Smooth Muscle Mechanosensitivity Generates and Maintains Pressure Gradients Across the Intestine

Richard J. Amedzrovi Agbesi performed experiments, analyzed data and revised the paper. Lucas Chassatte performed experiments and analyzed data. Nicolas R. Chevalier designed and supervised the research, performed experiments, analyzed data, wrote and revised the paper, and obtained funding.

We first characterized the mechano‐sensitive properties of the smooth muscle (Figure  1a ) in gut ring preparations, using a thin force‐calibrated glass cantilever to measure the contractile force and apply controlled stretch to the preparation (Figure  1b , Section  2 ). Smooth muscle rings presented cyclic spontaneous contractions (Video  S1 ). Stretching the ring elicited a contraction within a few seconds after stress application (Figure  1c , Videos  S2 and S3 ). The amplitude of the cyclic contractions increased with stretch of the preparation (Figure  1d ), while the contraction frequency slightly decreased (from 2.5 to 2 cycles‐per‐minute = cpm, Figure  S1a ). Contraction duration (width at half maxima) was 8 ± 2 s and did not significantly change with stretch (Figure  S1b ). We sought to determine the ion channels involved in mechanosensitivity. Nifedipine, an L‐type channel blocker, reduced the amplitude of the immediate contractile response to stretch by 76% ± 14% ( n  = 7, Figure  1e , Video  S2 ); 2‐APB, an inhibitor of the IP3R receptor at the endoplasmic reticulum involved in Ca 2+ ‐induced Ca 2+ cytosolic release, reduced the contractile response by 63% ± 35% ( n  = 10) at 100 μM and completely abolished it at 500 μM (Figure  1e , Video  S3 ). We further tested GdCl 3 , an inhibitor of cationic mechanosensitive TRP‐ and Piezo‐channels [ 30 ], lidocaine, an inhibitor of mechanosensitive sodium Nav1.5 channels [ 30 ] and ryanodine, an inhibitor of the potentially mechanosensitive [ 31 ] ryanodine receptors at the endoplasmic reticulum; none of these significantly changed the stretch‐induced contractile response (Figure  1e ). These results indicate that L‐type Ca 2+ channels mediate most (~80%) of the mechanosensitive response of the embryonic chicken smooth muscle, and that further amplification of intracellular Ca 2+ by IP3R receptors is essential. These results are consistent with previous findings in human jejunal smooth muscle [ 8 ]. The ~20% remaining contractile response after nifedipine application appears to result from phenomena independent of Ca 2+ shuttling at the membrane, perhaps due to stretch‐induced disruption of the Ca 2+ reservoirs at the endoplasmic reticulum; we could indeed elicit a stretch‐induced contraction, albeit weak, in Ca 2+ ‐deprived medium (Figure  S2 ). The stretch‐contractile force response curve (Figure  1d ) suggests that a myogenic smooth muscle tube contracts proportionally to the local pressure, and should therefore exhibit a differential behavior in a pressure gradient. To test this idea, we measured the pressures at the two extremities of ~2 cm long cannulated gut segments (Figure  2a , Section  2 ) while continuously recording the peristaltic waves propagating in the organ; syringes connected at the rostral and caudal ends allowed to impose the initial pressure applied rostrally (=oral, left) and caudally (=anal, right). We first confirmed that pressure levels equalized at the left and right end when a non‐contractile gut segment was cannulated (Figure  S3 )—this was achieved by working at room‐temperature in PBS without Ca 2+ and Mg 2+ . In PBS with Ca 2+ (0.9 mM) and Mg 2+ (0.5 mM) at room‐temperature, we found that no spontaneous contractility was present, but that contractile waves could be elicited by raising the pressure to ~2 cmH 2 O (Figure  S4 ). Finally, working in carbogen (95% O 2 –5% CO 2 ) saturated, glucose‐containing DMEM at 37°C (our standard condition), we measured a distinct increase of the contraction wave amplitude with increasing average luminal pressure (Figure  2b ). Contraction wave amplitude was significantly higher than when no pressure was applied to the preparation (Figure  2b ). These results parallel the mechanosensitive behavior measured in Figure  1 : the circumferential force F generated by P  = 1–5 cmH 2 O in a pressurized cylindrical vessel of radius r  = 0.4 mm and length l  = 2 mm (ring length) is F  =  Prl  = 80–450 μN, that is, comparable with the stretch force applied to elicit (Figure  1c,e ) or amplify (Figure  1d ) contractions in the ring preparation. E14 duodenal contractions are purely myogenic [ 27 ] and did not exhibit a preferred direction of propagation: rostro‐caudal (RC) and caudo‐rostral (CR) peristaltic waves were most often present simultaneously, their respective prevalence varying with time (Figure  2c , Video  S4 ). In contrast, waves in E16 duodenum segments propagated rostro‐caudally predominantly or completely (Figure  2d , Video  S4 ). We quantified this by computing the time‐and space‐averaged wave directionality index (see Section  2 : 0 pure CR waves, 1 pure RC waves, 0.5 equal mix of CR & RC waves): this index for initially equal rostral and caudal pressures was respectively 0.53 ± 0.12 ( n  = 15) and 0.84 ± 0.05 ( n  = 11) at E14 and E16. Motility at E16 features myogenic and neurogenic components [ 27 ]. Application of the neural action potential blocker tetrodotoxin (TTX) inhibited the neurogenic migrating motor complex (phases of low circular contractile activity in Figure  2d ), as we had previously shown [ 27 ], but did not however change the rostro‐caudal directionality of myogenic waves (Figure  2d , n  = 5/5, Video  S4 ). This result indicates that the rostro‐caudal directionality of waves at E16 is a myogenic phenomenon. We had previously shown that the onset of pacemaker interstitial cell of Cajal activity in the developing chicken gut gives rise to rhythmic waves, and tends to increase the propagation distance of waves (i.e., less inter‐wave collisions), and the amount of time during which they propagate in the same direction [ 32 ]. It is therefore reasonable to assume that interstitial cells of Cajal, being the only myogenic component involved in controlling directionality, are responsible for the shift to RC waves in the E16 duodenum. We surmise that higher ICC activity in rostral parts of the E16 duodenum initiate contractile waves at this end, which then travel rostro‐caudally. When initial equal pressures (Δ P i  = 0) were applied on the rostral and caudal ends, gut contractile activity spontaneously generated a pressure differential, which equilibrated to a constant value Δ P f after 30–60 min (Figure  3a ). In E16 guts presenting directional waves, pressure always ( n  = 11/11) increased at the caudal end and decreased at the rostral end (Δ P f  = 0.9 ± 0.5 cmH 2 O, Figure  3b ), consistent with the expected flow induced by RC waves (Figure  3b , orange dots). Surprisingly, we found that the omnidirectional E14 gut could also act as a pressure generator: pressure rose on the rostral side in n  = 8/15 samples (Δ P f  = − 1.5 ± 0.6 cmH 2 O, Video  S5 ); in n  = 7/15, the pressure increased at the caudal end (Δ P f  = 0.9 ± 0.5 cmH 2 O). The time‐averaged directionality index of waves at E14 was, however, not correlated to the direction of liquid flow Δ P f (Figure  3b , blue dots, Video  S5 ). To understand this, we injected a fluorescent dye tracer at E14 (Section  2 , Video  S6 ). Direction of dye movement was very difficult to predict based on the contractility pattern: annihilating waves could propel dye at high speed in one direction or the other [ 18 ], and dye was often observed to be transported in the opposite direction of the most adjacent contractile wave because of higher magnitude relaxation‐contraction movements occurring at more distant sites; similar observations have been made previously at stage E12 [ 33 ]. This erratic transport induced by multi‐directional waves rationalizes why we could not find a correlation between the wave direction index and average liquid flow in the multidirectional E14 samples. We further found that maximal Δ P f were reached in the range of initially applied pressure 4–6 cmH 2 O (Figure  3c ), decreasing both at lower and higher pressure. This optimal range of pressure for driving peristaltic contractions is consistent with previous findings in the rat [ 34 ]. We next examined the flow and pressure behavior when an initial pressure gradient was imposed (Δ P i  > 0 or Δ P i  < 0). Guts exhibited higher contractile wave amplitude at the high pressure end (Figure  3d ); this was conspicuous in longer gut segments (Figure  S5 ). At E14, the initially applied pressure gradient Δ P i was dampened over time to lower values, but its sign was maintained throughout the experiment (Figure  3e,f ); Δ P f was proportional to Δ P i (Figure  3f ). This shows that the random caudal or rostral pressure increase observed for Δ P i  = 0 could be coaxed by imposing Δ P i . E16 segments exhibited a similar behavior to E14, but initially applied rostral pressure (Δ P i   0) in n  = 5/11 samples (Figure  3f ). The assymetric shape of the Δ P f versus Δ P i characteristic at E16 indicates a competition between the effects of pressure on contractility, and the intrinsic RC directionality of waves at this stage. Interestingly, the RC wave direction at E16 could be reversed within ~1 h when a high Δ P i (6–8 cmH 2 O) was applied, leading to a subsequent sharp decrease of the caudal pressure (Figure  3f , n  = 3, red arrows). Distension of the gut wall induced the nucleation of waves at the high pressure caudal end, which annihilated with the physiological RC waves, first close to the caudal nucleation point, but with time at more and more rostral positions. After 30 min–1 h, CR waves became dominant: wave direction had reversed (Figure  3g , Video  S7 ). We finally measured how flow and pressure development depended on gut segment length and intrinsic contractility, working on E16 guts with initial equal pressure P ir  =  P ic  ~ 3 cmH 2 O. We found that the flow rate in the medium and long segments was significantly higher than in small segments (Figure  4a , Section  2 ). The flow‐rate tended however to level off at ~0.15 mm 3 /min, we could not find a significant difference between medium‐length and long segments, although the average number of contractile waves, i.e., the “power of the peristaltic pump” was proportional to the length of the gut segments (Figure  S6 ). We found this to be caused by the fact that the long E16 gut segments, i.e., that included a substantial portion of jejunum, presented multidirectional waves in the caudal regions (observed in n  = 6/6 long samples), that presumably mitigated the total flow rate. We further found that the generated Δ P f did not depend on gut length (Figure  4b ). Tetrodotoxin allowed us to modulate the contractile force generated by the smooth muscle at E16: because it blocks inhibitory, nitric‐oxide mediated neural relaxation of the smooth muscle [ 27 ], its application results in an increased smooth muscle contractile force and tone. This increased force drove a significant increase of Δ P f (Figure  4c , Video  S8 ).

In summary, we studied the mechanosensitive and pressure‐flow characteristics of two smooth muscle peristaltic tubes, one exhibiting randomly directed waves (E14), and the other featuring directional rostro‐caudal waves (E16). Wave directionality did not stem from neural activity at E16, but from higher rostral pacemaker interstitial cell of Cajal activity (Figure  2d ). Mechanosensitivity was characterized by increasing smooth muscle contractile force in response to stretch (ring experiment Figure  1 ) and pressure (cannulation experiment Figure  2b ), that was mediated by L‐type Ca 2+ channels at the membrane and required amplification of the Ca 2+ response by IP3 receptors at the endoplasmic reticulum (Figure  1e ). This mechanosensitive response is common to smooth muscle of the uterus [ 35 ], ureter [ 36 ] and gut [ 8 ]. Mechanosensitivity results in the higher pressure end of the gut developing higher tonic and/or phasic contractility (Figures  1 , 2b and 3d , Figure  S5 ): this induces a positive feedback mechanism that drives further increase of pressure at the same end (Figure  5a ). We suggest that this mechanism drives the spontaneous breaking of symmetry in E14 guts that do not exhibit preferred wave directionality (i.e., development of Δ P f  ≠ 0 from initial Δ P i  = 0, Figure  3b blue dots); the sign of the developed pressure Δ P f can be nudged by applying an initial pressure differential (Δ P i  ≠ 0, Figure  3e,f ). At E16, this mechanosensitive feedback mechanism competes (for P i  0, Figure  3f , orange dots) the natural tendency of RC waves to drive fluid flow in the rostro‐caudal direction (Figure  5a,b ). Importantly, high caudal pressure could reverse the RC direction of waves at E16 (Figure  3g , Video  S7 ) by generating a new caudal nucleation site that competes with the intrinsic directionality imparted by ICCs. The flow rate increased with tube length (Figures  4a and 5c ). This is consistent with the fact that longer gut segments present a higher total number of peristaltic waves, and that the flow induced by multiple co‐linear peristaltic waves sums up [ 18 ]. The maximum pressure developed did not depend on tube length (Figure  4b ), but on the maximum smooth muscle contractile force (the “leak‐tightness” of individual constrictions, Figure  5d ). This force can be modulated both by the resting pressure (Figure  3c ) and by applying pharmacological agents (Figure  4c , Video  S8 ). Schematic summary of our results on the mechanisms of fluid transport and pressure generation in a mechanosensitive, peristaltic smooth‐muscle tube. Translating our results in clinically relevant terms, our results show that peristaltic smooth muscle tubes like the gut (and by extension the ureter or the Fallopian tube) can spontaneously generate a pressure difference between two compartments, that is, between 2 gut regions, between the kidney and the bladder, or between the peritoneal cavity and the uterus. If the pressure in these two compartments is determined by other factors (i.e., uterus body contractions, osmotic effects in the peritoneal cavity), mechanosensitive feedback of the smooth muscle tubes acts to stabilize (i.e., contribute to) this externally imposed pressure gradient. If the direction of waves propagating in the tube is imprinted by other physiological phenomena (pacemaker or nervous activity), the mechanosensitive feedback and the directionality of the waves can, depending on the direction of the waves relative to the external pressure gradient, either act jointly to increase pressure at the same end, or antagonize (Figure  5a,b ). The maximum pressure difference depends on the intrinsic contractility of the smooth muscle, ~25 mN/mm 2 in embryonic [ 37 ] versus ~70–170 mN/mm 2 in adult gut [ 38 ], and on the smooth muscle layer thickness, ~50 μm in E14 embryonic chicken intestine versus ~750 μm in human duodenum ( https://histologyguide.com/ ). Scaling up from the measured Δ P f  ~ 1 cmH 2 O in the embryonic chicken, we find that adult human duodenum segments can generate pressure differentials of 31–75 mmHg, which is in line with reported values using capsule manometry [ 39 ]. Ureter and Fallopian tubes with a smaller circular smooth muscle thickness of ~100 μm can generate Δ P of ~4–10 mmHg: these values are significant in light of the pressure differences measured across these organs [ 1 , 2 , 6 ]. We finally found that the propagation direction of waves can be reversed in high pressure conditions because the tissue stretch it creates acts as a new wave nucleation site that can overrule the physiological (nervous or pacemaker imprinted) propagation pattern. Such high pressure conditions can occur in pathophysiology, for example, intestinal obstruction [ 40 ]. All together our result shed light on the intricate pressure‐flow behavior of biologic peristaltic smooth muscle tubes, with important implications for bolus, urine, sperm or menstrual debris transport and associated pathologies ranging from gut dysmotility to endometriosis.

Peristaltic smooth muscle tubes generate local, propagating constrictions that propel fluid flow in many of our organs. A first example is the ~10 cm long smooth‐muscle lined Fallopian tube that connects the ovaries (and the peritoneal cavity) to the uterus. Intrauterine pressure can reach up to 50 mmHg [ 1 ], whereas peritoneal pressure is usually 7–12 mmHg [ 2 ]; the Fallopian tubes have to work against the uterine pressure to ensure that endometrial menstrual debris do not flow towards the abdominal cavity—an occurrence known as retrograde menstruation, and which is thought to be an important factor in the development of endometriosis [ 3 , 4 , 5 ]. A second example is the ureter, a 25 cm long tube connecting the kidney to the bladder, and generating successive waves of contractions. The pressure at the renal pelvis is ~5 mmHg, but the bladder pressure can vary from 5 to 20 mmHg at filling states 10%–80% [ 6 ]. Again, the ureter has to actively work against this pressure gradient to avoid flow of urine from the bladder towards the kidney, a pathological phenomenon known as vesicoureteral reflux. A third and last example is the intestine, in which peristalsis is the driving force for bolus transport. The circular smooth muscle (CSM) is the main effector of peristalsis in these organs and can by itself, or in coordination with pacemaker cells, generate motor activity. The pacemaker cells in the intestine are called the interstitial cells of Cajal (ICCs). They generate together with the CSM the peristaltic activity known as “slow waves” [ 7 ], which consists of rhythmic, propagating circular smooth muscle constrictions. Extrinsic (ureter, Fallopian tube) or intrinsic (gut) innervation adds a second, neurogenic layer of complexity to the motility of these organs. Both the smooth muscle and the enteric nervous system are mechano‐sensitive, that is, they react to distension and pressure changes [ 8 , 9 , 10 ], and both systems interact as the propagation of nervous signal along the GI tract is dependent on muscle tone [ 11 , 12 ]. The contractile response of isolated GI segments is classically studied using the Trendelenburg preparation [ 13 ], where the explant is cannulated at both ends, and fluid is injected at the oral end to elicit a contractile response which can be measured from optical recordings using spatio‐temporal maps [ 14 , 15 ], and by measuring the pressure and fluid outflow [ 16 , 17 ]. Hydrodynamic transport in the GI tract has been investigated both experimentally [ 16 , 18 , 19 ] and computationally [ 18 , 20 , 21 , 22 ] for different contractile regimes. While the flow rate of single peristaltic waves can be accurately predicted from simple Poiseuille models [ 18 ], they also give rise to complex reverse vortical flow that contributes to bolus mixing [ 16 , 18 , 22 ]. The gastro‐intestinal tract is further characterized by the presence of four sphincters (the lower esophageal sphincter, the pyloric sphincter, the ileocecal sphincter, and the internal anal sphincter), that act as valves that can subdivide the GI tract in different compartments with very different tone and pressure levels [ 23 ]. In all these examples, a peristaltic, mechano‐sensitive, liquid‐infused tube connects two reservoirs at different pressures. This raises the question: what happens when we replace the capillaries used by Poiseuille in his seminal study on load loss [ 24 ] by a biologic, mechanosensitive peristaltic tube? What are the resulting flow and pressure characteristics? Can at least the myogenic component be grasped within a unifying picture for the different organs presented? Although the hydrodynamic transport induced by simple peristaltic wave patterns is well understood in engineering [ 25 ] and biological pumps [ 16 , 18 , 19 , 20 , 21 , 22 ], an integrated picture of how the mechanosensitive contractile response of muscle to bolus pressure feedbacks on flow [ 17 , 26 ] is today still missing. To address these questions, we resorted to a simple biologic model organ, the fetal chicken small intestine. Motility of this organ at 14 embryonic days of development (E14) consists of rhythmic CSM contractions that propagate along the length of the intestine [ 27 ]. The myogenic contractions do not exhibit a preferred direction, propagating alternately in the rostro‐caudal (stomach‐to‐anus) or caudo‐rostral direction. At E16, as we will show, waves exhibit a unidirectional rostro‐caudal propagation direction imparted by the ICCs. At both stages, the longitudinal smooth muscle presents only weak contractions, that do not cause fluid motion. The E14 and E16 chicken guts therefore represent excellent miniature models of biologic, myogenic peristaltic pumps, one omnidirectional and the other unidirectional; because of their small size, they can be adequately oxygenated by diffusion in carbogen‐saturated organ bath experiments for prolonged recording times (from hours to days). We designed the simplest possible experimental setup, inspired from open‐boundary Trendelenburg preparations [ 16 , 17 ], that allowed us to measure pressure at both extremities of the gut segment, overall fluid flow, and peristaltic activity.

The authors declare no conflicts of interest.

All experiments were carried out according to the guidelines of CNRS animal welfare committees and conform to their principles and regulations. Dissection of embryos to retrieve their gastrointestinal tract is a terminal procedure that does not require constitution of an ethics approval form. Fertilized eggs were purchased from EARL Morizeau and incubated at 37°C for 14 and 16 days. The gut was carefully dissected in PBS and the mesentery removed and cut in segments 1–2 mm (Figure  1 ), 2 cm (Figures  2 and 3 ) or 6 cm long (Figure  4 ). (a) Brighfield image of duodenum frozen transverse section, showing distinctly the ~60 μm thick circular smooth muscle ring (CSM), here at E16. (b) A calibrated thin glass cantilever attached to a z‐stepper motor allows to apply controlled stretch and measure the force exerted by the smooth muscle ring (diameter ~800 μm at E14, length ~2 mm) by tracking the tip deflection. (c) Example contraction elicited by stretching an E14 ring. (d) The force of spontaneous contractions (averaged over 4–6 contractions occuring in 2 min) increases with increasing stretch of the preparation, n  = 5 rings (embryos) at E14, error bars are SD. (e) Effect of nifedipine, GdCl 3 , ryanodine, lidocaine and 2‐APB on the stretch‐induced contractile response (c) relative to the response to stretch of the same sample before the drug was applied, n  = 4–10 different rings (embryos). Numbers in parentheses indicate the concentration of drugs in μM. *Significantly different from zero, p  < 0.05, Student t ‐test. (a) A 2 cm long gut segment is cannulated with elastane threads (knot) to syringe needles; gut contractile activity is monitored with a camera mounted on a binocular; another camera (not shown) monitors the water levels in the rostral and caudal capillaries. Syringes allow control of the initial pressures P ir and P ic . (b) Gut contraction amplitude (difference between contracted and relaxed diameter, cf.Section  2 ) increases when pressure is applied on E14 guts ( n  = 13 samples, each represented by a different symbol shape or color). The average pressure is defined as ( P ir  +  P ic )/2. (c) Kymograph of E14 spontaneous activity showing multi‐directional propagation with nucleation/annihilation sites along the gut. (d) Kymograph of E16 shows rostro‐caudal (RC) waves modulated at the frequency of the neurogenic migrating motor complex (MMC). The MMC is abolished after tetrodotoxin application, but waves remain RC. (a) Spontaneous pressure differential generation from initially equal pressures Δ P i  =  P ic  −  P ir  = 0, here an E16 duodenum generates Δ P f  =  P fc  −  P fr  = 2.9 cmH 2 O. The decrease of the total water level P c  +  P r of ~1 cmH 2 O over the course of the experiment is due to inflation of the lumen by the liquid (see Section  2 ). (b) Δ P f versus wave directionality index at E14 and E16 at Δ P i  = 0. At E14, waves do not exhibit directionality and the developed Δ P f can have either sign. At E16, waves are predominantly rostro‐caudal, and the developed Δ P f is always > 0, i.e. liquid flowed with the waves in the rostro‐caudal direction. (c) Absolute Δ P f versus initially applied pressure P ir  =  P ic . Maximal development of Δ P f was achieved in the range 4–6 cmH 2 O. (d, e) Contractile waves had higher amplitude at the high pressure (black arrowheads) than at the low pressure end (white arrowheads). The differential wave amplitude is defined as A caudal  −  A rostral . Each different symbol color corresponds to a different sample ( n  = 7 at E14). (e) Typical pressure relaxation after applying an initial pressure gradient Δ P i , E14 duodenum. (f) Δ P f versus Δ P i at E14 ( n  = 16) and E16 ( n  = 11 at Δ P i  > 0, n  = 11 at Δ P i  < 0), lines are linear fits. Red arrows indicate samples that underwent wave direction reversal upon further increasing pressure. (g) Kymograph of wave reversal dynamics by pressure at E16: Waves are initially RC (top‐panel), when Δ P i  = 8 cm H 2 O is applied (middle panel) a caudal nucleation site emerges (white arrowheads) and dominates after 1 h (bottom panel), causing almost complete wave reversal (Video  S7 ). Each point in (b), (c) and (f) represents a distinct sample (gut, embryo). Effect of gut length and increasing contractility on flow rate and developed pressure in E16 duodenum. (a) Flow rate increases with gut length for n  = 8/9 samples. Each sample is represented in a different color and its length varied: Small, medium and long samples were respectively 1.8–2, 3–3.5, and 4.5–6 cm long (Section  2 ). The flow rate is inversely proportional to the time it takes to reach a stationary pressure state, as described in Section  2 . (b) Δ P f is independent of gut length ( n  = 9) (c) Tetrodotoxin (TTX) application increases smooth muscle contraction force, leading to increased Δ P f within 15 min after application ( n  = 5, Video  S8 ). *Significantly different, p  < 0.05, paired Student t ‐test. For characterization of myogenic mechanosensitive properties (Figure  1 ), a ring of E14 gut (length ~ 1–2 mm) was doubly cannulated, on one side to a stiff short stainless steel pin (Ø = 200 μm, l  = 5 mm), on the other to a thin flexible force‐calibrated glass cantilever (Ø ~ 30 μm, l  = 3–4 cm). The cantilever was calibrated by attaching weights (metal wires) of known mass and recording the deflection prior to the experiment; typical sensitivities were in the range 0.7–10.8 N/m. The glass cantilever was attached to a stepper motor to apply stretch to the ring preparation with a constant strain rate (0.3 mm/s). The gut ring was kept in a 5 mL trough filled with DMEM at 37°C and constantly bubbled with carbogen (95% O 2 , 5% CO 2 ). Bubbles were released in a cylindrical chimney made out of a metal grid, so they did not disrupt the force measurement. Drugs applied and their stock concentration are: nifedipine (Sigma Aldrich, 10 mM stock in DMSO), lidocaine (Bio‐Techne 3057, 100 mM in DMSO), 2‐APB (Tocris 1224, 100 mM in DMSO), GdCl3 (Tocris 4741, 50 mM in water) and ryanodine (Tocris 1329, 10 mM in DMSO). Displacement of the endpoint of the cantilever was recorded with a camera at 15 fps and tracked with the “Tracker” ImageJ plugin, yielding the force generated by the gut ring. When a stretch was applied (i.e., z‐motor actioned), the pure displacement (i.e., not deflection) induced by the z‐displacement of the cantilever tip was substracted from the tracked tip position to yield the deflection (force). The prominence (amplitude) of the force‐time peaks were measured using automatic peak detection under Matlab. To measure the flow properties of the gut as a function of pressure applied at the extremities (Figure  2 ), a 2 cm long gut segment was cannulated between two blunted syringe needle tips (23G, 1.5 cm long), in a 10 mL bath of DMEM medium warmed to 37°C and constantly bubbled with carbogen. The syringe tip positions were kept fixed during the whole duration of the experiment. The gut‐syringe junction was made leak‐proof by tying tight knots with 0.2 mm elastane thread at the rostral and caudal ends. The outer and inner diameter of the gut at E14 was approximately 0.8 and 0.3 mm; at E16 1 and 0.4 mm. The rostral (proximal) end of the gut was always placed left. Glass capillary tubes (1 mm diameter, 10 cm long) were inserted in a hole as close as possible to the syringe tip, and sealed with silicon paste. They enabled us to measure the pressure from the height of the water column at the left (rostral) and right (caudal) ends of the gut segment. The syringes allowed to raise independently the initial pressure on the rostral or caudal side of the preparation. The whole setup was installed on a Leica Z16APO binocular equipped with a Stingray 1600 × 1200 px camera to record gut motility by 1 Hz time‐lapse imaging. The water level in the capillaries was synchronously monitored by a second camera, and was measured using the ImageJ Tracker plugin. In each recording, we checked for possible leaks by comparing the injected fluid volume inside the gut (from the level rise in the capillaries) to the total volume in the system after 30 min–1 h. RC or CR fluid flow were determined from the sign of the expression ∆ P f − ∆ P i = P fc − P fr − P ic − P ir = P fc − P ic − P fr − P ir . When the fluid travels rostro‐caudally, the caudal level increases and the rostral level decreases, so ∆ P f − ∆ P i > 0 ; conversely, when the fluid travels caudo‐rostrally, ∆ P f − ∆ P i < 0 . Because the diameter of the gut varies with both muscle tone and intra‐luminal fluid volume, we could not assess muscle tone optically from our experiments. We note that the duodenum segment itself had a non‐negligible liquid capacity of ~5 mm 3 /cm versus 8 mm 3 /cm for the capillaries. At the beginning of the experiment, the lumen is empty and fills up gradually during the experiment; for a 2 cm long gut (Figure  3 ), it was usual to observe a decrease of up to ~2 × 5/8 ~ 1 cmH 2 O during the experiment, corresponding to the capacity of the filled gut. The calculated total hydrodynamic resistance of the needles + gut was ~100–200 Pa mm 3 /s, i.e., we calculate that it would take ~2 s for the liquid column to drop by 1 cm from an initial height of 5 cm, corresponding to a volume flow of ~200 mm 3 /min; gravitational volume flows are much higher than the ones induced by the guts (< 1 mm 3 /min, cf, Figure  4a ). For experiments in which we analyzed flow rate and pressure as a function of gut length (Figure  4a,b ), a 6 cm segment was initially cannulated and gradually shortened in 1.5 cm steps by sliding the caudal segment over the syringe needle length (1.5 cm), tying a new elastane knot, readjusting liquid levels to 3 cm rostral and 3 cm caudal, and measuring Δ P f after it reached a stationary state (30–60 min). The flow rate was measured as Δ P f  × capacity of the capillaries (7.9 mm 3 /cm) divided by the time it took to reach the stationary state. This procedure was repeated until the gut reached the inter‐needle length of 1.8 cm (i.e., the gut segment was completely straight). Each experiment lasted ~3 h: samples had a constant activity in terms of frequency and amplitude of contraction over the duration of the experiment. Tetrodotoxin (abcam 120055) was added at a concentration of 1 μM from a 1 mM stock solution in water (Figure  4c ). 2D space–time kymographs of the gut were produced with the ImageJ Reslice function as previously described [ 28 ]. Each traveling wave appears in this diagram as a slanted line; the sign of the slope gives the direction of travel, while the slope is proportional to wave speed. The spatial average (over the whole gut segment length) and time average (over the whole recording period) of slope (defined as 0: caudo‐rostral; +1: rostro‐caudal) were measured by applying the OrientationJ plugin [ 29 ] to the kymographs. This plugin yielded high resolution vector maps, which faithfully followed the contours of the slanted lines in the kymographs, and allowed to compute the wave directionality index. The contraction amplitude was measured from kymographs obtained from thresholded videos, such that the 8‐bit gray value of the kymograph was directly proportional to gut diameter. Line plots of the kymographs (i.e., local diameter as a function of time) were generated at the rostral end, in the middle of the sample, and at the caudal end. The local amplitude A of the contractions, defined as the difference in diameter in a region between the contracted and relaxed state, was measured using the “findpeaks” function under Matlab (prominence of the minima), and averaged over 50–140 contractions. The amplitude measured in Figure  2b is the average of A caudal , A middle and A rostral . The amplitude differential measured in Figure  3d is A caudal  −  A rostral . We recorded dye motion (Video  S6 ) by pinning full‐length E14 gut to a Sylgard coated Petri dish filled with 50 mL carbogen‐saturated DMEM at 37C; fluorescein (1 mg/mL in PBS) was injected with a syringe needle at the rostral end of the gut and its motion recorded with a fluorescence binocular (Leica, GFP filterset) at 1 Hz for up to 10 min. The gut was free to move except at the pinned points (3 along the length). No external pressure was applied for this experiment before or after the injection.

Figure S1. Change of contraction frequency and duration with increasing stretch. Figure S2. Weak stretch‐induced contractions can be elicited in Ca 2+ ‐deprived medium. Figure S3. Pressures in the rostral and caudal capillaries equalize when the gut acts as a passive pipe, in PBS at RT without Ca or Mg. Figure S4. Pressure‐induced contractile waves in room‐temperature PBS Ca 2+ 0.9 mM, Mg 2+ 0.5 mM. Figure S5. The higher amplitude of contractile waves at the high‐pressure end is particularly striking in longer gut segments. Figure S6. The average number of peristaltic contractions is proportional to gut length. Video S1. E14 gut spontaneous rhythmic contractions can be measured with a glass cantilever. Video S2. Applying a stretch induces a contractile response within 10 s after stress application; nifedipine (10 μM) strongly diminishes this stretch‐induced contraction. Video S3. 2‐APB (500 μM) completely inhibits the stretch induced contraction. Video S4. Multidirectional wave pattern at E14 and unidirectional RC pattern at E16, which is preserved after tetrodotoxin application. Video S5. Simultaneous view of capillaries and of the contractile activity of an E14 gut, showing continuous increase of rostral pressure but contractile waves travel in RC and CR directions. Video S6. Dye tracing experiment after injection of fluorescein in E14 gut, showing the difficulty of relating dye movement to the contractile wave pattern at this stage. Video S7. Dynamics of RC wave reversal in E16 guts by application of high caudal pressure. Video S8. Simultaneous view of rostral and caudal pressure and of the contractile activity, showing the increase of tone and caudal pressure after tetrodotoxin application at E16.