Influence of bacterial components on the developmental programming of enteric neurons

Abstract Background Intestinal bacteria have been increasingly shown to be involved in early postnatal development. Previous work has shown that intestinal bacteria are necessary for the structural development and intrinsic function of the enteric nervous system in early postnatal life. Furthermore, colonization with a limited number of bacteria appears to be sufficient for the formation of a normal enteric nervous system. We tested the hypothesis that common bacterial components could influence the programming of developing enteric neurons. Methods The developmental programming of enteric neurons was studied by isolating enteric neural crest‐derived cells from the fetal gut of C57Bl/6 mice at embryonic day 15.5. After the establishment of the cell line, cultured enteric neuronal precursors were exposed to increasing concentrations of a panel of bacterial components including lipopolysaccharide, flagellin, and components of peptidoglycan. Key Result Exposure to bacterial components consistently affected proportions of enteric neuronal precursors that developed into nitrergic neurons. Furthermore, flagellin and D‐gamma‐Glu‐mDAP were found to promote the development of serotonergic neurons. Proportions of dopaminergic neurons remained unchanged. Proliferation of neuronal precursor cells was significantly increased upon exposure to lipopolysaccharide and flagellin, while no significant changes were observed in the proportion of apoptotic neuronal precursors compared to baseline with exposure to any bacterial component. Conclusions and Interfaces These findings suggest that bacterial components may influence the development of enteric neurons.


| INTRODUCTION
The enteric nervous system (ENS) plays a critical role in maintaining gut homeostasis and mediating interactions between the contents of the intestinal lumen and the internal environment of cells and tissues. The ENS participates in the regulation of blood flow (Neild et al., 1990), initiation of appropriate responses to sensory stimuli (Burns et al., 2016), crosstalk with the immune system , and controlling gut motility . Normal development of the ENS in early life is, therefore, critical for the postnatal function of the gastrointestinal tract.
As a result of its highly integrative and complex role, ENS development involves the tight regulation of precursor cell differentiation, neurite growth, and establishment of appropriate neural networks. Colonization of the gut is mediated by two populations of enteric neural crest-derived cells (ENCDCs) with unique patterns of migration (Burns and Le Douarin, 1998;Le Douarin & Teillet, 1973). Coordinated migration is essential for appropriate innervation throughout the gastrointestinal tract, with defects in these pathways giving rise to regions of aganglionosis and pathological manifestations (Young et al., 2006).
Enteric neurons of different phenotypes are born at different times of development (Pham et al., 1991). ENCDCs that colonize the stomach and proximal small intestines are the first to develop into phenotypes that define the mature ENS, and begin to arise as the migrating wavefront of ENCDCs continues to invade the caudal most regions of the gut (Conner et al., 2003;Young et al., 1999). Serotonergic, or 5-hydroxytryptamine (5-HT), neurons are considered early-born and their appearance coincides with the first wave of ENCDC invasion of the foregut (Bergner et al., 2014;Pham et al., 1991). Dopaminergic (TH) and nitrergic (nNOS) neurons both have dual populations of earlier and later-born neurons, with ongoing development into the postnatal period (Bergner et al., 2014;Branchek & Gershon, 1989;Hao et al., 2010;Li et al., 2004;Pham et al., 1991;Sang & Young, 1996).
This critical window of proliferation and phenotypic development of enteric neural precursors in the perinatal period coincides with microbial colonization of the gut. Previous work has shown that enteric neurons of germ-free (GF) mice exhibit structural (Collins et al., 2014) andelectrophysiological (McVey Neufeld et al., 2015) abnormalities which may be reversed upon conventionalization with specific pathogen-free (SPF) microbiota. Particularly intriguing are findings that demonstrate these abnormalities arise as early as postnatal day 3 in GF mice compared to altered Schaedler flora (ASF) mice (Collins et al., 2014). Further work has found that intestinal microbiota is not limited in its effects on influencing only the neurons of the ENS; instead, it also serves as an essential component for the normal development of the mucosal glial cell network (Kabouridis et al., 2015). It has been shown that GF mice exhibit a significant reduction in the number and density of enteric glial cell connections and that conventionalization after the perinatal period normalizes the glial cell networks (Kabouridis et al., 2015). These early life disruptions in ENS physiology support the concept that input from even a simple intestinal flora is sufficient for influencing ENS development.
While research to date suggests that intestinal microbiota can play a role in the normal development of the ENS, the mechanisms by which this occurs have not yet been elucidated. The current study tested the hypothesis that bacterial components interact directly with enteric neuronal precursors, and directly mediate the developmental programming of enteric neurons. We report that exposure of neuronal precursors to a panel of common bacterial components appears to influence developmental programming through direct interactions in vitro. Specifically, bacterial components appear to promote development into serotonergic and nitrergic, but not dopaminergic, phenotypes. Bacterial components stimulate proliferation under certain conditions; however, apoptosis remains unaffected. These findings suggest that bacterial components can influence the development of enteric neurons.

| Animals
Timed pregnant SPF C57Bl/6NTac mice were ordered from Taconic Biosciences, Inc. (Rensselaer, NY, USA), and maintained on ventilated racks in ultraclean units in the McMaster Central Animal Facility. The day of plug detection was considered E1. Timed pregnant dams (n = 6) were anesthetized using isoflurane, sacrificed by cervical dislocation, and male and female fetuses were dissected at E15.5. Treatment of animals and all experiments were conducted in accordance with the McMaster Animal Research Ethics Board (AUP: 140621).

| Preparation of cultured cells for immunocytochemistry
Cultured cells were grown in chamber slides at a density of 275,000 cells mL -1 of NSM + media. Cells were fixed in 4% paraformaldehyde (freshly prepared from paraformaldehyde, pH 7.4), and permeabilized by incubating in blocking buffer (1% PBS, 4% normal horse serum, 0.5% Triton TM X-100). Cells were exposed to primary antibody overnight at room temperature. Subsequent treatment with secondary antibodies for 3 hr at room temperature in the dark allowed for the visualization of targetted cells.

| Detection and quantification of apoptotic cells
Apoptosis was detected using terminal deoxynucleotidyl transferase (TdT)-mediated 2'-deoxyuridine 5'-triphosphate (dUTP) nick end labeling kit (TUNEL; Promega DeadEnd TM Fluorometric TUNEL System G3250, used according to manufacturer's instructions). This assay involves TdT binding to blunt ends of DNA fragments and catalyzing the incorporation of fluorescently labeled nucleotide, fluorescein-12-dUTP. Total apoptotic cells were manually counted and compared to total HuC/D positive cells. Bisbenzimide was used as a nuclear counterstain. The proportion of apoptotic cells after exposure to bacterial components were compared to NSM + vehicle control. Treatment with DNase (Promega M6101) was performed for the positive control, and the reverse TdT enzyme was omitted for the negative control.

| Exposure to bacterial components
Bacterial components were chosen to represent a range of common intestinal microbiota: lipopolysaccharide (LPS) is a component found in gram-negative bacteria, flagellin can be found in both gram-negative and gram-positive bacteria and the peptidoglycan derivatives muramyl dipeptide (MDP) and D-gamma-Glu-mDap (iE-DAP) are found primarily in gram-positive bacteria.

| Image analysis
Quantification of enteric neuronal precursors was performed by photographing five regions per slide at 20x magnification using a Leica DMRXA2 microscope with fluorescence (Leica Microsystems Inc, Concord, ON, Canada). Cells were manually counted using Volocity software (Improvision Inc., Montreal, QC, Canada) on a Macintosh computer (Apple Computers, Markham, ON, Canada). All images were coded to ensure investigator blinding of experimental conditions during image analysis.

| Statistical analyses
Data are presented as mean ± SEM. Statistical analyses were performed using a Mann-Whitney U with comparisons between culture characteristics of subculture 4 and 5. Data analyses of the effects of varying concentrations of bacterial components on culture characteristics were performed using Kruskal Wallis with Dunn's post hoc test. Significant outliers were removed. The level of statistical significance was set at p ≤ .05. All analyses were performed using GraphPad Prism Version 4 for Mac OS X (GraphPad Software Inc., La Jolla California, USA).

| Assessment of ENCDC culture purity and viability using flow cytometry
The duodenum, jejunum, ileum, and colon segments were harvested from fetuses at E15.5, dissociated into single cells, and immunolabeled using anti-p75 NTR as a marker for enteric neuron precursors (Binder et al., 2015). Cells were grown in culture for three subcultures to generate sufficient enteric neuronal precursors and allow development similar to a perinatal timepoint (Bergner et al., 2014). Flow cytometry analyses were conducted on subcultures 3 and 4 to determine culture purity ( Figure 1a) and culture viability using 7-AAD viability dye (data not shown). Analyses revealed a 95.1% p75 NTR+ -positive population ( Figure 1a) with a 97.3% viability (subculture 3). Similar findings were found when analysis was conducted on the subsequent culture (subculture 4), demonstrating 96.1% p75 NTR+ -positive cells ( Figure 1a) and a 98.2% viability.

| Characterization of ENCDC cultures using immunocytochemistry
Cultures were processed for immunocytochemistry to further identify proportions of p75 NTR+ cells, serving as a marker of enteric neuron precursors (Binder et al., 2015) (Figure 1b). The cultures were also characterized at this baseline timepoint by assessing the extent of differentiation and proliferation (Crone et al., 2003) (Figure 1c). All conditions were co-stained with pan-neuronal marker HuC/D as a control for neuronal differentiation, and bisbenzimide nuclear counterstain. Cells co-expressing p75 NTR+ and HuC/D composed the largest population (99.3% ± 1.18%), followed by nitrergic neurons (5.83% ± 1.77%), and serotonergic neurons (1.60% ± 0.850%). No dopaminergic neurons were detected. Proliferating precursor cells were present at 4.46% ± 1.56%.
Cultures treated with MDP demonstrated a significant increase in the proportion of serotonergic neurons between 0.01 µg/ml and 1 µg/ml of MDP (p = .0178), but no significance was compared to vehicle control (Figure 2c). A F I G U R E 1 Characterization of ENCDC cultures. Flow cytometry of ENCDC cultures stained with antibodies against the immature neuronal marker, p75 NTR+ -FITC demonstrated 95.1% p75 NTR+ -positive, 97.3% viability in subculture 3, and 96.1% p75 NTR+ -positive, 98.2% viability in subculture 4 (a). Proportion of ENCDC cultures characterized for expression of p75NTR+, pH3, 5-HT, nNOS, and TH. Scale bar = 20 µm (b). There was a significant increase in the proportion of serotonergic neurons and nitrergic neurons in subculture 5 (gray) compared to subculture 4 (black), but no significant difference in proliferating cells or dopaminergic neurons across the cultures. *p ≤ .05. Values are presented as mean ± SEM (c) significant increase in nitrergic neurons was found between vehicle control and 0.1 µg/ml of MDP (p = .0038), and vehicle control and 1 µg/ml of MDP (p = .0401). There was no significant difference in the proportion of proliferating cells at any concentration of MDP exposure.
Following iE-DAP exposure, there was a significant increase in the proportion of serotonergic neurons between vehicle control and 10 µg/ml of iE-DAP (p = .0233) (Figure 2d). This increase was also found between 1 µg/ml of iE-DAP and 10 µg/ml of iE-DAP (p = .0341). There was a significant increase in the proportion of nitrergic neurons between vehicle control and 0.1 µg/ml of iE-DAP (p = .0151), and vehicle control and 1 µg/ml of iE-DAP (p = .0153). There was no significant difference in the proportion of proliferation cells at any concentration of iE-DAP tested.
No significant differences were found in apoptosis or the proportion of dopaminergic neurons at any concentration of bacterial components tested.

| DISCUSSION
We tested the hypothesis that bacterial components directly interact with enteric neuronal precursors and influence the developmental programming of enteric neurons. We generated a model of enteric neuronal precursors in culture consistent with a perinatal period of development (Bergner et al., 2014). Subsequent treatment of the cultures with bacterial components provided evidence for direct interactions with developing enteric neurons. Specifically, the development of nitrergic neurons appears to be affected by all bacterial components tested in this study, including LPS, flagellin, and peptidoglycan derivatives, MDP, and iE-DAP. Furthermore, flagellin and iE-DAP appear to increase the proportion of serotonergic neurons. Finally, exposure to LPS and flagellin appear to stimulate proliferation in a population of enteric neuronal precursors.
Serotonergic neurons have been shown to be among the first phenotype of enteric neurons to emerge in the developing gut (Bergner et al., 2014;Pham et al., 1991). In contrast, dopaminergic neurons appear to consist of a heterogeneous population of both, early and late-born neurons (Bergner et al., 2014). A population of transient catecholaminergic (TC) neurons was first shown to arise in the foregut at E9.5 (Baetge & Gershon, 1989;Teitelman et al., 1981), which is ultimately replaced by mature neurons by E14-E15 (Baetge & Gershon, 1989). More recent evidence demonstrates the most abundance of dopaminergic neurons between E13.5-E15.5 F I G U R E 2 Role of LPS, flagellin, MDP, and iE-DAP on ENCDC programming. LPS had a significant effect on increasing enteric neuronal precursor proliferation and development into nitrergic neurons, but no significant effect on apoptosis or development into dopaminergic neurons. LPS exhibited a significant effect on increasing serotonergic neurons although this was not significant with the control (a). Flagellin had a significant effect on increasing enteric neuronal precursor proliferation and development into serotonergic and nitrergic neurons. Proliferation appeared to be concentration-dependent. Flagellin had no significant effect on apoptosis or development into dopaminergic neurons (b). MDP had a significant effect on promoting the development of nitrergic neurons. MDP had no significant effect on enteric neuronal precursor proliferation, apoptosis, or development into dopaminergic neurons. Development of serotonergic neurons was only significant between 0.01 µg/ml and 1 µg/ml, but not compared to the vehicle control (c). iE-DAP had a significant effect on promoting the development of serotonergic and nitrergic neurons. iE-DAP did not have a significant effect on enteric neuronal precursor proliferation, apoptosis, or development into dopaminergic neurons (d). *p ≤ .05. Values are presented as mean ± SEM (Bergner et al., 2014;Li et al., 2004), followed by a second smaller peak around P0, with continued development until approximately P10 (Bergner et al., 2014). Similarly, the emergence of nitrergic neurons spans both, prenatal and postnatal timepoints, with the first subset of mature neurons emerging at E11.5 (Bergner et al., 2014;Hao et al., 2010), and completion of development at P10 (Bergner et al., 2014). When we compare actual percentages of enteric neurons from our subcultures to previous detailed work in whole-mount preparations, we note that our nitrergic (10.00% ± 2.96%) and dopaminergic (0.180% ± 0.410%) populations in subculture 5 are consistent with the approximately 13.0% nitrergic neurons and < 1% dopaminergic neurons found at P0 (similar data at P0 not provided for serotonergic neurons) (Bergner et al., 2014). Given that the baseline characteristics of our cultures were suggestive of a perinatal timepoint, our findings suggest that populations of serotonergic and nitrergic neurons continue to be plastic and amenable to influences by intestinal bacteria past their previously established birthdate timelines.
We did not observe any significant effects of bacterial components on the development of dopaminergic neurons. This suggests that dopaminergic neurons may not be as responsive to microbial components as serotonergic and nitrergic neurons; instead, their development might be influenced by other endogenous factors. It has previously been shown that 5-HT may affect the development of perinatal dopaminergic neurons (Fiorica-Howells et al., 1998;Li et al., 2004), as mice lacking tryptophan hydroxylase 2 (TPH2) expression, an enzyme involved in serotonin synthesis, exhibited a reduction in dopaminergic populations (Li et al., 2011). Furthermore, 5-HT dependent increases in the total numbers of enteric neurons in culture have also been shown, suggesting that the effects of 5-HT on promoting enteric neuron survival and development are not restricted to dopaminergic neurons (Li et al., 2011). Indeed, 5-HT may also have effects on neurons born later in the developmental period such as GABAergic, CGRP-expressing, and even a subset of nitrergic neurons born after E15.5 (Li et al., 2011).
Enteric neuronal precursors that express both HuC/D and a marker of proliferation (pH3) comprise an intriguing population. We interpret this population of cells as being neuronal precursors that have yet to exit the cell cycle. Populations of early and late-born neurons arise due to variations in the timing of cell cycle exit (Chalazonitis et al., 2008;Pham et al., 1991). Hu proteins have been shown to affect neuronal differentiation (Akamatsu et al., 2005;Ratti et al., 2006;Sakakibara & Okano, 1997), with the further suggestion that HuD might commence its expression in proliferative neuronal progenitor cells and then become a driver of cell cycle exit (Akamatsu et al., 2005). We hypothesize, therefore, that the cells expressing HuC/D and pH3 are committed neuronal precursors at the transition from final phases of mitosis and nearing cell cycle exit with development into specific neuronal phenotypes. As our current study suggests that this population of neuronal precursors can be influenced by external factors, and in particular by LPS and flagellin, further studies are needed. It would be interesting to determine, for example, if analogous to the ability of bone morphogenetic protein to regulate enteric phenotype diversity by promoting the exit of precursors from the cell cycle (Chalazonitis et al., 2008), bacterial components might influence enteric phenotype diversity by delaying cell cycle exit and thus the timing of neuronal birth dates. Indeed, previous work has suggested that different enteric subtypes might be specified by extrinsic factors acting during the time period of enteric precursor exit of the cell cycle (Bergner et al., 2014).
Finally, we demonstrated a lack of enteric neuronal precursor apoptosis across all bacterial component conditions and concentrations. These findings consist of the previously established concept that apoptosis is not an inherent characteristic of ENCDC maturation and ENS development, and that myenteric neurons are largely static until factors such as the onset of disease result in cell death (Kulkarni et al., 2017). A marked lack of apoptosis at a range of developmental stages from E12.5 to adult has been demonstrated using cleaved caspase-3 staining in mouse myenteric plexus whole-mount preparations from mice (Gianino et al., 2003). Using a different approach, a limited response in apoptosis to neurotrophin-3 depletion on cultured ENCDCs has been demonstrated with no disturbances in the proportion of viable neurons using the TUNEL method (Chalazonitis et al., 2001).
While apoptosis is not a prominent process in the early development of the ENS, recent work suggests that apoptosis could play a role in the adult ENS. Kulkarni and colleagues (2017) showed that approximately a tenth of all myenteric neurons are tagged at all times for caspase-3 cleavage and that roughly a third of these cells undergoes apoptosis within 7 days (Kulkarni et al., 2017), which is consistent with previous findings in other neuron types (Schreiber et al., 2007). Interestingly, apoptosis has also been detected in pre-enteric vagal neural crest cells (NCCs) of an embryonic chick model, as they migrate toward the foregut and prior to foregut invasion (Wallace et al., 2009). Target cells were identified by immunohistochemistry with NCC marker HNK-1 and either TUNEL or activated caspase-3 to confirm apoptosis in NCCs that give rise to the ENS (Wallace et al., 2009). Taken together, these findings indicate that apoptosis is not a typical characteristic of enteric precursors during development after the invasion of the foregut; however, apoptosis may be a normal process in ENCDC precursors migrating from the vagal neural crest as well as adult myenteric neurons, as is typical of other nerves within the nervous system.
The findings presented here suggest that exposure of enteric neuronal precursors to bacterial components in perinatal development may directly influence the developmental programming of a subset of enteric neurons, particularly serotonergic and nitrergic neurons. We did not detect a significant influence on dopaminergic neurons; however, in accordance with previous research which demonstrated that serotonin may influence the proportions of dopaminergic neurons, it remains to be determined whether changes in chemical coding in early life could exert longer term effects. While our current work has focused on the development of enteric neurons, we recognize that enteric glia might also be similarly influenced by bacterial components (Kabouridis et al., 2015); this hypothesis could be an exciting area of future study. Overall, our work highlights the perinatal period as a critical timepoint during which the bacterial milieu can have direct effects on developing enteric neurons in early life, with the potential for longer term consequences, the cellular mechanisms of which remain to be elucidated.