Philosophical Transactions of the Royal Society B: Biological Sciences


    Peristalsis enables transport of the food bolus in the gut. Here, I show by dynamic ex vivo intra-cellular calcium imaging on living embryonic gut explants that the most primitive form of peristalsis that occurs in the embryo is the result of inter-cellular, gap-junction-dependent calcium waves that propagate in the circular smooth muscle layer. I show that the embryonic gut is an intrinsically mechanosensitive organ, as the slightest externally applied mechanical stimulus triggers contractile waves. This dynamic response is an embryonic precursor of the ‘law of the intestine’ (peristaltic reflex). I show how characteristic features of early peristalsis such as counter-propagating wave annihilation, mechanosensitivity and nucleation after wounding all result from known properties of calcium waves. I finally demonstrate that inter-cellular mechanical tension does not play a role in the propagation mechanism of gut contractile waves, unlike what has been recently shown for the embryonic heartbeat. Calcium waves are a ubiquitous dynamic signalling mechanism in biology: here I show that they are the foundation of digestive movements in the developing embryo.

    This article is part of the Theo Murphy meeting issue on ‘Mechanics of development’.

    1. Introduction

    The intestine is a peristaltic pump: smooth muscle located in the gut wall constricts and the propagation of this constriction along the gut tube pushes the food bolus forward. By actively replenishing the nutrient source required for growth, peristalsis enabled pluricellular organisms to grow beyond the size limit imposed by diffusive food transport. Peristalsis is most clearly evidenced in the gastrointestinal tract, but is also central to uterine fluid-, sperm-, urine-, and even blood-flow in some animals. Motility in the adult gut is a complex phenomenon featuring several different patterns of movement [1] and controlled by two cell networks, the enteric nervous system (ENS) and the interstitial cells of Cajal (ICC). The ENS presents a great variety of neuronal cell types (excitatory, inhibitory, mechanosensitive, etc.) that communicate with the help of more than 35 different neurotransmitters [2].

    In recent years, a few groups involved in the budding and exciting research field of developmental physiology have highlighted how peristalsis movements emerge during embryonic development. Using different animal models (chicken [3], mouse [4,5], zebrafish [6]), they have all shown that, unlike adult peristalsis, the earliest detectable motility patterns in an embryo are purely myogenic (muscular): they involve neither neurons nor ICCs. We have recently released a comprehensive description [3] of peristalsis wave propagation patterns in the chicken embryo and of their main phenomenological characteristics (see electronic supplementary material, video S1): wave nucleation sites give rise to two waves travelling in opposite directions at a constant speed in the range 10–50 µm s−1; when two oppositely travelling waves meet, they annihilate; waves are calcium-dependent, as they vanish in calcium-free medium and are affected by calcium channel blockers (cobalt chloride); waves propel luminal content at least from E10 in chick (E17.5 in mice [7]), because green bile is present in the lumen along the entire midgut from this age. Because of its simple geometry (a tube) and because only one cell-type is involved (smooth muscle), the embryonic gut provides a tractable experimental model of how contractile waves emerge and propagate in biological systems.

    2. Results

    (a) Calcium waves underlie contractions in the early embryonic gut

    I used the cell-permeant intracellular calcium concentration indicator Fluo4-AM. The different histological layers of the gut had to be exposed to allow for permeation of the indicator. To this end, I performed thin transverse embryonic chick gut slices and imaged them after dye loading on an inverted confocal microscope (electronic supplementary material, figure S1a and Material and methods). The sample exhibited spontaneous, radial contractions occurring with a frequency of approximately 2–3/min and lasting for at least 4 h (figure 1a and electronic supplementary material, video S1). The transverse view clearly shows how the contraction obliterated the lumen. Contractions started at a point in the smooth muscle ring and quickly propagated circumferentially around the ring, leading to an overall radial constriction of the gut section. This radial constriction then propagated longitudinally, along the long axis of the intestine (electronic supplementary material, video S2). A key advantage of the transverse-slice configuration is that displacements induced by the spontaneous contraction are radial, so that cells remain roughly in the optical plane of the confocal microscope during imaging. This minimizes fluorescence fluctuations due solely to displacements of the sample in and out of the confocal imaging plane.

    Figure 1.

    Figure 1. Calcium wave imaging on living embryonic gut explants. (a) Brightfield still shots (electronic supplementary material, video S2) of one cycle of circular smooth muscle constriction of an E9 gut transverse section embedded in agarose gel. l, lumen; ep, epithelium; m, mucosa; csm, circular smooth muscle; o, outer layer (myenteric nerve plexus, not yet differentiated longitudinal smooth muscle, serosa). (b) Confocal fluorescence imaging, E9 jejunum transverse section after loading with calcium indicator. Dashed white arrows indicate the propagation direction of calcium waves in this sample (cf. electronic supplementary material, video S3). Fluorescence intensity analyses presented in (d) and (e) were performed on cells located at positions A, B, B′ and C. (c) Immunohistochemical staining of smooth muscle (red, α-actin) and enteric neurons (Tuj, βIII-tubulin, cyan) of the slice presented in (b), after calcium imaging. (d) Three representative patterns of displacement and fluorescence for different smooth muscle cells: single-spike (point B in (b) and electronic supplementary material, video S3), double-spike (point B′ in (b) and electronic supplementary material, video S3), subthreshold spikes (from electronic supplementary material, video S5). (e) Fluorescence as a function of time at points A, B and C in (b). (f) Circumferential (n = 6) and longitudinal (n = 7) propagation speed of the contractile wave, E9 jejunum. Each point corresponds to a different sample. An asterisk indicates a statistically significant difference (p < 0.05, Mann–Whitney two-tailed test).

    Electronic supplementary material, video S3 shows a representative (n = 7 samples from n = 4 embryos, n = 14 movies examined) time-lapse movie of calcium dynamics in a slice of living E9 jejunum. The contraction is seen to initiate at a point at 10–11 o'clock and to propagate in clockwise and counter-clockwise directions in the explant (figure 1b). The propagation of the contraction is accompanied by propagating flashes and flickers of fluorescence in individual cells (electronic supplementary material, video S3 and figure S2). The cell layer in which flashes occurred was also the one that contracted circumferentially, indicating that smooth muscle cells were the most likely source of calcium transients. Groups of cells located at the periphery of the gut slice (figure 1b; electronic supplementary material, video S3) were also statically fluorescently labelled by Fluo4-AM. Post-calcium imaging immunohistochemical staining (figure 1c) shows that the latter were mostly enteric ganglia (Tuj staining, see superposition of figure 1b,c in electronic supplementary material, video S4). In contrast, the localization of the propagating calcium wave coincided with the α-actin positive circular smooth muscle ring (figure 1b,c; electronic supplementary material, video S4). I devised a plugin that tracks the position of a cell in the slice during the contraction, and reads out the fluorescence intensity in a small region in its immediate vicinity (figure 1d, Material and methods). In the majority of cases (n = 13 out of 22 cells analysed from n = 3 samples from n = 3 embryos), the displacement peak coincided with the fluorescence increase (single-spike, figure 1d, left). In some rarer cases (n = 6/22), two distinct calcium spikes (figure 1d, middle) occurred for a single contraction event. I also observed subthreshold calcium spikes (n = 4/22), i.e. fluorescence fluctuations that were not accompanied by a contraction (figure 1d, right, arrows and electronic supplementary material, video S5). The fluorescence intensity change I/I0 of these subthreshold spikes were typically lower (ΔI/I0 ∼ 0.2–0.4) than the ones that accompanied a contraction (ΔI/I0 ∼ 0.4–1). Figure 1e shows the propagation of the fluorescence signal: the signal reaches its maximum intensity first at the wave nucleation point A (figure 1b), and then at more distant points (B and C in figure 1b). When counter-propagating calcium waves met along the gut circumference, they annihilated (electronic supplementary material, video S6). The speed of the propagating circumferential calcium wave was equal to that of the contraction, vcirc = 29.2 ± 6.8 µm s−1 (figure 1f, n = 20 waves, n = 6 samples). This speed is significantly higher than the longitudinal propagation speed, vlong = 12.1 ± 1.3 µm s−1 (n = 7) measured at E9 in similar conditions (figure 1f, data reproduced from [3]).

    (b) Propagating contractions are mediated by gap junctions

    Embryonic heart contraction waves have recently been shown to be independent of gap junctions [8]. To determine the role of gap junctions in the propagation of embryonic gut contractile waves, I used two pharmacological inhibitors: enoxolone (18β-glycyrrhetinic acid) and heptanol. I placed E7 and E9 embryonic guts on Anodisc membranes on Dulbecco's modified Eagle's medium at 37.5°C (see electronic supplementary material, figure S1b,c and Material and methods) and, after recording the native activity, added stepwise increasing concentrations of inhibitor (electronic supplementary material, video S7). Figure 2a,b shows representative motiligrams (see Material and methods for details); figure 2c,d shows the quantitative contractile wave frequency changes induced by enoxolone (n = 5) and heptanol (n = 5) in the jejunum and hindgut. Enoxolone 33 µM triggered an increase of contractile frequency by a factor of approximately 2, followed 20–30 min after application by a gradual decrease of contractile wave amplitude (figure 2a; electronic supplementary material, video S7). The subsequent increase to 100 µM led to very quick fading of contraction amplitude and to the eventual cessation of wave generation and propagation (figure 2a; electronic supplementary material, video S7). The application of 100 µM heptanol at E7 led to a marked decrease of contractile frequency (2/5 samples), or to its cessation (3/5 samples). Heptanol at 500 µM completely abolished motility in all samples (5/5). The effects of enoxolone at stage E9 (n = 5, electronic supplementary material, figure S3) were identical to those at E7 (figure 1a,c). Importantly, the effects of enoxolone were reversible, i.e. motility was recovered after the drug was washed away (n = 6, electronic supplementary material, figure S4). This shows that the abolition of gut motility by 100 µM enoxolone was not the result of drug toxicity at these concentrations, but was due to gap junction inhibition.

    Figure 2.

    Figure 2. Effect of gap junction inhibitors on motility. (a) Representative motiligrams of E7 jejunum in control conditions, after addition of 33 µM and 100 µM enoxolone. (b) Motiligram of E7 jejunum in control conditions, after addition of 100 µM and 500 µM heptanol. (c,d) Change of contractile wave frequency in jejunum (black) and hindgut (red) in control conditions and after addition of enoxolone or heptanol. An asterisk indicates a statistically significant difference (p < 0.05, Mann–Whitney two-tailed test) with respect to control conditions.

    (c) The early embryonic gut is mechanosensitive

    Pressure can induce contractions in the adult gut [9]. To examine the response of embryonic guts to mechanical stimulation, I used a thin, pulled Pasteur pipette with a glass bead formed at the tip (figure 3a, top). I found that gently pressing the gut with this bead consistently (more than 25 pinches at stage E8–E9 on n = 6 samples) led to the nucleation of two waves travelling away from the point where pressure was applied (figure 3a,b and electronic supplementary material, video S8). Wave nucleation could be elicited from all gut segments (jejunum, ileum or hindgut). This reflex is extremely sensitive as even a light, delicate brush (approximately 1–2% deformation in an approximately 10 µm2 area) with a very thin pipette was sufficient (n = 10/10 brushes on n = 4 samples) to trigger wave nucleation (electronic supplementary material, video S8). I could elicit waves by mechanical stimulation at stage E6 (electronic supplementary material, video S8, n = 4/4 samples), the earliest stage at which physiological spontaneous propagating contractions occur [3]. I further found that pre-incubating embryonic guts for 30 min with tetrodotoxin (TTX, 1 µM), a potent neuronal inhibitor (Na+ channel blocker), did not alter the response of the guts to mechanical stimulation (figure 3c and electronic supplementary material, video S8, E9 n = 5/5 samples). We have previously shown that spontaneous contractile wave activity is TTX-insensitive at stages E7 and E9 [3].

    Figure 3.

    Figure 3. Pressure triggers propagating contractile waves. (a) Reaction to pinching of E9 gut (still shots from electronic supplementary material, video S8). Red arrowheads point at the two outgoing contractile waves. (b) Motiligram derived from electronic supplementary material, video S8, of the control gut. A pair of waves travels away (white arrows) from the point where the gut was pinched (black arrowhead). A second pair of waves nucleated at the same location approximately 30 s later (white arrowhead). (c) Reaction to pinching of E9 gut after 30 min treatment with tetrodotoxin (1 µM, still shots from electronic supplementary material, video S8).

    (d) Signalling by intercellular mechanical tension is not required for contractile wave propagation

    I have shown that the gut is sensitive to minute mechanical stimulation. It is therefore reasonable to hypothesize that the tensile forces generated by locally contracting cells may be felt by neighbouring cells that, in turn, contract. This elementary propagation mechanism has recently been shown to direct contractile waves in the embryonic heart [8] and may also play a role in the embryonic gut. To test this hypothesis, I took inspiration from an elegant experiment designed by Johansson & Ljung to test the role of mechanics in adult vein contraction [10]. To mechanically interfere with the contraction, I embedded an approximately 300 µm long gut segment in a drop of stiff agarose gel (figure 4a and Material and methods). Gel embedding effectively reduced the contraction amplitude Inline Formula (drest, dcstr: resting and constricted diameter) from 7–8% in agarose-free regions to less than 1% in the agarose embedded segment. The displacement of cells located within the gut was strongly reduced because of the new boundary condition imposed on the outer gut surface. I found that, in spite of the lack of gut wall displacement, peristalsis waves could travel unimpeded from one side of the agarose-embedded region to the other (electronic supplementary material, video S9). In the motiligram of figure 4a, wave propagation (slanted lines) in the non-embedded regions surrounding the agarose is clearly seen; I highlight propagation in the agarose with white dashed lines because the reduction of amplitude induced by the gel makes it is otherwise difficult to see. There was no discontinuity in the slope of the waves as they travelled through the agarose block (n = 5), indicating that blocking the displacement of the outer wall did not affect wave propagation velocity (figure 4a, bottom scheme).

    Figure 4.

    Figure 4. Testing the role of mechanics in embryonic gut contractile wave propagation. (a) Agarose gel embedding experiment. Top: scheme of the experiment. Middle: representative motiligram, from electronic supplementary material, video S9. Dashed white lines connect contractile waves propagating on either side of the gel. Bottom: explanatory scheme of expected motiligram depending on wave speeds in- and outside of the gel. (b) Effect of almost completely sectioning the gut on contractile wave propagation. Top: Scheme of the experiment. Bottom: Motiligram (E8 jejunum, from electronic supplementary material, video S10) showing that the site where the cut is performed generates contractile waves immediately following the injury. (c) Almost completely sectioned midgut segment, and position of wave as a function of time in regions 1–5. (d) Effect of collagenase on motility (E8, midgut): motiligram and comparison of wave speed before/after collagenase application. Wave speed was deduced from the slope (white dashes) of the lines in the motiligrams. An asterisk indicates a statistically significant difference (p < 0.05, Mann–Whitney two-tailed test).

    In a second set of experiments, I disrupted the mechanical continuity of the gut by sectioning it almost completely. The two pieces were connected only by a flap of cells at the outer border of the gut (figure 4b,c) and placed perpendicularly (figure 4c) to each other to minimize communication of tensile forces across the flap. I observed that immediately after the cut was performed, the site of wounding became the source of numerous contractile waves travelling away from the cut (figure 4b motiligram, electronic supplementary material, video S10, n = 5). After approximately 10 min the gut returned to a normal propagation pattern and waves travelling towards the cut in one segment could propagate through the flap of cells to the other gut segment (figure 4c, n = 5). The velocity of the wave remained constant as it travelled through the flap of cells (figure 4c, bottom).

    I next examined whether reducing the elastic modulus of the gut tissue using collagenase (cf. Material and methods) would influence contractile wave propagation speed. Softening with collagenase has been shown to decrease propagation velocity in the embryonic heart; this effect was backed by a mechanical model of contractile wave propagation [8]. I found that approximately 5 min after application, collagenase led to a near doubling (+80%) of the contraction frequency (figure 4d). The average speed of the waves (figure 4d bottom) before and after collagenase application was respectively 20.1 ± 1.8 µm s−1 (n = 6) and 28.3 ± 2.7 µm s−1 (n = 6), i.e. collagenase induced an approximately 40% wave velocity increase.

    I finally tested the influence of applying a longitudinal stress on contractile wave propagation. Changing the longitudinal pre-stress of muscle fibres could alter the propagation speed of contractile waves, much like increasing the tension of a violin string increases the velocity of deflection waves. Applying pressure has recently been shown to increase the frequency of smooth muscle contraction in the embryonic mouse lung [11]. I applied tensile force by stretching the gut with a thin glass cantilever [12] (figure 5a). Contractile wave frequency and speed of propagation in the jejunum were recorded for 10 min without strain, and then at 3–4 different, increasing strain values, for n = 5 guts. I found that stretching in the strain range 0–50% (corresponding to a stress of approximately 0–500 Pa) affected neither wave velocity nor wave generation frequency (figure 5b). In some instances I observed contractile wave generation upon strain application, consistent with the mechanosensitive properties of the gut described above.

    Figure 5.

    Figure 5. Influence of longitudinal strain on contractile wave propagation and generation. (a) Experimental set-up: the gut is pinned at one end and stretched at the other with a glass cantilever [12]. (b) Speed and frequency of contractile waves as a function of applied longitudinal strain. Each colour corresponds to a different gut (E9, n = 4 and E8, n = 1 in blue). Speed error bars represent the s.d. of n > 4 waves. Frequencies were computed from 10 min periods spent at each strain value.

    Embedding in gel (figure 4a) or cutting (figure 4b,c) did not prevent contractile wave propagation. In these experiments, I cannot exclude that some mechanical tension is still communicated from cell to cell, either by cells located in the gut interior (gel experiment) or through the cells located in the flap (cutting experiment). Wave velocity was however unaffected in either experiment, even though the mechanical properties of the medium in which the waves propagate were significantly modified. Modifying the longitudinal tensile pre-stress of muscle fibres did not affect longitudinal wave propagation speed or frequency (figure 5). This suggests that mechanical cues do not play a central role in the propagation mechanism of early gut contractile waves. This conclusion is further supported by the fact that softening of the gut by collagenase led to an increase in wave velocity (figure 4d), contrary to what is predicted by mechanical models of wave propagation in biological tissues [8].

    3. Discussion

    We previously reported that a calcium channel blocker (CoCl2) inhibits gut motility in the embryo, and that embryonic motility vanishes in calcium deprived medium [3]; these observations are consistent with the idea that calcium activity underpins early embryonic gut contractility. The shape of the calcium spikes I measured (sharp up-rise followed by a slower decay, figure 1d,e) is typical of calcium spikes in smooth muscle [13,14]. A wide literature has been devoted to calcium waves and several of their properties give straightforward explanations of the dynamics of the early gut, in particular:


    Early gut contractile waves propagate at constant speed, and with an undiminished amplitude [3]. These are known properties of Ca2+ waves [15]. The contractile wave propagation speeds we measure are in the range 10–40 µm s−1, i.e. similar to the speed of cytosolic calcium fertilization waves [16] and 2–3 orders of magnitude slower than embryonic heart (approx. 20 mm s−1 [17,18]) or arterial smooth muscle (5–15 mm s−1, [19]) calcium waves. The fact that the circumferential (along the fibre axis) propagation speed is higher than the longitudinal (perpendicular to the fibre axis) speed (figure 1f) is consistent with previous measurements on adult gut [20].


    When two counter-propagating contractile waves meet along the digestive tract, they annihilate (electronic supplementary material, videos S1 and S6). This is a direct consequence of the existence of a refractory period [21], i.e. the necessary time for CaATPase to reduce cytosolic Ca2+ concentration back to its pre-excitatory level so that the tissue can become excitable again. Annihilation of calcium waves has been observed in cardiomyocytes rings [17] and I observe them here directly on gut explants (electronic supplementary material, video S6).


    Wounding leads to the nucleation of numerous contractile waves at the injury site (figure 4b). It is known that calcium waves are emitted after wounding of epithelia [22] or whole organisms [23]. The calcium up-rise is probably due to influx from the extracellular medium due to disruption of membrane integrity after wounding [24,25], and has been shown to orchestrate wound healing [23,26].


    Calcium waves can be elicited in smooth muscle cell cultures by gentle mechanical stimulation [19,27]. Bayliss & Starling [28] discovered more than a century ago that peristalsis movements in the adult (dog) intestine can be triggered by pinching or inserting bolus. The insertion of a bolus results in contraction of the muscle coat above the bolus, and its relaxation downward of the bolus. This behaviour permits downward transport of the bolus and was coined the ‘law of the intestine’. Here, I show that an essential part of this reflex—namely the ability to sense the bolus by the mechanical pressure it exerts on the gut wall—is present at the earliest stages of gut development in the embryo; it only involves muscle cells as ICCs are not yet differentiated at E6–E8 [29] and I found this reflex to be resistant to tetrodotoxin, a potent neuronal inhibitor. Pinching the embryonic gut led to the symmetric nucleation of two waves travelling away from the point where pressure was applied. Further development of a full, asymmetric reflex with ascending contraction and descending relaxation very likely requires coordinated input from enteric nerves [30]. The stimulus I exert by pinching does not mimic pressure or distension within the lumen. I expect the latter to give rise to a similar reaction, because the gut was sensitive to minute stimulation (electronic supplementary material, video S8), and the reflex is unlikely to depend on the directionality of the mechanical excitation. It is reasonable to think that this reflex can be physiologically stimulated during embryogenesis, for example by the accumulation of amniotic fluid in the lumen. Note that contractile waves can also nucleate spontaneously (see all electronic supplementary material videos except video S8), i.e. their generation does not necessarily require mechanical stimulation.

    Hao et al. [31] recently discovered spontaneous calcium waves in enteric neural crest cells (ENCC) in embryonic mouse gut at stages E12.5–E16.5. The waves were transmitted via purinergic P2 receptors and not gap junctions. Contractions of the gut could occur without ENCC Ca2+ waves. The ENCC Ca2+ waves subsided after stage E16.5, whereas contractions became more vigorous. The calcium activity I report here is located in the circular smooth muscle and contractions were always accompanied by a calcium transient unless the dye had bleached. As Hao et al. used a mouse strain in which only ENCCs express the calcium indicator, they could not observe the calcium activity in smooth muscle. I did not observe calcium transients in enteric neurons. The Fluo4-AM fluorescence signal tended to be saturated in enteric neurons (figure 1b,c), which could have obscured readout of the calcium activity in these cells. However, in ganglia where the signal was not saturated (e.g. ganglia at 9.30, 10 and 11 o'clock in electronic supplementary material, video S3) or just partially saturated (e.g. ganglia at 8.30 o'clock in electronic supplementary material, video S3), I did not observe any fluorescence spikes.

    The fact that gap-junctional communication is required for contractile wave propagation/generation is consistent with previous work on Ca2+ communication in smooth muscle cell cultures [14,32] and adult gut motility [33]. Other small molecules that diffuse through gap junctions, like ATP, may play a role in the propagation of the calcium wave [31,34]. The blood vessels of a connexin-40 deficient mouse mutant have been shown to exhibit peristalsis-like activity [35]. This suggests that a gut and a vein may differ, from the point of view of their dynamics, only by the amount of gap-junctional coupling between adjacent circular smooth cells: high coupling results in synchronized contraction over long distances, i.e. vasomotion, whereas low coupling gives rise to local, propagative constrictions, i.e. peristalsis.

    Chiou et al. [8] found that gap junctions were not required for embryonic heartbeat propagation and that the heartbeat contraction velocity depended linearly on tissue stiffness. They concluded that early embryonic heart wave propagation is based on an active, local feedback to mechanical stretch. In contrast, my results indicate that, although the gut is mechanosensitive, the transmission of mechanical force from cell to cell during a contraction does not play a role in the propagation of the contractile wave. My experiments rather support a model in which propagation characteristics of the contraction are determined by the electrical conduction properties of the calcium wave across gap junctions. Johansson & Ljung [10] came to the same conclusion in their experiment on rat portal vein. In the adult, mechanical stimulation exerted by the bolus as it transits down the gut lumen has been shown [9] to be an integral part of its self-propulsion mechanism, by triggering sequential reflex contractions/relaxations. Development of this ‘neuromechanical loop’ [9] likely requires enteric nerve activity, and is therefore not yet present at the embryonic stages I examine here.

    Calcium waves propagating along a smooth muscle tube provide, at an embryonic stage, most of the essential dynamic features of peristalsis in the adult: the contractions obliterate the lumen, propagate, push luminal content and are mechanosensitive. The functional role of these calcium waves as a transient form of motility in the developing gastrointestinal tract remains to be elucidated: do these waves play a morphogenetic role? Do they transfer information to the already present, but quiescent enteric nervous system? These are some of the prominent questions that will require further investigation.

    4. Material and methods

    (a) Specimen preparation and ethics statement

    Fertilized chicken eggs (EARL Morizeau, France) were incubated from 6 to 9 days in a humidified 37.5°C chamber. The gut (hindgut to duodenum) was dissected out of the embryo and the mesentery was carefully removed. The experiments were conducted under European directive 2010/63/UE. The approval by an ethics committee is not required for research conducted on birds at embryonic stages.

    (b) Calcium imaging

    Chicken guts were embedded in 4% agarose type VII gel and thin (50–200 µm) transverse slices were cut at the level of the jejunum with a razor blade. For calcium indicator loading, 50 µg of Fluo4- AM (Thermofisher, F14201) were dissolved in 10 µl DMSO; this solution was added to 2 ml of DMEM GlutaMAX™-I (Thermofisher, 4.5 g l−1 D-glucose, sodium pyruvate, Ca2+ 1.8 mM, Mg2+ 0.8 mM) containing 1% penicillin-streptomycin (PS), 25 mM HEPES and 0.01% Kolliphor EL (Sigma). Kolliphor [20] was essential for permeation of the dye in the smooth muscle cells. Up to 4 slices were placed in 1 ml of the loading solution for 30 min at RT. After washing, deesterification was performed in DMEM at 37.5°C in a 5% CO2 incubator, for at least 30 min and at most 4 h. Samples were shielded from light during the whole procedure. Imaging was performed on individual slices placed in a grazing layer of DMEM (electronic supplementary material, figure S1a) on an inverted Olympus confocal microscope, at magnification ×40, laser excitation 488 nm, emission filter 512 nm, 1 Hz imaging frequency, 200 ms exposure time, at 37.5°C. Images were collected from one confocal plane (no z-stack). Irreversible bleaching of the sample occurred after approximately 3 min imaging, i.e. contractions continued but the calcium concentration transients could not be observed anymore. This was not due to dye leakage as the transients of samples that had meanwhile been kept in the dark for up to 4 h could still be observed.

    (c) Calcium fluorescence signal analysis

    Cells statically labelled by Fluo4-AM were used as fiducial markers. Coordinates x(t) and y(t) were retrieved with the Tracker ImageJ plugin (courtesy of O. Cardoso), yielding the displacement Inline Formula. An ImageJ macro computed the average fluorescence intensity in an elliptical region of interest (ROI) ∼ 5–10 µm in size (∼one cell) located at a fixed distance of no more than approximately 30 µm of the tracked point. The signal was normalized by I0, defined as the minimal intensity recorded in this ROI in the image stack (video). Fluorescence changes due to displacements of the sample in and out of the confocal imaging plane were small and can be discerned from calcium spikes because the fluorescence intensity change they induce is smooth and overlaps with the displacement curve (unlike the calcium spikes in figure 1d).

    (d) Immunohistochemistry

    The slides were fixed for 1 h in 4% PFA, washed, blocked for 1 h in a 1% bovine serum albumin (BSA) and 0.1% triton in PBS solution, and incubated overnight in anti-α smooth muscle actin (Abcam, 5694, dilution 1 : 1000) and anti βIII-tubulin antibody (Abcam 14545, dilution 1 : 1000). After washing, CY3- and A488-conjugated secondary antibodies (ThermoFisher, dilution 1 : 400 in PBS) were applied for 2 h at RT. The slides were washed and imaged at ×40 with the confocal microscope. The images were rotated to have the same orientation as on the calcium activity recordings. I used the shape of the epithelium and the localization of enteric ganglia as guides to find the correct rotation angle.

    (e) Whole-gut motility monitoring set-up and motiligram generation

    Guts were placed on a porous alumina membrane (Anodisc™, electronic supplementary material, figure S1b) resting on the edge of a tissue culture dish (Ø ∼ 20 mm) filled with DMEM + 25 mM HEPES as previously described [3], in a humidified Petri dish at 37.5°C. Motiligrams (figures 24) were generated from the image stacks by tracing a rectangular ROI along the edge of a midgut segment and by using the ImageJ function ‘Reslice’ (see [3] for details). Drugs and chemicals were added directly to the culture medium underneath the Anodisc.

    (f) Drugs and chemicals

    Tetrodotoxin 1 µM (Abcam). Collagenase-dispase 0.33 mg ml−1 (Roche): the gut can be dissociated by pipetting after 5 min at 37°C at this concentration, indicating that softening occurs within the first minutes after collagenase application. Enoxolone (18β-Glycyrrhetinic acid, Sigma-Aldrich) 33 µM (DMSO 1‰) and 100 µM (DMSO 3‰): DMSO alone does not affect gut motility at these concentrations. Heptanol 100 µM and 500 µM (Sigma-Aldrich): a 10 mM heptanol concentrate in PBS solution was prepared, vigorously agitated to homogenize the two-phase solution (milky white) and diluted in the sample medium.

    (g) Mechanical manipulations of the embryonic gut

    To block displacement of the gut wall, a drop of agarose type VII 4% w/w was placed with a thin syringe needle on a portion of midgut lying on an Anodisc membrane and allowed to solidify at RT for approximately 5 min. Exact position of the agarose-embedded segment could be visualized by phase-contrast imaging of the sample prior to time-lapse imaging. Precision cuts of the GI tract were realized with a surgical micro-scissor (Euronexia). Longitudinal stress/strain was applied to the gut as previously described [12], in DMEM at 37°C.

    Data accessibility

    All relevant data are within the paper and its electronic supplementary material files. Higher resolution videos are available from the author upon request.

    Competing interests

    I have no competing interests.


    This research was funded by a CNRS/INSIS Starting Grant ‘Jeune Chercheur’, by the CNRS ‘Défi Mécanobiologie 2018’ and by the Labex ‘Who Am I?’ (ANR-11-LABX-0071).


    I thank Vincent Fleury for proofreading the manuscript, Nicolas Dacher for help performing the experiment in figure 5 and Max Piffoux for suggesting to test the reversibility of enoxolone.


    Electronic supplementary material is available online at

    One contribution of 14 to a Theo Murphy meeting issue ‘Mechanics of development’.

    Published by the Royal Society. All rights reserved.