How do Dachsous and Fat polarise

10 We investigate the mechanisms of planar cell polarity (PCP) in the larval epidermis of 11 Drosophila. Measurements of the amount of Dachsous across the segment find a peak 12 located near the rear of the anterior compartment. Localisation of Dachs and 13 orientation of ectopic denticles reveal the polarity of every cell in the segment. We 14 discuss how well these findings evidence a zigzag gradient model of Dachsous activity. 15 Several groups have proposed that Dachsous and Fat fix the direction of PCP via 16 oriented microtubules that transport PCP proteins to one side of the cell. We test this 17 proposition in the larval cells and find that most microtubules grow perpendicularly 18 to the axis of PCP. We find no significant bias in the polarity of those microtubules 19 aligned close to that axis. We also reexamine published data from the pupal abdomen 20 and fail to find evidence supporting the hypothesis that microtubular orientation 21 draws the arrow of PCP. 22


23
Test 1 Testing the model in the wildtype and polarity modified larvae 117 (i) Polarity of typical cells 118 The model posits that the polarity of each cell, or part of cell, depends on a 119 comparison of mutually facing membranes. Thus, in the wildtype, a low activity of Ds 120 due in part to high expression of fj in T1 and T2, orients the denticles in the adjacent 2). In the polarity modified larvae, we engineer increased expression of ds in T1 and 125 T2 cells; this changes the landscape of Ds activity, making peaks (instead of troughs, 126 as in the wildtype) in T1 and T2. Consequently, the polarities of rows of cells 1, 2, 4 127 and 5, that abut T1 and T2, now point inwards, that is reversed from the wildtype 128 (Figure 2A-C). The other rows, 0, 3 and 6 could also be affected because polarity can  Table 1). However for these larvae, some 148 single atypical cells of row 2 have two anterior neighbours, -cells of T1 and row 2-149 that are higher and lower in Ds activity than the posterior neighbour of the atypical 150 cell, respectively. Consequently, the model predicts that their associated predenticles 151 should point forwards in that part of the cell that abuts T1 and backwards in that part 152 of the same cell abutting row 2, and they do (Figure2 G-I and Table 1). There are 153 some quantitative differences between the current data and the wildtypes we scored  assessed so long as the neighbour(s) does not contain any tagged D. 182 We examine the distribution of D in wildtype larvae in order to test the PCP 183 model and to reveal the molecular polarity of cells that lack denticles (Figure 5 and 6). 184 In the P compartment, all the denticulate and undenticulate cells show a consistent What is the distribution of D in larvae that lack the Ds or Ft protein (dsor ft -?

209
To our surprise, although D distribution was variable in location and was more diffuse 210 than the wildtype, D can be asymmetrically localised at the membrane, but in a 211 disorganised way. We found no consistent difference between dsand ftlarvae 212 (Figure 5-figure supplement 1B, C). Thus it appears that some asymmetric 213 localisation of D can occur independently of Ds-Ft heterodimers (which should be 214 missing in dsand ftlarvae).

215
Test 4. ovo-expressing clones reveal otherwise unseen polarity. 216 Small clones that overexpress ovo in naked areas often produce denticles in embryos . We made marked clones in larvae and these 218 also generally made denticles. The denticles showed a consistent orientation, pointing 219 forwards in P and backwards in most of A, exactly mirroring the polarity pattern as 220 identified by D localisation (Figure 7, compare with Figure 6). Thus, cell of row 11 at 221 the back of the A compartment mostly made denticles that pointed forwards ( Figure   222 7) as is characteristic of cells belonging to the P compartment. Just as signalled by the 223 localisation of D, in a minority of row 11 cells, polarity was ambiguous with denticles 224 pointing in various directions (Figure 7-figure supplement 1). The denticles 225 belonging to the cell row 10 in front of row 11 always pointed backwards and denticles 226 of the row behind row 11 (row -2 of the P compartment) always pointed forwards.

227
Test 5. Does the orientation of growing microtubules correlate with PCP? 228 We study the orientation of growing microtubules (using EB1 comets Schuyler and  240 In the wing, the predominant alignment of the microtubules is close to the axis 241 of PCP (Harumoto et al., 2010; Gomez et al., 2016). By contrast, in the larval 242 epidermal cells, in both A and P compartments, the majority of the microtubules are 243 aligned perpendicular to the anteroposterior axis, the axis of PCP (Figure 8A,B). To 244 analyse our data and following the approach in the wing, the comets of the larvae are 245 sorted into four 90 degree quadrants centred on the anteroposterior and mediolateral 246 axes and their frequencies plotted. The quadrants are described as "anterior",

247
"posterior", "medial" and "lateral" (Figure 8C,D). The axis of PCP lies in the 248 anteroposterior axis, but, in A compartment cells, 66% of the total angles of growth 249 fall within the medial and lateral sectors, while in the P compartment the comparable 250 figure is 71%. Clearly there is no overall correlation between microtubular orientation 251 and PCP and this belies the hypothesis that microtubular orientation is causal for 252 PCP.

253
However, we could look for a limited correlation between the orientation of 254 growing microtubules and the direction of PCP. For example, considering only the 255 minority of microtubules within the anterior and posterior sectors, we find small 256 differences in polarity but they lack statistical significance (Figure 8C,D) (Figure 8C,D). 265 How uniform are the individual cells? To answer we group all the growing 266 microtubules according to which cell (and larva) they come from and according to 267 which of four 90 degree quadrants they fall into ( Figure 8E). Remarkably, in all sets,    (Figure 8-figure supplement 2C,D). 312 We then plotted all the growing microtubules according to which pupa they   cell has a higher level of Ds, this Ds will bind more Ft in the abutting cell membrane, 376 and likely tend to exclude Ds from that membrane. These effects will tend to even out 377 the amounts of Ds in joint membranes and therefore tend to disguise any gradients, 378 local peaks or troughs.  Originally predicted to be at the A/P compartment border (Casal et al., 2002) 391 we conclude now that a Ds peak occurs two cells in front of that border, in row 10 ( Figure 9; a similar peak two cells from the A/P border has been described in the dorsal abdomen of the pupa Arata et al., 2017). This observation is supported by both 394 D localisation and the orientation of ectopic denticles formed by ovo-expressing 395 clones. There are interesting implications: the peak in Ds protein at the cell junctions 396 is in a cell that is flanked on both sides by A compartment cells, the most posterior of 397 which (row 11) has "P type" polarity. Why is this summit out of register with the 398 lineage compartments? It could be that that peak is specified by a signal emanating 399 from one compartment and crossing over to affect the next compartment. There are  448 and in different cell shapes in the A and P cells; these are more obvious at or close to 449 the A/P border (Umetsu et al., 2014). 450 The hypothesis of Uemura's group which proposes that microtubules transport   ( Figure 1-figure supplement 1B) (Figure 3, Figure 4) w; ds::EGFP FRT40A. 487 (Figure 4-figure supplement 1) Figure 5A,B, Figure 6) y w hs.FLP/ w; en.Gal4 UAS.DsRed/ +; act>stop>d::EGFP/ +. 490 ( Figure 5C,D, Figure 5-figure supplement 1A, Figure 6) (Figure 8, Figure 8-figure supplement 1,3, Figure 8-movie supplement 1,2)                The strong Ds accumulation on both sides of T3 tendon cells (Figures 3 and 4) 956 suggests that ds expression is high in T3 itself and/or its neighbours. In addition,

1A)
show that polarity of row 10 points backwards, away from T3, implying that Ds 959 activity is higher in row 11 than in T3. These two observations combined argue that ds 960 expression peaks in row 10, two cells in front of the A/P border, with Ds activity also 961 high in T3 and row 11. Graded ds expression forwards and backwards from this peak