Smooth muscle contractility causes the gut to grow anisotropically

Abstract

The intestine is the most anisotropically shaped organ, but, when grown in culture, embryonic intestinal stem cells form star- or sphere-shaped organoids. Here, we present evidence that spontaneous tonic and phasic contractions of the circular smooth muscle of the embryonic gut cause short-timescale elongation of the organ by a purely mechanical, self-squeezing effect. We present an innovative culture set-up to achieve embryonic gut growth in culture and demonstrate by three different methods (embryological, pharmacological and microsurgical) that gut elongational growth is compromised when smooth muscle contractions are inhibited. We conclude that the cumulated short-term mechanical deformations induced by circular smooth muscle lead to long-term anisotropic growth of the gut, thus demonstrating a self-consistent way by which the function of this organ (peristalsis) directs its shape (morphogenesis). Our model correctly predicts that longitudinal smooth muscle differentiation later in embryogenesis slows down elongation, and that several mice models with defective gut smooth muscle contractility also exhibit gut growth defects. We lay out a comprehensive scheme of forces acting on the gut during embryogenesis and of their role in the morphogenesis of this organ. This knowledge will help design efficient in vitro organ growth protocols and handle gut growth pathologies such as short bowel syndrome.

1. Introduction

Stem-cell derived intestinal organoids [1] form sphere- or star-shaped [2] structures instead of tubes. This observation shows that molecular information contained within the stem cells is not sufficient to trigger uniaxial anisotropic growth, making it necessary to factor in geometrical and mechanical cues [3] to fully understand organogenesis. Muscle contractility has been shown to be an essential actor of bone growth [4], lung branching [5], epithelial fold formation in the gut [6–9], oviduct [10], and even whole body-elongation in Caenorhabditis elegans [11]. We [12] and others [13] have shown that prolonged static longitudinal mechanical stress causes elongational growth of the gut. In a previous report [12], the mechanical stress we applied mimicked the in vivo force exerted by the vitelline duct on the herniating early (E6–E9) embryonic gut loop. In the present report, we elucidate how force generated by smooth muscle within the gut directs the growth of this organ.

The intestine presents at an early stage (day 6–7 of development in the chicken, week 7 in the human [14]) structural anisotropies at several levels: (1°) it is geometrically anisotropic, (2°) it has anisotropic non-muscle (passive) mechanical properties because collagen fibres wind circumferentially around the gut tract [15], (3°) it has anisotropic active contractile properties due to the presence of a circular smooth muscle (CSM) ring. CSM contractility has both static (muscle tone) and phasic components. Phasic contractions are caused by spontaneous, propagating calcium waves [16]. Both their frequency (in the 10–100 mHz range) and amplitude (0–20% diameter reduction) increases as development proceeds [17,18]. We aimed here at understanding how these structural and contractile anisotropies relate to organ growth. To this end, we measured gut passive and active mechanical properties at different developmental stages. We then correlated mechanical properties with the pattern of organ growth in an innovative culture set-up. We found that high gut contractility (at E10) was correlated with anisotropic elongational growth in culture whereas low contractility (at E7) resulted in isotropic growth. We further found that inhibiting smooth muscle contractility at E10 by pharmacological or microsurgical methods induced a switch from elongational to radial growth. We conclude that circular smooth muscle directs the anisotropic growth of the embryonic intestine, and discuss the implications of our findings for mammalian development (mouse, human), short-bowel syndrome (SBS) and organoid research.

2. Results

2.1. Circular smooth muscle activation triggers an immediate mechanical elongation of the gut which is significantly more pronounced at E10 than at E7

E7 and E10 midgut present the same tubular geometry and basic histology: they both have been colonized by enteric neural crest cells [19] and both present a circular smooth muscle ring that gives rise to spontaneous contraction waves [17]. Note that the contraction waves are only due to the circular smooth muscle layer, as the longitudinal layer only differentiates at E13 [7] and becomes contractile at E14 [20]. We recorded CSM phasic peristaltic activity by time-lapse imaging E7 and E10 guts lying on porous membranes on DMEM at 37°C (figure 1a, electronic supplementary material, video S1, Material and methods). The frequency of peristaltic waves at stages E7 (n = 12) and E10 (n = 5) was respectively 11.7 ± 2.7 mHz and 64.9 ± 15.2 mHz and their amplitude 4.4 ± 3.3% and 16.4 ± 1.2%. All values in this article are reported as mean ± standard deviation. Both frequency and amplitude of phasic CSM activity were significantly higher (by a factor approx. 4–6) at E10 than at E7 (figure 1a,b). To measure tonic CSM activity, we compared gut morphologies in DMEM at 37°C before and after application of the broad-spectrum calcium channel blocker CoCl₂, an efficient inhibitor of CSM activity [17]. Within minutes, CoCl₂ led to an increase of gut diameter and decrease of gut length (electronic supplementary material, video S2). After 1 h application the shapes of the guts were stable and spontaneous contractions were abolished. Morphological changes were slight at stage E7 but conspicuous at stage E10 (figure 1c). Length change ${\epsilon }_z\hspace{0.17em}$ at E7 (n = 7) and E10 (n = 5) were respectively −6.3 ± 5.5% and −27.2 ± 3.1%, the accompanying diameter change was respectively 4.5 ± 2.5% and 26.4 ± 7.6%; these differences were statistically significant (figure 1d). The morphological changes induced by CoCl₂ 1 mM applied for 1 h are not due to toxicity because these effects were reversible: smooth muscle contractility was recovered (n = 3) after washing out the drug and incubating E10 guts overnight.

Figure 1.

Gut circular smooth muscle (CSM) phasic and tonic properties. (a) Still shots of peristaltic contractions of E7 and E10 guts (electronic supplementary material, video S1). (b) Comparison of peristalsis wave frequency and amplitude at E7 (n = 12) and E10 (n = 5). All n values refer to different guts (different embryos). (c) Still shots of E10 and E7 guts in DMEM at 37°C before and after application of CoCl₂ 1 mM for 1 h (electronic supplementary material, video S2). The drug halts spontaneous contractions, as can be seen by comparing the corrugation of the E10 gut before and after application (and in electronic supplementary material, video S2). It also changes the morphology of the gut. (d) Change in length and diameter of E7 and E10 guts after 1 h CoCl₂ application, E7 (n = 7) and E10 (n = 5). *p < 0.05, Mann–Whitney two-tailed test. (e) Scheme of the deformation induced by circular smooth muscle tone. Muscle fibre tension is indicated with red arrows and the resulting centripetal pressure with black arrows. When CSM fibres contract (arrow 2), gut diameter decreases and the gut elongates. (Online version in colour.)

The morphological changes induced by CoCl₂ can be explained by a simple mechanical model (figure 1e): when CSM fibres are active, they contract (red arrowheads), exerting a net uniform, radially oriented pressure on the gut. When CSM tone is relieved by CoCl₂ application (arrow 1) the gut increases in diameter and, as a consequence of its incompressibility, shortens (electronic supplementary material, video S2). Conversely, when CSM contracts (arrow 2), the gut thins out and elongates. Crudely modelling the gut as a cylinder of Poisson ratio ${\nu }_{\theta z}$ and elastic modulus E_Z (see values at E7 and E10 below) an estimate of the pressure generated by smooth muscle $P=E_z{\epsilon }_z/{\nu }_{\theta z}$ can be computed: we find 370 ± 320 Pa at E7 and 1850 ± 370 Pa at E10. The embryonic gut is thus, during development, constantly subject to a centripetal mechanical stress generated by its own musculature, whose net mechanical effect is to cause elongation. We find that this effect is much more pronounced at stage E10 than stage E7, because circular smooth muscle is thicker and stronger at later stages [17].

2.2. Both E7 and E10 guts display anisotropic passive mechanical properties

We characterized non-muscle, passive mechanical properties of guts at stage E7 and E10 by performing stretch (figure 2a) and distension (figure 2d) assays in DPBS (Ca²⁺ 0.9 mM, Mg²⁺ 0.49 mM) at room temperature. In these conditions, the guts did not exhibit any spontaneous muscle activity, nor was stretching and/or distension able to elicit contractions [16]. We used thin, force-calibrated glass cantilevers (Material and methods) to measure the longitudinal elastic modulus E_Z (figure 2b,c, Material and methods). We found E_Z,E7 = 1570 ± 290 (n = 6) and E_Z,E10 = 2460 ± 60 (n = 6).

Figure 2.

Passive (non-muscle) mechanical properties of the gut. (a) Set-up to measure the longitudinal elastic modulus E_Z of the embryonic midgut by stretching with a glass microfibre. (b) Typical stress–strain curves and power law fits (dashed lines). Different symbol shapes correspond to data from consecutives stretches on the same gut (shown here two stretches for E7 and three stretches for E10). (c) Longitudinal elastic modulus of jejunum at E7 (n = 6) and E10 (n = 6). All n values refer to different guts (different embryos). (d) Still shots from gut pressurization experiments (electronic supplementary material, video S3). Upon pressurization, the gut elongates, increases in diameter and the wall thickness decreases. Inset E10: zoom showing wall-thickness. Note the different scale bars for E7 and E10. (e) Evolution of the wall thickness as a function of time (here at E10), dashed line: polynomial fit. (f) Representative diameter change ε_θ versus length change ε_z at E7 and E10. (g) Orthoradial elastic modulus E_θ at E7 (n = 7) and E10 (n = 9), phase II. (h) Second harmonic generation (SHG) images of longitudinal (left) and transverse (right) sections of E9 jejunum. Collagen I and III are the main molecules contributing to the SHG signal. *p < 0.05, Mann–Whitney two-tailed test. (Online version in colour.)

To measure the orthoradial elastic modulus $E_{\theta }$, we canulated a midgut segment with a micropipette, sealed the segment and monitored thickening ${\epsilon }_{\theta }$, lengthening ${\epsilon }_z$ and wall-thickness w as fluid pressure in the gut lumen was increased (electronic supplementary material, video S3, figure 2d). In the initial stages of pressurization, the gut wall thins out significantly as the lumen gets filled with fluid (figure 2d,e, I); as pressure is further increased, the change in wall thickness becomes small (figure 2d,e, II) and even negligible (figure 2d,e, III). In phases II and III the gut can be modelled as an orthotropic, thin-walled cylinder with anisotropic mechanical properties along two axes. The ratio of lengthening over thickening ${\epsilon }_z/{\epsilon }_{\theta }$ during phase II of the inflation (figure 2f) at stages E7 and E10 were respectively 2.4 ± 0.3 and 1.1 ± 0.8, yielding E_θ,E7 = 7730 ± 890 Pa and E_θ,E10 = 5340 ± 310 Pa (figure 2g, electronic supplementary material, figure S1, Material and methods). An isotropic pressurized cylindrical shell has ${\epsilon }_z/{\epsilon }_{\theta }=\mathrm{0.5.}$ From the values of ${\epsilon }_z/{\epsilon }_{\theta }$ measured above, we can conclude that both E7 and E10 guts are anisotropic. We found similar results analysing phase III data (electronic supplementary material, figure S2). Second harmonic generation imaging (SHG, cf. Material and methods) shows that collagen fibres [15] form a dense circumferential ring (figure 2h, right), whereas longitudinal fibres are scarce (figure 2h, left). This molecular-scale anisotropic structure is consistent with the macro-scale stiffness anisotropy we report.

2.3. In culture, E7 guts grow isotropically whereas E10 guts elongate anisotropically

We next examined how muscle and non-muscle mechanical properties relate to growth of the organ. We have previously demonstrated that, in ovo, the gut loop is stretched by the vitelline duct [12]. To isolate the effect of smooth muscle force on growth we therefore dissected the guts from their surrounding tissues. We found that an appropriate oxygen supply was critical to achieve growth of the isolated intestine. By constantly bubbling carbogen (95% O₂ – 5% CO₂) in the culture medium (figure 3a), we established an oxygen level of 24.2 ml l⁻¹ in the vicinity of the guts that is in the range of oxygen levels measured in the chorioallantoic vein of chicken embryos at similar stages (see Material and methods). Electronic supplementary material, video S4 shows the vivid peristaltic activity of E10 guts after 48 h culture in these conditions. Immediately after removing samples from culture, guts appeared very elongated because of CSM activity (see mechanism in figure 1e). After 1 h in DPBS at room temperature, CSM activity became extinct and the guts had relaxed to a stationary shape. In these conditions, the shape of the guts after culture could be compared in the same physico-chemical conditions (DPBS at RT) as before culture (i.e. after dissection). Figure 3b shows representative morphological changes of E7 and E10 midguts (duodenum to ileum) before and after culture. We quantitatively assessed these morphological changes using a Voronoi algorithm (Material and methods). We found that gut length, diameter and volume increased respectively by 12 ± 21%, 33 ± 12%, 94 ± 34% at stage E7 (n = 15) and 25 ± 14%, 7 ± 7%, 44 ± 20% at stage E10 (n = 32).

Figure 3.

Compared growth pattern of E7 and E10 guts in culture. (a) Scheme of the culture system (Material and methods). (b) Representative images of E7 and E10 guts before and after 48 h culture and subsequent relaxation of the guts for 1 h in PBS. (c) Length, diameter, volume and dry mass change of E7 (n = 15) and E10 (n = 32) guts after 2 day culture and 1 h relaxation in PBS. Each n-value and data point corresponds to a different gut (different embryo). (d) Relative contributions of lengthening $\mathrm{\Delta }{V}_l/\mathrm{\Delta }V$ and thickening $\mathrm{\Delta }{V}_d/\mathrm{\Delta }V$ to growth at stage E7 and E10. Note that $\mathrm{\Delta }{V}_l/\mathrm{\Delta }V$ and $\mathrm{\Delta }{V}_d/\mathrm{\Delta }V$ for each sample are related by $\mathrm{\Delta }{V}_l/\mathrm{\Delta }V+\mathrm{\Delta }{V}_d/\mathrm{\Delta }V=1$, i.e. the data have a mirror symmetry around the horizontal axis corresponding to isotropic growth $\mathrm{\Delta }{V}_l/\mathrm{\Delta }V=\mathrm{\Delta }{V}_d/\mathrm{\Delta }V=\mathrm{0.5.}$ (e) Immunohistochemistry for proliferating cells (anti-histone H3 phospho S10), smooth muscle α-actin (SMC) and βIII-tubulin (enteric neurons) on E7 and E10 transverse sections after 2 day culture. *p < 0.05, Mann–Whitney two-tailed test. (Online version in colour.)

Length increase was higher at E10 than at E7, while diameter and overall volume increase was higher at E7 than at E10; all differences were statistically significant (figure 3c). Culture led to a dry mass increase (Material and methods) of 159 ± 28% at stage E7 (n = 7) and 60 ± 29% at stage E10 (n = 21). The approximately three times higher volume and dry mass growth rate of E7 guts compared to E10 could be due to the higher proportion of proliferative, non-differentiated cells at younger stages or to the better oxygen tissue perfusion of the smaller E7 guts. Figure 3e shows that proliferating, histone 3 (phospho S10)-positive cells were found throughout the gut mesenchyme and epithelium after culture. Guts featured well defined enteric and smooth muscle layers (figure 3e, Tuj and SMA). The CSM ring was much thicker for E10+2 than for E7+2 guts. To quantify growth anisotropy, we computed the contributions of lengthening $\mathrm{\Delta }{V}_l$ and thickening $\mathrm{\Delta }{V}_d$ (Material and methods) to the overall volume change ΔV. We found that most (n = 11/15) E7 guts grew isotropically (i.e. $\mathrm{\Delta }{V}_l/\mathrm{\Delta }V\sim \mathrm{\Delta }{V}_d/\mathrm{\Delta }V\sim 0.5,$ figure 3d, blue circle); some samples (n = 4/15) thickened and shortened (figure 3d, dashed blue circle). In contrast, E10 guts anisotropically lengthened (figure 3d, red circle) in 80% of samples (n = 26/32). The remaining 20% of samples (n = 6/32) thickened rather than lengthened. We found that out-of-trend behaviours in both age groups were correlated with low overall growth (electronic supplementary material, figure S3), indicating that they may result from poor vitality of the organ after dissection or during culture.

We visualized the elongational growth of E10 guts in culture by time-lapse imaging (electronic supplementary material, video S5): the guts are initially pinned at their extremities and buckle as they progressively elongate, while undergoing constant peristaltic wave activity. This is to our knowledge the first live, organ-scale video of a growing embryonic gut.

2.4. Inhibition of smooth muscle contractions by pharmacological or surgical means induces a switch from elongational to radial growth

Weakly contractile E7 guts grow isotropically whereas strongly contractile E10 guts grow anisotropically (figures 1 and 3). To investigate the possibility that contractility is causal in directing growth anisotropy, we pharmacologically inhibited smooth muscle contractions with the l-type calcium channel inhibitor nicardipine [21]. This dihydropyridine derivative has been shown to inhibit contractility in the embryonic mouse gut [18] and is used for the treatment of hypertension and angina pectoris [22]. It was therefore a promising candidate to relax muscle activity without affecting tissue viability and growth. We found that 5 µM nicardipine consistently abolished motility in the whole gut within 10 min after application. Whereas motility in the caeca and hindgut was abolished for 48 h following application, motility in the midgut gradually recovered, albeit at a much reduced amplitude and frequency compared to control guts (electronic supplementary material, figure S4a,b). The volume and mass increase of control and nicardipine-treated guts were similar (figure 4b) after 48 h culture. However, whereas control guts featured anisotropic longitudinal growth (Δl = 42 ± 16%, Δd = 7 ± 5%, n = 8, figure 4b,c), nicardipine-treated guts grew mostly radially (Δl = 12 ± 17%, Δd = 24 ± 13%, n = 8, figure 4a–c).

Figure 4.

Disrupting smooth muscle contractions pharmacologically (a–c) or surgically (d–h) induces a switch from elongational to radial growth. (a) Example image of E10 gut cultured with L-type channel blocker nicardipine (5 µM) before and after culture. The gut thickened but length change was small. (b) Length, diameter, volume, dry mass change and (c) growth anisotropy of control (n = 8) and nicardipine treated (n = 8) guts after 2 day culture. Each n-value and data point corresponds to a different gut (different embryo). (d) Closed (control) and opened (e) E10 gut segments after 48 h culture, Δl indicates the length change. The cut was performed longitudinally, along one border of the gut (dashed line), causing the gut to spring open, revealing the epithelial grooves inside. (f) Length and dry mass change of control (n = 8) and opened (n = 8) segments. (g) Density of proliferating (PH3-positive) cells of closed (n = 4) and opened (n = 4) gut segments, assessed from (h) confocal images of anti-histone H3 phospho S10 immunolabelled transverse gut sections after 48 h culture. *p < 0.05, Mann–Whitney two-tailed test. (Online version in colour.)

As a further test of the muscle-driven anisotropic growth theory, we disrupted the continuity of the line of tension generated by circular smooth muscle by cutting longitudinally one gut wall with surgical micro-scissors (figure 4e). When the gut wall is cut, circumferential tension is released, causing the gut to open up and become an approximately 2D sheet of tissue. This procedure has been used to measure circumferential stress in arteries [23] and prepare 2D gut tissue for physiology [24] and developmental biology [25] experiments. We compared the growth characteristics after 48 h of intact (tubular) gut segments (figure 4d) with those of opened segments (figure 4e). We found that the length change (figure 4f) of the tubular and opened segments were respectively 27 ± 21% (n = 8) and 1.7 ± 4.4% (n = 8), i.e. tubular segments elongated whereas open segments did not. Open segments after 48 h were characterized by an exteriorized epithelium that covered the outer periphery of the explant (figure 4h). Although we could not directly quantify their diameter and volume from 2D pictures, transverse sectioning revealed that the opened segments after culture (figure 4h right) were noticeably and uniformly thicker (dashed white line) than one wall (i.e. the radius) of the closed tubes (figure 4h left). We investigated whether surgical opening of the segments could have induced growth defects by intrinsically reducing explant viability or proliferation. We found that all opened explants (n = 8) still displayed rhythmic smooth muscle contractions after 48 h culture, albeit not leading to elongation as described in figure 1e. We found that the dry mass increase of samples (figure 4f) and the density of proliferating cells (figure 4g) after 48 h culture were similar for tubular and opened preparations. This indicates that the elongation defect in opened preparations is due to the inhibited self-squeezing effect (figure 1e) following disruption of the circumferential tension generated by the circular smooth muscle.

3. Discussion

Together, the embryological (E7 versus E10, figure 3), pharmacological (nicardipine inhibited contractions at E10, figure 4a–c) and microsurgical (opened E10 segments, figure 4d–h) comparative growth experiments all indicate that anisotropic elongational growth is severely affected when circular smooth muscle contractions are weak or otherwise disrupted. These results show that the short-timescale (approx. 1 h), purely mechanical lengthening induced by CSM contractions we described in figure 1c–e leads to irreversible elongational growth when these deformations are cumulated on long timescales (48 h), as first hypothesized in 1958 by Coulombre & Coulombre [26].

3.1. Different mechanical forces guide the growth of the developing intestine

Importantly, although E7 guts presented anisotropic non-muscle mechanical properties (figure 2c,g), they grew isotropically in culture (figure 3d). This shows that the information leading to anisotropic gut growth is not encoded (1°) by molecular information contained within individual cells of the gut, (2°) by the initial cylindrical geometry of the gut or (3°) by passive, non-muscle tissue mechanical anisotropies (i.e. collagen fibre orientation). We have recently shown [12] that, from E5 to approximately E9, a tensile force transmitted via the vitelline duct pulls the early gut loop out of the coelom (white arrow in figure 5c, free-body diagram in figure 5d), causing physiological herniation. We had demonstrated [12] that applying a static, longitudinally directed force mimicking the vitelline duct tension to E7.5–E8 guts in culture was sufficient to induce anisotropic elongational growth. Here we find that E7 guts not subject to an externally applied mechanical force grow isotropically when perfused with O₂. Together, these experimental facts support the view that the tensile force transmitted via the vitelline duct is essential for the initial physiological constant-diameter lengthening (figure 5a) of the gut. This umbilical pull sets the initial high aspect ratio of the organ.

Figure 5.

Growth kinetics and mechanical forces acting on the gut during embryogenesis. (a) Physiological evolution of chicken midgut length (blue circles) and average diameter (red triangles). The midgut comprises the duodenum, jejunum and ileum up to caecal appendix tips. Sample numbers: E5 n = 8, E6 n = 7, E7 n = 5, E8 n = 6, E9 n = 7, E10 n = 8, E11 n = 7. Error bars are ± s.d. The length is fitted with an exponential (black line). The growth curves are consistent with those reported by other investigators [26,27]. (b) Length (blue circles) and diameter (red triangles) evolution of the duodenum from E6 to 3 days post-hatching, each point is the mean of at least 10 specimens. The experimental data were collected by Coulombre & Coulombre [26] and is adapted here. (c) Representative pictures of E6, E8 and E12 guts in the abdominal cavity of the embryo in PBS, immediately after it was extracted from the egg. The pulling force F exerted by the vitelline duct at stage E6 and E8 is indicated with an arrow. At stage E12, spontaneous in vivo CSM contractions are conspicuous (white arrowheads). (d) Proposed model of the main forces acting on the embryonic gut at different developmental stages and of their relation to organ growth. Bottom images: inflation of an anisotropic balloon or squeezing of a tube of toothpaste illustrate two possible mechanisms by which circular mechanical constraints can lead to anisotropic deformation or flow. (Online version in colour.)

As from approximately E10, vitelline duct tension becomes negligible as additional gut loops start forming by buckling [28] (figure 5c, E12). The dominant mechanical constraint acting on the gut as from this stage are compressive, radially oriented forces due to CSM contractions (figure 5c,d, E12, in vivo phasic contractions are pointed at with white arrowheads). We have demonstrated here that CSM forces alone can drive further anisotropic growth. CSM could act as a stiff corset [8] that skews proliferation along the longitudinal direction (figure 5d, anisotropic balloon analogy). Phasic contractions could also make the tissue creep in the longitudinal direction (figure 5d, toothpaste analogy), by actively remodelling cell–cell and cell–matrix junctions in the gut interior. Longitudinal smooth muscle differentiates in the chicken at E13 [7]. Our mechanical growth model correctly predicts that the build-up of tone and phasic contractions along the longitudinal direction shortly thereafter, at E14 [20], initiates a period of relatively slower longitudinal growth and relatively faster diameter growth (figure 4b, experimental data adapted from Coulombre et al. [26]).

The functional role of contractile waves generated by spontaneous calcium transients [16–18] in the developing gut has remained obscure, because the digestive tract does not transport any nutrients during embryogenesis. Our results show that these waves, together with smooth muscle tone, drive intestinal elongation. This provides a striking example of how the mechanical function of an organ (peristalsis) drives its shape (morphogenesis).

3.2. Mouse smooth muscle mutants exhibit growth defects

Several groups have reported gut morphological phenotypes in mice with genetically disrupted smooth muscle. Mutations included LMOD1 [29], laminin α5 [30], and Brahma (Brm) & Brahma-related gene 1 (Brg1) [31]. In all of these studies, weakened smooth muscle contractility, differentiation or organization resulted in significantly shorter (50–80% wild-type length), bulkier guts at embryonic stages. These findings support the role of smooth muscle as an essential driver of gut elongation in the embryo. Care must however be taken in comparing the morphogenesis in different animal species. Whereas epithelial villi formation in the chicken results from a mechanobiological buckling instability [7], their emergence in mice is independent of muscle activity and can be derived from a Turing field in which an inhibitor of Bmp signalling acts as the Turing activator [25]. It cannot therefore be excluded that mechano-biologically driven scenarios of organogenesis in the avian embryo may differ in mammals.

3.3. Implications for human gut development, short-bowel syndrome and organoids

As a rule of thumb, days of development in the chicken correspond to weeks of development in the human [20]. In the human embryo, MRI of the gut of embryos from the Kyoto collection [32] show it forming a hairpin-like umbilical hernia between Carnegie stages (CS) 14 and 18 (week 5–7), much like in the early chicken embryo (figure 5b, E8). Distinct undulations of the gut are seen at stages CS19–CS21 (week 7.5–8), which are similar to the phasic smooth muscle activity seen in the chicken in figure 1a. Although not commented upon by the authors [32], these undulations are, to the best of our knowledge, the first evidence of early gut motility in the human embryo. We believe that contractions of the gut can be seen because the embryos of the Kyoto collection were immediately fixed in paraformaldehyde following abortion. The presence of contraction waves at this stage is consistent with the fact that CSM is differentiated [33] in the human gut at least as from CS 18 (week 7). Vitelline duct tension and the squeezing effect due to CSM are thus present at early embryonic stages in the human embryo as well, raising the possibility that the mechanically driven growth mechanism we demonstrated in the chicken (figure 5) may also apply to humans. SBS is a devastating pathology resulting in dramatically reduced gut growth. Mutations in CLMP and FLNA have recently been found to cause SBS [34]. FLNA binds to actin and CLMP co-localizes with actin filaments. CLMP has moreover been found to strongly downregulate the expression of connexin 43 and 45 in mice [35], and we have recently shown that gap junctions (connexins) are essential for early gut motility [16]. Both actin-binding protein and connexin defects could severely hamper normal CSM contractility, relating it to the SBS phenotype. This causal link remains to be asserted however, because it is unclear whether SBS is paired with defective gut contractility or motility [34]. Mechanical distension is being considered to lengthen the intestine of SBS patients [36]. Our research suggests that an innovative therapeutic approach could involve boosting motility and circular smooth muscle tone.

Intestinal stem-cell derived organoids can adopt star-like shapes [2]. We believe that this may be due to a mechano-biological growth instability: if the initially spherical shell of smooth muscle within the organoid develops more in some locations than in others, these spots will experience greater smooth muscle strain, elongate, thus giving rise to protrusions (the arms of the star). Just like asymmetric chemical signalling cues [37], mechanical forces can provide a symmetry-breaking cue. Our study suggests that a way of growing anisotropic, cylindrical organoids would require breaking the initial spherical symmetry of the system, e.g. by depositing the stem cells on a rectangular pattern or in a cylinder. Our investigations reinforce the view that smooth-muscle generated forces [8,9] play a central role in establishing whole-organ shape, and thus functionality. Controlling the geometry and direction of contraction forces generated by muscle within organoids will prove essential for tissue engineering and for the interpretation and remediation of organ growth defects.

4. Material and methods

4.1. Specimen preparation

Fertilized chicken eggs were purchased from EARL Morizeau (Chartres, France, breeder Hubbard, JA57 hen, I66 rooster, yielding type 657 chicks). The eggs were incubated at 37.5°C in a humidified chamber for 5 to 12 days. The gastrointestinal tract was dissected out from the embryos from hindgut to proventrilicus and the mesentery carefully removed.

4.2. Characterization of smooth muscle contractility and passive mechanical properties

To characterize their peristaltic activity, guts at stage E7 and E10 were placed on Anodisc membranes (Whatman, pore size 0.2 µm) resting on cylindrical wells filled with DMEM GlutaMAX™-I (with 4.5 g l⁻¹ d-glucose and sodium pyruvate), as previously described [17]. To characterize static CSM tone, freshly dissected guts were incubated at 37°C in a 5% CO₂ atmosphere for 1 h in a thin (0.5 mm) layer of DMEM. In this configuration the whole gut is free to move and deform (unlike on Anodisc membranes, where the gut is held down by meniscus forces). One millimolar CoCl₂ was introduced and the morphological changes of the guts were monitored after approximately 1 h (at which time they had reached a stationary state, cf. electronic supplementary material, video S2). We characterized motility by computing the speed of the waves and their frequency$\hspace{0.17em}f=n/\mathrm{\Delta }t,$ where n is the number of waves passing through a given position in a time Δt. The amplitude was measured directly from the video as $A=(d_{\mathrm{rest}}-d_{\mathrm{constricted}})/d_{\mathrm{rest}},$ where d_rest is the resting diameter of a point along the gut and $d_{\mathrm{constricted}}$ is the diameter at the same position when it is constricted by a contraction wave. Characterization of passive mechanical properties relies on previously established protocols [38] and microscopy techniques [15] that we describe in the electronic supplementary material.

4.3. Organotypic culture

For E7-E10 compared growth (figure 3), the stomach of demesenterized guts was pinned with a needle to the side of a Sylgard coated rectangular trough (H × W × D 50 × 80 × 32 mm) filled with 140 ml DMEM supplemented with 1% penicillin–streptomycin. The medium was saturated with carbogen (95% O₂, 5% CO₂) by continuous bubbling. For reference, blood in the vitelline artery (supplying the gut) in the chicken embryo at E6 has a total oxygen concentration (dissolved + carried by haemoglobin) of 13.5 ml l⁻¹ [39], and of 89 ml l⁻¹ [40] in the blood of the chorioallantoic vein at E10. The equilibrium dissolved O₂ concentration of carbogen at 37.5°C is 24.2 ml l⁻¹, which lies in this physiological range. We verified this value (electronic supplementary material, figure S5a) by direct measurement with a Clark-type electrode (Unisense, Denmark). We set the carbogen bubbling rate so that rising bubbles generated a light convection current in the trough. This insured that oxygen was replenished in the vicinity of the guts, without exerting significant hydrodynamic forces on the guts. The density of the guts is almost that of the culture medium, so that gravity effects can be neglected. In this culture system, the only forces acting on the guts are the phasic and tonic components of the smooth muscle. The trough was covered with a cap and the guts cultured for 2 days in a humidified incubator (Thermos) at 37.5°C. For the experiments with nicardipine (figure 4a–c), gut growth was achieved by putting individual samples in a shallow layer of medium (1 ml in a 35 mm diameter Petri dish) and incubating all samples in an atmosphere of 95% O₂, 5% CO₂ at 37°C. Diffusion of the oxygen through the shallow layer of DMEM resulted in a concentration of 21 ± 5 ml l⁻¹ O₂ (n = 3) at a distance of approximately 1 cm from the organ (measured with Clark electrode, see electronic supplementary material, figure S5b). Nicardipine was prepared as a stock solution (10 mM) in DMSO and diluted in the medium to 5 µM. Control guts were incubated in the equivalent quantity of vehicle, i.e. 0.5 ‰ DMSO. Drug and medium were refreshed after 24 h culture. For open gut preparations (figure 4d–h), approximately 1 cm long E10 gut segments were pinned at one end to a Sylgard-covered tray. The opening was performed with Vannas microscissor (CM-85-AS30, Euronexia, France) and with the help of a binocular. Samples were then pinned to the other end and cultured in a shallow layer of DMEM (1 ml in a 35 mm diameter Petri dish) in a carbogen atmosphere for 48 h. Morphology of all guts pre-culture was measured in DBPS at room temperature. Morphologies of all guts post-culture (figures 3 and 4) was measured after relaxation of the samples for 1 h in DBPS at room temperature.

4.4. Analysis of gut morphometric changes

We follow a previously described procedure [12]. Raw images of the guts lying in a shallow depth of DPBS (so they were in 2D plane) were processed to erase the hindgut, stomach and any residual tissue around the axially symmetric midgut (segment comprised between caecal appendix tip to duodenum–stomach junction). We then thresholded these images and extracted the midgut contour (Matlab routine). We next applied a Voronoi Matlab algorithm to this contour to obtain the gut midline length, average diameter, and volume. We neglect the volume occupied by the lumen, which is close to 0 at E7, and only approximately 5% at E10 [7]. In approximately 20% of samples after culture, certain regions of the guts presented local (along max approx. 10% of gut length) accumulation of intraluminal fluid or DMEM—we erased these bulges to exclude them from the total volume. Lengthening $\mathrm{\Delta }{V}_l$ and thickening $\mathrm{\Delta }{V}_d$ were defined as $\mathrm{\Delta }V=\mathrm{\Delta }{V}_l+\mathrm{\Delta }{V}_d\hspace{0.17em}=(\mathrm{\partial }V/\mathrm{\partial }l)\mathrm{\Delta }l+(\mathrm{\partial }V/\mathrm{\partial }d)\mathrm{\Delta }d=\pi d^2\mathrm{\Delta }l/4+\pi dl\mathrm{\Delta }d/2,$ where ΔV is the total volume change, d and l are respectively the initial gut average diameter and length and Δd and Δl the gut diameter and length change induced by culture. Isotropic growth corresponds to $\mathrm{\Delta }{V}_l/\mathrm{\Delta }V=\mathrm{\Delta }{V}_d/\mathrm{\Delta }V=0.5$.

4.5. Dry mass determination

Water was removed from the midguts by bathing them successively in 1 : 2, 2 : 1 and 1 : 0 absolute ethanol : water mixtures, for 30 min in each solution. This procedure warrants that all ethanol-miscible components like water, salts and any residual intraluminal fluid are washed away. Guts were then dried in air either on steel pins or on a weighing cantilever (E7). Dry masses of native and cultured E10 guts were measured with a precision balance (Sartorius). For E7 guts, the balance did not provide the required precision, we weighed them using calibrated cantilevers as previously described [12]. We measured a dry mass density of native E7 and E10 guts of ρ_E7 = 0.097 ± 0.01 mg mm⁻³ (n = 4) and ρ_E10 = 0.125 ± 0.002 mg mm⁻³ (n = 7). We used these densities to compute the initial dry mass m_i of the guts before culture, $m_i={\rho }_{\mathrm{E}7\hspace{0.17em}\mathrm{or}\hspace{0.17em}\mathrm{E}10}{V}_i$, where V_i is the initial volume. The dry mass change induced by culture (figure 3c and 4b,f) is $(m_f-m_i)/m_i$, where m_f is the measured dry mass of the gut after culture.

4.6. Immunohistochemistry

Guts were fixed for 1 h in a 4% PFA in PBS solution, washed in PBS, then left overnight in 30% sucrose in water solutions, and embedded the next day in OCT compound (VWR) on dry ice. Thin (14 µm) slices were cut at −20°C in a Leica cryotome and deposited on Thermofrost glass slides. After rehydration, the slides were blocked for 15 min in a 1% BSA and 0.1% triton in PBS solution, the slides were then incubated overnight in anti-α smooth muscle actin antibody (Abcam, ref5694, dilution 1 : 2000), anti βIII-tubulin antibody (Abcam, ref14545, dilution 1 : 1000) or anti-histone H3 (phospho S10) (Abcam, ref14955, dilution 1 : 200) solution composed of 1% BSA in PBS; the following day, after washing, CY3- and A488-conjugated secondary antibodies (ThermoFisher, dilution 1 : 400 in PBS) were applied for 2 h. The slides were washed in PBS, and immediately imaged in a layer of PBS with a confocal microscope (Olympus).

4.7. Statistical analysis

All samples are included in the presented data, no randomization or blinding was used, median, upper (75%) and lower (25%) interquartile range and means (empty square) are presented in figures 1–5. All pairwise statistical analysis were performed using a two-tailed Mann–Whitney test. Differences were considered statistically meaningful at p < 0.05 (indicated by a star).

Ethics

The experiments were conducted under European law article 2016/63/UE. The approval of experimental protocols by an ethics committee is not required for research conducted on chickens at embryonic stages. All experiments were performed in accordance with the ethics guidelines of the INSERM and CNRS.

Data accessibility

All data, code and materials are presented within the paper.

Authors' contributions

D.K., Y.K., N.D. and N.R.C. performed experiments and analysed data; N.R.C. conceptualized and supervised the research and wrote the article.

Competing interests

We declare we have no competing interests.

Funding

This research was funded by a CNRS/INSIS Starting Grant ‘Jeune Chercheur’ and by the CNRS ‘Défi Mécanobiologie 2018’.