GPR39 Is Coupled to TMEM16A in Intestinal Fibroblast-Like Cells

GPR39 is a GPCR implicated as a regulator of gastrointestinal motility, although the mechanism remains elusive. Here, we report that GPR39 is expressed by a specific cell population cultured from mouse small intestine muscle layers, which was subsequently identified as fibroblast-like cells (FLCs) that have recently been shown to modulate gut motility. Application of the GPR39 agonist, Zn2+, induced large currents and membrane depolarization in FLCs cultured from wild-type mice, but not Gpr39 −/− mice. This Zn2+-induced current could be suppressed by application of a TMEM16A antagonist, CaCCinh-A01, or by silencing Tmem16a expression. These data suggest that GPR39 might modulate gut motility via regulating TMEM16A function in FLCs.


Introduction
GPR39 is a G protein-coupled receptor (GPCR) found in all vertebrates. Together with Gpr38, Gpr39 was cloned as a novel member of the ghrelin and neurotensin receptor families [1]. While GPR38 has since been deorphanized as the receptor for motilin [2], an important peptide regulator of GI tract motility [3], the natural ligand of GPR39 is still a subject of discussion. In 2005, Zhang et al. reported GPR39 to be the receptor for obestatin [4]. However, studies by others have contradicted this finding [5,6], and demonstrated the inability of obestatin to bind GPR39 in vitro. Instead, Zn 2+ has been proposed as an endogenous agonist for GPR39 [5,7,8]: exogenously applied Zn 2+ stimulates inositol 1,4,5trisphosphate production, as well as induces Ca 2+ signals in GPR39-expressing cells; the potency and efficacy of Zn 2+ on GPR39 suggests that it could well be physiologically relevant modulator of GPR39 in vivo.
GPR39 is most highly expressed in the pancreas, gastrointestinal (GI) tract, liver and kidney [9,10]. Other tissues that express GPR39 include adipose tissue, thyroid, and heart, as well as some regions of the brain [9]. As to its physiological function, GPR39 is involved in insulin release and pancreatic function [5,11]. It has also been implicated as a regulator of GI motility since Gpr39 2/2 mice display an acceleration of gastric emptying and a more effective colonic expulsion [10]. However, little is known about the mechanism by which GPR39 regulates gut motility.
Regulation of GI motility takes place through the coordinated activities of enteric and sensory neurons, smooth muscle cells (SMCs), as well as interstitial cells of Cajal (ICC). Endocrine or paracrine factors associated with secretory glands and the gut flora also play a role [12]. More recently, FLCs are suggested as a new class of excitable cells that might also be involved in the control of GI motility [13]. In the gastrointestinal tract, FLCs form a cellular network with their processes and mirror the anatomical distributions of ICC [14]. They can be distinguished from ICC by their robust expression of platelet derived growth factor receptor a (PDGFRa) and lack of c-Kit expression. It has been suggested that FLCs help with spreading of the slow waves generated by the ICC [15]; or, because FLCs locate near terminals of enteric motor neurons and form a syncytium with SMCs via gap junctions [13,14], they might be involved in motor neurotransmission in GI tract.
To increase understanding of the role of GPR39 in the gut, we first conducted extensive work using commercially available and in-house generated GPR39 antibodies to localise expression. However, we were unable to generate robust and convincing data on the cell-types expressing GPR39, and therefore focussed on functionally localising GPR39 in isolated gut cells. In the present study, we report that functional GPR39 is highly expressed by a specific cell population cultured from GI muscle layers, which were subsequently identified as intestinal FLCs. Activation of GPR39 by Zn 2+ not only resulted in evident Ca 2+ signals in the cultured FLCs, but also induced large TMEM16A-dependent currents and membrane depolarization. These data suggest that GPR39 is functionally coupled to TMEM16A channels in cultured intestinal FLCs.

Results
Gastrointestinal (GI) motility is tightly controlled by the enteric nervous system. Therefore, GPR39 might regulate gut motility by influencing the activity of enteric neurons. However, no Zn 2+ - induced Ca 2+ signals were observed in cultured enteric neurons (Fig. S1). Furthermore, we observed no changes in cAMP or IP3 signaling pathways when the cultures were challenged GPR39 agonist (data not shown), suggesting the expression of functional GPR39 in enteric neurons is low. To further explore the cell type expressing GPR39, we set up a culture using dissociated muscle layers from mouse small intestine. Over the culture period, a highly proliferative cell population gradually dominated the culture, accounting for more than 90% of the cells in the culture. These proliferating cells showed a distinct morphology of cobblestones under the microscope (Fig. 1A). Real time PCR analysis showed that expression of Gpr39 was up-regulated more than 160-fold in the cultures as a result of the enrichment of this cell-type after 7 days in vitro (163.9610.5 fold, n = 3, P,0.0001, Fig. 1B). When challenged with Zn 2+ , a robust but transient Ca 2+ signal was evident (n = 176, Fig. 1C and D). In contrast, none of those cobblestone-like cells cultured from Gpr39 2/2 mice showed any response to Zn 2+ (n = 565).
To illustrate the origin of the Ca 2+ signals, Ca 2+ was removed from the external bath solution before challenging the GPR39expressing cells with Zn 2+ : the [Ca 2+ ] i rises remained unaffected (Fig. 1E). In contrast, Ca 2+ signals were completely attenuated by the pre-treatment with thapsigargin (Tg) to deplete intracellular Ca 2+ stores (Fig. 1F). Thus, as predicted for such a Gq-coupled GPCR, Zn 2+ -dependent Ca 2+ responses in the cobblestone-like cells were mediated by release of Ca 2+ from intracellular stores via GPR39 activation.
Over the culturing period, in parallel to up-regulation of Gpr39 mRNA expression, both the percentage of Zn 2+ responding cells and the total Ca 2+ response are increased in the cultures established from wild-type (WT) mice (Fig. 2), suggesting the domination of these cells in the culture. To further confirm the correlation between the cobblestone morphology and GPR39 function in the primary cultured cells, we analyzed Ca 2+ responses to Zn 2+ application in the cultures from wild-type mice: indeed, Zn 2+ elicited Ca 2+ signals in all cobblestone-like cells, while cells with other morphologies only responded to ATP (Fig. S2). Together, it is concluded that Zn 2+ induces Ca 2+ signals in these cobblestone-like cells via GPR39 activation.
The muscularis externa of the GI tract contains SMCs, enteric neurons, glial cells, immune cells and several classes of interstitial cells, including ICC and FLCs [13]. In order to determine the identity of the cobblestone-like cells, we performed immunostaining with various cellular markers. The cells were not immunopositive for neuronal (bIII tubulin) or glial (GFAP) markers, the marker for SMCs (smooth muscle a-actin) (data not shown), nor for the ICC marker c-Kit (Fig. 3), but were strongly immunopositive for CD34 and PDGFRa. Similarly, up-regulation of CD34 and pdgfra expression, as well as down-regulation of c-kit was observed in these cultures (Fig. 3). Also, these cobblestone-like cells showed positive immunostaining for a small-conductance Ca 2+activated K + channel (SK3) (Fig. S3). The expression of CD34, PDGFRa and SK3 has previously been used to differentiate FLCs from other cell types in the gut [13][14][15]. Collectively, these data suggest that the cobblestone-like cells described here are likely to be FLCs, and referred to as such in the rest of the manuscript.
FLCs form an excitable network within muscle layers, and have gap junctions with circular and longitudinal SMCs. Recently Kurahashi et al. proposed that FLCs might be involved in motor neurotransmission in GI tract [13]. We therefore examined whether a GPR39 agonist could affect the membrane potential of cultured FLCs. Perforated patch was used to preserve intracellular integrity. To ensure the maximum Ca 2+ response in FLCs, we challenged the cells with 200 mM Zn 2+ (EC 50 = 30 mM in FLCs as measured with Ca 2+ imaging, Fig. S4). For FLCs cultured from WT or Gpr39 2/2 mice, no significant difference was observed between their resting membrane potentials (24061.48 mV vs 23761.31 mV, p.0.05, n = 23 and 19). As shown in Figure 4A and 4B, brief exposure to Zn 2+ led to a rapid depolarization of FLCs (16.962.49 mV, n = 5), while no such effect was observed in FLCs cultured from Gpr39 2/2 mice (n = 6). Next, we investigated whether whole-cell currents were stimulated by Zn 2+ in FLCs. Shortly after Zn 2+ application, an outward rectifying current developed that peaked within 2 minutes and rapidly decreased thereafter ( Fig. 4B and C, 17366198.3 pA, n = 21) (again, all FLCs identified by the cobblestone morphology showed responses to Zn 2+ ). As expected, no such current was observed in cells cultured from Gpr39 2/2 mice ( Fig. 4D and E, n = 11). These observations demonstrate that activation of GPR39 resulted in stimulation of currents in FLCs.
At its peak, the Zn 2+ -induced current reversed at around 240 mV indicating that it is likely to be carried by either potassium or chloride channels. To determine the type of channel involved, we used CsCl in pipette solutions to block potassium channels. Large currents were observed after Zn 2+ application (Fig. 5A, 1590.56442.9 pA, n = 5, p.0.05 compared with Fig. 4C), suggesting potassium currents were unlikely to be the major component of the Zn 2+ -induced current in FLC. To further confirm that the Zn 2+ -induced current is a chloride current, whole-cell recordings were performed with symmetrical N-methyl-D-glucamine chloride (NMDG-Cl) solutions: both extracellular and pipette solutions contained NMDG-Cl, so that cation currents were minimized and the chloride current was the only channel activity that could be recorded. Upon Zn 2+ stimulation, a prominent current was observed (Fig. 5B, n = 4). The reversal potential was around 0 mV, which reflects the symmetrical Cl 2 concentrations across the membranes. Together, these data imply that activation of GPR39 leads to the opening of a chloride channel in FLCs.
The GI tract contains numerous chloride channels, including cystic fibrosis transmembrane regulator (CFTR) [16]. However, the lack of CFTR antagonist activity excluded the involvement of this channel in Zn 2+ -induced currents (Fig. 5C, n = 3). TMEM16A is another chloride channel that has been reported to be highly expressed in the GI tract [17]. To explore whether the Zn 2+induced Cl 2 current was carried by TMEM16A, a TMEM16A antagonist, CaCC inh -A01, was tested [18]. It blocked the channel acutely when it was applied after the channel had been activated by Zn 2+ (the current decrease at the rate of 610.676107.8 pA/min before, and 2595.856241.1 pA/min after CaCC inh -A01 application, p,0.001) (Fig. 5D, bottom trace, n = 7). Furthermore, when CaCC inh -A01 was applied prior to Zn 2+ administration, no current was observed (16366214.3 pA in control cells, 281.4682.1 pA in CaCC inh -A01 pre-treated cells, p,0.001, n = 8) (Fig. S5). Thus, CaCC inh -A01 effectively blocked the Zn 2+induced Cl 2 current in FLCs, suggesting this current could be indicated on a rainbow scale with blue representing low [Ca 2+ ] i and orange/red high [Ca 2+ ] i . (D) Ca 2+ signals were elicited in cells cultured from wildtype mice, but not Gpr39 2/2 mice. (E) Zn 2+ induces Ca 2+ signals in Ca 2+ free bath solution. (F) Cells were pre-incubated with or without 1 mM thapsigargin to deplete intracellular Ca 2+ stores. Thapsigargin treatment not only abolished Zn 2+ elicited Ca 2+ signals, but also resulted in a decrease in [Ca 2+ ] i after Zn 2+ application, the mechanism of which is still unknown. doi:10.1371/journal.pone.0047686.g001 carried by TMEM16A channels. Indeed, Tmem16a expression was readily detected in cultured FLCs by reverse transcription PCR (Fig. 5E).
As shown in Figure 4, the Zn 2+ induced Cl 2 current is absent in Gpr39 2/2 FLCs. However, if this current is indeed carried by TMEM16A, and given TMEM16A is a Ca 2+ -activated Cl 2 channel, such a current should be evoked in Gpr39 2/2 FLCs by raising intracellular [Ca 2+ ]. To test this, we examined the effect of ionomycin, a Ca 2+ -specific ionophore, on FLCs. As shown in Figures 6A and 6B, in FLCs cultured from Gpr39 2/2 mice, Zn 2+induced currents are absent, whereas ionomycin-induced currents are similar to those in WT FLCs (826.46147 pA in WT vs 825.56131 pA in Gpr39 2/2 , p = 0.83, n = 11 and 6). Therefore, as predicted for TMEM16A, Ca 2+ alone is sufficient to activate this current in FLCs.
To further confirm the identity of the channel, we knocked down Tmem16a expression in FLCs by siRNA transfection. Three days after transfection, Tmem16a expression was down-regulated by 94% in the transfected cells, while no significant changes in Gpr39 expression were observed (n = 3, Fig. 6C). Accordingly, GPR39 and TMEM16A activities were also measured in those cells: Zn 2+ -induced Ca 2+ responses were unaffected by Tmem16a silencing (DR 340/380 : 1.460.06, n = 54, in control; 1.4260.06, n = 67, in transfected cells, p = 0.83), while Zn 2+ -induced Cl 2 currents were reduced to 32% in transfected FLCs (16366214.3 pA in control vs 529.56147.1 pA in transfected cells, p,0.001, n = 7 and 12) (Fig. 6D). Similar results were observed when another Tmem16a siRNA was used (data not shown). Collectively, our data suggest that activation of GPR39 is functionally linked to the opening of TMEM16A channels in FLCs.

Discussion
Despite the evidence that GPR39 might be involved in regulation of GI motility, most GPR39 expression was suggested to be in enterocytes within the epithelium of GI tracts [10]. Expression in enteric neurons was also reported [10], however we did not observe any Zn 2+ induced responses in cultured myenteric neurons. Thus, it is difficult to explain how GPR39 affects GI motility. Using a culture established from the intestinal muscular layer, we identify a cobblestone-like cell population that expresses functional GPR39. They are likely to be FLCs or PDGFRa+ cells as suggested by immunocytochemistry results [13,14]. Therefore, functional GPR39 expression in intestinal FLCs might account for its role in regulation of GI motility. Indeed, we observed changes of membrane potentials in the cultured FLCs upon GPR39 activation, which is in line with the idea that FLCs might be excitable cells and able to modulate SMC activity via a syncytium [13].
Ca 2+ -activated Cl 2 channels (CaCC) play a basic or at least a modulatory role in many tissues, including various sensory cells, different types of smooth muscles, heart, endothelium, neuronal tissues, and epithelial organs [19]. They mediate Ca 2+ -dependent Cl 2 secretion in glands and epithelia, and modify cellular responses to various stimuli in muscle, nerve and receptors. [19]. As a major component of CaCC, the importance of TMEM16A in GI physiology is underscored by the diminished rhythmic contraction in GI tracts of Tmem16a knockout mice [17,20]. Immunochemistry results suggest TMEM16A is expressed by SMCs in various tissues such as the airway, reproductive ducts and oviduct. However, in the GI tract, robust expression of TMEM16A has only been reported in the ICC network [17,20], which serves as the pacemaker of GI tract contractility. Although these earlier studies did not identify any other TMEM16A expressing cell-type, our data suggest that in addition to ICCs, prominent TMEM16A currents are also present in FLCs.
Activation of TMEM16A can be achieved by application of ionomycin, a Ca 2+ -specific ionophore. It has been also reported that various GPCRs can be functionally coupled with TMEM16A. When over-expressed in HEK293 cell, TMEM16A has been shown to link to M3 or P2Y2 receptors [21,22]. In small nociceptive neurons from rat dorsal root ganglia, activation of B 2 receptors by Bradykinin robustly opens TMEM16A [23]. Zhu et al. also demonstrated that TMEM16A current in ICCs can be elicited by muscarinic activation [22]. In line with these reports, our results suggest that TMEM16A is functionally linked with GPR39 in the intestinal FLC population, and might serve as a downstream effector of GPR39. Interestingly, we noticed that Zn 2induced currents are significantly larger than those induced by ionomycin (Fig. 6A), which suggests that other factors might be involved in TMEM16A activation in these cells.
How GPR39 regulates gut motility remains unclear. However, given the potential role for FLCs in GI function [13] and the importance of TMEM16A in regulation of GI motility [20], it is tempting to speculate that the motility changes observed in Gpr39 2/2 mice are the result of disrupted coupling between GPR39 and TMEM16A in the intestinal FLCs. Meanwhile, overexpression of GPR39 contributes to development of esophageal squamous cell carcinoma [24]; while TMEM16A also serves as a marker for gastrointestinal stromal tumors (GIST) [25]. Although GISTs are thought to originate from ICCs, a subset of GIST display a similar expression pattern as observed in cultured FLCs (TMEM16A, CD34 and PDGFRa positive, but c-Kit negative) [25,26]. Therefore whether these tumors are originated from FLCs, and whether coupling between GPR39 and TMEM16A plays a role in GI tumorigenesis may be worth further investigation.
In conclusion, our data shows that GPR39 is highly expressed in intestinal FLCs isolated from the muscularis externa. Its activation leads to the opening of TMEM16A channels and depolarization of these cells. The coupling mechanism between GPR39 and TMEM16A in the intestinal FLCs might provide a framework for further studies in GPR39 biology.

Isolation and Culture of Fibroblast-like Cells
Intestinal fibroblast-like cells were isolated from wild-type C57BL/6 mice (Charles River) or Gpr39 2/2 mice (Deltagen). All the procedures were performed according to UK Home Office regulations. Briefly, animals (8,12 weeks) were deeply anesthetized before being sacrificed. The small intestines were removed and the muscularis externa stripped. The tissue was cut into pieces 5-6 mm in length, and digested at 37uC for 2 hours in Ca 2+ / Mg 2+ -free Dulbecco's Phosphate-Buffered Saline (Invitrogen) containing 1.25 mg/ml collagenase, type IV and 0.125 mg/ml DNase (Worthington Enzymes). After the incubation, tissues were dissociated into single cells by passing through fire-polished glass   pasteur pipettes. Aliquots were placed onto glass coverslips and cultured in a humidified atmosphere containing 5% CO 2 at 37uC. Cells were cultured in Minimum Essential Medium (Invitrogen), supplemented with 10% fetal calf serum, antibiotics and antimycotics. Culture medium was changed every 2-3 days. Immunocytochemistry Cells were stained according to standard protocols. Briefly, cells were washed 3 times in PBS prior to being fixed in ice-cold 4% paraformaldehyde for 15 mins. The cells were washed a further 3 times in PBS and then blocked in PBS containing 5% donkey serum, 1% BSA and 0.3% triton X100. The cells were then incubated overnight at 4uC in PBS containing 1% BSA and 1:100 dilution of primary antibodies: anti-CD34 (BD Biosciences), anti-SK3 (Alomone) or anti-PDGFRa (Abcam). Cells were then rinsed 3 times in PBS before being incubated in a FITC or Cy3conjugated secondary antibody for 3 hours at 4uC. Finally, the cells were rinsed a further 3 times in PBS before being mounted in Vectashield containing DAPI (Vector).

PCR
RNA was extracted from cultured enteric neurons and animal tissues using an RNeasy mini kit (Qiagen) according to the manufacturer's recommended protocol. The samples were tested for concentration/RNA integrity using an Agilent 2100 Bioanalyser. 1 mg of total RNA from each sample was reverse transcribed using SuperScript TM III reverse transcriptase and a random hexamer primer (Invitrogen) according to the manufacturer's instructions. All reagents used for real-time PCR were purchased from Applied Biosystems and were used according to the manufacturer's instructions. Individual PCR reactions were set up with 20 ng of cDNA and samples were analyzed on an Applied Biosystems 7500Fast system. Samples were run as uniplex in triplicate using the following gene-specific probes: Gpr39: Mm01308380_s1; Tmem16a: Mm0072442424_m1 (Applied Biosystems).
In Figure 5E, Tmem16a was detected by 30 cycles of amplification with primer sets AATCGCGCACGAGGCACAG and AAGCGTTTGTCATCTTCATGGTAATCC. The expected size of PCR product is 459 base pair.
In voltage clamp configuration, currents were measured by ramping from 2100 mV to +100 mV (applied at 0.1 Hz over 200 ms period) from a holding potential of 280 mV. Perforation was achieved with either 10 mg/ml gramicidin or 60 mg/ml of amphotericin-B. No difference was observed between the currents recorded with these two reagents. For Fig. 4A, membrane potentials were continuously recorded in I = 0 mode (zero current clamp).
Tmem16a Silencing using siRNA For gene silencing experiments, cells were reverse transfected with either Tmem16a siRNA (s97830, Applied Biosystems) or dH 2 O (mock transfection) using RNAiMAX (Invitrogen) as described by the manufacturer. The final concentration of siRNA in the medium was 20 nM. 72 h after transfection, cells were subjected to gene expression analysis or functional analysis.

Statistical Analysis
All statistical analyses were conducted using the Origin 7.5 software (OriginLab). Data were expressed as the mean6SEM from n $ 3 separate experiments, and subjected to two-sample t test. A p value of p#0.05 was considered statistically significant with * indicating p,0.05, ** p,0.01 and *** p,0.001. Figure S1 Cultured myenteric neurons [27] were loaded with Fura2-AM for 1 hour at 37uC. Cells were then challenged with 100 mM Zn 2+ . No apparent Ca 2+ signal was observed. The neuronal status of the cells was confirmed by depolarizing the cells with 75 mM K + at the end of each experiment. (TIF)