Ion Channel Expression in the Developing Enteric Nervous System

The enteric nervous system arises from neural crest-derived cells (ENCCs) that migrate caudally along the embryonic gut. The expression of ion channels by ENCCs in embryonic mice was investigated using a PCR-based array, RT-PCR and immunohistochemistry. Many ion channels, including chloride, calcium, potassium and sodium channels were already expressed by ENCCs at E11.5. There was an increase in the expression of numerous ion channel genes between E11.5 and E14.5, which coincides with ENCC migration and the first extension of neurites by enteric neurons. Previous studies have shown that a variety of ion channels regulates neurite extension and migration of many cell types. Pharmacological inhibition of a range of chloride or calcium channels had no effect on ENCC migration in cultured explants or neuritogenesis in vitro. The non-selective potassium channel inhibitors, TEA and 4-AP, retarded ENCC migration and neuritogenesis, but only at concentrations that also resulted in cell death. In summary, a large range of ion channels is expressed while ENCCs are colonizing the gut, but we found no evidence that ENCC migration or neuritogenesis requires chloride, calcium or potassium channel activity. Many of the ion channels are likely to be involved in the development of electrical excitability of enteric neurons.

A growing body of evidence indicates that ion channels regulate cell motility, neurite outgrowth and axon guidance in a variety of cell types [17][18][19]. Ion channels regulate cell migration by mediating changes in cell shape and volume through osmotic activity, and by interactions with components of the cell migration machinery including the cytoskeleton and integrins [19][20][21][22]. For example, glioma cell migration requires activity of K + and Clchannels [20,[23][24][25][26], platelet migration requires activity of Ca 2+ and K + channels [27], neuroblast migration along the rostral migratory stream involves K + channels [28], and the migration of postmitotic granule cells in the cerebellum [29] and trunk neural crest cells in the salamander [30] are regulated by Ca 2+ channels. In growing axons, ion channels can regulate the rate of axon growth, the response to different guidance cues and target recognition. For example, voltagedependent K + channels have been reported to regulate the rate of growth and guidance of retinal ganglion cell axons [31], activity mediated by voltage-dependent Ca 2+ channels regulates transcription of receptors for axon guidance molecules, which in turn regulates the growth rate of the neurites [32], Clchannels are involved in nerve growth factor-induced neurite outgrowth in PC12 cells and cortical neurons [33] and Ca 2+ channels can interact with laminin to inhibit neurite growth after arrival at the target tissue [34].
We have previously shown the expression of several sodium channels in the ENCCs at E11.5, namely Na v 1.3, Na v 1.5, Na v 1.6, Na v 1.7 and Na v 1.9 [35]. Other studies have detected expression of the calcium channel genes Cacng4 [36] and Cacng2 [37] and the chloride channel Clcn4-2 [36] in E14.5 ENCC. However, there has been no comprehensive study of the expression of ion channels by ENCCs, and little is known about whether ion channels play roles in ENCC migration and/or neurite formation during ENS development. Therefore, we first investigated the expression of ion channels by ENCCs using a PCR-based array. We found that many ion channels, including Cl -, Ca 2+ , K + and Na + channels are already expressed by ENCCs at E11.5, and there is an increase in the expression of numerous ion channel genes between E11.5 and E14.5. As this time period coincides with population of the gut by ENCCs and the first extension of neurites by enteric neurons, we then examined the effects of pharmacological inhibition of many of the ion channels on ENCC migration and neurite formation. None of the Ca 2+ or Clblockers examined had significant effects on migration or neurite formation. The non-selective K + channel blockers, TEA and 4-AP, retarded ENCC migration and inhibited neurite formation, but only at concentrations that also resulted in significant cell death. collected in phosphate buffered saline (PBS), pelleted, excess PBS removed and immediately frozen at -80°C. The small intestine was isolated from postnatal day (P)0 and adult mice in sterile DMEM/F12, and the mucosa removed with forceps, between 9 AM-3 PM. The remaining muscle, myenteric plexus and serosa were immediately transferred into 1ml of RNAlater (Qiagen).
Total RNA was extracted from approximately 1x10 6 freshly dissociated and purified E11.5 and E14.5 FACS-sorted ENCCs using Qiashredder and RNeasy mini kit (Qiagen), including the on-column DNase treatment, according to manufacturer's instructions. Total RNA was extracted from P0 and adult gut using Trizol (Life Technologies Invitrogen), then purified further using RNeasy mini columns and on-column DNase treatment (Qiagen), according to manufacturer's instructions. RNA quality and quantity were tested by spectrophotometry using a NanoDrop 1000 and electrophoresis, and only RNA meeting the criteria detailed by SABiosciences RT 2 Profiler PCR Array System was used in the arrays.

PCR array
0.2 μg of total RNA was converted to cDNA for each age, using the RT 2 First Strand kit (SA Biosciences). Real time PCR was performed on a 384 well RT 2 Profiler PCR array for Mouse Neuroscience Ion channels and Transporters (PAMM-036, 2011, SA Biosciences) using SA Biosciences RT 2 qPCR Master Mix, and run on an ABI 7900HT Real time instrument. Three separate PCRs were performed, in which cDNA from each age was loaded onto 96 wells of the 384 well PCR plate. Real time PCR was run and analysed according to SA Biosciences recommended protocols, and data analysed using the SA Biosciences web portal data analysis.

Reverse transcription-polymerase chain reaction (RT-PCR)
RNA was extracted from E14.5 freshly dissociated and purified ENCCs, and from adult whole brain as described above. The concentration of total RNA in each sample was measured using a NanoDrop ND-1000 spectrophotometer. cDNA was synthesised using the iScript Advanced cDNA Synthesis Kit for RT-qPCR (Bio-Rad); 100-350ng of total RNA was used in a final reaction volume of 20 μl according to the manufacturer's instructions. Control reactions using no reverse transcriptase or substituting cDNA with water were run in parallel for each tissue.

Catenary culture
Mid and hindgut from E11.5 mice were dissected and strung across a "V" cut into a piece of filter paper [44]. These catenary cultures were placed into 20 μl of TCM in wells of a Terasaki plate. The explants were grown for 48 h at 37°C in control TCM or TCM with drugs (selected at random), and then processed for immunohistochemistry using the ENCC marker, Sox10 [45], and the neuronal marker, Tuj1 [46] (S3 and S4 Tables). The distance migrated was determined as described previously; in brief the distance between the most caudal Sox10+ cell and the ileocaecal junction was measured using Fiji software [47]. These experiments were set up at 3-5 PM.

Neurite outgrowth and cell death assays
The small and large intestines were dissected from E14.5 mice, and dissociated as described previously [38]. Following the addition of TCM and gentle trituration, the cell suspension was filtered through a 40 μm cell strainer, and centrifuged for 2 min at 2000 rpm. Cells were resuspended in TCM and 200 μl aliquots of cells were placed as a drop on sterilised 18 mm round coverslips within the wells of a 12-well plate at a density of 4x10 5 cells/well. The cells were left for 3 h at 37°C to adhere, then the media with any non-attached cells was removed, and replaced with either control TCM or TCM containing drug (selected at random). Cultures were incubated for a further 9 h, and then fixed in 4% paraformaldehyde (PFA) for 15 min before being processed for Tuj1 immunostaining to identify neurons. Images of randomly chosen fields of view were taken using a X10 objective lens, and then the proportions of Tuj1+ cell bodies possessing one or more neurites that were >50% of the diameter of the cell body was counted using the cell counter plug-in on Fiji software. These analyses (imaging and counting) were all done blinded. Some cultures were also grown in the presence of drugs for 21 h prior to fixation. To examine apoptosis, gut from E14.5 Ednrb-hKikGR mice was dissociated, the cells allowed to adhere to coverslips for 3 h and then exposed to drugs for 9 h or 21 h prior to fixation and processing for immunohistochemistry using antisera to activated caspase-3 (S3 and S4 Tables). Random fields of view containing KikGR+ ENCCs were imaged and the percentage of KikGR+ cells that was also activated caspase-3+ determined. Data were analysed using t tests or one way ANOVAs, as appropriate. These experiments were set up at 7-9 AM.

Immunofluorescence
Wholemount preparations of embryonic gut, cultured explants or cultured cells were processed for immunohistochemistry as described previously [38,48] using primary and secondary antisera shown in S3 and S4 Tables.
For Kv7.3 immunolabelling [49], embryonic mouse guts distal to the stomach were dissected quickly, cryopreserved in 20% sucrose and embedded in OCT, which was rapidly frozen in isopentane. Cryostat sections (10 μm) were thaw-mounted onto Superfrost slides and air dried for 3 minutes before being immersed in ice-cold acetone for 10 minutes. Sections were then processed for immunohistochemistry as described above.

E11.5 and E14.5 ENCCs express a variety of ion channels
We performed a QPCR array and compared the expression of ion channels and transporters in E11.5 and E14.5 purified ENCCs. Many ion channels were already expressed at E11.5; over 30 were up-regulated more than 3-fold in E14.5 ENCCS compared to E11.5, three were down-regulated (Table 1), and a further 25 ion channels were expressed at similar levels at both E11.5 and E14.5 (less than 3-fold difference in expression; Table 2). The ion channels that were upregulated were predominately K + , Na + and Ca 2+ channels (Table 1), and those expressed at similar levels at E11.5 and E14.5 included multiple Cl -, K + and Ca 2+ channels and a variety of transporters ( Table 2). We then prioritized several Cl -, K + and Ca 2+ channels that showed substantial gene expression changes, either up or down, between E11.5 and E14.5, and/or that had previously been reported to be expressed by or play a role in migrating cells, for further expression and functional analysis (see below). A number of genes encoding Clchannels were highly expressed at both E11.5 and E14.5, so their functional roles in ENCC migration and neuritogenesis were also examined.
Gene expression levels were also examined in the P0 and adult gut but could not be quantitatively compared to E11.5 or E14.5 ENCCs, because myenteric plexus with attached external muscle layers and serosa was used for P0 and adult gut for technical reasons. Nonetheless, most genes that were expressed in P0 and adult tissue were also expressed at both E11.5 and E14.5 (S5 Table).

Inhibition of Clchannels does not affect ENCC migration or neurite outgrowth
A number of chloride channels were highly expressed at both E11.5 and E14.5 (Table 2). We verified the expression of most of these channels in FACS-sorted ENCCs using RT-PCR ( Fig  1A). Inhibition of Clchannels impairs the ability of glioma cells to migrate by perturbing their ability to make the necessary changes in shape and volume required to migrate through small spaces [23,26,51]. As migrating ENCCs undergo changes in shape to move through the extracellular space between mesenchymal cells and to maintain chain formation [13], we assessed the effect of chloride channel blockers on the migration of ENCC along intact explants of E11.5 gut. NPPB is a non-selective Clchannel blocker, while bumetanide blocks the activity of NKCC1 and NKCC2 cotransporters, which play a major role in Claccumulation [23,52]. Neither NPPB nor bumetanide had a significant effect on the rate of migration of ENCC in cultured explants of gut cultured for 48 h (Fig 1B).
Neuritogenesis was assessed by dissociating the small and large intestines of E14.5 mice; after 3 h to allow cells to adhere to the coverslips, the cells were exposed to bumetanide or NPPB for 9 hours, then fixed and processed for immunohistochemistry using an antibody to Tuj1 to identify neurons. Blocking Clchannels did not affect the proportion of Tuj1+ cells extending neurites in dissociated cultures of E14.5 gut (Fig 1C).

Inhibition of Ca 2+ channels does not affect ENCC migration or neurite outgrowth
The array showed that some Ca 2+ channels were expressed at similar levels by E11.5 and E14.5 ENCCs (Table 2), while others increased expression between E11.5 and E14.5 (Table 1). Of particular interest was the Ca v 2.2 subunit of the N-type Ca 2+ channel, which is involved in neuronal migration and axon guidance in some parts of the nervous system [29,34]. Moreover, we have previously shown that E11.5 enteric neurons respond to electrical field stimulation with Ca 2+ transients that can be blocked by the N-type Ca 2+ channel blocker, ω-conotoxin GVIA [39]. Cacna1b, which encodes Ca v 2.2, was upregulated 4.7 fold between E11.5 and E14.5 (Table 1). Other Ca 2+ channels were also expressed at E11.5 and E14.5, including Cacna1a, which encodes the P/Q channel Ca v 2.1 ( Table 2). Expression of most of the Ca 2+ channels was confirmed by RT-PCR of sorted ENCCs (Fig 2A). Immunohistochemical studies showed expression of Ca v 2.1 (Fig 2B and 2B') and Ca v 2.2 ( Fig 2C and 2C') by neurites of cultured E14.5 enteric neurons. Pharmacological inhibition of N-, P-, L-and T-type Ca 2+ channels using ω-conotoxin GVIA (5 μM), ω-agatoxin IVA (200 nM), nicardipine (2.5 μM) and mibefradil (1 μM) respectively, did not affect ENCC migration in E11.5 gut explants (Fig 2D), nor did it affect neuritogenesis of dissociated, cultured E14.5 gut cells (Fig 2E). As the IC 50 for ω-agatoxin IVA acting Only genes showing an average cycle length less than 30 at either E11.5 or E14.5 are shown. Expression of some genes was confirmed at E14.5 by RT-PCR (bold); expression of KCa3.1 (also called IK1 or IK Ca ), which is encoded by Kcnn4, by ENCCs was also previously reported using immunohistochemistry [50]. * Expression was not confirmed by RT-PCR at E14.5. Expression of other channels was not examined. Adult mouse brain (Br) was used as a positive control, and-RT was a negative control (-). B. Migration assay to assess effects of blocking chloride channels on ENNC migration. Explants of gut were removed from E11.5 mice, when the wavefront of ENCCs (grey) is just beyond the caecum, and grown in culture for 48 hours, during which time the ENCCs migrate into the distal hindgut. The preparations were fixed and the distance from the ileo-caecal junction to the most distal ENCC was then measured (dotted line). There was no significant difference in the distance migrated by ENCCs in explants grown in the presence of bumetanide (n = 9) or NPPB (n = 8) compared to controls (n = 10) (mean ± SEM; one way ANOVA). C. Effects of chloride channel blockers on neurite outgrowth. The small and large intestines from E14.5 mice were dissociated, allowed to adhere to coverslips for 3 hours and then exposed to bumetanide or NPPB for 9 hours. The cells were then fixed and processed for immunohistochemistry using an antibody to Tuj1. There was no significant difference in the percentage of Tuj1+ cells that extended neurites between control and drug-treated cultures (one way ANOVA; a minimum of 1750 Tuj1+ cells was examined from 6 coverslips from 2 experiments). There was no significant difference in the distance migrated by ENCCs in explants grown in the presence of the N-type blocker, ωconotoxin GVIA (n = 9), the P/Q-type blocker, ω-agatoxin IVA (n = 10), the L-type blocker, nicardipine (n = 9) or the T-type blocker, mibefradil (n = 10) compared to controls (n = 11) (mean ± SEM; one way ANOVA; minimum of 2 experiments). E. Effects of calcium channel blockers on neuritogenesis. There was no significant difference in the percentage of Tuj1+ cells that extended neurites between control and drug-treated cultures of E14.5 dissociated gut (for controls and each drug, a minimum of 950 Tuj1+ cells was examined from 5 or 6 coverslips from 2 experiments). doi:10.1371/journal.pone.0123436.g002 at Q-type Ca 2+ channels is 120 nM [53], it is also unlikely that Q-type channels are involved in neuritogenesis or ENCC migration.

Effects of inhibition of potassium channels on ENCC migration and neurite outgrowth
Many potassium channels of various different subfamilies were expressed at E11.5 and most were upregulated significantly by E14.5, including Kcnq3, upregulated 170-fold, and Kcnc4, upregulated 16-fold (Tables 1 and 2). RT-PCR confirmed expression of these and other potassium channels at E14.5 (Fig 3A). Immunohistochemistry was used to localize K v 3.4 (Kcnc4), and K v 7.3 (Kcnq3). Double staining of wholemount preparations of gut from E11.5, E12.5 and E13.5 mice with antibodies to K v 3.4 and the neurite marker, Tuj1, showed that most Tuj1+ neurites close to the ENCC migratory wavefront showed K v 3.4 immunoreactivity (Fig 3B-3B"). However, rostral to the wavefront, only a small sub-population of Tuj1+ neurites was K v 3.4+. Only rarely was K v 3.4 immunostaining detectable in Tuj1+ cell bodies. Frozen sections of embryonic gut immunostained using antibodies to K v 7.3 revealed that Tuj1+ neurites showed K v 7.3 staining (Fig 3C-3C"). Most cells in the outer mesenchymal layer were also K v 7.3+, but were less intensely stained than the neurites (Fig 3C). Studies in the CNS have shown that K v 7.2 and K v 7.3 interact with ankyrin-G at the axon initial segment [54]. We therefore also examined the immunolocalization of ankyrin-G in the E11.5 and E12.5 gut. Double labelling with the neurite marker, Tuj1, showed that ankyrin-G immunostaining was present along the entire length of most neurites, and some nerve cell bodies were also ankyrin-G+ (Fig 3D-3D").
We examined the effects of the non-selective K + channel blockers, TEA and 4-AP [55], as well as the K v 3.4 blocker, BDS-I, and the K v 7 channel blocker, linopirdine, on ENCC migration and neurite outgrowth. The functional roles of K v .4.2, K v 4.3, K v 10.1 and K v 11.1 were not examined in this study.
Previous studies showed that 4 mM 4-AP or 30 mM TEA inhibit the growth of retinal ganglion cells into the Xenopus brain [31]. Following exposure to 5 mM 4-AP or 30 mM TEA for 48 h, ENCCs in explants of E11.5 gut halted their caudal migration (Fig 4A), but there was also a striking decrease in the number of Sox10+ ENCCs (Fig 4B). In dissociated cultures of E14.5 gut grown in the presence of 30 mM TEA or 5 mM 4-AP for 9 h, the proportion of Tuj1+ cells extending neurites was significantly lower than controls (Fig 4D and 4E). Although there was no significant change in the proportion of activated caspase-3+ ENCCs in dissociated gut cultures after exposure to 30 mM TEA or 4 mM 4-AP for 9 h, there was a significant increase in the number of activated caspase-3+ ENCCs after 21 h exposure to the drugs (Fig 4F). The osmolarity of 30 mM TEA in TCM is~16% higher than control TCM; however, the toxic effects of 30 mM TEA is not due to high osmolarity as the migration and survival of ENCCs in intact explants of E12.5 gut grown in TCM with 19% higher osmolality due to added sucrose did not differ from explants grown in control TCM (data not shown). Our data are consistent with very recent studies showing that exposure to high concentrations of TEA (>2 mM) kills HeLa cells [56] and cultured hippocampal neurons [57]. Lower concentrations of TEA (2 mM, 10 mM) or 4-AP (0.1 mM) had no significant effect on ENCC migration (Fig. 4C), neuritogenesis (Fig 4D) or cell death (Fig 4F). Moreover, as in control gut explants, Tuj1+ neurites were closely associated with the most caudal Sox10+ cells. We could not quantify ENCC migration in explants exposed to 30 mM TEA or 5 mM 4-AP due to the small number of surviving Sox10+ cells (see Fig 4A).
Kcnc4 (K v 3.4) is upregulated~16 fold between E11.5 and E14.5 (Table 1), and is expressed by newly extended neurites of young enteric neurons (see above). Although the K v 3 subfamily of potassium channels is blocked by TEA and 4-AP [55], we also examined the effects of BDS-I, a specific blocker of K v 3.4 [58]. BDS-I did not retard ENCC migration (Fig 4C) or neurite formation (Fig 4D).
Kcnq3, which encodes K v 7.3, was upregulated 170 fold between E11.5 and E14.5 ( Table 1). The K v 7 subfamily of channels is not blocked by TEA or 4-AP; we therefore examined the effects of linopirdine, a blocker of K v 7 channels [55], on ENCC migration and neurite extension. Linopirdine had no significant effect on ENCC migration or neuritogenesis (Fig 4C and 4D).

Discussion
Using a PCR-based array and FACS-sorted ENCCs, we showed that a large number of ion channel genes are already expressed by ENCCs at E11.5, and expression of many ion channels increases significantly by E14.5. These ages span the time during which the colon is colonized by ENCCs and a subpopulation of ENCCs starts to differentiate into neurons and extends neurites [3,8,59,60].
The PCR array and RT-PCR studies were conducted on the entire population of ENCCs, including undifferentiated ENCCs and neurons. We therefore do not know whether particular ion channels are expressed by subpopulations of, or all, ENCCs. The up-regulation of numerous ion channels between E11.5 and E14.5 correlates with large increases in both the proportion of ENCCs expressing neuronal markers [7] and in enteric neuron density [9], so it is likely that many of the ion channels upregulated by E14.5 are expressed mainly or exclusively by neurons. This is supported by our immunohistochemical data showing that ion channels encoded by some of the most highly upregulated genes between E11.5 and E14.5 were localized predominantly to neurites; these include K v 7.3 (Kcnq3, upregulated 170 fold), K v 3.4 (Kcnc4, upregulated 16 fold) and Ca v 2.2 (Cacna1b, upregulated 5 fold). However, although Ca v 2.1 immunostaining was also predominantly on neurites, Cacna1a was expressed at similar levels at both E11.5 and E14.5. The ENS is one of the first parts of the nervous system to exhibit electrical activity [61]. We have previously shown that some ENCCs in the mouse gut express voltage-gated Ca 2+ and Na + channels that are involved in the electrical activity of early enteric neurons [35,39].

Ion channels and ENCC migration
Ion channel activity is required for the migration of various types of cells, including glioma cells, platelets, neuroblasts and granule cells in the cerebellum [20,[23][24][25][26][27][28][29]. Interestingly, there are some common mechanisms controlling the migration of CNS neural progenitors and the migration of glioma cells, including roles for several different ion channels [26]. It is thought that one mechanism by which ion channels facilitate cell migration is by promoting hydrodynamic volume and shape changes that enable cells to squeeze between small spaces [20,62]. Since ENCCs undergo rapid changes in shape as they migrate [13,[63][64][65], we hypothesized that ion channel activity might be required for ENCC migration. We found that high concentrations of the non-selective K+ channel inhibitors, TEA and 4-AP, halted ENCC migration in cultured explants, but killed the vast majority of ENCCs. In contrast, addition of drugs that Expression of potassium channels and ankyrin-G by ENCCs. A. RT-PCR confirming expression of transcripts encoding 21 potassium channels in purified (FACS-sorted) ENCCs from E11.5 gut. Adult mouse brain (Br) was used as a positive control and-RT was a negative control (-). Dotted lines indicate different gels. B. Immunoreactivity for K v 3.4 (B) (encoded by Kcnc4) in a wholemount preparation of E13.5 gut overlaps with Tuj1 immunostaining (B', B"). C. Frozen transverse section of E14.5 gut: Immunoreactivity for K v 7.3 (encoded by Kcnq3) overlaps with Tuj1 immunostaining of neurites (arrows, C-C"). Kv7.3 staining is also present in outer mesenchymal cells (out mes), but is not detectable in the mucosa (C). D. Immunolocalization of ankyrin-G in the E11.5 gut. Ankyrin-G immunostaining was present along the entire length of most Tuj1+ neurites (D-D"), and some nerve cell bodies were also ankyrin-G+ (asterisk).
doi:10.1371/journal.pone.0123436.g003 block the activities of a range of Ca 2+ , Cland K + channels at concentrations that did not cause cell death did not perturb the migration of ENCCs in gut explants. The lack of effect of the channel-blocking drugs on ENCC migration is unlikely to be due to poor penetration of the drugs as intact E11.5 gut explants are permeable to antibodies and large (250 kDa) toxins [47,66], and often similar concentrations of drugs are used to affect ENCCs in intact embryonic gut explants to those used in dissociated ENCCs [64,67]. We had previously shown that tetrodotoxin, which blocks voltage-gated sodium channels, and clotimazole, which blocks the intermediate conductance Ca 2+ -dependent K + channel (K Ca 3.1), also do not affect ENCC migration [50]. It is possible that it is necessary to block multiple ion channels simultaneously to perturb ENCC migration and/or other types of ion channels not tested are involved in ENCC migration. For example, Kv11.1 is involved in the migration of some non-neuronal cell types [68], but its function was not examined in the current study. It is also feasible that ENCCs do not require ion channel activity but use other mechanisms to migrate through the gut mesenchyme. For example, blocking matrix metalloproteinase activity has been shown to retard ENCC migration [69].

Ion channels and neurite extension
At the same time as ENCCs are migrating along the gut, a sub-population of cells starts to differentiate into neurons and extends neurites [10,12,59]. During formation of the ENS, it is essential that developing neurons project their axons in the correct direction and to the correct targets. Little is known about the mechanisms controlling axon extension and navigation in the developing ENS, although the planar cell polarity (PCP) pathway has been shown to play a role in the direction in which axons project along the gut in vivo [12], and retinoic acid retards neurite extension from enteric neurons in vitro [70]. Ion channels, including Ca 2+ channels [32,34], Clchannels [33] and K + channels [71], have been reported to regulate axon outgrowth in a variety of regions of the nervous system.
Many voltage-dependent K + channels are expressed in the developing Xenopus brain [71], and K v 3.4 is expressed by pioneer axons in the developing rat forebrain [72]. Similarly, we showed that K v 3.4 immunoreactivity in the developing gut was most prominent on neurites close to the migratory wavefront.
were close to the end of the explants. In explants grown in the presence of 30 mM TEA, there were very few Sox10+ ENCCs caudal to the caecum (middle panel). The red staining in the distal gut of the control explant is non-specific staining of the gut epithelial cells. Scale bars: 200 μm. B. High magnification images of the pre-caecal regions of E11.5 explants following 48 h in culture. There was a dramatic reduction in the number of Sox10+ cells (green) in gut explants grown in the presence of 30 mM TEA compared to control explants and explants grown in the presence of 2 mM TEA. Scale bars: 20 μm. C. Quantification of the location of the most caudal Sox10+ ENCC in E11.5 explants cultured for 48 h. Preparations exposed to 30 mM TEA or 5 mM 4-AP were not analysed due to the small number of Sox10+ cells in the post-caecal gut. There was no significant difference in the distance migrated by ENCCs in explants grown in the presence of 2 mM TEA (n = 8), 0.1 mM 4-AP (n = 10), the K v 7 blocker, linopirdine (n = 11) or the K v 3.4 blocker, BDS-I (n = 13) compared to controls (n = 12 or n = 14 for BDS-I experiments) (mean ± SEM; one way ANOVA; minimum of 2 experiments). The experiments using BDS-I were done separately from the other experiments and so had separate controls. Measurements were not performed on explants exposed to 30 mM TEA or 5 mM 4-AP due to the small number of surviving Sox10+ cells. D. Effects of potassium channel blockers on neurite outgrowth. The small and large intestines from E14.5 mice were dissociated, allowed to adhere to coverslips for 3 hours and then exposed to drugs for 9 h. The cells were then fixed and processed for immunohistochemistry using an antibody to Tuj1. A lower proportion of Tuj1+ cells extended neurites in the presence of 30 mM TEA or 5 mM 4-AP compared to controls (one way ANOVA followed by Tukeys test; for controls and each drug treatment, a minimum of 680 Tuj1+ cells was examined from 6 coverslips from 2 experiments). E. Examples of Tuj1+ cells in cultures of dissociated E14.5 gut grown under control conditions, or in the presence of 5 mM 4-AP or 30 mM TEA for 9 h. Under control conditions, most Tuj1+ cells extended neurites, whereas in the presence of 5 mM 4-AP or 30 mM TEA, many Tuj1+ cells did not possess any neurites (arrows). Scale bar: 20 μm. F. Gut from E14.5 Ednrb-hKikGR mice was dissociated, the cells allowed to adhere to coverslips for 3 hours and then exposed to drugs for 9 h or 21 h prior to fixation and processing for immunohistochemistry using the apoptosis marker, activated caspase-3. Although there was no difference in the proportions of KikGR+ ENCCs that were also activated caspase-3+ following 9 h exposure to 5 mM 4-AP or 30 mM TEA (one way ANOVA), there was a significant increase in activated caspase-3+ ENCCs after 21 h exposure to high concentrations of TEA or 4-AP (one way ANOVA followed by Tukeys test). Exposure of dissociated ENCCs to high concentrations of the non-selective K + channel blockers, TEA (30 mM) and 4-AP (5mM) for 9 hours resulted in a smaller percentage of Tuj1+ cells extending neurites in vitro. However, although similar concentrations of these drugs have been used previously in studies of axon growth and guidance in the Xenopus visual system, we found that TEA and 4-AP at high concentrations caused death of ENCCs; no increase in cell death (activated caspase-3+ cells) was detectable after only 9 hours exposure to 30 mM TEA or 5 mM 4-AP, but exposure to the drugs for 21 hours did result in a significant increase in cell death. Furthermore, there was a dramatic reduction in the number of ENCCs in intact explants of E11.5 gut following exposure to 30 mM TEA or 5 mM 4-AP for 48 h for the ENCC migration assays (see above). It is therefore highly likely that the decreased neuritogenesis observed after 9 h exposure to 30 mM TEA or 5 mM 4-AP is due to ENCCs being unhealthy. Other recent studies using HeLa cells and cultured hippocampal neurons have also reported that exposure to high concentrations of TEA for 24 h or more results in cell death [56,57]. Proteomics analyses of HeLa cells exposed to high concentrations of TEA revealed changes in proteins involved in a variety of biological functions including oxidative stress responses, protein synthesis and degradation, metabolism and signal transduction [56].
Decreased neuritogenesis was observed after 9 hours exposure to high concentrations of TEA or 4-AP, but no significant cell death was detected at this time point. Although we think the decreased neuritogenesis observed after 9 hours is most likely due to the cells being unhealthy, we cannot rule out the possibility that the perturbed neuritogenesis is a primary effect of exposure to high concentrations of TEA or 4-AP, with prolonged drug exposure then leading to extra effects and ultimately cell death.
Different neurochemical classes of neurons are present in the E14.5 gut [9,50]. Although there was an overall reduction in the proportion of Tuj1+ cells extending neurites after 9 hours exposure to high concentrations of TEA or 4-AP, some neurons still extended neurites (see Fig  4E). As Tuj1 is a pan-neuronal marker, it is possible that different classes of neurons are differentially sensitive to TEA and 4-AP.
Despite numerous studies showing roles for ion channel activity in neurite extension in a variety of regions of the nervous system, we were unable to demonstrate a role for ion channel activity in neuritogenesis in cultured embryonic enteric neurons. We also did not observe any obvious changes to the projections of Tuj1+ neurites in intact explants of gut in the ENCC migration assays. We cannot rule out the possibilities that other ion channels are involved, that it is necessary to block multiple ion channels simultaneously to perturb neurite formation or that ion channels play a role in enteric neuritogenesis (both axon and dendrites) in vivo but not in vitro. The axons of enteric neurons have to navigate to specific targets, including the external muscle and other ganglia and it is possible ion channels are involved in axon navigation or synaptogenesis in vivo. Furthermore, enteric neurons undergo significant morphological changes in their dendrites during development [41] but the mechanisms underlying these dendritic changes are unknown.

Ion channels and electrical excitability
We have previously shown functional roles for several Na + , Ca 2+ and K + channels in the development of electrical excitability in early enteric neurons [35,39], and it is possible that the vast majority of ion channels expressed in the fetal ENS are involved in the excitability of all, or specific subtypes of, enteric neurons [60].
In the CNS, ankyrin-G is concentrated at axon initial segments and plays an essential role in the clustering and function of K v 7.2 and K v 7.3 [54]. In the current study, we showed that ankyrin-G and K v 7.3 were expressed by many embryonic enteric neurons, but appeared to be present along the entire length of the neurites. Na v 1.6 has been reported to localize to the axon initial segments of enteric neurons in the colon of adult guinea-pigs [73], but little else is known about the axon initial segments in the ENS. Even though embryonic enteric neurons can fire action potentials, our ankyrin-G data suggest that they may not yet possess axon initial segments.

Conclusions
Our array data showed that many ion channels are expressed early during the development of the ENS. Many ion channels are likely to be involved in the development of neuronal electrical excitability, but it remains to be determined which ion channels are expressed by undifferentiated ENCCs and whether they also play functional roles.
Supporting Information S1 Table. Primer sequences and the TD-PCR cycling program used for ion channel amplification.