LR defects associated with E-cad and Myo1D modifications coincided with cell sliding in the hindgut

During the LR asymmetric development of the embryonic hindgut, its epithelial tube concurrently rotates and elongates through cell sliding and convergent extension, respectively [22]. However, whether these two epithelial dynamics operate independently of each other or have mutual interdependence remains unclear. To distinguish between these two possibilities, we genetically altered the activities of E-cad and Myo1D that have been shown to play essential roles in the LR asymmetric development of the hindgut and then examined whether such genetic modifications affect either cell sliding or convergent extension or both.

We previously found that the LR asymmetry of the hindgut is mostly randomized in E-cad-mutant embryos at stage 14 [9]. Consistent with the previous idea that cell chirality of the hindgut epithelial cells drives the LR asymmetric rotation of the hindgut, cell chirality was found to be abolished in these E-cad-mutant cells [9]. In this study, to determine the roles of E-cad in cell sliding and convergent extension, we performed live imaging of the hindgut in an E-cad-null mutant, shotgun^R69 (shg^R69) [27]. UAS-RedStinger (encoding a nuclear marker) and UAS-myr-GFP (encoding a membrane marker) were specifically expressed in the hindgut epithelium by the Gal4–UAS system [28–30]. From the dorsal side of the embryo, three-dimensional time-lapse images of the hindgut were taken for 120 min from late stage 12, which is sufficient time for the wild-type hindgut to complete its 90-degree rotation (Figs 1A1–3, S1 Video) [22]. However, the E-cad-mutant hindgut of the hardly rotated under this condition (Figs 1B1–3, S2 Video). To visually detect cell sliding, each column of nuclei that lined up along the anterior–posterior axis at 0 min was indicated by different colored circles (Fig 1D1). After 30 and 60 min, the relative positions of the nuclei did not change markedly in E-cad-mutant embryos (Fig 1E1–1E3 and S3 Video); however, the columns of nuclei slanted to the left in the wild type, as reported previously (Fig 1D1–1D3 and S4 Video) [22]. Thus, cell sliding was severely diminished in E-cad-mutant embryos.

Download:

• PNG
  larger image
• TIFF
  original image

Fig 1. E-cad and Myo1D perturbation affected hindgut rotation and cell sliding.

(A1–C4) Still shots from time-lapse Videos of the hindgut epithelium visualized using myr-GFP (membrane in green) and RedStinger (nuclei in magenta) from 0 to 120 min in the wild type (WT) (A1–A3), E-cad homozygote (B1–B3), and WT overexpressing Myo1D (Myo1D O/E) (C1–C4). (D1–F3) Displacement of nuclei (visualized by RedStinger) from 0 to 60 min in the hindgut epithelium of the WT (D1–D3), E-cad homozygote (E1–E3), and Myo1D O/E (F1–F3). Three central columns of nuclei are marked by green, yellow, and blue circles. (G) Schematic to quantify x displacement of a nucleus. The coordinates of two nuclei located along the anterior–posterior axis at two time points (t = t1 and t = t2) were observed. The subjacent nucleus was set at (0, 0), and the relative displacement of the upper nucleus in the x direction was measured every 30 min until 60 min. (H) Boxplots of x displacement in the E-cad mutant (n = 175, N = 7), WT (n = 164, N = 7), and Myo1D O/E (n = 180, N = 5). n and N in parentheses represent cell and embryo numbers, respectively. Black dots are outliers. ***p < 0.001. *p < 0.05. For all images, the anterior is on top. From A1 to F3, the time elapsed from the start of the Video is shown on the lower right. Scale bars in A1 and D1 are 25 and 10 μm, respectively.

https://doi.org/10.1371/journal.pgen.1011422.g001

To confirm this observation quantitatively, we measured relative cell displacement in the x direction, as described previously [22]. In brief, we set a subjacent (posterior in the hindgut) nuclear position at (0, 0) coordinates and measured the relative x displacement of the above nucleus against the subjacent nucleus every 30 min (Fig 1G). The wild-type hindgut epithelium had negative x displacement, showing leftward relative movement, as reported previously. In contrast, the E-cad epithelium hardly showed x displacement (Fig 1H) [22]. These results demonstrated that cell sliding requires E-cad.

Myo1D exhibits dextral activity to define the enantiomorphic states of organ and cell chirality [9,25,26]. Thus, organ and cell chirality become the mirror images of those noted in the wild-type counterparts in the absence of Myo1D [9,25,26]. Consistent with these findings, compared with the wild type, the direction of cell sliding is reversed in Myo1D mutants [22]. In contrast, the overexpression of wild-type Myo1D induces de novo chirality in organs and cells [25,31]. Therefore, in this study, we overexpressed wild-type Myo1D in the hindgut epithelium driven by byn-Gal4 and analyzed the potential reinforcement of cell sliding. As shown in Fig 1A, the wild-type hindgut took approximately 120 min to complete 90-degree rotation (Fig 1A1–1A3). However, upon the overexpression of Myo1D, the hindgut occasionally showed overrotation (six out of 13 embryos). Under this condition, the hindgut completed its 90-degree rotation in 90 min and continued to rotate further at least until 120 min (Fig 1C1–1C4, S5 Video).

We found that such acceleration of hindgut rotation coincided with an increase in cell sliding (Fig 1F and 1H). In the Myo1D overexpression condition, the columns of nuclei at 0 min gradually slanted to the left after 30 and 60 min; the degree of sliding was greater than that noted in the wild type (Fig 1F1–1F3 and S6 Video). This result was quantitatively supported by the result of relative cell displacement analysis, which revealed significantly more negative x displacement than that in the wild type (Fig 1H). These results suggest a positive correlation between the degree of cell sliding and the expression level of Myo1D. Based on the results of E-cad and Myo1D analyses, we established conditions to reduce and facilitate cell sliding, which could be used to assess whether cell sliding is coupled with convergent extension.

E-cad and Myo1D did not affect convergent extension

Under the aforementioned conditions, we next analyzed whether cell sliding is concomitant with convergent extension in the hindgut epithelium. To quantitatively analyze the degrees of hindgut rotation and extension simultaneously in a live embryo, we developed a novel procedure. The hindgut has a hook-like structure, and the peak of the elbow-shaped bend in this structure was defined to evaluate hindgut rotation (Fig 2A). The root of the hook-like structure is largely straight; therefore, the direction of the hindgut lumen is parallel to its rotation axis, which was defined as the pivot line (S1 Fig). Thus, as the hindgut rotates, the distance between the peak of the elbow-shaped bend and the pivot line increases (S1 Fig). The distance between the pivot line and the center of the hindgut lumen at the peak of the elbow-shaped bend was measured at 0 and 60 min (Fig 2A). The change in these values from 0 to 60 min was referred to as the rotational movement index. Moreover, to quantify hindgut elongation, a line passing through the most anterior peak of the hook-like structure and perpendicular to the pivot line was drawn. The distance between the line and the posterior end of the curved part in the hook-like structure was measured at 0 and 60 min (Figs 2C and S1). The ratio of the values at 60 min to those at 0 min was defined as the elongation rate.

Download:

• PNG
  larger image
• TIFF
  original image

Fig 2. E-cad and Myo1D perturbation did not affect hindgut elongation and cell intercalation.

(A–D) Quantification of hindgut deformation. (A) Schematic diagram of rotational movement index (indicated by a two-way arrow). (B) Boxplots of rotational movement index (μm) in the E-cad mutant (N = 10), wild type (N = 9), and wild type overexpressing Myo1D (N = 15). (C) Schematic diagram of elongation rate. The relative length of the hindgut at 60 min (indicated by a right two-way arrow) to its initial length (indicated by a left two-way arrow with “1”) was defined as the elongation rate. (D) Boxplots of elongation rate in the E-cad mutant (N = 10), wild type (N = 9), and wild type overexpressing Myo1D (N = 14). In b and d, black dots are outliers. **p < 0.01. *p < 0.05. (E) Schema of T3 to T1 transition. (F1–G3) Still shots from time-lapse Videos of cell nuclei (magenta) and boundaries (green) in the hindgut epithelium of the E-cad mutant (F1–F3) and wild type overexpressing Myo1D (G1–G3), visualized as described in Fig 1B and 1C, respectively. The time elapsed from the start of the Video is shown on the lower right. Yellow circles indicate cells next to each other in the first frame of the Video (F1–G3). A cell boundary showing T3 to T1 transition indicated by white arrowheads (F1–F3). The anterior is on top. (H, I) Frequencies of cell boundaries diminished during junction remodeling in the hindgut of the E-cad mutant (H) (n = 20, N = 4) and wild type overexpressing Myo1D (I) (n = 12, N = 5) with every 20-degree angle bin. (J, K) Angle change in the cell boundaries from 0 to 60 min. (J) Schemes for the analysis of cell boundary rotation. Initial cell boundaries (0 min) are shown in orange. Boundaries rotating counterclockwise (ccw) and clockwise (cw) are shown in red and blue, respectively. (K) Frequency of boundaries rotating ccw (red) and cw (blue) and remaining unchanged (uc, white) in the E-cad mutant (n = 50, N = 7), wild type (n = 55, N = 5), and wild type overexpressing Myo1D (n = 57, N = 5). n and N in parentheses represent cell and embryo numbers, respectively. ***p < 0.001. **p < 0.01.

https://doi.org/10.1371/journal.pgen.1011422.g002

Using this novel procedure, the rotation and elongation of the hindgut were compared among wild-type embryos, E-cad-mutant embryos, and Myo1D-overexpressing embryos in the hindgut epithelium. In the E-cad mutant, the rotational movement index was almost negligible. Moreover, the reduction of the value was significant compared with that in the wild type, which coincided with the elimination of cell sliding under this condition, as described above (Fig 2B). However, in these embryos, the elongation rate was largely comparable to that in the wild type (Fig 2D). Therefore, no correlation was noted between hindgut rotation and elongation. These values were also assessed in the hindgut of Myo1D-overexpressing embryos. Under this condition, the rotational movement index was significantly higher than that in the wild type, which coincided with the increase in cell sliding, as described above (Fig 2B). In contrast, the elongation rate did not differ markedly between the wild-type and Myo1D-overexpressing embryos, demonstrating the discordance between these two events (Fig 2D). These results suggest that E-cad and Myo1D affect cell sliding independently of convergent extension.

E-cad and Myo1D affected cell sliding independently of cell intercalation

Convergent extension is driven by cell intercalation, which is observed as a type of anisotropic junction remodeling, previously designated as T3 to T1 transition (Fig 2E) [4]. This transition drives these epithelial cells to intercalate with each other, resulting in the elongation of the hindgut epithelial tube [10]. Junctional remodeling has been considered unrelated to cell sliding or hindgut rotation because most cell boundaries were maintained during cell sliding, and the cell boundaries diminished by junctional remodeling did not exhibit LR bias [22]. Additionally, a simulation without junctional remodeling successfully recapitulated hindgut rotation [22]. Based on our aforementioned results, we speculated that cell intercalation is not affected in E-cad-mutant embryos or Myo1D-overexpressing embryos. To test this prediction, we examined the frequencies of the cell intercalation in the hindgut epithelium of these embryos. In the wild type, the cell intercalation was detected in 9.3% of examined cells in 0–60-min intervals, as reported previously (Table 1) [22]. In the E-cad mutant, in which cell sliding was largely abolished, the cell intercalation was still observed at a frequency of 12.7% in the same interval, supporting the independent regulation of these two cellular dynamics (shg^69B in Table 1 and Fig 2F). Furthermore, in the embryos overexpressing Myo1D in their hindgut, the cell intercalation was observed at a frequency of 7.8% in the same interval, which was largely comparable with that in the wild type (Table 1 and Fig 2G). These results suggest that the deceleration and acceleration of cell sliding do not affect the frequency of cell intercalation.

Download:

• PNG
  larger image
• TIFF
  original image

Table 1. The frequency of cell intercalation.

https://doi.org/10.1371/journal.pgen.1011422.t001

As the T3 to T1 transition may exclusively contribute to hindgut elongation, we predicted that this anisotropic junction remodeling should not exhibit an LR bias. Previously, we analyzed the angle θ between the diminishing cell boundaries and the vertical line to the pivot line in the wild type and reported that cell boundaries with an angle θ less than 10 degrees were selectively diminished, although the angle θ of diminishing cell boundaries did not exhibit a detectable LR bias in the wild type [22]. In this study, we analyzed the angle θ of diminishing cell boundaries in E-cad-mutant embryos and Myo1D-overexpressing embryos. Although cell sliding activity significantly differed in these embryos (Fig 1H), we did not observe a marked difference in the angle distribution of diminishing cell boundaries (Fig 2H and 2I). Furthermore, we did not detect an LR bias at these frequencies under these two conditions (Fig 2H and 2I). These results further suggest that cell sliding is not coupled with cell intercalation.

Although we did not observe marked changes in cell junction remodeling upon the modulation of cell sliding, we revealed that the chiral dynamics of cell junctions, detected as cell boundary rotation, were coupled with cell sliding but not with junction remodeling. As described above, we measured the angle θ of cell boundaries at 0 and 60 min and the rotational angle of each cell boundary from 0 to 60 min [22]. Then, we assessed the frequency of cell boundaries (%) that showed counterclockwise or clockwise rotation (Fig 2J and 2K) [22]. Compared with the wild type, we found that the LR bias of cell boundary rotation largely disappeared in the E-cad mutant (Fig 2K). In contrast, compared with the wild type, the LR bias of cell boundary rotation was augmented in Myo1D-overexpressing embryos (Fig 2K). Therefore, cell sliding correlated with cell boundary rotation but not with cell junction remodeling under these two genetic conditions. These results further suggest that cell sliding and cell intercalation are distinct epithelial dynamics and that E-cad and Myo1D preferentially contribute to the former.

Par-3/Bazooka (Baz) affected convergent extension but not cell sliding

We next analyzed whether genes affecting convergent extension also play roles in cell sliding. Lengyel et al. identified various mutants affecting the convergent extension of the Drosophila hindgut [17,18]. However, as these mutants, including bowl, lines, and drumstick mutants, showed severe defects in hindgut development, it was difficult to properly evaluate the rotation and elongation of the hindgut using our procedure (S2A–S2C Fig). In a recent study, a mutant of Par-3/baz, encoding Drosophila Par-3, showed defects in convergent extension during germband elongation [32]. In this study, we found that embryos homozygous for a classical allele of baz, baz⁴, a presumptive null mutant, showed elongation defects in the hindgut (Fig 3A and S7 Video) [33,34]. Approximately 25% of these embryos had obscure hindgut morphology (designated as deformation), which led us to exclude them from analyzing LR asymmetry (S2D Fig). However, in the remaining embryos (approximately 75%), the hindgut had a hook-like structure. Most of them showed normal dextral LR asymmetry, based on the curving direction, although they occasionally showed delayed germband retraction, as reported previously (S2D and S2E Fig). In the hindgut of these embryos, the x displacement (cell sliding) and rotational movement index were equivalent to those in the wild type, demonstrating normal hindgut rotation (Fig 3B, 3D and 3E and S8 Video). However, their elongation rate was significantly reduced in the Par-3/baz mutant (Fig 3C), which coincided with a severe decrease in the cell intercalation to a frequency of 2.0% compared with 9.3% in the wild type (Table 1 and Fig 3F). However, the observed reduction in hindgut elongation detected in our assay could be explained if the hindgut leaned ventrally. To exclude this possibility, we measured the depth of the hindgut along the dorsal-ventral axis (DV depth) at 0 and 60 min (S3A–S3D Fig). Our results demonstrated that DV depth was similar between wild type and Par-3 mutant hindgut, and the ratios at 0–60 min were equivalent for both genotypes (S3A, S3C and S3D Fig). Conversely, the DV depth of E-cad mutant differed from that of the wild type due to suppressed hindgut rotation (S3B Fig). These findings indicate that wild-type Par-3 regulates hindgut elongation through cell intercalation without affecting cell sliding and hindgut rotation. Based on these results, we speculate that cell sliding and cell intercalation are concurrently executed via two distinct and independent mechanisms in the hindgut epithelium.

Download:

• PNG
  larger image
• TIFF
  original image

Fig 3. Par-3 mutation impaired hindgut elongation.

(A1–A3) Still shots from a time-lapse Video of hindgut rotation in the Par-3 homozygote visualized as described in Fig 1A. (B1–B3) Displacement of nuclei in the hindgut epithelium of the Par-3 homozygote visualized as described in Fig 1D. (C1–C3) Still shots from a time-lapse Video of cell nuclei (magenta) and cell boundaries (green) in the Par-3-homozygote. Yellow circles indicate cells next to each other from 0 to 60 min. (D–F) Quantification of Par-3-mutant phenotypes. (D) Boxplots of x displacement in the wild type (WT) (n = 164, N = 7) and Par-3 homozygote (n = 296, N = 9). (E, F) Boxplots of rotational movement index (μm) (E) and elongation rate (F) in the WT (N = 9) and Par-3 homozygote (N = 7). In A1–C3, the anterior is on top, and the time elapsed from the start of the Video is shown on the lower right. In D–F, black dots are outliers. **p < 0.01.

https://doi.org/10.1371/journal.pgen.1011422.g003

Computer model recapitulated that cell sliding and convergent extension are independently but concurrently executed through contractile force with distinct polarities

Our analyses demonstrated that cell sliding and convergent extension are independently but concurrently regulated in the hindgut epithelium. To understand the underlying mechanisms by which these two epithelial dynamics are integrated, we performed numerical simulations using vertex dynamics models [35–38]. Previous studies have shown that cell sliding and convergent extension rely on two distinct contractions of cell boundaries; contractions with chirality and anterior–posterior/distal–proximal anisotropy are responsible for cell sliding and convergent extension, respectively [4,22]. Therefore, we hypothesized that these two contractions account for the rotation and extension of the hindgut, which occur independently of each other but simultaneously (Fig 4A–4C).

Download:

• PNG
  larger image
• TIFF
  original image

Fig 4. Simulations verified that cell sliding and cell intercalation are independently executed.

(A–C) Schemas showing that cell sliding (A) and cell intercalation (B) regulate hindgut rotation and elongation, respectively, and both together cause hindgut rotation and elongation simultaneously (C). (D, E) Schemas of the vertex dynamics model. (D) The model tube represents the apical surfaces of the hindgut epithelial cells, which are approximately 16 cells in circle and 28 cells in length. (E) Biased contraction, dependent on edge angle was introduced (indicated by two arrows). (F) To recapitulate cell sliding, initial cell chirality was formed by rightward-tilted-edge biased contraction and its subsequent dissolution were introduced (upper). To recapitulate cell intercalation, horizontally biased edge contraction (two arrows) was introduced, which occasionally resulted in the T3 to T1 transition of the edges (lower). (G–J) Simulations of gut deformation. Initial condition with cell chirality (G), and final shapes (H–J) of the model tube in 3D (upper) and 2D (lower) resulting from simulations with cell sliding only (H), cell intercalation only (I), and both simultaneously (J). Coordinate axes in 3D (x, y, and z; upper) and 2D (x and y; lower) are shown on the extreme left. Model cells forming a line at the front face in the initial phase (G) are labeled in blue and are chased until the end (H, I, J). (K, L, M) Quantification of rotation angle, elongation, and intercalation rates was performed in silico from 5 independent simulations for each condition. (K) Boxplots of rotation angle (degree). (L) Boxplots of elongation rate. (M) Boxplots of intercalation rate (%). Black dots are outliers. CS, model implementing only cell sliding; CE, model implementing only convergent extension; CS + CE, model implementing both cell sliding and convergent extension.

https://doi.org/10.1371/journal.pgen.1011422.g004

Our previous studies showed that introducing cell chirality by contracting rightward-tilted cell edges and subsequently dissolving this chirality can induce cell sliding, resulting in hindgut rotation [9,22]. In this study, we enhanced this model by incorporating the axial and radial expansion and contraction of the hindgut, which facilitates hindgut elongation (refer to Eq 1 in Materials and Methods). In brief, a regular cylinder with 448 regular hexagons (16 hexagons in circle and 28 hexagons in length; the area is 1) was created as a model tube representing the apical (luminal) surfaces corresponding to the straight region of the hindgut (Fig 4D). Anisotropic contraction of the cell boundaries was introduced as edge tension biased at oriented angles (Fig 4E). Cell chirality was represented by initial cell shape strain (Fig 4F top left, and 4G), induced by biased edge tension maximal at 45 degrees from the x-axis. As the simulation progressed, this tension was dissolved (Fig 4F top right) and a horizontal biased edge tension, maximal at 0 degrees was added (Fig 4F bottom).

In our current model, cell boundary contractions with chirality or anterior–posterior/distal–proximal anisotropy were separately introduced. The introduction and subsequent dissolution of chiral tension caused the model tube to rotate approximately 90 degrees counterclockwise, as viewed from the bottom of the tube (Fig 4G and 4H). Additionally, applying edge tension with anterior–posterior/distal–proximal anisotropy resulted in axial elongation of the model tube (Figs 4I and S4). Notably, integrating both tensions led to simultaneous tube rotation and elongation, effectively recapitulating hindgut morphogenesis in vivo (Figs 4J and S4).

The deformations of model tubes were quantified by averaging the results of five independent simulations for each condition. The incorporation of cell chirality with or without the anterior–posterior/distal–proximal anisotropic tension induced approximately 90 degrees of rotation, whereas the anterior–posterior/distal–proximal anisotropic tension alone induced almost no rotation (Fig 4K). Moreover, the anterior–posterior/distal–proximal anisotropic tension induced approximately 1.4-fold axial elongation with or without the incorporation of cell chirality, whereas cell chirality alone did not induce any axial elongation (Fig 4L). As the wild-type hindgut tube exhibited approximately 1.2-fold elongation for 1 h (Fig 2D) and 90-degree rotation for 2 h (Fig 1A) in vivo, our in silico models quantitatively agreed with the in vivo behaviors of the hindgut tube. In the wild-type hindgut, cell intercalation occurred at 9.3% of the cell boundaries every 30 min (Table 1). Consistent with these observations, the wild-type model exhibited an average cell intercalation rate of 35.4% over a rotation and elongation period of approximately 2 h in vivo (Fig 4M). These results further supported our hypothesis that contraction forces with distinct polarities cause two independent tissue deformations, which are integrated by occurring in the same developmental window.

MyoII contributed to both cell sliding and cell intercalation

Our in silico models predicted that contraction forces with distinct polarities, namely chirality and anterior–posterior/distal–proximal anisotropy, are crucial for the rotation and elongation of the hindgut. Although it is difficult to specifically inhibit only one of these two contraction forces, it is known that MyoII regulates the contractile force of cell boundaries in various epithelial dynamics [4,6,39]. Therefore, the validity of our in silico model could be evaluated by examining whether the disruption of MyoII-dependent contraction forces in the hindgut and in the in silico models has comparable consequences.

In Drosophila, the MyoII heavy chain (MHC) is encoded by zipper (zip). We performed live imaging of the embryonic hindgut homozygous for zip², a null allele of zip [40]. However, the zip²-homozygous embryo showed normal hindgut rotation and elongation, probably because the maternal supply of zip is sufficient for supporting the normal morphogenesis of the hindgut (S3A Fig) [41]. Hence, to further inhibit MyoII, we misexpressed a dominant-negative form of zip, DN-zip, in the hindgut epithelium of zip²-heterozygous embryos [42]. We noted LR inversion and bilateral phenotypes of the hindgut, suggesting that MHC is required for proper hindgut rotation (S3B and S3C Fig). We also examined a null mutant of spaghetti squash (sqh), which encodes the myosin regulatory light chain (MRLC) [43]. In embryos homozygous for sqh^AX3, a null allele of sqh [44], we noted LR inversion and no laterality phenotypes of the hindgut, suggesting that MRLC is also important for hindgut rotation (S3B Fig). However, the percentages of LR defects in these mutants were too low to conduct quantitative analyses at a cellular level, probably because of the residual activity of MyoII.

To solve this problem, we developed an ex vivo assay, which was derived from an ex vivo organ culture system reported previously [22]. In brief, we dissected the most posterior part of the embryo, which is mainly composed of the hindgut, before the initiation of its rotation [22]. In this new system, we treated the cultured hindgut with chemical inhibitors and quantitatively analyzed its morphological changes ex vivo (Fig 5). We found that the hindgut rotated and elongated ex vivo, as observed in vivo (S9 Video). Using this ex vivo system, we administered an inhibitor against MyoII, Y-27632, which blocks the MyoII activator Rho kinase, into the cultured hindgut (S10 Video) [45]. We assessed the rotational movement index ex vivo under this condition using the same procedures as those used in vivo. The rotational movement index was almost zero in the wild-type hindgut treated with Y-27632 ex vivo. However, the rotational movement index in the mock-treated hindgut (control) was equivalent to that in the wild type in vivo (Fig 5G). Therefore, hindgut rotation was diminished by the suppression of MyoII activity. We then analyzed the effect of Y-27632 treatment on cell sliding. In the mock-treated hindgut cultured ex vivo, the columns of nuclei lined up along the anterior–posterior axis at 0 min slanted to the left after 30 and 60 min, as noted in the wild type in vivo (Fig 1D1–1D3 and 5B1–3, S11 Video). However, the columns of nuclei in the wild-type hindgut treated with Y-27632 ex vivo did not change their direction for 60 min, demonstrating that cell sliding was abolished (Fig 5C1–5C3 and S12 Video). This result was quantitatively supported by our findings that x displacement was negligible in the wild-type hindgut treated with Y-27632 ex vivo, whereas x displacement in the mock-treated hindgut ex vivo was largely equivalent to that in the wild type in vivo (Fig 5F). These results suggest that the activity of MyoII is essential for cell sliding.

Download:

• PNG
  larger image
• TIFF
  original image

Fig 5. Inhibition of Myo II activity stopped the rotation and elongation of the hindgut.

(A) Schema of drug treatment ex vivo. Late-stage 12 embryos were dissected using a glass pipette needle (top), treated with a drug or DMSO (control) (middle), and encapsulated with a coverslip to perform time-lapse imaging (bottom). (B1–C3) Displacement of nuclei in the wild-type hindgut epithelium cultured ex vivo, visualized as described in Fig 1D. The cultured hindgut was treated with DMSO (B1–B3, control) or an inhibitor against MyoII, Y-27632 (C1–C3, Y-27632). (D1–E3) Still shots from a time-lapse Video of cell nuclei (magenta) and boundaries (green) in the hindgut cultured ex vivo, visualized as described in Fig 1A. The cultured hindgut was treated with DMSO (D1–D3, control) or an inhibitor against MyoII, Y-27632 (E1–E3, Y-27632). In B1–E3, the anterior is on top, and the time elapsed from the start of the Video is shown on the lower right. (F) Boxplots of x displacement in the wild-type hindgut cultured ex vivo with DMSO (control) (n = 155, N = 5) or with Y-27632 (Y-27632) (n = 232, N = 5). (G, H) Boxplots of rotational movement index (μm) (G) and elongation rate (H) in the wild-type hindgut cultured ex vivo with DMSO (control) (G, N = 8; H, N = 7) or with Y-27632 (Y-27632) (G, N = 9; H, N = 8). (I–K) Results of simulation in which edge tension was reduced after the initial introduction of chirality. (I) Final shape of the model tube in 3D (right) and 2D (left). Model cells forming a line at the front face in the initial phase are labeled in blue and are chased until the end. (J, K) Boxplots of rotation angle (degree) (J) and extension rate (K) in models implementing both cell sliding and convergent extension with normal (control) or reduced (low tension) edge tensions. In F, H, and K, black dots are outliers. ***p < 0.001. **p < 0.01.

https://doi.org/10.1371/journal.pgen.1011422.g005

Using the same ex vivo assay, we assessed the requirement of MyoII for hindgut elongation. We found that the elongation rate was close to one (no elongation) in the wild-type hindgut treated with Y-27632 ex vivo. However, the mock-treated hindgut elongated to a similar extent as the wild type in vivo (Fig 5H). These results suggest that the activity of MyoII is also essential for the convergent extension of the hindgut. Therefore, we predicted that T3 to T1 cell junction remodeling, which drives convergent extension, is impaired under this condition. We found that the T3 to T1 transition (cell intercalation) was severely inhibited by the addition of Y-27632 compared with that in the mock-treated wild type ex vivo (Table 2, Fig 5D and 5E). Collectively, our results suggest that MyoII-dependent activity, which induces contraction force along cell boundaries, is required for both cell sliding and convergent extension.

Download:

• PNG
  larger image
• TIFF
  original image

Table 2. The frequency of cell intercalation in ex vivo experiments.

https://doi.org/10.1371/journal.pgen.1011422.t002

We next validated these results in our in silico model. We concurrently diminished the dissolution of cell chirality and contractions of cell boundaries with anterior–posterior/distal–proximal anisotropy by reducing edge tensions in the model, although chirality was introduced before edge tension reduction (Fig 5I). We found that this model gut showed decreases in rotation and elongation (Fig 5I–5K), which is similar to the findings in the hindgut in which MyoII activity was inhibited ex vivo (Fig 5C, 5E, 5G and 5H). Thus, our in silico model qualitatively recapitulated the situations. Taken together, our results suggest that cell sliding and convergent extension are distinct epithelial dynamics induced by chiral and anterior–posterior/distal–proximal anisotropic contractions, which operate parallelly in the hindgut epithelium in the same developmental window.

Fly strains

The following fly strains were used: shg^R69 [27], zip² (BL8739) [40], sqh^AX3 (BL25712) [44], baz⁴ (BL3295) [33,34], byn-gal4 [56], NP2432-gal4 (Kyoto 104201), UAS-myrGFP-p10 (JFRC29) [30], UAS-RedStinger (BL8547), UAS-Myo31DF [25], UAS-DN-zip [42], and UAS-sqh-GFP [22]. They were raised at 18°C or 25°C. The genotypes used for live imaging of shg homozygotes were shg^R69/shg^R69, NP2432; UAS-RedStinger/UAS-myrGFP-p10 and those used for live imaging of baz homozygotes were baz⁴/baz⁴; UAS-RedStinger/byn-gal4, UAS-myrGFP-p10. For wild-type embryos overexpressing Myo31DF, the genotypes UAS-Myo31DF/+; UAS-RedStinger/byn-gal4, UAS-myrGFP-p10 were used to record time-lapse Videos. For the selection of shg^R69 and zip² mutants, Cyo-sChFP (BL35523) and Cyo-GFP (BL5195) balancers were used. The FM7c-GFP (BL5193) balancer was used for selecting the baz⁴ mutant.

Live imaging and quantification of parameters for characterizing hindgut deformation

Live imaging of the hindgut epithelium was performed as described previously [22]. Late-stage 12 embryos of the appropriate genotype were mounted with the dorsal side on top in oxygen-permeable halocarbon oil 27 (Sigma-Aldrich) using 0.17–0.25-mm-thick coverslips (thickness no. 2, Matsunami Glass) as spacers. Time-lapse Videos were recorded every 5 min for 2 h using a scanning laser confocal microscope (LSM700, Zeiss) at 23°C–25°C.

For quantifying cell displacement in these time-lapse Videos, centers of nuclei detected by the expression of UAS-RedStinger were manually defined. The root part of the hindgut, the region 20%–40% away from the bottom, and the central three columns of cells were analyzed to avoid extraneous effects of tissue movement (Fig 1D) [22]. To track the movement of nuclei, the depth of the z stack was adjusted at each time point to make the time-lapse Videos clearer because these nuclei changed their depth from the object glass over time. The x- and y-coordinates of nuclei in these time-lapse Videos were measured every 30 min until 60 min after the beginning of hindgut rotation using ImageJ software (https://imagej.net/ij/). The position of the subjacent nucleus was set at (0, 0) coordinates, and the relative displacement of the upper nucleus in the x direction was quantified every 30 min until 60 min after the beginning of hindgut rotation (Fig 1G). Welch’s t-test was used for statistical analysis.

To analyze the rotation of cell boundaries detected by the expression of UAS-myrGFP-p10 in the hindgut epithelium, the angle changes in the boundaries between cells aligned along the central column were measured every 30 min until 60 min after the beginning of hindgut rotation. The changes in these angles were then classified into counterclockwise, unchanged, and clockwise (Fig 2J), and the frequencies of each class were calculated. Statistical analysis was performed using χ² test.

To quantify the frequency of cell intercalation in hindgut epithelial cells, cell boundaries that were separated through the intercalation of other cells were counted every 30 min until 60 min after the beginning of hindgut rotation. This number was divided by the total number of examined cells, which was designated as the intercalation index. Moreover, the angle of cell boundaries diminished by cell intercalation against the anterior–posterior axis of the hindgut was measured, and the percentage of these cell boundaries with every 20-degree angle bin was calculated.

Hindgut rotation was quantitatively assessed as an increase in the distance between the peak of the elbow-shaped bend and the pivot line from 0 to 60 min, which was designated as the rotational movement index (S1 Fig). To quantify hindgut elongation, a line passing through the most anterior peak of the hook-like structure and perpendicular to the pivot line was drawn. The distance between the line and the posterior end of the hook-like structure was measured at 0 and 60 min. The ratio of the distance at 60 min to that at 0 min was designated as the elongation rate (Figs 2C and S1). The DV depth was measured as the distance between the most dorsal part of the gut lumen and the most anterior peak of the hook-like structure along the z (dorsal-ventral-axis) direction at 0 and 60 min (S1 and S3A–S3D Figs). Mann–Whitney U-test was used for statistical analysis.

ex vivo culture and drug treatment

The hindgut with the most posterior part of the embryo was dissected in M3 medium (S8398, Sigma-Aldrich) with 10% fetal bovine serum (FBS, Biowest) and 1% trehalose (34413–92, Nakarai) on double sticky tape placed on a glass slide. The dissected hindgut was covered with a coverslip with 0.17–0.25-mm-thick spacers and continuously cultured for 2 h at 23°C–25°C under a microscope (LSM700). To inhibit MyoII in the culture, Y-27632 (Y0503, Sigma-Aldrich), an inhibitor against MyoII, dissolved in water was added to the culture medium at a final concentration of 100 μM after the dissection. Time-lapse Videos were recorded and analyzed as described above.

Two-dimensional (2D) vertex model for hindgut extension and torsion

To analyze the effects of polarized contractions of cell boundaries, such as anterior–posterior/distal–proximal anisotropy and chirality, on hindgut deformation, we performed numerical simulations of epithelial torsion by enhancing a 2D vertex model.

In our model, the hindgut epithelium was assumed to be a deformable cylinder with variable perimeter and axial length represented by L_x and L_y, respectively. The multicellular dynamics within the cylinder were calculated by expanding them to those in a rectangular sheet of width L_x and length L_y in the xy-coordinates, where the radial and longitudinal axes of the cylinder correspond to the x- and y-axes of the sheet, respectively. Periodic boundary conditions were applied at boundaries x = L_x and y = L_y. To represent the longitudinal extension of the cylinder, the perimeter L_x and length L_y of the cylinder can vary depending on the tissue stress. Additionally, to represent the torsion of the cylinder, the x-coordinate can be shifted at the boundary from y = L_y to y = 0 with length L_yx according to the tissue stress.

The rectangular sheet forming the cylinder is densely lined with numerous cells. Each epithelial junction between neighboring cells was represented by an edge, and each cell shape was represented by a polygon containing the edges. The edges and vertices were shared with neighboring cells. Cell dynamics were calculated by tracking the motion of vertices, where the location vector of the ith vertex is represented by r_i. In addition, cell rearrangement was represented by reconnecting the edges when the length of the edges fell below a threshold value Δl, based on the motion of vertices. The properties of all cells were assumed to be the same, and cell growth and division were not considered [7]. For simplicity, we ignored the effects of epithelial thickness.

To combine cylindrical deformation with multicellular dynamics, we followed a sequential calculation process: first, we calculated the geometry of cells r_i, from which we obtained the stress tensor of the entire tissue, represented by σ_αβ. Using this stress tensor, we then calculated the geometry of the cylinder L_x and L_y. The iterative calculations of these processes allowed us to calculate the combined dynamics of the cylinder and cells.

Motion equation of a deformable tube

Hindgut torsion occurs independently of surrounding tissues [22]. Additionally, abnormalities in chirality and torsion can be rescued by expressing Myosin1D in the posterior intestinal epithelium [9,25]. These findings suggest that hindgut torsion is driven autonomously by actomyosin-dependent junctional contraction of the hindgut epithelium. Therefore, we focused on the effects of junctional contractions, disregarding external forces such as luminal pressure and other forces from the hindgut environment.

To describe the cylindrical extension, the motion on the size of the tube, L_x and L_y, was given by: (Eq 1)

In Eq (2), σ_αβ is a stress tensor of the system. Moreover, to describe the cylindrical torsion, the motion on the shift length L_yx was given by: (Eq 2)

The coefficient ξ/3 was given assuming that the hindgut is an isotropic material. Vertex positions were shifted in an affine manner according to the changes in L_y.

Motion equation of cells

The location vector of the ith vertex, represented by r_i, was given by: (Eq 3) where η is a friction coefficient and U is a potential energy. The potential energy U was given by: (Eq 4)

The first term describes the energy of area elasticity with the elastic coefficient K, for which A_i is the area of the ith cell and A_eq is the preferred area. The coefficient ζ(1−A_eq/A_i)² in the first term regulates the individual cell size, preventing excessive shrinkage or expansion, where ζ represents the constraint strength. The second term describes the energy of perimeter elasticity with the elastic coefficient Γ, for which P_i is the perimeter of the ith cell and P_eq is the preferred perimeter. This term reflects the summation of the contractility of the actomyosin ring of individual cells and cell–cell adhesion at junctions between cells. The third term describes the line energy of junctions with active line tension, Λ_j. This term represents both polarized contractions of cell boundaries, including anterior–posterior/distal–proximal anisotropy and chirality. The detailed implementation of these polarized contractions is described in the following simulation procedure. Here, L_j denotes the length of the jth edge. Line tensions can be changed by actin–myosin contractility. The line tension Λ_j, depending on the angles of edges, was given by: (Eq 5) where θ_j is the angle between the polar axis and the jth edge. The constant θ_ref is the angle of cell polarity that maximizes Λ_j (= Λ_α) when θ_j = θ_ref. The third term denotes the fluctuation force of actomyosin contractility with strength Λ_r, where w_α satisfies 〈w_α(t)〉 = 0 and , where δ_α is the Kronecker delta and τ_R is the correlation time of the noise as used previously in cell flow models [36,37].

Stress tensor under periodic boundary condition

The stress tensor under the periodic boundary condition was given by: (Eq 6) where A_i and are the area and stress tensor of the ith cell, respectively. The stress tensor of the ith cell was given by: (Eq 7) Here, the α-component of the distance vector from the center of the ith cell. is the α-component of the force of the jth vertex. The potential energy U was divided into those of individual cells as follows: (Eq 8) where the linear energy in the third term of Eq (4) is equally distributed between two adjacent cells. Assuming the quasistatic process, the force was given by: (Eq 9)

Simulation procedure

Simulations were performed via two processes, i.e., precalculation (0≤t<τ₁) and torsion (τ₁≤t≤τ₂). In both these processes, the boundary conditions of the x- and y-axes were set to periodic boundary conditions. During the precalculation, the angle of the polarized contraction of cell boundary was set to θ_ref = π/4 to represent the chirality causing chiral cell sliding. Additionally, during torsion, the angle was set to θ_ref = 0 to form the anterior–posterior/distal–proximal anisotropy of cell shapes, which drives convergent extension.

Nondimensional values and parameter setting

was set to unit length, KA_eq to unit energy, and η/K to unit time. All the parameters used are listed in Table 3 [22,37,38]. For the simulation using reduced tension, Γ and Λ_a were reduced by 0.01 holds and Λ_r and Λ_r were reduced by 0.1 holds in torsion.

Download:

• PNG
  larger image
• TIFF
  original image

Table 3. Physical parameters.

https://doi.org/10.1371/journal.pgen.1011422.t003

Evaluation functions

To evaluate tube deformation, the normal strains of the system along the x- and y-axes, represented by γ_xx and γ_yy, respectively, and the shear strain of the system, represented by γ_yx, were given by: (Eq 10)

The ratio of the rotation angle to the length of the hindgut, represented by r_θ, was given by: (Eq 11)

The intercalation rate was calculated using a column of cells located in the middle (part of blue-colored cells, 23–28 cells per tube, Fig 4G). This rate was determined by dividing the number of cell boundaries diminished by cell intercalation by the total number of observed cell boundaries.

Other mechanical factors possibly affecting hindgut extension and torsion

To concentrate on the effects of polarized contractions of cell boundaries, including anterior–posterior/distal–proximal anisotropy and chirality, our model made several simplifying assumptions. 1) The hindgut morphology was simplified to a cylinder. 2) Forces maintaining and deforming the hindgut were simplified to those from the apical junctions between cells, excluding external forces such as surrounding tissue and internal pressure. 3) Tissue viscosity was simplified to isotropic. While these simplifications help predict the effects of polarized contractions of cell boundaries, they may also affect hindgut extension and torsion. For instance, internal pressure, which counteracts epithelial extension, can delays the extension speed. Additionally, if tissue viscosity was anisotropic, it may either accelerate or decelerate the rate of extension and torsion.