Deletion of choline acetyltransferase in enteric neurons results in postnatal intestinal dysmotility and dysbiosis

Abstract

Acetylcholine (ACh)-synthesizing neurons are major components of the enteric nervous system (ENS). They release ACh and peptidergic neurotransmitters onto enteric neurons and muscle. However, pharmacological interrogation has proven inadequate to demonstrate an essential role for ACh. Our objective was to determine whether elimination of ACh synthesis during embryogenesis alters prenatal viability, intestinal function, the neurotransmitter complement, and the microbiome. Conditional deletion of choline acetyltransferase (ChAT), the ACh synthetic enzyme, in neural crest–derived neurons (ChAT-Null) was performed. Survival, ChAT activity, gut motility, and the microbiome were studied. ChAT was conditionally deleted in ENS neural crest–derived cells. Despite ChAT absence, mice were born live and survived the first 2 wk. They failed to gain significant weight in the third postnatal week, dying between postnatal d 18 and 30. Small intestinal transit of carmine red was 50% slower in ChAT-Nulls vs. WT and ChAT-Het. The colons of many neonatal ChAT-Null mice contained compacted feces, suggesting dysmotility. Microbiome analysis revealed dysbiosis in ChAT-Null mice. Developmental deletion of ChAT activity in enteric neurons results in proximal gastrointestinal tract dysmotility, critically diminished colonic transit, failure to thrive, intestinal dysbiosis, and death. ACh is necessary for sustained gut motility and survival of neonatal mice after weaning.—Johnson, C. D., Barlow-Anacker, A. J., Pierre, J. F., Touw, K., Erickson, C. S., Furness, J. B., Epstein, M. L., Gosain, A. Deletion of choline acetyltransferase in enteric neurons results in postnatal intestinal dysmotility and dysbiosis.

A high proportion of neurons in the enteric nervous system (ENS) exhibit a cholinergic phenotype. Acetylcholine (ACh) is found in cholinergic motor-, inter-, and sensory neurons (1). These neurons can be identified by expression of choline acetyltransferase (ChAT), the enzyme responsible for ACh synthesis. In mice, cholinergic enteric neurons appear early in development at embryonic day (E)10.5 and increase in number to reach adult proportions between E13.5 and postnatal day (P)0 (2). ACh plays an integral role in propulsion of intestinal content (3, 4). This is seen in the peristaltic reflex, which is triggered by stimulation of sensory neurons that synapse on descending and ascending interneurons, both of which use ACh as a neurotransmitter. Descending interneurons project distally from the stimulus site and synapse on inhibitory motor neurons, which relax the smooth muscle in preparation for a forthcoming contraction. Ascending interneurons project proximally and synapse on excitatory motorneurons, which initiate a contractile response that travels aborally (5, 6).

Despite the large number of cholinergic neurons in the ENS, both fast and slow excitatory transmission at neuro-neuronal synapses have significant residual activity after block of nicotinic (fast) or muscarinic (slow) transmission (7–10). Excitatory transmission to the muscle also exhibits a significant component of transmission that is resistant to the block of cholinergic receptors (10–14). Moreover, after cholinergic transmission is blocked there is recovery of nerve-mediated peristaltic reflex activity (13). This suggests that other transmitters can substitute for ACh at neuro-neuronal and neuromuscular synapses in the intestine. In addition, in humans, muscarinic antagonists can depress, but do not block, motility in the intestine (15). In agreement with this, evidence has accumulated for the roles of other excitatory transmitters, including 5-hydroxytryptamine and tachykinins at neuro-neuronal synapses (16–18) and tachykinins at neuromuscular synapses (6, 19).

Resistance of transmission to pharmacological block may be because antagonists do not achieve sufficient concentrations to surmount fully the actions of neurotransmitter at synapses. In view of these data, the purpose of this study was to determine the dependence of motility control on ACh by conditional deletion of the gene for the ACh synthetic enzyme ChAT. We show that ChAT activity is absent in the gut of neural crest cell conditional ChAT knockout (ChAT-Null) mice but that other enteric neurotransmitter phenotypes [calcitonin gene-related peptide (CGRP) and substance P] continue to be present. ChAT-Null animals demonstrate significantly reduced proximal intestinal motility compared with control animals, and the colons of many ChAT-Null mice became impacted with fecal material. These animals fail to gain weight during the third postnatal week, and most die by P30. Furthermore, ChAT-Null animals displayed specific bacterial taxonomic changes in their cecal and colonic microbiota. These data provide a direct in vivo demonstration of the effect of the absence of ACh on intestinal motility.

MATERIALS AND METHODS

Animals

All animal procedures were approved by the University of Wisconsin Animal Care and Use Committee (M01392) or by the University of Tennessee Health Science Center Animal Care and Use Committee (16-021). Mice carrying loxP sites flanking exons 3 and 4 of the gene encoding ChAT were provided by Dr. Joshua Sanes (Harvard University, Cambridge, MA, USA). The production and properties of these mice have been described (20). Floxed-ChAT animals were mated with mice expressing Cre recombinase under the control of a Wnt-1 enhancer element as well as the fluorescent reporter gene tdTomato in the ROSA26 locus with an upstream STOP cassette flanked by LoxP sites (2, 21). Herein, animals homozygous for the mutant ChAT allele are referred to as ChAT-Null, heterozygotes are referred to as ChAT-Het, and wild types are referred to as WT. Animals were genotyped by PCR amplification of DNA isolated from tail biopsies using previously described primers (20), which differentiate the ChAT-Null and WT alleles. Control mice were identified by the lack of expression of Cre recombinase and were additionally negative for tdTomato expression.

Dissection of tissues

Mice (age P22–26) were anesthetized with isoflurane and then humanely euthanized by cervical dislocation. The gastrointestinal (GI) tract was removed and placed in ice-cold PBS along with the remainder of the body. The small intestine and colon were measured, and the luminal contents were flushed with PBS. After removal of the mesentery and external nerve fibers, the myenteric ganglia and accompanying smooth muscle were dissected from the rest of the intestinal wall from the distal ileum and midcolon. Segments of spinal cord and brain were also removed for analysis. Dissected tissues were either immediately frozen on dry ice and stored at −80°C in microcentrifuge tubes or fixed in 4% paraformaldehyde overnight at 4°C in preparation for immunostaining.

Frozen tissues, generally 20–50 mg, were weighed and then homogenized in a 0.2-ml tapered ground glass tissue grinder (357848; Wheaton, Millville, NJ, USA) combined with small amounts (50–100 μl) of homogenization buffer [50 mM Tricine (pH 8.1), 5 mM EDTA, 5 mM 2-ME, 1 mM sodium azide] to a total of 2.5 volumes (tissue w/v). The homogenate was transferred to a 1.5-ml Eppendorf tube (MilliporeSigma, Burlington, MA, USA) and thoroughly mixed with an equivalent amount of homogenization buffer containing, in addition, 1% Triton X-100 and 10% glycerol. After 30–60 min on ice, the homogenate was spun at 14,000 rpm for 5 min in an Eppendorf 5415 (MilliporeSigma) centrifuge, and the supernatant was removed to a fresh tube.

Measurement of ChAT activity

ChAT activity was determined with the single-vial liquid extraction method of Rand and Johnson (22), which measures the conversion of [³H]choline to acetyl-[³H]choline. Assays were carried out by multiple additions to 1-dram glass vials (03-339-26B; Thermo Fisher Scientific, Waltham, MA, USA). First, 100 μl of substrate mix, comprised of 50 mM tricine (pH 8.1), 0.4 mM neostigmine bromide (N2001; MilliporeSigma), and 4 mM acetyl CoA (Ac-CoA) (A2181; MilliporeSigma) and containing 50,000–100,000 counts per minute (cpm) of [³H]choline (NET109; PerkinElmer, Waltham, MA, USA) was carefully placed in the bottom of the vial. The assay was initiated by addition of 10 μl of tissue supernatant. After incubation at room temperature for 1 or 2 h, the assay was terminated by addition of 100 μl of 50 mM Tris (pH 8.1), 10 mM MgCl₂, and 5 mM ATP (A2383; MilliporeSigma) containing an amount of partially purified yeast choline kinase determined to convert ∼90% of unreacted [³H]choline to phospho-[³H]choline within 1 min. The incubation with choline kinase proceeded for 10–15 min to produce a nearly complete conversion of unreacted [³H]choline to phospho-[³H]choline. Then, 200 μl of H₂O and 3.5 ml of scintillation fluid comprised of 90% toluene and 10% isoamyl alcohol with 0.5% polyphenol oxidase, 0.03% POPOP (Research Products International, Mt. Prospect, IL, USA) and 2 mg/ml sodium tetraphenylboron (T4125; MilliporeSigma) were added. The vial was capped and shaken, and the contents were allowed to settle for 30 min and then counted in a Beckman LS6510 Scintillation Counter (Beckman-Coulter, Brea, CA, USA). In the resulting 2-phase system, the acetyl-[³H]choline forms an ion pair with the organic anion tetraphenylboron and efficiently partitions into the organic phase where it is detected by scintillation counting. The negatively charged phospho-[³H]choline remains in the aqueous phase and is not counted.

Quantitation of ChAT activity

To maximize production of acetyl-[³H]choline, the assay of ChAT activity uses [³H]choline undiluted with unlabeled choline. Given the specific activity of the [³H]choline (78.3 Ci/mmol) and 65% counting efficiency, 50,000–100,000 cpm in 100 μl would be 5–10 nM choline, a concentration that is well below the K_m of mouse central or peripheral isoforms of ChAT for choline. Under these conditions, the rate of conversion of [³H]choline to acetyl-[³H]choline is proportional to V_max/K_m in units of reciprocal time per unit volume of enzyme or milligram of protein (22).

Assays in which the choline kinase was omitted were used to determine the expected maximal conversion to ACh (total). Background counts (blanks) were performed using assays in which the tissue supernatant was substituted with homogenization buffer or in which Ac-CoA was omitted from the substrate mix. ChAT activity is dependent upon the presence of Ac-CoA, the acetylcholinesterase inhibitor neostigmine, and a source of active enzyme (Supplemental Table 1).

Activity is reported as fractional conversion (F_c) of [³H]choline to acetyl-[³H]choline corrected for substrate depletion: F_c = −ln(1 – F), where F = [cpm – cpm (blank)]/[cpm(total) – cpm(blank)]. Assuming that the ChAT activity is stable under assay conditions and that the Ac-CoA concentration remains well above its K_m throughout the assay period, F_c will increase linearly with time. A full description of this calculation is provided in Rand and Johnson (22).

Choline kinase preparation

Choline kinase was prepared from Sacchromyces pastorianus (brewers bottom yeast) obtained from a local brewing supply store. The enzyme was solubilized by extraction with 0.1 M sodium carbonate and then partially purified by ammonium sulfate precipitation and chromatography on diethylaminoethyl-cellulose using minor modifications of the procedure of McCaman et al. (23).

Measurement of food consumption and fecal pellets

Mice were individually housed in cages in which the floors were covered with paper. Mice were provided mouse food pellets. The weight of the food and the number of fecal pellets was determined at t = 0 and again after 19 h.

Immunostaining

Segments of ileum and colon were removed from mice, pinned out in Sylgard dishes, and fixed in 4% paraformaldehyde overnight at 4°C. Tissues were washed in PBS, and the mucosa was stripped off and processed as previously described (2). Tissues were immunostained with antisera specific for either human anti-HuC/D (1:1000, serum obtained from a patient), calbindin (1:500; Immunostar, Hudson, WI, USA), substance P (1:500; Accurate Chemical Co., Westbury, NY, USA), or CGRP (1:500; MilliporeSigma). Secondary antibodies donkey anti-rabbit Dylight 488 (1:500; Jackson ImmunoResearch, West Grove, PA, USA) or donkey anti-rat Cy2 (1:500; Jackson ImmunoResearch) were diluted in PBS containing 0.1% Triton-X 100, 2% donkey serum, 3% bovine serum albumin, and 0.1% sodium azide.

Image analysis

To determine neuronal cell number, ×20 magnification images were collected randomly from three to seven regions of P23 ChAT-Het and ChAT-Null colon (n = 3–4 preparations per genotype). Images were captured on a Nikon A1 Confocal Microscope (Nikon, Melville, NY, USA) and neurons were identified when both the nucleus and cytoplasm were clearly visualized within a single plane from a Z series. For neurotransmitter analysis, immunostained tissues were imaged at ×40 (calbindin) or ×20 (CGRP and substance P).

Gut motility measurement

A 2% carmine red solution (Acros; Thermo Fisher Scientific) suspended in 0.5% methyl cellulose was delivered by oral gavage using a 24-gauge, round-tipped feeding needle to P22–P26 animals. The volume delivered was equal to 10% of the body weight. Mice were euthanized 20 min after carmine red administration, and the entire GI tract was dissected and placed in cold PBS. The distance from the pylorus to the most distal position of the carmine red wavefront and the distance between the pylorus and the ileal-cecal junction were determined (24).

Gut microbiota analysis

Mouse cecal and colonic samples were collected and homogenized in 1 ml extraction buffer [50 mM Tris (pH 7.4), 100 mM EDTA (pH 8.0), 400 mM NaCl, 0.5% SDS) and 20 μl proteinase K (20 mg/ml) containing 500 μl of 0.1‐mm‐diameter zirconia/silica beads (Biospec Products, Bartlesville, OK, USA). Samples were placed in a Mini‐Beadbeater‐8 cell disrupter (Biospec Products) for 5 min to lyse bacterial cells and incubated overnight at 55°C. Bacterial DNA extraction was performed with phenol/chloroform/isoamyl alcohol and precipitation with ethanol. Isolated DNA was dissolved in nuclease‐free water.

To assess bacterial community structure, primers specific for 16S rRNA V3-V4 region (forward: 338F, 5f‐GTGCCAGCMGCCGCGGTAA‐3G and reverse: 806R, 5a‐ GGACTACHVGGGTWTCTAAT‐3) that contained Illumina 3 adapter sequences as well as a 12‐bp barcode were used. Sequences were generated by an Illumina MiSeq DNA platform at Argonne National Laboratory and analyzed with QIIME software (v.1.9.1) (25). Operational taxonomic units (OTUs) were picked against the Greengenes database at 97% sequence identity. OTUs were quality filtered on the basis of default parameters set in the open-reference OTU command in QIIME, and sequences were rarified to an equal sampling depth of 9000 reads per sample. Representative sequences were aligned using PyNAST, and taxonomy was assigned in the RDP Classifier. Alpha diversity was calculated as Shannon Diversity. Beta diversity is represented with UniFrac distances calculated using weighted algorithms and visualized with PCoA plots generated in Emperor. Significant differences in β diversity were identified using ADONIS in Qiime. OTUs generated in QIIME were then analyzed using linear discriminant analysis (LDA) effect size (LEfSe) (26), where nonparametric factorial Kruskal-Wallis sum-rank testing (P < 0.05) identified significantly abundant taxa followed by unpaired Wilcoxon rank-sum test to determine LDA scores >2 (27). Differentiated taxa determined by LDA are displayed in plots as well as cladograms, where taxonomic trees display differential taxa in hierarchical layers.

RESULTS

Mice are viable in utero and postpartum despite absence of ChAT in enteric neurons

Many enteric neurons appear to use ACh based on their expression of ChAT. We previously reported that ChAT expression appears in the ENS as early as E10.5 and reaches adult proportions by E13.5 in the small intestine and by birth in the colon. These cholinergic neuronal numbers are maintained in the adult animal (2). To investigate the role of ChAT in ENS function, we conditionally deleted ChAT in neural crest cells by mating Wnt-1 Cre–expressing mice with animals homozygous for floxed ChAT. Despite the elimination of ChAT in neural crest–derived cells, including enteric neurons, ChAT-Null and ChAT-Het animals were born at the expected frequencies.

Measurement of ChAT activity

To confirm that ChAT was successfully deleted within the bowel of ChAT-Null mice, we measured ChAT activity in laminar preparations of myenteric ganglia and accompanying smooth muscle from WT, ChAT-Het, and ChAT-Null mice. We carried out experiments to show that the detection of activity required tissue extract, Ac-CoA, and the acetylcholinesterase inhibitor neostigmine (Supplemental Table 1). ChAT activity measured in supernatants of tissue homogenates was shown to quickly plateau when using a concentration of 0.5 mM Ac-CoA in the substrate mix (data not shown). When the Ac-CoA concentration was increased to 2 mM, activity was substantially enhanced. It was increased by only a small additional amount at 6 mM Ac-CoA. Based on this, we used 4 mM Ac-CoA in the substrate mix and found that activity measured in supernatants from WT cortex and colon homogenates was close to linear with both time (<2 h) and volume (<10 μl) assayed. We also found that ChAT activity was considerably lower in Tris buffer compared with tricine and was inhibited by 100 mM sodium chloride (data not shown), confirming previous observations (22).

ChAT activity was robustly detected in tissue from the colon of WT and ChAT-Het animals, whereas it was essentially absent in colon of ChAT-Null animals. Ten microliters of extract from WT colon converted 0.60 ± 0.08 F_c of [³H]choline to acetyl-[³H]choline in 2 h, whereas the activity level of a comparable volume of Chat-Null colonic extract was 0.003 ± 0.002 F_c in 2 h (Fig. 1). Similarly, in the small intestine we found that the ChAT activity was extensively reduced in the ChAT-Null compared with WT tissue (0.46 ± 0.08 vs. 0.005 ± 0.002; P < 0.004). In contrast, ChAT activity was not significantly different in the spinal cord of ChAT-Null (0.24 ± 0.005 F_c) compared with WT (0.40 ± 0.09 Fc) or ChAT-Het animals (0.30 ± 0.03 F_c) (Fig. 1). These data confirm that we have achieved conditional deletion of ChAT within intestinal tissue from ChAT-Null mice. The absence of a reduction of ChAT activity in the spinal cord supports the restriction of the Wnt1-Cre promotor to peripheral neurons.

Figure 1.

ChAT activity is absent in the GI tract of ChAT-Null mice. ChAT activity in extracts of P22–P26 WT, ChAT-Het, and ChAT-Null gut. Assays were performed in triplicate. Blank controls (absence of sample), absence of Ac-CoA (No Ac-CoA), and total counts made in the absence of choline kinase were performed in duplicate. Incubation times were 1 h for spinal cord samples and 2 h for intestinal samples. F_c values of [³H]choline to acetyl-[³H]choline corrected for substrate depletion were calculated and recorded (n = 2–5 preparations/genotype). Error bars represent sem. *P < 0.05.

Loss of enteric ChAT results in no reduction in birth weight but results in progressive weight reduction in the third week of life and eventual death by P30

To document the effects of loss of ChAT, we recorded the weight of neonatal animals during the first 3 wk of life. We found that the weights of WT, ChAT-Het, and ChAT-Null mice were similar in wk 1 and 2 when their diets consisted of mother’s milk (Fig. 2). However, during wk 3, the amount of available mother’s milk is normally reduced, weaning occurs, and the neonates begin to consume solid food pellets. Approximately 50% (9/17) of the ChAT-Null animals failed to thrive during this week and subsequently died. The remaining ChAT-Null mice did not significantly gain weight during this week and were, at the end of wk 3, about half the weight of the WT (4.85 ± 0.13 vs. 9.43 ± 0.15; P < 0.001) or ChAT-Het neonates (4.85 ± 0.13 vs. 9.06 ± 0.59; P < 0.001) (Fig. 2).

Figure 2.

Loss of ChAT results in reduced weight over the first 3 wk of life. Weights of P21 WT, ChAT-Het, and ChAT-Null mice. Means ± sem of WT (n = 9), ChAT-Het (n = 9), and ChAT-Null (n = 7) animals measured during the first 3 wk of life. Error bars represent sem. *P < 0.001.

The ability of the ChAT-Null animals that survived beyond wk 3 to consume solid food was examined by placing the animals into individual cages, measuring the food weight at the beginning (t = 0) and end of the experiment (t = 19 h), and counting the number of fecal pellets produced during this 19-h period. The body weight of P21–-28 ChAT-Null mice was about 50% that of WT or ChAT-Het animals. Over this period of observation, WT and ChAT-Het mice consumed almost 3 g of food, whereas the ChAT-Null mice ingested almost no food (0.1 g) (Table 1). Consistent with reduced intake, after 19 h only a small number of fecal pellets were present in the cages of the ChAT-Null animals (3 ± 2 pellets), compared with that found in WT (171 ± 20 pellets) or ChAT-Het cages (175 ± 26 pellets) (Table 1).

TABLE 1.

Weight, food consumption, and fecal output

Genotype	Average weight (g)	Food consumption (g)	Fecal pellets (n)
WT	12.6 ± 0.5	3.07 ± 0.10	171 ± 20
ChAT-Het	12.3 ± 0.3	2.91 ± 0.15	175 ± 26
ChAT-Null	6.7 ± 0.5*	0.09 ± 0.02*	3 ± 2*

For mice surviving beyond 3 wk of age, weight, food consumption, and number of fecal pellets produced within 19 h by P25 WT, ChAT-Het, and ChAT-Null mice (n = 3/genotype) are shown. Data are means ± sem. *P < 0.05.

We dissected ChAT-Het and ChAT-Null animals during wk 4 to get a clearer picture of why ChAT-Null animals were failing to thrive. In comparison to WT or ChAT-Het mice, which show fecal material distributed evenly along the small intestine and pelleted stool within the colon, ChAT-Null mice display compacted fecal material within the cecum and distal colon (Fig. 3). This resulted in obstruction of the colon and subsequent death usually by P30.

Figure 3.

ChAT-Null mice have colonic fecal impaction. Bright field image of representative ChAT-Het and ChAT-Null GI tracts dissected from P4 animals. Note the reduced length of the intestine in the ChAT-Null compared with the ChAT-Het. Fecal material is compacted in the cecum and colon of ChAT-Null mice. Scale bar, 1 cm.

A separate subset of ChAT-Null animals was fed a diet of moistened, powdered chow. These animals survived for an extended period (range, 139–353 d; n = 4), indicating that the small intestine continues to function in nutrient digestion and absorption in the absence of ACh in neurons.

Small intestinal propulsive gut motility is reduced in ChAT-Null mice

We measured a combination of gastric emptying and small intestinal transit by delivering carmine red dye into the stomach of P22–26 mice. After 20 min, we determined the distance traversed by the dye in the small intestine. In WT animals, carmine red moved an average distance of 10.2 cm, which represented 38.1 ± 7.3% of the small intestinal length (Table 2). The carmine red traveled a similar distance in ChAT-Het mice (10.9 cm; 42.6 ± 6.1%) (Table 2). In ChAT-Null animals there was a reduction in total intestinal length compared with WT and ChAT-Het animals. Although carmine red moved a significantly shorter distance in ChAT-Null animals (4.4 cm; 22.4 ± 2.9%), the dye was not stationary, suggesting that propulsive activity was retained in the small intestine of these animals (Table 2).

TABLE 2.

Small intestinal gut motility

Genotype	Average weight (g)	SI length (cm)	Carmine transit (% total SI length)
WT	10.75 ± 0.5	26.85 ± 1.5	38.1 ± 7.3
ChAT-Het	9.82 ± 1.1	26.32 ± 1.2	42.6 ± 6.1
ChAT-Null	5.5 ± 0.3*	19.83 ± 1.0*	22.4 ± 2.9*

Average weight, length of small intestine (SI), and percentage of small intestinal length occupied by carmine red 20 min after its administration in P22 WT (n = 4), ChAT-Het (n = 5), and ChAT-Null (n = 4) mice. Data shown are means ± sem. *P < 0.05.

Calbindin, CGRP, and Substance P neuronal phenotypes are maintained in the ENS in the absence of ChAT

The absence of ChAT activity in enteric neurons correlated with slowed intestinal transit and the presence of impacted feces in the colon. To investigate whether conditional deletion of ChAT altered the enteric neuronal number, we immunostained the colon of ChAT-Het and ChAT-Null animals with the pan-neuronal marker Hu and visualized the labeled cells using confocal microscopy. We counted the number of Hu+ neurons within three to seven randomly chosen fields along the colon. We found that the numbers of Hu+ neurons were similar between genotypes (562 ± 66 vs. 560 ± 175/region; P = not significant).

To determine whether the dysmotility we detected resulted from the loss of other neurotransmitters normally coexpressed within cholinergic neurons, we performed additional immunostaining (Fig. 4). Enteric sensory neurons express ChAT, CGRP, and the calcium-binding protein calbindin, whereas excitatory motor neurons produce both ChAT and substance P (1, 3). Abundant calbindin-immunoreactivity (IR) and CGRP-IR was identified by colocalization with endogenously labeled tdTomato+ neurons in the myenteric ganglia (Fig. 4). In the gut, CGRP is found in both extrinsic spinal afferent fibers and intrinsic sensory neurons. Although we did not obtain clear localization of CGRP within neuronal cell bodies, some of the CGRP-IR likely arises from intrinsic sensory neurons (28). Numerous substance P–IR nerve fibers were found in the smooth muscle of the colon (Fig. 4). Such fibers only arise from intrinsic substance P expressing neurons in which cholinergic markers are normally located (28). These data confirm that, in the absence of ChAT, calbindin, CGRP, and substance P neuronal phenotypes are present in the myenteric ganglia of ChAT-Null mice.

Figure 4.

Expression of calbindin, CGRP, and substance P is preserved in ChAT-Null mice. Upper row: calbindin-IR neurons and neuronal processes (green) are present within the myenteric ganglia, which are endogenously labeled with tdTomato (red, center panel) of a P25 ChAT-Null cecum. Right panel shows a merged image with the neuronal marker, Hu-IR, (blue). Scale bar, 50 µm. Middle row: CGRP-IR processes (green) are abundant within ChAT-Null small intestine. Center panel shows tdTomato+ neurons; a merged image can be seen in the right panel. CGRP-IR processes project between myenteric ganglia. Scale bar, 100 µm. Bottom row, left panel: substance P–IR processes (green) appear throughout the smooth muscle of a P25 ChAT-Null colon. Most substance P–IR processes also express tdTomato, as shown in the merged right panel. Scale bar, 100 µm.

Microbial community composition is altered in ChAT-Null mice

Cecal and colonic samples from WT, ChAT-Het, and ChAT-Null mice were analyzed by 16S rRNA sequencing to determine whether the dysmotility resulting from ChAT deletion affected the composition of microbial communities (Fig. 5). Alterations in the relative abundance of multiple bacterial phyla, specifically a decrease in Firmicutes and increases in Bacteroidetes and Proteobacteria, were noted in the ChAT-Null cecal microbial communities (Fig. 5A). ChAT-Null cecal and colonic microbiota displayed lower α diversity, which is a measure of diversity in the sample, by the Shannon Diversity Index (Fig. 5B). Beta diversity, which compares diversity between samples, was determined with principal coordinate analysis weighted UniFrac distances. Here, ChAT-Null animals segregated from WT and ChAT-Het animals (Fig. 5C), a finding indicating that the loss of ChAT expression profoundly alters community composition. Significant differences in β diversity were detected across genotypes in cecal (R² = 0.224; P > 0.01) and colon (R² = 0.174; P < 0.01) communities using ADONIS on unweighted UniFrac distance matrices. Differentially represented taxa within each group were determined by LEfSe for both the cecal and colonic microbial communities (Fig. 5D, E). Finally, differentially represented taxa identified in LEfSe were plotted by cladograms (Fig. 5F, G) to visualize the differences in phylogenetic structure across taxonomic levels. These analyses revealed that Enterobacteriaceae, found only in ChAT-Null mice, was the major taxa responsible for differences found between the ChAT-Null and the other genotypes.

Figure 5.

Dysbiosis in the gut of ChAT-Null mice. 16S rRNA sequencing of bacterial communities in WT, ChAT-Het, and ChAT-Null cecum and colon samples. A) Relative abundance of cecal and colonic bacterial phyla is shown for each genotype. B) Alpha diversity was measured by the Shannon Diversity Index, where ChAT-Null displayed lower diversity in both microbiota communities. C) Beta diversity dissimilarity was calculated with weighted UniFrac distances, where PC1 explained 33% of variability, and PC2 explained 20% of variability observed in cecal and colonic communities. D, E) To determine differentially represented taxa within each group, LDA LEfSe was calculated. Significantly differential taxa reaching logarithmic LDA scores of >2 are displayed for each group. Tenericutes were consistently represented by WT animals in both microbial communities, whereas Enterobacteriaceae were enriched in ChAT-Null samples. F, G) The differentially represented taxa determined by LEfSe were then graphed as cladograms to display phylogenic structure across taxonomic levels. Cladogram taxa abbreviated by letters are shown in the key in (F, G). These results demonstrate the shifted microbiota taxa were taxonomically related between genotypes, suggesting the altered luminal environment with deficient ChAT-directed specific shifts in the bacterial kingdom. Error bars represent sem.

DISCUSSION

We investigated in vivo the loss of synthesis of ACh in neural crest–derived cells, including enteric neurons, by deleting enteric ChAT using the Wnt1-Cre driver. We observed a number of alterations in ChAT-Null mice. ChAT-Null animals are born in the expected numbers and appear similar to WT or ChAT-Het mice during the first 2 wk of life. Therefore, although the cholinergic phenotype appears in enteric neurons halfway through gestation, it is not necessary for mouse survival or nutrition in the first 2 wk. However, ChAT-Null mice failed to gain weight after weaning in postnatal wk 3. During postnatal wk 4, they show reduced intestinal transit along the GI tract, and many had an enlarged colon filled with fecal material, suggesting dysmotility and functional obstruction. This intestinal obstruction may contribute to the failure of the ChAT-Null mice to gain weight and their death around P30. Despite these findings, some observations, such as carmine dye movement in the small intestine and the prolonged survival of a few ChAT-Null animals fed a moistened chow diet, indicate that some motor activity in the gut persists in the absence of ACh. The observation of fecal impaction in the colons of animals that died, along with evidence of slowed, but maintained, small intestinal transit, suggests that enteric ACh is essential for colonic, but not small intestinal, esophageal or gastric motility. Carmine red dye advanced in the ChAT-Null small intestine even though gut motility was reduced by about 50% of that observed in WT and ChAT-Het animals. Because we observed the presence of fecal material obstructing the colon, we were unable to perform measurements of colonic motility using the expulsion of a glass bead method (29). It is possible that tachykinins support contractile activity in the ChAT-Null intestines because excitatory enteric motor neurons express both acetylcholine and a tachykinin (substance P) (1). Tachykinins contribute to excitatory transmission to the muscle (19), and we have demonstrated that expression of substance P is maintained in the absence of ChAT. However, the levels of specific neurotransmitters and their contribution to overall ENS function in the presence or absence of ChAT remain to be defined.

We were surprised that the absence of ChAT within neural crest–derived cells did not produce a more extreme phenotype during embryonic development or earlier in postnatal life. To confirm that the Wnt1-Cre driver sufficiently excised the regions flanked by LoxP sites, thereby resulting in the absence of ChAT in our ChAT-Null mice, we used a method with high sensitivity for detection of ChAT activity (22). Direct measurement of ChAT activity in homogenates of myenteric ganglia indicates that we achieved a reduction of >99% in ChAT-Null animals compared with WT littermates. The absence of a reduction of ChAT activity in the spinal cord supports the restriction of the Wnt1 Cre promotor to peripheral neurons. In contrast, Misgeld et al. (20) used the β-actin Cre promotor to ubiquitously delete ChAT. In that study, ChAT-knockout animals were unable to breathe and died at birth, presumably due to lack of ChAT in respiratory motorneurons.

The interrelationship between intestinal motility and microbiome composition is an area of recent focus. Wiles et al. (30) used live intestinal imaging in a gnotobiotic zebrafish model with and without intestinal aganglionosis to elucidate the contribution of motility to the intestinal microbiota. They found that different bacterial strains are differentially affected by changes in intestinal motility. In further work from the same group, Rolig et al. (31) demonstrated that the presence or absence of the ENS was a primary determinant of microbiota composition and intestinal inflammation. Additionally, using fecal microbiota transplant experiments and a model of loperamide-induced constipation, Touw et al. (32) found that reduced motility altered the microbial community of the gut and that this resulted in a feed-forward loop whereby the slowed transit was exacerbated or maintained by microbial metabolites. In our study, WT and ChAT-Het animals had similar microbiota α and β diversity in both the colon and cecum, whereas only ChAT-Null animals displayed dysbiosis at both these sites. Specifically, ChAT-Null animals display significantly more Enterobacteriaceae, Blautia, and Clostridiaceae in both cecal and colonic samples, whereas WT animals were characterized by significantly elevated Tenericutes. The bacterial family Enterobacteriaceae includes many well-defined gram-negative pathogens, including Salmonella, Escherichia coli, and Shingella, that are associated with nosocomial and intra-abdominal infection. Our microbiota results also resemble dysbiotic patterns observed in murine models of Hirschsprung disease, in which the composition of the microbiota clustered based on presence or absence of intestinal motility (33, 34). In animal models of Hirschsprung disease and in human patients, an elevated risk of mucosal enterocolitis is observed in the absence of treatment. Furthermore, ChAT-Het animals, which have reduced ChAT activity but do not have altered transit, display differences from WT in both cecal and colonic taxa (by LEfSe), suggesting that ACh could contribute to the maintenance of microbial composition. Together, these studies contribute to the growing understanding of a bidirectional relationship between gut microbiota, intestinal motility, and the ENS.

In summary, we find that developmental deletion of ChAT activity in enteric neurons results in proximal gastrointestinal tract dysmotility, critically diminished colonic transit, failure to thrive, intestinal dysbiosis, and death. Although other enteric neurotransmitter phenotypes are preserved, we conclude that ACh is necessary for sustained gut motility and survival of mice on a normal rodent diet after weaning.

Supplementary Material

This article includes supplemental data. Please visit http://www.fasebj.org to obtain this information.

Glossary

Ac-CoA

acetyl co-A

ACh

acetylcholine

CGRP

calcitonin gene–related peptide

ChAT

choline acetyltransferase

cpm

counts per minute

ENS

enteric nervous system

Fc

fractional conversion

GI

gastrointestinal

Het

heterozygote

IR

immunoreactivity

LDA

linear discriminant analysis

LEfSe

linear discriminant analysis effect size

OTU

operational taxonomic unit

AUTHOR CONTRIBUTIONS

M. L. Epstein and A. Gosain developed the study concept and design; C. D. Johnson, A. J. Barlow-Anacker, C. S. Erickson, M. L. Epstein, and A. Gosain acquired the data; C. D. Johnson, A. J. Barlow-Anacker, J. F. Pierre, K. Touw, C. S. Erickson, J. B. Furness, M. L. Epstein, and A. Gosain analyzed and interpreted the data; C. D. Johnson, A. J. Barlow-Anacker, J. F. Pierre, J. B. Furness, M. L. Epstein, and A. Gosain drafted and revised the manuscript; and all authors approved the final manuscript.