Chirality in Planar Cell Shape Contributes to Left-Right Asymmetric Epithelial Morphogenesis

See allHide authors and affiliations

Science  15 Jul 2011:
Vol. 333, Issue 6040, pp. 339-341
DOI: 10.1126/science.1200940


Some organs in animals display left-right (LR) asymmetry. To better understand LR asymmetric morphogenesis in Drosophila, we studied LR directional rotation of the hindgut epithelial tube. Hindgut epithelial cells adopt a LR asymmetric (chiral) cell shape within their plane, and we refer to this cell behavior as planar cell-shape chirality (PCC). Drosophila E-cadherin (DE-Cad) is distributed to cell boundaries with LR asymmetry, which is responsible for the PCC formation. Myosin ID switches the LR polarity found in PCC and in DE-Cad distribution, which coincides with the direction of rotation. An in silico simulation showed that PCC is sufficient to induce the directional rotation of this tissue. Thus, the intrinsic chirality of epithelial cells in vivo is an underlying mechanism for LR asymmetric tissue morphogenesis.

Directional left-right (LR) asymmetry is widely found in animals, such as in the position and structure of the heart, spleen, gut, and lung in vertebrates (1). The mechanisms of LR axis formation are well understood in some vertebrates (1), and the cellular basis for LR symmetry breaking, including cell polarities, is beginning to be elucidated (2, 3). Drosophila shows a directional LR asymmetry of certain organs, including the embryonic hindgut (4, 5). Although some unique features of Drosophila laterality development have been revealed, such as the involvement of myosin ID (MyoID) (4, 5), the detailed mechanisms of its LR asymmetric development remain largely unknown.

The Drosophila embryonic hindgut begins as a symmetric midline structure that curves ventrally at stage 12 (Fig. 1A and fig. S1A). It subsequently makes a 90° left-handed rotation, forming a rightward curving structure by stage 13 (Fig. 1A) (6). The hindgut epithelium, but not the overlying visceral muscles, is responsible for this rotation, which is not accompanied by cell proliferation or cell death (6). Therefore, we speculated that the hindgut epithelial cells themselves might have LR polarity, which could contribute to the rotation.

Fig. 1

Polarization of centrosome positions in hindgut epithelial cells. (A) Schematic drawing of the left-handed rotation of the wild-type hindgut. (B) Centrosomes (green) visualized by UASp-GFP-cnn expression driven by NP2432. Cell boundaries were detected by antibody against PY20 (anti-PY20) (magenta). (C) Diagram representing the position of a centrosome relative to the cell centroid in two-dimensional coordinates (anterior-posterior and left-right). (D) Representative result showing the positions of centrosomes (blue dots) in a wild-type hindgut. (E) Percentages of centrosomes plotted on the four areas of (D). Bars show standard errors among the means of 10 embryos.

To analyze LR polarity in the hindgut epithelial cells, we examined the locations of the centrosomes, which reflect cell polarity in other systems (7, 8). We calculated each cell’s centroid with respect to its boundaries and plotted the relative position of the centrosome, labeled with green fluorescent protein (GFP)–centrosomin (Fig. 1, B and C). In wild-type animals, the relative position of the centrosome was significantly biased to the right-posterior region (Fig. 1, D and E). These results suggest that hindgut epithelial cells adopt a LR polarity within their plane before the hindgut rotates.

We speculated that this LR polarity would be reflected in the cell shape and participate directly in the left-handed rotation. To address this, we measured the angle between apical cell boundaries and the antero-posterior (AP) axis of the hindgut epithelial tube before rotation (late-stage 12) (x° in Fig. 2A). These apical cell boundaries corresponds to the zonula adherens (ZA). Cell-boundary angles of −90° to 0° to the AP axis were more frequent than those of 0° to 90°, indicating that hindgut epithelial cells have a LR-biased planar cell shape (Fig. 2B). We designated this LR bias as planar cell-shape chirality (PCC), because the mirror image of the cell’s planar shape does not overlap with its original cell shape.

Fig. 2

Requirement for MyoID and DE-Cad to form normal PCC in hindgut epithelial cells. (A) The angle between a cell boundary (magenta) labeled with anti-PY20 and the AP axis (yellow arrow) is defined as x° (turquoise arc). Images of cell boundaries at the dorsal apical surface of the hindgut epithelial tube were used for analysis. (B to H) Percentage of cell boundaries with angle x° in the ranges indicated at the bottom of (B). Genotypes of the analyzed embryos are indicated at top. At left, graphs show the means obtained from the number of embryos indicated (N), and error bars indicate standard errors among the means. At right, graphs show the percentage of cell boundaries (N) with angle x° at 15° intervals.

We previously demonstrated that the hindgut rotates right-handedly in embryos homozygous for Myo31DF, which encodes MyoID (4). In Myo31DFL152 homozygotes, the distribution of angle x° was reversed from that of wild type, although the LR bias became less prominent (compare Fig. 2, B and C). The reversed PCC in Myo31DFL152 was rescued by the overexpression of Myo31DFGFP (Fig. 2D). Rho family guanosine triphosphatases, including Rho1 and Rac1, regulate the organization of the actin cytoskeleton (9). We previously showed that overexpression of a dominant-negative Rho1 (Rho1.N19) or Rac1 (Rac1.N17) in the hindgut epithelium disrupts the hindgut’s LR asymmetry (9). We observed no PCC in these epithelial cells, suggesting that PCC formation depends on the actin cytoskeleton (Fig. 2, E and F). These results support our suggestion that PCC could determine the subsequent laterality of the hindgut.

To identify genes involved in PCC formation, we screened for mutations affecting LR asymmetry of the hindgut. We found that shotgun (shg) mutations (shgR758, shgR1232, and shgR69, a null allele) disrupted the laterality of the hindgut (fig. S1). shg encodes DE-Cad, a conserved transmembrane protein that mediates cell-cell adhesion in the epithelium (10). Genetic analyses suggested that DE-Cad functions downstream of MyoID, and both are required in the hindgut epithelium just before its rotation for normal LR asymmetric development (figs. S1D and S2). In shgR69 homozygotes, the angle x° did not demonstrate LR asymmetry, indicating that PCC was not formed in this mutant (compare Fig. 2, B and G). This PCC defect in shgR69 homozygotes was rescued by the overexpression of shgDECH (Fig. 2H).

To understand how DE-Cad contributes to PCC formation, we examined whether the distribution of DE-Cad showed LR polarity in hindgut epithelial cells. For this, we calculated the mean of DE-Cad’s relative intensity at the ZA of each cell boundary in the hindgut epithelium at late-stage 12. In wild type, the mean intensity was significantly greater at the cell boundaries with an angle x° of −90° to 0° than in those with 0 to 90° angles (Fig. 3, A and B), whereas this situation was reversed in Myo31DFL152 homozygotes (Fig. 3C). Rose diagrams depicting the intensity of DE-Cad in the cell boundaries bundled for 30° intervals showed that DE-Cad was enriched in cell boundaries with an angle x° of −90° to −30° (Fig. 3B′). Conversely, in Myo31DFL152 homozygotes, this situation was reversed (Fig. 3C′). This reversed bias was restored to the wild-type situation by overexpressing Myo31DFGFP in the hindgut epithelium (Fig. 3D).

Fig. 3

Polarized distribution of DE-Cad. (A) Hindgut epithelium stained with an antibody against DE-Cad (anti-DE-Cad). See Fig. 2A for angle x°. (B to D) Bar graphs showing the mean of the normalized DE-Cad fluorescent intensities at the cell boundary in the hindgut epithelium of the indicated genotypes representing the indicated ranges of angle x°. (B′ and C′) Rose diagrams showing the mean of normalized DE-Cad intensity in the hindgut epithelium of wild-type and Myo31DFL152 homozygous embryos at 30° intervals of angle x°. (E) Mosaic hindgut epithelium including Myo31DFmEGFP-expressing (+, green) and nonexpressing cells (−) in a Myo31DFL152 homozygote. Cell boundaries composed of +/+, +/−, and −/− cells are indicated in green, yellow, and magenta, respectively (right). (F to H) Bar graphs showing the mean of normalized DE-Cad intensities in +/+ (F), +/− (G), and −/− (H) cell boundaries. (F′ to H′) Schematic representation of cell boundaries composed of +/+ (green), +/− (yellow), and −/− (magenta) cells. Myo31DFmEGFP-expressing cells are shown in light blue. The number of cell boundaries and embryos analyzed are indicated as N = (Ncell boundary, Nembryo) in (B) to (D) and (F) to (H).

We then asked whether the LR bias of DE-Cad distribution was attributable to a cell-autonomous function of MyoID. To address this, we developed a new system (fig. S3) for generating a mosaic hindgut epithelium composed of Myo31DFL152 homozygous cells with (+) or without (−) Myo31DFmEGFP overexpression (Fig. 3E and fig. S3B). In the hindgut epithelium, the cell boundaries were classified into three types according to the cell type on either side (Fig. 3E): +/+, green; +/−, yellow; −/−, magenta. Cell boundaries of +/+ showed the wild-type LR bias of DE-Cad localization, which was reversed in the −/− boundaries (Fig. 3, F and H). The +/− cell boundaries did not show a statistically significant LR bias (Fig. 3G). Thus, the LR asymmetry of DE-Cad distribution at each cell boundary is attributable to the concordance of LR polarity in two adjacent cells.

To gain insight into how MyoID reverses the LR asymmetric distribution of DE-Cad, we looked for defects in endocytic trafficking, because DE-Cad’s localization to the ZA is controlled by its recycling (11). Rab11-positive recycling endosomes became fewer in the apical-middle part of cells in Myo31DFL152 homozygotes compared with wild type, and this defect was restored by Myo31DFGFP expression (fig. S4). These results may suggest that MyoID is involved in the recycling of DE-Cad.

Besides the angle x°, we also measured the length of cell boundaries at the ZA of the hindgut epithelial cells (fig. S5). In wild-type animals, the cell boundaries gradually expanded from late-stage 12 to late-stage 13 (fig. S5G). In homozygous Myo31DFL152 or shgR69 embryos, the length was greater than in wild type at all stages examined (fig. S5, B, E, and G). This increase was rescued by the overexpression of Myo31DFGFP or shgDECH in the respective mutant background (fig. S5, C, F, and G). Thus, DE-Cad and MyoID appear to restrict the expansion of these cell boundaries, suggesting that these proteins introduce cortical tension, possibly with LR asymmetry.

To evaluate our idea that PCC is involved in the hindgut LR asymmetric development, we built an in silico simulation model of the PCC of the hindgut epithelial cells and the directional rotation of the tube composed of these cells [see supporting online material (SOM) text for details] (figs. S6 to S9 and table S1). This model consisted of two epithelial sheets composed of model cells, forming the dorsal and ventral arcs of a tube with boundary cells separating the sheets, as found in vivo (Fig. 4A). In this simulation, the number of cells along the AP and LR sides was set to mimic the in vivo situation, and a statistical LR shape bias was not introduced initially (Fig. 4D). In vivo, DE-Cad was enriched at cell boundaries with an angle x of −90° to 0° and might restrict the cell-boundary expansion (Fig. 3, B and B′, and fig. S5G). Therefore, in this simulation, the constriction of cell boundaries was maximized at −45° to the AP axis of the hindgut epithelial tube, and the maximized value was twofold greater than at −135° or +45° (Fig. 4, B and D′). This parameter introduced PCC in the modeled epithelial cells (corresponding to late-stage 12) (Fig. 4, C and D′).

Fig. 4

Computer simulation suggesting that PCC may account for the left-handed rotation of the hindgut epithelial tube. (A) Diagram of the hindgut tube composed of epithelium (green) and two rows of boundary cells (orange). The direction of tube rotation is indicated by the curved arrow. (B) Diagram of the LR-biased constriction of cell boundaries (green). Arrows (magenta) show the constriction, whose magnitude corresponds to the width of the arrows (maximum at an angle x° of −45°). (C) Percentage of modeled cell boundaries formed in silico with angle x° of the ranges indicated at the bottom. (D to D″) Outputs of the simulation program. Green and red polygons are epithelial cells and boundary cells, respectively. Insets show higher-magnification images of framed areas. Steps in silico and the corresponding stages in vivo are indicated at the bottom of each panel.

Because DE-Cad and MyoID were required before but not during the left-handed rotation of the hindgut epithelium, we did not add a LR bias to the cell-boundary constriction during the epithelial remodeling (simulation from the state of Fig. 4D′ to Fig. 4D″). The removal of LR bias subsequently led the modeled epithelial cells to assume stable cell shapes that were mostly regular hexagons (corresponding to late-stage 13) (Fig. 4D″). This progressive transition in cell shape was also observed in vivo (fig. S10). This simulation reproduced the 90° left-handed rotation of the epithelial tube in silico, suggesting that PCC is sufficient to explain this rotation in vivo (Fig. 4, D′ and D″, and movie S1). In addition, LR asymmetric changes in the cell-boundary length observed in the hindgut epithelium in vivo were recapitulated in the simulation, supporting the validity of our model (fig. S11).

We report PCC as a previously unknown mechanism of LR asymmetric morphogenesis. Various mutants of genes encoding the core components of planar cell polarity (PCP), a well-understood type of epithelial planar polarity, did not affect the laterality of the Drosophila embryonic gut (table S2), suggesting that PCC is not simply a variant of PCP. Although the importance of single-cell chirality has not been studied in multicellular organisms in vivo, intrinsic cell chirality has been found in the LR-polarized protrusion of neutrophil-like cells in vitro (8). Therefore, cell chirality may be a general property of animal cells. Our findings demonstrate a contribution of such chirality to LR asymmetric morphogenesis.

Supporting Online Material

Materials and Methods

SOM Text

Figs. S1 to S11

Tables S1 and S2

References (1235)

Movie S1

References and Notes

  1. Acknowledgments: We thank U. Tepass, S. Noselli, H. Oda, P. Adler, the Drosophila Genetic Resource Center at Kyoto Institute of Technology, and the Bloomington Drosophila Stock Center at Indiana University for flies and A. Nakamura and the Developmental Studies Hybridoma Bank for antibodies.
View Abstract

Stay Connected to Science

Navigate This Article