Wound healing recapitulates morphogenesis in Drosophila embryos

brief communications

Wound healing recapitulates morphogenesis in Drosophila embryos William Wood*, Antonio Jacinto*§, Richard Grose*¶, Sarah Woolner*, Jonathan Gale†, Clive Wilson‡ and Paul Martin*# *Department of Anatomy and Developmental Biology, University College London, Gower Street, London WC1E 6BT, UK †Department of Physiology, University College London, Gower Street, London WC1E 6BT, UK ‡Department of Human Anatomy and Genetics, University of Oxford, South Parks Rd, Oxford, OX1 3QX, UK §Current address — Instituto Gulbenkian de Ciência, Rua da Quinta Grande, 6, 2780-156 Oeiras, Portugal ¶Current address — Cancer Research UK, Lincolns Inn Fields, London, WC2 3PX, UK #e-mail: [email protected]

Published online: 28 October 2002; DOI: 10.1038/ncb875

The capacity to repair a wound is a fundamental survival mechanism that is activated at any site of damage throughout embryonic and adult life1. To study the cell biology and genetics of this process, we have developed a wounding model in Drosophila melanogaster embryos that allows live imaging of rearrangements and changes in cell shape, and of the cytoskeletal machinery that draws closed an in vivo wound. Using embryos expressing green fluorescent protein (GFP) fusion proteins, we show that two cytoskeletal-dependent elements — an actin cable and dynamic filopodial/lamellipodial protrusions — are expressed by epithelial cells at the wound edge and are pivotal for repair. Modulating the activities of the small GTPases Rho and Cdc42 demonstrates that these actin-dependent elements have differing cellular functions, but that either alone can drive wound closure. The actin cable operates as a ‘purse-string’ to draw the hole closed, whereas filopodia are essential for the final ‘knitting’ together of epithelial cells at the end of repair. Our data suggest a more complex model for epithelial repair than previously envisaged and highlight remarkable similarities with the well-characterized morphogenetic movement of dorsal closure in Drosophila.

A

t the end of Drosophila embryogenesis, a naturally occurring epithelial hole — the result of germ band retraction — is sealed by the morphogenetic movement of dorsal closure, which, at a gross level, resembles re-epithelialization of a wound2. Lateral epithelium from the two sides of the embryo is drawn up and over the extra-embryonic amnioserosa to form a neat, and subsequently invisible, midline seam where the two epithelial edges meet3–5. Live studies of dorsal closure in embryos expressing GFP–actin reveal an actin cable, and dynamic filopodia and lamellipodia expressed by leading-edge epithelial cells6. Blocking assembly of the actin-rich filopodial protrusions by modulating Cdc42 activity results in a failure of the zipping together of opposing epithelial sheets6. In addition, modulation of the small GTPase Rho to cause disassembly of the actin cable demonstrates that the cable is contractile and operates both as a purse-string and to restrain cells so that the epithelial advance is uniform7–10. Studies of wound healing in the adult mammalian cornea, in embryonic skin and in tissue culture monolayers show that an actin cable also rapidly assembles in epithelial wound-edge cells and suggest that it too may operate like a purse-string to draw the epithelial wound hole closed11–13. Indeed, inactivation of Rho signalling by loading wound-edge cells in the chick embryo with the Rho

blocker, C3 transferase, prevents cable assembly and results in severely impaired healing14. However, in the chick wound model, live analysis of actin dynamics is not possible and neither is a genetic approach to dissect the precise functions of each element of the actin machinery. To study the epithelial movements of wound healing and directly compare them with dorsal closure in the Drosophila embryo, we made wounds to the epithelium of embryos at dorsal closure stages, or shortly thereafter at stage 15. Wounds were made either by laser ablation of circular epithelial patches on the ventral aspect of the embryo or by manual wounding with a microinjection needle. The subsequent healing process was visualized by time-lapse confocal microscopy. Dorsal closure proceeds during a period when cell division is not naturally occurring in the embryonic epithelium15. Our live studies of wounded embryos found no activation of cell division at the epithelial margin (Fig. 1), so wound closure must also be achieved entirely by changes in cell shape and rearrangements. By ablating circular patches of epithelium with a laser (as previously described for analysis of tissue tensions during dorsal closure16) we were able to reproducibly create wounds of approximately five cell diameters across, which took on average just over 2 h to close (136 ± 11 min; n = 6). To study the changes in cell shape that occur in the wound-edge epithelium during the repair process, we performed live confocal time-lapse analysis of wounded embryos expressing a GFP–αcatenin fusion protein under the control of the epithelial driver e22c-GAL4. We observed that the cells around the wound margin underwent marked alterations in shape as the wound closed (Fig. 1a–d). The front-row cells constricted their leading edges and elongated as the wound drew closed. Occasionally, cells withdrew from the front row and became accommodated in rows further back so that the number of cells with an edge forming part of the wound circumference decreased as the hole closed (Fig. 1a–d). Furthermore, during dorsal closure, leading-edge cells exhibited polarized constriction and marked elongation, and were progressively lost from the leading edge as cells zipped together at anterior and posterior ends of the dorsal hole (Fig. 1e, f). We wondered whether wound-edge epithelial cells used similar actin machinery to epithelial cells of the dorsal closure leading edge, which, unlike the naked wound margin, is continuous with its underlying amnioserosal substratum. During dorsal closure, leading-edge cells assemble an actin cable, which extends the circumference of the epithelial margin10,17. A similar actin cable was identified in laser-generated epithelial wounds stained with fluorescently tagged phalloidin to visualize filamentous actin (Fig. 2a). The cable

NATURE CELL BIOLOGY ADVANCE ONLINE PUBLICATION www.nature.com/naturecellbiology

©2003 Nature Publishing Group

1

brief communications a

c

e

b

b

c

d

e

f

d

f

Figure 1 Changes in cell shape and rearrangements during wound closure. a–d, Images were taken at approximately 30-min intervals from a movie of a laser wound made to the ventral epithelial surface of an α-catenin–GFP Drosophila embryo. As re-epithelialization proceeds, the free margins of all wound-edge cells shrink (for example, cell indicated in red) and occasional cells withdraw completely from the wound margin (for example, blue cell), so that fewer and fewer cells contribute to the wound circumference (broken white line). No cells are lost or gained in the wound epithelium, indicating that no cell division or cell death occurs during the repair process. e, f, Images from a movie of the ‘zipping’ phase of dorsal closure. These images also show that cells leave the free epithelial margin (for example, blue cell), as hole closure proceeds. Arrows indicate α-catenin-rich adherens junctions linking intercellular segments of the actin cable. Scale bars represent 10 µm in a–d and 20 µm in e and f.

assembled rapidly, within minutes of wounding, and non-muscle myosin colocalized within leading-edge cells, as observed in GFP–Spaghetti-squash (GFP–sqh) embryos (sqh encodes nonmuscle myosin ATPase; Fig. 2b). Intercellular segments of the cable were anchored into adherens junctions where neighbouring cells abut one another, as revealed in wounded GFP-α-catenin-expressing embryos (Fig. 2c). To visualize live actin dynamics during wound closure, we used the e22c-GAL4 driver to express a GFP–actin fusion protein in all epithelial cells. Time-lapse confocal movies demonstrated that leading-edge cells assemble not only an actin cable, but also dynamic filopodia that extend up to 5 µm in length (Fig. 2d; see also Supplementary Information, Movie 1). Assembly of both these actin-containing structures occurred within minutes of laser wounding. The same actin structures were also observed after 2

a

Figure 2 The cytoskeletal machinery of wound closure. a, A fixed Alexa594–phalloidin-stained wound visualizes a dense cable of actin that encompasses the full circumference of the wound. b, Similarly, non-muscle myosin localizes to the leading edge of epithelial cells at the wound margin, as seen in sqh–GFP embryos. c, Wounded α-catenin–GFP embryos have bright spots of positive staining where neighbouring cells abut one another around the wound circumference (arrowheads). These sites probably represent adherens junctions, where intercellular segments of the actin cable are inserted. d, An image from a live GFP–actin wound, revealing both the actin cable and filopodial protrusions (arrows) extending from leading-edge cells. e, A TEM section through a repairing embryo wound. The region shown is magnified from the boxed region in the inset. The two epithelial fronts (arrows) push the intervening wound debris (bracket) ahead of them as they advance towards one another over the wound substratum (WS). f, A higher-magnification TEM showing filopodial protrusions (arrows) from the leading-edge epithelial cell (white asterisk). Adjacent wound debris is indicated with black asterisks. Scale bars represent 5 µm in a–e and 1 µm in f.

mechanical wounding of the epithelium with a microinjection pipette (Fig. 3). The wound filopodia seemed to have similar dynamic properties to their dorsal closure counterparts, extending and retracting at up to 1 µm min−1, as well as sweeping from side to side as if sampling the substratum ahead of them. Transmission electron microscopy (TEM) demonstrated that filopodia make contacts with the underlying wound substratum and wound debris ahead of them, but also demonstrated that filopodia apparently flail free as the epithelium advances (Fig. 2e, f). During the adhesion phase of repair, just as during dorsal closure, filopodia from opposing epithelial cells made contact and the wound was subsequently tugged closed (Fig. 3). We observed few lamellipodial protrusions until this final phase, when the epithelial fronts make contact with one another. Wound topology seems to be important in determining precisely the mode of re-epithelialization. In mechanically created incisional wounds, in which opposing epithelial fronts are within filopodial reach of one another (4 out of 17 mechanical wounds observed), we consistently observed lamellipodia formation and zipping together of wound edges from the outset (see Supplementary Information, Movie 2). To test the function of the actin cable in the repair process, we




b

c

d

Figure 3 Filopodial interactions during repair. Images were captured at approximately 20-min intervals from a repairing wound created mechanically in an actinGFP embryo. Note how filopodial projections reach across from opposing epithelial faces (arrow) and precede the tugging together of the two faces at these points. Scale bar represents 10 µm.

wounded Drosophila embryos that were mutant for Rho1 (ref. 18), as Rho1 has been shown to mediate the formation of cable-like actin stress fibres in tissue-culture fibroblasts19 and the actin cable during Drosophila dorsal closure7,10. To minimize contributions from maternal Rho, we made laser wounds in the ventral epithelium at a stage when dorsal closure was almost complete. Indeed, when embryos were stained with an antiDRho1 monoclonal antibody, no Rho protein was detected from the mid-dorsal closure stages onwards (Magie, C. and Parkhurst, S., personal communication; also see insets in Fig. 4d, e). We observed that Rho mutant embryos failed to assemble a continuous actin cable in wound-edge cells and consequently there was negligible, if any, initial contraction of the leading edge of these cells. However, the leading-edge of these cells did extend filopodia (Fig. 4a, b). Moreover, filopodia in mutant embryos were longer (extending up to 12 µm) and approximately three times more common than in wild-type embryos (Fig. 4d, e). In many instances, several filopodia coalesced to form a lamellipodium. These data support a model for actin cable function during dorsal closure in which the cable functions both as a purse-string and as a restraint on filopodial number and activity10. Surprisingly, although Rho mutant embryos are unable to assemble an actin cable, they are able to close their wounds, suggesting that purse-string contraction is not the only means by which a wound hole can close. Laser wounds made in Rho mutant embryos take on average almost twice as long to repair as their wild-type counterparts (247 ± 30 min, n = 5; Fig. 4a–c and Supplementary Information, Movie 4). Quantitative video analysis of such wounds identified a lag phase of nearly 2 h (the time in which an equivalent wild-type wound can close fully) before the disorganized leading edge begins to move forward significantly. During this initial period, no obvious changes in cell shape occur in the leading-edge epithelial cells. However, once forward movement begins, the wound closes at a rate that is not significantly different to that of a wild-type wound (7.0 ± 1.9 µm2 min−1; n = 6 in the

mutant compared with 9.0 ± 2.6 µm2 min−1; n = 5 in the wild type; Fig. 4c). We wondered how the epithelium draws forward in the absence of an actin cable. At various points around the Rho mutant wound margin, we observed interactions between the filopodia and lamellipodia of neighbouring cells (Fig. 4f). Movie analysis demonstrated that lamellipodia from adjacent leading-edge cells apparently tug on one another, essentially resulting in the formation of several local zipping fronts around the wound margin (Fig. 4g). Together, these fronts seem to exert sufficient driving force to close the wound. In wild-type wound healing, such interactions were only observed in the last moments of wound closure, when opposing epithelial fronts are driven close enough together (through purse-string contraction) to allow their actin protrusions to make contact. To test the function of filopodia, we modulated the levels of Cdc42, as this small GTPase has been shown to regulate filopodial assembly in fibroblasts20 and during dorsal closure6. However, Cdc42 mutant embryos or those expressing dominant-negative Cdc42 throughout the epithelium undergo developmental arrest long before the completion of dorsal closure21, so we wounded fly embryos expressing dominant-negative Cdc42N17 and GFP–actin under the control of the engrailed-GAL4 driver. These embryos complete dorsal closure but fail to survive to larval stages. We made small laser wounds (approximately two cell diameters in width) to patches of dominant-negative Cdc42-expressing epidermis in actin–GFP-labelled stripes of tissue and made movies of the subsequent repair process. These wounds assembled an actin cable in leading-edge cells with a similar time-course to that of their wildtype counterparts. However, no filopodial or lamellipodial protrusions extended from any of the leading-edge cells throughout the wound closure period (see Supplementary Information, Movie 5). At 30 min after wounding, the hole was almost closed (Fig. 5b) but, unlike wild-type epithelium, which undergoes this final closure phase in less than 15 min, the mutants were consistently unable to seal the remaining tiny hole closed (Fig. 5a, b); even 2 h after wounding, the hole was not fully repaired (see Supplementary Information, Movie 5). Finally, we were able to address the potential role of Rac, the third of the paradigm actin regulatory small GTPases, because of a recently constructed triple mutant of the three Drosophila Rac genes, Drac1, Drac2 and Mtl22. Although no antibodies are currently available to test whether maternal Rac levels had diminished, our wounds were made long after protein levels of the related small GTPase Rho had diminished to an undetectable degree in Rho mutants. However, laser wounds to triple Rac mutant embryos, in which dorsal closure was clearly perturbed by a lack of Rac activity, showed no significant retardation in wound closure (data not shown). This is consistent with our earlier observations in chick embryos, where Rac activity was inhibited by loading wound-edge cells with dominant-negative Rac14. In this study, no alteration of re-epithelialization rate or sealing efficiency was detected. Our studies in Drosophila embryos have allowed us to highlight the actin dynamics of epithelial repair in a genetically tractable in vivo wound-healing model. We show how a complex arrangement of interdependent actin elements, rather than a simple purse-string, operate together to close the wound. Although a contractile actin cable is normally pivotal in closing the epithelial hole, genetic ablation of this structure identifies redundancy in the cytoskeletal machinery, such that through filopodial/lamellipodial zipping, the hole can also be closed by the localized tugging of neighbouring epithelial cells on one another. Our data also clearly demonstrate the importance of filopodia in the final knitting together of epithelial sheets at the termination of wound healing. Preliminary data from our laboratory suggests that similar actin machinery is assembled at later stages of Drosophila development (unpublished observation); for example, after wounding tissues of the wing imaginal disc.



3

brief communications

a

0 min

60 min

120 min

Wound area (µm2)

c 900 800 700 600 500 400 300 200 100 0

Rho mutant

WT

0

50

100 150 200 Time (min)

250

WT 240 240min min

b

rho1

d

WT

rho1

g

f

Figure 4 Analysis of wound repair in Rho mutant embryos. Images from movies of equivalent wounds as they repair in wild-type (a) and Rho mutant (b) embryos that each express actin–GFP. The elapsed time in minutes is displayed in the top right corner. c, Graph of wound area against time plotted at 4-min intervals from the same movie data as in a and b. Note the 2-h lag phase before commencement of significant closure in the mutant embryo. d, A high-magnification view of a representative wild-type wound, revealing a typical actin cable and filopodial protrusions (arrows). The inset shows a wild-type embryo stained with anti-DRho1 antibody. High levels of Rho protein are observed throughout the epithelium at the developmental stage when wounds are made. e, In a Rho mutant embryo, an equivalent wound edge to that shown in d has little or no cable. Instead, numerous

4

e

filopodia appear to coalesce to form lamellipodia (arrows). The inset shows a Rho mutant embryo stained with anti-Rho1 antibody. No maternal Rho1 protein is detected in the embryo at the time of wounding (compare with the inset in d). f, Images taken at 2-min intervals from a Rho mutant actin–GFP wound showing how lamellipodia from neighbouring cells (arrows) interact and seem to aid in the zipping together of these cells. g, A fixed Alexa–phalloidin-stained Rho mutant wound lacks both an actin cable and leading edge cell constrictions. However, several regions of intense actin staining are detected where local epithelial zips seem to be in operation (arrows). Scale bars represent 10 µm in a and b, 5 µm in d and e, 5 µm in f and 10 µm in g.




0 min

3 min

6 min

9 min

12 min

b

Cdc42N17

Figure 5 Analysis of repair in dominant-negative Cdc42-expressing wounds. a, The last 12 min of wound closure in an embryo expressing actin–GFP in engrailed stripes, but that is otherwise wild-type. Elapsed time in minutes is displayed in the top-right corner. b, An equivalent wound in an embryo expressing

Cdc42N17 in engrailed stripes. Leading-edge cells in these stripes do not express filopodia. Such a wound closes to this almost final stage, but no further. Even after a further 2 h, the wound edges fail to knit together (see Supplementary Information, Movie 5). Scale bar represents 10 µm.

Furthermore, our previous studies in chick and mouse embryos (reviewed in ref. 2) suggest that this is also true for higher vertebrates. The parallels shown here between epithelial wound healing and morphogenetic closure of holes, together with the observations that adult and larval wounds in Drosophila activate the same Jun N-terminal kinase (JNK) signalling pathway as embryonic dorsal closure (ref. 23; W.W., A.J. and P.M., unpublished observations), suggest that the embryo may be a valuable source of further clues to enhance our understanding of the repair process. These observations may be invaluable in uncovering therapeutic strategies for enhancing wound healing in the clinic.

TEM and resin histology

Methods Fly stocks GFP–actin and α-catenin–GFP were expressed in the embryonic epidermis using the Gal4-UAS system24. w1118;UAS-GFP–actin25 or w1118;UAS-α-catenin-GFP26 stocks were crossed with y w1118; e22cGAL4 to express the GFP fusion proteins throughout the epidermis. Co-expression of dominant-negative Cdc42 and GFP–actin was achieved by crossing w1118; UAS-Cdc42N17 (ref. 27) flies to a stock carrying both en-GAL4 and UAS-GFP–actin on a recombined chromosome. GFP–actin was visualized in a Rho mutant background by crossing flies carrying both e22c-Gal4 and rhoA72R to a rhoA72O, UASGFP–actin stock10,18, and in a zygotic rac1, rac2, mtl triple mutant22 by crossing e22c-Gal4/Cyo;racJ10 rac2 mtl/ TM6 flies with a UAS-GFP–actin; racJ10 rac2 mtl/ TM6 stock. Homozygous rac1 rac2 mtl mutants from this cross were identified by their dorsal closure phenotype. The sqh-GFP transgenic stock was obtained from Roger Karess28, and OrR stocks were used as wild-type controls for the Rho antibody staining experiment.

Wounding and imaging of embryos Embryos were collected during stage 15 of development, hand-dechorionated with forceps and mounted in Voltalef oil under a coverslip before being subjected to laser ablation from a nitrogen laser-pumped dye laser (model no. VSL-337ND-S; Laser Science Inc., Franklin, MA) connected to a Zeiss Axioplan 2 microscope using the Micropoint system (Photonic Science, Arlington Heights, IL). Embryos for live analysis were then imaged using a Leica TCS SP confocal system (Milton Keynes, UK). Images were compiled from four confocal optical sections (each averaged two times) and collected at one of three time-intervals (30 s, 1 min or 2 min). The time-lapse series were assembled and analysed using ImageJ imaging software. Embryos to be wounded mechanically were hand dechorionated with forceps, stuck to double-sided sticky tape on a slide and covered with Voltalef oil before being wounded using a microinjection pipette. A coverslip was then placed on top and live analysis was carried out as described above. Embryos seemed largely unharmed by the wounding and imaging process and, if left unperturbed, will generally survive until pupal stages, although they rarely hatch.

Phalloidin staining of filamentous actin Embryos for staining with phalloidin were first wounded as described above before fixation for 30 min at room temperature in a 1:1 mix of heptane and 8% formaldehyde. After fixation, embryos were hand-devitellinized in PBS, blocked in 1% BSA/0.1% Triton in PBS for 30 min, incubated in 1 µg ml−1 Alexa594–phalloidin for 30 min, washed three times for 15 min in PBS containing 0.1% Triton and analysed on the confocal microscope.

Embryos for TEM were wounded in the usual way and fixed in 1:1 mix of heptane and half-strength Karnovsky fixative29 containing 1 µg ml−1 phalloidin for 15 min before being hand-devitellinized and refixed in the same fixative overnight. Next, embryos were rinsed in 0.1 M sodium cacodylate, postfixed in 1% osmium tetroxide in 0.1 M sodium cacodylate and rinsed in water before being incubated in 2% uranyl acetate in water for 15 min. Embryos were washed further in water and then dehydrated through a graded ethanol series. Specimens were finally embedded in agar resin mix and ultrathin sections were cut and examined on a Jeol 1010 transmission electron microscope (Peabody, MA).

Antibody staining Embryos for antibody staining were dechorionated before being fixed for 20 min in a 1:1 mix of heptane and 4% paraformaldehyde in PBS. Specimens were then hand-devitellinized before being incubated at room temperature for 4 h in PAT (1% BSA and 0.1% Triton X-100 in PBS) and then transferred to fresh PAT containing anti-Rho1 P1D9 monoclonal antiserum (gift from C. Magie, Fred Hutchinson Cancer Research Centre, Seattle) at a 1:50 concentration and rolled overnight at 4 °C. Samples were then washed several times in PBT (0.1% Triton X-100 in PBS) and incubated for 3 h at room temperature in fresh PAT containing rhodamine-conjugated anti-mouse IgG secondary antibody (Jackson Labs, West Grove, PA) at a dilution of 1:200. After several washes in PBT, embryos were mounted on slides using Citiflour and visualized by confocal microscopy. RECEIVED 17 MAY 2002; REVISED 22 AUGUST 2002; ACCEPTED 30 SEPTEMBER 2002; PUBLISHED 28 OCTOBER 2002.

1. Martin, P. Science 276, 75–81 (1997). 2. Jacinto, A., Martinez-Arias, A. & Martin, P. Nature Cell Biol. 3, E117–E123 (2001). 3. Martinez-Arias, A. in The Development of Drosophila melanogaster (eds Bate, M. and MartinezArias, A.) 517–607 (Cold Spring Harbor Laboratory Press, New York, 1993). 4. Jacinto, A., Woolner, S. & Martin, P. Dev. Cell 3, 9–19 (2002). 5. Harden, N. Differentiation 70, 181–203 (2002). 6. Jacinto, A. et al. Curr. Biol 10, 1420–1426 (2000). 7. Magie, C. R., Meyer, M. R., Gorsuch, M. S. & Parkhurst, S. M. Development 126, 5353–5364 (1999). 8. Harden, N., Ricos, M., Ong, Y. M., Chia, W. & Lim, L. J. Cell Sci. 112, 273–284 (1999). 9. Bloor, J. W. & Kiehart, D. P. Development 129, 3173–3183 (2002). 10. Jacinto, A. et al. Curr. Biol. 12, 1245–1250 (2002). 11. Martin, P. & Lewis, J. Nature 360, 179–183 (1992). 12. Bement, W. M., Forscher, P. & Mooseker, M. S. J. Cell Biol. 121, 565–578 (1993). 13. Danjo, Y. & Gipson, I. K. J. Cell Sci. 111, 3323–3332 (1998). 14. Brock, J., Midwinter, K., Lewis, J. & Martin, P. J. Cell Biol. 135, 1097–1107 (1996). 15. Foe, V. E. Development 107, 1–22 (1989). 16. Kiehart, D. P., Galbraith, C. G., Edwards, K. A., Rickoll, W. L. & Montague, R. A. J. Cell Biol. 149, 471–490 (2000). 17. Young, P. E., Richman, A. M., Ketchum, A. S. & Kiehart, D. P. Genes Dev. 7, 29–41 (1993). 18. Strutt, D. I., Weber, U. & Mlodzik, M. Nature 387, 292–295 (1997). 19. Ridley, A. J. & Hall, A. Cell 70, 389–399 (1992). 20. Nobes, C. D. & Hall, A. Cell 81, 53–62 (1995). 21. Genova, J. L., Jong, S., Camp, J. T. & Fehon, R. G. Dev. Biol. 221, 181–194 (2000). 22. Hakeda-Suzuki, S. et al. Nature 416, 438–442 (2002). 23. Ramet, M., Lanot, R., Zachary, D. & Manfruelli, P. Dev. Biol. 241, 145–156 (2002). 24. Brand, A. H. & Perrimon, N. Development 118, 401–415 (1993). 25. Verkhusha, V. V., Tsukita, S. & Oda, H. FEBS Lett. 445, 395–401 (1999). 26. Oda, H. & Tsukita, S. Dev. Genes Evol. 209, 218–225 (1999). 27. Luo, L., Liao, Y. J., Jan, L. Y. & Jan, Y. N. Genes Dev. 8, 1787–1802 (1994). 28. Mizuno, T., Tsutsui, K. & Nishida, Y. Development 129, 1215–1223 (2002).



5

brief communications 29. Karnovsky, M. J. J. Cell Biol. 27, 137–138 (1965).

ACKNOWLEDGEMENTS We thank all in the Martin lab for constant encouragement and support, K. Barrett and K. Nikolaidou for helpful discussions, M. Turmaine for help with electron microscopy and A. Martinez-Arias for continued enthusiasm and advice. The GFP fusion fly stocks were generously supplied by H. Oda (GFP–actin and α-catenin) and R. Karess (GFP–sqh). The anti-Rho1 antiserum was kindly provided

6

by C. Magie and S. Parkhurst. Thanks also to the Bloomington Stock Center, D. Strutt, B. Dickson and N. Harden for other fly stocks. This work is funded by the Medical Research Council, The Wellcome Trust, The Royal Society and a Pfizer UK studentship to W.W. Correspondence and requests for material should be addressed to P.M.

COMPETING FINANCIAL INTERESTS The authors declare that they have no competing financial interests.



supplementary infomation Movie 1 shows the final hour of epithelial repair in a wild type embryo expressing GFP-actin. Frames are captured every minute. An actin cable is clearly apparent extending the full wound circumference, as well as occasional actin protrusions from the leading edge cells. Note how these cells elongate and constrict at their leading edge during the period of the movie. Movie 2 illustrates repair of a mechanically-created incisional wound made to an embryo expressing GFP-actin. Frames are captured every 30 seconds (total movie time = 35mins). In this wound the epithelial fronts are sufficiently close for filopodial interactions right from the outset of closure. Consequently, repair here is driven more by these zippering interactions than by actin purse-string contraction as occurs in our standard circular wounds (compare with Movie 1). Movie 3 reveals actin dynamics during wound closure in a Rho mutant embryo expressing GFP-actin. Frames are captured every minute. This movie shows the final 90mins of wound closure after the initial lag period of more than 2 hours. There is no actin cable as seen in wild type wounds, but instead the leading edge cells express many filopodia that have coalesced to form large lamellipodia, which enable neighbouring cells to tug on one another and close the wound. Movie 4 shows a direct comparison of wild type (right) and Rho mutant (left) wound closure in embryos expressing GFP-actin. Wounds were made at the same stage and location in the two embryos and frames captured every 2 minutes (total movie time = 240 mins). The wildtype wound has a taut leading edge with a contractile actin cable and occasional filopodial protrusions and takes two hours to close. In contrast the Rho mutant wound has a more disorganized leading edge and does not form a cable. The mutant shows an initial lag phase of about 2 hours before it begins to close at a rate not dissimilar to that of the wild type. Movie 5 is a direct comparison of wild type (right) and Cdc42N17-expressing (left) wounds. Both embryos express GFP-actin in engrailed stripes. The wounds are smaller than in other experiments to ensure that in the mutant embryo only Cdc42defective cells are at the leading edge. Frames are captured once a minute (total movie time = 111mins). The movies commence 30 minutes after laser ablation. From this point the wild type wound closes rapidly through filopodial interaction of opposing epithelial cells. By contrast, Cdc42N17 expressing wound cells, which assemble an actin cable but no filopodial/lamillipoidial protrusions, cannot finally seal the wound closed, even after more than 2 hours.

1
