The microbiome is essential for normal gut intrinsic primary afferent neuron excitability in the mouse


  • K. A. McVey Neufeld,

    1. McMaster Brain-Body Institute at St Joseph’s Healthcare, Hamilton ON, Canada
    2. Department of Psychiatry and Behavioural Neurosciences, McMaster University, Hamilton ON, Canada
    Search for more papers by this author
  • Y. K. Mao,

    1. McMaster Brain-Body Institute at St Joseph’s Healthcare, Hamilton ON, Canada
    Search for more papers by this author
  • J. Bienenstock,

    1. McMaster Brain-Body Institute at St Joseph’s Healthcare, Hamilton ON, Canada
    2. Department of Pathology and Molecular Medicine, McMaster University, Hamilton ON, Canada
    Search for more papers by this author
  • J. A. Foster,

    1. McMaster Brain-Body Institute at St Joseph’s Healthcare, Hamilton ON, Canada
    2. Department of Psychiatry and Behavioural Neurosciences, McMaster University, Hamilton ON, Canada
    Search for more papers by this author
  • W. A. Kunze

    1. McMaster Brain-Body Institute at St Joseph’s Healthcare, Hamilton ON, Canada
    2. Department of Psychiatry and Behavioural Neurosciences, McMaster University, Hamilton ON, Canada
    Search for more papers by this author

Address for Correspondence
Karen-Anne McVey Neufeld, McMaster Brain-Body Institute at St Joseph’s Healthcare, T3338, 50 Charlton Ave E., Hamilton, ON, Canada L8N 4A6.
Tel: 905-522-1155 x34329; fax: 905-540-6593;


Background  The role of intestinal microbiota in the development and function of host physiology is of high interest, especially with respect to the nervous system. While strong evidence has accrued that intestinal bacteria alter host nervous system function, mechanisms by which this occurs have remained elusive. For this reason, we have carried out experiments examining the electrophysiological properties of neurons in the myenteric plexus of the enteric nervous system (ENS) in germ-free (GF) mice compared with specific pathogen-free (SPF) control mice and adult germ-free mice that have been conventionalized (CONV-GF) with intestinal bacteria.

Methods  Segments of jejunum from 8 to 12 week old GF, SPF, and CONV-GF mice were dissected to expose the myenteric plexus. Intracellular recordings in current-clamp mode were made by impaling cells with sharp microelectrodes. Action potential (AP) shapes, firing thresholds, the number of APs fired at 2× threshold, and passive membrane characteristics were measured.

Key Results  In GF mice, excitability was decreased in myenteric afterhyperpolarization (AH) neurons as measured by a lower resting membrane potential and by the number of APs generated at 2× threshold. The post AP slow afterhyperpolarization (sAHP) was prolonged for GF compared with SPF and CONV-GF animals. Passive membrane characteristics were also altered in GF mice by a decrease in input resistance.

Conclusions & Inferences  Here, we report the novel finding that commensal intestinal microbiota are necessary for normal excitability of gut sensory neurons and thus provide a potential mechanism for the transfer of information between the microbiota and nervous system.


There has been a significant recent increase in our understanding of the complexity and diversity of the intestinal microbiota.1 There is evidence that microorganisms are directly involved in the development and function of the intestinal, respiratory, and nervous systems, as well as in modulating the immune, metabolic, and endocrinologic functions they subserve.2–6 Of particular current interest is the impact of intestinal microbiota on both the enteric (ENS) and central nervous systems (CNS) as reflected by a number of recent reviews focusing on the role of intestinal bacteria in the communication known to exist between the gut and nervous system, communication now commonly referred to as the microbiome-gut-brain axis.7–11

The bacterial contents of the gut have become important for research into the etiology of functional bowel disorders, with a number of studies pointing to variations in the composition of gut microbiota in patients suffering from irritable bowel syndrome (IBS) compared with controls.12,13 Clinical work with probiotics has demonstrated that ingestion of certain strains may improve gastrointestinal symptoms in IBS14,15 with the mechanisms of action yet to be determined. Abdominal pain is a hallmark symptom of IBS, and feeding rodents probiotics or altering the gut microbiome with antibiotics16–20 changes pseudoaffective responses evoked by colorectal distension (CRD). Active smooth muscle contraction is necessary for nociceptive afferent signaling in response to CRD21 and we have previously demonstrated that probiotics can affect the excitability of myenteric neurons,22 thereby altering the intensity of propagated peristaltic contractions. Taken together, the data thus suggest that the change in excitability of myenteric neurons after exposure to certain bacterial species could be modulating the nociceptive response.

The ENS can control the functioning of the intestine independently from the CNS 23 while still allowing for the modulation of interchange with the central and autonomic nervous systems. Neurons in the ENS can be divided into two sub-types, based on both morphologic and functional categories24 and can also be identified by their electrophysiological properties that correlate with the morphotype.25,26 Afterhyperpolarization (AH) cells have multiple long processes and a large oval soma (Dogiel Type II neurons), are chemo- and mechanosensory intrinsic primary afferent neurons (IPANs)27,28 and their activity is necessary for normal peristalsis and viability.29 Intrinsic primary afferent neurons project to other IPANs, inter- and motor neurons, and send long processes into the gut mucosa, terminating near the epithelial layer.27,30,31 Myenteric AH cells are thus in a favorable position to transmit information from the intestinal lumen to the nervous system as a whole. Previous work in our laboratory has demonstrated that rats fed Lactobacillus rhamnosus (JB-1) for 9 days prior to electrophysiological recording show an increase in myenteric AH cell intrinsic excitability and a related decrease in duration of the inhibitory slow afterhyperpolarization (sAHP).32 This finding suggests a cellular neural mechanism by which bacteria influence motility and possibly pain perception.

We hypothesized that the total absence of bacteria might result in a corresponding decrease in excitability of AH cells possibly associated with an enhanced sAHP, and we have thus designed experiments to examine AH cell electrophysiology in germ-free (GF) mice for comparison with specific pathogen-free (SPF) mice of identical strain. We also hypothesized that conventionalizing adult GF (CONV-GF) mice with normal commensal intestinal microbiota would normalize any observed differences in IPAN physiology between GF and SPF groups. Accordingly, we have carried out intracellular recordings on IPANs within the myenteric plexus in GF and SPF male and female, adult Swiss Webster mice and sought to determine if there were differences in excitability and sAHP parameters, and if so, could these be normalized after conventionalization.



Male and female, Swiss Webster, 8-week-old, SPF or GF mice were obtained from the Animal Research Facility at McMaster University. Germ-free mice were bred and raised in the gnotobiotic facility, while SPF mice were conventionally housed. CONV-GF mice were obtained as 8-week-old GF adults from the gnotobiotic facility and then housed for 4 weeks in conventional housing. Conventionalization was accomplished by transferring dirty bedding from SPF cages into GF cages on alternate days as described by Harris & Macpherson.33 All mice were maintained in cages with free access to food and water until the beginning of testing. All animals were housed under a 12 h light–12 h dark schedule, with lights on at 7 AM. Housing room temperature was maintained at 20 °C and humidity at 60–70%.


All experimental procedures followed the guidelines of the Canadian Council on Animal Care and were approved by the Animal Research Ethics Board, McMaster University, Hamilton, Ontario, Canada.


Mice were received from the Animal Facility at McMaster University daily at 9 AM. Animals were transported to the laboratory and immediately killed by cervical decapitation. A 2-cm segment of jejunum was removed within 30 min of receipt from the gnotobiotic facility, and tissue placed in a 2-mL recording dish lined with silastic and filled with carbogenated Krebs buffer of the following composition (in mM): NaCl 118.1, KCl 4.8, NaHCO3 25, NaH2PO4 1.0, MgSO4 1.2, glucose 11.1, and CaCl2 2.5, gassed with carbogen (O2 95%, CO2 5%). Nicardipine (2–3 μM) and scopolamine (1 μM) were routinely added to the saline to minimize spontaneous muscle contraction. The segment was opened along a line parallel to the mesenteric attachment and pinned flat, under moderate tension, mucosa uppermost. The myenteric plexus was exposed by dissecting away the mucosa, submucosa, and circular muscle. The recording dish was then mounted on an inverted Nikon T 2000 microscope (Tokyo, Japan) and the tissue continuously superfused (4 mL min−1) with carbogenated Krebs that had been warmed to 35–37 °C.


Intracellular voltage signals were measured in current-clamp mode using an Axon Instruments Multiclamp 700A computer-controlled amplifier (Molecular Devices, Sunnyvale, CA, USA) and a Digidata 1322A (Molecular Devices) A/D digitizer. Current clamp commands were delivered to the amplifier using Clampex 8 Molecular Devices software. Thin-walled borosilicate glass tubes were pulled on a Flaming-Brown P97 (Sutter Instruments, electrode puller to produce sharp pipettes. The pipettes were filled with 1 M KCl and 0.5% neurobiotin to produce sharp electrodes with resistances of 100–120 MΩ.

After impalement, AH neurons (IPANs) were identified electrophysiologically by the presence of a hump on the repolarization phase of the action potential (AP) and by the presence of a sAHP of ≥2 s duration following the AP (Fig. 1). The hump could also be detected as an inflection in the first-order time derivative of the AP (Fig. 1C and D). A single AP for measuring the hump was evoked by injecting a short (<1 ms duration) depolarizing current pulse which just evoked the spike on the falling phase of the electrotonic potential.34 These data were Bessel prefiltered at 5 and acquired at 20 kHz. For other experiments, data were prefiltered at 2 and acquired at 10 kHz. After neurons were identified as being AH cells, 500 ms duration depolarizing current pulses of increasing intensity were injected until threshold for AP firing was reached (Fig. 1A). Having determined the threshold current intensity, a 500 ms depolarizing current pulse of 2× threshold was injected (Fig. 1B). Finally, the sAHP was evoked by delivering three suprathreshold 50 ms duration depolarizing current pulses separated by 100 ms intervals. Following recording cells filled with neurobiotin were processed to reveal their shape34; cells were classified as having multiple long processes (Dogiel Type II cells) or as having one long uniaxonal process.

Figure 1.

 Traces of intracellular recordings were made from a representative AH cell. (A) Electrotonic voltage deflection and suprathreshold responds with action potential to subthreshold and suprathreshold 500 ms square wave positive current pulses injected into the cell via the recording electrode. (B) Two APs were discharged when depolarizing current at 2× threshold intensity was injected. (C) Single AP showing hump on repolarizing phase (downstroke), (D) first-order time derivative of AP reveals the hump as an inflection. (E) Fast afterhyperpolarization shown as continuation of AP downstroke. AP truncated at ★. (F) Slow afterhyperpolarization (sAHP) lasts for seconds and peaks with a delay after AP and fAHP.

Data were stored on a computer and analyzed off-line; for each neuron, analyses included measurements of excitability, that is AP threshold, and the number of APs fired at 2× threshold (Fig. 1A and B). The resting membrane potential (RMP) and input resistance (Rin) were measured using Clampfit software (Molecular Devices). The following AP properties were measured (Fig. 1C–F): spike amplitude (AP amp.), AP width at half amplitude (1/2 width), maximum rate of depolarization (max dV/dt), the amplitude of the fast afterhyperpolarization (fAHP) that continues the AP downstroke, and sAHP area under the curve (AUC), duration and amplitude (Fig. 1E and F). It was not always possible to measure the fAHP duration as the fAHP frequently merged into the much longer sAHP. Descriptive statistics are given as mean ± SD and number of neurons (n) in the sample. Errors on bar graphs are SEM where *, **, and *** and ns denotes significance at the = 0.05, 0.01, 0.001 or >0.05 (not significant) level. One-way anova was used to compare the measured parameters between the different treatments described above. When a statistically significant (P ≤ 0.05) treatment was identified, post hoc tests were used to test for hypothesized differences. GraphPad Prism version 5 (GraphPad Software, San Diego, CA, USA) was used for all descriptive statistics, including one-way anova, and Bonferroni or Dunn’s multiple comparison tests.



No differences were observed between male and female mice in any of the electrophysiological parameters measured below; results were therefore pooled across sexes.

Cell Classification

Cells were differentiated based on both morphotype and electrophysiological properties and then categorized as either AH cells or S cells. Afterhyperpolarization cells were unequivocally identified by the presence of a hump on the AP relaxation phase and a post AP sAHP lasting longer than 2 s. S cells were identified by the clear absence of these properties. In agreement with previous work, S cell APs were never followed by a sAHP (0/25 S cells) and all 25 were monoaxonal and lacked Dogiel Type II morphology. S cell membrane potential was not altered by the absence of intestinal microbiota −57 ± 7 (12) SPF vs−53 ± 8 mV (13) GF (= 0.3). Similarly, input resistance for S cells was unaffected with Rin 225 ± 49 (12) SPF vs 136 ± 21 MΩ (13) GF (= 0.08). Threshold current required to evoke an AP was 204 ± 125 (12) SPF vs 235 ± 208 pA (13) GF (= 0.7) and number of APs evoked at twice threshold current intensity 1.7 ± 1 (12) SPF vs 1.3 ± 0.6 (13) GF (= 0.3) were unaffected by bacterial status (data for S cells not shown). As no differences were observed in S cells across any of the electrophysiological parameters measured, the following results refer to AH cells only. We carried out several experiments without neurobiotin in the electrodes and observed no differences in any of the parameters recorded so that all of the electrophysiological data presented were obtained from neurons in which morphotype was ascertained.


The threshold current required to evoke an AP using control (SPF) animals was 175 ± 82 (16), for GF animals 289 ± 271 (9), and for CONV-GF animals 188 ± 135 pA (19) (Fig. 2A); there was no statistical difference in population means (anova, = 0.2). The number of APs evoked at twice threshold current intensity differed between treatments (Kruskal Wallis test, = 0.01), for SPF animals this was 3.6 ± 2.2, for GF animals 1.9 ± 0.6, and CONV-GF 3.7 ± 2.8 APs (Fig. 2B). Post hoc Dunn’s multiple comparison test showed that SPF differed from GF (< 0.01, rank sum difference = 15) and GF differed from CONV-GF (< 0.05).

Figure 2.

 Afterhyperpolarization cell excitability: summary statistics. (A) Sample differences for AP thresholds between treatment groups did not reach statistical significance. (B) Number of APs evoked by intracellular stimulus current at 2× threshold intensity differed between treatment groups. Post hoc analysis with Dunnett’s multiple comparison test revealed that number of APs discharged for GF differed from those of SPF mice, and those from CONV-GF differed from GF mice. For this and subsequent figures post hoc multiple comparison tests were always between GF vs SPF and CONV-GF vs GF treatment groups, as these were hypothesized to be different. Numbers inside bars denote n number.

Passive membrane properties

Resting membrane potentials were −56 ± 3 (15), −66 ± 7 (9), and −59 ± 7 (19) mV for SPF, GF, and CONV-GF, respectively (Fig 3A), giving a statistical difference between treatments (anova, = 0.001). A post hoc Bonferroni’s multiple comparison test indicated that SPF differed from GF (< 0.001) and CONV-GF differed from GF (< 0.05). Intrinsic primary afferent neuron Rin for SPF, GF, and CONV-GF mice were 206 ± 62 (16), 146 ± 45 (8), and 212 ± 64 (16) MΩ (Fig. 3B) with a significant difference between treatments (anova, = 0.04). Specific pathogen-free mice differed from CONV-GF mice (< 0.05) and CONV-GF differed from GF mice (< 0.05; Bonferroni’s multiple comparison test).

Figure 3.

 Passive membrane characteristics: summary statistics. (A) Resting membrane potential (RMP) and (B) input resistance differed between treatment groups, with statistically significant differences between GF and SPF and CONV-GF and GF groups. Selected post hoc tests, with Bonferroni’s correction, were between GF vs SPF and CONV-GF vs GF. Numbers inside bars represents n number.

Action potential parameters

Action potential amplitudes were 81 ± 9 (14), 89 ± 22 (8), and 76 ± 12 (18) mV for SPF, GF, and CONV-GF mice, respectively (Fig. 4A; anova, = 0.1). The maximal rates of AP depolarization (dV/dt) were also not affected by treatments (81 ± 10 (15), 89 ± 22 (8), and 76 ± 12 (19) for SPF, GF, and CONV-GF treatment groups (anova, = 0.1; data not shown). Action potential (1/2 widths were 2.2 ± 0.5 (13), 2.6 ± 0.7 (8), and 2.4 ± 0.4 (17) (anova, = 0.2; Fig. 4B), and amplitudes of the fast afterhyperpolarization (fAHP) were −5.5 ± 6.2 (12), −3.6 ± 4.2 (8), and −4.9 ± 7 (18) (Fig. 4C; anova, = 0.8), all for SPF, GF, CONV-GF treatments, respectively.

Figure 4.

 Action potential properties: summary statistics. (A–C) AP amplitude (AP amp), width at 1/2 amplitude (1/2 width) and the amplitude of the fast afterhyperpolarization (fAHP amp) were not significantly altered by the treatment conditions. (D) Treatments affected the area under the curve of the slow afterhyperpolarization (sAHP AUC) with post hoc analysis showing a statistically significant difference between CONV-GF vs GF treatments, but for the sample size used only a non-significant trend was evident when SPF was compared with GF treatment. (E) sAHP duration was significantly affected by the treatments, with SPF vs GF and CONV-GF vs GF treatments producing highly significant differences in the durations. (F) sAHP amplitude (sAHP amp) was not statistically affected by the treatments.

Slow sAHPs areas under the curve (AUC) were altered by the treatments. SPF, GF, and CONV-GF AUCs were 26 874 ± 16 969 (16), −58 637 ± 67 143 (9), and −24 539 ± 17 288 (19) (Fig. 4D; anova, = 0.04). Post hoc comparisons between SPF vs GF and CONV-GF vs GF revealed that the former comparison had > 0.05 while the latter had < 0.05 (Bonferroni’s multiple comparison test). Slow afterhyperpolarization durations were SPF 3.0 ± 1.5 (16), GF 6.2 ± 3.2 (9), and CONV-GF 2.8 ± 1.6 (19) s (Fig. 4F; anova, = 0.0003). The difference between SPF vs GF and CONV-GF vs GF samples were both statistically significant with < 0.001 (Bonferroni’s multiple comparison test). However, the treatments had no effect on sAHP amplitudes (Fig. 4F), sample values were SPF −6.3 ± 2.4 (16), GF −9.7 ± 6.1 (9), and CONV-GF −7.3 ± 2.6 (19) (anova = 0.08).


An altered gut microbiome has been associated with inflammatory bowel diseases such as Crohn’s disease and ulcerative colitis35 and IBS.12,13 Functional changes in motility and pain perception are well described in these conditions. Indeed, in spite of both motility and pain perception being influenced by ENS-controlled contractions,21 very little information exists about the role that the normal microbiota of the gut play in regulating the development and function of the ENS. This is somewhat surprising because it is becoming clear that acquisition of a normal microbiome in infancy has far-reaching implications, for instance playing a major role in the development of the immune and metabolic systems3,4,33 and even influencing the set points of the hypothalamic-pituitary-adrenal axis.6 Recent data have shown that conventionalization of germ-free mice influenced the development of the brain5 as well as its neurochemistry, and we have observed significant differences between the behavior and brain neurochemistry of germ-free and conventionally housed animals.36,37

Given the role of the gut microbiome in establishing normal adult brain structure and function, we have begun to study the function of the ENS in the germ-free state and compared this to that found in normal SPF and conventionalized GF mice. Specifically, in GF mice, we observed a lowered excitability of mouse myenteric IPANs. This effect was associated with both a more negative RMP and decreased Rin; that is, the leak (background) conductance was increased in GF mice. With respect to action potential characteristics, only the duration and AUC of the sAHP were significantly increased. The functional effects of these changes would be to decrease IPAN discharge in response to adequate sensory stimuli. Exposing adult GF mice to intestinal microbiota for 4 weeks (conventionalization) normalized levels of excitability in myenteric AH cells and reversed the effects on passive membrane characteristics and sAHP duration that were associated with the germ-free condition. These findings directly demonstrate that in the absence of intestinal microbiota, ENS neurons have altered electrophysiological properties that can be changed to normal when the animals are no longer microbe-free.

The ENS provides sensory innervation of the gut mucosa, and nerve fiber endings of the myenteric plexus terminate in close proximity to epithelial cells lining the lumen.24,27 Given that these neuronal endings are thus close to gut luminal contents, including trillions of commensal bacteria, it is reasonable to hypothesize that bacteria and their products, including those of fermentation, or factors released by the intervening epithelium, may signal to the central nervous system via these enteric neurons.38 For the current experiment, it was notable that differences were found in the myenteric sensory AH neurons, which is consistent with their rich innervation of the mucosal epithelium. Afterhyperpolarization cell firing critically influences peristaltic reflexes and migrating motor complexes31,39 and indeed previous work in our own laboratory has demonstrated changes in these particular neurons after the feeding of live probiotic L. rhamnosus JB-1 to conventionally housed rats.32 Rats fed the JB-1 as opposed to vehicle control for 9 days demonstrated increased neuronal excitability via a number of altered electrophysiological measures including; lower threshold for eliciting action potentials, increased number of elicited action potentials during a suprathreshold depolarizing pulse, and decreased duration of sAHP in AH neurons. A follow-up study using mice showed that JB-1 likely worked to increase excitability of AH neurons by blocking the calcium-dependent potassium channel IKCa.22 These results are all consistent with the hypothesis that in healthy animals under SPF conditions, Lactobacilli and other commensals influence myenteric neurons to maintain their excitability. This contrasts with the current work demonstrating that AH neurons show less excitability in the absence of colonizing microbiota. Interesting too is our observation that similar to the present findings, the probiotic appeared to have no persisting effect on electrical activity of S cells and is therefore specific for this subpopulation of AH enteric neurons, which unlike S cells, has been shown to possess a long-term cellular memory of excitability.40,41 While patch clamp techniques were not carried out in the current work, and thus direct measures of single ion channel activity were therefore not possible, the fact that GF mice differ in the duration of the sAHP in AH neurons when compared with SPF mice, and that normal duration is apparently achieved after conventionalizing GF mice, indicates that the IKCa channel activity could also be altered in the absence of gut bacteria.

Earlier work in GF rats has demonstrated altered gut motility in the absence of intestinal microbiota. Fewer migrating motor complexes (MMC) reached the midpoint of the small intestine, and the duration between MMCs was increased in GF mice as compared with controls.42 In a follow-up study, this research group demonstrated that colonizing GF rats with a single bacterial species was sufficient to alter the fasting myoelectric activity of the small intestine with the specific changes observed dependent upon the particular bacterial species used for colonization.43 These data suggest that the effects seen in GF rats and our own experiments with mice were unlikely due to LPS. Indeed, we have done similar experiments to those recorded herein with different doses of LPS and not found any effects on electrophysiological recordings from AH cells (data not shown). Taken together, the overall findings are indicative of decreased motility in the GF gut and suggest that the enteric neurons are responding to the bacterial status of the intestinal lumen. Our findings of reduced excitability in AH neurons of the myenteric plexus in GF mice are entirely consistent with these reports of decreased motility.

The current study is the first examination of the electrophysiological properties of myenteric neurons in the complete absence of colonizing commensal intestinal microbiota. Our novel findings of decreased excitability in myenteric sensory neurons in the absence of intestinal microbiota, and that the acquisition of a normal gut microbiome increased the excitability of AH sensory neurons in GF mice, offers a window into how intestinal bacteria can induce changes in the functions of the enteric nervous system. It also provides a preliminary look into how the nervous system as a whole may be initially responding to alterations and acquisition of the normal balance of intestinal bacterial populations.


The authors are grateful for the start-up and continuing support received from the McMaster Brain-Body Institute, St. Joseph’s Healthcare, Hamilton, and operating funds from National Science and Engineering Research Council of Canada (NSERC). Graduate stipend support (to K.A. McVey Neufeld) was provided by Ontario Graduate Scholarship and Ontario Graduate Scholarship in Science and Technology. This work was conducted in partial fulfillment of requirements for PhD in which J.A. Foster was the supervisor.


The study was supported by the McMaster Brain-Body Institute at St. Joseph’s Healthcare, Hamilton, and operating funds from National Science and Engineering Research Council of Canada (NSERC #371955-2009). Graduate stipend support (to K.A. McVey Neufeld) was provided by Ontario Graduate Scholarship and Ontario Graduate Scholarship in Science and Technology.


The authors have no competing interests to declare. An abstract of this data was published in Neurogastroenterology & Motility, Special Issue: Abstracts of the 16th Neurogastroenterology & Motility Meeting, September 16–18 2011, St Louis, Missouri, USA.

Author Contribution

JF, WK, and JB were responsible for the study design; KMN was responsible for conducting experiments; Data analysis was completed by KMN and WK. All investigators contributed to manuscript preparation.