Two separate molecular systems, Dachsous/Fat and Starry night/Frizzled,act independently to confer planar cell polarity

Most organisms are built of epithelia consisting of cells that are both asymmetric in the apicobasal axis and within the plane of the cell sheet(Fanto and McNeill, 2004; Grebe, 2004). Planar cell polarity (PCP) is shown by the orientation of structures such as hairs in insects (Lawrence, 1966; Strutt, 2003; Saburi and McNeill, 2005), and cilia (Eaton, 1997) and stereocilia in vertebrates (Lewis and Davies, 2002). PCP is also implicated in convergent extension in vertebrate embryos (Wallingford et al.,2002). Genetic and molecular studies in Drosophila have identified proteins essential for PCP; these are generally conserved in vertebrates (Klein and Mlodzik,2005). Here, we use Drosophila and build a new logical structure for PCP.

There are two sets of genes involved in PCP: the Stan system and the Ds system. The Stan system depends on a cadherin receptor-like molecule, Starry Night (Stan) (Chae et al.,1999; Usui et al.,1999) and Frizzled (Fz) (Adler et al., 1997), a receptor for Wnts(Wodarz and Nusse, 1998). Other proteins in the Stan system are Diego, Dishevelled, Van Gogh (Vang -FlyBase; also called Strabismus) and Prickle. There are several ideas about how the Stan system might function. A popular model proposes that PCP is determined by asymmetrically localised complexes of Stan system proteins in cell membranes (Strutt, 2002). This asymmetry, which has been observed in some epithelial cells, would be oriented by an unknown graded signal ['factor X'(Struhl et al., 1997a)]. Propagation of PCP would be driven by feedback between proteins, the asymmetrical arrangement of proteins in one cell affecting localisation in neighbouring cells (Tree et al.,2002; Amonlirdviman et al.,2005). We have argued(Lawrence et al., 2004) that this view is largely incorrect, and base our opinion mainly on two pieces of evidence. First, cells that completely lack the Fz protein can be polarised by their neighbours - yet, in the asymmetry model the orientation of each cell depends on the differential accumulation and activity of Fz(Tree et al., 2002). Second,flies that lack a crucial component of the feedback mechanism of the Stan system, Prickle, lose the asymmetric localisation of other core proteins -yet, in these flies, disparities in the amounts of Fz still propagate polarity from cell to cell (Lawrence et al.,2004). This result with prickle mutant flies has been confirmed in the wing and even extended to wings mutant for dishevelled (Strutt and Strutt,2006).

Our alternative model for the Stan system has four main tenets(Lawrence et al., 2004): (1)Fz activity is normally gently graded from one cell to the next as a response to factor X; (2) a cell becomes polarised by comparing its own level of Fz activity with that of its various neighbouring cells, and pointing hairs towards neighbours with lower levels and away from neighbours with higher levels; (3) the level of Fz activity in any one cell is subject to feedback that adjusts its level to an average of its neighbours - this `averaging'mechanism explains how and why experimentally induced disparities in Fz activity can induce changes in polarity that propagate for several cells; (4)cells perceive differences in their level of Fz activity relative to that of their neighbours through intercellular homodimers made by Stan - hence, Stan is required for cells both to send and to receive this information.

The second set of genes that acts in PCP, the Ds system, encodes two atypical cadherins, Dachsous (Ds) and Fat (Ft), as well as a resident Golgi protein, Four-jointed (Fj) (Strutt et al.,2004). The Ds system, like the Stan system, orients cellular outgrowths. However, unlike the Stan system, it also affects the orientation of cell divisions and organ shape, as well as having some input into growth(Bryant et al., 1988; Baena-López et al.,2005). In an important paper, Yang et al.(Yang et al., 2002) proposed that the polarity genes constitute a linear pathway in which morphogens, such as Wingless (Wg), orient the Ds system. In the eye, this system consists of opposing gradients of Fj and Ds controlled by Wg(Simon, 2004) with Fj first repressing Ds activity and Ds then repressing Ft activity. Yang and colleagues argued that Ft then activates Fz to polarise the Stan system. Thus, the graded activity of the Ds system constitutes factor X, and the Stan system transduces X to polarise cells. This single pathway model of PCP has become accepted and now prevails in the literature on PCP(Adler, 2002; Strutt and Strutt, 2002; Ma et al., 2003; Uemura and Shimada, 2003) (but see Klein and Mlodzik, 2005; Strutt and Strutt, 2005a; Strutt and Strutt, 2005b).

Experiments in the Drosophila abdomen give comparable results with those in the eye. A morphogen, Hedgehog (Hh), appears to be responsible for activity gradients of Fj, Ds and Ft (Casal et al., 2002). As in the eye, the Stan system acts in PCP but there is no evidence as to whether there is a single pathway: Hh→Ds system→Stan system. Experimentally, the abdomen has some advantages over the eye. For example, in the eye, PCP is revealed in the arrangement of cells in entire ommatidia: each an ensemble of photoreceptors, lens and pigment cells. In the abdomen, the polarity of each cell is shown directly by the orientation of hairs produced by that cell alone. Here, we use this advantage to test whether the Ds and Stan systems act as part of a single linear pathway. Our main conclusion is that they do not and that each system deploys a different mechanism to polarise cells and to propagate polarity from cell to cell.

Unless stated otherwise, FlyBase(Grumbling et al., 2006)entries of the mutations and transgenes referred in the text are as follows. CD2y⁺: Rnor\CD2^hs.PJ. hs.FLP: Scer\FLP1^hs.PS. act.fz::GFP: fz^{P278L.Act5C.T:Avic\GFP-EGFP.}tub.Gal80: Scer\GAL80^{alphaTub84B.PL}. tub.Gal4: Scer\GAL4^{alphaTub84B.PL}. ptc.Gal4: Scer\Gal4^ptc-559.1. UAS.GFP: Avic\GFP^{Scer\UAS.T:Hsap\MYC,T:SV40\nls2.} UAS.fz:fz^Scer\UAS.cZa and fz^Scer\UAS.cSa. UAS.ft: ft^Scer\UAS.cMa. UAS.ds: ds^Scer\UAS.cTa. UAS.fj: fj^Scer\UAS.cZa. UAS.Nrt::wg: Nrt::wg^{Scer\UAS.T:Ivir\HA1}. UAS.fz2DN: fz2^{GPI.Scer\UAS.T:Hsap\MYC}. UAS.Wnt2: Wnt2^Scer\UAS.cSa. UAS.Wnt4: Wnt4^{Scer\UAS.cSa.} UAS.Wnt6: Wnt6^Scer\UAS.cSa. UAS.Wnt8: wntD^Scer\UAS.cSa. UAS.Wnt10: Wnt10^Scer\UAS.cSa. fz^-: fz¹⁵ or fz^21. Df(3L)fz2. fz2^-: fz2^C1. ds^-: ds^UA071. ds^38K. ft^-: ft¹⁵. ft¹². fj^-: fj^d1. fj^N7. ptc^-: ptc^IIW. en^-: Df(2)en^E. FRT39: P{FRT(w^hs)}39. FRT40: P{neoFRT}40A. FRT42: P{neoFRT}42D. FRT2A: P{FRT(w^hs)}2A. FRT80: P{neoFRT}80B. The following are derivatives of P{UAS-ds.T}and P{UAS-ft.M} (Matakatsu and Blair, 2004), in which the amino acid sequence of the joins are UAS.ectoDs:...FLFIHMRSRKPRprp. UAS.ectoFt:...LGSYVIYRFRprprp. UAS.ectoDs::endoFt:...FLFIHMRSRKPRGKQEKIGSL.... UAS.ectoFt::endoDs:...LGSYVIYRFRPRNAVKPHLAT...(ds sequences are in bold, ft sequences are in italics,added sequences are in lower case and transmembrane sequences are underlined). UAS.endoFt: As in P{UAS-wg.flu}(Zecca et al., 1996), the wg signal peptide is followed by three copies of the HA1 epitope tag,joined to the Ft transmembrane and cytoplasmic domains. The amino acid sequence at the join is...[YPYDVPDYA]sAAQVADPLSIGFTLVI... UAS.endoDs:...[YPYDV-PDYA]sAGGSSGGSIGDWAIGLL... (the sequence in brackets corresponds to the last flu epitope; the beginning of the transmembrane domains of both proteins is underlined).

The stan³/stan^E59 allelic combination, the `stan^-' genotype, was chosen for the following reasons: the amorphic allele stan^E59 is lethal homozygous, owing to a requirement for Stan activity in the nervous system. stan^E59 mutant flies can be rescued by neural expression of the stan gene (Lu et al.,1999), but doing this was impractical with our complex genotypes and also open to the criticism that low level expression of the rescuing UAS.stan transgene in the epidermis might alleviate the PCP phenotype. We therefore used a hypomorphic allele, stan³,in trans to stan^E59, a combination devoid of PCP activity in the abdomen (Lawrence et al.,2004) (see Fig. 2). For the key experiments where we generated UAS.ft and ectoDsclones in stan^- flies (genotypes 4, 5, 12), the cells within the clones are stan^E59/stan^E59(the null genotype); only the surrounding cells are stan³/stan^E59 and yet polarisation still occurred. Nevertheless, under the same conditions, both UAS.fzand UAS.fz UAS.stan clones failed to repolarise(Lawrence et al., 2004;genotypes 2, 3). Finally, stan^E59UAS.ft and stan^E59 UAS.ectoDs clones repolarise surrounding stan³/stan^E59cells, even in flies that were also fz^- (genotypes 18, 19) and indubitably lacking the Stan system.

Clones were induced by heat-shocking third instar larvae for 1 hour at 34,35 or 37°C. Abdominal cuticles were dissected, mounted in Hoyer's and images captured with Auto-Montage (Syncroscopy) and processed with Adobe Photoshop (Adobe Systems).

1. stan^- fz^- clones: y w hs.FLP; FRT42 act.fz::GFP CD2y⁺/FRT42 pwn stan^E59 sha; fz^-ri FRT2A/fz^- CD2y⁺ ri FRT80

2. tub.Gal4 UAS.stan UAS.fz clones in stan^-: y w hs.FLP; FRT42 stan³ tub.Gal80 CD2y⁺/FRT42 pwn stan^E59; UAS.fmi UAS.fz/tub.Gal4

3. tub.Gal4 UAS.fz clones in stan^-: y w hs.FLP; FRT42 tub.Gal80 stan³ CD2y⁺/FRT42 pwn stan^E59; UAS.fz/tub.Gal4

4. tub.Gal4 UAS.ft clones in stan^-: y w hs.FLP; FRT42 tub.Gal80 stan³ CD2y⁺/FRT42 pwn stan^E59; UAS.ft/tub.Gal4 and

5. y w hs.FLP tub.Gal4 UAS.GFP/y w hs.FLP; FRT42 tub.Gal80 stan³ CD2y⁺/FRT42 pwn stan^E59;UAS.ft/+

6. tub.Gal4 UAS.ft clones: y w hs.FLP/w; FRT42 tub.Gal80 CD2y⁺/FRT42 pwn; UAS.ft/tub.Gal4 and

7. y w hs.FLP tub.Gal4 UAS.GFP/y w hs.FLP; FRT42 pwn/ds^- FRT42 tub.Gal80 CD2y⁺; UAS.ft/+ and

8. y w hs.FLP; FRT42 pwn/ds^- FRT42 tub.Gal80 CD2y⁺;UAS.ft/tub.Gal4

9. ft^- clones in stan^-: y w hs.FLP; ft^- stc FRT39 stan^E59/CD2y⁺ FRT39 stan³

10. tub.Gal4 UAS.ds clones: y w hs.FLP/w; FRT42 tub.Gal80 CD2y⁺/FRT42 pwn; UAS.ds/tub.Gal4

11. tub.Gal4 UAS.ectoDs clones: y w hs.FLP; FRT42 tub.Gal80 CD2y⁺/FRT42 pwn sha; UAS.ectodDs/tub.Gal4

12. tub.Gal4 UAS.ectoDs clones in stan^-: y w hs.FLP tub.Gal4 UAS.GFP/y w hs.FLP; FRT42 pwn stan^E59 sha/FRT42 tub.Gal80 stan³ CD2y⁺; UAS.ectoDs/+

13. tub.Gal4 UAS.fj clones in stan^-: y w hs.FLP tub.Gal4 UAS.GFP; FRT42 tub.Gal80 stan³CD2y⁺/FRT42D pwn stan^E59; UAS.fj/+

14. tub.Gal4 UAS.ft clones in fz^-: y w hs.FLP tub.Gal4 UAS.GFP/y hs.FLP; FRT42 tub.Gal80 CD2y⁺; UAS.ft FRT42 pwn;fz^- ri FRT2A/fz^- CD2y⁺ ri FRT2A and

15. y w hs.FLP tub.Gal4 UAS.GFP/y w hs.FLP; FRT42 tub.Gal80 CD2y⁺/FRT42 pwn sha; fz^- CD2y⁺UAS.ft/fz^- ri FRT2A

16. ft^- clones in fz^-: y hs.FLP;ft^- stc FRT39/CD2y⁺ FRT39; fz^-/fz^-trc FRT2A

17. tub.Gal4 UAS.ectoDs clones in fz^-: y w hs.FLP tub.Gal4 UAS.GFP/y w hs.FLP; FRT42 tub.Gal80 CD2y⁺/FRT42 pwn sha; fz^- CD2y⁺ UAS.ectoDs/fz^- ri FRT2A

18. tub.Gal4 UAS.ft clones in stan^-fz^-: y w hs.FLP tub.Gal4 UAS.GFP/y w hs.FLP; FRT42 pwn stan^E59 sha/FRT42 tub.Gal80 stan³ CD2y⁺;fz^- CD2y⁺ UAS.ft ri FRT2A/fz^-CD2y⁺ ri FRT80

19. tub.Gal4 UAS.ectoDs clones in stan^-fz^-: y w hs.FLP tub.Gal4 UAS.GFP/y w hs.FLP; FRT42 pwn stan^E59 sha/FRT42 tub.Gal80 stan³ CD2y⁺;fz^- CD2y⁺ UAS.ectoDs ri FRT2A/fz^-CD2y⁺ ri FRT80

20. fz^- clones in ds^-: y w hs.FLP12; ds^- FRT39/In(2LR)bw^V1; fz^- trc ri FRT2A/CD2y⁺ hs.GFP ri FRT2A

21. tub.Gal4 UAS.fz clones in ds^-: y w hs.FLP tub.Gal4 UAS.GFP/y w hs.FLP; ds^- FRT42 pwn/ds^- FRT42 Gal.80 CD2y⁺; UAS.fz CD2y⁺/+ and

22. y w hs.FLP122 tub.Gal4 UAS.GFP/y w hs.FLP122; ds^- ck FRT40/ds^- tub.Gal80 FRT40; UAS.fz fz^- fz2^C1FRT2A/+

23. tub.Gal4 UAS.stan clones in ds^-: y w hs.FLP tub.Gal4 UAS.GFP/y w hs.FLP; ds^- FRT42 pwn/ds^-FRT42 tub.Gal80 CD2y⁺; UAS.stan CD2y⁺/+

24. tub.Gal4 UAS.fz clones in ft^-: y w hs.FLP;ft^- FRT42 pwn sha/ft¹² FRT42 tub.Gal80 CD2y⁺;UAS.fz/tub.Gal4

25. hh.Gal4 UAS.fz in ds^-: y w hs.FLP122;ds^- ck FRT40/In(2LR)bw^V1, ds^-; hh.Gal4/UAS.fz fz^- fz2^- FRT2A

26. 2xfz⁺ clones in ds^-/ds^-;fz⁺/fz^-: y w hs.FLP; ds^-;CD2y⁺ trc ri FRT2A/fz^- Df(3L)fz2 FRT2A

27. fz^- tub.Gal4 UAS.ft clones: y w hs.FLP tub.Gal4 UAS.GFP/y w hs.FLP; UAS.ft FRT42 pwn; tub.Gal80 FRT2A/fz^- trc ri FRT2A

28. tub.Gal4 UAS.fz clones in ds^-fz^-: y w hs.FLP tub.Gal4 UAS.GFP/y w hs.FLP;ds^- FRT42 sha/ds^- FRT42 tub.Gal80 CD2y⁺;fz^- CD2y⁺ ri FRT2A UAS.fz/fz^-CD2y⁺ ri FRT80

29. tub.Gal4 UAS.ectoDs in ds^-: y w hs.FLP tub.Gal4 UAS.GFP/y w hs.FLP; ds^- tub.Gal80 FRT40/ds^- ck FRT40; UAS.ectoDs/+ and

30. y w FL122; ds^- CD2y⁺ FRT42 pwn sha/ds^-FRT42 tub.Gal80 CD2y⁺; UAS.ectoDs/tub.Gal4

31. tub.Gal4 UAS.ds in ds^-: y w hs.FLP tub.Gal4 UAS.GFP/y w hs.FLP; ds^- tub.Gal80 FRT40/ds ck FRT40;UAS.ds/+ and

32. y w hs.FLP tub.Gal4 UAS.GFP/y w hs.FLP; ds^- FRT42 pwn/ds^- FRT42 tub.Gal80 CD2y⁺; UAS.ds/+

33. tub.Gal4 UAS.ft in ds^-: y w hs.FLP tub.Gal4 UAS.GFP/y w hs.FLP; ds^- tub.Gal80 FRT40/ds^- ck FRT40; UAS.ft/+ and

34. y w hs.FLP tub.Gal4 UAS.GFP/y w hs.FLP; ds^- FRT42 pwn/ds^- FRT42 tub.Gal80 CD2y⁺; UAS.ft/+

35. tub.Gal4 UAS.ectoDs clones in ft^-: y w hs.FLP; ft^- FRT42 pwn sha/ft¹² FRT42 tub.Gal80 CD2y⁺; UAS.ft/tub.Gal4

36. tub.Gal4 UAS.ft clones in ft^-: y w hs.FLP;ft^- FRT42 pwn sha/ft¹² FRT42 tub.Gal80 CD2y⁺;UAS.ft/tub.Gal4

37. tub.Gal4 UAS.fz UAS.ft clones in ds^-: y w hs.FLP; ds^- CD2y⁺ FRT42 pwn sha/ds^- FRT42 tub.Gal80 CD2y⁺; UAS.fz UAS.ft/tub.Gal4

38. tub.Gal4 UAS.fz UAS.ds clones in ds^-: y w hs.FLP; ds^- CD2y⁺ FRT42 pwn sha/ds^- FRT42 tub.Gal80 CD2y⁺; UAS.fz UAS.ds/tub.Gal4

39. tub.Gal4 UAS.ft UAS.fz clones in ft^-: y w hs.FLP; ft^- FRT42 pwn sha/ft¹² FRT42 tub.Gal80 CD2y⁺; CD2y⁺ UAS.ft UAS.fz/tub.Gal4

40. stan^-: y w hs.FLP/+;stan³/ds^- CD2y⁺ FRT42 pwn stan^E59sha

41. ds^-: y w hs.FLP; ds^- tub.Gal80 FRT40/ds^- CD2y⁺ FRT42 pwn stan^E59sha

42. ds^- stan^-: y w hs.FLP tub.Gal4 UAS.GFP/y w hs.FLP; ds^- tub.Gal80 FRT40 stan³/ds^-CD2y⁺ FRT42 pwn stan^E59 sha

43. ds^- ft^- clones: y w hs.FLP/y;ds^- ft^- stc FRT39/CD2y⁺ FRT39

44. ds^- ft^- tub.Gal4 UAS.ds clones: y w hs.FLP; ds^- ft^- stc FRT39/tub.Gal80 CD2y⁺FRT39; UAS.ds/tub.Gal4

45. ds^- ft^- tub.Gal4 UAS.ft clones: y w hs.FLP; ds^- ft^- stc FRT39/tub.Gal80 CD2y⁺FRT39; UAS.ft/tub.Gal4

46. ds^- tub.Gal4 UAS.ectoDs clones: y w hs.FLP;ds^- stc FRT39/tub.Gal80 CD2y⁺ FRT39;UAS.ectoDs/tub.Gal4

47. ft^- tub.Gal4 UAS.ectoDs clones: y w hs.FLP;ft^- stc FRT39/tub.Gal80 CD2y⁺ FRT39;UAS.ectoDs/tub.Gal4

48. ds^- ft^- tub.Gal4 UAS.ectoDs clones: y w hs.FLP; ds^- ft^- stc FRT39/tub.Gal80 CD2y⁺FRT39; UAS.ectoDs/tub.Gal4

49. tub.Gal4 UAS.ectoFt clones: y w hs.FLP; FRT42 pwn sha/FRT42 tub.Gal80 CD2y⁺; UAS.ectoFt/tub.Gal4

50. ft^- tub.Gal4 UAS.ft clones: y w hs.FLP;ft^- stc FRT39/tub.Gal80 CD2y⁺ FRT39;UAS.ft/tub.Gal4

51. ft^- tub.Gal4 UAS.ectoFt clones: y w hs.FLP;ft^- stc FRT39/tub.Gal80 CD2y⁺ FRT39;UAS.ectoFt/tub.Gal4

52. ds^- ft^- tub.Gal4 UAS.ectoFt clones: y w hs.FLP; ds^- ft^- stc FRT39/tub.Gal80 CD2y⁺FRT39; UAS.ectoFt/tub.Gal4

53. tub.Gal4 UAS.ectoDs::endoFt clones: y w hs.FLP; FRT42 pwn sha/FRT42 tub.Gal80 CD2y⁺; UAS.ectoDs::endoFt/tub.Gal4

54. ds^- tub.Gal4 UAS.ectoDs::endoFt clones: y w hs.FLP;ds^- stc FRT39/tub.Gal80 CD2y⁺ FRT39; UAS. ectoDs::endoFt/tub.Gal4

55. ft^- tub.Gal4 UAS.ectoDs::endoFt clones: y w hs.FLP;ft^- stc FRT39/tub.Gal80 CD2y⁺ FRT39; UAS. ectoDs::endoFt/tub.Gal4

56. ds^- ft^- tub.Gal4 UAS.ectoDs::endoFt clones: y w hs.FLP; ds^- ft^- stc FRT39/tub.Gal80 CD2y⁺ FRT39; UAS. ectoDs::endoFt/tub.Gal4

57. tub.Gal4 UAS.ectoFt::endoDs clones: y w hs.FLP; FRT42 pwn sha/FRT42 tub.Gal80 CD2y⁺; UAS.ectoFt::endoDs/tub.Gal4

58. ds^- tub.Gal4 UAS.ectoFt::endoDs clones: y w hs.FLP;ds^- stc FRT39/tub.Gal80 CD2y⁺ FRT39;UAS.ectoFt::endoDs/tub.Gal4

59. ft^- tub.Gal4 UAS. ectoFt::endoDs clones: y w hs.FLP; ft^- stc FRT39/tub.Gal80 CD2y⁺ FRT39;UAS.ectoFt::endoDs/tub.Gal4

60. ds^- ft^- tub.Gal4 UAS. ectoFt::endoDs clones: y w hs.FLP; ds^- ft^- stc FRT39/tub.Gal80 CD2y⁺ FRT39; UAS.ectoFt::endoDs/tub.Gal4

61. ptc.Gal4 UAS.endoDs: y w hs.FLP; Sp/fj^- ptc.Gal4;UAS.endoDs/+

62. ptc.Gal4 UAS.endoFt: y w hs.FLP; Sp/fj^- ptc.Gal4;UAS.endoFt/+

63. ds^- fj^-: ds^-fj^-/ds^38K fj^N7

64. tub.Gal4 UAS.fj clones in ds^-: y w hs.FLP tub.Gal4 UAS.GFP/y w hs.FLP; ds^- tub.Gal80 FRT40/ds^- ck FRT40 UAS.fj

65. ft^- tub.Gal4 UAS.fj clones: y w hs.FLP;ft^- stc FRT39/tub.G80 CD2y⁺ FRT39;UAS.fj/tub.Gal4

66. ds^- ft^- tub.Gal4 UAS.fj clones: y w hs.FLP; ds^- ft^- stc FRT39/tub.G80 CD2y⁺FRT39; UAS.fj/tub.Gal4

67. ds^- tub.Gal4 UAS.fj clones: y w hs.FLP tub.Gal4 UAS.GFP/y w hs.FLP; tub.Gal80 FRT40/ds^- ck FRT40 UAS.fj

68. tub.Gal4 UAS.fj UAS.ds clones: y w hs.FLP tub.Gal4 UAS.GFP/y w hs.FLP; FRT42 tub.Gal80 CD2y⁺/FRT42D pwn UAS.fj; UAS.ds/+

69. tub.Gal4 UAS.fj UAS.ectoDs clones: y w hs.FLP tub.Gal4 UAS.GFP/y w hs.FLP; FRT42 tub.Gal80 CD2y⁺/FRT42D pwn UAS.fj;UAS.ectoDs/+

70. tub.Gal4 UAS.ft UAS.ds clones: y w hs.FLP122 tub.Gal4 UAS.GFP/y w hs.FLP; FRT42 tub.Gal80/FRT42 pwn; UAS.ft/UAS.ds

71. tub.Gal4 UAS.ft UAS.ectoDs clones: y w hs.FLP tub.Gal4 UAS.GFP/y w hs.FLP; FRT42 tub.Gal80/FRT42D pwn sha; UAS.ft/UAS.ectoDs

72. tub.Gal4 UAS.ft clones in fj^-: y w hs.FLP tub.Gal4 UAS.GFP/y w hs.FLP; FRT42 pwn fj^-/FRT42 tub.Gal80 fj^-; UAS.ft/+

73. tub.Gal4 UAS.ectoDs clones in fj^-: y w hs.FLP tub.Gal4 UAS.GFP/y w hs.FLP; FRT42 pwn fj^-/FRT42 tub.Gal80 fj^-; UAS.ectoDs/+

74. tub.Gal4 UAS.ds clones in fj^-: y w hs.FLP tub.Gal4 UAS.GFP/y w hs.FLP; FRT42 pwn fj^-/FRT42 tub.Gal80 fj^-; UAS.ds/+

75. ptc^- en^- clones in stan^-: y w hs.FLP tub.Gal4 UAS.GFP/y w hs.FLP; FRT42 pwn ptc^- stan^E59 en ^-/FRT42 tub.Gal80 stan³ CD2y⁺

76. ptc^- en^- clones in fz^-: y w hs.FLP tub.Gal4 UAS.GFP/y w hs.FLP; FRT42 pwn cn ptc^- en ^-/FRT42 tub.Gal80; fz^- CD2y⁺ ri FRT2A/fz^- ri FRT2A

77. ptc^- en^- clones in ds^-: y w hs.FLP/w; ds^- FRT42 CD2y⁺/ds^- FRT42 pwn ptc^- en ^-

78. ptc^- en^- clones in ds^-stan^-: y w hs.FLP tub.Gal4 UAS.GFP/y w hs.FLP;ds^- CD2y⁺ FRT42 pwn ptc^- stan^E59en ^-/ds^- FRT42 tub.Gal80 stan³CD2y⁺

79. tub.Gal4 UAS.wg clones in ds^-: y w hs.FLP;ds^- ck FRT40/ds^- tub.Gal80 FRT40; UAS.wg/+ and

80. y w hs.FLP; ds^- FRT42 pwn sha/ds^- FRT42 tub.Gal80 CD2y⁺; UAS.wg/tub.Gal4

81. tub.Gal4 UAS.Nrt::wg clones in ds^-: y w hs.FLP; ds^- ck FRT40/ds^- tub.Gal80 FRT40;UAS.Nrt::wg/+

82. tub.Gal4 UAS.fz2DN clones in ds^-: y w hs.FLP; ds^- ck FRT40/ds^- tub.Gal80 FRT40;UAS.fz2DN/+

83. tub.Gal4 UAS.Wnt2 clones in ds^-: y w hs.FLP; ds^- ck FRT40/ds^- tub.Gal80 FRT40;UAS.Wnt2/+ and

84. y w hs.FLP; ds^- FRT42 pwn sha/ds^- FRT42 tub.Gal80 CD2y⁺; UAS.Wnt2/tub.Gal4

85. tub.Gal4 UAS.Wnt3 clones in ds^-: y w hs.FLP; ds^- FRT42 pwn sha/ds^- FRT42 tub.Gal80 CD2y⁺; UAS.Wnt3/tub.Gal4

86. tub.Gal4 UAS.Wnt4 clones in ds^-: y w hs.FLP; ds^- ck FRT40/ds^- tub.Gal80 FRT40;UAS.Wnt4/+ and

87. y w hs.FLP; ds^- FRT42 pwn sha/ds^- FRT42 tub.Gal80 CD2y⁺; UAS.Wnt4/tub.Gal4

88. tub.Gal4 UAS.Wnt6 clones in ds^-: y w hs.FLP; ds^- FRT42 pwn sha/ds^- FRT42 tub.Gal80 CD2y⁺; UAS.Wnt6/tub.Gal4

89. tub.Gal4 UAS.Wnt8 clones in ds^-: y w hs.FLP; ds^- FRT42 pwn sha/ds^- FRT42 tub.Gal80 CD2y⁺; UAS.Wnt8/tub.Gal4

90. tub.Gal4 UAS.Wnt10 clones in ds^-: y w hs.FLP; ds^- FRT42 pwn sha/ds^- FRT42 tub.Gal80 CD2y⁺; UAS.Wnt10/tub.Gal4

91. ptc^- en^- stan^- clones in ds^-: y w hs.FLP; ds^- FRT42 pwn ptc^- en ^- stan^E59/ds^- FRT42 tub.Gal80; tub.Gal4/+

The dorsal epidermis of the adult abdomen is segmented and divided into a chain of anterior (A) and posterior (P) compartments. The epithelium secretes pigmented plates (tergites), made by the A compartments and separated by strips of more flexible cuticle; most of the cells make cuticular hairs or bristles that point posteriorly. Cells in the P compartment secrete the morphogen Hh that controls cell polarity (and cell type) in the A compartment. Here, we focus on the A compartment (Fig. 1). The vectors and extents of the gradients shown in Fig. 1 are derived from experiments with genetic mosaics: for example, just in front of a clone of ds^- cells, wild-type hairs point the `wrong' way(forwards). This, we argue, is because the normal grade of Ds activity (high at the back of the A compartment, low at the front), is locally reversed across the clonal border. At the back of the clone, the effects are concordant with the normal grade and therefore polarity is not altered. Similarly, clones of cells in which ds is overexpressed (henceforth called UAS.ds clones) make the hairs behind the clone point forwards,because, there, the normal grade of Ds activity is reversed. The corresponding experiments with fj and ft give similar results, except that the sign is opposite (ft^- and fj^-clones cause the polarity of wild-type cells to reverse behind the clones, and UAS.fj (Casal et al.,2002) and UAS.ft clones reverse in front of the clones). For the experiments described below, the genotypes are referred to by number(1-91; see also Fig. 8 for a summary of all results).

If the linear relationship were correct, cells that lack the Stan system should not support propagation of polarity changes caused by disparities in the Ds system. Indeed, in the eye, the repolarising abilities of fj^-, ds^- and ft^- clones all appear to be blocked in the absence of fz(Yang et al., 2002) (but see Discussion). However, experiments in the abdomen lead to a different conclusion. Stan is required in both `sending' and `receiving' cells for the transmission of polarising information induced by differences in Fz activity: stan^-, stan^- fz^- and stan^- UAS.fz clones do not repolarise their wild-type neighbours (genotype 1) (Lawrence et al.,2004) and neither UAS.fz nor UAS.fz UAS.stanclones repolarise surrounding stan^- cells (genotypes 2, 3, Fig. 2B). These experiments show that, with respect to PCP, the Stan system is completely disabled by the stan^- genotypes we have used (see Materials and methods). Nevertheless, we find UAS.ft clones in stan^-flies reverse the polarity of cells anterior to the clone, particularly posteriorly within the A compartment (genotypes 4, 5, Fig. 2D), as they do in wild-type flies (genotypes 6-8, Fig. 3A). In addition, ft^- clones in stan^- flies (genotype 9) can reverse the polarity of cells behind the clone, as they do in wild-type flies(Casal et al., 2002). The repolarisations caused by gain or loss of Ft in clones have a similar range in both wild-type and in stan^- flies, extending a few cell diameters away from the clones.

We find comparable results for Ds: UAS.ds clones have only weak effects in wild-type flies (genotype 10). However, a form of Ds that lacks the cytosolic domain (`ectoDs') is more potent, so that UAS.ectoDs clones usually reverse the polarity of wild-type cells behind the clone, with a range of several cells (genotype 11, Fig. 3C). We have used ectoDs to test whether repolarisation caused by ectopic Ds activity depends on the Stan system, and find that it does not: in stan^- flies, UAS.ectoDs clones reverse cell polarity strongly behind the clone (genotype 12, Fig. 2F). UAS.fjclones in stan^- flies (genotype 13) also repolarise in front, as they do in wild-type flies. Thus, at least in the A compartment,signals coming from UAS.ft, ft^-, UAS.ectoDs and UAS.fj clones are effective and can propagate over several cell diameters through stan^- territory. It follows that the Ds system has an intrinsic capacity to repolarise cells, even when the Stan system is incapacitated.

Our conclusion in the abdomen using stan^- contrasts with results in the eye, using fz^-(Yang et al., 2002). We therefore repeated the UAS.ft, ft^- and UAS.ectoDsexperiments described above in a fz^- background (genotypes 14-17). For UAS.ft clones in fz^- flies, we find that hairs in front are disturbed or reversed, although the effects are less consistent than in stan^- flies. For ft^- and UAS.ectoDs clones in fz^- flies, hairs behind are reversed, as observed in stan^- flies. We then made UAS.ft and UAS.ectoDs clones in stan^- fz^- flies(genotypes 18, 19) and, again, the clones repolarise nearby hairs in front and behind, respectively - the UAS.ectoDs clones have the strongest effects, reorienting the hairs around the clone over a long range (see Fig. S1 in the supplementary material). These results show that the Ds system can initiate and propagate PCP, even when the functions of both key components of the Stan system are abolished, indicating that the Ds system can confer and propagate PCP without the participation of the Stan system.

If the two systems were independent, the extent of repolarisation caused by disparities in one system might be limited or overcome by the normal and opposing action of the other system. Indeed, in ds^- wings, fz^- clones repolarise surrounding cells over a longer range than they do in wild-type wings(Adler et al., 1998). Similarly, an ectopic gradient of Fz expression repolarises cells over an increased range when Ft is absent (Ma et al., 2003). In agreement, we find that in the abdomen,repolarisations induced by fz^-, UAS.fz or UAS.stan clones show a longer range in ds^-(Fig. 2A; genotypes 20-23), and also when UAS.fz clones are made in ft^- (genotype 24), than they do in wild-type flies. In addition, if UAS.fz is driven in the entire P compartment (genotype 25), reversal at the back of the A compartment is greater in ds^- than in wild-type flies. Finally, a weak disparity in Fz activity that does not repolarise cells in wild-type flies is sufficient to repolarise cells over several cell diameters in ds^- flies (genotype 26, Fig. 4A). This same disparity can even induce a little repolarisation in ds^-/ds⁺ flies(Fig. 4B).

Normally, UAS.ft clones in the A compartment reverse the polarity of cells in front of the clone and do so most strongly when located at the rear of the compartment, where endogenous Ft is least active. Conversely, fz^- clones reverse the polarity of cells behind the clones, wherever they arise (Lawrence et al., 2004). Thus, in the A compartment, clones of fz^- UAS.ft cells (genotype 27) will create opposing disparities in the Ds and Stan systems, and send conflicting outputs to the adjacent wild-type cells. We find that, at the front of the A compartment,they reverse posteriorly, behaving like fz^- clones. At the back of the A compartment, however, fz^- UAS.ft clones reverse anteriorly, as do UAS.ft clones. This can be explained as follows. For the Stan system, repolarisation is driven by the difference in Fz activity across the clone/background interface,which appears to be of similar strength all along the AP axis(Fig. 1). For the Ds system,the strength of the disparity in Ft activity between UAS.ft clones and the surround depends on position, being least at the front and greatest at the back of the A compartment (Fig. 1). Thus, in the anterior region, the repolarisation caused by Fz overcomes the weaker opposing influence of UAS.ft. At the rear of the A compartment, the effect caused by the Ds system is the stronger.

UAS.fz clones in wild-type flies reverse polarity in front of the clone, creating a conflict with the Ds system: this conflict appears to limit the range of repolarisation caused by such clones, as that range increases in ds^- flies. In fz^- flies, UAS.fz clones change the polarity of only the adjacent cells(Lawrence et al., 2004). If UAS.fz clones in ds^- flies were using only the Stan system to drive long-range repolarisation, then UAS.fz clones in ds^- fz^- flies should behave exactly as they do in fz^- flies, and they do: only one cell is repolarised(genotype 28).

Disparities in the Ds system do not bias the Stan system The experiments above show that the Ds system can polarise cells independently of the Stan system. However, the Stan system might still be biased by the Ds system. To assess whether there is normally any input from the Ds system into the Stan system, we generated clones expressing UAS.ectoDs, UAS.ds or UAS.ft in ds^- flies (genotypes 29-34), and also clones expressing UAS.ectoDs or UAS.ft in ft^- flies (genotypes 35-36), and asked whether such clones repolarise surrounding mutant cells. The responding mutant cells are particularly sensitive to small disparities in activity of the Stan system(Fig. 4A); hence, if these types of clones were to bias the Stan system, either within the clone or across the border, they should repolarise the surround, in either ds^- or in ft^- animals. Nevertheless,they do not, not even changing the polarity of one cell in either ds^- (Fig. 2C,E) or in ft^- flies. We know that UAS.ds, UAS.ectoDs and UAS.ft are effective constructs even in the absence of endogenous Ds and Ft - when these constructs are expressed in ds^- ft^- clones, they repolarise surrounding wild-type cells (see below). As positive controls, we added UAS.fzseparately to both UAS.ft and UAS.ds clones in ds^- flies (genotypes 37, 38) and then the long-range repolarisation normally induced by UAS.fz clones in ds^- flies (Fig. 2A) was seen. Likewise, when UAS.fz was added to clones expressing UAS.ft in ft^- flies, these clones again caused long-range repolarisation (genotype 39). Thus, the failure of UAS.ds, UAS.ectoDs, and UAS.ft clones to repolarise surrounding cells in ds^- or ft^-animals argues that the Stan system is not biased by the Ds system.

Cell polarity in the absence of both the Ds and Stan systems If the Ds and Stan systems give independent inputs into PCP, the loss of either system might compromise polarity, but the loss of both systems should cause more damage. This is so: stan^- flies have almost normal hair polarities in the tergite, apart from near the front and near the rear(genotype 40; Fig. 5C); and in ds^- tergites, hair polarities are normal apart from whorls in the middle (genotype 41; Fig. 5A). The phenotype of ds^- stan^-flies is more extreme than in either ds^- or stan^-, and hair and bristle polarity is randomised throughout the tergite (genotype 42, Fig. 5B). Similar results are observed for the ventral dentical pattern of the third instar larva: the double mutant condition is more severe than in either single mutant (Fig. 5E-G).

We now ask how the Ds system, when acting on its own, can affect PCP. The Ds system has three components and all appear to be graded in activity(Fig. 1). Either ds^- or ft^- clones can initiate polarity changes that spread into wild-type territory(Casal et al., 2002), but clones that lack both ds and ft do not cause repolarisations(genotype 43). Adding back either UAS.ds or UAS.ft to ds^- ft^- clones restores their ability to repolarise, with UAS.ds reversing polarity behind the clone and UAS.ft in front (genotypes 44, 45). These results suggest that an imbalance (from the normal ratio) of Ds and Ft proteins in the `sending' cells changes polarity in the wild-type `receiving' cells that then spreads further. The sending cell, in particular, does not need both Ds and Ft in order to repolarise nearby wild-type cells - the presence of either protein alone will do so.

ds^- or ft^- clones both cause polarity changes in neighbouring wild-type cells. However, inside regions of such clones, the hairs are oriented in whorls, resembling small regions of entire ds^- or ft^- flies(Casal et al., 2002) (J.C.,P.A.L. and G.S., unpublished), suggesting that the polarity outside the clone cannot propagate into territory lacking either Ds or Ft. Other experiments confirm this: as we have seen, UAS.ds, UAS.ectoDs and UAS.ftclones in ds^- flies all fail to repolarise, not even changing the polarity of those ds^- cells adjacent to the clone (Fig. 2C,E). Moreover, UAS.ectoDs and UAS.ft clones in ft^-flies also fail to repolarise any ft^- cells outside the clone. Together, these experiments show that cells need both Ds and Ft in order to receive and respond to a polarity signal initiated by the Ds system,even when that signal comes from immediate neighbours.

As described above, UAS.ectoDs clones repolarise surrounding cells as do UAS.ds clones (reversing behind), only more potently(Fig. 2F, Fig. 3C). The same is true for UAS.ectoDs clones that are also ds^-,ft^- or ds^- ft^- and therefore lack one or both the endogenous proteins (genotypes 46-48) - presenting the Ds ectodomain on the surface of the sending cell is alone sufficient to change the polarity of the receiving cells. However, the Ft ectodomain cannot act alone: although UAS.ectoFt clones (genotype 49) behave similarly to UAS.ft and ft^- UAS.ft clones (genotype 50), ft^- UAS.ectoFt and ds^- ft^-UAS.ectoFt clones (genotypes 51, 52) behave, respectively, like ft^- or ds^- ft^- clones. Thus, the capacity of ectoFt to repolarise nearby cells also requires endogenous Ft in the sending cell, supporting suggestions that Ft may form cis-homodimers (Matakatsu and Blair,2006).

Can the cytosolic domains influence the sign of the signal? We swapped them to make two chimaeric molecules, ectoDs::endoFt and ectoFt::endoDs and found the answer to be no. Clones expressing these proteins behaved as if they expressed the native protein with the same ectodomain, reversing hairs behind strongly (ectoDs::endoFt, genotypes 53-56) or in front (ectoFt::endoDs,genotypes 57-60), either when expressed in cells that were otherwise wild type, or were ds^-, ft^- or ds^-ft^-. However, the Ds and Ft endodomains are not always interchangeable: the endodomain of Ft cannot substitute for that of Ds in limiting the potency of the signal (UAS.ectoDs and UAS.ectoDs::endoFt clones repolarise strongly, whereas UAS.ds clones repolarise weakly). Nevertheless, the endodomain of Ds can substitute for the endodomain of Ft to allow the ectoFt protein to signal in the absence of endogenous Ft: ds^- ft^-UAS.ectoFt::endoDs clones reverse the polarity of cells in front of the clone, whereas ds^- ft^- UAS.ectoFt clones do not. We also made forms of Ds and Ft that lack the ectodomains(UAS.endoDs and UAS.endoFt). If endoDs or endoFt are expressed in wild-type cells (genotypes 61, 62), we see no alteration in polarity - however, some rescue of polarity was reported when endoFt was expressed in a ft^- mutant background(Matakatsu and Blair, 2006). The key finding is that Ds and Ft can each signal on their own, and that the nature of that signal is governed by the ectodomain.

Fj acts in a graded fashion and appears to repress Ds and promote Ft activity (Zeidler et al.,1999; Casal et al.,2002; Yang et al.,2002). In the abdomen, ds^- fj^-flies (genotype 63) resemble ds^- flies, and UAS.fj clones have no effect on polarity in ds^-flies (genotype 64). UAS.fj clones in the tergite normally repolarise wild-type cells in front (Casal et al.,2002), but UAS.fj clones that are also ft^- or ds^- ft^- do not(genotypes 65, 66). These findings indicate that Fj works through Ds and/or Ft to polarise cells.

However, other results argue that Fj works specifically through Ft and not Ds: unlike ft^- UAS.fj clones, ds^-UAS.fj clones repolarise strongly in front (genotype 67), more strongly than clones that are simply ds^-. In addition, UAS.fj clones behave like UAS.ft clones and reverse the polarity of cells in front, even when they co-express UAS.ds(genotypes 68,70) or UAS.ectoDs (genotypes 69,71). Thus, Fj appears to promote Ft to signal, irrespective of whether Ds is absent, or whether it is overexpressed.

To gain more insight into Fj, we made UAS.ft, and UAS.ectoDs clones in fj^- flies (genotype 72 and 73). The lack of Fj enhances the effects of both proteins: repolarisations can spread further than in any other situation we have seen, with a range of up to about 10 cells (Fig. 3B,D). By contrast, the action of UAS.ds clones is not enhanced in fj^- flies (genotype 74).

According to the linear model of PCP, morphogens such as Hh in the abdomen or Wg in the eye, control polarity by establishing gradients of the Ds system,which then bias the Stan system. But, if the Ds and Stan systems are independent, we must now ask does Hh signalling bias both systems, or only one? To answer this, we used clones of patched^-(ptc^-) cells in which the Hh transduction pathway is constitutively activated in all cells within the clone. Unfortunately, ptc^- clones can cause complex effects by ectopically inducing engrailed (en), leading to a Hh-secreting P compartment forming near the middle of the A compartment(Struhl et al., 1997b; Lawrence et al., 1999)! We avoided these problems by using ptc^- en^- clones(Lawrence et al., 1999; Lawrence et al., 2002). Such clones reverse the polarity of wild-type cells behind the clone, allowing us to test whether activation of the Hh transduction pathway can polarise cells via either, or both, the Stan and Ds systems.

ptc^- en^- clones cause reversal of polarity behind in stan^- (genotype 75, Fig. 6C), fz^- (genotype 76), and ds^- flies(genotype 77, Fig. 6A). However, ptc^- en^- clones do not reverse polarity in ds^-stan^- flies (genotype 78, Fig. 6B). It follows that Hh signalling polarises cells in the tergite largely, or only, via the Ds and Stan systems, and that it does so by means of two distinct inputs into PCP.

For the Ds system it seems that Hh governs cell polarity, at least in part,by driving the graded expression of the transcription factor Omb(Lawrence et al., 2002), which(probably) controls transcription of ds. For the Stan system, Hh presumably biases the activity of Fz(Lawrence et al., 2004) but it is not clear how it does so. It did not escape anyone's notice that Fz is a Wnt receptor and therefore many suggested that Wg or some other Wnt might be an intermediary. Several experiments argued against this possibility(Wehrli and Tomlinson, 1998; Lawrence et al., 2002), but they were all carried out in wild-type flies, where an active Ds system might have blocked any effect. Therefore, we made clones of cells that express UAS.wg, UAS.Nrt::wg (a membrane-tethered form of Wg), UAS.fz2DN (a membrane-tethered form of the Wg-binding domain of Fz2 to manipulate the distribution of Wg) and the remaining six Drosophila Wnts (UAS.Wnt2, 3, 4, 6, 8 and 10) in ds^- flies, but they induced no repolarisation (genotypes 79-90). These results argue against all known Wnt genes, notably Wg itself, as being polarising factors for the Stan system.

Many epithelia exhibit planar cell polarity (PCP), but examples from Drosophila have been studied in most depth (reviewed by Klein and Mlodzik, 2005). It was proposed long ago (Lawrence,1966; Stumpf,1966) that the vectors of a pervasive gradient orient PCP and here we examine how this is achieved. In the current and prevailing model, a morphogen gradient (for example, Hh or Wg) organises the expression of fj and ds to set up Ds system gradients(Casal et al., 2002; Simon, 2004). Then, small differences in Ds system activity from one cell to the next are thought to feed into Fz and bias the Stan system. The Stan system is then thought to act more directly on the cell to orient structures, such as ommatidia or hairs(Yang et al., 2002; Ma et al., 2003). Here, we test this model in the abdomen and find our results do not support the main part of it; instead they argue that the morphogen gradient acts separately on the Ds and Stan systems to generate two independent inputs into PCP.

The case for the Ds system polarising cells via the Stan system rested on epistasis experiments in the eye: disparities in the Ds system, such as clones of ds^- or ft^- cells, repolarise cells in wild-type flies, but not in fz^- flies. This requirement for Fz suggested that the Ds system might act via Fz(Yang et al., 2002). However,we find that, in the dorsal abdomen, the Ds system can polarise cells without the Stan system. We present several lines of evidence, but the most crucial is that clones of UAS.ft or UAS.ectoDs cells, both of which repolarise surrounding wild-type cells up to several cell rows away, also do so in stan^-, fz^- or stan^-fz^- flies. It follows that the Ds system, acting alone and using Ds and Ft, can drive changes in the polarity of surrounding cells. This conclusion raises new questions: how does the Ds system produce and propagate polarising information without any involvement of the Stan system? What polarises the Stan system? How do cells integrate the two separate inputs from the Ds and Stan systems?

The discovery that fz^- clones can change the polarity of nearby wild-type cells was important(Gubb and Garcia-Bellido,1982; Vinson and Adler,1987) and many attempts have been made to explain it: most models invoke feedback to amplify initial biases in Fz activity, within or between cells. Now we have shown that, independently of the Stan system, disparities in the Ds system can repolarise cells; yet the two systems employ fundamentally different molecules. How does the Ds system act?

First, morphogen gradients (Hh in A, Wg in P)(Lawrence et al., 2002) appear to polarise the Ds system by grading the amount and/or state of activity of three components of the system: Ds, Ft and Fj(Casal et al., 2002). Second,we find that cells can `send' information by presenting either Ds or Ft to`receiving' neighbours. Thus, both Ds and Ft appear to have ligand-like signalling activities that can repolarise surrounding cells. This signal appears to depend on the ratio of Ds to Ft in the sending cell (in the tergite, hairs made by the receiving cell point towards neighbours with a higher Ds/Ft ratio). It is not clear how this ratio is encoded but it presumably determines how much free Ds or Ft the sending cell presents to neighbours (Fig. 7). Third, we have shown that in order to respond to this signal by changing their polarity,the receiving cells need both Ds and Ft, indicating that Ds and Ft both have receptor-like and ligand-like properties and defying any simple categorisation of Ds as a ligand and Ft as a receptor. More relevant, perhaps, is the evidence that Ds and Ft can form trans-heterodimers that bridge adjacent cells both in culture and in vivo, and that Ds or Ft proteins become concentrated along cell interfaces in which the abutting cell presents only Ft or Ds,respectively (Strutt and Strutt,2002; Ma et al.,2003). Furthermore, accumulation of either Ds or Ft along one cell surface, in response to excess Ft or Ds presented on the abutting surface, may lead to the depletion of Ds or Ft along the remaining surfaces of the same cell (Strutt and Strutt, 2002; Ma et al., 2003), localising and limiting the potential to form trans-heterodimeric bridges with other cells. These properties suggest a model in which Ds and Ft are required in the receiving cells both to respond to and to propagate polarising information(Fig. 7). For example, in UAS.ft clones, the more active Ft is presented by the sending cell,the greater amount of Ds would be drawn to the facing membrane of the receiving cell, leaving less Ds and more free Ft on the opposite face of the receiving cell (Fig. 7). Fourth, we ask how the amplitude of the signal is determined. The range depends on where (in the compartment) the clones are made, indicating that the degree of discrepancy between Ft and Ds levels in the clone and in the surrounding cells is a key factor. The range of repolarisation also depends on Fj, possibly acting on Ft to promote the formation of heterodimers. Thus, with UAS.ft clones in a fj^- background, in which heterodimers should be sparse because the activity of Ft is low, there would be a large discrepancy across the clone border that should produce a long-range effect, as observed. The same clones in a wild-type background should have a smaller discrepancy and therefore a shorter range(Fig. 7). In both wild-type and fj^- flies, excess ectoDs sends a much stronger signal than excess Ds, suggesting that the cytosolic domain may have an inhibitory function.

At first sight, the tergites might seem exceptional, for here the Ds system can polarise cells in the absence of the Stan system - yet neither in the ventral pleura nor in the wing do UAS.ft or UAS.ectoDsclones repolarise cells that lack the Stan system. Thus, we now ask whether our results represent a fundamental property that is obscured in other places,or a special case that applies only to the tergite. Our results tell that the Ds system has an inherent capacity to confer and propagate PCP, and we rate this positive result as decisive, suggesting that the apparent failure of the Ds system to act independently in other parts of the fly could be explained in other ways. There are several possible explanations.

First, if cells normally integrate separate inputs from the Ds and Stan systems, the lack of one system might, in some places, interfere with the response to the other system. For example, in the pleura, as in the eye,polarity is randomised in the absence of the Stan system(Zheng et al., 1995; Wehrli and Tomlinson, 1998; Yang et al., 2002; Lawrence et al., 2004) and it may be impossible for the Ds system to reorganise polarity where there is such a strong requirement for the Stan system. Second, there are qualitative differences in the outputs of the two systems: the Ds system being involved in growth, cell shape and cell affinity(Bryant et al., 1988; Adler et al., 1998; Matakatsu and Blair, 2006);the Stan system not affecting these properties and instead possibly placing asymmetric structures, such as actin filaments. These differences might help explain why the Ds system can, even in the absence of the Stan system,reorient hairs in some tissues. Third, experiments that create conflicts between the Ds and Stan systems can lead to varying outcomes even in the tergite, depending on the location of the clones (e.g. fz^-UAS.ft clones, see Results). Perhaps cell polarity is a composite property (like height in humans!): the orientation of hairs being the deceptively simple outcome of diverse inputs. At the least our results show the linear pathway, Ds system→Stan system, is wrong in the tergite and challenge its universality.

The behaviour of ptc^- en^- clones is pertinent because they repolarise surrounding cells by means of both systems. In wild-type flies, these clones reverse behind in the A compartment. The type of cuticle made by ptc^- en^- clones corresponds to the back of the A compartment and it is here that we believe the Ds activity should normally peak and Ft activity should be minimal(Casal et al., 2002) - thus, it makes sense for ptc^- en^- clones to resemble UAS.ectoDs or ft^- clones. Similarly, as cells in the tergite make hairs that point towards neighbours with lower Fz activity,it makes sense that ptc^- en^- clones behave like fz^- clones: this is because all hairs in the wild-type A compartment point towards the back of the compartment, where Hh signalling peaks and where our model calls for Fz activity to be minimal(Lawrence et al., 2004).

The ability of ptc^- en^- clones to repolarise surrounding cells in ds^- flies provides an intriguing hint as to how Hh signalling might feed into the Stan system: we have made ptc^- en^- stan^- clones and these clones do not repolarise in ds^- flies (genotype 91), in contrast to ptc^- en^- clones. This result suggests that Hh might polarise the Stan system by acting via Ptc to regulate Fz activity, a mechanism that would depend on the ptc^-en^- cells communicating their altered level of Fz activity to their wild-type neighbours via Stan. If this were so, then Hh would be a component of the elusive Factor X!

Finally, we need to address why the Stan system proteins can be induced to form abnormal asymmetric distributions by manipulating the Ds system; for example, ft^- clones in the wing contain abnormally polarised cells that also show corresponding changes in the distribution of Dishevelled (Strutt and Strutt,2002; Ma et al.,2003). For us this presents no problem, as we have argued that the asymmetric accumulation of the Stan system proteins is an outcome not a cause of polarity (see Introduction) (see also Lawrence et al., 2004). Hence,if cells are reoriented by perturbing the Ds system, whatever polarity they adopt will show in both the asymmetric localisation of Stan system proteins and in the orientation of the hairs.

The Ds and Stan system gradients are not congruent - yet another argument that they are independent. The Ds system consists of two gradients with opposing slopes: the Ds activity peaking at the back of the A compartment, and declining forwards into the A compartment and backwards into P(Fig. 1)(Casal et al., 2002). By contrast, the Stan system appears to be a monotonic gradient of Fz activity with A and P cells both pointing down the gradient. An unsolved problem is the registration of a Fz activity gradient that presumably repeats once per metamere: do its borders coincide with segmental or parasegmental borders? We do not know, but two systems with different spatial registrations may solve the tricky problem of how cell polarity is maintained across boundaries.

We thank Simon Bullock, David Strutt and Jean-Paul Vincent for comments on the manuscript; and Seth Blair and Bloomington for stocks. David Strutt has been very generous with both advice and stocks. We thank Atsuko Adachi, Kit Bonin and Xiao-Jing Qiu for assistance in New York. Birgitta Haraldsson andthe Zoology Department, University of Cambridge have given invaluable support. P.A.L. and J.C. have been supported by the MRC; G.S. is an HHMI Investigator.