Dispatch

Current Biology, Vol. 14, R716–R718, September 7, 2004, ©2004 Elsevier Ltd. All rights reserved. DOI 10.1016/j.cub.2004.08.047

Animal Development: Crowd Control

Buzz Baum

To shape a developing animal, individual cellmovements must be coordinated over long distances.Two recent studies help show how this is achievedduring convergence and extension of the Drosophilagerm-band, where polarity within the plane of theembryonic epithelium biases junction remodeling topolarize cell intercalation.

Many animals begin life as a ball of cells beforerearranging to form a long narrow embryo with a headand tail at opposing ends. In the fruit fly Drosophilamelanogaster, the morphogenetic event responsible iscalled germ-band extension. During this process, cellsat the centre of the action undergo relatively smallchanges in shape as the epithelium in which theyreside doubles in length and halves in width in ~100minutes [1]. This is possible because the developinggerm-band is remodeled entirely by orchestratedchanges in cell–cell interactions. Two recent papers[2,3] have furthered our understanding of the molecu-lar and cellular processes involved in this remodeling.

As developing animals are shaped by the collectiveefforts of cells, knowing what individual cells are doingduring a specific morphogenetic process is essential ifthe process is to be properly understood. Irvine andWieschaus [4] were the first to follow the movements ofcells in live embryos during germ-band extension. Inthis way, they observed individual cells in the germ-band epithelium forcing their way between pairs ofneighboring anterior and posterior cells. As aconsequence, the number of cells along the embryonicdorsal-ventral (D-V) axis decreases as cell number alongthe anterior-posterior (A-P) axis increases (Figure 1).

This type of cellular behaviour is called intercalationand underlies the convergence and extension oftissue in many systems [5,6]. As conclusively shownfor Xenopus explants [6], the forces required appearto be generated within the reorganizing tissue itself.Moreover, when the space available for the extendingDrosophila germ-band is limited, these changing cellinteractions generate sufficient force to throw theentire embryo into impressive folds [4]. Pinning downthe source of global polarity in the system, however,has proved difficult.

At this early stage of development, Drosophilaembryos are divided along their A-P axis intometameric units, three to four cells wide, which remaincoherent throughout embryogenesis [7]. Surprisingly,this segmental patterning is required for intercalationand for germ-band extension, whilst D-V patterning isnot [8]. To explain these observations, Irvine andWieschaus [4] proposed a model in which adhesion

differences between stripes of adjacent tissue drivemorphogenesis [9]. If adjacent domains of cells alongthe embryonic A-P axis express different homophilicadhesion molecules, and if individual cells within eachunit are free to maximize their adhesive interactions,these parallel stripes of tissue will have a natural ten-dency to become shorter and fatter over time — asobserved [10] — elongating the embryonic A-P axis.Although the hypothetical cell adhesion moleculesinvolved were not identified, this model remainselegant in its simplicity. It also explains why, inembryos which develop with an excess of the anteriormorphogen Bicoid, the decrease in width of A-Pstripes leads to a corresponding increase in the extentof convergent-extension [4].

It was thus something of a surprise when Lecuit etal. [11] discovered that a GFP-tagged protein, Slam,preferentially localizes at D-V oriented adherens junc-tions when ectopically expressed in the extendinggerm-band. This fortuitous observation revealedhidden polarity within the plane of the embryonicepithelium. Inspired by this finding, Zallen et al. [2] andBertet et al. [3] decided to test whether the observedaxis of planar polarity plays a role in the unidirectionalextension of the germ-band. Importantly, both groupsfound that the polar localization of GFP-Slam mirrorsthat of its endogenous binding partner Myosin II. Inaddition, GFP-Slam and Myosin II only appear polar-ized in tissues actively undergoing intercalation [2]and become concentrated at D-V oriented junctionsas they begin to shorten [3].

As Myosin II is a bipolar, actin-based motor thatpowers the interdigitation of anti-parallel actin filaments[12], this correlation between Myosin II polarity andjunction remodeling suggests that Myosin II mayactively promote the contraction of D-V oriented celljunctions to drive germ-band extension: a notion sup-ported by the demonstration that mutations in MyosinII, or drugs that inhibit Rho kinase-induced Myosin acti-vation, block the dynamic exchange of junctions in thesystem and germ-band extension [3].

To gain a more detailed understanding of thejunction remodeling process, Bertet et al. [3] filmedGFP-labeled junctions during germ-band extension,keeping track of their orientation with respect to theembryonic A-P and D-V axes. Interestingly, junctionswere found to follow a stereotypical path (Figure 1).First, D-V oriented cell junctions shorten. This createsan X-shaped junction at which four cells meet, includ-ing the two cells that were previously separated alongthe D-V axis. Then, the new cell–cell interface expandsin the A-P direction, restoring hexagonal packing. Itwas also clear from this analysis that interfaces linkinga pair of cells in the epithelium always contract orextend together, and that each of a cell’s junctionsacts as an autonomous unit. Although this type ofidentical junctional remodelling has been seen duringconvergent extension movements in fly imaginal discs[5], similar changes have not been observed in manyother epithelial remodeling events.

Ludwig Institute for Cancer Research, University CollegeLondon branch, 91 Riding House Street, London W1W 7BS,UK. E-mail: buzz@ludwig.ucl.ac.uk

This study of planar polarity in the Drosophila germ-band identified two sources of bias that contribute tothe directionality of intercalation [3]. First, there is thetendency of Myosin II to accumulate preferentiallyalong D-V oriented cell junctions, inducing theircontraction. Second, there is the fact that, in its initialstate, the system has an excess of D-V orientedjunctions and a paucity of junctions parallel to the A-P axis. How then do these asymmetries arise?

Zallen et al. [2] and Bertet et al. [3] both found thatthe preferential accumulation of GFP-Slam andMyosin II at D-V junctions requires A-P patterning ofthe embryo. The pair-rule genes Runt and Even-skipped were already known to be required for propergerm-band extension and to inhibit the process whenuniformly expressed [4]. Zallen et al. [2] furthered thisobservation by showing that planar polarity can beartificially induced, in any orientation, simply by juxta-posing cells expressing different levels of Runt orEven-skipped. This is possible even in the headregion, a tissue that does not usually participate ingerm-band extension. Strikingly, the polarity informa-tion induced by these tissue boundaries was found topropagate within the plane of the embryonic epithe-lium, generating patterns of GFP-Slam reminiscent ofthe swirls of hairs seen in the wings of flies carryingmutations in planar polarity genes [13].

These observations led Zallen et al. [2] to examinethe role of known regulators of planar cell polarityduring germ-band extension. Although Dishevelled andFrizzled are required for planar polarity in flies and forconvergent extension in Xenopus [13], they do notappear to play a role in germ-band extension inDrosophila (although polarity in the developing germ-band is bi-polar rather than mono-polar). Nevertheless,mutations in another conserved regulator of epithelialpolarity, Bazooka/Par3 [14–16], were found to preventfull extension of the germ-band. Moreover,Bazooka/Par3 protein is concentrated at A-P orientedjunctions, which tend to lack Myosin II. These datashow that the embryonic epithelium is planar polarizedand that this is important for directed intercalation andgerm-band extension. Even so, the cycle of convergentextension is achieved without cells from different com-partments mixing [1], arguing that adhesion still playsa key role in the process. In fact, it has been suggestedthat refining and increasing the width of segmentalunits may be the raison d’être of germ-band extension[1]. Perhaps differential adhesion and polarised Myosin

II-induced contraction collaborate to bring about con-vergent extension in the germ-band.

Why might two parallel systems, adhesion andplanar polarity, be involved in the process? In thinkingabout this question, it should be noted that althoughMyosin II appears absolutely necessary for the fluidexchange of junctions during germ-band extension,intercalation can still occur in patterning mutants inwhich Myosin II is no longer polarized [3]. A key role ofMyosin II may therefore be to promote rapid cell–cellsampling (Figure 1), encouraging cells, previouslyseparated along the D-V axis, to meet across thecrowded epithelium and to compare adhesionmolecules. If two cells prove to be from the samecompartment, they can then adhere to one another,expanding their common boundary to complete around of intercalation.

Adhesion differences downstream of Even-skippedand Runt will also tend to polarize the epithelium,because cells from different compartments willminimize the extent of their mutual interface. This willstraighten the intervening junctions to increase the pro-portion that are oriented perpendicular to the A-P axis(type I junctions in [3]). As Myosin II and Bazooka/Par3are known to be involved in the establishment of com-plementary cortical domains [14,15], they may relay thispolarity information from segmental boundaries to cellswithin compartments. When confined to complemen-tary cortices Bazooka/Par3 and Myosin II will then biasjunction remodeling, with Bazooka/Par3 stabilizing A-Pjunctions [16], while Rho kinase and Myosin II promoteD-V junction disassembly [17].

In summary, we can reconcile adhesion and planarpolarity models by imagining that, although adhesiondifferences make convergent extension thermody-namically favourable, Bazooka/PAR3 and Myosin IIare vital to facilitate partner swapping, helping to over-come the energy barrier that limits reorganization ofthe epithelium. The planar polarity inherent in thesystem will help break junctional symmetry and willspeed up the process by ensuring that most newinteractions are productive, occurring between cellsfrom the same compartment. The combined use ofadhesion differences and polarized junction remodel-ing could also help ensure that germ-band extensionis robust. In other systems, the relative contributionsof planar polarity, junctional fluidity, adhesion, cellgrowth and division to morphogenesis are likely to bevery different. For example, during elongation of the

Current BiologyR717

Figure 1. Polarized junctional rearrange-ments help drive convergent extension.Cells within the Drosophila germ-bandepithelium form a hexagonal array, eachwith an average of six partners. Prior togerm-band extension, Myosin II, an actin-based motor (purple) becomes concen-trated at D-V oriented junctions, whereasBazooka/Par3 (green) preferentially accu-mulates at A-P oriented junctions. Asintercalation begins, Myosin II promotesD-V junction shortening, which leads to an accumulation of X-shaped junctions (as seen between cells with black nuclei). The result-ing defects in hexagonal packing are then resolved as newly formed A-P oriented junctions grow, perhaps under the influence ofplanar polarized Bazooka/Par3. As the epithelium narrows and elongates, neighboring cells along the A-P axis become separatedwhile stripes of D-V tissue widen, retaining their coherence.

Current Biology

A AP P0' 30'

DispatchR718

Caenorhabditis elegans embryo, Rho kinase activatesMyosin II at D-V oriented adherens junctions tosqueeze the embryo into shape [18]. Remarkably, inthis case, junctions remain intact throughout theprocess, preventing cell intercalation and causing thecells involved to become stretched as the animal elon-gates. In contrast, in zebrafish, polarity-inducedchanges in the orientation of cell divisions contributeto axial elongation [19]. Thus, related molecular mech-anisms appear to be involved in the elongation ofmany types of embryo [18–20], even though the cellbiology underlying morphogenesis appears pro-foundly different in each case.

References1. Costas, M., Sweeton, D., and Wieschaus, E. (1993). Gastrulation in

Drosophila, Volume 1 (Cold Spring Harbor Laboratory Press).2. Zallen, J.A., and Wieschaus, E. (2004). Patterned gene expression

directs bipolar planar polarity in Drosophila. Dev. Cell 6, 343-355.3. Bertet, C., Sulak, L., and Lecuit, T. (2004). Myosin-dependent junc-

tion remodelling controls planar cell intercalation and axis elonga-tion. Nature 429, 667-671.

4. Irvine, K.D., and Wieschaus, E. (1994). Cell intercalation duringDrosophila germband extension and its regulation by pair-rule seg-mentation genes. Development 120, 827-841.

5. Fristrom, D. (1988). The cellular basis of epithelial morphogenesis.A review. Tissue Cell 20, 645-690.

6. Keller, R., Davidson, L., Edlund, A., Elul, T., Ezin, M., Shook, D., andSkoglund, P. (2000). Mechanisms of convergence and extension bycell intercalation. Phil. Trans. R. Soc. Lond. B 355, 897-922.

7. Martinez Arias, A. (1993). Development and patterning of the larvalepidermis, Volume 1 (Cold Spring Harbor Laboratory Press).

8. Wieschaus, E., Sweeton, D., and Costas, M. (1991). Convergenceand extension during germband elongation in Drosophila embryos.(Plenum Press, NY).

9. Steinberg, M.S., and Poole, T.J. (1981). Strategies for specifyingform and pattern: adhesion-guided multicellular assembly. Phil.Trans. R. Soc. Lond. B 295, 451-460.

10. Carroll, S.B., and Scott, M.P. (1986). Zygotically active genes thataffect the spatial expression of the fushi tarazu segmentation geneduring early Drosophila embryogenesis. Cell 45, 113-126.

11. Lecuit, T., Samanta, R., and Wieschaus, E. (2002). slam encodes adevelopmental regulator of polarized membrane growth duringcleavage of the Drosophila embryo. Dev. Cell 2, 425-436.

12. Jacinto, A., and Baum, B. (2003). Actin in development. Mech Dev120, 1337-1349.

13. Mlodzik, M. (2002). Planar cell polarization: do the same mecha-nisms regulate Drosophila tissue polarity and vertebrate gastrula-tion? Trends Genet. 18, 564-571.

14. Benton, R., and St Johnston, D. (2003). Drosophila PAR-1 and 14-3-3 inhibit Bazooka/PAR-3 to establish complementary corticaldomains in polarized cells. Cell 115, 691-704.

15. Severson, A.F., and Bowerman, B. (2003). Myosin and the PARproteins polarize microfilament-dependent forces that shape andposition mitotic spindles in Caenorhabditis elegans. J. Cell Biol.161, 21-26.

16. Wodarz, A., Ramrath, A., Grimm, A., and Knust, E. (2000).Drosophila atypical protein kinase C associates with Bazooka andcontrols polarity of epithelia and neuroblasts. J. Cell Biol. 150, 1361-1374.

17. Sahai, E., and Marshall, C.J. (2002). ROCK and Dia have opposingeffects on adherens junctions downstream of Rho. Nat. Cell Biol. 4,408-415.

18. Wissmann, A., Ingles, J., McGhee, J.D., and Mains, P.E. (1997).Caenorhabditis elegans LET-502 is related to Rho-binding kinasesand human myotonic dystrophy kinase and interacts geneticallywith a homolog of the regulatory subunit of smooth muscle myosinphosphatase to affect cell shape. Genes Dev. 11, 409-422.

19. Gong, Y., Mo, C., and Fraser, S.E. (2004). Planar cell polarity sig-nalling controls cell division orientation during zebrafish gastrula-tion. Nature 430, 689–693.

20. Ninomiya, H., Elinson, R.P., and Winklbauer, R. (2004). Antero-pos-terior tissue polarity links mesoderm convergent extension to axialpatterning. Nature 430, 364-367.