So far, we have presented evidence against cell-scale polarity relying on a ‘depletion’ mechanism (i.e. not highly sensitive to core protein levels) or an MT transport mechanism. We have also observed that at the cellular level, over a period of 6 hr an antero-posterior boundary of Fz overexpression can only repolarise core protein localisation for about one to two cell rows (Figure 4O, S, Figure 4—figure supplement 1H). This short range of repolarisation was unexpected, especially as the same induction regime results in repolarisation of trichomes in the adult wing over at least five rows (Figure 5C), consistent with previous reports (Wu and Mlodzik, 2008). To investigate further, we examined the sites of trichome emergence in pupal wings at 33 hr APF under similar induction conditions. With this longer induction time, we observed trichomes being repolarised for about three to four cell rows (Figure 5—figure supplement 1A), but still less far than observed in adult wings (see Discussion).
Reorientation of planar polarity from a boundary under control and de novo conditions.
Images of dorsal surface of adult Drosophila wings, taken proximal to the anterior cross vein between longitudinal vein 3 and longitudinal vein 4, showing trichome orientation in control condition with constitutive Fz::mKate2-sfGFP expression (A), in de novo Fz::mKate2-sfGFP induction condition (B), in repolarisation from the hh-GAL4 boundary with constitutive Fz::mKate2-sfGFP expression condition (C), and in repolarisation from the hh-GAL4 boundary under de novo Fz::mKate2-sfGFP induction condition (D). Repolarisation conditions are under UAS-GAL4 control for Fz overexpression in the posterior wing (hedgehog expression domain), with induced Fz expression from UAS-FRT-STOP-FRT-Fz in the presence of hsFLP. Note, the swirling pattern of polarity generated by de novo induction generates a proximo-distal trichome orientation in this analysed wing area (B) but not in surrounding wing regions (Figure 5—figure supplement 1B–D). Trichome polarity is reoriented in the antero-posterior direction in both repolarisation conditions. See Figures 4M and 1C’ for vein locations and imaged wing area. See Supplementary file 1 for the full genotypes. Scale bar, 50 µm (E–H) Planar polarity measurement at the cellular scale in the region proximal to the anterior cross vein, between longitudinal vein 3 and longitudinal vein 4 in fixed pupal wings at 28 hr APF. The length and orientation of cyan bars denote the polarity magnitude and angle for a given cell, respectively. De novo condition with 6 hr induction of Fz::mKate2-sfGFP (E), constitutive expression of Fz::mKate2-sfGFP (F), repolarisation condition with constitutive expression of Fz::mKate2-sfGFP (G), and repolarisation condition with 6 hr de novo induction of Fz::mKate2-sfGFP (H). In repolarisation conditions, induced Fz is over-expressed in the posterior wing, whereas in the anterior wing only Fz::mKate2-sfGFP is expressed. In the posterior wing, the two Fz populations (over-expressed Fz and Fz::mKate2-sfGFP) compete for the same membrane locations and Fz::mKate2-sfGFP signal is not evident. Scale bar, 10 µm (I–L) Circular plots of quantified total Fz (Fz::sfGFP) magnitude and orientation relative to horizontal axis in the region proximal to the anterior cross vein, between longitudinal vein 3 and longitudinal vein 4 at 28 hr APF in fixed pupal wings. Small dots show polarity angle and magnitude for individual wings, arrows show average polarity and magnitude across all wings. Cells are grouped in rows relative to their location relative to the Fz overexpression domain in Fz repolarisation condition or relative to longitudinal vein 4 without Fz repolarisation, with row 0 in contact with Fz overexpression boundary and row 3 furthest away. (I–I’’’) de novo condition with 6 hr to establish core protein polarity, (J–J’’’) control condition with constitutive Fz::mKate2-sfGFP expression, (K–K’’’) repolarisation condition for 6 hr to re-orient core protein polarity, and (L–L’’’) repolarisation under de novo condition for 6 hr induced Fz::mKate2-sfGFP expression; in row 0 (I–L), in row 1 (I’–L’), in row 2 (I’’–L’’), and in row 3 (I’’’–L’’’). Vertical black lines associated with p values on right of each column represent comparisons between different rows of the same polarisation condition. Horizontal pale blue lines associated with p values represent comparison between the two adjacent polarisation conditions for the same cell row. On the far right, horizontal red underlined p values represent comparison between repolarisation in de novo condition versus de novo condition (far left) for the respective rows of cells. Hotelling’s T-square tests were used to compare total Fz polarity (orientation and magnitude).
We hypothesised that the limited ability of a Fz overexpression boundary to repolarise may be due to three factors. First, there might be a strong and persistent influence of a global cue specifying proximo-distally oriented polarity. Second, there might be strong cell–cell coupling of polarity between neighbours. Hence, a local perturbation, e.g. overexpression of Fz in a neighbouring row of cells, has little effect on the polarity of neighbouring cells, as their polarity is already strongly coupled to their neighbours in the adjacent region of non-overexpression. Third, cells might have a strong intrinsic ability to polarise, while cell–cell coupling of polarity with neighbours might be weak. In this case, once established, polarity is highly robust to the effects of Fz overexpression in neighbouring cells due to relatively weak effects of cell–cell coupling.
In terms of examining these possibilities, we have already attempted to find conditions that weaken (but do not break) the cell-intrinsic polarisation (Figure 3) by altering core protein levels, but this was unsuccessful. Moreover, there is no reported way to weaken cell–cell coupling of polarity. Interfering with global cues also presents a challenge, given that they are poorly characterised. However, current evidence suggests that de novo induction at times following the onset of hinge contraction results in a swirling polarity pattern that is not influenced by global cues (see Introduction), a possibility that we investigate further below.
Hence, we decided to compare repolarisation, induced as previously with hh-GAL4, in a control condition with constitutive Fz::mKate2-sfGFP expression (where global cues set up proximo-distal polarity), to repolarisation in the de novo condition produced by Fz::mKate2-sfGFP induction at 22 hr APF, where cells are in the process of polarising. Our prediction was that de novo polarity would be easy to orient/repolarise by a boundary of Fz overexpression, as cells (and their neighbours) should not have a pre-existing polarity.
We first looked at the polarity of trichomes in the adult wing. In the control condition of constitutive Fz::mKate2-sfGFP expression, proximal to the anterior cross vein between longitudinal vein 3 and longitudinal vein, as expected, trichomes are oriented along the proximo-distal axis (Figure 5A). We found that in this wing region, trichomes also point proximo-distally after de novo polarisation induction of Fz::mKate2-sfGFP expression at 22 hr APF (Figure 5B). To investigate the possibility that this proximo-distal polarity in the de novo condition might be due to the presence of a previously unappreciated proximo-distal global cue, we examined the trichome polarity in surrounding regions of the wing. Notably, regardless of whether Fz::mKate2-sfGFP expression was induced at 18, 20, or 22 hr APF, we observed a similar trichome swirling pattern in the proximal wing, with regions proximal, distal, anterior and posterior to the experimental region showing non-proximo-distal adult trichome polarity (Figure 5—figure supplement 1B–D). Moreover, quantification of polarity at 28 hr APF in the experimental region showed no difference in the degree of proximo-distal polarisation with longer induction times (Figure 5—figure supplement 1E). These observations argue against a proximo-distal global cue being active in this region of the wing at this developmental stage. Instead, we surmise that the proximo-distal polarity in the experimental region is part of the normal trichome swirling pattern in this wing region. Moreover, we can expect to see repolarisation of this proximo-distal polarity to antero-posterior polarity if we overexpress Fz in the posterior compartment.
Interestingly, induction of repolarisation in both the control and de novo conditions resulted in similar degrees of partial trichome repolarisation on the antero-posterior axis (Figure 5C, D). In neither case was trichome repolarisation seen beyond longitudinal vein 3. This result does not fit our prediction of longer-range repolarisation in de novo.
We then compared Fz distribution in this region of pupal wings at 28 hr APF in the same four conditions: control constitutive Fz::mKate2-sfGFP expression, 6 hr of de novo Fz::mKate2-sfGFP polarity establishment, 6 hr of repolarisation caused by Fz overexpression in the posterior compartment, and 6 hr of de novo polarity with 6 hr of repolarisation. As we saw in the adult wing, de novo polarity was broadly proximo-distal and similar in magnitude to the control polarity establishment in this region (Figure 5E, F, I–J’’’).
Surprisingly, in both the repolarisation (Figure 5G, K–K’’’) and repolarisation in de novo conditions (Figure 5H, L–L’’’), only row 0 was strongly repolarised (Figure 5J vs K p = 0.0014, Figure 5L vs I p = 0.0011). In the repolarisation only condition, there was no significant change in polarity in row 1 or 2 (Figure 5J’ vs K’ p = 0.2945, Figure 5J’’ vs K’’ p = 0.2258) although unexpectedly row 3 showed a significant displacement from proximo-distal polarity (Figure 5J’’’ vs K’’’ p = 0.0095). In the repolarisation in de novo condition, row 1 was modestly repolarised towards the antero-posterior axis (Figure 5L’ vs 5I’ p = 0.0018), but again row 2 showed no detectable repolarisation (Figure 5L’’ vs 5I’’ p = 0.2634) and unexpectedly row 3 also showed antero-posterior displacement (Figure 5L’’’ vs 5I’’’ p = 0.0197).
The strong repolarisation of row 0 (in cells touching Fz overexpressing cells) is expected, as Stbm is known to be strongly recruited to cell boundaries with high apposing levels of Fz (Bastock et al., 2003). However, the variable pattern of weak repolarisation between rows 1 and 3 indicates a failure of this strong repolarisation to propagate from cell to cell, and we suspect the variation may be simply due to sampling noise. In particular, these data indicate that there is no dramatic increase in repolarisation from a boundary in de novo as compared to repolarisation in the control condition, with only a modest difference in polarity in row 1 in the de novo condition.
Overall, our results show that it is hard to repolarise from a boundary of Fz overexpression in both control and de novo polarity conditions, consistent with de novo polarity being rapidly and robustly established and not easily perturbed. This provides evidence in favour of an effective cell-intrinsic polarisation mechanism.