Membrane potential regulates Hedgehog signaling and compartment boundary maintenance in the Drosophila wing disc

The Drosophila wing imaginal disc is composed of two lineage-restricted populations of cells separated by a smooth boundary. Hedgehog (Hh) from posterior cells activates a signaling pathway in anterior cells near the boundary which is necessary for boundary maintenance. Here, we show that membrane potential is patterned in the wing disc. Anterior cells near the boundary, where Hh signaling is most active, are more depolarized than posterior cells across the boundary. Elevated expression of the ENaC channel Ripped Pocket (Rpk), observed in these anterior cells, requires Hh. Antagonizing Rpk reduces depolarization and disrupts the compartment boundary. Using genetic and optogenetic manipulations, we show that membrane depolarization promotes membrane localization of Smoothened and augments Hh signaling. Thus, membrane depolarization and Hh-dependent signaling mutually reinforce each other in this region. Finally, clones of depolarized cells survive preferentially in the anterior compartment and clones of hyperpolarized cells survive preferentially in the posterior compartment.

altered, leading to decreased GFP fluorescence. Therefore, depolarized cells appear less 128 fluorescent than cells that maintain their resting potential. We expressed UAS-ArcLight in the 129 dorsal compartment of the wing disc using apterous-Gal4 (ap-Gal4). Using an anti-GFP antibody 130 in fixed tissue, we observed uniform ArcLight expression in the ap-Gal4 expression domain (Fig.  131 1H). When GFP fluorescence was visualized in live discs, a dorsoventral stripe of cells appeared 132 less fluorescent, indicating that they are more depolarized than surrounding tissue (Fig. 1I). hinge also were more fluorescent upon DiBAC staining (Fig. 1J), but for the purposes of this 147 work we focused on the wing pouch. As with ptc-Gal4 expression, the domain of relative depolarization was broader in early third-instar wing discs, becoming more and more restricted 149 to cells just anterior to the compartment boundary over the course of developmental time (Fig.  150

1K-K'). 151 152
We examined other imaginal discs using DiBAC (Figure 1, Figure Supplement 1). We did not 153 detect increased fluorescence either at the compartment boundary of the leg disc or of the 154 antennal disc. In the eye disc, we observed increased fluorescence in the region of the "ocellar 155 spot", the primordium of the ocelli, which also expresses ptc>RFP. Thus, it is possible that the 156 altered depolarization at the compartment boundary might be specific to the wing disc. 157 Alternatively, the changes in Vmem in these other discs might be more subtle and not detectable 158 using DiBAC. We therefore focused our efforts on studying the wing disc. 159 160 Altered expression of endogenous ion channels anterior to the compartment boundary 161 162 The resting potential, Vmem, results from the activity of a large number of different transporters 163 of charged molecules, as well as the permeability of the membrane to each of those molecules. 164 Thus, the relative depolarization of the region immediately anterior to the compartment 165 boundary is unlikely to result simply from a change in the activity of a single pump or channel. 166 However, by identifying transporters expressed in this region, it should be possible to 167 manipulate Vmem by altering their expression or properties. To that end, we examined a 168 published transcriptome dataset (Willsey et al., 2016), comparing the abundance of transcripts 169 in ptc-expressing cells with those of cells in the posterior compartment. In this dataset, we 170 noticed several ion channels with differential expression between the two populations of cells. 171 Among these are two members of the Degenerin Epithelial Na + Channel (DEG/ENaC) family of 172 channels, ripped pocket (rpk) (C. M. Adams et al., 1998), and pickpocket 29 (ppk 29) (Thistle et 173 al., 2012). DEG/ENaC channels are members of a diverse family of amiloride-sensitive cation 174 channels. An antibody that recognizes the Rpk protein has been characterized previously 175 (Hunter et al., 2014), which allowed us to examine its pattern of expression. In late L3 wing 176 discs, we found that Rpk was indeed expressed anterior to the compartment boundary, in a

RFP
stripe of cells that also express ptc>RFP ( Fig. 2A-A'). In addition, we observed expression of Rpk 178 in cells near the dorsoventral (D-V) boundary. Expression was most obvious in two rows of cells 179 flanking the D-V boundary in the anterior compartment, which are likely to be the two rows of 180 cells arrested in the G2 phase of the cell-cycle (Johnston & Edgar, 1998). Indeed, the pattern of 181 DiBAC uptake in this portion of the wing disc also suggests a very thin stripe of low-fluorescence 182 flanked by two regions of higher fluorescence (see inset in Fig. 1D). This is consistent with 183 previous work showing that cells in culture become more depolarized as they progress through 184 S-phase, and peaks at the onset of mitosis (Cone, 1969), reviewed in (Blackiston et al., 2009). 185 Thus, at least in principle, the increased expression of Rpk could contribute to the 186 depolarization observed in these regions of the wing disc. 187

188
Since Rpk, and possibly Ppk29, are expressed anterior to the compartment boundary, we tested 189 the effect blocking these channels by treating discs with amiloride. Amiloride is predicted to 190 reduce the permeability of DEG/ENaC-family channels (Garty, 1994). Addition of amiloride did 191 not alter the pattern of Rpk protein expression (Fig. 2C-C'). However, amiloride addition 192 abolished the stripe of increased DiBAC fluorescence anterior to the compartment boundary 193 ( Fig. 2B-B''), as well as the stripe of decreased fluorescence observed with ArcLight (Fig. 2D). 194 These findings suggest that a conductance mediated by one or more channels of the DEG/ENaC 195 family contributes to the relative depolarization of this region. Amiloride addition did not, 196 however, seem to affect the increased DiBAC fluorescence observed at the D-V boundary (Fig.  197 2B''). Since amiloride likely targets multiple ENaC channels, we also depleted Rpk using an RNAi 198 transgene expressed in the dorsal compartment of the disc using ap-Gal4. In these discs, we 199 observed reduced DiBAC fluorescence in the dorsal portion of the disc (Fig. 2E). However, due 200 to altered disc morphology, it was difficult to assess whether the stripe of In most cells, the Na + /K + ATPase is primarily responsible for setting a negative Vmem, since it uses 208 ATP hydrolysis to extrude three Na + ions and bring in two K + ions per cycle of activity (Morth et 209 al., 2007). RNA of ATP (Lebovitz et al., 1989), which encodes a subunit of the Na + /K + ATPase, 210 was also detected at higher levels in ptc-expressing cells (  In order to investigate the short-term consequences of altering V mem we used 292 channelrhodopsin ChR2, which when exposed to blue light causes membrane depolarization  ). In addition to providing a second, independent way of 294 depolarizing cells, this approach allowed us to examine short-term changes in Hh signaling that 295 occur in response to depolarization. Dissected, ChR2-expressing salivary glands were either 296 kept in darkness or exposed to activating light for variable intervals of time. Compared to 297 glands kept in the dark, light-exposed glands had higher levels of Smo at the cell membrane 298 (Fig. 5G-I)  altering Vmem would impair compartment boundary function. We set out to knock down rpk in 323 cells anterior to the compartment boundary using dpp-Gal4. Imaginal discs expressing both 324 dpp-Gal4 and UAS-rpk RNAi had more irregular compartment boundaries, indicating that 325 patterned expression of Rpk in this region is necessary for maintaining a compartment 326 boundary of normal appearance (Fig. 6A-B''). 327 328

Manipulation of Vmem has different effects on cell survival in the two compartments 329 330
The anteroposterior compartment boundary appears to represent not just the boundary 331 between two lineage-restricted domains of cells, but also the boundary between two 332 populations maintaining disparate Vmem. Although A cells just anterior to the boundary appear 333 more obviously depolarized than neighboring P cells, the pattern of DiBAC fluorescence in 334 younger discs suggests that the zone of depolarization could be broader at earlier stages of 335 development (Fig. 1K-K'). We therefore investigated the properties of clones of cells in each of 336 the two compartments that are more depolarized or hyperpolarized than their neighbors. We 337 first generated discs containing clones of cells that overexpressed the bacterial sodium channel 338 NaChBac (Ren et al., 2001;Luan et al., 2006;Nitabach et al., 2006), which would be predicted to 339 be more depolarized than their neighbors. When clones were induced late in development, and 340 given only 48 hours to grow in the tissue before dissection, they were recovered with equal 341 frequency in both compartments (Fig. 7A, B). However, when clones were induced at 342 progressively earlier times in development, we observed fewer and fewer surviving clones in 343 the P compartment (Fig. 7A'-B). 344

345
The elimination of clones of cells over time in a context-dependent manner is reminiscent of 346 cell competition (Morata & Ripoll, 1975;Johnston, 2009;Baker, 2017). Cell competition is a 347 phenomenon by which cells of lower relative fitness are eliminated from a tissue due to 348 heterotypic interactions with cells of higher fitness (Baker, 2017). We wondered whether cells 349 expressing NaChBac were eliminated by a similar mechanism in the posterior compartment. We 350 drove expression of NaChBac in the entire posterior compartment, forming a homotypic  The amount of NaChBacexpressing tissue in a compartment was quantified by measuring the ratio of the total area of GFP+ tissue to the area of the compartment, to control for the difference in size between the anterior and posterior compartments. A striking preference for clone recovery in the anterior compartment was observed in clones induced 96 hours prior to dissection, while clones induced later in development survived with relatively equal frequency between compartments, n=12 discs for each time point. Data were compared using an upaired t test (**p<0.01). (C-E) Elimination of clones from the posterior compartment is dependent on heterotypic interactions with wildtype cells. No difference in apoptosis was observed when NaChBac was expressed in the entire posterior compartment (E), n=6 discs. Data were compared using an unpaired t test. (F-G) Clones of cells expressing RNAi against the ENaC rpk are recovered preferentially in the posterior compartment, using the quantification scheme described above, n=15 discs, data were compared using an unpaired t test (**p<0.01). Clone area ratios are lower in this experiment because the heat shock exposure was decreased in order to mitigate lethality associated with widespread knockdown of rpk. Scale bars are 100µM in all panels.

environment in which cells should survive if NaChBac expression is not cell-autonomously 352
lethal. In these discs, we observed no increase in apoptosis in the NaChBac-expressing posterior 353 compartments as compared to control discs (Fig. 7C-E), suggesting that elimination of NaChBac-354 expressing clones in the posterior compartment is dependent on heterotypic interactions with 355 wild-type cells. 356

357
In order to generate clones of cells that are predicted to be more hyperpolarized than their 358 neighbors, we generated clones expressing UAS-rpk RNAi . Such clones were recovered 359 preferentially in the P compartment, suggesting that these have lower relative fitness in the A 360 compartment (Fig. 7 F-G). Taken together, these data indicate that more depolarized cells 361 survive better in the A compartment and that more hyperpolarized cells survive better in the P 362 anterior compartment ( Fig. 2A''), indicating that increased expression of Rpk could contribute 399 to the depolarization observed in those cells. We note, however, that the increased DiBAC 400 fluorescence in these cells was not entirely eliminated by exposing discs to amiloride, indicating 401 that other factors are also likely to contribute. 402 403 While we observed a clear increase in DiBAC fluorescence anterior to the compartment 404 boundary in the wing disc, we did not observe this in the leg disc (Figure 1 Supplement 1B-B'). 405 This could either be because the change in Vmem at the compartment boundary is below that 406 threshold of detection with DiBAC, or that the phenomenon is specific to the wing disc. In the 407 eye-antennal disc, the region immediately anterior to the region of Hh expression is the 408 morphogenetic furrow. Because of the change in disc architecture in this region, it is difficult to evaluate alterations in DiBAC fluorescence in unfixed preparations. We note, however, that a 410 portion of the eye disc which represents the primordium of the ocelli has increased DiBAC 411 fluorescence and increased Ptc expression (Figure 1 Supplement 1A-A'). As more sensitive 412 reagents for detecting changes in Vmem become available, some of these issues may be clarified 413 with greater confidence.  Interestingly, it has recently been proposed that Ptc might function in its inhibitory capacity by 436 a chemiosmotic mechanism where it functions as a Na + channel (Myers et al., 2017). An early 437 outcome of Smo activation is its localization to the membrane where its C-terminal tail   The compartment-specific mechanism of cell elimination that we have uncovered has some 487 similarities to the group of phenomena referred to as cell competition that was first described 488 for clones of cells that were heterozygous for a Minute mutation (Morata & Ripoll, 1975). A 489 gene that has been implicated in cell competition, flower (fwe) encodes a Ca 2+ channel (Yao et   Larvae were washed with 70% EtOH and PBS prior to dissection. Live tissue was dissected in 531 Schneider's media (#21720001, Gibco), and care was taken to not damage or stretch tissue. For 532 DiBAC staining, imaginal discs were incubated in 1.9µM DiBAC4(3) (bis-(1,3-dibutylbarbituric 533 acid) trimethine oxonol; DiBAC4(3); Molecular Probes)) in Schneider's media for 10 minutes 534 with gentle rotation. A small amount media was used to mount the discs, such that addition of 535 a coverslip did not destroy the tissue, and discs were imaged right away. 100µM amiloride 536 (#A7419, Sigma-Aldrich) and 100µM ouabain (#O3125, Sigma-Aldrich) were added to DiBAC 537 solution for pharmacology experiments. Discs imaged in FM4-64 dye (#T13320, ThermoFisher) 538 were incubated in 9µM FM4-64, and imaged without washing, to preserve staining of the cell 539 membrane. Discs expressing ArcLight were dissected, mounted, and imaged in Schneider's. 540

541
For optogenetics experiments, carcasses were dissected and cleaned (fat body removed) in 542 Schneider's medium. Carcasses were loaded onto a glass cover slip in a large drop of 543 Schneider's, and either kept in the dark (control condition) or exposed to activating light 544 (480nm for ChR2 and ChloC experiments, 647nm for ReaChR experiments). After exposure, 545 carcasses were immediately fixed and prepared for immunohistochemistry. 546 547 Immunohistochemistry 548 Imaginal discs were dissected in phosphate buffered saline, fixed for 20min in 4% PFA at room 549 temperature, permeabilized in phosphate buffered saline with 0.1% Triton X-100, and blocked Experiments using heat shock-controlled FLPase and act<<Gal4 were maintained at 25C. 569 To generate clones overexpressing NaChBac, ywhs-FLP;;Act<<Gal4,UAS-RFP/S-T virgin females 570 were crossed to yw;;UAS-NaChBac (BL:9466) males. Crosses were set up in cages, and females 571 were allowed to lay eggs on grape plates with yeast paste for 8 hour increments. L1 larvae were 572 picked and seeded into vials at a density of 50 larvae/vial. Vials were subjected to a 10 minute 573 heat shock in a 37ºC water bath 96, 72, and 48 hours prior to dissection. To generate clones 574 expressing RNAi against Rpk, hs-FLP;;act<STOP<UAS-RFP/S-T virgin females were crossed to 575 yw;;UAS-Rpk-RNAi (BL:39053, BL:25847) males. Larvae were collected as described above, but 576 vials were subjected to five minute heat shocks in order to mitigate lethality associated with 577 broad expression of Rpk-RNAi. To generate clones expressing Ci3m, hs-FLP;;act<STOP<UAS-578 RFP/S-T virgin females were crossed to UAS-Ci3m males. Larvae were collected as described 579 above, and vials were subjected to a 10 minute heat shock in a 37ºC water bath 48 hours 580 before dissection and live imaging in DiBAC. 581