An Enteric Occult Reflex Underlies Accommodation and Slow Transit in the Distal Large Bowel

Transcription

1 GASTROENTEROLOGY 2007;132: An Enteric Occult Reflex Underlies Accommodation and Slow Transit in the Distal Large Bowel EAMONN J. DICKSON, NICK J. SPENCER, GRANT W. HENNIG, PETER O. BAYGUINOV, JIM REN, DANTE J. HEREDIA, and TERENCE K. SMITH Department of Physiology & Cell Biology, University of Nevada School of Medicine, Reno, Nevada Background & Aims: Transit of fecal material through the human colon takes >30 hours, whereas transit through the small intestine takes 24 hours. The mechanisms underlying colonic storage and slow transit have yet to be elucidated. Our aim was to determine whether an intrinsic neural mechanism underlies these phenomena. Methods: Recordings were made from circular muscle (CM) cells and myenteric neurons in the isolated guinea pig distal colon using intracellular recordings and Ca 2 imaging techniques. Video imaging was used to determine the effects of colonic filling and pellet transit. Results: Circumferential stretch generated ongoing oral excitatory and anal inhibitory junction potentials in the CM. The application of longitudinal stretch inhibited all junction potentials. N- nitro-l-arginine (100 mol/l) completely reversed the inhibitory effects of longitudinal stretch suggesting that nitric oxide (NO) inhibited interneurons controlling peristaltic circuits. Ca 2 imaging in preparations that were stretched in both axes revealed ongoing firing in nnos ve descending neurons, even when synaptic transmission was blocked. Inhibitory postsynaptic potentials were evoked in mechanosensitive interneurons that were blocked by N- -nitro-l-arginine (100 mol/l). Pellet transit was inhibited by longitudinal stretch. Filling the colon with fluid led to colonic elongation and an inhibition of motility. Conclusions: Our data support the novel hypothesis that slow transit and accommodation are generated by release of NO from descending (nnos ve) interneurons triggered by colonic elongation. We refer to this powerful inhibitory reflex as the intrinsic occult reflex (hidden from observation) because it withdraws motor activity from the muscle. The large bowel has evolved to serve several major functions: to recover water and electrolytes from the contents of the intestinal lumen, to use bacteria to digest nutrients (in instances in which appropriate enzymes are lacking), and to prevent being tracked by scent-orientated predators and mark territories. 1 To perform these functions, transit of intraluminal contents through the human colon is slow, taking 30 hours, whereas transit through the small intestine takes 2 4 hours. 1 4 As intraluminal contents move down the large bowel, water and electrolytes are absorbed, causing the contents to become more viscous leading to stool formation. However, the mechanism underlying colonic accommodation and slow transit to promote absorption is unknown, 1 5 although increased segmental contractions and extrinsic sympathetic reflexes have been proposed. 1 4 Recently, an intrinsic neural mechanism involving an excess production of nitric oxide (NO) within the myenteric plexus, the ganglionated network that controls motility, 5,6 has been proposed from immunohistochemical studies to have a role in the pathology underlying slow transit constipation in the human large bowel. 7,8 However, how an excess of NO might alter transit is unknown. We have recently proposed that there are multiple sensory neurons in the myenteric plexus of the guinea pig large intestine. 9 There appears to be 2 intrinsic sensory systems underlying peristaltic type motor patterns that are analogous to tension-sensitive Golgi tendon organs and lengthsensitive muscle spindles in skeletal muscle. Activation of these systems by circumferential stretch applied to the colon generates neurally mediated peristaltic waves that are dependent on muscle tone (duration seconds; frequency, 0.3/min) 10 and ongoing peristaltic reflex activity (duration, 3 seconds; frequency, 11/min), which is independent of muscle tone but generated by increases in the diameter of the bowel These 2 motor patterns appear to be generated by myenteric after-hyperpolarization (AH)- type sensory neurons, which sense muscle tension, 14 and mechanosensory S-type ascending and descending interneurons that detect changes in muscle length around the circumference of the bowel. 11,13,15 Multipolar AH neurons, which were until recently thought to be the only sensory neuron in the gut, are defined for their long-lasting AH following an action potential. Unipolar synaptic (S) neu- Abbreviations used in this paper: CM, circular muscle; CMMP, circular muscle myenteric plexus; EJP, excitatory junction potential; FEPSP, excitatory postsynaptic potential; IJP, inhibitory junction potential; IPSP, inhibitory postsynaptic potential; LM, longitudinal muscle; L-NA, N- -nitro-l-arginine; PPADS, pyridoxal phosphate-6-azophenyl-2,=4=disulfonic acid; TS, transmural stimulation by the AGA Institute /07/$32.00 doi: /j.gastro

2 May 2007 ENTERIC OCCULT REFLEX AND LARGE BOWEL TRANSIT 1913 rons, on the other hand, receive fast excitatory synaptic input and comprise the mechanosensory interneurons and excitatory and inhibitory motor neurons. 11,15 Recently, low-threshold extrinsic afferent mechanoreceptors in the rectum have been shown not to exhibit directional sensitivity because they respond equally to both circular and longitudinal stretch. 16 Given this finding, we wanted to determine whether peristaltic activity in the colon is affected not only by circumferential stretch but also by longitudinal stretch. We also investigated how intrinsic release of NO might affect these reflexes. We demonstrate that a third myenteric sensory neural system, which is inhibitory, is activated by colonic elongation. These mechanosensory neurons release NO within the myenteric plexus resulting in a depression of activity of mechanosensitive interneurons driving peristaltic nerve pathways activated by circumferential stretch. Materials and Methods Intracellular Microelectrode Recordings From Smooth Muscle Male and female guinea pigs ( g) were killed in accordance with the Animal Ethics Committee at the University of Nevada-Reno, and a segment of distal colon (15 20 mm long) was removed. The mucosa and submucosa were removed to reveal the circular muscle (CM), and the segment was threaded through greased (High Vacuum Grease; Dow Corning, Midland, MI) rubber diaphragms in 2 partitions (7 mm apart; see Figure 1A). Each chamber was separately perfused with warmed oxygenated Krebs solution (36.0 C 0.5 C) containing nifedipine (1 2 mol/l) to paralyze the smooth muscle Simultaneous microelectrode recordings were made from CM cells at both the oral and anal ends of the Figure 1. Effects of circumferential and longitudinal stretch. (A) Circumferential stretch applied to the distal colon activated ongoing peristaltic reflex activity consisting of oral excitatory junction potentials (EJPs) that were coordinated with anal inhibitory junction potentials (IJPs) in the circular muscle. Transmural nerve stimulation (transmural stimulating single pulse 0.5 ms, 30 V) applied to the middle of the preparation evoked a robust oral EJP and anal IJP. (B) When the same preparation was stretched only in the longitudinal axis, only low-amplitude neural activity was observed and gave only a small response to TS. preparation. Preparations were stretched and pinned either in the circumferential (12 mm, twice their resting diameter) or longitudinal directions or in both directions simultaneously. Once control activity was recorded, the preparation was then stretched and repinned in both axes, which took approximately 2 3 minutes. After a 10-minute equilibration period, microelectrodes were reinserted into CM cells at either end of the tissue. Longitudinal stretch was normalized to the slack segment length. Simultaneous intracellular recordings were made from CM cells using 2 independently mounted micromanipulators (WPI; World Precision Instruments, Inc. Sarasota, FL; model M3301R). Microelectrodes (ID, 0.5 mm) were filled with 1.5 mol/l KCl solution and had tip resistances of approximately M. Electrical signals were amplified using a dual input Axoprobe 1A amplifier (Axon Instruments, Foster City, CA) and digitized at 2 5 khz for neuronal recordings (see below) and 500 Hz to 1 khz for smooth muscle recordings on a PC running Axoscope software (version 8.0; Axon Instruments). CM Myenteric Plexus Preparations CM myenteric plexus (CMMP) preparations were prepared by removing the longitudinal muscle (LM). The LM was removed by teasing up muscle fibers at the oral end, pinching these fibers with a pair of curved forceps and peeling the LM off the preparation, in an oral to anal direction. This easily removed the LM, which often came off as wide strips of muscle, without damaging the myenteric plexus that remained intact upon the CM. The integrity of the enteric circuitry was assessed by the fact that the CM generated spontaneous coordinated oral excitatory junction potentials (EJP) and anal inhibitory junction potentials (IJP). This would not happen if there was any damage to the myenteric plexus from removing the LM. Intracellular Microelectrode Recordings From Myenteric Neurons The mucosa and submucosa and the LM were removed from a segment of distal colon 20 mm in length to create a CMMP preparation. The segment, which was stretched circumferentially to twice its slack width, was pinned in an organ bath with the myenteric plexus uppermost. Retaining the CM ensures that neurons in peristaltic nerve pathways are spontaneously active. 13,15 Transmural stimulating wires were placed across the oral end of the preparation approximately 10 mm from the recording site. Once a neuron was impaled, transmural stimuli (TS) were applied (0.5-ms duration, 10 Hz for 5 to 10 pulses). Standard intracellular recording techniques were used to record from myenteric neurons at the anal end of the segment. 13,15 The organ bath was separately perfused with warmed oxygenated Krebs solution (36.0 C 0.5 C) containing nifedipine (1 2 mol/l) to

3 1914 DICKSON ET AL GASTROENTEROLOGY Vol. 132, No. 5 washed overnight in 1% PBS (ph 7.2); (3) blocked with bovine serum albumin (BSA; 2%); (4) incubated in antineuronal nitric oxide synthase (nnos; 1/500) raised in sheep (a generous gift from P. C. Emson, University of Cambridge, United Kingdom) with 0.5% Triton X-100 and PBS for 48 hours at 4 C; (5) washed with PBS for 2 hours; (6) incubated with Alexa Fluor 594 (donkey antisheep IgG secondary antibody; Molecular Probes) for 1 hour at 21 C; (7) washed in PBS overnight at 4 C; and (8) mounted on glass slides (Fisher Scientific) with Aquamount (VWR International, West Chester, PA). Figure 2. Effect of blocking nitric oxide production. (A) Control: ongoing oral EJPs and anal IJPs in another circumferentially stretched preparation. Black arrow indicates onset of TS (single pulse 0.5 ms, 30 V). (B) Ongoing peristaltic reflex activity evoked by circumferential stretch and oral EJPs and anal IJPs evoked and TS were almost abolished by 60% longitudinal stretch. (C) L-NA (100 mol/l) added to the middle chamber restored all ongoing and evoked activity. paralyze the smooth muscle 9 12 and guanethidine (0.5 mol/l) to block norepinephine release. 17 Calcium Imaging The LM was peeled from a segment of colon to create the CMMP preparation to expose the myenteric plexus. 18 The preparation was stretched in both axes and viewed under an upright microscope (Eclipse E600FN; Nikon Inc., Melville, NY) fitted with water immersion lenses ( 10X, 20, 40: Nikon Plan Fluor). 6,18 Fluo 4-AM (FluoroPure, AM, Molecular Probes, Eugene, OR) was loaded into neurons at room temperature as previously described. 18 The final loading concentration of Fluo-4 in the bath was 50 nmol/l. The tissues were then perfused with warm (36.0 C 0.5 C) Krebs solution and illuminated using a Lambda DG-5 light source (Sutter Instrument Company, Novato, CA). Appropriate filters produced excitation of Fluo-4 between 460 and 490 nm and collected emissions 515 nm. Image sequences were captured using a Cascade 512B digital camera (Roper Scientific Inc., Trenton, NJ) and MetaMorph 6.26 software (Universal Imaging Corporation; Molecular Devices Corporation, Downington, PA). Image sequences were analyzed using custom-written software (Volumetry G5; G.W.H.). 18 Immunohistochemistry Following calcium imaging experiments, whole mount tissues were then (1) fixed for 15 minutes in acetone ( 20 C) (Fisher Scientific, Pittsburgh, PA); (2) Figure 3. Correlation between EJPs and IJPs. (A) Correlation coefficient (R 2 ;n 6) for temporal coordination between ongoing oral EJPs and anal IJPs evoked by circumferential stretch (1) for control-circumferential stretch only; (2) following the addition of 60% longitudinal stretch, and (3) following removal of longitudinal stretch but not circumferential stretch. (B) Amplitude of oral EJPs plotted against the amplitude of anal IJPs (1) during circumferential stretch, (2) following the addition of 60% longitudinal stretch, and (3) after removing only longitudinal stretch. Note that the noise bubble shows that EJPs and IJPs 5 mv were uncoordinated. (C) Correlation coefficient (R 2 ) for temporal coordination between ongoing oral EJPs and anal IJPs evoked by circumferential stretch (1) for circumferential stretch only, (2) following 25% longitudinal stretch, (3) following the addition of 60% longitudinal stretch, and (4) following 60% longitudinal stretch and L-NA (100 mol/l) in middle chamber.

4 May 2007 ENTERIC OCCULT REFLEX AND LARGE BOWEL TRANSIT 1915 Effect of Inhibitory Reflex on Pellet Propulsion A segment of colon 4 to 5 cm in length was opened for 15 mm at the oral end. The preparation was pinned adjacent to where the free opened end of colon joined the intact tube. The tube was connected to a tension transducer under 0.5g of tension at the anal end to keep the intact segment straight. The loose oral flap of colon was connected via a rake system to the string of another pulley system attached to its oral end, to which weights could be applied. The rake, which ensured a more even longitudinal pull, consisted of a stainless steel triangle to which dissecting pins had been welded. 19 An epoxy-coated pellet was inserted into the oral end of the intact segment and allowed to propel spontaneously down the bowel. Once several control runs 3 minutes apart were obtained, a 1-g weight was applied to stretch the oral flap when the pellet started to move in a continuous manner. The diameter of the colon was used to create a spatiotemporal map that visualized the progression of the pellet over time (see below). 10,20,21 Colonic Filling A segment of distal colon (30 40 mm long) was mounted in an organ bath and secured by O rings (see Figure 4. Effects of transmural nerve stimulation and removal of longitudinal muscle. (A) Evoked oral EJP and (B) anal IJP amplitude plotted against different strengths of TS (single pulses 0.5 ms, 0 75 V; n 5) (1) in a circumferential stretched preparation (solid black line with diamond); (2) following the addition of 60% longitudinal stretch (dashed line with square); and (3) following the further addition of L-NA (100 mol/l) (dashed line with triangle) in the middle chamber. (C) Following removal of longitudinal muscle, circumferential stretch evoked normal ongoing coordinated oral EJP and anal IJP activity. (D) Despite removal of the longitudinal muscle, stretching the same segment of distal colon as used in Panel C to 60% beyond the slack length still depressed activity generated by circumferential strength. (E) Following the removal of longitudinal stretch, the activity evoked by circumferential stretch returned.

5 1916 DICKSON ET AL Figure 7A).10,20 The oral end was fixed and the anal end connected to a moveable outflow tube, under 2g of tension in an organ bath containing oxygenated Krebs solution at 36.0 C 0.5 C (see Figure 7A). A force transducer (TST125C; Biopac Systems Inc., Santa Barbara, CA) was glued (Vet bond) to the serosal surface at the oral end. Marker (eye glitter) arrays were glued along the wall of the colon to measure longitudinal movements and viewed with a video camera (WV-BP330; Panasonic CCTV) and recorded on a VHS tape recorder (VWM-390; Sanyo). Warmed Krebs solution was injected into the oral end with a syringe (Hamilton, Reno, NV; L). Changes in diameter and length of the intact segment GASTROENTEROLOGY Vol. 132, No. 5 were followed over time by converting video images into a silhouette and then plotting the diameter along the colon at each time point (33 ms) as a spatiotemporal diameter or D map.10,20,21 Drugs and Solutions Guanethidine was used to block postganglionic sympathetic nerve transmission. Nifedipine was used to paralyze the muscle. Stock solutions of hexamethonium bromide, ondansetron, and phosphate-6-ozophenyl-2=-4=-disulfonic acid (PPADS) were made in distilled water and used to block all known forms of fast synoptic transmission. A combination of these drugs Figure 5. Ca2 imaging of myenteric neurons following blockade of synaptic neurotransmission. A segment of colon, with the longitudinal muscle removed, was pinned in both axes. Hexamethonium (100 mol/l), ondansetron (1 mol/l), and PPADs (30 mol/l) were added to the Krebs solution to block fast synaptic transmission. (A) Average Ca2 fluorescence in myenteric neurons. (B) An overlay of nnos immunoreactivy (red) superimposed on averaged Ca2 fluorescence. (C) nnos immunoreactivy (red) overlaid with average Ca2 fluorescence in L-NA (100 mol/l). (D) Before the addition of L-NA, only 1 neuron in this ganglia exhibited ongoing activity (see arrowhead 1 in A C). This neuron was nnos ve and anally projecting (see arrowhead 1 in A and B). (E) L-NA (100 mol/l) did not effect the activity of the NOS ve neuron (black trace) but revealed activity in other neurons (boxes 2 5; C and E) that were NOS ve.

6 May 2007 ENTERIC OCCULT REFLEX AND LARGE BOWEL TRANSIT 1917 was diluted in Krebs solution at a final concentration of hexamethonium (100 m/l), ondansetron (1 mol/l), and PPADS (30 mol/l). Krebs solution (mmol/l): NaCl, ; KCl, 5.9; NaHCO 3, 15.5; NaH 2 PO 4, 1.2; MgSO 4, 1.2; CaCl 2, 2.5; and glucose, 11.5 (gassed with 3% CO 2-97% O 2 (vol/vol), ph ). Statistical Analysis Analysis of spontaneous junction potentials (sjps) was performed using custom-written routines on short samples of recordings of electrical activity ( 2 3 minutes in duration) that were resampled to 100 Hz. 22 A smoothing algorithm (100-ms average, 5 iterations) and an average baseline algorithm (10 seconds, 3 iterations) were applied to reduce high frequency noise and for calculation of amplitudes of spontaneous junction potentials, respectively (see Figure 3). The first recording was used to identify the peaks of spontaneous EJPs (sejps) and spontaneous IJPs (sijps) by locating inflexion points. The amplitude, calculated as the difference between the peak of the junction potential and the average baseline was calculated for each sjp. To determine whether sjps in the second recording were coordinated in time and had similar amplitudes to sjps in the first recording, 2 different search routines were used. The first routine used the time at the peak of an sjp in the first recording and looked for any inflexion (sijp or sejp) in the second recording that was closest in time to that point. Once the peak of an sjp was located in the second trace, the amplitude, maximum slope, and time to half-maximum amplitude of the second peak were calculated. Results from this analysis were used to compare the similarity of the amplitudes of sjps between the 2 recordings. Regression analysis was used to summarize the variability in the amplitude of sjps between the 2 recordings and was expressed as the correlation coefficient r. The second routine, instead of searching for any inflexion in the second recording, looked only for sjps that were similar to those in the first trace (eg, if there was an IJP in the first recording, the routine would only return the closest IJP in the second recording). Results from this analysis were used to compare differences in time between similar sjps and were expressed as standard deviations calculated at 3 different amplitude levels (1 5, 5 9, and 9 mv). All results are expressed as means SEM. One-way and 2-way factorial analysis of variance statistical tests were used to compare differences between groups. A probability of less than 0.05 was considered significant. Scheffé post hoc tests were used. N refers to the number of animals on which observations were made. Results Effects of Circumferential vs Longitudinal Stretch Alone We recently reported that maintained circumferential stretch of the guinea pig distal colon generated ongoing peristaltic reflex activity. This activity consists of an ongoing discharge of oral EJPs and anal IJPs, which are temporally synchronized, in the circular and longitudinal smooth muscle layers (Figure 1A) This stretch-evoked ongoing peristaltic reflex activity, which can be recorded for several hours, is stretch sensitive and independent of muscle tone because it occurs in the presence of smooth muscle relaxants such as nifedipine or papavarine (Figure 1B). 13 Transmural nerve stimuli applied to the middle of the preparation evoked robust polarized responses, consisting of an oral EJP and an anal IJP in the circular muscle (Figure 1A). We examined whether this ongoing peristaltic reflex activity in the distal colon is affected not only by circumferential stretch but also by longitudinal stretch. We found that maintained longitudinal stretch by itself when applied to the same preparations to which circumferential stretch was not applied failed to generate prominent neural activity in the circular muscle (n 6, Figure 1B). When transmural nerve stimuli were applied to the middle of a segment of colon that was stretched only in the longitudinally axis, only low-amplitude junction potentials were evoked, which were similar in amplitude to those in Figure 2B as described below. Figure 6. NO-mediated synaptic events. In the presence of guanethidine (0.5 mol/l) and nifedipine (1 mol/l), recordings were made from myenteric neurons in a circumferentially stretched CMMP preparation that were situated approximately 10 mm anal of transmural stimulating (TS) electrodes. (A) IPSP evoked in S neuron exhibiting spontaneous action potentials (0.5-ms duration, 5 pulses at 10 Hz, 30 V). (B) A similar stimulus evoked FEPSPs in a neuron with ongoing FEPSPs. (C) IPSP evoked in another neuron in the presence of quanethidine (0.5 mol/l). This neuron exhibited spontaneous action potentials, FEPSPs, and proximal process potentials. (D) Following the further addition of L-NA (100 mol/l), the IPSP in the same neuron was blocked.

7 1918 DICKSON ET AL GASTROENTEROLOGY Vol. 132, No. 5 A B Oral Anal Oral Anal Oral Stretch 5mm Effects of Longitudinal Stretch on Ongoing Peristaltic Reflex Activity Remarkably, longitudinal stretch (to 25% and 60% of the relaxed length of colon) reduced in a graded fashion the amplitude (Figure 2) and coordination of both oral EJPs and anal IJPs (n 6; Figure 3) generated by circumferential stretch. Longitudinal stretch had no effect on the resting membrane potential of circular muscle cells (circumferential stretch, mv; plus 60% longitudinal stretch mv; n 7; P.01). Most notably, the inhibitory effects of longitudinal stretch were reversible because activity returned to control levels following removal of longitudinal stretch (Figure 3). Furthermore, polarized responses activated by transmural nerve stimulation applied to the center of the preparation, which evokes an oral EJP and an anal IJP, were also reduced by longitudinal stretch (Figures 2 and 3A and B). Inhibitory Effects of Longitudinal Stretch Are Mediated by NO We were interested in the mechanism that could withdraw ongoing reflex activity from the smooth muscle. Previously, we had shown that mucosal stimulation of the guinea pig distal colon that was stretched in both axes activates myenteric interneurons in peristaltic reflex pathways that can be inhibited by exogenously applied and neurally released NO. 23 However, as we have also shown previously, 12,22 the NO synthesis inhibitor L-NA (100 mol/l, n 4) had no effect on the frequency, amplitude, and correlation of oral EJPs and anal IJPs when added to the middle chamber of a preparation 5s 5mm 5s Pre Oral Stretch Post 5mm Figure 7. Effects of colonic elongation on pellet propulsion. (A) As shown in the spatiotemporal map, an epoxy-coated pellet was allowed to propel spontaneously down the distal colon. Once propulsion was initiated, the pellet propelled at a constant velocity down the colon. (B) In another run, once the pellet started to move at a constant velocity, colonic elongation was applied to the oral flap. Elongation of the flap to approximately twice its resting length almost stopped pellet propulsion. Following removal of the stimulus, the flap gradually relaxed back to its resting length, and the pellet then continued to move down the colon at its prestretched velocity. stretched in only the circumferential axis (R 2 control, ; L-NA, ; P.01; n 4). Therefore, we were most surprised to find that L-NA (100 mol/l; n 6) applied to the middle chamber between the 2 recording sites completely reversed the inhibitory effects of longitudinal stretch on both the amplitude and the coordination of ongoing junction potentials activated by circumferential stretch (Figures 2C and 3C) and polarized responses evoked by transmural nerve stimulation (Figures 2C and 4A and B). This suggests that NO was being released by myenteric neurons to inhibit the enteric neural circuitry. Effect of Removing Longitudinal Muscle Because colonic elongation inhibits ongoing peristaltic activity, we were interested in determining whether the sensory receptors responsible for this inhibitory reflex were located in the LM. Removal of the LM by sharp dissection to create a CMMP preparation had no effect on the junction potentials activated by circumferential stretch, as we have recently reported (Figure 4C). 13 Remarkably, however, in these CMMP preparations, the coordinated oral EJPs and anal IJPs were still inhibited by longitudinal stretch to 60% of the relaxed length following repinning of the same preparation (Figure 4D; n 5). Removal of longitudinal stretch restored control activity (Figure 4E). This suggests that the mechanosensitive processes (dendrites) of neurons underlying this reflex are in either the myenteric plexus or in the underlying circular muscle.

8 May 2007 ENTERIC OCCULT REFLEX AND LARGE BOWEL TRANSIT 1919 Activity of Myenteric Neurons in Preparations Stretched in Both Axes Clearly, the myenteric neurons releasing NO to inhibit this ongoing reflex activity respond selectively to longitudinal stretch. We reasoned that such neurons are likely to be NOS ve descending interneurons. 24 To test this hypothesis, we spatially monitored the activity of many myenteric neurons simultaneously using Ca 2 imaging 18 in preparations devoid of LM. These preparations were stretched in both axes so that the excitability of myenteric motor neurons and interneurons in peristaltic pathways was inhibited. In these experiments, hexamethonium (100 mol/l; nicotinic receptor antagonist), ondansetron (1 mol/l; 5-HT 3 receptor antagonist), and PPADs (30 mol/l; P 2X receptor antagonist) were added to block all known forms of fast synaptic transmission in the myenteric plexus. 25 Under these circumstances, any active neurons are likely to be mechanosensory and activated by longitudinal stretch. We found that only 1 to 3 neurons per ganglia (mean, , n 6) exhibited ongoing firing of Ca 2 transients ( Hz), suggesting that they are likely to be mechanosensory (see Figure 6D). Later immunohistochemistry of the preparations revealed that these active neurons were nnos ve and anally projecting, suggesting that they are likely to be interneurons (Figure 5B). L-NA (100 mol/l) had no effect on the frequency (L-NA, Hz; P.01) of Ca 2 transients in these nnos ve neurons (Figure 5C and E, black traces). In addition, in L-NA (100 mol/l), 2 populations of NOS ve neurons also became active (Figure 5C and E). In one group ( neurons per ganglia), Ca 2 transients occurred at a frequency of Hz, whereas, the other group of teardropshaped unipolar neurons ( neurons per ganglia) elicited a Ca 2 wave (duration, seconds; 22 neurons from 6 ganglia), which was initiated at the apical end of the neuron, in any given 30-second period. Therefore, these NOS ve neurons are probably mechanosensitive interneurons in peristaltic pathways 13,15 because fast synaptic transmission was blocked (Figure 5E). NO-Mediated Synaptic Events We attempted to determine the synaptic events underlying NO-mediated inhibition of reflex activity by longitudinal stretch. In circumferentially stretched CMMP preparations from which the LM had been removed, we made intracellular electrical recordings from 5 myenteric neurons that fired spontaneous action potentials and from 5 myenteric neurons that exhibited ongoing spontaneous bursts of fast excitatory postsynaptic potentials (FEPSPs) (Figure 6A and B). The neurons exhibiting spontaneous action potentials are likely mechanosensory interneurons driving ongoing peristaltic reflex activity, whereas the neurons showing bursts of FEPSPs are likely to be muscle motor neurons. 15 Repetitive transmural nerve stimulation (duration, 0.5 seconds; 10 Hz for 5 10 pulses) applied 10 mm oral of the recording site evoked inhibitory postsynaptic potentials (IPSPs; duration, seconds; amplitude, mv) in the 5 neurons exhibiting spontaneous action potentials (Figure 6A) and bursts of FEPSPs in motor neurons showing ongoing bursts of FEPSPs (Figure 6B). 15 IPSPs, although rare in myenteric neurons, can be readily evoked in submucosal neurons and are largely mediated by the release of norepinephrine from extrinsic nerve fibers acting on 2 receptors on the soma of submucosal neurons. 17 Therefore, our experiments were carried out in the presence of guanethindine (1 mol/l) to block norepinephrine release from extrinsic nerve terminals. L-NA (100 mol/l) completely blocked IPSPs in 3 neurons exhibiting spontaneous action potentials after a period of 15 to 20 minutes (n 3) (Figure 6C and D). Inhibition of Transit of a Pellet We reasoned that, if descending interneurons activated by elongation were involved in inhibiting transit of a pellet, then elongating the oral flap of colon should inhibit pellet propulsion. In control segments, epoxycoated pellets, after a brief delay, traveled smoothly down the entire length of the intact colon at a velocity of mm/s (n 4; Figure 7A). When the pellet had assumed a constant velocity at a distance of mm (n 4) from the entrance of the intact segment, we found that elongating the oral flap to twice its resting length significantly reduced the velocity of pellet propulsion down the colon to mm/s, which was significantly different from control (P.01, n 4; Figure 7B). Pellet propulsion was inhibited throughout the duration of the stimulus ( seconds), moving only mm (n 4) down the colon. Following release of oral longitudinal stretch, the oral flap gradually relaxed back to its resting length after a period of seconds, and then the pellet resumed its propulsion down the colon at mm/s (n 4). L-NA (100 mol/l, n 4) applied to the bath blocked pellet propulsion in 3 experiments, as previously described. 26 In the presence of LNA, longitudinal stretch of the oral flap had no effect on pellet propulsion. In 1 preparation, however, the application of L-NA (100 mol/l) had no effect on pellet propulsion down the colon (velocity in L-NA, 1.7 mm/s). However, in the presence of L-NA, oral stretch for 8.7 seconds also had no detectable effect on pellet propulsion in this preparation, suggesting that the descending inhibitory reflex was blocked. Effects of Colonic Filling We further investigated whether longitudinal stretch might have physiologic significance by examining the effects of filling an intact isolated segment of distal colon with fluid (Figure 8A). The colon distended circumferentially in response to increasing volumes of injected

9 1920 DICKSON ET AL GASTROENTEROLOGY Vol. 132, No. 5 fluid and generated an increase in contractile activity, including peristaltic waves (Figure 8B and C). Surprisingly, when the colon reached a near maximal degree of circumferential stretch, further intraluminal injections of fluid resulted in colonic elongation. This elongation was associated with a dramatic suppression of motor activity of both the CM and the LM (Figure 8B and C). We found that the maximum inhibition of motility occurred when the colon increased its relaxed length by 18.7% 1.6% (n 5). When fluid was withdrawn from the colon, the preparation slowly relaxed back to its original length (Figure 8C). These impeding effects on motility were also reversed by the addition of L-NA (100 mol/l), which increased contractility (Figure 8B). Discussion In this study, we initially investigated the effects of longitudinal stretch on ongoing peristaltic reflex activity in the guinea pig distal colon, which we have previously shown to be independent of muscle tone and activated by circumferential stretch. Ongoing peristaltic reflex activity consists of ongoing oral EJPs and anal IJPs that is driven by mechanosensitive ascending and descending interneurons. This class of intrinsic sensory neuron detects changes in circumferential length rather than changes in muscle tension A dendrite of these mechanosensitive interneurons enters and runs parallel to the circular muscle, suggesting that it Figure 8. Colonic filling leads to elongation. (A) Organ bath for filling a segment of colon with fluid and video imaging its movements. A tension transducer was attached to the circular muscle at the oral end of the colon. (B) Graph showing diameter, length, and tension changes in the circular muscle during increases in intraluminal volume (n 5). L-NA (100 mol/l) increased tone during colonic elongation. (C) An example of colonic motor activity during filling expressed as a spatiotemporal map. Once the colon reached its near maximal level of circumferential distension, it elongated with a consequent drop in contractile activity in both the longitudinal and the circular muscle. Withdrawal of injected fluid increased contractions of colon, which gradually assumed its resting length.

10 May 2007 ENTERIC OCCULT REFLEX AND LARGE BOWEL TRANSIT 1921 Figure 9. Proposed neural circuit underlying colonic elongation. (A) Schematic shows that circumferential stretch activates mechanosensitive ascending excitatory (AEPs) and descending inhibitory nerve pathways (DIPs) underlying peristalsis. AEPs and DIPs, which reinforce one another, activate excitatory motor neurons (EMNs) orally and inhibitory motor neurons (IMNs) anally to both the longitudinal and circular muscle layers. Activation of EMNs produces an excitatory junction potential (EJP), and activation of IMNs produces an inhibitory junction potential (IJP) in the circular muscle, at the oral and anal ends of the segment, respectively. During circumferential stretch, descending NOS neurons are quiescent. (B) Following colonic elongation, descending NOS ve interneurons are activated to release NO, which likely generates IP- SPs in mechanosensitive interneurons in AEPs and DIPs and a consequent withdrawal of activity from both EMNs and IMNs. responds to circumferential stretch. 13,15 Ascending interneurons trigger excitatory motor neurons to both the CM and LM, whereas descending interneurons activate inhibitory motor neurons to the CM and LM at the same time. Longitudinal Stretch and NO We were most surprised to find that longitudinal stretch decreased both the amplitude and the coordination between oral EJPs and anal IJPs, activated by circumferential stretch and transmural nerve stimulation, in a graded manner at the same time. These effects of longitudinal stretch were completely reversed by the NO synthesis inhibitor L-NA when applied to the middle chamber between the 2 recording sites. This suggests that NO is being released by myenteric neurons, which respond selectively to longitudinal stretch, to inhibit synaptic neurotransmission within the myenteric plexus (Figure 9). This also suggests that the mechanosensitive interneurons driving this reflex activated by circumferential stretch are being inhibited by NO because they are the only neurons within the middle chamber that are likely to drive the short excitatory and inhibitory motor neurons at the oral and anal ends of the preparation, respectively. 13,15 In the large intestine, motor neurons to the circular muscle have short projections ( 1 mm), whereas interneurons have projection lengths of 7 mm along the bowel. 27,28 Interneurons in the Colon We hypothesized that myenteric, NOS ve, descending interneurons are likely to be mechanosensory and selectively respond to longitudinal stretch by releasing NO. Furthermore, the neurons responding to longitudinal stretch cannot be AH sensory neurons because they are not NOS ve. In the guinea pig distal colon, immunohistochemical coding studies 24 demonstrate that there are 3 classes of ascending interneurons and 3 classes of descending interneurons that contain acetylcholine (ACh), suggesting that they may be involved in peristaltic reflex pathways. A fourth descending interneuron is immunoreactive only for NOS, whereas one of the above descending interneurons contains in addition to ACh, vasoactive intestinal polypeptide (VIP)/gastrin-releasing peptide (GRP)/ NOS/ Calbindin. 24 The NOS ve neuron responding to colonic elongation is likely to be the class containing only NOS because ACh, calcitonin gene-related peptide, and VIP excite myenteric neurons. 25,29 Similar complex classes of ascending and descending interneurons have been observed in the human colon. 30 Calcium Imaging of Activity in Myenteric Neurons Because this inhibitory reflex triggered by longitudinal stretch is still present even when the LM has been

11 1922 DICKSON ET AL GASTROENTEROLOGY Vol. 132, No. 5 removed suggests that the mechanosensory dendrites of these NOS-positive neurons must either be in myenteric ganglia or in the underlying circular muscle. We used Ca 2 imaging of myenteric neurons in preparations that were stretched in both axes so that peristaltic pathways activated by circumferential stretch should be inhibited by longitudinal stretch. Our rationale was that in these preparations, the intrinsic sensory neurons that respond to longitudinal stretch should still be active. In these preparations we used a combination of drugs to block all known forms of fast synaptic transmission. Under these circumstances, only 1 or 2 neurons per ganglia exhibited ongoing activity, suggesting that they are mechanosensory in response to longitudinal stretch. These were found to be nnos ve and anally projecting, suggesting that they were likely to be descending interneurons. The number of ganglia in 1-cm length of colon is approximately 1600 (40 40) ganglia, 13 suggesting that the number of these mechanosensory nnos ve interneurons ( 2 per ganglia) is likely be approximately 3200 per cm length of guinea pig distal colon. L-NA had no effects on the frequency of Ca 2 transients in these nnos ve neurons, which is not surprising because myenteric neurons that release NO are not themselves targets for NO. 31 NO-Mediated Synaptic Potentials Clearly, NO released during longitudinal stretch inhibits synaptic transmission in peristaltic nerve pathways. Previous studies in the guinea pig ileum showed that sodium nitroprusside (NaNO; NO donor) had no effect on evoked FEPSPs, although it did produce an occasional decrease in input resistance and hyperpolarization in some myenteric neurons. 32 IPSPs can be evoked readily in submucous neurons, 17 but they are rare in myenteric neurons. 33 IPSPs have, however, been observed in some AH neurons and in some unidentified S neurons in the guinea pig distal colon. 33 In this study, we show for the first time that IPSPs can be generated in spontaneously firing mechanosensitive S interneurons that underlie peristaltic nerve pathways but not apparently in motor neurons. 15 Presumably, the IPSPs were generated by the descending nnos ve interneurons because our stimulus was applied oral to the ganglia from which recordings were made and the IPSPs were blocked by L-NA, suggesting that they are mediated by NO. IPSPs would be expected to reduce the probability of stretch-evoked action potentials in mechanosensory interneurons underlying the muscle tone independent of ongoing peristaltic reflex activity. Interestingly, in the reciprocal tendon reflex of the central nervous system, the motor neurons to the extensor muscle are excited by FEPSPs while, at the same time, motor neurons to flexor muscle are inhibited. This inhibition is produced by interneurons that release -aminobutyric acid, rather than NO, which generates IPSPs to reduce action potential firing in motor neurons. 34 Pellet Propulsion We showed that applying longitudinal stretch on the oral segment of colon inhibited pellet propulsion for at least as long as the stimulus was maintained. This suggests that colonic elongation activates a descending inhibitory reflex that can dramatically reduce transit. This reflex is an acute reflex because pellet propulsion was inhibited immediately following application of the stimulus. This descending reflex can be directly correlated with our electrophysiological findings. Previously, it has been shown that the velocity of a single pellet along the guinea pig distal colon is reduced in a dose-dependent manner by inhibiting NO production, suggesting that these effects are caused by blocking descending inhibition in front of a pellet. 35 Interestingly, it has also been demonstrated that the presence of multiple pellets in the guinea pig distal colon leads to a dramatic slowing in the velocity and rate of emptying of pellets. 21 In light of our current findings, this is not surprising because Gabella 36 found that the colon was elongated over each pellet by approximately 29% compared with the region between pellets. In addition to the possible activation of descending inhibitory pathways by the preceding pellet, 21 we propose that the elongation of the colon by pellets also activates mechanosensitive descending (NOS ve) interneurons that release NO to inhibit enteric neurons in peristaltic circuits, thus slowing transit and promoting storage (Figure 9). Colonic Filling When a segment of distal colon was filled with fluid, it first distended circumferentially and generated peristaltic waves, and then, as it reached its near maximum level of circumferential stretch, it began to elongate. This elongation was associated with a dramatic suppression of motility that was shown to involve NO. This elongation is likely a natural consequence of the cross-ply lattice arrangement of collagen fibers in the submucosal layer that constrains the gut to distend in either the circumferential or longitudinal axis. 36 This collagen matrix, which changes with the movements of the intestinal wall, appears to protect the bowel from overstretching in both the circumferential and longitudinal axes. It should be noted that accommodation in the isolated small intestine is considered to be the phase of fluid accumulation (filling phase), which is characterized by radial distension accompanied by shortening of the bowel, before the threshold for peristalsis is reached. 37 In contrast, our studies in the colon suggests that accommodation is associated with elongation of the bowel. In summary, colonic filling by fluid leads to elongation of the colon that is associated with a suppression of motility. This is also supported by the observation that colonic elongation applied oral to a pellet inhibits pellet propulsion. Therefore, our data are consistent with the hypothesis that colonic elongation activates mechano-

12 May 2007 ENTERIC OCCULT REFLEX AND LARGE BOWEL TRANSIT 1923 sensitive NOS ve descending interneurons that respond to longitudinal stretch by releasing NO. The release of NO by these neurons is likely to generate IPSPs that inhibit action potential firing in mechanosensitive interneurons and possibly other neurons underlying peristaltic pathways (Figure 9). We refer to this elongation reflex as an intrinsic occult reflex (hidden from observation) because it withdraws all activity from the muscle. Our evidence suggests that this intrinsic neural circuit is critical for slow transit and accommodation of colonic contents. The occult reflex is probably deactivated following feeding because this causes extrinsic vagal excitation of intrinsic cholinergic motor neurons (gastrocolonic reflex) and an increase in motility. 1 4 This excitation would be expected to produce contraction of the longitudinal muscle leading to a reduction in colonic elongation. Clinically, the intrinsic occult reflex provides a negative feedback system that would be expected to exacerbate the inhibition of motility associated with constipation in an impacted colon. In support of this concept, an excess production of NO within the myenteric plexus has been recently proposed to underlie slow transit constipation in the human large bowel. 7,8 References 1. Christensen J. The motility of the colon. In: Johnson LR, ed. Physiology of the gastrointestinal tract, 4th ed. New York: Raven, 1994;24: Smith TK, Sanders KM. Motility of the large intestine. In: Yamada T, Alpers DH, Owyang C, Powel DW, Silverstein FE, eds. Textbook of gastroenterology, vol 1. ch 10. Hagerstown, MD: Lippincott, Williams and Wilkins, 1995; Camilleri M, Ford MJ. Colonic sensorimotor physiology in health, and its alteration in constipation and diarrhoeal disorders. Aliment Pharmacol Ther 1998;12: Phillips SF. Functions of the large bowel: an overview. Scand J Gastroenterol Suppl 1984;93: Sanders K, Smith TK. Neural regulation of colonic motor function. Colonic Dis 2003;3: Stevens RJ, Publicover NG, Smith TK. Induction and regulation of Ca 2 waves by enteric neural reflexes. Nature 1999;399: Cortesini C, Cianchi F, Infantino A, Lise M. Nitric oxide synthase and VIP distribution in enteric nervous system in idiopathic chronic constipation. Dig Dis Sci 1995;40: Tomita R, Fujisaki S, Ikeda T, Fukuzawa M. Role of nitric oxide in the colon of patients with slow-transit constipation. Dis Colon Rectum 2002;45: Smith TK, Hennig GW, Spencer NJ. Sensory transduction in the ENS. Physiol News 2005;58: Smith TK, Oliver GR, Hennig GW, O Shea DM, Vanden Berghe P, Kang SH, Spencer NJ. A smooth muscle tone-dependent stretchactivated migrating motor pattern in isolated guinea-pig distal colon. J Physiol 2003;551: Spencer NJ, Hennig GW, Smith TK. A rhythmic motor pattern activated by circumferential stretch in guinea-pig distal colon. J Physiol 2002;545: Spencer NJ, Hennig GW, Smith TK. Stretch activated neuronal pathways in longitudinal and circular muscle of guinea-pig distal colon. Am J Physiol 2003;284:G231 G Spencer NJ, Dickson EJ, Hennig GW, Smith TK. Sensory elements within the circular muscle are essential for mechanotransduction of ongoing peristaltic reflex activity in guinea-pig distal colon. J Physiol 2006;576: Kunze WA, Furness JB, Bertrand PP, Bornstein JC. Intracellular recording from myenteric neurons of the guinea-pig ileum that respond to stretch. J Physiol 1998;506: Spencer NJ, Smith TK. Mechanosensory S-neurons rather than AH neurons appear to generate a rhythmic motor pattern in guinea-pig distal colon. J Physiol 2004;558: Lynn P, Zagorodnyuk V, Hennig GW, Costa M, Brookes S. Mechanical activation of rectal intraganglionic laminar endings in the guinea pig distal gut. J Physiol 2005;564: North RA, Surprenant A. Inhibitory synaptic potentials resulting from -adrenoceptor activation in guinea-pig submucous plexus neurones. J Physiol 1985;358: Park KJ, Hennig GW, Lee HT, Spencer NJ, Ward SM, Smith TK, Sanders KM. Spatial and temporal mapping of pacemaker activity in interstitial cells of Cajal in mouse ileum in situ. Am J Physiol Cell Physiol 2006;290:C1411 C Brookes SJH, Chen BN, Costa M, Humpheys CM. Initiation of peristalsis by circumferential stretch of flat sheets of guinea-pig ileum. J Physiol 1999;516: Hennig GW, Costa M, Chen BN, Brookes SJ. Quantitative analysis of peristalsis in the guinea-pig small intestine using spatio-temporal maps. J Physiol 1999;517: Hennig GW, Costa M, Humphreys CM, Brookes SJ. Analysis of motor patterns in the isolated guinea-pig large intestine by spatiotemporal maps. Neurogastroenterol Motil 2001;13: Spencer NJ, Hennig GW, Smith TK. Spatial and temporal coordination of junction potentials in circular muscle of guinea-pig distal colon. J Physiol 2001;535: Smith TK, McCarron S. Nitric oxide modulates cholinergic reflex pathways to the longitudinal and circular muscle in the isolated guinea-pig distal colon. J Physiol 1998;512: Lomax AE, Furness JB. Neurochemical classification of enteric neurons in the guinea-pig distal colon. Cell Tissue Res 2000; 302: Galligan JJ. Pharmacology of synaptic transmission in the enteric nervous system. Curr Opin Pharmacol 2002;2: Foxx-Orenstein AE, Grider JR. Regulation of colonic propulsion by enteric excitatory and inhibitory neurotransmitters. Am J Physiol 1996;271:G433 G Neunlist M, Michel K, Aubé A-C, Galmiche J, Schemann M. Projections of excitatory and inhibitory motor neurones to the circular and longitudinal muscle of the guinea pig colon. Cell Tissue Res 2001;305: Spencer NJ, Hennig GW, Dickson E, Smith TK. Synchronization of enteric neuronal firing during the murine colonic MMC. J Physiol 2005;564: Wood JD. Physiology of the enteric nervous system. In: Johnson LR, ed. Physiology of the gastrointestinal tract. 3rd ed. New York: Raven, 1994: Wattchow DA, Porter AJ, Costa M. The polarity of neurochemically defined myenteric neurons in the human colon. Gastroenterology 1997;113: Shuttleworth CW, Xue C, Ward SM, de Vente J, Sanders KM. Immunohistochemical localization of 3,5 -cyclic guanosine monophosphate in the canine proximal colon: responses to nitric oxide and electrical stimulation of enteric inhibitory neurons. Neurosci 1993;56: Tamura K, Schemann M, Wood JD. Actions of nitric oxide-generating sodium nitroprusside in myenteric plexus of guinea pig small intestine. Am J Physiol 1993;265:G887 G Lomax AE, Sharkey KA, Bertrand PP, Low AM, Bornstein JC, Furness JB. Correlation of morphology, electrophysiology and