FLUJO DE TRABAJO DEL PROCESO UNIFICADO

In this chapter I describe the dynamic cell shape changes and rearrangements of re- epithelialisation as a wound closes in the 50% epiboly and 24 hour Zebrafish embryo. In addition, I outline the actin-cytoskeletal changes occurring in wound edge cells that probably drive these cell shape changes, and the signals that may in turn trigger cytoskeletal reorganisations. I present results implicating a kinase downstream of the small GTPase Rho, Rho-kinase, which appears to be responsible for directing actin cable assembly. As in single cell repair I demonstrate a dramatic re-orientation of microtubules following wounding - this time within the front three or four rows of cells at the wound edge - but unlike single cell repair, microtubules seem unnecessary for cable assembly or subsequent healing of the wound.

• Cells at the wound edge undergo a reproducible programme of cell shape changes and movements

In order to dissect out the signals directing re-epithelialisation of a wound it is first critical to have a full description of precisely the cell shape changes and cell movements that underlie the concerted tissue movement of epithelial forward movement. This I have elucidated from SEM snap-shots of the process and by timelapse imaging of lectin-stained embryos as they repair a wound. Essentially it appears that the front 2 or 3 rows of cells back from the wound edge participate in the movement by actively changing their shapes. By far the most dramatic shape change - apical constriction and cell elongation - appears to

C hapter 3 Discussion

be largely restricted to the front row cells. It seems that epithelial purse-string closure is achieved by a combination of concerted cell shape changes and cell shufflings, with all the wound margin cells constricting their front edge, and one-by-one altering their cell.'cell contacts to withdraw from the front row and move into rows further back. Consequently, as the wound closes there are fewer and fewer of the initial wound edge cells remaining in the front row, until the front edges of the last few cells meet and close a tiny pin-hole wound at the end of the repair process. The cell shape changes in leading edge cells appear to be coincident with, and presumably dependent on, assembly and contraction of an actin purse-string, just as has been shown in several other embryonic re-epithelialisation models (Brock et al., 1996; McCluskey and Martin, 1995).

• What signals might initiate these actin-driven cell shape changes?

The Rho family of small GTPases have long been implicated in controlling actin polymerisation and driving cell shape changes. In other wound healing models, Rho has been shown to direct assembly of the actin-cable. In both single-cell wounds (Dement et al., 1999) and multi-cellular wounds (Brock et al., 1996), blocking Rho activity with C3 transferase caused a failure to assemble an actin cable and subsequent repair but in these studies there was no attempt to analyse the effector signals downstream of Rho. 1 find that incubation of 50% epiboly embryos with Y-27632, a specific inhibitor of Rho-kinase (Uehata et al., 1997), prevents wound-induced cable assembly and the subsequent closure of a wound, suggesting that Rho operates via the effector Rho-kinase, in order to direct assembly of the actin-cable.

Chapter 3 Discussion

In addition to mediating stress fibre formation, Rho-kinase has been implicated in transcriptional activation of the serum response element of c-fos since constitutively-active Rho-kinase also stimulates the transcriptional activity of c-fos SRE (Chihara et al., 1997). C-fos message and protein have been detected within wound edge cells following mouse hindlimb amputation (Martin and Nobes, 1992) and whether this is in response to Rho activation in wound edge cells has not been tested. However c-fos can be activated by several other pathways, one of which, calcium, is also thought to play an important role during tissue repair (discussed in Chapter 5).

In addition to the Rho-kinase family of proteins, Rho activates several other signalling pathways which may also be important in other aspects of the wound response, although there must be some control of specificity in downstream effectors. Other proteins which could be involved downstream of Rho are those encoded by the Formins, e.g. pl40Dia in mice (related to Drosophila diaphanous and the yeast budding polarity genes Bn Ip and Pus Ip). These formins have been shown to bind Rho and may play key roles in actin reorganisation and assembly. The FH l domain has been implicated in binding profilin which can promote the elongation of actin filaments, and therefore could be important in stress fibre and cable formation (Wasserman, 1998; Watanabe et al., 1997). Also downstream of Rho is CitronK which is involved in cytokinesis. Whilst the healing of an embryonic wound appears to be largely achieved by epithelial cell shape changes and cell rearrangements, cell division is also required after the hole has closed to replace any

C hapter 3 Discussion

damaged or lost tissue. Therefore although is it unlikely that Rho is acting at that time to activate CitronK some feedback loop might later trigger cell proliferation.

• Membrane remodelling may play a role in re-epithelialisation

My timelapse studies of wounded lectin-stained embryos show that significant membrane turnover occurs in many of the cells which alter their cell shape surrounding the wound. This is not surprising since membrane traffic and remodelling have been shown to play a key role in motility of several cell types (reviewed in Bretscher and Aguado-Velasco, 1998) and it may well be that this is true to purse-string re-epithelialisation also. Exocytosis appears to be at least partially responsible for driving forward movement at the leading edge of some advancing cells. For example, in the plasmodium Physarum, vesicles have been observed fusing with the leading edge and producing a protrusion with a surface area equivalent to the original vesicle (Sesaki and Ogihara, 1997; Sesaki and Ogihara, 1997). In mammalian cells, experiments demonstrate that newly synthesised proteins and the endocytic cycle both emerge at the leading edge of polarised cells suggesting this may be a conserved mechanisms in mammalian cells (Marcus, 1962).

Almost certainly this endocytic machinery is dependent on the cytoskeleton for directing vesicle traffic and for scaffolding the dynamic membrane and indeed there is some evidence that membrane turnover via the endosome cycle is dependent on the small GTPase Rac.

Chapter 3 Discussion

• What triggers Rho activity?

The primary cues at the time of wounding that might activate Rho remain largely elusive. Immediately following wounding, release of tension in the epithelium leads to the wound edges gaping open, and so a potential cue for activation of Rho might be a stretch signal. Indeed, there is evidence that related mechanical cues can activate the small GTPases in tissue culture models (Li et al., 1999). Alternatively, the signal could be chemical; any one of the cocktail of cytokines and growth factors in the amniotic fluid bathing the embryo might be the initial stimulus for triggering a wound response. Another potential signal is calcium which is at a concentration 10,000 times higher extracellularly than within cells and so enters damaged cells as they become transiently leaky upon wounding. This Ca^+ influx has been shown to be important in several wound healing models (Stanisstreet, 1982; Tran et al., 1999) (discussed in chapter 5).

In yeast, genetic studies of the response to cell wall damage show that Rho acts to direct local actin reorganisations at sites adjacent to fluid regions of plasma membrane, and Rho also triggers redistribution of a biosynthetic enzyme for rebuilding the yeast cell wall. Upstream of Rho is a plasma membrane protein, W SCl, which apparently operates as a damage sensor both controlling and responding to the actin cytoskeleton (Delley and Hall, 1999). Clearly several components of this cell repair signalling machinery are likely to be specialised for patching up parts of the yeast cell that have no equivalent in vertebrate cells, but there are also certain to be several conserved elements and yeast offers a superb genetic tool by which to discover these parallels.

Chapter 3 Discussion

• How does this machinery switch off when two edges meet?

When two wound edges confront one another, advancement of the epithelial edges stops in a poorly understood process known as contact inhibition (Abercrombie, 1979). In chick and mouse embryos the wound epithelium appears to slightly run-over the endpoint, as epithelial fronts make contact, suggesting that the motility machinery takes a while to shut down. Indeed, the actin cable still persists for at least several minutes post contact before fully disassembling leaving a pool of disorganised actin, perhaps as a response to the changing tension in the newly-closed epithelium (Brock et al., 1996).

Since morphogenetic movements like dorsal closure in flies and ventral enclosure in worms halt as the epithelia meet and do not over-run, model systems like this are not only perfect hunting grounds for signals that might activate and initiate tissue repair movements, but also for the cellxell signals that read contact inhibition cues and brake tissue movement so that they come to a smooth halt just as the wound is closed. During fly dorsal closure it is clear that a dual activity phosphatase, puckered, is acting as some sort of brake on the advancing lateral epithelial fronts. Puckered acts as a negative regulator on the JNK pathway, and presumably such negative feedback loops may be acting during the final movements of wound closure also. In C. elegans, the Eph family of membrane receptors and ligands appear to be important also during the final stages of ventral enclosure, as worms null for Vab-I, an Eph receptor tyrosine kinase, are also disrupted in this movement (George et al., 1998). Whatever stop signals are being activated it is likely that these will

Chapter 3 Discussion

be relayed through the small GTPases in order to disassemble cytoskeletal motility machinery that was assembled for closure of the wound.

• How is the hole finally sealed closed?

While the mechanisms of tissue movements of the advancing wound edges has been well studied, the actual adhesion of the two opposing epithelial edges once they meet and seam together has only recently begun to be characterised. Studies in primary tissue culture cells have demonstrated that epithelial cells extend calcium-dependent filopodia which project towards adjacent cells and embed into their plasma membranes (Vasioukhin et al., 2000). Various adhesion proteins localise to the tips of the filopodia, including E-cadherin and other components of adherens junctions. In this way the two opposing epithelial edges form confronting rows of puncta, each comprising junctional proteins, and thus forming a so-called adhesion zipper. Once the opposing membranes have been brought together by the filopodia, the two membranes become secondarily attached by desmosomes which form at flanking sites in between the adherens puncta. The row of puncta then merge into a single row, and then into a continuous line over a period of several hours as they form mature adherens junctions.

More recently, analogous epithelial adhesion machinery has been shown to occur during dorsal closure in Drosophila, where the two lateral epithelia meet at the dorsal midline and zipper closed (Jacinto et al., 2000). Live time-lapse imaging of transgenic GFP-actin Drosophila embryos undergoing the final zippering stages of dorsal closure show that

C hapter 3 D iscussion

filopodia extend from the leading edge cells of the lateral epithelia as it sweeps forward over the amnioserosa. As filopodia make contact with their counterparts on the opposing epithelial face, the cells also exhibit lamellae, as if stimulated by the contact recognition, and these lamellae draw the fronts together. As in tissue culture, TEM studies show that the filopodia interdigitate with those of the opposing cells, as a primer to forming mature adherens junctions with them. If filopodial assembly is blocked by expression of dominant negative Cdc42 then the epithelial sheets fail to seam together correctly and dorsal closure fails.

Filopodia have also been detected at the leading edge during repair of epithelial wounds made in the fly embryo suggesting that this mechanism of epithelial sheet adhesion may be a fundamental process that occurs during the fusion of two opposing wound fronts, as well as during morphogenesis. To date no live studies of dynamic actin activities in wound edge cells have been performed in any vertebrate embryo wound model but clearly the Zebrafish offers the best opportunities as soon as transgenic GFP actin lines become available.

• What role do microtubules have during wound closure?

In this chapter I have presented data suggesting that microtubules are not required for assembly of the actin cable and subsequent closure of a multicellular wound. These results apparently directly contrast with those of Bement showing that microtubules are transiently required for actinomyosin trafficking towards the margin of single-cell wounds in the Xenopus oocyte (Bement et al., 1999).

Chapter 3 Discussion

Interestingly, in Bement’s earlier experiments on multicellular wounds in Caco-2 cells, microtubules were not found to be associated with the cable, and nor were the microtubules in leading edge cells orientated with respect to the wound (Bement et al., 1993). During cell migration of comeal endothelial cells in tissue culture, wound closure is prevented by treatment with colchicine and wound edge cells no longer display extensions typical of migrating cells (Gordon and Staley, 1990). Potentially, microtubules could be required for polarity during directed migration or for directing intracellular vesicles to the leading edge. However, several tissue culture fibroblast scrape-wound models show that nocodazole treatment causes loss of cell polarisation at wound edges and a loss of directed cell migration (Bershadsky and Futerman, 1994).

1 propose that the reorientation of microtubules after wounding an epithelium is simply a result of passive stretching of the cells that are elongating; as the wound closes and leading edge cells resume their normal shape, the microtubules would re-establish the random network of a relaxed cell.

If microtubules are not required for actin-cable assembly, how does actin locate to the leading edge? As the cell is damaged various factors leak in through the damaged membrane and may set up a gradient of signals, with higher concentrations at the damaged membrane closest to the extracellular medium and low concentrations at the edge of the cell away from membrane damage. One of these gradients may be a signal for actin polymerisation, so that actin is specifically polymerised into cable-like structures only at

Chapter 3 Discussion

sites of membrane damage. Another possibility is that factors are released from the membrane at the site of damage, much like the W SCl protein in yeast which acts as a membrane bound damage sensor. These issues may well be best resolved in large single cells such as the sea urchin embryo where minor perturbations and the study of influxes of molecules can more easily be observed (Terasaki et al., 1997).

• Similar sets of concerted ceil shape changes occur during some epithelial morphogenetic processes

As previously mentioned, dorsal closure in the Drosophila embryo looks remarkably like re-epithelialisation of an embryonic wound, with a free epithelial edge apparently being drawn forward by contraction of an actin cable with numerous lamellae and filopodia extending from the leading edges. Much of the signalling driving dorsal closure has been characterised and shares several similarities with wound healing where formation of the actin cable appears to be dependent on one of the small GTPases, Rho, and downstream of this the AP-1 family are activated leading to TGF-P upregulation in a paracrine fashion in the contracting mesenchyme. These similarities suggest that many more signals may be conserved between the two processes, including signals controlling cytoskeletal reorganisations, signals controlling epithelial migration and signals preventing the over­ running of the two opposing epithelial sheets at the end of wound closure. Almost certainly there is much more to dorsal closure than simply closure of an epithelial hole as in wound healing, but it seems likely that some of the signalling machinery activated during dorsal closure will be re-activated at the wound site. In addition, an early morphogenetic

Chapter 3 Discussion

movement in the Zebrafish superficially resembles wound closure, and this will be discussed in the following chapter.

C hapter 4 Introduction