Neurotransmitters responsible for purinergic motor neurotransmission and regulation of GI motility

Abstract

Classical concepts of peripheral neurotransmission were insufficient to explain enteric inhibitory neurotransmission. Geoffrey Burnstock and colleagues developed the idea that ATP or a related purine satisfies the criteria for a neurotransmitter and serves as an enteric inhibitory neurotransmitter in GI muscles. Cloning of purinergic receptors and development of specific drugs and transgenic mice have shown that enteric inhibitory responses depend upon P2Y₁ receptors in post-junctional cells. The post-junctional cells that transduce purinergic neurotransmitters in the GI tract are PDGFRα⁺ cells and not smooth muscle cells (SMCs). PDGFRα⁺ cells express P2Y₁ receptors, are activated by enteric inhibitory nerve stimulation and generate Ca²⁺ oscillations, express small-conductance Ca²⁺-activated K⁺ channels (SK3), and generate outward currents when exposed to P2Y₁ agonists. These properties are consistent with post-junctional purinergic responses, and similar responses and effectors are not functional in SMCs. Refinements in methodologies to measure purines in tissue superfusates, such as high-performance liquid chromatography (HPLC) coupled with etheno-derivatization of purines and fluorescence detection, revealed that multiple purines are released during stimulation of intrinsic nerves. β-NAD⁺ and other purines, better satisfy criteria for the purinergic neurotransmitter than ATP. HPLC has also allowed better detection of purine metabolites, and coupled with isolation of specific types of post-junctional cells, has provided new concepts about deactivation of purine neurotransmitters. In spite of steady progress, many unknowns about purinergic neurotransmission remain and require additional investigation to understand this important regulatory mechanism in GI motility.

1. Introduction

The autonomic nervous system is composed of 3 divisions, the sympathetic, parasympathetic and enteric nervous systems, with signaling to post-junctional cells mediated by a variety of neurotransmitters. Work begun by Professor Geoffrey Burnstock and his colleagues at the University of Melbourne in the early 1960s revealed neurons that are neither adrenergic nor cholinergic in the autonomic nervous system. Initially the challenge facing this concept was to demonstrate unequivocally that inhibitory neural responses in GI muscles were not due to sympathetic neurons and the excitatory innervation of the bladder was due to more than just parasympathetic, cholinergic neurons. Examples of neurotransmission were observed in visceral smooth muscles that were blocked by tetrodotoxin (TTX) but not by adrenergic or cholinergic receptor antagonists (Burnstock et al., 1964). The search for non-adrenergic, non-cholinergic (NANC) neurotransmitters spanned the period from the early 1960s to 1990 during which time a variety of novel neurotransmitter substances, including purines, peptides and nitric oxide (NO), were shown to serve as inhibitory or excitatory neurotransmitters in various tissues and organs. The roles of various peptides and NO in enteric neurotransmission have been reviewed elsewhere (Shuttleworth and Keef, 1995; Sanders and Ward, 2019), and this review will concentrate on purine substances that have been linked to enteric inhibitory neurotransmission.

Burnstock will be remembered as the scientist who advanced the idea that purines serve as neurotransmitters in visceral smooth muscles (Burnstock et al., 1970; Burnstock, 1972; Burnstock et al., 1972). NANC nerve responses were compared to the effects of exogenous ATP in sixteen gastrointestinal (GI) muscles from 4 vertebrate animals and one bladder preparation. ATP mimicked the responses of NANC nerve stimulation, inhibitory in GI muscles and excitatory in the bladder, and it was concluded that “…it would be surprising if this close correlation observed in the seventeen preparations examined was only a coincidence.” ATP and other purines were released from GI muscles and enteric nerve preparations. However, more work was needed to satisfy the criteria set out by prominent neuroscientists in the 1950s–1960s for establishing a substance as a neurotransmitter (Paton, 1958; Curtis, 1961; McLennan, 1963; Eccles, 1964). At the time, measurements of purines resulting from nerve stimulation were crude, and specific receptors and their affinities for purine compounds were unknown. With time, receptors for ATP and ADP (P2X and P2Y receptors) were distinguished from adenosine (P1) receptors (Burnstock and Kennedy, 1985; Kennedy and Burnstock, 1985) and pathways for extracellular biotransformation of purines were discovered (Zimmermann et al., 1998). Measurements of purines released from neurons have improved significantly by utilization of high-performance liquid chromatography (HPLC) and etheno-derivatization of purines in tissue superfusates that facilitates fluorescence detection (Davis and White, 1980; Levitt et al., 1984; McConalogue et al., 1996; Bobalova et al., 2002; Smyth et al., 2004). This review will describe the experimental observations that advanced the concept of ‘purinergic’ neurons in the GI tract, techniques used to identify purines released from neurons and more recent findings that have investigated the sources, identity of purine neurotransmitters, post-junctional cells and receptors that mediate purinergic responses and metabolic pathways involved in deactivating purine neurotransmitters.

2. Neurons that release purines in the GI tract

Stimulation of intrinsic neurons in the guinea pig taenia coli by electrical field stimulation (EFS) resulted in hyperpolarization of membrane potentials in smooth muscle cells (SMCs) and inhibition of contractions (Burnstock et al., 1963). The inhibitory neurons in the GI tract were found to be distinct from sympathetic neurons (Burnstock et al., 1964; Burnstock et al., 1966). The hyperpolarization responses were linked to the activation of neurons because they were blocked by TTX that has little or no effect on SMCs (Tomita, 1972). Single inhibitory junction potentials (IJPs) were activated with a latency of 150–300 ms by electrical pulses too short to activate SMCs directly. IJPs disrupted the normal rhythms of spontaneous activity and caused transient relaxation. Multiple pulses of electrical stimulation caused summation of IJPs, and responses up to about 15 pulses per sec were not affected by guanethidine, an inhibitor of norepinephrine (NE) release. In the presence of atropine, to block intrinsic excitatory inputs, EFS of taenia coli with single 200 μsec pulses caused rapidly developing IJPs in SMCs impaled with microelectrodes that could reach 25 mV in amplitude (Fig. 1) (Bennett et al., 1966). Inhibitors of sympathetic neurotransmission, such as guanethidine or bretylium, did not inhibit the IJPs. Other early experiments identified NANC IJPs in preparations of jejunum (Hidaka and Kuriyama, 1969), colon (Furness, 1969), and stomach (Beani et al., 1971).

Fig. 1.

Effects of increasing strength of EFS of taenia coli on the amplitude of inhibitory junction potentials (IJPs). The pulse durations were 200 μsec and stimulus strengths were 37% (A), 47% (B), 73% (C) and 100% (D) of the pulse strength that produced the maximum amplitude IJP (D). Each stimulus caused hyperpolarization below the resting potential (dotted lines) and a post-stimulus rebound response (arrows). Figure is redrawn from Bennett et al., 1966 with permission.

Significant numbers of enteric neurons with cell bodies in the myenteric plexus or in Henle's plexus in the colon (Gunn, 1968; Sanders and Smith, 1986) are enteric inhibitory motor neurons that innervate the muscle layers (Furness, 2012). While excitatory neurons were recognized as cholinergic very early, it was many more years before the nature of enteric inhibitory neurons and the NANC neurotransmitters were discovered. Inhibitory motor neurons are important for regulation of motility in the GI tract. For example, the lower esophageal sphincter (LES) is contracted tonically and must relax transiently to allow the passage of swallowed food into the stomach. An early study to understand the neurotransmitter responsible for neurally-mediated relaxation of the LES reported that immunoantagonism of vasoactive intestinal polypeptide (VIP) blocked inhibitory neuromuscular transmission in the LES (Goyal et al., 1980). Thus, VIP was proposed to be a primary enteric inhibitory neurotransmitter in this tissue (Behar et al., 1989). How exactly these experiments were successful and indicated such an important role for VIP is mysterious, since it was later shown convincingly that NO is the predominant enteric inhibitory neurotransmitter in the LES of several species including humans (Murray et al., 1991; Tottrup et al., 1991; Oliveira et al., 1992; Tøttrup et al., 1993; Boeckxstaens, 2005; Farré and Sifrim, 2008). Enteric inhibitory neurons are common and important throughout the GI tract, and as in the LES, NO has important effects in most regions and in most species tested to date (Bult et al., 1990; Sanders and Ward, 2019). However, there is another class of neurotransmitters that contribute to enteric inhibitory regulation.

The neurotransmitter responsible for fast IJPs in taenia coli (Bennett et al., 1966) was unknown when the inhibitory responses were first recorded (Fig. 1). IJPs in taenia coli were correctly surmised to be due to a transient increase in K⁺ conductance because IJP amplitude depended upon the external K⁺ concentration ([K⁺]_o). The equilibrium potential for K⁺ ions, estimated to be −89 mV (Casteels, 1969) and thus lying quite negative to the resting potentials of taenia coli SMCs, made it likely that the hyperpolarization during IJPs was due to an increase in K⁺ conductance. Transient activation of a K⁺ conductance by the NANC inhibitory neurotransmitter was later confirmed more explicitly by demonstrating that a linear relationship existed between membrane potential and the amplitude of the IJPs (Tomita, 1972). Strong evidence for the nature of the K⁺ conductance activated during IJPs was provided by showing that apamin blocked the IJPs (Banks et al., 1979; Maas and Den Hertog, 1979). The sensitivity of IJPs to apamin was due to its effects on small-conductance Ca²⁺ activated K⁺ (SK) channels (Köhler et al., 1996). An interesting aspect of this conductance is that its open probability is not tied to voltage, but to sub-micromolar concentrations of Ca²⁺ applied to the cytoplasmic aspect of the channels (1/2 maximal activation at 0.3 μM). Thus, understanding the mechanism of NANC IJPs also required determining the identity of the neurotransmitter and how it couples to transient rises in intracellular Ca²⁺ ([Ca²⁺]_i), which at the time seemed counterintuitive for an inhibitory response in SMCs.

3. ATP is suggested as a neurotransmitter in the gut

Part of the initial evidence that purinergic neurons exist in GI muscles was that similarities were observed in post-junctional responses to NANC nerve stimulation and responses to exogenous purines (Burnstock et al., 1970). Release of ATP during stimulation of enteric neurons was suggested by experiments in which tritiated ([³H]) adenosine was taken up and [³H] was released by EFS of guinea pig taenia coli (Su et al., 1971). These experiments, however did not prove unequivocally that [³H]-ATP was the substance released or verify that enteric neurons were the cellular source of [³H] released. Several years later, release of ATP in response to “purinergic” nerve stimulation of the guinea-pig taenia coli was confirmed by the luciferin-luciferase assay (Burnstock et al., 1978). The release of ATP was blocked by TTX, but remained intact when 6-hydroxydopamine was employed to destroy adrenergic nerves in these preparations. It was concluded that ATP was released from a specialized population of NANC nerves that became known as purinergic nerves. In another study using the same detection method (i.e., luciferin-luciferase assay) and the same test system (i.e., the guinea-pig taenia coli) measurable release of ATP was achieved only at stimulation parameters that were not affected by TTX (White et al., 1981). The conclusions of this study were that release of ATP induced by EFS was not mediated by propagation of action potentials in neurons. Issues with the luciferin-luciferase detection method that might explain the conflicting results are discussed in Methods of measuring purines from neurons.

Definitive labeling of purinergic neurons has been elusive. One technique used to identify purinergic neurons in GI muscles was based on the binding of quinacrine to ATP (Crowe and Burnstock, 1981). Quinacrine fluorescence was viewed as a label for NANC neurons in mouse, rat and guinea-pig myenteric plexus (Olson et al., 1976; Alund, 1978). Quinacrine labeling of enteric neurons decreased upon depolarization with elevated extracellular K⁺, and this was taken as evidence that quinacrine binds to a neurotransmitter substance (Alund and Olson, 1979). Because quinacrine binds to ATP and other nucleotides (Irvin and Irvin, 1954) and quinacrine labeled neurons in tissues with NANC innervation and purinergic nerve responses (Crowe and Burnstock, 1981), it was suggested that quinacrine labels purinergic neurons. However, labeling of myenteric neurons in rabbit, rat and guinea pig small intestine and stomach was variable with some ganglia showing a high percentage of labeled neurons and others showing very few or no positive neurons (Crowe and Burnstock, 1981). Quinacrine-positive varicose nerve fibers were also observed in the circular muscle layer. However, it is possible that quinacrine labeling does not distinguish neurons that utilize ATP (or other nucleotides) as neurotransmitters. All neurons (and cells) generate ATP as a primary source of energy and all secretory vesicles contain ATP (Borges, 2013). For example, measurements using expressible luciferase have estimated that 1–2 mM ATP is present in hypothalamic neurons and glia (Ainscow et al., 2002). ATP levels could be higher in neurons utilizing this compound as a neurotransmitter, and labeling might be concentrated in vesicles containing ATP. Quinacrine was found in varicose processes and cell bodies labeling of enteric neurons (Crowe and Burnstock, 1981). However, labeling was diffuse, apparently cytoplasmic, and not restricted to secretory vesicles. That some neurons are labeled, and others are not, suggests that uptake of quinacrine or access of the compound to all ganglia within a muscle preparation affects labeling. Finally, quinacrine labeling may not be specific for ATP since it binds other adenine nucleotides and guanosine monophosphate (Irvin and Irvin, 1954), plasma proteins (Irvin and Irvin, 1952), DNA (Sumner, 1986), acetylcholine receptors (Yu et al., 2003), and prion proteins (Zawada et al., 2013).

Release of enteric neurotransmitters is Ca²⁺ dependent, and thus it is thought that ATP or a related purine is contained in vesicles and released by exocytosis (Rizo, 2018; Vessey et al., 2020). An exciting possibility for definitive labeling of purinergic neurons and settling questions about whether NO/purines/VIP are co-transmitters in the same populations of motor neurons was suggested by the cloning of the vesicular nucleotide transporter (VNUT) that is believed to be responsible for the vesicular storage of ATP and encoded by SLC17A9 (Sawada et al., 2008). One study reported VNUT labeling of neural varicosities in the circular muscle of murine stomach and co-localization of neuronal nitric oxide synthase (nNOS) in VNUT⁺ varicosities (Chaudhury et al., 2012). However, both nNOS and vesicular acetylcholine (ACh) transporter positive neurons, as well as other neural phenotypes were VNUT⁺, suggesting that the VNUT antibodies used were not specific for purinergic neurons, and VNUT immunoreactivity may not be a specific means of detecting purinergic neurons. This may also be true because ATP is present in all secretory vesicles and may serve non-neurotransmitter roles (discussed in Neurotransmitter identification). If this is the case, however, one would expect more abundant labeling of vesicles in all enteric and autonomic neurons in the gut wall than was observed in murine stomach (Chaudhury et al., 2012). Our own experience with VNUT antibodies labeling of enteric motor neurons has not been successful, with either the original antibody provided to us by Yoshi Moriyama (Sawada et al., 2008) or with antibodies generated commercially from likely-antigenic peptide sequences (Sanders KM, unpublished observations). It should also be noted that RNAseq of cells isolated from murine colon showed very low expression of Slc17a9 (Ha et al., 2017; Lee et al., 2017). These data are interesting because other experiments have suggested that β-NAD⁺ is a more likely neurotransmitter in enteric purinergic neurons than ATP (Mutafova-Yambolieva et al., 2007; Hwang et al., 2011), and β-NAD⁺ is a poor substrate for VNUT (Moriyama et al., 2017). Thus, it seems likely that another, as yet unidentified, vesicular transport protein is required for uptake and storage of β-NAD⁺. Such a molecule could be an excellent biomarker for purinergic neurons in the gut. SLC25A51 was recently identified as the first mammalian mitochondrial NAD⁺ transporter (Luongo et al., 2020; Kory et al., 2020; Girardi et al., 2020). It remains to be determined if this protein also transports NAD⁺ in other cellular organelles (e.g., synaptic vesicles). In fact, the likelihood of another vesicular purine transporter is hinted at by the puzzling observation that mice with global deletion of the Slc17a9 gene (VNUT^−/− mice) are viable and appeared to be normal “in terms of weight gain, body size, morphology, food intake, water intake, oxygen consumption, carbon dioxide emission, respiratory exchange ratio (RER), locomotor activity, and open field and maze behaviors” (Sakamoto et al., 2014). It seems remarkable that genetic deactivation of a protein so widely distributed throughout the body and essential for vesicular storage of ATP (Moriyama et al., 2017) does not produce serious functional defects.

4. Concept of co-transmission in motor neurons

It was thought classically that each type of neuron makes and releases a single type of neurotransmitter substance (known as Dale's Principle) (Eccles, 1976). However, purines are co-stored with classical neurotransmitters in autonomic neurons (Helle et al., 1971), and the realization that purines have post-junctional effects consistent with neurotransmission suggested that more than one neurotransmitter substance could be released from the same neuron (Burnstock, 1976). While this idea was speculative when first enunciated, development of antibodies specific for neurotransmitters or synthetic enzymes revealed that multiple transmitters co-localize in the same neurons. Functional support for this hypothesis was obtained first for the autonomic innervation of the guinea-pig vas deferens. Adrenergic neurons in this tissue elicited complex excitatory junction potentials (EJPs) and two-phase contractions. The first phase of the contractile response was blocked by a photoaffinity label arylazido aminoproprionyl adenosine triphosphate (ANAPP₃), a P2 receptor antagonist, and the second phase was blocked by prazosin, an α₁ adrenoceptor antagonist (Sneddon et al., 1982; Sneddon and Westfall, 1984). These data suggested that the first phase of the complex action potential was mediated by ATP acting via P2 receptors and the second phase was mediated by NE, acting via α₁ receptors. These observations were confirmed by experiments on rat tail arteries. Electrophysiological responses to nerve stimulation were biphasic and consisted of a fast transient depolarization followed by a much slower and prolonged depolarization (Sneddon and Burnstock, 1984). α,β-methylene adenosine 5′ triphosphate, used to desensitize P2 receptors, blocked the initial fast depolarization but did not affect the slow phase. The slow phase was blocked by the α-adrenoceptor antagonist phentolamine, and combination of these drugs blocked the entire response. Guanethidine blocked both the fast and slow depolarizations suggesting that ATP and NE are released from sympathetic neurons.

Co-transmission also occurs in enteric inhibitory neurons. A population of enteric motor neurons expresses both NOS and VIP (Costa et al., 1992; Keef et al., 1994; Sang and Young, 1996; Guo et al., 1997). While there is no definitive label for purinergic neurons (see discussion above), it is likely that NOS⁺/VIP⁺ neurons represent enteric inhibitory neurons, and these neurons may co-release purine neurotransmitters. Complex IJPs are recorded from many GI muscles that consist of a fast transient phase followed by a more slowly-developing, longer-duration hyperpolarization (Shuttleworth et al., 1997; Gallego et al., 2008a; Mañé et al., 2014). The fast phase of the IJP is purinergic and blocked by P2Y₁ receptor antagonists. The slow phase is nitrergic and blocked by inhibitors of NOS. An interesting observation regarding inhibitory cotransmission is that there are inverse gradients for purinergic and nitrergic regulation along the length of the colon (Mane et al., 2016). Purinergic neural regulation is dominant in the distal colon, and nitrergic neural regulation is dominant in the proximal and mid colon.

5. Neurotransmitter identification (advantages and disadvantages of common experimental approaches)

Ascertaining the identity of a neurotransmitter requires verifying (1) its presence in the presynaptic neurons, (2) its release in response to neural activation, (3) its postsynaptic receptors, and (4) its elimination from the neuroeffector junction (NEJ) (e.g., Eccles, 1964; Burnstock, 2012; Purves et al., 2012). The challenge is to prove that the candidate neurotransmitter meets all four criteria or at the very least a combination of presynaptic and postsynaptic requirements.

Criterion 1 is commonly validated by demonstrating that either the substance or precursors and enzymes for its biosynthesis are present in presynaptic neurons. Finding a substance in neurons, however, is not sufficient evidence to establish it as neurotransmitter. For instance, intracellular ATP, β-NAD⁺ or ADP-ribose (ADPR) are involved in functions of the mitochondria and other organelles, and in regulation of membrane ion channels. A superior proof is finding the substance in synaptic vesicles. In a study of NGF-differentiated PC12 pheochromocytoma cells, a cell line that has an embryonic origin from the neural crest, we demonstrated that β-NAD⁺ and ADPR are present in both small synaptic-like vesicles and large dense-core vesicles (Yamboliev et al., 2009). ATP, ADP, AMP, UTP, and diadenosine polyphosphates (e.g., Ap4A and Ap5A) are also stored in synaptic vesicles and likely function as extracellular signaling molecules (Zimmermann, 1994). It is estimated that the concentrations of purines in vesicles is in the millimolar range, which is sufficient to serve as a store for their release (Sperlagh and Vizi, 1996). ATP, in particular, appears to be present in every synaptic and secretory vesicle (Sperlagh and Vizi, 1996; Abbracchio et al., 2009; Borges, 2013) and it is sometimes used as a universal tracer of cellular secretion events that are not limited to neurotransmitter release (Aspinwall and Yeung, 2005). It is suggested that ATP in vesicles might be important for the vesicular proton gradient driving neurotransmitter uptake into vesicles, for preserving principal neurotransmitters by acidification of the vesicular milieu, for formation of osmotically inactive complexes with primary transmitters, for maintaining colligative and osmotic balance of vesicles that contain peptides, and for steps of vesicle exocytosis itself (Johnson Jr, 1987; Johnson Jr, 1988; Sperlagh and Vizi, 1996; Estevez-Herrera et al., 2016).

Criterion 2 is met by demonstrating that the substance is released from neurons upon action potential firings, and that its release is blocked by neurotoxins. This is relatively easy to achieve in simple systems such as cultured cells or isolated brain synaptosomes. We demonstrated that β-NAD⁺ and ADPR are released along with dopamine and ATP in response to high-K⁺ depolarization of PC12 pheochromocytoma cells. Cleavage of the SNARE protein SNAP-25 with Clostridium botulinum neurotoxin A abolished the depolarization-evoked release of β-NAD⁺, ADPR and dopamine, but not of ATP, suggesting that β-NAD⁺ and ATP are released by different mechanisms in these cells (Yamboliev et al., 2009). We also found that high-K⁺ evokes release of β-NAD⁺ from rat brain synaptosomes, a subcellular fraction containing synaptic terminals of neurons (Durnin et al., 2012). The release of β-NAD⁺ was attenuated by ω-conotoxin GVIA, bafilomycin, and botulinum neurotoxin A that inhibit the Ca_V2.2 neural voltage gated Ca²⁺ channels, vesicular neurotransmitter uptake, and SNAP-25-dependent vesicular exocytosis, respectively. ATP was released from rat (White, 1978) and mouse (Fiedler et al., 1992) brain synaptosomes following depolarization by high-K⁺ solution and influx of Ca²⁺. Likewise, ATP was released from both crude and purified varicosities isolated from myenteric plexus of the guinea-pig ileum in response to elevated extracellular K⁺ and Rb⁺ and to veratridine in a Ca²⁺ dependent manner, but whether ATP was released from synaptic vesicles was not determined.

Definitive measurement of neurotransmitter release in complex tissues consisting of multiple types of cells is more challenging, as detection methods with high sensitivity and superior spatial and temporal resolution are needed. Instead of direct measurements of neurotransmitter substances, most studies with this objective have characterized post-junctional responses to nerve stimulation such as miniature synaptic currents, changes in membrane potentials of post-junctional cells or changes in contractile tone (Edwards et al., 1992; Evans and Surprenant, 1992; Ramme et al., 1987). It should be realized that studies utilizing this approach characterize highly integrated pre- and post-junctional responses to nerve stimulation that depend upon neurotransmitter release, kinetics of metabolism in interstitial spaces between nerve terminals and post-junctional cells, pre-junctional neuromodulation, the expression and status of post-junctional receptors, other post-junctional factors regulating the excitability of the post-junctional cells, and relative availability of signal transduction pathways and cellular effectors. Studies of IJPs elicited by EFS in GI smooth muscles have been instrumental in characterizing the purinergic receptor(s) and intracellular pathways that mediate the effects of the inhibitory neurotransmitter (Gallego et al., 2006; Gil et al., 2010). However, studies of this type do not reveal the identity of the neurotransmitter(s) released and require additional assessments of the compounds released and compounds formed via metabolism during neurotransmission.

Criterion 3 is validated by showing that: i) Application of exogenous candidate transmitters mimic postsynaptic effects of nerve stimulation. It is difficult to compare responses to neurotransmitters released from nerve terminals and responses to exogenous candidate transmitters in a tissue as complex as GI muscles because exogenous compounds may have a much broader exposure to receptors on other cells types, some electrically coupled to SMCs, than neurotransmitters released from neurons. Metabolism, as discussed below, may also generate additional bioactive compounds. ii) Agonists and antagonists that alter the postsynaptic response to neurotransmission have the same effect when the exogenous candidate transmitter is applied. iii) Response to nerve stimulation and application of the candidate neurotransmitter are both eliminated after deletion of the gene encoding the receptor that mediates post-junctional effects. These experiments require knowing the identity of receptor(s) responsible post-junctional effects, availability of highly specific and selective ligands (i.e., agonists and antagonists) and laboratory animals with specific gene deletions. The last three decades have been particularly fruitful for the field of extracellular purinergic signaling, and for the cloning and characterization of seven ligand-gated ion channel P2X receptors, eight G-protein coupled P2Y receptors, and four G-protein coupled adenosine receptors. In early studies, pharmacological characterization of P2 purinergic receptors was based on establishing the relative potency order of exogenous purines, which frequently produces variable results. In later studies, P2 purinergic receptor antagonists such as suramin (Dunn and Blakeley, 1988) and pyridoxalphosphate-6-azophenyl-2′,4′-disulfonic acid (PPADS) (Lambrecht et al., 1992) were introduced. However, suramin and PPADS are not selective for P2 receptors (Jacobson et al., 2004) and cannot determine the contribution of specific P2 receptor subtypes in neurotransmitter responses. Development of potent and selective P2Y₁ receptor antagonists, 2′-deoxy-N6-methyladenosine 3′, 5′-bisphosphate (MRS 2179) (Moro et al., 1998), 2-chloro N(6)-methyl-(N)-methanocarba-2′-deoxyadenosine-3′,5′-bisphosphate (MRS 2279) (Boyer et al., 2002), and 2-iodo-N6-methyl-(N)-methanocarba-20-deoxyadenosine 30,50-Bisphosphate (MRS 2500) (Hechler et al., 2006; Jacobson et al., 2004; Jacobson et al., 2020), was particularly instrumental for discerning that P2Y₁ receptors are the principal receptors mediating enteric purinergic inhibitory neurotransmission in GI muscles (discussed below).

Criterion 4 is supported by demonstrating that exogenously applied candidate neurotransmitter is either degraded to inactive metabolites by enzymes on cells of the NEJ, the transmitter is transported back in the presynaptic neuron (reuptake) or it is “washed” from the NEJ by the circulation. The case of purinergic neurotransmission is particularly convoluted by the complex metabolic pathways for extracellular purines that produce both inactive and biologically active metabolites (Zimmermann et al., 2012; De Flora et al., 2004). The situation is further complicated by the fact that the enzymes for purine metabolism are expressed by multiple cell types that may or may not be involved in neurotransmission. Therefore, demonstrating that a substance is metabolized in contact with tissues composed of multiple types of cells is not satisfactory evidence to establish the substance as neurotransmitter. More solid evidence would be to demonstrate that enzymes that metabolize the candidate neurotransmitter are present in cell EJ (discussed below). The subject of neurotransmitter metabolism is important not only to understanding how the neurotransmitter action is terminated. Bioactive purine metabolites may serve as neuromodulators or neurotransmitters on their own (e.g., Durnin et al., 2012). Simultaneous evaluation of neurotransmitters and metabolites is essential for a more complete understanding of chemical neurotransmission.

Fulfilling the four principal criteria establishes with high certainty that a substance is used as a transmitter at a given synapse/NEJ. Practical difficulties discussed above, have prevented these standards from being satisfied rigorously at many types of NEJs, including the NEJs innervated by enteric motor neurons. A comprehensive approach that combines: i) direct measurements of purines released by nerve stimulation, ii) characterization of cellular sources of extracellular purines including neurons, and iii) evaluation of post-junctional responses to nerve stimulation and underlying mechanisms is superior to pharmacological characterization of integrated post-junctional responses to nerve stimulation alone, which is currently the most common approach in studies of enteric inhibitory purinergic neurotransmitters.

6. Methods of measuring purines released from neurons (accuracy and advantages or disadvantages of techniques)

6.1. Paper chromatography

In the original study suggesting that “ATP or a related nucleotide” is the neurotransmitter released from non-adrenergic inhibitory nerves in the gut Burnstock et al., 1970 used paper chromatography to demonstrate increased venous efflux of adenosine and inosine in response to stimulation of vagal non-adrenergic inhibitory nerves in GI muscles. It was reasoned that adenosine and inosine in vascular perfusates likely originated from the release of ATP. In the same study, extracts of unstimulated myenteric plexus of turkey gizzard showed spots that had similar rates of migration across sheets of paper as exogenous ATP, ADP, and AMP. The spots corresponding to ATP and ADP appeared to be diminished in extracts of stimulated myenteric plexus preparations, whereas the spot that corresponded to the migration rate of AMP increased in size. It is interesting to note that there were also spots in the chromatograms of plexus extracts that migrated at different rates than ATP, ADP, and AMP, but these spots were not analyzed in spite of the fact that might have been generated by enteric neurotransmitters (Fig. 2). It was concluded that ATP was likely released from the myenteric plexus preparations upon stimulation and was rapidly degraded to AMP. The results of this study differed from those of another study that also employed paper chromatography and UV absorption to detect released nucleosides in vascular perfusates of guinea-pig stomach in response to stimulation of vagal non-adrenergic innervation. The subsequent study failed to demonstrate an increase in nucleosides after stimulation of non-adrenergic inhibitory nerves (Ishizuka et al., 1978). These results did not support the idea that ATP is the inhibitory transmitter released by stimulation of nerves in the stomach. Discrepancies between the two studies may have occurred from technical differences.

Fig. 2.

Release of nucleotides from isolated Auerbach's plexus (aka myenteric plexus) of turkey gizzard. Paper chromatography of extracts of unstimulated myenteric plexus (control nerve, lane 4) and extracts of myenteric plexus stimulated with EFS (2 ms pulse; at 50 Hz, trains of 45 s every 2 min for 30 min). Voltage was progressively increased from 5 to 50 V (stimulated nerve, lane 5). Media collected from unstimulated plexus (control medium, lane 6) and from stimulated plexus (stimulated medium, lane 7). Lanes 1, 2, 3, and 8 are spotted with authentic AMP, ADP, ATP, and adenine, respectively. The authors emphasized that stimulation caused the release of AMP and traces of ADP and ATP into the medium. Note that there are spots with unknown identity (designated with red arrows and question marks) in extracts of tissue and in media collected from stimulated plexus. These might indicate that additional purines were released upon EFS. Redrawn from Burnstock et al., 1970 with permission.

6.2. Radiolabeled purines

Use of radiolabeled compounds for purine release consist of incubating cells, tissues or brain slices with radiolabeled adenosine or radiolabeled adenine (i.e., [³H]-adenosine or [³H]-adenine). It is presumed that the labeled compounds are taken up by cells and converted to [³H]-AMP, [³H]-ADP, and [³H]-ATP. It is further assumed that [³H]-ATP is loaded into synaptic vesicles, and therefore release of ATP from neurons can be determined by the counts of [³H] released into the medium during nerve stimulation. This approach suffers from low fidelity because: 1) radiolabel adenine labels all intracellular adenine nucleotides and nucleosides and, therefore, the released label (i.e., [³H]) represents a mixture of substances (Sperlagh et al., 2003), 2) radiolabeled purine precursors load into various cell types present in tissues, and therefore, [³H] can originate from multiple cellular sources, including SMCs (Fredholm and Hedqvist, 1980), 3) release of [³H] purines is typically expressed as a percentage of the amount of radioactivity in the tissue at the time of sample collection (i.e. fractional release) and does not provide accurate quantification of transmitter release. This method was used in the original study that concluded that ATP is released from enteric neurons (Su et al., 1971). However, due to significant shortcomings, the method of radiolabeled purines was later viewed as unsuitable for determination of the chemical nature and the cellular sources of purine neurotransmitters.

6.3. Chemiluminescence detection of ATP via the luciferin-luciferase assay

Chemiluminescence was the first method used to measure the release of ATP from nerves (Holton, 1959), and it is still used widely to measure extracellular ATP. The luciferin-luciferase chemiluminescence assay is based on analyzing the bioluminescence of the ATP-dependent luciferase-mediated oxidation of luciferin to oxyluciferin, which yields light at 560 nm. Since the reaction depends on the presence of ATP, the resulting luminescence is proportional to the ATP concentration (Manfredi et al., 2002; Lundin, 2014). The assay is rapid, relatively simple, highly sensitive (femtomolar range) and with low toxicity. However, this method also has shortcomings. First, interference from the sample matrix can cause variable results. The reaction inherently depends on O₂ concentration that can be inhibited by cellular factors and pharmacological agents, including anions, anion transport inhibitors, ion channel blockers, cationic surfactants, salts, bacterial contamination, or P2 receptor antagonists (Tani et al., 2008; Praetorius and Leipziger, 2009; Rajendran et al., 2016). Second, product inhibition is an often unrecognized complication making it difficult to achieve quantification of ATP (Leitao and Esteves da Silva, 2010; Lundin, 2014). Third, the luciferin-luciferase system has limited applicability in real time measurements of ATP since bioluminescence imaging typically requires long exposure times that limits spatiotemporal resolution and diminishes luminescence (Rajendran et al., 2016). Finally, the method is restricted to ATP and overlooks possible release of other purines and generation of metabolites that might affect neurotransmission. This method was used to demonstrate that ATP was released by stimulation of nerves in guineapig taenia coli and bladder (Burnstock et al., 1978) and to characterize release of ATP from isolated varicosities of guinea-pig ileum myenteric plexus by depolarization with high-K⁺ solution or the Na⁺ channel opener, veratridine (White and Leslie, 1982). Since the 1980s, the luciferin-luciferase chemiluminescence assay has been used very little in studies of enteric motor neurotransmission, because ATP was accepted to be the enteric inhibitory neurotransmitter, eliminating further need to investigate the identity of the transmitter.

In recent years, imaging of single photons generated during the chemiluminescent reaction between luciferin and luciferase has been used to monitor vesicular and non-vesicular release of ATP from axons (Fields, 2011). The technique has been enhanced by developing single photon detectors with avalanche photodiodes in conjunction with optical fiber-based systems for local detection of ATP release from single cells (Iinuma et al., 2016), but it has not yet been used in complex tissues such as smooth muscles. Attempts have been made to improve the sensitivity and utility of luciferase as a sensor of ATP release into localized or restricted extracellular spaces by selectively targeting the luciferase to the extracellular surface of platelets that were coated with a protein A-luciferase chimera (proA-luc) stably absorbed onto the surface of intact platelets (Beigi et al., 1999). Local extracellular ATP levels can be continuously assayed with high sensitivity and excellent temporal and spatial resolution using either suspended platelets or adherent cells that are coated with proA-luc. This method might be useful for the online detection of ATP released from cells on tissue surface such as epithelia but has not been augmented to measure ATP release from cells within the walls of organs.

6.4. High-performance liquid chromatography

High performance liquid chromatography (HPLC) represents another frequently used technique for detection of extracellular purines. The HPLC method has the advantage of high efficiency as it is able to separate and identify all purine compounds released to the extracellular space in one run. Different HPLC techniques such as ion-exchange HPLC or reversed-phase HPLC coupled with UV detection or fluorescence detection are commonly used for separation of nucleotides and nucleosides (Zakaria and Brown, 1981; Manfredi et al., 2002). Ion-exchange HPLC offers good separation of nucleotides, however the process is relatively long and ion-exchange columns are less stable than reversed-phase columns. The reversed-phase HPLC-UV method detects UV absorbance with detection sensitivity in the nanomolar range (Manfredi et al., 2002). The method is suitable for detection of extracellular purines during inflammation, cell lysis, tumor development, and apoptosis, but this method lacks the sensitivity to detect physiological extracellular concentrations of ATP. Highly sensitive methods are particularly required for measurements of released neurotransmitters that are diluted in cell or tissue medium and in superfusates. The HPLC method that is coupled with fluorescence detection (HPLC-FLD) involves 1,N⁶-etheno-derivatization of the endogenous purines and was first used to measure ATP and metabolites in the guinea-pig vas deferens (Levitt et al., 1984). In our own research, we advanced the technique and achieved unprecedented sensitivity of purine detection in the high attomolar (10⁻¹⁶) to low femtomolar (10⁻¹⁵) range (Bobalova et al., 2002). This methodology has been instrumental for identifying novel purines released from enteric neurons and their metabolites (discussed in the sections Metabolism of extracellular purines and Novel purines that act as enteric motor neurotransmitters). The main advantages of the HPLC-FLD methodology are high sensitivity, specificity, and efficiency. To analyze coeluting substances, we developed the method of HPLC fraction analysis to complement the HPLC-FLD methodology. Using this methodology, we were able to determine that β-NAD⁺ was the primary substance that yielded 1,N⁶-etheno-ADPR in the samples collected during nerve stimulation in many smooth muscles (Smyth et al., 2004; Breen et al., 2006; Mutafova-Yambolieva et al., 2007; Hwang et al., 2011; Durnin et al., 2012). The most significant shortcoming of HPLC analysis of bulk solution samples is its low spatial and temporal resolution. Such disadvantage is shared by all methodologies that analyze neurotransmitters in extracellular media, sampled after stimulation of cells or tissues. This methodology is most valuable for validation of a putative neurotransmitter when it is combined with studies of cellular sources of purine release, mechanisms of extracellular metabolism, and post-junctional signal transduction mechanisms. We used this combination of approaches in studies of purinergic neurotransmission in the colon, as discussed in more detail below (Mutafova-Yambolieva et al., 2007; Hwang et al., 2011; Durnin et al., 2012; Durnin et al., 2013; Durnin et al., 2014; Durnin et al., 2017; Durnin et al., 2020).

6.5. Biosensors

The principle of ATP biosensors exploits the binding or reacting property of ATP to biological molecules/receptors and a secondary signal coupled to this binding process. As of now, such methodologies are restricted mainly to ATP or adenosine (Dale, 2021).

Electrochemical microelectrode biosensors:

The basic principle of electrochemical biosensors is that an enzymatic cascade recognizes the purine of interest (e.g., ATP or adenosine) and makes a product (e.g., H₂O₂) that can be detected electrochemically by a microelectrode. The electrochemical signal is proportional to the ATP or adenosine concentration (Llaudet et al., 2003; Dale et al., 2005). ATP-sensitive microelectrodes provide rapid sampling and the ability to monitor ATP concentration deep in the brain in vivo or in brain slices (Dale, 2021). The threshold sensitivity of ATP assay by microelectrodes is about 100 nM, which is far less than the sensitivity of ATP detection by chemiluminescence or HPLC-FLD. There are also problems with selectivity of this approach. It is recognized that electrochemical sensors can respond to many electroactive species in the milieu in addition to the analyte of interest (e.g., ATP or adenosine). There have been attempts to remove interreferences by utilizing a “null” electrode (Llaudet et al., 2005; Masson et al., 2008), but the success of such approaches is variable and depends on the size of the interference signal and the stability of sensor and “null” electrodes. These biosensors operate at rather positive potentials necessary for the oxidation of H₂O₂ and hence many other compounds will also oxidize at these potentials causing electrochemical interference. The ATP-selective microelectrodes are also subject to degradation by protein fouling and difficulties in calibration for accurate measurements. Finally, electrochemical approaches to measure ATP have relatively low spatial resolution (Dale and Frenguelli, 2012). Electrochemical biosensors for ATP and adenosine have been used in studies of brain tissues (Dale and Frenguelli, 2012; Dale, 2021), gastrointestinal mucosa (Patel et al., 2011), and myenteric ganglia (Brown et al., 2016), but have not been used to measure release of purines from motor neurons within GI muscles.

Cell-based biosensors:

The sensing elements can also be the entire cell or an excised membrane patch that have built-in naturally occurring receptors for specific target molecules (Brown and Dale, 2002; Pangrsic et al., 2007; Dale, 2021). Activation of these receptors elicits cellular responses that can be monitored optically or electrochemically. P2X receptors are ATP-gated cation channels and thus produce ATP-dependent currents that can be measured with the patch clamp techniques. PC12 cells (Praetorius and Leipziger, 2009) or HEK-293 cells (Hayashi et al., 2004) that express P2X receptors in the plasma membrane can be used to detect ATP by monitoring receptor-mediated Ca²⁺ influx. A bullfrog fibroblast cell line expressing G protein coupled P2Y receptors has been used to monitor ATP release by activation of surface P2Y receptors and increased intracellular Ca²⁺. This technique has been applied in primary cell cultures, established cell lines, freshly isolated single cells or on surface of freshly isolated tissue slices (Hayashi et al., 2004; Fujii et al., 2017). This method provides good temporal resolution and detects ATP concentrations above 10 μM (Hayashi et al., 2004; Fujii et al., 2017). There are limitations regarding the use of receptors due to desensitization of the receptor and loss of cellular responses. The method is appropriate to detect local ATP release at the surface of cell or tissue, but it is not suitable for measuring neurotransmitter release in complex tissues such as GI muscles.

Genetically-encoded optical sensors:

Genetically-encoded fluorescence resonance energy transfer (FRET)-based indicators for ATP, ATeams, have been used to measure intracellular levels of ATP in single HeLa cells in the range from 2 μM to 8 mM (Imamura et al., 2009). ATeams could also be targeted to the nucleus and mitochondria and probably other organelles. Other tools for measurements of intracellular ATP have also been described (Rajendran et al., 2016). Genetically-encoded fluorescent biosensors for detecting intracellular NAD⁺ concentrations in subcellular compartments of cultured cells are based on measuring NAD⁺ dependent fluorescence changes in the cells (Cambronne et al., 2016; Zou et al., 2020). A NAD-Snifit sensor for measuring intracellular levels of NAD⁺ has recently been developed (Sallin et al., 2018). The sensor consists of two synthetic fluorophores attached to self-labeling proteins and a NAD binding protein, sulfapyridine (SPR). Cofactor-dependent binding of the intramolecular ligand to the SPR leads to a ratiometric FRET signal. Though exploited in intracellular contexts, these sensors have not been used to detect extracellular NAD⁺. A ratiometric ATeam FRET-based ATP sensor, ecAT3.10, was developed for monitoring real-time changes in extracellular ATP levels by targeting the sensor to the cell surface of Neuro2A cells in culture (Conley et al., 2017). More recently, a genetically encoded ADP sensor that reports fluctuations in extracellular ADP through changes in FRET between a cyan fluorescent protein and a yellow fluorescent protein has been used to detect osmotic shock-induced extracellular nucleotide release in HEK293A cells (Trull et al., 2019). A potential challenge for such sensors is the large concentration range of extracellular nucleotides in basal and stimulated conditions that require adequate sensitivity over slower timescale as well as to study faster events. Genetically encoded sensors for extracellular nucleotide measurements in complex tissues remain to be developed.

Progress has been made to detect intracellular concentrations of ATP and NAD⁺ or extracellular release of ATP and ADP in cell cultures. However, the methodologies for measuring regulated release of purines in the extracellular space, and from neurons in particular, have been slower to develop. No single measurement system can currently fulfill the entire range of requirements in terms of sensitivity, spatial resolution, temporal resolution, and quantitative accuracy. Multifaceted approach that evaluates the release, degradation and post-junctional action of the entire range of purinergic neurotransmitters will be a goal for future studies of purinergic neurotransmission.

7. Metabolism of extracellular purines

The extracellular metabolism of purines is remarkably complex, and part of this complexity comes from the fact that multiple cell types make up the post-junctional targets of enteric motor neurotransmission. Smooth muscle tissues of the GI tract contain extensive populations of interstitial cells that are in close proximity to the varicose processes of motor neurons. These cells, interstitial cells of Cajal (ICC) and platelet-derived growth factor receptor α-positive (PDGFRα⁺) cells (Fig. 3), make gap junctions with SMCs (Horiguchi and Komuro, 2000), and thus form an electrically coupled network of cells known as the SIP syncytium (Sanders et al., 2012; Sanders et al., 2014). PDGFRα⁺ cells have been referred to as ‘fibroblast-like’ cells by morphologists for many years.

Fig. 3.

Double immunolabeling for PDGFRα (green) and c-Kit (red) in muscles from human colon. (A–C) Circular muscle layer of the ascending colon (C shows merged files). (D–F) Myenteric plexus region of the sigmoid colon (F shows merged files). PDGFRα⁺ cells and c-Kit⁺ cells (ICC) are distinct cells and form networks in the same anatomical regions. Scale bar is 10 μm in all panels. Redrawn with permission from Kurahashi et al., 2012.

Metabolism of purines is also complicated by the existence of several groups of enzymes including the ecto-nucleoside triphosphate diphosphohydrolases (ENTPDases), NAD-glycohydrolases, ecto-5′-nucleotidase (NT5E)/CD73, ecto-nucleotide pyrophosphate phosphodiesterases (E-NPPs), and alkaline phosphatases (Fig. 4). ENTPDases and NT5E/CD73 represent the enzyme cascades that perform the sequential metabolism of nucleoside triphosphates and diphosphates to adenosine and are the most extensively studied enzymes in extracellular purinergic signaling. Metabolism of extracellular purines can lead not only to termination of a biological action, but also to formation of by-products, notably ADP, ADPR and adenosine, that can have signaling functions on their own through different receptors. Triggering of purinoceptors by their ligands regulates important physiological functions such as GI motility and regulation of the development of inflammation. Therefore, to understand the extracellular purinergic signaling in GI muscles, including purinergic enteric motor neurotransmission, it is necessary to determine what enzymes are present and what cell types are involved in metabolic processes.

Fig. 4.

Biotransformation pathways for extracellular purines. CD38 and CD157 belong to the family of ADP-ribosyl cyclases and convert β-NAD⁺ to ADPR and nicotinamide (Nam) and to cyclic ADPR (cADPR) and Nam. The primary activity of NAD-glycohydrolases (NADase) is the degradation of β-NAD⁺ to ADPR. Cell-surface ADP-ribosyl transferases (ARTs) transfer ADP-ribose groups from β-NAD⁺ on target proteins. ADPR is catabolized to AMP by ENPPs. ENPPs can also directly degrade β-NAD⁺ to AMP and nicotinamide mononucleotide (NMN). ATP is degraded sequentially to ADP and AMP or directly to AMP by ENTPDases and ENPPs. Up4A is metabolized to ATP plus UMP in the human colon (h) and to ADP plus UDP in the murine colon (m). AMP is the convergence point of catabolism pathways for β-NAD⁺, ADPR, ATP, and Up4A. AMP is then degraded to adenosine (ADO) and phosphate.

ENTPDase1, 2, 3, and 8 are surface-located enzymes that hydrolyze ATP/ADP into AMP. ENTPD2 however displays preferential ecto-ATPase activity (Kukulski et al., 2005). NTPDases 4, 5, 6, and 7 are intracellular proteins, with ENTPD5 and 6 being secreted upon heterologous expression. E-NTPDases are glycosylated proteins with two transmembrane domains. Enzyme and immune histochemistry studies of the gastrointestinal tract have shown that ENTPDase1 is expressed in mucosa (Sévigny et al., 1998), SMCs (Sévigny et al., 1998), blood vessels (Sevigny et al., 2002), and immune cells (Neshat et al., 2009) whereas ENTPDase2 and 3 are colocalized in certain epithelial cells of the upper digestive tract (Grubišić et al., 2019). ENTPDase2 is expressed in mouse and guinea-pig enteric glia cells (Braun et al., 2004; Grubišić et al., 2019) and in mouse and human enteric neurons (Feldbrügge et al., 2017) whereas ENTPDase3 appears to be expressed in enteric neurons of myenteric and submucosal plexuses and in tunica muscularis (Grubišić et al., 2019). A recent study reported that ENTPDase8 is expressed in the epithelium of mouse colon and might be protecting the intestine from development of inflammation (Salem et al., 2021). Therefore, ENTPDases are likely involved in regulation of secretory and barrier functions of the GI mucosa, in gut motility and in neuromuscular communication. Transcriptome analyses of mouse colon and fluorescence-activated cell sorting (FACS)-purified SMCs, ICC and PDGFRα⁺ cells revealed that Entpd1 is expressed chiefly in SMCs, Entpd2 is primarily expressed in PDGFRα⁺ cells, Entpd3 has low expression in SIP cells, Entpd4 is enriched in SMCs and ICC, Entpd5 is enriched in PDGFRα⁺ cells, whereas Entpd6,7,8 have low expression in colon muscularis and SIP cells (Lee et al., 2015; Breland et al., 2019). The significance of differential distribution of ENTPDases in different cells of the colon for extracellular purinergic signaling and colonic motility remains to be elucidated.

NT5E/CD73, which converts AMP into adenosine, has been described both as GPI-anchored protein or soluble enzyme (Airas et al., 1997; Yegutkin et al., 2000). Immunohistochemistry studies in gastrointestinal preparations have demonstrated that NT5E/CD73 is expressed in membranes of SMCs in rat ileum (Giron et al., 2008), but not in enteric neurons of the guinea-pig GI tract (Lavoie et al., 2011). NT5E/CD73 in the GI tract has been studied mainly in the context of inflammation that is improved by adenosine (Bynoe et al., 2012; Kaku et al., 2014). Our own research on extracellular metabolism of β-NAD⁺ in colon muscularis of wild-type and Nt5e^−/− mice demonstrated that NT5E is the enzyme that is responsible for adenosine production from β-NAD⁺ (Durnin et al., 2020). Studies of cells purified by FACS indicated that NT5E is expressed mainly by SMCs and moderately by PDGFRα⁺ cells in the murine colon (Durnin et al., 2020). Since PDGFRα⁺ cells are the chief cell target of the enteric purine neurotransmitter (Hwang et al., 2012), discussed in Cells and receptors involved in enteric inhibitory motor neurotransmission, NT5E appears to be involved in termination of the transmitter action of β-NAD⁺ and producing adenosine may also contribute to relaxation of SMCs during enteric inhibitory nerve stimulation.

E-NPPs hydrolyze 5′-phosphodiester bonds in nucleotides and their derivatives, resulting in the release of 5′-nucleotide monophosphates (Bollen et al., 2000). Thus, E-NPPs can produce AMP from β-NAD⁺, ADP-ribose or ATP. So far, five members (NPP1–5), characterized by a similar modular structure, have been cloned and intensively studied. NPPs can also directly degrade β-NAD⁺ to AMP, bypassing the production of ADPR. AMP that is produced by metabolism of either ATP/ADP or β-NAD⁺/ADPR is then degraded to adenosine by NT5E/CD73 (Fig. 4). We also found that the new putative enteric inhibitory neurotransmitter uridine adenosine tetraphosphate (Up4A) produces adenosine in both human and murine colon muscularis (Durnin et al., 2014). Therefore, adenosine can be produced by cascade metabolism of ATP, β-NAD⁺, ADPR or Up4A that are released in the extracellular space. This is particularly relevant to the enteric inhibitory motor neurotransmission, since ATP (Burnstock et al., 1970; Burnstock, 2008), β-NAD⁺ (Mutafova-Yambolieva et al., 2007; Hwang et al., 2011), ADPR (Durnin et al., 2012), and Up4A (Durnin et al., 2014) have been proposed to be enteric inhibitory neurotransmitters in the colon.

CD38 and CD157/BST1 function as both receptors and ectoenzymes and belong to the same family of NAD⁺ converting enzymes with NAD-glycohydrolase and ADP-ribosyl cyclase activities (Berthelier et al., 1998). CD38 can be expressed on cell surfaces where it faces the extracellular milieu or in intracellular compartments such as endoplasmic reticulum, nuclear membrane and mitochondria (De Flora et al., 2004). The hydrolysis of NAD⁺ to ADPR and nicotinamide is the primary extracellular catalytic reaction that is mediated by CD38. CD38 is mainly expressed in hematopoietic and immune cells. However, CD38 might be involved in functions of the neuroeffector junction in smooth muscles, because it is present in postganglionic perivascular nerve terminals (Smyth et al., 2006). In the tunica muscularis of primate colons degradation of β-NAD⁺ to adenosine was greater in circular muscle containing no ganglia but a higher proportion of nerve processes and nerve terminals per unit weight than whole muscle preparations (Durnin et al., 2012). In contrast, the catabolism of ATP was similar in both tissue preparations. We concluded, therefore, that β-NAD⁺ was likely degraded near nerve terminals/site of release whereas ATP appeared to be degraded at sites different from sites of neurotransmitter release. Indeed, we also found that ganglionic stimulation of motor neurons in mouse and monkey colons caused release of β-NAD⁺ from nerve varicosities within the circular muscle, whereas the main source of ATP appears to be from ganglia (Durnin et al., 2013). In experiments performed on colon muscularis of wild type and CD38^−/− mice, β-NAD⁺ produced adenosine with equal efficiency, suggesting that the degradation of β-NAD⁺ was likely accomplished by at least one additional pathway besides a CD38-mediated mechanism (Durnin et al., 2012). Our recent study confirmed these observations and demonstrated that in the murine colon β-NAD⁺ is degraded directly to AMP by ENPP1, whereas CD38 likely metabolizes β-NAD⁺ to ADPR, which in turn is metabolized to AMP by ENPP1 (Durnin et al., 2020). CD157/BST1 is unlikely to have a prominent role in the catabolism of extracellular β-NAD⁺ in the gut since transcriptome analyses showed very low expression of Bst1, the gene encoding CD157, in mouse small and large intestines (Lee et al., 2015; Breland et al., 2019). The family of ecto ADP-ribosyl transferases (ARTs) is another group of cell-surface enzymes that use extracellular NAD⁺ as a substrate and transfer ADP-ribose groups onto target proteins (Seman et al., 2004). This family of enzymes contains five known mammalian members, ART1-ART5, which are GPI-anchored membrane proteins (ART1-ART4) or secreted enzymes (ART5) (Glowacki et al., 2002). Deep sequencing analyses revealed that Art3 is expressed highly in SMCs of mouse jejunum and colon and in PDGFRα⁺ cells in jejunum whereas Art4 is highly expressed in PDGFRα⁺ cells of both jejunum and colon (Lee et al., 2015; Breland et al., 2019). The roles of ARTs in the SIP syncytium or elsewhere in the GI tract are unknown.

Clearly, the extracellular space is a metabolically active compartment where several enzyme families capable of catabolizing nucleotides exist. Some of these enzymes have broad and partially overlapping substrate specificities and cellular localizations and may yield biologically active metabolites. Although frequently unappreciated, it is important to identify the cell type expressing specific enzymes to understand how extracellular purine metabolism in multicellular environment shapes the final physiological response (Fung and Vanden Berghe, 2020). We used a cell-based approach to investigate how the primary inhibitory neurotransmitter, β-NAD⁺, is catabolized in the mouse colon (Durnin et al., 2020). Our goal was to determine the key enzymes and cell types of the SIP syncytium that contribute to the degradation of β-NAD⁺ to ADPR, AMP, and adenosine. We used 1,N⁶-etheno-NAD (eNAD), a highly fluorescent derivative of β-NAD⁺, as a substrate and investigated eNAD metabolism in colon muscularis isolated from mice lacking major ecto-enzymes such as CD38, ENPP1, and NT5E and found that eNAD is catabolized to ADPR by CD38, to AMP by ENPP1, and to adenosine by NT5E in tissues. However, after FACS isolation of SMCs, ICC and PDGFRα⁺ cells, we found that SMCs and PDGFRα⁺ cells are the cell types that are involved in eNAD degradation to etheno-adenosine by ENPP1 and NT5E. NT5E appears to be expressed predominantly in SMCs and moderately in PDGFRα⁺ cells, whereas ICC do not participate in the degradation of eNAD. CD38 contributes to the sequential degradation of eNAD, but its primary location is outside of the SIP syncytium and likely to be enteric neurons. A model showing cell-specific degradation of β-NAD⁺ in colonic NEJ is shown in Fig. 5. It is possible that the enzymes responsible for purine metabolism could become therapeutic targets in motility disorders.

Fig. 5.

Schematic of the SIP syncytium showing mechanisms of co-transmission at play in enteric inhibitory neurotransmission. SIP syncytium is composed of SMCs, ICC, and PDGFRα⁺ cells. These cells are electrically coupled through gap junctions (GJ). Conductance changes in any SIP cell affects the excitability of other cells in the syncytium. The current concept is that both a purine neurotransmitter and NO are released from enteric inhibitory neurons (EIN). Current data favor the following concept for purinergic neurotransmission: β-NAD (and possibly ADPR and Up4A; latter not shown) are released from EIN and bind to post-junctional P2Y₁ receptors expressed dominantly by PDGFRα⁺ cells in the SIP syncytium. P2Y₁ receptors couple through G (G_q/G₁₁) proteins to activate PLCβ and production of inositol-1,4,5-trisphosphate (IP₃). IP₃ enhances release of Ca²⁺ from the endoplasmic reticulum (ER). Release of Ca²⁺ causes activation of SK3 channels in the plasma membrane and hyperpolarization of PDGFRα⁺ cells. The hyperpolarization response conducts to SMCs via gap junctions, causes fIJPs and exerts a relaxation effect on the muscle. Released purine neurotransmitters come into contact with membrane-bound enzymes in cells that participate in the neuroeffector junction. β-NAD is metabolized to ADPR by CD38 possibly located in the plasma membrane of EIN. ADPR is sequentially metabolized to AMP by ENPP1 and to adenosine (ADO) by CD73. Both enzymes are located in plasma membranes of PDGFRα⁺ cells. ADO that is formed by the degradation of both β-NAD and ADPR binds to adenosine receptors (AR) in PDGFRα⁺ and SMCs. ATP is primarily released from nerve cell bodies and non-neural cells. Extracellular ATP can reach SMCs and is sequentially degraded to ADP, AMP and ADO by NTPDases and CD73 in plasma membranes of SMCs. ATP and its metabolites can activate purinergic receptors in SMCs. NO is thought to be released from the same EIN. NO diffuses through the ICC membrane and binds to soluble guanylate cyclase dimers (GCa and GCb). cGMP is produced and leads to inhibition of Ca²⁺ release from the ER. Inhibition of Ca²⁺ release decreases the activation of ANO1 channels in ICC, also produces the sIJP that conduct to SMCs and cause inhibition of contractions. Released NO is rapidly oxidized to ${\text{NO}}_3^{-}$ or forms S-nitrosothiols (S-NO). VIP and PACAP are also released from EIN, however their pathway(s) of action in the SIP syncytium has not been described. The figure also depicts the many metabolic pathways for the metabolism of the purines released and NO. See text for the details about metabolic pathways. Redrawn with permission from Durnin et al., 2020.

8. Why bioassays to identify neurotransmitter substances provide confusing results

The inherent complexity of nucleotide signaling raises significant challenges for determining the actual compounds involved in purinergic neurotransmission. For example, single cell types can express a variety of receptors. Immunohistochemistry and functional studies suggest that myenteric neurons express P2X₂ (Castelucci et al., 2002), P2X₃ (Poole et al., 2002), P2X₇ (Gulbransen et al., 2012) and P2Y₁ receptors (Thornton et al., 2013) whereas SMCs of guinea-pig taenia coli appear to express P2Y₁ and P2Y₁₁ receptors (King and Townsend-Nicholson, 2008). Moreover, purinergic receptors can be expressed in several cell types. For example, different P2X isoforms are expressed in circular and longitudinal muscles of the canine colon (Lee et al., 2005). P2Y₁ receptors are highly expressed in PDGFRα⁺ cells of the murine colon (Peri et al., 2013; Breland et al., 2019) and to a degree in enteric glia (Brown et al., 2016; Fung et al., 2017), myenteric neurons (Gwynne and Bornstein, 2009), and SMCs (King and Townsend-Nicholson, 2008). The contributions of specific receptors and cell types in purinergic signaling in complex tissues remain difficult to resolve.

Understanding purinergic neurotransmission is further complicated because several endogenous ligands exist for the same receptors, including those generated by extracellular purine metabolism. For example, P2Y₁ receptors are activated by ADP (Waldo and Harden, 2004; Jacobson et al., 2020), β-NAD⁺ (Mutafova-Yambolieva et al., 2007), ADP-ribose (Gustafsson et al., 2011), Up4A (Durnin et al., 2014), and Ap4A (Paquola et al., 2019). The complexities surrounding ATP are particularly interesting. Based on inhibition of responses to EFS in GI smooth muscles by specific P2Y₁ receptor antagonists, ATP is commonly assumed to be the endogenous activator of P2Y₁ receptors (Gallego et al., 2006; Wood, 2006). In fact, use of purified human P2Y₁ receptors, under conditions in which no nucleotide metabolism occurs, showed that ATP acts as a partial agonist of the P2Y₁ receptors with significantly lower efficacy and affinity than ADP or 2MeSADP (Waldo and Harden, 2004). Other studies demonstrated that ATP acts either as a partial agonist or even as a competitive antagonist of P2Y₁ receptors (Hechler et al., 1998; Palmer et al., 1998), perhaps depending on the number of P2Y₁ receptors in the test system, leading to the conclusion that ATP is “not of major importance as a physiological regulator of the P2Y₁ receptor” (Waldo and Harden, 2004). So the question arises as to whether ATP is the only (or the main) inhibitory purine neurotransmitter in the gut, because P2Y₁ receptors mediate the purinergic part of enteric inhibitory neurotransmission (Gallego et al., 2006; Mutafova-Yambolieva et al., 2007; Gallego et al., 2012; Hwang et al., 2012). Studies that compare ADP and ATP as ligands for P2Y₁ receptors suggest that ADP is the preferred ligand. However, it should be noted that ADP also activates P2Y₁₂ and P2Y₁₃ receptors (Jacobson et al., 2020), which may contribute to the integrated effects of ADP. Responses to exogenous purines in complex tissues are likely mediated by mixed purine receptors. Functional studies of tissues using receptor agonists and antagonists to determine the potency order of purine ligands or to determine the receptor that mediates a particular response are predestined to yield perplexing results.

Another major difficulty in the study of purinergic signaling in intact tissues originates from the fact that released purines are metabolized by a cohort of enzymes located on the surfaces of many cell types (discussed in Metabolism of extracellular purines). Some of the products of purine catabolism (e.g., ADP, ADP-ribose, and adenosine) are biologically active and likely participate in integrated responses. Enzymes that participate in extracellular purine metabolism control the levels of biologically active metabolites in the vicinity of purine receptors. Studies with intact tissues are complicated by metabolism and interconversion of extracellular nucleotides and nucleosides. Adenosine is perhaps the most studied ATP metabolite in GI smooth muscles. Adenosine activates four G-protein coupled receptors, namely A₁, A_2A, A_2B, A₃ (Fredholm et al., 2011). Expression of these receptors is found throughout the GI tract (Christofi et al., 2001). In functional studies receptors are commonly distinguished by the relative potency of adenosine receptor agonists and antagonists. Such studies have suggested that adenosine inhibits neuronal excitability/synaptic transmission (Christofi and Wood, 1993; Christofi and Wood, 1994) and the release of ACh (Nitahara et al., 1995; Lee et al., 2001), substance P (Moneta et al., 1997) and other tachykinins (Broad et al., 1992) in the myenteric plexus of guinea-pig ileum. In isolated rat longitudinal muscle with the myenteric plexus attached, bath-applied adenosine decreased the EFS-evoked release of [³H]-acetylcholine. In the presence of the A₁ receptor antagonist 8-cyclopentyl-1,3-dipropylxanthine (DPCPX), the inhibitory effect of exogenous adenosine was eliminated but an increase of EFS-evoked overflow of [³H] was revealed. The EFS-evoked overflow of [³H]-acetylcholine was enhanced by the A_2A receptor agonist 2-p-(2-carboxyethyl) phenylethylamino-5′-N-ethylcarboxamido-adenosine (CGS 21680), by increasing the endogenous level of adenosine by adding its precursor AMP or by inhibition of its uptake and degradation. This increase of ACh release was eliminated by the A_2A receptor antagonist 4-(2-[7-amino-2-(2-furyl) [1,2,4]-triazolo[2,3-a][1,3,5]triazin-5-yl amino]ethyl)phenol (ZM 241385). Removal of adenosine by adding adenosine deaminase mimicked the effects of A_2A receptor inhibition. It was proposed that adenosine has dual effects on ACh release in this preparation mediated by junctional facilitatory A_2A receptors and extrajunctional inhibitory A₁ receptors on myenteric neurons of this preparation (Duarte-Araújo et al., 2004). Similar conclusions were reached in a study of the twitch contractile responses to EFS of guineapig ileum preparations in the presence of A₁ and A_2A receptor antagonists (Tomaru et al., 1995). Other studies have suggested that adenosine A₁ receptors mediate direct inhibition of the smooth muscle contractile activity in rat duodenum (Nicholls et al., 1996), and rat ileum (Nicholls and Hourani, 1997). In human colon, responses of circular muscle preparations to EFS were enhanced by inhibition of A₁ receptors with DPCPX and of A_2A receptors with ZM 241385. Conversely, the A₁ receptor agonist CCPA and the A_2A receptor agonist CGS 21680 reduced the contractile responses to EFS. The effects of the A_2A receptor drugs were abolished by a nNOS inhibitor (N^ω-propyl-l-arginine) whereas the effects of the A₁ receptor agonist and antagonist remained intact. The contractile responses to carbachol in the presence of tetrodotoxin were enhanced by DPCPX or reduced by CCPA but were unaffected by the A_2A receptor drugs. It was concluded that adenosine directly stimulates inhibitory A₁ receptors on SMCs of human colon whereas A_2A receptors stimulate inhibitory nitrergic neurotransmission. A₁ receptors appeared to influence the human colon contractility with lower efficiency than A_2A receptors (Fornai et al., 2009). Adenosine also appears to cause relaxation of intestinal smooth muscle by direct activation of adenosine A_2B receptors on SMCs of guinea-pig taenia coli (Prentice and Hourani, 1997) and distal colon (Kadowaki et al., 2000). Clearly, the receptors that mediate the effects of adenosine on GI motility are difficult to study due to the great variability of adenosine effects depending on species, region of GI tract, and use of pharmacological tools. However, the fact that adenosine is biologically active is undisputable. Interestingly, qPCR and transcriptome analyses demonstrated high expression of Adora1, the gene for the adenosine A₁ receptor, in PDGFRα⁺ cells isolated from mouse colonic muscles (Peri et al., 2013; Breland et al., 2019). The role of the A₁ receptor in these cells that are the main target of the purine neurotransmitter in the colon remains to be determined. In any event, adenosine present in the NEJ and in the vicinity of cells of the SIP syncytium, likely contributes to the integrated responses to nerve stimulation or exogenous application of ATP (as depicted in Fig. 5).

A fact commonly ignored in functional studies is that ATP is not the only, and perhaps not even the main, source of endogenous adenosine. Extracellular β-NAD⁺, ADPR and Up4A also metabolized to adenosine (Zimmermann et al., 2012; Durnin et al., 2012; Durnin et al., 2014) (Fig. 4). Therefore, adenosine effects must be anticipated in every system in which these substances are released into the extracellular space with the machinery for purine metabolism. Responses of GI muscles to bath-applied purines can be confusing and difficult to interpret. For example, a study of guinea-pig ileum and human jejunum reported that bath-applied β-NAD⁺ had little or no effect on SMC membrane potential, but it inhibited smooth muscle contractions in response to EFS. The effect of β-NAD⁺ on contractions was abolished by the A₁ receptor antagonist 2-chloro-N6-cyclopentyladenosine but not by MRS 2179, an inhibitor of P2Y₁ receptors. From these observations the authors concluded that β-NAD⁺ is not a neurotransmitter in these preparations but it is a direct ligand of presynaptic adenosine A₁ receptors (Wang et al., 2015). We have found that very brief contact (1–5 s) of β-NAD⁺ with intestinal tissues results in generation of β-NAD⁺ metabolites, including adenosine (Durnin et al., 2012). Thus, it is not surprising that effects of bath-applied β-NAD⁺ might be dominated by its breakdown products such as adenosine.

Another issue with bath application of candidate neurotransmitters is that these substances may not bind to the same population of receptors that neurotransmitters released from neurons encounter in the more confined volume of the NEJ. Responses to bath-applied substances may be mediated largely by extrajunctional receptors whereas effects of neurotransmitters are likely mediated by junctional receptors. Such an example was provided above in which the effects of adenosine on ACh release in rat intestine appear to be mediated by extrajunctional A₁ receptors and intrajunctional A_2A receptors (Duarte-Araújo et al., 2004). In the murine gastric fundus different Ca²⁺ sensitization mechanisms are recruited by bath-applied muscarinic agonists and ACh released from neurons (Bhetwal et al., 2013). The discrepancy between responses to bath application and release from neurons is also seen with β-NAD⁺. Bath-applied β-NAD⁺ causes modest hyperpolarization or no effect in intestinal smooth muscle preparations (e.g. Gallego et al., 2011; Gallego et al., 2012; Wang et al., 2015), but when β-NAD⁺ is applied in a focused manner by picospritzing, it mimics responses to nerve stimulation (discussed in Novel purines that act as enteric motor neurotransmitters). Neurotransmitters are stored in synaptic vesicles at concentrations that can reach up to 150–1000 mmol/L (Van der Kloot, 2003; Zimmermann, 2008). Thus, rapid localized application of small volumes of exogenous neurotransmitter candidate are likely to better simulate neurotransmitter release than less direct application of substances in solutions bathing the tissue.

9. Cells and receptors involved in enteric inhibitory motor neurotransmission

As discussed above, the nature of the cells and receptors responsible for post-junctional effects of enteric inhibitory motor neurons was probably obscured by the use of bioassays to assess both the neurotransmitter substance and its post-junctional effects. Complex responses are possible in intact GI muscles due to the widespread expression of P2 and P1 receptor subtypes by a variety of cells. Before selective agonists and antagonists were developed, immunohistochemistry and in situ hybridization were used to attempt to identify the cell-specific localization of receptors. P2Y₁ receptor mRNA was detected only in the myenteric plexus and submucosal plexus in the mouse ileum (Giaroni et al., 2002). In the same study, P2Y₁ and P2Y₂ receptors were detected in the circular and longitudinal muscle layers by immunohistochemistry, however the specific cells with P2Y₁-LI were not identified. A few neurons in the myenteric plexus also expressed P2Y₁-LI. P2Y₂ receptors were also noted in a subpopulation of myenteric neurons, however with a lower level of immunoreactivity. The P2Y₁ receptor antagonist, MRS 2179 (Camaioni et al., 1998) caused partial block of relaxation responses of mouse small intestinal longitudinal muscles to ATP and 2-MeSATP (Giaroni et al., 2002). This study concluded that P2Y₁ receptors on SMCs are likely to mediate purinergic relaxation responses. However, a portion of the relaxation effects were suggested to occur through P2Y₂ receptors as exogenous UTP also caused relaxation.

More extensive pharmacological studies were performed on human colonic muscles and showed that ATP and the P2Y₁ receptor agonist ADPβS caused hyperpolarization of SMCs and inhibited spontaneous contractile activity (Gallego et al., 2006). MRS 2179 antagonized the inhibitory responses to ATP and ADPβS in a concentration-dependent manner (Fig. 6), and apamin inhibited the IJPs in response to EFS. Responses to EFS were completely blocked by MRS 2179, while responses to exogenous ADPβS were only partially blocked. It was suggested that bath applied transmitter analogues may affect a broader population of receptors than the receptors mediating responses to purines released from neurons. These experiments were the first support for the idea that P2Y₁ receptors mediate purinergic enteric inhibitory neurotransmission in the human GI tract. A highly selective P2Y₁ agonist, MRS 2365 caused hyperpolarization of SMCs and inhibited spontaneous contractile activity (Gallego et al., 2011). Immunohistochemical studies confirmed the presence of P2Y₁ receptors in circular and longitudinal muscle layers.

Fig. 6.

Intracellular microelectrode recordings showing IJPs induced by EFS of human colonic muscles at different stimulus intensities (5, 10, 12, 15, 17, 20, 25, 30, and 40 V) (A) shows control recordings and B–D show concentration-dependent inhibition of IJPs by MRS 2179 (1 (B), 3 (C), 5 (D) μM). Concentration-response effects are summarized in (E). Redrawn with permission from Gallego et al., 2006.

Many laboratory mammals also display purinergic IJPs and relaxation responses mediated by P2Y₁ receptors, making them useful for studies of this pathway in muscles and intact organs. For example, in guinea-pigs, pigs, rats and mice small bowels and colons, EFS elicits compound IJPs, consisting of a fast component that is blocked by P2Y₁ receptor antagonists (e.g. MRS 2179 and MRS 2500) and reduced by apamin (Mutafova-Yambolieva et al., 2007; Wang et al., 2007; Gallego et al., 2008b; Grasa et al., 2009; Zhang et al., 2010), and a 2nd component, slower to develop and longer in duration, mediated by NO (Shuttleworth et al., 1997; Gil et al., 2010). The role of P2Y₁ receptors in mediating purinergic enteric inhibitory neurotransmission was further confirmed in muscles of animals in which P2ry1 was deactivated genetically (Gallego et al., 2012; Hwang et al., 2012). Typical 2 phase IJPs were elicited by EFS in wildtype mouse colon, but IJPs were monophasic in P2ry1^−/− colons and blocked completely by antagonists of NOS (Fig. 7). Furthermore, the P2Y₁ antagonist, MRS 2500, was without effect in P2ry1^−/− mice. Hyperpolarization and relaxation responses to ATP and ADP were intact in colons of P2ry1^−/− mice. Video recording of colonic transit showed that fecal pellets were retarded in the colons of P2ry1^−/− mice. Taken together these data provide strong evidence that purinergic inhibitory responses are mediated through P2Y₁ receptors.

Fig. 7.

Spontaneous electrical activity and responses to enteric inhibitory nerve stimulation in colons from wildtype and P2ry1^−/− mice. (A–E) spontaneous action potential complexes and responses to EFS in a wildtype colon. EFS responses were blocked by TTX (not shown). (A) The action potential complexes occurred with regularity in colonic muscles. (B) Single pulses (0.5 ms; arrow) evoked an excitatory junction potential (EJP, *) and a bi-phasic inhibitory junction potential (IJP). The complex IJP consisted of a fast (fIJP) phase and a slow (sIJP) phase. (C) The EJP was blocked by atropine (1 μM). (D) L-NNA (100 μM) blocked the sIJPs and elicited depolarization and continuous action potential activity. (E) the fIJP was blocked by MRS2500 (1 μM). These observations are consistent with the responses to EFS being mediated by acetylcholine (EJP), a purine neurotransmitter (fIJP), and nitric oxide (sIJP). (F–J) Electrical activity and responses to EFS in a colonic muscle from a P2ry1^−/− mouse. (F) Action potential complexes in colons of P2ry1^−/− mice were similar to wildtype activity. (G) EFS evoked EJPs with enhanced amplitude than in wildtype muscles. The IJP following the EJP lacked the fIJP phase, but the sIJP was maintained. (H) the EJP was blocked by atropine (1 μM), and (I) the sIJP was blocked by L-NNA (100 μM). (J) After addition of atropine and L-NNA, MRS2500 had no further effect. These observations show that the fIJP in murine colons are mediated by P2Y₁ receptors. Redrawn with permission from Hwang et al., 2012.

Cellular expression of P2ry1 and response to purines released from motor neurons suggest a more complicated NMJ than the simple model proposed in many of Professor Burnstock's papers. As described above, GI smooth muscles contain ICC and PDGFRα⁺ cells that are in close proximity to the varicose processes of motor neurons and electrically coupled to SMCs via gap junctions, forming the SIP syncytium (Sanders et al., 2012; Sanders et al., 2014).

Alterations in the conductances in any of the cells of the SIP syncytium can regulate the excitability of SMCs. PDGFRα⁺ cells express SK3 channels as resolved by immunohistochemistry (Klemm and Lang, 2002; Vanderwinden et al., 2002; Fujita et al., 2003; Iino and Nojyo, 2009). Expression of SK3 was intriguing because of the possibility that fibroblast-like cells respond to purine neurotransmitters. Close inspection of immunohistochemical images reporting expression of P2Y₁ receptors in the tunica muscularis of human and animal GI muscles indicate an uneven distribution of P2Y₁-LI (Giaroni et al., 2002; Gallego et al., 2006), suggesting that either not all SMCs express these receptors or expression occurs preferentially in cells other than SMCs.

The discovery that fibroblast-like cells express PDGFRα provided a breakthrough for understanding the functions of this type of cell in GI muscles (Iino et al., 2009). We used a reporter strain of mice engineered to express a histone 2B/eGFP fusion protein driven by the endogenous promoters for Pgfra (Hamilton et al., 2003). The reporter, specific for PDGFRα⁺ cells, allowed unequivocal identification of these cells in the mixture of cells resulting from enzymatic dispersions of colonic muscles. We evaluated expression of genes that might be involved in purinergic neurotransduction and measured currents activated by purines (Kurahashi et al., 2011). PDGFRα⁺ cells express SK3 and are closely aligned with enteric motor neurons. PDGFRα⁺ cells constitute a population of cells distinct from c-Kit⁺ cells (ICC) and SMCs. A similar population of cells identified by immunolabeling with PDGFRα and SK3 antibodies was found in human colon muscles (Fig. 8) (Kurahashi et al., 2012). P2ry1 is also highly expressed in PDGFRα⁺ cells as determined by quantitative PCR (Kurahashi et al., 2011).

Fig. 8.

Immunolabeling of PDGFRα (green; A) and nNOS (red; B) in circular muscle of human sigmoid colon. (A–C) PDGFRα⁺ cells and enteric inhibitory neurons, labeled with nNOS antibody, are closely associated varicose processes (white arrows in C which shows merged files). Scale bar is 10 μm in panels A–C. Immunolabeling of PDGFRα (green; D&G) and SK3 (red; E&H). F&I are merged files. (D–F) are images from circular muscle of human transverse colon. (G–I) are images from the plane of the myenteric plexus in human transverse colon. PDGFRα⁺ cells expressed SK3-like immunoreactivity. Scale bar is 10 μm in panels D–I. Redrawn with permission from Kurahashi et al., 2012.

Using specific reporters for SMCs, ICC and PDGFRα⁺ cells allowed fluorescence activated cell sorting (FACS) to purify these populations of cells. Deep sequencing of mRNA transcripts in each type of SIP cell confirmed that P2ry1 transcripts are 16-fold and 22-fold more highly expressed in PDGFRα⁺ cells than in SMCs and ICC, respectively (Ha et al., 2017). Kcnn3 (encoding SK3 channels) was similarly highly expressed, showing 310-fold and 39-fold higher expression in PDGFRα⁺ cells than in SMCs and ICC, respectively. Thus, PDGFRα⁺ cells have the molecular apparatus to transduce purine signals and mediate hyperpolarization responses in the SIP syncytium. Further functional testing confirmed this role for these cells.

Patch clamp studies of identified single PDGFRα⁺ cells showed functional expression of a Ca²⁺-activated K⁺ conductance with the properties and pharmacology of SK3 channels (Kurahashi et al., 2011). Dialysis of cells with 500 nM Ca²⁺ caused development of large outward currents that reversed close to the equilibrium potential for K⁺ ions (E_K) (Fig. 9).

Fig. 9.

Ca²⁺-activated K⁺ currents in PDGFRα⁺ cells. (A) Voltage clamp of PDGFRα⁺ cells. Ramp potentials (−80 to +80 mV) were applied in the whole cell configuration. Pipette solutions contained: 500 nM Ca²⁺ (a), 500 nM Ca²⁺ with (300 nM) apamin in the bath (b), and <10 nM Ca²⁺ (c). Cells dialyzed with 500 nM Ca²⁺ displayed large amplitude, apamin-sensitive outward currents that reversed near E_K. The outward currents activated by high Ca²⁺ were reduced significantly by apamin (b). When cells were dialyzed with low Ca²⁺ (c) small linear currents were evoked by the ramp potentials. Averaged densities of currents evoked at 0 mV in PDGFRα⁺ cells are shown in (B). Even with low Ca²⁺ in the pipette solution, some PDGFRα⁺ cells generated STOCs; average amplitude of these events was 42 ± 13.8 pA pF⁻¹ and frequency was 3.2 ± 0.8 min⁻¹. The STOCs were blocked by apamin (300 nM). Redrawn with permission from Kurahashi et al., 2011.

This current was not available when cells were dialyzed with a pipette solution buffered to contain [Ca²⁺] = 10 nM. The outward current that developed with 500 nM Ca²⁺ dialysis was inhibited by apamin. Some cells developed spontaneous transient outward currents (STOCs) that were also blocked by apamin. Single channel recordings from off-cell patches revealed the presence of 10 pS, Ca²⁺-sensitive K⁺ channels (Fig. 10). The conductance of these channels and the Ca²⁺ sensitivity mimicked the properties of SK3 expressed in HEK293 cells (Barfod et al., 2001). ATP or β-NAD⁺ activated large amplitude transient outward currents in PDGFRα⁺ cells (i.e. 37 pA/pf average peak currents), and these responses were inhibited by apamin or MRS 2500. These findings are consistent with receptor/effector coupling between purines binding to P2Y₁ receptors and activation of SK3 channels in PDGFRα⁺ cells. ATP activated small amplitude outward currents in SMCs isolated from the same muscle preparations, and in fact ATP activated small inward currents in some cells, suggesting the expression of P2X receptors in SMCs which have been shown to be expressed in SMCs of other GI muscles (Bo et al., 2003; Lee et al., 2005; Zhang and Paterson, 2005).

Fig. 10.

PDGFRα⁺ cells express ion channels consistent with the presence of SK3 channels. (A) Excised patches were made from PDGFRα⁺ cells and single channel currents were recorded at various holding potentials (−60 to +20 mV in symmetrical KCl (140 mM). Single channel openings were observed that reversed at 0 mV. The bath solution contained 100 nM Ca²⁺. Currents at −60 mV are shown in the expanded trace with closed (c) and open (o) states noted. Amplitude histograms (not shown) at the test potentials were plotted in (B). The current amplitudes were fit with a straight line showing a single channel conductance of 10 pS. (C) Ramp potentials (−80 to 80 mV from a holding potential of −60 mV) were applied to excised patches in symmetrical KCl. Bath [Ca²⁺] (thus exposed to the intracellular surface of the patch) was varied from 10⁻⁸ to 10⁻⁵ M. (D) Ca²⁺ concentration-response curves plotted, and the data were fitted with a Boltzmann function. The EC50 for the effects of Ca²⁺ was 364 nM. Redrawn with permission from Kurahashi et al., 2011.

The question of coupling between P2Y₁ receptors and SK channels in PDGFRα⁺ cells and SMCs must be considered. P2Y₁ receptors couple through G_q/G₁₁, phospholipase Cβ and production of IP₃ and diacylglycerol (von Kügelgen and Harden, 2011). IP₃ activates Ca²⁺ release from stores which is a viable signal for activation of SK channels in cells. Coupling of IP₃ directly to inhibitory responses in SMCs is questionable. In many SMCs Ca²⁺ transients activate large-conductance Ca²⁺-activated (BK) K⁺ channels (Nelson et al., 1995; Wellman and Nelson, 2003), but there is no evidence that iberiotoxin, a highly selective antagonist of BK channels, blocks purinergic IJPs or inhibitory effects. Diacylglycerol causes phosphorylation of CPI-17 in SMCs via activation of PKC (Somlyo and Somlyo, 2003), however this would be a potent excitatory mechanism for SMC contraction, as CPI-17 inhibits myosin phosphatase and leads to enhanced phosphorylation of myosin (Somlyo and Somlyo, 2003). SK2 channels are expressed in GI SMCs (Ro et al., 2001; Klemm and Lang, 2002), but the current density from these channels activated by high concentrations of ATP is minor compared to the current density from SK3 channels in PDGFRα⁺ cells (i.e. only 1–2% of the current density in SMCs vs. PDGFRα⁺ cells; Kurahashi et al., 2011). Thus, post-junctional hyperpolarization and inhibitory responses developing directly in SMCs from release of purines from enteric inhibitory neurons are unlikely. Expression of SK channels in ICC is also very low, and responses coupled to Ca²⁺ transients in ICC are due to large amplitude inward currents mediated by Ano1 channels (Zhu et al., 2009; Zhu et al., 2015).

Does purinergic stimulation enhance Ca²⁺ release in PDGFRα⁺ cells? This question was addressed in muscles of the murine fundus, a tissue with a rich population of these cells (Baker et al., 2013). It has been hard to determine the responsiveness of PDGFRα⁺ cells in intact muscles because these cells represent less than 10% of the cells in the tunica muscularis and there have been no means of recording the specific responses of these cells. Basal levels of Ca²⁺ signaling in PDGFRα⁺ cells and responses to purinergic stimulation were investigated using mice with histone 2B/eGFP fusion protein expressed in PDGFRα⁺ cells (Baker et al., 2013). As in other regions of the GI tract, PDGFRα⁺ cells in the fundus express P2ry1 and Kcnn3. Muscles were loaded with Oregon Green 488 BAPTA-2 AM to visualize Ca²⁺ transients in PDGFRα⁺ cells identified in situ as cells with nuclei expressing eGFP. Ongoing spontaneous, stochastic Ca²⁺ transients were observed in PDGFRα⁺ cells. Addition of ATP, ADP, β-NAD⁺ and the specific P2Y₁ agonist, MRS 2365, enhanced Ca²⁺ transients (Fig. 11), and these responses were blocked by MRS 2500. The reporter (histone 2B/eGFP fusion protein) was also bred into P2ry1^−/− mice. Fundus muscles of these mice displayed no response to the P2Y₁ agonists that elicited robust Ca²⁺ transients in wildtype mice. CPA and thapsigargin blocked Ca²⁺ transients and responses to purines, demonstrating that release of Ca²⁺ from intracellular stores is responsible for Ca²⁺ transients.

Fig. 11.

Purines elicit Ca²⁺ transients in PDGFRα⁺ cells. In these experiments a reporter strain of mice (PDGFRα^{tm11(EGFP)Sor}/J) was used that utilizes the endogenous, cell-specific promoter for Pdgfra to express eGFP in cell nuclei. After loading murine fundus muscles with a Ca²⁺ sensor, in this case Oregon green 488 BAPTA-2 AM, changes in cytoplasmic Ca²⁺ were monitored selectively in cells with eGFP tagged nuclei. Spontaneous Ca²⁺ transients were observed in PDGFRα⁺ cells (A–D), however this activity was greatly enhanced by ATP (100 μM; A), ADP (100 μM; B), β-NAD (100 μM; C) or MRS 2365 (1 μM; D). The Ca²⁺ transients were phasic regular and mimicked the STOCs elicited by exogenous purines in PDGFRα⁺ cells (see Kurahashi et al., 2011). Redrawn with permission from Baker et al., 2013.

The primary response to purinergic nerve stimulation is rapid and transient hyperpolarization responses in post-junctional cells (Bennett et al., 1966; Mutafova-Yambolieva et al., 2007; Gallego et al., 2008a). SMCs are the cells typically impaled by microelectrodes in GI muscles (they are the dominant cell type present). It was thought that post-junctional responses to purine neurotransmitters were due to transduction and development of hyperpolarizing currents in SMCs. Experiments described above showed that the molecular apparatus for transducing signals via P2Y₁ receptors are expressed dominantly in PDGFRα⁺ cells rather than SMCs. Therefore, direct tests of P2Y₁ agonists were performed on isolated and identified SMCs and PDGFRα⁺ cells from murine colon (Kurahashi et al., 2014). Application of purines to isolated cells under current clamp conditions elicited large amplitude hyperpolarization responses in PDGFRα⁺ cells and either no response or small hyperpolarization responses in SMCs (Fig. 12). Responses to putative neurotransmitters (ATP, ADP, β-NAD⁺ and ADPR; section on Novel purines that act as enteric motor neurotransmitters) were the same as with MRS 2365, with the exception that ATP (10 μM) elicited depolarization responses in SMCs. Hyperpolarization responses of PDGFRα⁺ cells were inhibited by MRS 2500 and by UCL 1684 (an SK channel antagonist; Malik-Hall et al., 2000; Wulff and Zhorov, 2008). An SK2/SK3 agonist, CyPPA (Hougaard et al., 2007) caused hyperpolarization of PDGFRα⁺ cells to E_K. CyPPA caused only small hyperpolarization responses in SMCs, consistent with the low current density from SK channels in these cells (Kurahashi et al., 2014). These data demonstrate that SMCs are unlikely to mediate the signature fast, transient hyperpolarization responses (IJPs) elicited by stimulation of purinergic neurons in intact muscles. The SIP cells responsible for purinergic hyperpolarization responses (IJPs) are PDGFRα⁺ cells. Purinergic IJPs are conveyed to SMCs via gap junctions between SMCs and PDGFRα⁺ cells.

Fig. 12.

Purines cause hyperpolarization of colonic PDGFRα⁺ cells but little to no effect on SMCs. (A) shows a patch clamp, whole-cell recording in current clamp mode from a PDGFRα⁺ cell (I = 0). MRS2365 (100 nM) caused transient hyperpolarization of PDGFRα⁺ cells that reached a peak of about −80 mV (≌ E_K). The hyperpolarization response was blocked by MRS 2500 (1 μM). The effects of MRS 2365 recovered when MRS 2500 was washed out. (B) MRS 2365 (1 μM) did not affect the membrane potentials of SMCs isolated from the same colonic muscles and studied under the same experimental conditions. (C) shows a summary of the effects of MRS 2500 on hyperpolarizations elicited in PDGFRα⁺ cells by MRS 2365 (n = 5; *P = 0.0321). In the original study from which this experiment was reproduced, several additional purines, including β-NAD⁺) had similar effects on PDGFRα⁺ cells and little or no effect on SMCs. Redrawn with permission from Kurahashi et al., 2014.

Imaging experiments were also performed on colonic muscles loaded with Oregon Green 488 BAPTA-2 AM to investigate the sequence of events activated by purinergic nerve stimulation (Baker et al., 2015). The animals used for these studies expressed histone 2B/eGFP in cell nuclei of PDGFRα⁺ cells so it was possible to monitor Ca²⁺ events unequivocally in these cells. Spontaneous Ca²⁺ transients were observed in PDGFRα⁺ cells, and as in the gastric fundus (see above) purines stimulated Ca²⁺ oscillations. Responses to purines were inhibited by MRS-2500. In muscles with nitrergic and cholinergic neural inputs blocked, EFS (1 pulse and 5–20 Hz pulses for 1 s) evoked an immediate increase in Ca²⁺ transients in nerve bundles, and these events were followed by development of Ca²⁺ transients in PDGFRα⁺ cells after a 280 ms latency from the initiation of EFS (Fig. 13). SMCs responded initially with a small decrease in Ca²⁺ and Ca²⁺ only increased upon termination of EFS. The latency between initiation of EFS and the onset of the rise in Ca²⁺ in SMCs was 1941 ms. Ca²⁺ transients evoked in nerve fibers by EFS were unchanged after addition of MRS 2500, but this compound blocked postjunctional responses in PDGFRα⁺ cells and SMCs (Fig. 13). Postjunctional responses were absent in PDGFRα⁺ cells and SMCs of P2y1r^−/− mice, and responses were blocked in SMCs but retained in PDGFRα⁺ cells after treatment of muscles with the gap junction blockers, 18-β glycyrrhetinic acid or octanol. These data show that postjunctional activation of P2Y₁ receptors, triggered by the release of purines from enteric neurons, causes enhanced Ca²⁺ transients in PDGFRα⁺ cells and reduction in Ca²⁺ in SMCs during nerve stimulation. A large ‘rebound’ increase in Ca²⁺ occurs in SMCs upon cessation of stimulation. This sequence of events depends upon gap junction coupling between PDGFRα⁺ cells and SMCs. Fig. 5 summarizes the cells, receptors, ion channels and metabolic pathways responsible for purinergic and nitrergic inhibitory neurotransmission in the SIP syncytium.

Fig. 13.

Temporal sequence of Ca²⁺ transients in nerve bundles, PDGFRα⁺ cells and SMCs. (A) Raw images of cells in PDGFRα^{tm11(EGFP)Sor}/J mice. Muscles were loaded with Oregon green 488 BAPTA-2 AM. PDGFRα⁺ cells displaying bright nuclei are denoted by white arrows. EFS (10 Hz) evoked Ca²⁺ transients first in nerve fibers (NF) and then, after a brief delay, in PDGFRα⁺ cells. (B) Response elicited in PDGFRα⁺ cells after firing of NF were inhibited by MRS-2500 (1 μM). (C) Time sequence images showing sequence of Ca²⁺ transients in NF and PDGFRα⁺ cells evoked by EFS (10 Hz) before (C) and in the presence of MRS-2500 (1 μM; E). L-NNA (100 μM) and atropine (1 μM) were present during all recordings. (D&F) show an expanded trace of the responses in C&E. EFS failed to evoke Ca²⁺ transients in PDGFRα⁺ cells and SMCs in the presence of MRS-2500, but responses in NF were not affected by MRS-2500. (D) Ca²⁺ responses in SMCs were delayed until after cessation of EFS. A slight drop in Ca²⁺ was noted during and shortly after EFS (region below dotted line in D). EFS was delivered in 1 s trains (denoted by the grey box and the dotted lines through the traces). Redrawn with permission from Baker et al., 2015.

10. Novel purines that act as enteric motor neurotransmitters

In the original paper that introduced the concept of purinergic neurotransmission, it was acknowledged that the novel transmitter could be ATP or a related nucleotide (Burnstock et al., 1970). The exact language at the conclusion of the paper was: “In fact the evidence presented would hold equally well for AMP, ADP and probably other nucleotides. There is no valid reason for choosing between these nucleotides, but we favor ATP and ADP in view of their high inhibitory potency.” This classic study used the criteria summarized by Eccles to determine whether a substance was a neurotransmitter (Eccles, 1964). As discussed above, the most significant limitation of the early studies was the lack of reliable methods for accurate identification of the purines released during stimulation of intrinsic motor neurons.

As inhibitory neurotransmission in GI muscles was investigated in more detail, some observations were not strictly compatible with the concept that ATP was the neurotransmitter. Therefore, many investigators referred to the purinergic neurotransmitter either as purine-like or a related purine (De Man et al., 2003; Serio et al., 2003; Gallego et al., 2006; Gallego et al., 2008a; Mutafova-Yambolieva and Durnin, 2014). For example, drugs that blocked P2Y₁ receptors blocked purinergic IJPs in murine colon (Gallego et al., 2006; Mutafova-Yambolieva et al., 2007; Wang et al., 2007; Grasa et al., 2009), but MRS2179 did not block hyperpolarization responses to ATP applied near the site of intracellular recording by a spritz pipette (Mutafova-Yambolieva et al., 2007). β-NAD⁺ is also released in murine colonic muscles during EFS, and it comes from intrinsic neurons because its release was blocked by TTX and ω-conotoxin GIVA (Mutafova-Yambolieva et al., 2007). Purinergic IJPs are also blocked by ω-conotoxin GIVA in colonic muscles (Bridgewater et al., 1995). β-NAD⁺ caused hyperpolarization of colonic muscles, and these responses, like IJPs, were inhibited by MRS2179. The amount of β-NAD⁺ released by EFS increased as a function of stimulation frequency, however ATP release did not increase with frequency. ATP release was also less sensitive than β-NAD⁺ to TTX and ω-conotoxin GIVA. The release of β-NAD⁺ was more consistent with what is known about mechanisms controlling neurotransmitter release.

β-NAD⁺ was also found to be a novel agonist for P2Y₁ receptors expressed in HEK293 cells, and responses to β-NAD⁺ were blocked by MRS2179. In intact colonic muscles purinergic neural input caused inhibition of spontaneous action potentials and contractions. β-NAD⁺ mimicked these responses, and the effects of β-NAD⁺ were not affected by pre-treatment with a NOS antagonist (L-NNA). Thus, β-NAD⁺ does not activate inhibitory responses by pre-junctional stimulation of NO release. Taken together β-NAD⁺ met the key presynaptic and postsynaptic criteria set out by Eccles (1964): i) β-NAD⁺ and the enzymatic apparatus to synthesize this compound is universally expressed in cells. ii) β-NAD⁺ is released in a frequency-dependent manner by stimulation of intrinsic enteric inhibitory neurons. Its release is sensitive to Na⁺ and Ca²⁺ channel antagonists known to block purinergic neurotransmission; iii) β-NAD⁺ is an agonist for P2Y₁ receptors that are known to mediate post-junctional effects of purinergic nerve stimulation; iv) metabolic mechanisms are expressed and functional in colonic muscles that deactivate β-NAD⁺ (Durnin et al., 2020); v) receptor antagonists that block IJPs also inhibit responses to β-NAD⁺. Importantly these experiments suggest that ATP is less likely to be the purinergic inhibitory motor neurotransmitter because hyperpolarization responses to spritz application of this compound were not blocked by MRS2179, however this compound was effective in blocking IJPs. Furthermore, the release of ATP in tissues did not display frequency-dependence or sensitivity to the neurotoxins tested. Such findings suggest that ATP might be released from sites that are different from the sites of neurotransmitter release (discussed below).

The anatomical components of purinergic neurotransmission, enteric inhibitory motor neurons and PDGFRα⁺ cells, are present in human and non-human primate (Cynomolgus monkeys) GI muscles (Blair et al., 2012; Kurahashi et al., 2012). Thus, the role of β-NAD⁺ as a purinergic inhibitory neurotransmitter was also evaluated in human and monkey colonic muscles (Hwang et al., 2011). ATP and β-NAD⁺ and several metabolites were measured in tissue superfusates. β-NAD⁺ release increased in a frequency-dependent manner with EFS, and similar observations were made with circular muscle strips after removal of the myenteric plexus. β-NAD⁺ was the dominant purine released in response to nerve stimulation (15–25 fold higher than ATP). Release of ATP did not increase as a function of EFS frequency in either whole muscles or isolated circular muscle. β-NAD⁺ release was blocked by TTX and ω-conotoxin GIVA, however ATP release was not significantly affected by these neurotoxins (Fig. 14). Large amplitude IJPs were elicited by EFS in monkey and human colons, and these responses were blocked by PPADS, MRS2179 or MRS2500. Spritz application of equal concentrations of putative neurotransmitters, ATP and β-NAD⁺, caused transient hyperpolarizations in colons of both species. Responses to β-NAD⁺ were blocked in human and monkey colons by PPADS or MRS2179, however the hyperpolarizations elicited by ATP were not blocked by these antagonists. ATP and β-NAD⁺ evoked inward currents in isolated human and monkey SMCs, suggesting that SMCs could not be responsible for the hyperpolarization responses to purinergic neurotransmitter release or application of endogenous purines in intact muscles. As in murine colon, β-NAD⁺ is a better candidate for purinergic neurotransmission than ATP in human and non-human primate colon.

Fig. 14.

Neuronal release of purines in human colon tunica muscularis. Panel A shows original chromatograms of tissue superfusate samples collected during electrical field stimulation at 4 and 16 Hz, 0.1 ms, for 30 s, in the absence and presence of the fast Na⁺ channel blocker tetrodotoxin (TTX, 0.5 μM) and the Ca_V2.2 channel inhibitor ω-conotoxin GVIA (ω-Ctx 50 nM). LU, luminescence units. B and C show summarized data (means ± SEM) of EFS-evoked neurotransmitters release; number of experiments in parentheses. (o) denote significant difference from 4 Hz controls (P < 0.05); asterisks denote significant differences from 16 Hz controls (* < 0.05, **P < 0.01). Note that the EFS-evoked release of β-NAD⁺, but not of ATP, increased with stimulation frequency. The release of β-NAD⁺ at 16 Hz, but not of ATP, was significantly reduced by both TTX and ω-conotoxin GVIA. D indicates that β-NAD⁺ is the primary purine in the mixture β-NAD⁺ + ADPR + cADPR. E shows that the amounts of β-NAD⁺ released by EFS exceeded significantly the amounts of ATP in the same samples collected during EFS. Redrawn with permission from Hwang et al., 2011.

It is difficult to discriminate which neurons in GI muscles are stimulated by EFS, as both cell bodies in ganglia and neural processes might be activated by current passing through the tissue. Therefore, the source of purines released by stimulation of neurons in mouse and monkey colonic muscles was investigated by stimulating nicotinic receptors (nAChRs) and 5-HT₃ receptors (5-HT₃Rs), which are expressed on cell bodies of enteric neurons (Zhou and Galligan, 1999; Galligan, 2002a; Galligan, 2002b; Mazzia et al., 2003; Gershon, 2004; Kapeller et al., 2011). Epibatidine and 1,1-dimethyl-4-phenylpiperazinium iodide (DMPP), used to stimulate nAChRs, and SR57227, used to stimulate 5-HT₃Rs, enhanced release of ATP, β-NAD⁺ and metabolites ADP, AMP, and adenosine (ADO). Release of β-NAD⁺ was reduced by TTX or by ω-conotoxin GIVA, but release of ATP was unaffected by these neurotoxins (Fig. 15). Epibatidine and SR57227 failed to evoke release of ATP or β-NAD⁺ in circular muscles of monkey colon devoid of myenteric plexus. These observations demonstrate that release of β-NAD⁺ requires information transfer from ganglia to nerve terminals using axonal Na⁺ action potentials and Ca²⁺-dependent neurotransmitter release mechanisms. However, release of ATP did not depend upon axonal and neurotransmitter release mechanisms and suggested that the ATP release in response to nerve activation occurs in ganglia, perhaps from cell bodies of neurons or glia, while β-NAD⁺ is released from varicosities of motor neurons that innervate the circular muscle layer (Fig. 15).

Fig. 15.

Stimulation of nicotinic acetylcholine receptors on neuronal cell bodies causes release of purines in monkey colon tunica muscularis. (A) shows original chromatograms of tissue superfusates collected before treatment with epibatidine (Control), during stimulation with epibatidine alone (Epib, 500 μM), epibatidine in preparations pretreated for 30 min with 500 μM hexamethonium (Hex + Epib), with 0.5 μM tetrodotoxin (TTX + Epib), and with 50 nM ω-conotoxin GVIA (ω-Ctx + Epib). Epibatidine elicited release of ATP, ADP, β-NAD⁺, AMP, and adenosine (ADO). (B&C) show averaged data (means ± SEM) for ATP and β-NAD⁺. Asterisks denote significant differences from epibatidine-evoked release (*P < 0.05, **P < 0.01, ***P < 0.001); number of experiments in parenthesis. The ganglionic blocker hexamethonium reduced the epibatidine-evoked release of purines. The neuronal blockers TTX and ω-Ctx inhibited the epibatidine-evoked release of β-NAD⁺ but not of ATP. (D) shows model of purine release from ganglia and within the tunica muscularis. Stimulation of ganglionic receptors nAChR (data shown in this figure) or 5HT3R (data shown in original publication) activate action potentials (AP) in axons of motor neurons that innervate the circular muscle. Stimulation with either EPI (nAChR agonist) or SR57227 (5HT3 agonist) cause release of ATP and β-NAD⁺. TTX, which inhibits action potential blocks release of β-NAD⁺ but did not significantly reduce the release of ATP. These data suggest that ATP is released primarily from ganglionic sources not dependent upon Na⁺ action potentials and β-NAD⁺ is released from nerve terminals of motor neurons. Redrawn with permission from Durnin et al., 2013.

Two additional substances, measured in tissue superfusates and in response to nerve stimulation, are adenosine diphosphate ribose (ADPR) (Durnin et al., 2012) and uridine adenosine tetraphosphate (Up4A) (Durnin et al., 2014). ADPR is a primary metabolite of β-NAD⁺ (Munshi et al., 2000; Graeff et al., 2009), the major purine released in a frequency-dependent and TTX-sensitive manner by stimulation of enteric neurons. ADPR is itself a ligand for P2Y₁ receptors (Gustafsson et al., 2011). Therefore, the potential for ADPR as a participant in purinergic neurotransmission must be considered. ADPR accounted for about 20–30% of the single peak that was generated by β-NAD⁺, ADPR and cyclic ADPR in superfusates of monkey colon (Fig. 14, Hwang et al., 2011). β-NAD⁺ is rapidly metabolized to ADPR when added to colonic muscles of monkey and mouse or circular muscles from monkey devoid of the myenteric plexus (Durnin et al., 2012). ADPR is further metabolized to ADO in monkey and mouse colons. Having shown that ADPR is present in tissue superfusates and β-NAD⁺ is converted to ADPR, the effects of ADPR on membrane potentials in the proximal colon were tested. ADPR, applied by spritz pipette, caused transient hyperpolarization responses that were inhibited by MRS2500 and apamin. These data suggest that ADPR might participate in purinergic post-junctional responses in colonic muscles. The results of this study were unable to determine whether ADPR is a primary neurotransmitter (i.e. stored and released from purinergic vesicles) or whether it is produced only by metabolism of β-NAD⁺ after it is released from motor neurons.

Up4A is another interesting endogenous substance that appears to play a role in enteric neurotransmission. Up4A was originally described as an endothelium-derived vasoconstricting factor (Jankowski et al., 2005) that can also cause vasorelaxation by activation of P2Y₁ receptors (Tölle et al., 2010). We found that Up4A is also released during nerve stimulation of human and mouse colonic muscles (Durnin et al., 2014). Release of Up4A is inhibited by TTX, suggesting that at least a portion of the released compound comes from neurons. Up4A caused hyperpolarization responses upon spritz application and inhibition of contractions with bath application in proximal and distal colonic muscles. These responses were not affected by atropine or L-NNA but were inhibited by apamin or MRS2500, suggesting that Up4A meets the post-junctional criteria for the purinergic neurotransmitter. Actually, Up4A is a more potent agonist for P2Y₁ receptors than ATP, ADP, β-NAD⁺ or ADPR. Actions of Up4A on P2Y₁ receptors were confirmed by showing that Ca²⁺ responses were evoked by Up4A and blocked by MRS2500 in 1321N1 astrocytoma cells expressing human P2Y₁ receptors (Durnin et al., 2014). Up4A also caused hyperpolarization of isolated PDGFRα⁺ cells under current-clamp and activated large outward currents that reversed near E_K under voltage-clamp. These responses were inhibited by MRS2500 and UCL1684, an SK channel antagonist. Up4A was metabolized efficiently in both human and mouse colons. The presence, release and efficacy of Up4A make it likely that this substance also contributes to purinergic regulation of colonic motility.

In summary, experiments using a variety of techniques have cast doubt that ATP is a primary inhibitory neurotransmitter in GI muscles. In fact the data suggest that enteric inhibitory purinergic neurotransmission is mediated by multiple purines (Mutafova-Yambolieva and Durnin, 2014) that bind to post-junctional P2Y₁ receptors expressed by PDGFRα⁺ cells. Perhaps the existence of multiple purine neurotransmitters represents backup mechanisms to support enteric inhibitory purinergic neurotransmission, underscoring its importance for regulation of intestinal motility.

11. Examples of the role of purines in GI motility

An extensive summary of the known and/or hypothesized roles for purines in GI motility will not be provided here, as these topics are the foci of contributions from Professors Costa and Galligan in this monograph. A curious topic is why enteric inhibitory neurons employ multiple substances that generate post-junctional inhibition of SMC excitability and contraction: e.g. purines that activate P2Y₁ receptors in PDGFRα⁺ cells, NO that utilizes soluble guanylyl cyclase in ICC and SMCs, and inhibitory peptides, VIP and PACAP that bind potentially to multiple receptors and activate cAMP-dependent mechanisms in SIP cells. The redundancy in inhibitory transmitters seems to underscore the importance of enteric neural inhibition for generation of GI motility patterns, and multiple neurotransmitters working though different receptors and transduction mechanisms expressed in different cells of the SIP syncytium may provide a safety factor for inhibitory regulation. It is interesting that the parallel pathways providing such a safety factor are due to expression of receptors and transduction mechanisms in different post-junctional cells rather than by multiple neurons each releasing a unique neurotransmitter. This arrangement would appear to offer protection against lesions in SIP cells and not defend quite so effectively against neuropathies. Of course, it is possible that enteric neuropathies might include lesions in production, storage or release of specific types of neurotransmitters rather than loss of an entire group of enteric neurons, which would presumably reduce the availability of all of the redundant inhibitory motor neurotransmitters.

The significance of redundant inhibitory neurotransmission was investigated using pharmacological techniques to block purinergic (apamin) or nitrergic (nitric oxide synthase inhibitors) neurotransmission in the guinea-pig small intestine (Waterman and Costa, 1994). Blocking both inhibitory pathways disrupted peristalsis, but blocking either the purinergic or nitrergic pathways independently altered, but did not block, peristalsis. The effects of blocking the purinergic pathway in the small intestine were fairly minor. For example, there was no change in the threshold pressure required for initiation of peristalsis, but the maximum pressures during small bowel emptying were increased. As described above, purinergic inhibitory responses were abolished in mice with genetic deactivation of P2ry1 (Gallego et al., 2012; Hwang et al., 2012). We tested the effects of the lesion in P2Y₁ receptors on colonic transit to better understand how loss of a portion of inhibitory regulation might affect colonic motility. Colonic transit was characterized by measuring the movement of artificial fecal pellets by digital video recording in ex vivo colons (Hwang et al., 2012). Normal rates of pellet movement in wildtype mice averaged 1.12 mm/s. Colonic transport was significantly delayed in the colons of P2ry1^−/− mice. The pellets were transported to a point about 60% along the length of the colon and then transport was stopped or delayed for a period of time that exceeded recording sessions. Similar observations were made when MRS 2500 was added to the solution bathing the colons. In spite of the delay in colonic transit in ex vivo colons, P2ry1^−/− mice survive, grow into adulthood and have normal body weights and gross appearance. Thus, additional mechanisms, such as sympathetic input to the distal colon (Kurahashi et al., 2020a; Kurahashi et al., 2020b), may support colonic transit in the intact animal.

Purine neurotransmission also appears to be under pre-junctional regulation via NO which may be a mechanism of restraining inhibitory input. NO inhibited the release of purine neurotransmitters (i.e., β-NAD⁺/ADPR, Up4A, and ATP) in human, monkey and mouse colons via activation of soluble guanylyl cyclase (sGC) and subsequent activation of cGMP-dependent kinase I (cGK1/PRKG1) (Durnin et al., 2017). Neural release of purines was decreased by a NO donor (i.e., SNAP) and was increased by a nitric oxide synthase (NOS) inhibitor (i.e., L-NNA), by a sGC inhibitor (i.e., 1H-[1,2,4]oxadiazolo[4,3-a]quinoxalin-1-one, ODQ), and in colonic muscularis of mice lacking Nos1 and Prkg1 genes. Therefore, inhibition of purine release by NO may act to prevent overinhibition of colonic motility when the activity of inhibitory motor neurons is abnormally high. We also found that purine neurotransmitter release was increased in colons of W/W^V mice that have compromised nitrergic neurotransmission (Sanders et al., 2010; Durnin et al., 2017). Thus, purinergic neurotransmission may be enhanced when nitrergic neurotransmission is deficient to preserve the important functions of enteric inhibitory neurotransmission in GI motility.

Enteric neuroplasticity leading to deleterious effects on motility can occur in GI disease models (Spear and Mawe, 2019). For example, colitis induced by treatments of guinea-pig colon with trinitrobenzene sulfonic acid (TNBS) results in selective decrease in the amplitude and frequency of purinergic IJPs (Strong et al., 2010); nitrergic IJPs were not inhibited. This effect was not due to neural damage, as changes in nerve fiber density were not detected, and responses to exogenous purines were also unaffected. These observations suggested that defects in purinergic neurotransmission may result from reduced purine release from enteric inhibitory neurons or abnormal local metabolism of purine neurotransmitters. As with P2ry1^−/− mice, described above, MRS2179 retarded colonic transit in inflamed guinea-pig colon. In a later study inhibition of purinergic IJPs in TNBS colitis was linked to oxidative stress, a condition reported in inflammatory models of the colon (Shi et al., 2011). The effects of inflammation on purinergic IJPs could be simulated by addition of H₂O₂ or by treatment with ATP synthase inhibitors, oligomycin or dicyclohexylcarbodiimide (Roberts et al., 2013). Consistent with the hypothesis that the reduction in IJPs in TNBS colitis was due to reduced release of purines, the release of ATP, ADP and β-NAD⁺, as well as total purines, was reduced in TNBS-inflamed colonic muscles. IJPs were rescued by addition of a free radical scavenger, tempol. Colonic motility, reduced in TNBS colitis, was also improved by treatment with tempol. These studies offer an excellent example of how enteric neuropathies can be neurotransmitter specific, and are not necessarily due to loss of a whole population of enteric neurons.

12. Conclusions and needs for further research on purines in the gut

As discussed above, there is still no suitable biomarker to distinguish purinergic neurons definitively in the enteric nervous system. In spite of some success claimed with using antibodies to VNUT (Chaudhury et al., 2012), this approach has not been applied widely. Expression of Slc17a9 (VNUT) was low in the unsorted cell populations resulting from enzymatic dispersions of murine colonic muscle (Lee et al., 2015; Lee et al., 2017). The same populations of cells showed high expression of Vip, suggesting that cell bodies of inhibitory neurons were present in this population of cells. Low success with immunostaining for VNUT also tends to suggest that this transporter is not highly utilized in purinergic neurons in the GI tract. Another transporter is possibly responsible for storage of β-NAD⁺, ADPR or Up4A, and no protein with these properties has been identified. Finding a reliable biomarker would be a useful means of clarifying definitively whether enteric inhibitory neurons are a single population of enteric motor neurons.

Another area of fundamental importance is development of technologies for precise evaluation of release of purines from neurons and other cellular sources in complex tissues and from subcellular organelles such as synaptic vesicles. HPLC is a powerful means of assaying purines released in smooth muscle preparations, as discussed. However, the field of purinergic neurotransmission is in need of new technologies to measure purine neurotransmitters with high sensitivity, specificity and spatial and temporal resolution. Development of such methodologies will enable study of detailed mechanisms of neurotransmitter release, compartmentalized metabolism of extracellular purines, and mechanisms of presynaptic neuromodulation. Such techniques might facilitate discoveries of differential mechanisms of release of various inhibitory neurotransmitters from the same neurons at different levels of neural activation. As discussed in Methods of measuring purines released from neurons, genetically-encoded biosensors for detection of ATP and β-NAD⁺ in subcellular compartments such as the nucleus or mitochondria have been developed. However, no methodology exists for measuring these putative neurotransmitters in synaptic vesicles. Detection of purines in synaptic vesicles would be a major advancement that would reinforce the notion that key intracellular constituents can function as extracellular mediators. Resolving the complexities of prejunctional mechanisms of enteric inhibitory neurotransmission could lead to fundamental advancements in physiology or pathophysiology of the peripheral neuroeffector junction.

Presynaptic neuromodulation is another area that deserves further in-depth studies. Enteric motor neurons express receptors and membrane channels for endogenous substances (Lay et al., 2016; Fichna et al., 2015) that alter neurotransmitter release and concentration profiles of neurotransmitters in the NEJ. Very little is known about mechanisms of presynaptic neuromodulation of purine neurotransmitter release in the gut. It is an open question whether aberrant presynaptic neuromodulation underlies motility dysfunction in inflammation or diabetes.

An aspect of co-transmission that may be important in developing a more extensive understanding of enteric neuropathies is that different Ca²⁺ conductances have been linked to the release mechanisms of different neurotransmitters. Reduced expression of one of these conductances or its subunits and support proteins, could cause reduced neural output of one neurotransmitter without affecting the release of others. There have been reports that Ca²⁺ entry via N-type (Ca_V2.2), P/Q type (Ca_V2.1) and R-type (Ca_V2.3) channels can be linked to neurotransmitter release in enteric neurons (Takahashi et al., 1992; Bridgewater et al., 1995; Borderies et al., 1996; Reis et al., 2000; Bian and Galligan, 2007; Rodriguez-Tapia et al., 2017). Two phases of IJPs in guinea-pig taenia caeci were differentially blocked by Ca²⁺ channel antagonists; the fast phase, due to purinergic neurotransmission, was blocked by ω-conotoxin GVIA, whereas the secondary, slower peak, due to NO, was insensitive to ω-conotoxin GVIA and blocked by Cd²⁺, a nonspecific Ca²⁺ channel antagonist (Bridgewater et al., 1995). Immunohistochemistry showed that the α1E subunit of Ca_V2.3 channels is expressed in guinea-pig ileal longitudinal muscle motor neurons and in mouse colonic, and immunoreactivity co-localized with nNOS immunoreactivity (Rodriguez-Tapia et al., 2017). α1E-like immunoreactivity was not resolved in α1E^−/− mice. Electrophysiological and contractile recordings in this study suggested that T-type Ca²⁺ channels regulate nitrergic inhibitory responses. We observed differential regulation of neurotransmitter release by different neuronal Ca²⁺ channels in blood vessels (Smyth et al., 2009). For example, the release of norepinephrine was mediated by Ca_V2.2 channels, the release of ATP was mediated by the Ca_V2.1 channel, and the release of β-NAD⁺ was mediated by both Ca_V2.2 and Ca_V2.1 channels. It remains to be elucidated whether different neuronal Ca²⁺ channels are associated with different sites of neurotransmitter release or different populations of synaptic vesicles. Using selective expression of fluorescent reporters in specific types of enteric neurons and FACS to purify the neurons expressing the reporters, it may be possible to dissect the Ca²⁺ channels expressed in inhibitory motor neurons and show how various Ca²⁺ channel antagonists affect Ca²⁺ transients in nerve varicosities using optogenetic approaches. Evaluations of neurons from GI motility disorders may reveal that Ca²⁺ transients due to specific Ca²⁺ conductance may be affected, thus selectively reducing nitrergic, purinergic or peptidergic neurotransmitter release. A great deal has been published about the loss of nitrergic regulation in diabetes (Takahashi et al., 1997; Sanders and Ward, 2019). At least part of this defect is attributed to reduced expression of nNOS, however NO is thought to be released from the same neurons that release purines. Thus, whether diabetes leads to selective reduction in nitrergic responses, or whether purinergic neurotransmission is also compromised in diabetes is an important question. It is also possible that defects in NO release could be linked to changes in the expression and function of the specific Ca²⁺ entry mechanism controlling NO synthesis.

Relatively high expression of P2ry14 was observed in isolated PDGFRα cells in colon and jejunum (Ha et al., 2017). Receptors encoded by P2ry14 are activated by UDP-glucose and other sugar-nucleotides (Chambers et al., 2000; Abbracchio et al., 2003). These receptors have structural similarities with the P2Y receptor family, but GI motor responses mediated by these receptors are unknown. Pertussis toxin blocking the responses of P2Y₁₄ suggests that coupling occurs through G_i/o (Wittenberger et al., 2001). It is unclear whether a source of UDP-glucose (or other sugar-nucleotide compounds that might be agonists) exists in the tunica muscularis of the gut. But cellular damage is known to release nucleotides, and compounds like UDP-glucose and ATP have been suggested to constitute damage associated molecular pattern (DAMPs) molecules (Elliott et al., 2009). P2Y₁₄ receptors and DAMPs (Chen and Nuñez, 2010) have been associated with the development of a pro-inflammatory response, called sterile inflammation (Azroyan et al., 2015), and techniques described in this study could be used to investigate whether inflammatory effects in the GI tract are also initiated through generation of DAMPs and activation of P2Y₁₄ receptors. Another suggestion has been that P2ry14 might be a useful biomarker for high-risk gastrointestinal stromal tumors (Jin et al., 2018).

P2x7r is another purinergic receptor gene that shows relatively high expression in colon muscularis and in cells of the SIP syncytium (Breland et al., 2019). The P2X₇ receptor is a trimeric ligand-gated ion channel that is activated by binding of unusually high concentrations of ATP (EC50 ≥ 1 mM) (Karasawa and Kawate, 2016). On the other hand, low micromolar concentrations of β-NAD⁺ activate P2X₇ by ART-dependent ADP-ribosylation of the receptor protein. Mechanisms of NAD-induced ADP-ribosylation of P2X₇ receptor are mostly studied in immune cells of mice (Seman et al., 2004). Given that β-NAD⁺ is present in the extracellular space in colonic tunica muscularis, it would be of interest to determine whether such mechanisms exist in the gut. Immunohisto-chemistry studies have shown robust expression of P2X₇ receptors in the myenteric plexus and glial cells of GI muscles (Vanderwinden et al., 2003; Gulbransen et al., 2012; Mendes et al., 2019). Increased activation of P2X₇ receptors by ATP in enteric neurons (Gulbransen et al., 2012) and in mast cells (Kurashima et al., 2012) has been demonstrated in models of colitis. A role for P2X₇ receptors in septic and sterile inflammation has been suggested (Adinolfi et al., 2018). However, it is currently unclear whether extracellular β-NAD⁺ can also activate this receptor under physiological or pathophysiological conditions, and what role P2X₇ receptors might have in cells of the SIP syncytium is unknown.

In GI muscles, single, bare varicose nerve fibers are rare, as processes of motor neurons coursing through GI muscles are arranged in multiaxonal bundles ensheathed by enteric glia (Bałuk and Gabella, 1987; Gabella, 1987). Innervation is thought to occur at points where varicosities are not covered by the glial sheath, but at many points along axons and at sites of bared varicosities glial cells are in close association with presumed sites of neurotransmitter release. There has been considerable interest lately in the role of glial cells in regulating GI motility and responses to inflammatory mediators (Gulbransen and Sharkey, 2009; Gulbransen and Sharkey, 2012; Neunlist et al., 2014; McClain et al., 2015). Most studies of enteric glia have focused on the glial network surrounding ganglia, and these cells have dramatic intracellular Ca²⁺ responses to substances, including purines, released in the ganglia (Gulbransen and Sharkey, 2009). Application of ATP or activation of interganglion nerve fibers activate Ca²⁺ transients in glial cells that are blocked by the non-selective P2Y receptor antagonist, PPADS. P2Y₁ receptors are found in the myenteric plexus of mouse ileum (Cerantola et al., 2020) and appear to be present in a subpopulation of glial cells in myenteric ganglia (Brown et al., 2016). Activation of P2Y₁ receptors with ADP and ADPβS elicited Ca²⁺ transients in glial cells in myenteric ganglia of human jejunum and mouse colon (Brown et al., 2016; McClain et al., 2014). If P2Y receptors are also expressed by glial cells enveloping nerve fiber bundles, then it is possible that there is interplay between neurons and glia at or near points of motor innervation. Whether activated glia modulate enteric motor neurotransmission is a question that does not appear to have been addressed.

A final very important question for future investigation is the effects of inflammation on purinergic motor neurotransmission. The cells that mediate post-junctional purinergic motor responses, PDGFRα⁺ cells, were originally considered to be fibroblasts, and therefore might be considered a class of innate immune cells within the wall of the gut, and might respond to cytokines by release of prostaglandins, additional cytokines, enhanced collagen formation and proteases (Newton and Dixit, 2012). How immune factors might affect responses of PDGFRα⁺ cells, enteric inhibitory nerve-PDGFRα⁺ cell interactions and electrical connectivity between PDGFRα⁺ cells and SMCs are all unanswered questions. Furthermore, inadequate levels of extracellular purines and aberrant purinergic receptor signaling have been associated with excessive inflammation or fibrosis in many systems (Idzko et al., 2014), including the GI tract (Vuerich et al., 2019). In broad terms, ATP and ADP are considered to be pro-inflammatory mediators whereas adenosine is largely considered as having immunosuppressive potential. In this context, strategies for treatment of inflammation might promote extracellular conversion of purines to adenosine and activation of adenosine receptors. As discussed above, PDGFRα⁺ cells in the colon express high transcripts of Adora1, the A1 adenosine receptor gene (Breland et al., 2019). However, the role of this receptor in PDGFRα⁺ cells is unresolved. Deletion of Entpd1, Entpd2 and Entpd3 (Feldbrügge et al., 2017) or Entpd8 (Salem et al., 2021) exacerbates the acute phase of dextran sodium sulfate (DSS)-induced colitis in mice. Entpd2^−/− mice have reduced density of myenteric neurons and reduced relaxations of colonic circular smooth muscle (Grubišić et al., 2019). Cell-specific catabolism of purines has been suggested (Grubišić et al., 2019; Durnin et al., 2020), but whether this is associated with dysregulation of GI motility during inflammation is currently unknown.