Neurally Released GABA Acts via GABAC Receptors to Modulate Ca2+ Transients Evoked by Trains of Synaptic Inputs, but Not Responses Evoked by Single Stimuli, in Myenteric Neurons of Mouse Ileum

γ-Aminobutyric Acid (GABA) and its receptors, GABAA,B,C, are expressed in several locations along the gastrointestinal tract. Nevertheless, a role for GABA in enteric synaptic transmission remains elusive. In this study, we characterized the expression and function of GABA in the myenteric plexus of the mouse ileum. About 8% of all myenteric neurons were found to be GABA-immunoreactive (GABA+) including some Calretinin+ and some neuronal nitric oxide synthase (nNOS+) neurons. We used Wnt1-Cre;R26R-GCaMP3 mice, which express a genetically encoded fluorescent calcium indicator in all enteric neurons and glia. Exogenous GABA increased the intracellular calcium concentration, [Ca2+]i of some myenteric neurons including many that did not express GABA or nNOS (the majority), some GABA+, Calretinin+ or Neurofilament-M (NFM)+ but rarely nNOS+ neurons. GABA+ terminals contacted a significantly larger proportion of the cell body surface area of Calretinin+ neurons than of nNOS+ neurons. Numbers of neurons with GABA-induced [Ca2+]i transients were reduced by GABAA,B,C and nicotinic receptor blockade. Electrical stimulation of interganglionic fiber tracts was used to examine possible effects of endogenous GABA release. [Ca2+]i transients evoked by single pulses were unaffected by specific antagonists for each of the 3 GABA receptor subtypes. [Ca2+]i transients evoked by 20 pulse trains were significantly amplified by GABAC receptor blockade. These data suggest that GABAA and GABAB receptors are not involved in synaptic transmission, but suggest a novel role for GABAC receptors in modulating slow synaptic transmission, as indicated by changes in [Ca2+]i transients, within the ENS.

INTRODUCTION γ-Aminobutyric Acid (GABA), a prominent neurotransmitter in the central nervous system (CNS) is expressed by neurons of the enteric nervous system (ENS), enteroendocrine cells in the mucosa and nerve terminals in the smooth muscle layers of the gastrointestinal tract (Jessen et al., 1986;Krantis et al., 1994;Sang and Young, 1996;Krantis, 2000). GABA has been proposed to be a neurotransmitter in the ENS for decades, as early studies on the guinea-pig intestine showed that GABA can excite enteric neurons (Krantis et al., 1980;Cherubini and North, 1984a,b), and that electrical field stimulation of cultured enteric neurons releases [ 3 H]GABA (Jessen et al., 1983). Nonetheless, its role in synaptic transmission has remained elusive.
GABA is expressed in the ENS of several species; but significant interspecies differences in the distribution of GABA neurons and fibers have been reported (Furness et al., 1989;Messenger and Furness, 1990;Sang and Young, 1996;Timmermans and Scheuermann, 1998;Krantis, 2000). Cell bodies of GABA-containing neurons are predominantly found in the myenteric plexus of the ENS and have terminals in the myenteric plexus and the longitudinal and circular muscle, and so they are thought to be interneurons and motor neurons (Hills et al., 1987;Furness et al., 1989;Sang and Young, 1996). In the large intestine of rats and mice, GABA is expressed in the cell bodies of 5-14% of all myenteric neurons (Sang and Young, 1996;Krantis, 2000), while GABA cell bodies comprise about 5% of all myenteric neurons in the guinea-pig and mouse small intestine (Williamson et al., 1996;Li et al., 2011). In the mouse small intestine, expression of GABA is reported to be sparse and hence excluded from co-localisation studies Young, 1996, 1998). GABA may have a role in synaptic transmission in this region as there are many GABA-immunoreactive myenteric terminals in the mouse small intestine (Jessen et al., 1986;Hills et al., 1987;Sang and Young, 1996), but these have not been examined in depth.
All three major subtypes of GABA receptors have been identified within the ENS. GABA A , GABA B , and GABA C receptors are all expressed in the rat small and large intestine (Krantis et al., 1995;Nakajima et al., 1996;Poulter et al., 1999;Fletcher et al., 2001). In mouse, GABA A receptor subunits and GABA B receptors are expressed in the large intestine (Casanova et al., 2009;Seifi et al., 2014); the presence of the 3 receptor subtypes in the small intestine has been deduced from functional contractility studies (Sanger et al., 2002;Zizzo et al., 2007) although GABA C receptor expression and function has been particularly elusive in comparison to its counterparts (Auteri et al., 2015).
Unlike its predominant inhibitory role in the CNS, GABA can have both an excitatory and inhibitory influence on enteric neurons depending on the receptor it activates. GABA can act through GABA A and GABA C receptor subtypes, which are both pentameric chloride channels, to excite enteric neurons (Auteri et al., 2015). This due to the high intracellular Cl − concentration in enteric neurons, which is maintained by a Na + -K + -2Cl − symporter (Liu et al., 2013). In contrast, activation of GABA B receptors causes neuronal inhibition, as they are metabotropic G protein-coupled receptors that inhibit presynaptic voltageactivated Ca 2+ channels and activate postsynaptic inwardly rectifying K + channels (Auteri et al., 2015). The net functional output of GABA is difficult to predict as it depends on the receptor subtypes involved and their species-and region-specific localization (Auteri et al., 2015). Even within the same gut region, varying concentrations of GABA can alter the receptor subtype that is recruited leading to changes in the behavioral output (Auteri et al., 2014). While several studies have reported modulatory roles for GABA and its receptors on contractile activity in the gut (Tonini et al., 1989a,b;Sanger et al., 2002;Zizzo et al., 2007;Auteri et al., 2014), a role for GABA in synaptic transmission has not been identified.
In this study, we employed Ca 2+ -imaging from Wnt1-Cre;R26R-GCaMP3 mice which express a genetically encoded fluorescent calcium indicator in all enteric neurons and glia as a high throughput assay to elucidate the elusive role of GABA in synaptic transmission in the murine small intestine. We identified myenteric neuronal subtypes that express functional GABA receptors via the use of specific GABA receptor antagonists and post-hoc immunohistochemistry. Our data suggest that GABA A and GABA B receptors do not have a synaptic function within the ENS, but demonstrate a novel role for GABA C receptor in modulating slow synaptic transmission within the ENS.

Expression of GABA and Other Neuronal Subtype Markers
Pinned and stretched segments of ileum of mice from a C57/Bl6 background were fixed overnight in 4% formaldehyde in 0.1 M phosphate buffer, pH 7.2, at 4 • C, and then the tissue was given three washes with phosphate-buffered saline (PBS). Longitudinal muscle-myenteric plexus (LMMP) whole-mount preparations were obtained by firstly removing the mucosasubmucosa layer, then stripping away overlying circular muscle via microdissection. These preparations were incubated for 30 min with 1% triton X-100 (ProSciTech, Thuringowa, QLD, Australia) at room temperature. The tissue was then given three washes with PBS, followed by overnight incubation with a combination of primary antibodies (Table 1) at 4 • C. After three washes with PBS, the tissue was incubated with a combination of relevant secondary antibodies (Table 1) for about 2.5 h at room Preparations of myenteric plexus were viewed using a Zeiss Axio Imager M2 microscope and images were acquired with a Axiocam 506 mono camera using Zen 2.3 (blue edition) software (all from Zeiss, Australia). Images were taken using 20x or 40x objectives. The proportion of Calretinin and/or nNOS and/or GABA expressing neurons was obtained by examining co-expression with the pan-neuronal marker, Hu. GABA staining within neuronal cell bodies in this study was found to be highly variable, but unambiguous. Myenteric ganglia were imaged and a neuron was considered to be immunoreactive for GABA if the staining of its cytoplasm was higher than the background such that an unstained nucleus could be seen. At least 200 Hu+ cell bodies were counted in each preparation. The mean proportion of each neuronal subtype was determined by calculating averages from 3 animals. The data are expressed as mean ± SEM and n = the number of cells examined. Statistical analyses were performed using unpaired t-tests with P < 0.05 considered statistically significant. Comparisons were performed using using GraphPad Prism 5.0 (GraphPad Softwares, San Diego California).
To quantify GABA-immunoreactive terminals apposing Calretinin-and nNOS-immunoreactive neurons, highresolution z-stacks of preparations co-stained with GABA and Calretinin or nNOS were imaged using a Zeiss LSM800 confocal microscope. The images were sampled at a resolution of 1,024 × 1,024 pixels using the 63x/1.40 Oil DIC M27 objective, with a 1.8x software zoom and z steps of 0.43 µm. Fourteen to seventeen Calretinin-or nNOS-immunoreactive neurons were examined from each preparation. A total of 3 preparations, each from a different animal, was examined. 3D rendered surfaces of the neuronal cell-bodies and GABA terminals were generated using the 3D image analysis software Imaris x64 (Bitplane, version 9.0.0). The percentage of total cell surface area covered by GABA-ergic terminals was determined for each Calretinin-or nNOS-immunoreactive neuron. Data are presented as the mean % surface area ± SEM, where n = number of neurons examined. Statistical analyses were performed using unpaired t-tests with P < 0.05 considered statistically significant. Comparisons were performed using using GraphPad Prism 5.0 (GraphPad Softwares, San Diego California).

Tissue Preparation
Segments of ileum were removed from Wnt1-Cre;R26R-GCaMP3 mice, cut along the mesenteric border and pinned flat, mucosa side up in a dish lined with silicone elastomer (Sylgard 184; Dow Corning). The mucosa and submucous plexus were removed from the underlying smooth muscle and myenteric plexus layers, then strips of circular muscle were carefully peeled off and the resultant LMMP preparations were stretched over a small inox ring and immobilized by a matched rubber O-ring (Vanden Berghe et al., 2002). A maximum of 5 rings was prepared from each segment of ileum. The rings were transferred to an organ bath for imaging. The bath was constantly superfused (1 ml/min) with 95% O 2 , 5% CO 2 physiological saline at room temperature throughout the experiment via a gravity-fed inflow system that included a manual valve to switch between saline and drugcontaining saline solutions.

Imaging
The ring preparations were imaged using a 20x (NA 0.5) water dipping objective on an upright Zeiss Axioskope microscope with a Zeiss AxioCam MRm camera, and images (278 × 278) were acquired at 1 Hz.
Neurons within the myenteric ganglia were stimulated either chemically or electrically. Chemical stimulation involved the pressure ejection (spritz, 2 s duration; 9 psi) of GABA (1 mM) using a micropipette (tip diameter ∼20 µm) situated right at the edge of the imaged ganglion. During time control experiments GABA was applied to each ganglion via 2 repeated local applications, each separated by a 10 min time period. In initial studies, the amplitudes of GABA-evoked [Ca 2+ ] i responses were significantly reduced at the second time point suggesting desensitization of receptors, in accordance with previous reports (Cherubini and North, 1984a,b). To prevent potential desensitization of GABA receptors, the spritz pipette was moved away from the ganglion in between applications. To electrically stimulate the neurons, a focal stimulating electrode (tungsten wire; 50 µm diameter) was placed on an interganglionic fiber tract leading into the ganglion of choice where a single pulse and a train of 20 pulses (20 Hz) were elicited. For time control experiments, ganglia were stimulated first with a single pulse and then a train of 20 pulses 5 min later. This stimulation regime was repeated 2 times 10 min apart.
To test the effects of antagonists, drug-containing saline was superfused for 10 min into the organ bath after either the first GABA spritz or electrical stimulation regime, so that the first spritz or electrical time point was taken as the control response. Each ringed preparation was only exposed to an antagonist, or a combination of antagonists, once.
Following live-imaging experiments, selected preparations tissues were fixed overnight with 4% formaldehyde at 4 • C and immunostained using primary antisera to the neuron subtype markers, GABA, nNOS, neurofilament-M (NFM) and/or calretinin together with secondary appropriate antisera (Table 1), to identify the neurochemistry of the responding neurons.

Data Analysis and Statistics
Analyses were performed using custom-written directives in IGOR Pro (WaveMetrics, Lake Oswego, Oregon, USA). Regions of interest were drawn over a selected area of the cytoplasm for each neuron, excluding the nucleus because GCaMP3 is absent from the nuclei (Tian et al., 2009;Yamada and Mikoshiba, 2012). The intensity of the intracellular calcium ([Ca 2+ ] i ) transient signal for each response was calculated and expressed as the maximum increase in [Ca 2+ ] i from the baseline signal ( Fi/F0).
[Ca 2+ ] i transients were only considered if the intensity of the transient signal was more than 5 times the intrinsic noise. For both time control and antagonist experiments the Fi/F0 of the second GABA spritz, or second electrical stimulation response, was normalized and expressed as a fraction of the first (% Fi/F0). A minimum of 3 animals were examined for each experimental set; unless stated otherwise.
Ganglia of interest that were examined with Ca 2+ -imaging were processed for post-hoc immunohistochemistry and relocated using 20x or 40x objectives of a Zeiss Axio Imager M2 microscope. Images were acquired with a Axiocam 506 mono camera using Zen 2.3 (blue edition) software (all from Zeiss, Australia). Responding GCaMP3+ neurons were identified immunohistochemically by matching the micrographs with the Ca 2+ -imaging videos.
Data are presented as the mean % F i /F 0 of the control ± SEM where n = number of neurons examined or as mean ± SEM and n = the number of cells examined. Statistical analyses were performed using one-way analysis of variance (ANOVA) followed by Dunnett's post-hoc test with P < 0.05 considered statistically significant. Comparisons were performed using GraphPad Prism 5.0 (GraphPad Softwares, San Diego California).

Drugs Used
Drugs used included GABA, hexamethonium bromide (both from Sigma Aldrich, Castle Hill, (CGP 54626) and (1,2,5,6-Tetrahydropyridin-4-yl)methylphosphinic acid (TPMPA) (all from Tocris Bioscience, Avonmouth Bristol UK). All drugs were diluted in distilled water to make stock solutions and then again in physiological saline to their working concentration on the day of experimentation.

Expression of GABA in the Distal Ileal Myenteric Plexus
GABA+ neurons are relatively uncommon in the myenteric plexus of the mouse ileum and very little is known about their functional classification in this gut region (Sang and Young, 1996;Li et al., 2011). nNOS+ neurons have been described mainly as inhibitory motor neurons, while Calretinin+ neurons are intrinsic sensory neurons, interneurons and excitatory neurons to the muscle layers (Sang and Young, 1996;Li et al., 2011). In this study, GABA immunoreactivity was expressed in the somata of 8.2 ± 0.7% of Hu+ cells (n = 3 animals, Figure 1). GABA+ terminals and varicosities were also present within the myenteric ganglia and muscle layers ( Figure 1A). nNOS immunoreactivity was observed in 24.3 ± 2%, while Calretinin immunoreactivity Frontiers in Physiology | www.frontiersin.org was found in 33 ± 1.4% of Hu+ neurons (both n = 3 animals, Figure 1). This is similar to previous reports in the mouse ileum (Sang and Young, 1996;Qu et al., 2008). Furthermore, of the total number of Hu-immunoreactive neurons examined, we found that GABA colocalised with some of the nNOS+ (2.0 ± 0.1%) and Calretinin+ (3.7 ± 0.6%) neurons.
Calretinin and nNOS immunoreactivity reveals the whole shape of the neuron (Figures 4A,B) and has been previously used to examine numbers of varicosities contacting their cell bodies using confocal microscopy (Neal and Bornstein, 2007). In this study, we used high resolution confocal microscopy and Imaris surface rendering software to quantify GABAimmunoreactive terminals and varicosities that apposed cell bodies of Calretinin-and nNOS-immunoreactive neurons. Unlike synaptic proteins such as alpha-synuclein, synaptophysin, synaptotagmin and synaptobrevin which only stain varicosities (Sharrad et al., 2013), in our preparations, GABA stained axons and varicosities (Figures 4A,B). This made accurate counts of individual GABA+ varicosities onto Calretinin and nNOS cell bodies difficult, so we analyzed the % surface area of the neuronal cell bodies that were contacted by GABA-ergic axons and varicosities instead. We found that the % surface area of Calretinin+ cell bodies that was contacted by GABA-ergic axons and varicosities was significantly greater than the contacted surface area of nNOS+ cell-bodies (% surface area contacted by GABA+ structures: Calretinin 1.2 ± 0.1, 47 neurons; nNOS 0.8 ± 0.1, 44 neurons; P = 0.03) (Figure 4).

Involvement of Endogenous GABA in Synaptic Transmission
Electrical stimuli consisting of a single pulse or a train of 20 pulses (20 Hz) were applied via a stimulating electrode to an internodal strand leading to the ganglion imaged, which are respectively expected to evoke fast-intermediate, and fast-intermediate-slow excitatory postsynaptic potentials in the ENS (Shuttleworth and Smith, 1999;Nurgali et al., 2004;Gwynne and Bornstein, 2007;Foong et al., 2012Foong et al., , 2015. Bicuculline, TPMPA and CGP54626 did not affect the number of neurons that exhibited electrically-evoked [Ca 2+ ] i transients relative to time controls (all P > 0.1 Fisher's exact test; Table 3). The amplitude of single pulse-evoked [Ca 2+ ] i responses was unaffected by TPMPA (n = 33, P > 0.4; Figure 3B), but the amplitudes of train-evoked [Ca 2+ ] i transient responses were significantly potentiated by TPMPA (n = 52 P < 0.01; Figures 3C,D). Bicuculline and CGP54626 had no effect on the amplitudes of single or train-evoked responses (all P > 0.1; Figures 3B-D).

DISCUSSION
The physiological roles of GABA and its receptors in the gastrointestinal tract have proved elusive as they can have excitatory or inhibitory effects on enteric neurons and their expression is highly specific to gastrointestinal regions and species (Auteri et al., 2015). In this study, we used a multifaceted approach to examine the role of GABA and its receptors in the myenteric plexus of the murine ileum. We characterized the expression of GABA in neurons and varicosities surrounding enteric neurons, and identified the neuronal subtypes that express GABA receptors using specific antagonists and Ca 2+imaging. We showed that GABA receptors are unlikely to mediate fast or intermediate EPSPs and revealed a novel role for GABA C receptors activated by endogenously released

GABA Colocalises with Some Calretinin and nNOS Neurons in the Myenteric Plexus of the Mouse Ileum
In accordance with previous studies (Sang and Young, 1996;Li et al., 2011), we found that GABA neurons constitute 8% of all myenteric neurons in the murine ileum. Further, we found that GABA colocalised with a proportion of nNOSimmunoreactive neurons, which are often interneurons and inhibitory motor neurons, and some Calretinin-immunoreactive neurons which include intrinsic sensory, excitatory interneurons and motor neurons in the murine small intestine (Qu et al., 2008). Thus, GABA neurons may include subpopulations of such neurons. Indeed, we found GABA-immunoreactive varicosities in the circular muscle and within ganglia, where they closely apposed Calretinin-immunoreactive neurons and some nNOS-immunoreactive neurons. Thus, some GABA-containing neurons are probably interneurons contacting excitatory and inhibitory motor neurons as previously suggested for the rat colon (Krantis, 2000). This is consistent with findings from the murine colon, where widespread GABAergic innervation of neurons throughout the myenteric plexus has been identified by immunolabelling of the vesicular GABA transporter (vGAT) (Seifi et al., 2014).

Myenteric Neuronal Subtypes That Respond to GABA in the Mouse Ileum
Exogenous application of GABA evoked [Ca 2+ ] i transients in some myenteric neurons. The vast majority of these neurons were GABA-/nNOS-, including some Calretinin-and NF-M-immunoreactive neurons. nNOS-immunoreactive neurons rarely responded to GABA. Moreover, while GABA+ terminals and varicosities made close contacts to many Calretinin+ and nNOS+ neurons, the % surface area of Calretinin+ cell bodies contacted by GABA-ergic terminals was significantly greater than the contacted area of nNOS+ cell bodies. Although most GABA-immunoreactive neurons responded to GABA, these neurons made up only a small proportion of all responding neurons, reflecting their overall numbers. Nonetheless, we cannot exclude the possibility that GABA neurons make functional contacts onto other myenteric neurons including nNOS+ neurons that we did not detect.
Earlier intracellular studies of guinea-pig ileum showed that GABA excites intrinsic sensory myenteric neurons (which display distinctive AH-type electrophysiology) via GABA A receptors. Conversely, all other types of neurons that display S-type electrophysiology, which includes interneurons and motor neurons, were insensitive to GABA (Cherubini and North, 1984a). Our study is congruent with this, as some neurons that expressed Calretinin and NF-M, both markers of intrinsic sensory neurons, responded to GABA (Qu et al., 2008). But Calretinin also labels excitatory interneurons and motor neurons, and some nNOS and GABA neurons responded, albeit in smaller numbers. Moreover, half the GABA-evoked responses were in secondary neurons of the circuitry as they were blocked by, hexamethonium indicating involvement of a cholinergic synapse. That neurons other than intrinsic sensory neurons respond to GABA, differs from the earlier findings in the guinea-pig ileum (Cherubini and North, 1984a), which may be due to differential GABA receptor expression, which has not been detailed across species.
GABA A , GABA B , and GABA C Receptors Are Expressed in the Myenteric Circuitry GABA-mediated chloride conductance via GABA A and GABA Creceptors is known to produce depolarisations in enteric neurons, while GABA B receptor activation causes neuronal inhibition (Liu et al., 2013;Auteri et al., 2015). We found that GABA A , GABA B , and GABA C antagonists each prevent GABA-evoked calcium responses in many neurons. This suggests that some myenteric neurons, including nearly all GABA-immunoreactive neurons possess GABA A , GABA B , and GABA C receptors; some may possess more than one subtype. The concentrations of GABA receptor antagonists used were taken from previous publications (Sanger et al., 2002;Zizzo et al., 2007;Auteri et al., 2014) that showed selectivity in functional studies. However, a point to consider is that GABA antagonists have non-specific effects. In particular, bicuculline can block nicotinic receptors (Zhang and Feltz, 1991) and can act on SK (small conductance) Ca 2+activated K + channels (Debarbieux et al., 1998;Khawaled et al., 1999;Johansson et al., 2001). Each of these actions may have implications for neuronal excitability. However, such "off-target" effects cannot account for the failure of bicuculline to alter the responses to electrical stimulation of interganglionic connectives, as nicotinic blockade would depress such responses  and there is no evidence for involvement of SK channels in the regulation of enteric neuron excitability. Further, at higher concentrations (∼EC 50 500 µM), TPMPA may weakly activate GABA B receptors (Ragozzino et al., 1996), but as 5-fold lower concentrations were used in this study, an action on GABA B receptors is unlikely.
A response to GABA that was not blocked by either GABA A or GABA C antagonists or both was identified, potentially mediated via GABA B receptors. GABA B receptor blockade abolishing responses in a significant number of neurons is a surprising finding, but application of a GABA B receptor agonist induced [Ca 2+ ] i transients in cultured guinea-pig myenteric neurons (Reis et al., 2006). It is also possible that GABA B receptors are located on inhibitory inputs to the responding neurons, such that when they are blocked, inhibition is restored thereby decreasing the number of GABA responders.
Responses that were not abolished by a GABA B receptor antagonist were significantly larger than control responses, revealing a potential inhibitory action of GABA. Indeed GABA B receptors in the mouse gut act at the presynapse to prevent ACh release from excitatory cholinergic neurons (Sanger et al., 2002;Hyland and Cryan, 2010), which may account for the increase in neuronal activity observed following GABA B receptor blockade in this study.
In the mouse gut, GABA A and GABA B receptors have been identified in the colon (Casanova et al., 2009;Seifi et al., 2014;Auteri et al., 2015) and GABA A receptors are expressed on neurochemically diverse cell types including nNOS, ChAT, 5-HT, and SOM neurons (Seifi et al., 2014). Expression of GABA C receptors in the mouse was only deduced by contractility studies in the longitudinal muscle of the small intestine (Zizzo et al., 2007). We attempted to examine expression of GABA receptors on enteric neurons using immunohistochemistry, but failed to obtain satisfactory immunostaining with the available antisera. Thus, this component of the study was not pursued further.

Involvement of Endogenous GABA in Synaptic Transmission and a Role for GABA C Receptors
Electrical stimulation to evoke endogenous neurotransmitter release was used in this study to ascertain whether any GABA receptors and therefore endogenous GABA contribute to synaptic transmission. Using a reporter mouse which expresses GCaMP3 in a Cre-dependent manner in the Rosa26 locus (Tian et al., 2009;Zariwala et al., 2012) has revolutionized research into the examination of ENS excitability, allowing highthroughput analysis in previously challenging or impossible situations Hao et al., 2017). With the omission of tissue loading steps, tissue viability and resolution are increased, background signals are reduced with the GCaMP3 indicator being specifically Wnt-1 driven to be selectively expressed in neural crest derivatives (enteric neurons and glia) within the heterogenous preparation (Boesmans et al., 2013). GCaMP3, although not the current fastest variant of genetically encoded calcium indicators in its family, can detect calcium transients with amplitudes linearly dependent on action potential number in the brain of intact mice (Tian et al., 2009). The stimulus regime and sampling frequency we used will detect [Ca 2+ ] i transients due to neuronal activity. While the relatively slow inactivation kinetics of GCaMP3 will not resolve fast events like action potentials and subthreshold fast EPSPs (Tian et al., 2009;Michel et al., 2011), we can detect their longer term consequences as neuronal activity. The recorded single pulse-evoked [Ca 2+ ] i transients in our study include responses to fast-intermediate synaptic events that are mediated by several neurotransmitters, but particularly include acetylcholine activating nicotinic receptors, the typical fast neurotransmitter in the ENS . While [Ca 2+ ] i transients evoked by a 20 Hz train of pulses include responses to neurotransmitters mediating fast, intermediate and slower synaptic events.
In our study, single pulse-evoked [Ca 2+ ] i responses were unaffected by any of the GABA receptor antagonists. Responses evoked by trains of stimuli were unaffected by GABA A or GABA B blockade, but potentiated by the GABA C antagonist, TPMPA, which has a strong selectivity (>100-fold) for GABA C receptors compared with GABA A or GABA B (Ragozzino et al., 1996;Johnston, 2002). Hence despite neural expression of all 3 GABA receptors, our data do not support a role for GABA in fast neurotransmission, but indicate that release of endogenous GABA modulates slow synaptic transmission, where GABA acting at GABA C receptors tonically inhibits the system. The exclusion of GABA A and GABA B receptors is surprising as there is strong evidence in the guinea-pig myenteric plexus that GABA excites neurons via GABA A receptors North, 1984a,b, 1985;Zhou and Galligan, 2000), and in cultured guineapig myenteric neurons, application of GABA and agonists for GABA A and GABA B receptors induced [Ca 2+ ] i transients, while GABA C agonist only elicited small responses in fewer neurons (Reis et al., 2006). Moreover, several studies have implicated functional roles for GABA A and GABA B receptors (Sanger et al., 2002;Zizzo et al., 2007;Auteri et al., 2014;Seifi et al., 2014), but there is only one report of involvement of GABA C receptors (Zizzo et al., 2007) in the control of smooth muscle contractility in the mouse gut. While interspecies and gut regional differences can be postulated, we cannot exclude a potential role of GABA A and GABA B receptors in synaptic transmission, as [Ca 2+ ] i is only one indication of neuronal excitability so their contribution may be revealed using other types of measurements, sampling frequencies or stimulus regimes.
While GABA C receptors mediate excitation via a depolarising conductance, intrinsic differences in receptor kinetics may account for their inhibition of responses to trains of stimuli. GABA A receptors mediate rapid transient currents, while GABA C receptors produce more prolonged responses (Qian and Dowling, 1995;Lukasiewicz and Shields, 1998), where the deactivation rate or closing phase of GABA C receptors is much slower (Amin and Weiss, 1994). In addition, unlike GABA A receptors, GABA C receptors display little desensitization (Matthews et al., 1994). GABA C receptors, like those located on terminals of retinal bipolar cells produce a (tonic) standing current due to these properties (Hull et al., 2006;Palmer, 2006). Slow excitatory transmission in intrinsic sensory neurons displaying AH-type electrophysiology, is associated with a net decrease in membrane conductance due largely to inhibition of two potassium conductances; resting or "leak" K + conductance (gK) and calcium-dependent K conductance (gK Ca ) (Grafe et al., 1980). While this applies to AH neurons, the situation involving S neurons is more complex as the K + channels have not been clearly identified and there is potential for contribution by Mcurrent channels like the KCNQ (Kv7) channels (Brown and Passmore, 2009;Hirst et al., 2015). In addition, slow transmission may also be due to an increase in chloride conductance (gCl) in myenteric neurons (Bertrand and Galligan, 1994). Inactivation of resting gK and activation of gCl leads to depolarisation of the neuron relative to the resting potential to a membrane potential approximating the reversal potential of Cl − which is estimated to be between −39 (Cherubini and North, 1984a) and −17 mV (Bertrand and Galligan, 1994) in guinea-pig myenteric neurons. Therefore, activation and prolonged opening of GABA C receptors and thus chloride conductance, would lead to a new membrane potential approximating the reversal potential of Cl − . The initial depolarisation of the cell at this membrane potential is likely to activate delayed rectifier K + channels which are active when the membrane is depolarised to values above −30 mV (Hirst et al., 1985), producing a repolarising inhibition. This stabilization of the membrane would prevent the activation of voltage gated Na + channels and further depolarisation, therefore allowing GABA to shunt action potential firing by a "clamping" of the membrane potential. Thus, following repetitive stimulation of interganglionic connectives, prolonged opening of these channels by endogenous GABA may be responsible for this shunting inhibition which would be released with GABA C receptor blockade, producing the potentiated response observed in this study.

CONCLUSIONS
Our study extended understanding GABA within the murine myenteric plexus in the small intestine, showing that some GABA-immunoreactive neurons are probably interneurons as in other species. Broadly surveying the ENS via Ca 2+imaging has enabled us to observe the previously elusive endogenous effects of GABA, to show expression of all 3 GABA receptor subtypes, and for the first time reveal a role for modulating synaptic transmission in the myenteric plexus via GABA C receptors. The specific actions of GABA C receptors will require more detailed analysis of the membrane potential, which would be via intracellular recording. However, recording individual neurons via sharp electrodes from a sufficient sample size of neurons is technically difficult and outside the scope of study. Predictions of neuronal behavior via realistic computational modeling experiments can also be employed in the future to examine the net effect of the activation of various receptors at neuronal synapses.