, clear and large, dense core vesicles were identified. Third, the process was packed with mitochondria. Fourth, running down the core of the process was a ribbon-like band absent of ribosomes that upon immunohistochemical staining was found to contain neurofilament proteins, the structural component of neuronal axons. Finally, this axon-like process came into contact with glia in the submucosa. These findings indicated that EECs possess many neuron-like properties and the cytoplasmic process became known as a neuropod (Figure 2).

Figure 1

An enteroendocrine cell neuropod. A confocal microphotograph shows an enteroendocrine cell within the mouse small intestine. An enteroendocrine cell (green) stained with a fluorescently labeled PYY antibody possesses a neuropod that extends below the basal surface of surrounding mucosal cells (eg, enterocytes). Nuclei are stained blue with 4′,6-diamidino-2-phenylindole. Scale bar: 10 μm.

Figure 2

Enteroendocrine cell communication within the gut. A model of an enteroendocrine cell (green) residing in the epithelium of the gut shows that EECs contain both large, dense (blue) and small, clear (yellow) secretory granules that are believed to contain peptide hormones (eg, CCK, PYY, and GLPs) and neurotransmitters (eg, glutamate), respectively. Neurotransmitters are released at synaptic connections with sensory neurons. Peptides such as PYY or GLP-2 are released locally and bind to proliferative cells in the crypt or intestinal subepithelial myofibroblasts, respectively. Peptides (eg, CCK, PYY, and GLPs) also may bind locally to nerves.

Paracrine Actions of EECs

Somatostatin

The prototypical paracrine cell is the somatostatin-containing D cell found in the intestine and pancreatic islets. In the gut, somatostatin cells were notable for their dendritic-like processes that extended to adjacent cells, including other EECs. The dictum that structure determines function was keenly illustrated by somatostatin cells when it was recognized that the broad inhibitory actions of somatostatin were not well suited for precise control of individual cells. In contrast, when cellular extensions were seen to innervate other cells it was easy to appreciate that somatostatin could be delivered at a specific site. Thus, the cellular processes on somatostatin cells are ideal for targeted delivery of a transmitter onto a single cell.

In the stomach, cellular processes extend from the cell body of D cells through which somatostatin exerts a tonic inhibitory action on gastrin-containing cells to block gastrin release in the antrum. From somatostatin cells in the body and fundus of the stomach, cellular processes extend to parietal cells to directly inhibit gastric acid secretion. Dual feedback loops regulate somatostatin cell function. First, somatostatin cells that are open to the gastric lumen are stimulated by gastric acid. Somatostatin released from these cells then inhibits both G cells and parietal cells to reduce gastrin and gastric acid secretion. Second, somatostatin cells possess gastrin (CCK2) receptors, activation of which also stimulates somatostatin release, resulting in reduced gastrin and acid secretion.

Glucagon-Like Peptides

GLP-1 and GLP-2 are synthesized in L cells, which are most abundant in the ileum and colon. These are the same cells that produce PYY. It is believed that stimulation of L cells causes the release of all 3 peptides. GLP-1 and GLP-2 are products of post-translational processing of the prohormone proglucagon. GLP-1 and GLP-2 are secreted in response to ingested nutrients, including glucose and fat, and their main biological actions are to enhance insulin secretion and intestinal epithelial growth, respectively. The growth-stimulating effects of GLP-2 accompany nutrient ingestion and result from both increased epithelial proliferation and decreased apoptosis. GLP-2 also enhances nutrient digestion and absorption, increases intestinal transit time, increases intestinal blood flow, improves epithelial barrier function, and enhances immune protection by increasing Paneth cell function. GLP-2 also has protective effects in animal models of intestinal inflammation.

Despite the prominent effects of GLP-2 on the intestinal epithelium, these actions appear to be indirect because GLP-2 receptors are not expressed on intestinal stem cells (either Lgr5+ crypt base or in Bmi1+ cells that comprise the proliferative reserve stem cell population). Although GLP-2 receptors are expressed almost exclusively in the gastrointestinal tract, they are found on 3 distinct cell types, but not intestinal epithelial cells. The most prominent cell type expressing GLP-2 receptors are intestinal subepithelial myofibroblasts (iSEMFs), which lie below the intestinal mucosa.21, 22, 23 GLP-2 receptors also are expressed in enteric neurons,24, 25, 26 including vagal afferent nerves and a subpopulation of EECs. It is quite remarkable that despite the lack of receptors on intestinal stem cells or cells in the transit-amplifying region of the crypt, GLP-2 increases mucosal and crypt cell proliferation, increases the length of microvilli on enterocytes, improves intestinal barrier function and decreases intestinal permeability, facilitates nutrient digestion and absorption, and enhances enterocyte survival.29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40, 41

Similar to most gastrointestinal hormones, upon release from its EEC, GLP-2 enters the blood stream or is taken up in lymphatic vessels.42, 43 However, based on the lessons learned from somatostatin cells, together with the discovery of neuropods in L cells, it is just as likely that the local effects of GLP-2 are the result of paracrine actions, but the actions are likely on a cell type expressing GLP-2 receptors that reside in the intestinal submucosa. A strong case can be made for iSEMFs mediating the growth-promoting effects of GLP-2. Not only do iSEMFs express GLP-2 receptors, but they produce the growth factors insulin-like growth factor-1 (IGF-1) and epidermal growth factor (EGF).23, 44, 45, 46 Both IGF-1 and EGF receptors are expressed in intestinal stem cells where they could affect intestinal growth. Importantly, administration of IGF-1 and EGF each have been shown to exert mucosal growth-promoting effects.23, 47, 48, 49 Interestingly, EGF, but not IGF-1, restores intestinal growth in GLP-2–receptor knockout mice.

These findings show that GLP-2 has prominent local, growth-promoting effects on the intestinal epithelium. GLP-2 receptors are expressed on cells that lie in close proximity to the GLP-producing EECs, and the target cells (such as iSEMFs) are capable of mediating the growth-promoting effects of GLP-2. Thus, it appears that L cells release GLP-2, which stimulates iSEMFs to produce IGF-1 and EGF, which exert their own paracrine effects on the intestinal mucosa.

Even though GLP-2 receptors have been identified on a subpopulation of EECs, there are insufficient data to invoke a role for another EEC type to mediate the growth-promoting effects of GLP-2 on the intestine. Nevertheless, it is interesting that in Drosophila, EECs provide local cues for maintaining intestinal stem cells. Therefore, it is possible that EECs exert as yet unrecognized direct effects on intestinal stem cells.

GLP-1 receptors are expressed on β cells of the pancreatic islet and GLP-1 increases insulin synthesis and secretion.53, 54, 55 Even though the effects of GLP-1 on the pancreas are well established and therapeutic GLP-1 mimetics are used to treat patients with diabetes mellitus, there is evidence that endogenously released GLP-1 may influence insulin secretion through a paracrine action by acting locally on afferent nerves.56, 57 Local paracrine actions of GLP-1 also may influence immune function through effects on intraepithelial lymphocytes and afferent neurons. Thus, despite our historical focus on the actions of circulating glucagon-like peptides and their exploitation for therapeutic purposes, locally released GLP-1 and GLP-2 may exert important and previously unappreciated physiological effects.

Peptide YY

A notable feature of some L cells is their long neuropods, which are packed with both large, dense and small, clear secretory vesicles that are believed to contain hormones and neurotransmitters, respectively. One function of neuropods is neurotransmission, with glutamate as a candidate neurotransmitter. However, that does not explain the functional significance of the abundant peptide-containing vesicles that reside in neuropods.

Although these, too, could function as neuropeptides, it is striking that EECs are found as single cells in the mucosa, surrounded by enterocytes. In this orientation, only a few enterocytes actually come into contact with an EEC. Neuropods lie between the base of enterocytes and the lamina propria, but their long extension allows EECs to reach many more cells in the mucosa than those that lie next to the EEC cell body. Similar to the prototypical paracrine cell—the somatostatin cell with its cellular extensions—the neuropods of L cells provide a convenient mechanism for delivering peptide to enterocytes throughout the mucosa.

This possibility took on more meaning when it was discovered that PYY induces proliferation of intestinal cells in vivo and in vitro and activates mitogenic signaling pathways, and gut epithelial cells express Y1-type PYY receptors. Y1-receptor messenger RNA is expressed in basal crypt cells consistent with the transit-amplifying cell region of the intestinal mucosa. In this location, the receptor is in a position to mediate the growth-promoting effects of PYY. Expression of the Y1 receptor protein has been found throughout the colonic crypt epithelium, suggesting that the receptor protein is long lived even though the proliferative effects of PYY would be realized only in the intestinal stem cell or transit-amplifying cell region. Thus, it appears that PYY, unlike GLP-2, exerts direct effects on intestinal epithelial cells to stimulate growth. In addition, confocal microscopy of fluorescently labeled Y1 receptor localized the receptor to the basolateral surface of human colonocytes. In this location, cells are ideally positioned to respond to locally released PYY.

Serotonin

One of the earliest recognized EEC signaling actions onto nerves in the gut was the regulation of peristalsis initiated by the local release of serotonin from enterochromaffin cells (a major EEC subtype).63, 64 Locally released serotonin also stimulates nerves in the gut to transmit signals to the brain and mediates chemotherapy-induced nausea. This pathway is the basis for the use of 5-hydroxytryptamine-3 antagonists in the treatment of nausea. Although much of the serotonin effect was believed to be paracrine, a direct functional connection between enterochromaffin cells and gut nerves recently was shown. Upon activation, enterochromaffin cells were shown to transmit an electrical signal onto serotonin-sensitive primary sensory afferent nerve fibers via synaptic connections, enabling them to transduce information from the gut lumen directly to the nervous system. Thus, what once was believed to be exclusively a paracrine action between serotonin-producing cells and cells within the gut has taken on a new level of communication because the local actions of serotonin may be either paracrine or neural.

This pathway has implications for gut–brain communication and diseases in which the brain perceives sensory signals from the gut. Abnormal serotonin signaling has been implicated in intestinal hypersensitivity and irritable bowel syndrome. Tissue concentrations of serotonin appear to be more abundant in patients with irritable bowel syndrome and 5-hydroxytryptamine-3 antagonists have been used to ameliorate symptoms in patients with irritable bowel syndrome.

Neuron-Like Features

EECs are terminally differentiated and, after in vitro isolation, single cells do not grow in culture. However, if maintained in intestinal organoids, EECs arise from crypt stem cells and in organoids prepared from the intestine of CCK-GFP or PYY-GFP mice, fluorescent EECs are not only present, but they flourish. They even express neuropods. EECs were found to express neurotrophin receptors including TrkA and glial-derived neurotrophic factor family receptor α-3 and the neurotrophins, nerve growth factor-β, and artemin were discovered to induce neuropod elongation and branching, respectively.

The neural network within the intestinal villus is extensive and neuropods come into close contact with neurons lying beneath the intestinal mucosa. It seemed natural to ask if EECs connect with neurons. This possibility was investigated using several complementary approaches. First, using a variety of neuronal cell markers, immunohistochemical staining showed that EECs contacted nerves expressing neurofilament proteins, the pan-neuronal marker protein gene product 9.5, calbindin, and calcitonin gene-related peptide, a marker of sensory neurons. Thus, it appeared that EECs were capable of connecting to sensory neurons. EECs were not seen connecting to vasoactive intestinal peptide-containing nerves, which has been used as a marker of motor neurons.

Next, the connection of EECs to sensory neurons was recapitulated in vitro. Despite EECs not growing in culture, a small number survive cell isolation and can be maintained in vitro, making it possible to co-culture EECs from CCK-GFP mice together with sensory neurons from trigeminal or dorsal root ganglia. By using time-lapse photomicroscopy it was observed that fluorescent EECs interacted with sensory neurons labeled with the lipophilic dye, DiI. With an EEC and sensory nerve lying in close proximity, the nerve extended a small neurite toward the EEC. The EEC responded by elongating a process to contact the nerve’s neurite. The EEC then withdrew from the nerve but the cells remained connected through an axon-like structure. These studies show that EECs and neurons have an affinity for one another that likely results from the release of chemical signals. We propose that chemical attractants are released from EECs and neurons and form chemical gradients that induce directional extensions of cellular processes (either neuropods or neurites) to produce axon-like connections.

EECs express a number of genes that encode neuronal proteins, including those for presynaptic and postsynaptic proteins. For example, genes for the presynaptic proteins synapsin 1, piccolo, bassoon, MUNC13B, RIMS2, latrophilin 1, and transsynaptic neurexin 2 have been identified in EECs. In addition, these cells express postsynaptic genes for neuroligins 2 and 3, homer 3, and postsynaptic density 95. Thus, the connection between EECs and neurons appeared to be a synapse.

A synaptic connection between EECs and neurons was proven using monosynaptic rabies virus neurotracing. By using a modified rabies virus in which the envelope glycoprotein rabG was replaced with green fluorescent protein (rabG-GFP), it was possible to show that rabies virus placed into the lumen of mouse colon infected PYY-containing EECs. Because rabG is responsible for the spread of rabies from neuron to neuron through synapses, the modified rabies virus could not move beyond the EEC. In separate experiments, using a mouse with targeted expression of the envelope protein rabG exclusively in PYY cells, instillation of rabG-GFP into the colon infected EECs, and because the virus could be packaged with the viral coat expressed within the PYY cells, the virus jumped onto underlying nerves, indicating that EECs connected to nerves through a synaptic connection. These studies established that a neural circuit exists connecting EECs to the nervous system.

This new neural circuit has been shown to connect EECs directly to the brainstem. By using modified rabies tracing, some EECs from both the small intestine and colon connected to nerves in the vagal nodose ganglia and projected into the nucleus tractus solitarius of the brainstem, indicating that the EECs communicate with vagal neurons and connect the gut to the brain through a single synapse. Interestingly, labeled nerve fibers also were seen in the dorsal root ganglia, indicating that EECs connect with nerves innervating the spinal cord.

EECs have been characterized by the hormones they express and because individual EECs were thought to produce only a single hormone, the possibility of an EEC producing other transmitters, not to mention neurotransmitters, had been largely overlooked. However, when it was appreciated that EECs synapse with neurons, it was natural to ask why and what transmitters were involved. RNA analysis of GFP-expressing EECs provided a hint when it was discovered that EECs express the vesicular glutamate transporter 1 protein. Subsequently, glutamate was shown to be released from EECs in vitro and to transmit a synaptic signal in EECs co-cultured with sensory neurons. Thus, it appears that the synaptic signal emanating from EECs involves the classic neurotransmitter glutamate.

Electrical recordings from synaptically connected EECs and vagal nodose neurons in a co-culture system was used to dissect the synaptic transmission properties of the EEC-neuron connection. By using selective CCK and glutamate-receptor blockers, it was shown that CCK-GFP EECs release both CCK and glutamate, which stimulate vagal neurons in a time-dependent manner. Glutamate activation of nodose neurons occurred within milliseconds of EEC stimulation, indicating synaptic transmission. In contrast, the CCK-receptor blocker devazepide inhibited a slower phase of nerve activation consistent with CCK having a paracrine action. Thus, EECs use both CCK and glutamate as transmitters to activate vagal sensory nerves through both paracrine and synaptic pathways.

Gut signaling through the vagus nerve generally has been thought to mediate negative feedback mechanisms to control meal size. Recently, however, using selective optogenetic techniques to stimulate select fibers in the vagus, it was shown that vagal sensory pathways from the gut project to reward regions in the brain including the substantia nigra. Thus, not only do vagal signals from the gut control satiety, but select neurons can affect motivation and pleasure.

The function of the EEC-neural circuit is only beginning to be appreciated, however, it is evident that nutrient signals such as glucose can activate a signal within an EEC that is transmitted onto sensory neurons that extend to the brain. Therefore, it is highly likely that other signals from ingested nutrients or microbiota in the gut may be transmitted to the brain in the same way. This circuit establishes a direct sensory neural link between the gut and the brain.

EEC Life Span

In general, it had been believed that EECs, similar to other epithelial cells of the intestinal mucosa, turn over every 4–5 days. If correct, the dynamics of EEC migration from the crypt to the villus tip and the rapid turnover of cells may restrict the ability of EECs to connect with nerves and limit the duration of the connection. However, some studies have suggested that EECs live longer than enterocytes. By using bromodeoxyuridine labeling of cells, it was observed that some EECs reside in the intestinal mucosa for at least 60 days, and likely longer. These findings indicate that EECs have a life span similar to that of other sensory cells such as taste cells and olfactory neurons.78, 79

Final Note on EEC Sensing

Accumulating evidence has indicated that EECs respond not only to food in the lumen of the intestine, but also are stimulated by absorbed nutrients. Because neuropods extend throughout the gut mucosa, they are ideally positioned to receive signals from enterocytes and other neighboring cells. These signals may be in the form of absorbed nutrients, such as lipids that stimulate CCK secretion. Although an attractive hypothesis, it remains to be determined if neuropods receive and transmit signals from the paracellular space to the EEC.

It recently became apparent that EECs respond to a variety of chemical and mechanical stimuli and transmit sensory signals from the gut to the brain to inform us of what we eat and warn us of ingested dangers. We interpret those signals by modifying what, when, and how we eat.

In addition, EECs act locally in the gut to regulate intestinal secretion, motility, and growth to facilitate the ingestion, digestion, and absorption of nutrients. However, it is not yet understood how EECs know to do all of these things. Although we understand a little bit about how EECs send signals from the gut, we know EECs also express postsynaptic proteins, implying that they receive efferent signals from the nervous system. On the frontier is understanding how the EEC learns from the brain.