Extrinsic Primary Afferent Neurons Link Visceral Pain to Colon Motility Through a Spinal Reflex in Mice

Abstract

BACKGROUND & AIMS:

Proper colon function requires signals from extrinsic primary afferent neurons (ExPANs) located in spinal ganglia. Most ExPANs express the vanilloid receptor TRPV1, and a dense plexus of TRPV1-positive fibers is found around myenteric neurons. Capsaicin, a TRPV1 agonist, can initiate activity in myenteric neurons and produce muscle contraction. ExPANs might therefore form motility-regulating synapses onto myenteric neurons. ExPANs mediate visceral pain, and myenteric neurons mediate colon motility, so we investigated communication between ExPANs and myenteric neurons and the circuits by which ExPANs modulate colon function.

METHODS:

In live mice and colon tissues that express a transgene encoding the calcium indicator GCaMP, we visualized levels of activity in myenteric neurons during smooth muscle contractions induced by application of capsaicin, direct colon stimulation, stimulation of ExPANs, or stimulation of preganglionic parasympathetic neuron (PPN) axons. To localize central targets of ExPANs, we optogenetically activated TRPV1-expressing ExPANs in live mice and then quantified Fos immunoreactivity to identify activated spinal neurons.

RESULTS:

Focal electrical stimulation of mouse colon produced phased-locked calcium signals in myenteric neurons and produced colon contractions. Stimulation of the L6 ventral root, which contains PPN axons, also produced myenteric activation and contractions that were comparable to those of direct colon stimulation. Surprisingly, capsaicin application to the isolated L6 dorsal root ganglia, which produced robust calcium signals in neurons throughout the ganglion, did not activate myenteric neurons. Electrical activation of the ganglia, which activated even more neurons than capsaicin, did not produce myenteric activation or contractions unless the spinal cord was intact, indicating that a complete afferent-to-efferent (PPN) circuit was necessary for ExPANs to regulate myenteric neurons. In TRPV1-channel rhodopsin-2 mice, light activation of ExPANs induced a pain-like visceromotor response and expression of Fos in spinal PPN neurons.

CONCLUSIONS:

In mice, ExPANs regulate myenteric neuron activity and smooth muscle contraction via a parasympathetic spinal circuit, linking sensation and pain to motility.

Graphical Abstract

The gastrointestinal tract is innervated by intrinsic neurons (the enteric nervous system [ENS]), extrinsic autonomic neurons, and extrinsic primary afferent neurons (ExPANs). In addition to the well-described afferent role of sensory neurons, ExPANs are hypothesized to have local effector functions on myenteric neurons via neuropeptides (eg, substance P and CGRP) released from their peripheral terminals in the colon.^1,2 Previous studies indicate that the majority of colon ExPANs express transient receptor potential vanilloid 1 (TRPV1),^3,4 a nonselective cation channel activated by capsaicin, and that applied capsaicin modulates colon contractility in rodents,⁵ large mammals,⁶ and humans.⁷ Supporting the idea that these effects are mediated by neuropeptides released from ExPANs, CGRP- and TRPV1-immunoreactive ExPAN fibers have been localized to myenteric ganglia, where they form dense networks around myenteric neurons.^8–12

Furthermore, a recent calcium imaging study showed capsaicin-evoked transients in ExPAN endings within myenteric ganglia that were temporally correlated with increased activity in myenteric ganglion neurons.¹³ These findings indicate that peptidergic ExPANs can locally and directly influence myenteric neuron activity to alter colon contractility. A significant and acknowledged caveat for this conclusion is that the peptides implicated in neurotransmission from ExPANs to myenteric neurons are also expressed in myenteric neurons^14,15 and that TRPV1-expressing myenteric neurons have been reported.¹⁶ Additionally, functional TRPV1 expression has been reported in other nonneuronal cell populations that reside in the colon (eg, epithelial cells, endothelial cells, and immune cells).^17–20 Thus, the goal of this study was to produce direct evidence confirming that ExPANs make functional, efferent connections with myenteric neurons and, through this connection, affect colon motility.

We used the genetically encoded calcium indicator, GCaMP6, to image the activity of myenteric neurons while simultaneously tracking stimulus-evoked movement of colon tissue, indicative of smooth muscle contraction. Several recent reports have pioneered the use of GCaMP to image activity of myenteric neurons^21–23 in ex vivo intestinal preparations. Using the embryonic promoter E2a to drive GCaMP expression in neuronal and nonneuronal populations,²⁴ we first characterized GCaMP signals (spontaneous and stimulus-evoked) from myenteric neurons ex vivo and then validated these findings in a newly developed in vivo model. Myenteric responses induced by stimulation of ExPANs and preganglionic parasympathetic neurons (PPNs) were also measured in an ex vivo preparation in which the L6 dorsal and ventral roots, containing ExPANs and PPN axons, respectively, were dissected in continuity with the pelvic nerve and colon. We found that (1) direct stimulation of the colon, either oral or anal to the imaging field, activates different neural circuits and evokes distinct patterns of smooth muscle contractility; (2) parasympathetic pathways drive colon motility via ascending myenteric pathways; and (3) ExPANs influence myenteric neuron activity indirectly by engaging parasympathetic pathways through a spinal reflex rather than through direct interaction via neuropeptide release from peripheral terminals in the colon. Although we replicated previous findings that capsaicin applied to the colon produces myenteric neuron activity and changes in contractility, this was not due to activation of ExPANs because capsaicin applied to the dorsal root ganglion (DRG), specifically, did not produce responses. Despite the close approximation of ExPAN axons to myenteric neurons, there are functional/anatomic constraints that prevent sensory information from directly altering myenteric neurons, ensuring that autonomic pathways are the final site for integration of extrinsic sensory information from the colon. Our findings also indicate that disease states (eg, irritable bowel syndrome or inflammatory bowel disease) or injury (eg, trauma or surgery) that sensitize colon afferents will not only produce pain but can also have profound effects on motility via an ExPAN-to-parasympathetic spinal reflex.

Materials and Methods

See the Supplementary Material for a detailed description of the materials and methods.

Animals

Male and female mice aged 8–12 weeks were analyzed in these studies. E2a-Cre mice (RRID:IMSR_JAX:003724) were crossed with mice containing a floxed-STOP-GCaMP6s sequence in the Rosa26 locus (Ai96 mice, RRID:IMSR_JAX:028866). TRPV1-Cre mice (RRID:IMSR_JAX:017769) were crossed with mice containing a floxed-STOP-ChR2 sequence (Ai32 mice, RRID:IMSR_JAX:012569). Animals were housed in a facility approved by Assessment and Accreditation of Laboratory Animal Care (AAALAC), with a 12-hour light/dark cycle and free access to water and standard chow. Animal use protocols were approved by the Institutional Animal Care and Use Committee at the University of Pittsburgh.

Ex Vivo GCaMP6 Imaging (Colon Only)

Colons were removed from E2a-GCaMP6 mice, cut open longitudinally, and pinned out (mucosa facing down) in a Sylgard-lined dish (Dow Corning, Midland, MI), superfused with carbogenated (95% O₂, 5% CO₂) artificial cerebrospinal fluid (ACSF) and maintained at 35°C-37°C. ACSF was prepared on the day of the experiment containing (in mmol/L) 117.9 NaCl, 4.7 KCL, 25 NaHCO₃, 1.3 NaH₂PO₄, 1.2 MgSO₄·7H₂O, 2.5 CaCl₂, 11.1 d-glucose, 2 sodium butyrate, and 20 sodium acetate. Nifedipine (4 μmol/L) (Sigma-Aldrich, St. Louis, MO), an L-type calcium channel blocker, was added to ACSF in all ex vivo experiments to improve stability for analysis of calcium imaging data. In full-thickness colon preparations, 4 μmol/L nifedipine consistently reduced spontaneous contractions but did not abolish activity in interstitial cells of Cajal; importantly, movement patterns evoked by electrical stimulation were nearly identical in a subset of experiments comparing 1 (routinely used in stripped preparations) and 4 μmol/L nifedipine (Supplementary Figure 1). Indomethacin (3 μmol/L) (Sigma-Aldrich), a cyclooxygenase inhibitor, was also added to ACSF to suppress potential actions of endogenous prostaglandins. GCaMP signals in myenteric neurons were imaged with an upright DM6000FS Leica fluorescent microscope (Leica, Buffalo Grove, IL) and a Prime 95B Scientific Complementary Metal-Oxide-Semiconductor (CMOS) camera (Photometrics, Roper Scientific, Tucson, AZ) using 20× or 40× objective lens, and images were collected with Metamorph software (Molecular Devices, San Jose, CA) at a 40-Hz sampling rate and 25-ms exposure time. To examine evoked activity, concentric electrodes were placed on the colon 5 mm oral and anal to the imaging field, and electrical pulses were delivered to the colon during image acquisition of myenteric neurons. A pulse duration of 100 μs was chosen because it is too short to elicit muscle fiber contractions but reliably induces neuronal action potential firing.

In Vivo GCaMP6 Imaging

Mice were maintained in a nose cone administering 1.5%–2% isoflurane (vaporized with 95% O₂/5% CO₂). The abdominal wall was cut longitudinally, and a reservoir was created by securing the abdominal wall to a metal ring positioned over the abdomen, allowing superfusion of oxygenated ACSF without nifedipine. Due to a slower flow rate, the temperature of the reservoir was maintained at slightly lower temperatures compared with ex vivo (33°C–35°C). A Sylgard-lined platform was lowered into the reservoir, and the proximal colon was pinned flat onto the platform. GCaMP imaging was performed with the same methods as for ex vivo preparations.

Ex Vivo Colon Nerve Preparation

By using a protocol modified from Brierley et al,²⁵ the distal colon was isolated along with the intact pelvic nerve, L6 DRG, and L6 dorsal and ventral roots. (DRG and roots remained in the spinal column.) The spinal column and attached colon were pinned to a dish superfused with ACSF maintained at 35°–37°C. For ex vivo preparations in which the spinal cord was kept intact (from T12–S3), the dissection was performed in ice-cold sucrose ACSF, and the preparation was maintained at 31°C during imaging, which improved spinal cord viability during experiments. In some experiments, a special recording chamber was used that contains 2 compartments separated by a removable wall sealed off with grease. The L6 DRG and L6 dorsal and ventral roots occupied one compartment, and the colon was in the other, allowing us to apply drugs to the DRG without exposing the colon. GCaMP signals in DRG neurons and myenteric neurons were imaged as previously described.

Electromyographic Recording of Visceromotor Response to Laser Stimulation

Visceromotor response (VMR) to laser stimulation of the colon in TRPV1–channelrhodopsin-2 (ChR2) mice or control littermates was measured using previously described protocols.²⁶

Immunohistochemistry in Spinal Cord

After laser stimulation, mice were deeply anesthetized with isoflurane and perfused with 4% paraformaldehyde in phosphate-buffered saline; lumbosacral spinal cord segments were dissected, immunolabeled, and quantified as previously described.²⁷

Drug Preparation

Tetrodotoxin (TTX) 0.5 μmol/L (Sigma-Aldrich), hexamethonium (HEX) 300 μmol/L (Sigma-Aldric), or capsaicin 10 μmol/L (Sigma-Aldrich) were dissolved into freshly oxygenated ACSF on the day of the experiment. Responses were imaged 10 minutes after the addition of TTX or HEX. For experiments using capsaicin, responses were imaged immediately after addition to bath solution for a total of 10 minutes or until the imaging field moved out of the visual field and could not be relocated.

Data Analysis

Image files collected in Metamorph were exported to ImageJ (National Institutes of Health, Bethesda, MD). The amplitudes of GCaMP signals were analyzed and quantified as previously described²⁸ by calculating ΔF/F₀ as % = [(F − F₀)/F₀] × 100, where F is the peak fluorescence signal and F₀ is the mean fluorescence signal at baseline. Tissue movement in response to stimuli was determined using a Template-Matching plugin in ImageJ, which quantifies movement along the x- and y-axes, representing the circular and longitudinal muscle, respectively. Statistical tests were performed in Excel (Microsoft, Redmond, WA) and GraphPad Prism (GraphPad Software, San Diego, CA). Data are expressed as mean ± standard error of the mean, where n represents mice used, unless indicated otherwise. Statistical tests are indicated in the Results section; significance is defined as P < .05.

Results

Comparison of Spontaneous and Evoked GCaMP Activity in Myenteric Neurons in Ex Vivo and In Vivo Colon Preparations

In the absence of stimulation, approximately 20% of myenteric neurons per ganglion exhibited GCaMP signals that did not appear to be synchronized, based on the absence of any discernable pattern over the course of recordings lasting 30 seconds (Figure 1A and B). The percentage of cells displaying spontaneous activity were not significantly different in ex vivo and in vivo preparations (ex vivo: 19.2% ± 3.14%, n = 5; in vivo: 15.88% ± 1.38%, n = 5; P = .3611) (Figure 1C). The main excitatory neurotransmitter involved in synaptic communication in the ENS is acetylcholine,²⁹ and therefore we used HEX (300 μmol/L), a nicotinic receptor antagonist, to measure the effects of blocking cholinergic synaptic transmission on spontaneous activity. HEX significantly reduced the percentage of spontaneously active myenteric neurons to 7.9% ± 3.0% in the ex vivo preparation (P = .0301, n = 3) (Figure 1D and Supplementary Figure 2) and to 8.6% ± 0.6% in vivo (P= .0466, n = 30) (Figure 1E). TTX (0.5 μmol/L), a voltage-gated sodium channel blocker, also significantly decreased the percentage of myenteric neurons that displayed spontaneous GCaMP activity to 7.1% ± 1.2% in the ex vivo colon preparation (P = .004, n = 6; not tested in vivo). These pharmacologic studies indicate that the majority of spontaneous GCaMP signals are the result of synaptic communication.

Figure 1.

Characterization of spontaneous and evoked GCaMP activity in myenteric neurons. (A, B) Time-lapse color-coded image (left) and ΔF/F₀ traces (right) of myenteric neurons without stimulation in (A) ex vivo and (B) in vivo preparations. Each pixel in a collection of GCaMP images is assigned a color corresponding to when that particular pixel reached peak fluorescence. (C) There were no significant differences in percentage of spontaneously active cells between ex vivo and in vivo preparations. (D, E) HEX significantly decreased the proportion of myenteric neurons that displayed spontaneous GCaMP activity (D) ex vivo and (E) in vivo. (F) Amplitudes of responses were not significantly different across frequencies. (G) Percentage of responsive neurons per myenteric ganglion increased with increasing frequency. (H) Time-lapse color-coded image (H_i), ΔF/F₀ traces of responsive myenteric neurons (H_ii), and resulting tissue movement (H_iii: black, circular muscle axis; gray, longitudinal muscle axis) to 20-Hz electrical stimulation of the colon (red arrow). (I) TTX significantly decreased the percentage of responsive neurons per ganglion. (J) TTX significantly decreased evoked movement. (K–M) No significant differences in the amplitude of (K) evoked responses or (L) response latency were measured in ex vivo and in vivo experiments, but (M) movement latency was significantly increased in vivo. *P < .05, **P < .01, (C, K–M) unpaired and (D, E, I–J) paired Student t test, 1-way analysis of variance (ANOVA), Tukey post hoc test (F, G).

We then characterized GCaMP signals from myenteric neurons evoked by focal electrical stimulation (5 mm from the imaging field) of the colon at various frequencies in the ex vivo preparation. Although the amplitude of response (ΔF) was not significantly different across frequencies (1 Hz: 16.3 ± 3.04 ΔF, 5 Hz: 16.9 ± 0.96 ΔF, 10 Hz: 16.28 ± 1.54 ΔF, 20 Hz: 18.9 ± 1.77 ΔF; P = .6350; n = 3) (Figure 1F), the percentage of responsive neurons per ganglion increased with increasing stimulation frequency (1 Hz: 7.0% ± 1.8%, 5 Hz: 19.0% ± 2.8%, 10 Hz: 23.9% ± 3.4%, 20 Hz: 28.2% ± 3.0%; P = .0055; n = 3) (Figure 1G). Stimulation at 20 Hz activated the most myenteric neurons in a ganglion; therefore, this frequency was used in the remainder of experiments. Neural activity was consistently followed by smooth muscle contraction (0.802 ± 0.202 seconds after stimulation), as indicated by movement within the imaging field (Figure 1H). In a separate set of experiments, we recorded changes in tension (mN) while imaging the colon to validate that tissue displacement in the x- and y-axes was directly correlated to circular and longitudinal muscle contraction (Supplementary Figure 3). TTX significantly decreased the proportion of responding neurons (P = .0002, n = 6) (Figure 1I) and decreased evoked movement (P = .0002, n = 6) (Figure 1J), confirming that tissue movement is neuronally mediated and is not the consequence of direct smooth muscle cell activation by the electrical stimulus. In vivo, the GCaMP signal in myenteric neurons and subsequent muscle contractions evoked by 20-Hz stimulation were similar to that seen ex vivo. There were no significant differences in the amplitude (P = .4119, n = 5) (Figure 1K) or latency of response (P = .2816, n = 5) (Figure 1L), but tissue movement occurred later in vivo (P = .0115, n = 5) (Figure 1M). Regardless, in both preparations, more than half a second separates neural activity and muscle contraction, suggesting additional processing between these events (eg, engagement of slow-wave activity from interstitial cells of Cajal).⁸

Distinct Ascending and Descending Myenteric Circuits Produce Different Patterns of Contraction

The movement of fecal matter through the colon is thought to be mediated by excitatory ascending and inhibitory descending pathways.^30–32 To determine how these pathways differentially regulate myenteric neuron responses and motility, we imaged GCaMP signals of myenteric neurons and measured contractile responses produced by electrical stimulation of the colon oral and anal to the imaging field in both the ex vivo and in vivo preparations (Figure 2A and B). Stimulation of the colon either anally or orally of the imaging field activated different subsets of myenteric neurons within a ganglion. Of the myenteric neurons that were responsive to colon stimulation ex vivo, 35.4% ± 4.0% responded only to oral stimulation, another 33.7% ± 4.9% responded only to anal stimulation, and 30.8% ± 1.8% responded to stimulation at both locations (n = 5). In vivo, slightly more myenteric neurons responded only to anal stimulation (44.7% ± 6.5%) compared with the percentage that responded only to oral stimulation (29.2% ± 4.4%) or stimulation at both sites (26.1% ± 5.0%); however, this difference was not significant (P = .1022, n = 5). There were distinct temporal patterns of activation due to oral and anal stimulation: although relatively early responses were observed after both oral and anal stimulation (2C_i, 2E_i), anal stimulation led to a second, delayed wave of responses that was rarely observed after oral stimulation (for example, see traces in Figure 2A_ii). Because of the variability in the ex vivo preparation, this trend was only significant in vivo (ex vivo: P = .1148, n = 5, Figure 2C_ii; in vivo: P = .0018, n = 5, Figure 2E_ii). Many of the neurons that exhibited late responses to anal stimulation also responded to oral stimulation (53%), suggesting that the late response is due to reverberation in enteric circuits such that the ascending stimulus activates proximally located neurons that then send information back down the colon to produce the late response.

Figure 2.

Stimulation oral and anal to the imaging field activates distinct neural circuits and leads to different patterns of smooth muscle contractility. (A, B) Time-lapse color-coded images (i) and ΔF/F₀ traces (ii) of responses to stimulation oral (left) and anal (right) stimulation in (A) ex vivo and (B) in vivo preparations. (C, E) Percentage of neurons per ganglion that exhibit early (i) and late (ii) responses due to oral and anal stimulation in (C) ex vivo and (E) in vivo preparations. (D, F) Movement traces in the circular (i) and longitudinal (ii) muscle axes evoked by oral (black) and anal (gray) stimulation in (D) ex vivo and (F) in vivo preparations. Positive values represent lengthening (circular) or movement (longitudinal) in the oral direction; negative values represent shortening (circular) or movement (longitudinal) in the anal direction.(G, H) Scatterplot of movement coordinates from (G) ex vivo and (H) in vivo preparations. (I–L) HEX significantly decreased the percentage of responsive neurons to (I, K) oral and (J, L) anal stimulation in vivo and ex vivo. (C, E, I–L) *P < .05, **P < .01, paired Student t tests.

Oral and anal colon stimulation evoked movement in opposing directions along the circular and longitudinal muscle axes ex vivo (Figure 2D_i, D_ii, and G) and in vivo (Figure 2F_i, F_ii, and H). As shown in the scatterplots (Figure 2G and H), oral stimulation typically led to longitudinal movement in the oral direction and circular relaxation, whereas anal stimulation produced longitudinal movement in the anal direction and circular contraction. In in vivo preparations, evoked muscle movement was not as stereotyped as movement observed ex vivo, but it still followed the overall trend. The addition of HEX significantly decreased the percentage of cells per ganglion that responded to oral and anal stimulation in both ex vivo and in vivo preparations (ex vivo: oral stimulation, P = .0329, n = 3, Figure 2I; anal stimulation, P = .0215, n = 3, Figure 2J; in vivo: oral stimulation, P = .0250, n = 3, Figure 2K; anal stimulation, P = .0243, n = 3, Figure 2L), confirming that the majority of responses were synaptic in nature rather than the result of direct electrical activation. These data support the notion that ascending and descending pathways exist in the colon and that specific myenteric circuits can be activated by using external stimulation to produce distinct patterns of functional output.

Comparison of Myenteric Responses Produced by Capsaicin Applied to Colon vs Capsaicin Applied to Extrinsic Primary Afferent Neurons-Containing Dorsal Root Ganglion Neurons

We next sought to determine whether ExPANs make functional connections with myenteric neurons, and if so, which myenteric pathways they activate. Recent calcium imaging studies have provided evidence that the TRPV1 agonist, capsaicin, applied to the colon activates ExPAN terminals and leads to increased activity in myenteric neurons.³ To confirm that capsaicin responses in ExPANs directly activate myenteric neurons, we compared myenteric responses produced by capsaicin applied specifically to the DRG in our ex vivo preparation, in which the L6 DRG, pelvic nerve, and colon are dissected in continuity. Because the DRG was kept in a chamber separate from the colon (Figure 3A), we were able to use capsaicin to activate ExPANs without exposing the colon to the drug. Capsaicin applied to the DRG evoked widespread activation of L6 DRG neurons, but to a lesser extent than electrical stimulation (Figure 3B); this was expected because capsaicin-sensitive TRPV1 neurons make up a smaller subset of DRG neurons. When applied to the DRG, capsaicin did not increase activity in myenteric neurons above spontaneous levels (Figure 3C–E) and did not evoke muscle contraction (Figure 3F). However, capsaicin applied to the colon significantly increased activity (Figure 3C–E) and produced strong muscle contractions that oftentimes caused the tissue to move completely out of the imaging field (Figure 3F). Thus, we were able to reproduce the results from previous studies for capsaicin applied directly to the colon but could not phenocopy this result when capsaicin was applied to the L6 DRG that contains the neurons thought to be responsible for the response when capsaicin was applied directly to the colon. These results indicate that capsaicin application to the colon is activating cells other than ExPANs. Furthermore, after capsaicin application to the colon, spontaneous and evoked myenteric neuron activity was significantly decreased for up to 30 minutes after washout (Supplementary Figure 4). The intensity of the acute response to colon-applied capsaicin and the accompanying loss of spontaneous and evoked activity was never seen after any of the paradigms using electrical stimulation. These observations suggest that the colon-applied capsaicin response was the result of a supernormal stimulus that may not have physiologic relevance.

Figure 3.

Specific activation of ExPANs with capsaicin (CAP) does not reproduce myenteric neuron responses and smooth muscle contractions induced by capsaicin applied to the colon. (A) Schematic of ex vivo preparation in which 2 chambers keep the L6 DRG and dorsal and ventral roots separate from the colon. (B) Time-lapse color-coded images of DRG responses to electrical stimulation of the dorsal root (left) and CAP (10 μmol/L, right); electrical stimulation produced responses significantly greater in amplitude compared with CAP (B_ii). (C) GCaMP traces of spontaneous activity in myenteric neurons (top) in response to CAP applied to the DRG (middle) or applied to the colon (bottom); arrow indicates the addition of CAP to bath solution. (D, E) CAP applied to the DRG had no effect on myenteric neuron activity, whereas CAP applied to the colon significantly increased myenteric neuron activity: (D) percentage of responding neurons and (E) response amplitude compared with levels of spontaneous activity. (F) CAP applied to the DRG had no effect on tissue movement, whereas CAP applied to the colon significantly increased tissue movement. *P < .05, **P < .01, (B) paired Student t test or (D–F) repeated-measures 1-way ANOVA. M, mol/L; Max, maximum.

Parasympathetic Pathways, but not ExPANs, Directly Influence Myenteric Activity and Produce Smooth Muscle Contraction by Engaging Ascending Pathways

Our data indicated that our electrical stimulus was more efficacious in activating the majority of ExPANs, and therefore we next used electrical stimulation rather than pharmacologic means to determine if efferent communication exists between ExPANs and myenteric neurons. We selectively activated ExPANs by stimulating the L6 dorsal root in our ex vivo colon preparation (Figure 4A), and as a positive control, we also stimulated the ventral root, which contains PPN axons that have been shown to have excitatory effects on colon motility.³³ Activation of ExPANs with dorsal root stimulation (100 μs at 1, 5, 10, or 20 Hz for 1 or 5 seconds) did not produce responses in myenteric neurons and did not cause tissue movement, although virtually all cell bodies in the DRG exhibited a saturating calcium signal (n = 5) (Figure 4B and D). However, in the same preparations, activation of PPNs using ventral root stimulation (100 μs at 20 Hz for 1 second) produced responses in 25.44% ± 5.0% of myenteric neurons along the distal colon (n = 5) (Figure 4C and D). This increase in GCaMP activity was followed by smooth muscle contraction 1.51 ± 0.28s later, as indicated by movement within the imaging field. HEX completely abolished myenteric neuron responses (P = .0014, n = 5) (Figure 4E) and subsequent tissue movement (P = .0225, n = 5) (Figure 4F) evoked by ventral root stimulation, suggesting that these responses were likely due to the canonical, disynaptic parasympathetic circuit. Figure 4G also indicates that stimulation of the L6 ventral root produces a pattern of muscle contraction that is similar to that evoked by focal anal stimulation, suggesting that stimulation of PPNs preferentially engages ascending (anal to oral) pathways.

Figure 4.

Extrinsic parasympathetic pathways, but not ExPANs, directly influence myenteric neuron activity and produce smooth muscle contractions in the colon. (A) Image of colon nerve ex vivo preparation with dorsal and ventral roots cut and spinal cord removed from spinal column. Black dots indicate imaging fields. (B) GCaMP signal in L6 DRG neurons (B_i), time-lapse color-coded image of myenteric neuron activity (B_ii, B_iii), and evoked tissue movement (B_iv) due to dorsal root stimulation (red arrow). (C) GCaMP signal in L6 DRG neurons (note the presence of a few ventral root afferents) (C_i), time-lapse color-coded image of myenteric neuron activity (C_ii and C_iii), and tissue movement (C_iv) due to ventral root stimulation (red arrow). (D) Percentage of neurons that respond to dorsal and ventral root stimulation when spinal cord is removed. (E, F) HEX significantly decreased the (E) percentage of responses and (F) tissue movement due to ventral root stimulation. (G) Scatterplot of movement coordinates due to ventral root stimulation (red) overlaid on the oral/anal movement scatterplot from Figure 2 for comparison. *P < .05, **P < .01, using paired Student t test (E, F).

Extrinsic Primary Afferent Neurons Indirectly Influence Myenteric Neuron Activity and Produce Smooth Muscle Contraction via a Spinal Reflex

These data suggest that ExPANs do not directly influence myenteric neurons and confirm that PPNs drive colon motility via myenteric neuron activation. We were surprised by the lack of local communication from ExPANs to myenteric neurons, but this raised the question of whether ExPANs could indirectly influence ENS activity through spinal cord circuits. To test this, we modified our ex vivo preparation so that the portion of the spinal cord containing PPNs remained viable and intact (Figure 5A). Dorsal root stimulation in preparations with an intact spinal cord activated approximately 10% of myenteric neurons and produced movement of the imaging field similar to that produced by ventral root stimulation (Figure 5B, C, and G). Compared with ventral root stimulation, the percentage of myenteric neurons that responded to dorsal root stimulation was significantly lower (P = .0030, n = 6) and the amplitude of responses to dorsal root stimulation was lower than responses to ventral root stimulation (dorsal root, 11.55 ± 1.1 ΔF; ventral root, 16.69 ± 1.1 ΔF; P = .0071, n = 6) (Figure 5D), likely due to divergence of afferent input within the spinal cord. The latency of responses to dorsal root stimulation tended to be longer than the latency of responses to ventral root stimulation; however, this difference was not significant (dorsal root, 1.12 ± 0.57 s; ventral root, 0.178 ± 0.07 s; P = .1301, n = 6) (Figure 5E). Importantly, responses and movement were abolished when the dorsal or ventral root was cut from the spinal cord. Therefore, our data strongly suggest that ExPANs indirectly influence the activity of myenteric neurons by engaging spinal reflexes that activate parasympathetic pathways to the colon.

Figure 5.

ExPANs indirectly influence myenteric neuron activity and produce smooth muscle contractions in the colon. (A) Image of colon nerve ex vivo preparation with the spinal cord intact. White dots indicate fields. (B) Time-lapse color-coded image (B_i), ΔF/F₀ traces of responsive myenteric neurons (B_ii), and tissue movement (B_iii) due to dorsal root stimulation with spinal cord intact (red arrow). (C) Time-lapse color-coded image (C_i), ΔF/F₀ traces of responsive myenteric neurons (C_ii), and tissue movement (C_iii) due to dorsal root stimulation (red arrow). (D) Amplitude (ΔF/F₀) of GCaMP responses to ventral and dorsal root stimulation. (E) Latency of responses to ventral and dorsal root stimulation. (F) Percentage of active myenteric neurons during recordings (per 1-second bin) in different experimental conditions. (G) Scatterplot of movement coordinates due to dorsal root stimulation (blue) overlaid on oral/anal/ventral root scatterplot from Figures 2 and 5 for comparison. **P < .01 paired Student t test (D).

In Vivo Confirmation That Stimulation of Extrinsic Primary Afferent Neurons Activates Preganglionic Parasympathetic Neurons and Produces Visceral Pain-Like Responses

Our previous studies show that optogenetic stimulation of ExPANs (via ChR2) evokes a VMR,²⁶ a surrogate for visceral pain-like responses.³⁴ We used the same paradigm combined with Fos staining to identify PPNs in the spinal cord involved in the spinal reflex shown by GCaMP imaging. In TRPV1-ChR2 mice, blue-light stimulation of ExPANs induced a VMR (response rate, 93.3% ± 6.7%) (Figure 6D) and led to the activation of cells within the sacral parasympathetic nucleus (SPN) and dorsal commissure (DCM), an autonomic relay nucleus³⁵ (n = 3) (Figure 6B). The same stimulation in control littermates did not produce a VMR (Figure 6C) and led to significantly fewer cells with Fos expression in the SPN and DCM (SPN: P = .0362, Figure 6E; DCM: P = .0240, Figure 6F). These studies provide in vivo evidence linking visceral pain to colon motility via activation of an ExPAN to parasympathetic spinal reflex as modeled in Figure 7.

Figure 6.

Optogenetic stimulation of ExPANs activates neurons in the SPN and DCM. (A) Lumbosacral spinal cord section from littermate control after blue-light stimulation with no c-fos expression in the SPN (A_ii and A_iv) or DCM (A_iii). (B) Lumbosacral spinal cord section from TRPV1-ChR2YFP mouse shows yellow fluorescent protein expression in sensory afferents and c-fos expression in activated cells in response to optogenetic stimulation. Note the close apposition of afferent fibers to c-fos-positive cells in the SPN (Bii, Biv) and DCM (B_iii). (C) Electromyography recordings of the visceromotor response evoked by blue-light stimulation in TRPV1-ChR2 (C_i) and control mice (C_ii). (D, E) Numbers of c-fos-positive cells in the (D) SPN and (E) DCM were significantly higher in TRPV1-ChR2 mice compared with controls. *P < .05, unpaired Student t test.

Figure 7.

Diagram of hypothesized and newly proposed pathways of lumbosacral ExPANs to the colon. (A) ExPAN terminals localized to the myenteric plexus have led to the hypothesis that ExPANs can directly influence myenteric neuron activity via release of neuropeptides at their peripheral terminals. (B) Our findings indicate that rather than local modulation, ExPANs indirectly influence myenteric neuron activity and colon motility via central modulation through a sensory-parasympathetic spinal reflex. M, mucosa; MP, myenteric plexus; SP, submucosal plexus.

Discussion

The primary aim of this study was to investigate how colon ExPANs influence myenteric neuron activity and regulate motility. Therefore, we developed GCaMP imaging methods to manipulate both intrinsic and extrinsic components of the ENS. To interrogate intrinsic pathways, we imaged spontaneous and evoked activity in the myenteric plexus. Using direct electrical stimulation of the colon, we showed that myenteric neurons in a given ganglion respond differentially to activation of ascending and descending pathways, resulting in distinct patterns of contractility. Ex vivo findings were then validated in an in vivo system. Although recording and GCaMP imaging of myenteric neurons in vivo has been previously described,³⁶ our newly developed model enables imaging of myenteric neurons with unprecedented spatial resolution and simultaneous measure of colon motility. The ex vivo colon preparation has proven to be a valuable tool, used for decades to build our current knowledge and understanding of the ENS and colon motility.^{30,31,37–39} The studies described here are the first to directly compare ENS activity and colon function in ex vivo and in vivo preparations.

To investigate extrinsic connections to myenteric neurons, we used a colon–pelvic nerve–L6 DRG ex vivo preparation. Surprisingly, capsaicin-mediated activation of ExPANs had no effect on myenteric neuron activity. This was unexpected, given previous evidence that ExPANs have local effector functions via neuropeptide release at their terminals,^1,40–43 which are in close proximity to myenteric neurons.^8,12,13 A previous study¹³ reported myenteric neuron calcium influx after capsaicin was applied directly to the colon, but it is unclear whether this activation occurs via ExPANs or via another mechanism. We saw similar myenteric activity upon adding capsaicin to the colon but did not see activity after activating ExPANs with capsaicin. We then activated ExPANs in the L6 DRG to an even greater extent using various electrical stimulation parameters (1, 5, 10, and 20 Hz; 1-second and 5-second duration). Despite the observation that 20-Hz stimulation activated the majority of DRG neurons, this was not sufficient to activate myenteric neurons or elicit muscle contractions, suggesting that ExPANs were not responsible for the observed myenteric activity elicited when capsaicin was applied directly to the colon.

There are several reasons why direct application of capsaicin to the colon resulted in myenteric neuron activity but capsaicin activation of ExPANs did not. It is possible that myenteric responses to colon-applied capsaicin is due to TRPV1 expression on other cell types^16–20 that can, in turn, activate myenteric neurons. Another possibility is that the addition of capsaicin to the colon may be more efficient in promoting local neurotransmitter and peptide release from ExPAN terminals. However, if that is the case, it is difficult to imagine the conditions under which this would happen in vivo because the resulting activation could not be repeated within the time frame of our experiments and was followed by dramatic suppression of both myenteric activity and muscle contractions. This suggests that if this circuit is activated in vivo, it is likely to shut down colon function, a response that is hard to reconcile with normal colon activity. Our observation is instead consistent with a capsaicin-induced injury response of ExPAN terminals. Application of high concentrations of capsaicin are used both experimentally and medically to desensitize and/or induce degeneration of afferent endings as a way to block neurotransmission in nociceptors.^44–47 A capsaicin-induced damage response would produce a one-time release of neuromodulatory substances that could produce the impressive, nonrepeatable response seen in myenteric neurons.

To interrogate whether ExPANs could indirectly influence myenteric neuron activity, we repeated our ExPAN stimulation experiments with the spinal cord intact in the ex vivo preparation. Indeed, electrical stimulation of the dorsal root in this preparation resulted in myenteric neuron responses and smooth muscle contractions, indicating that ExPANs act as indirect effectors of myenteric neuron activity via a spinal cord–parasympathetic pathway. This is consistent with previous studies that have suggested that ExPANs form connections with preganglionic parasympathetic neurons in the SPN.^33,48,49 To further test these functional connections in vivo, we used transgenic mice that expressed ChR2 in ExPANs under the TRPV1 promoter (TRPV1-ChR2 mice) and used blue light (473 nm) from a fiberoptic cable inserted into the colon to activate ExPAN fibers. We have shown that optogenetic activation of ExPANs mimics natural stimulation and evokes visceral pain-like responses (ie, visceromotor response).²⁶ In the present experiments, brief bursts of blue-light stimulation in TRPV1-ChR2 mice, but not in control mice, induced Fos activation in the SPN, consistent with our hypothesis. The ability of activated ExPANs to drive SPN activity was predicted by studies of Forrest et al,⁴⁹ who showed dense ExPAN projections to the SPN.⁴⁹ The notion that ExPANs can regulate colon motility via a spinal parasympathetic reflex is further supported by studies in cats. In the work of De Groat and Krier,³³ ExPAN stimulation evoked activation of colon efferents (ie, PPNs) and produced propulsive contractions, even in animals that underwent a T10 spinal cord transection. Our experiments confirmed that activation of the ExPAN-parasympathetic fiber reflex results in contractile activity in the colon and also showed that myenteric neurons are a target of this reflex.

In gastrointestinal disorders such as irritable bowel syndrome and inflammatory bowel disease, dysmotility (both constipation and diarrhea) is often accompanied by pain. Previously, the simplest explanation for these coexisting symptoms was that inciting factors, such as infection and stress, acted independently on the ENS and on colon spinal afferents. For example, it has been shown in separate studies that such factors can alter the function of enteric neurons⁵⁰ and can lead to visceral pain via sensitization of colon afferents.⁵¹ Our studies provide evidence that changes in ExPANs alone may directly affect parasympathetic output to the colon and, through this pathway, dysregulate colon motility. This might explain why, in some functional bowel diseases, pain and dysmotility symptoms may be present without overt organ damage or pathology; if the primary insult selectively affects visceral afferents, changes in organ function could occur via a spinal reflex and be accompanied by pain. Furthermore, these results suggest that once visceral afferents become sensitized, autonomic dysfunction will be proportional to the level of visceral pain throughout disease progression. Mapping the functional connectivity among extrinsic and intrinsic ENS components in mouse models of inflammatory bowel disease/irritable bowel syndrome by using the optogenetic techniques described here will enable us to determine how symptoms arise in these disorders and provide valuable insights into the most effective ways to alleviate pain and dysmotility in patients.

WHAT YOU NEED TO KNOW.

BACKGROUND AND CONTEXT

It is unclear how extrinsic primary afferent neurons (ExPAN; ie, pain-sensing neurons) of the colon influence myenteric neuron activity and resulting colon motility.

NEW FINDINGS

ExPAN stimulation initiates myenteric neuron activity and subsequent colon contractions, but only in the presence of an intact spinal cord. This suggests that ExPANs regulate colon motility via a sensory-parasympathetic spinal circuit.

LIMITATIONS

More studies are required to define the myenteric neuron subtypes that are the target of the described circuit, and to investigate the details of the spinal portion of this circuit.

IMPACT

The described circuit demonstrates how visceral pain is linked to colon motility. Techniques introduced in this study will enable the investigation of how pain and dysmotility co-occur in GI disorders.

Abbreviations used in this paper:

ACSF

artificial cerebrospinal fluid

ChR2

channelrhodopsin-2

DCM

dorsal commissure

DRG

dorsal root ganglion

ENS

enteric nervous system

ExPAN

extrinsic primary afferent neuron

HEX

hexamethonium

PPN

preganglionic parasympathetic neurons

SPN

sacral parasympathetic nucleus

TRPV1

transient receptor potential vanilloid 1

TTX

tetrodotoxin

VMR

visceromotor response