Hh-producing cells in the P compartment of the wing imaginal disc extend cytonemes that transport Hh to the A compartment cells and that are essential for the restricted distribution of Hh during Drosophila epithelial development (Callejo et al., 2011; Bilioni et al., 2013; Bischoff et al., 2013). In addition, the Hh-receiving cells of the anterior compartment also extend cytonemes towards the Hh-secreting cells of the P compartment. Here we have characterized the cytonemes from the signal-receiving cells and investigated their role in Hh morphogen reception.

In previous studies on Hh signalling filopodia in the abdominal histoblasts we showed that the P compartment generated highly dynamic protrusions that reached anteriorly the Hh-receiving cells (Bischoff et al., 2013). The Hh-receiving cells also produce highly dynamic protrusions oriented towards the Hh-producing cells, easily visualized when expressing the actin-binding domain of moesin (GMA) fused to GFP (Figure 1A, Video 1A). These GMA-labelled filopodia are less dynamic when they co-express Ihog (Figure 1B, Video 1B), as was previously described for the Hh-producing histoblasts (Bischoff et al., 2013). Here we show that both Hh-presenting and Hh-receiving histoblast cells emit protrusions with similar dynamics (Video 1 and Video 2). In a more detailed analysis of filopodia dynamics, we have been able to distinguish two different dynamic behaviours: one of filopodia that elongate and immediately retract, which we have classified as ‘triangle dynamics’ and another one with a ‘stationary’ interphase between the elongation and retraction phases, which we have classified as ‘trapezoid dynamics’ (Figure 1—figure supplement 1; see Materials and methods). Both Hh-producing and Hh-receiving cell filopodia have similar values of average maximum extent, lifetime, elongation (Ve) and retraction (Vr) velocities (Figure 1C–E, Figure 1—figure supplement 2).

Figure 1 with 3 supplements see all

Download asset Open asset

Video 1

Download asset

Video 2

Download asset

The maximum extent of filopodia is similar with or without overexpression of Ihog (Figure 1C, Figure 1—figure supplement 2). This overexpression, however, changes cytoneme dynamics. Filopodia lifetime values are higher for filopodia with high levels of Ihog than those with low levels (Figure 1D, Figure 1—figure supplement 2). Hh-producing and Hh-receiving cell filopodia elongate and retract at similar rates (Figure 1E, Figure 1—figure supplement 2A). In summary, Ihog overexpression leads to more stable cytonemes also in the A compartment cells.

To further analyse the role of A compartment cytonemes, we manipulated actin cytoskeletal components implicated in filopodia formation within the wing imaginal disc, monitoring the expression of Ptc and Cubitus interruptus (Ci) as a read-out of Hh signalling. We knocked down SCAR, Diaphanous (Dia), Filamin (Cheerio, Cher) or actin Capping protein alpha (Cpa) function in the A compartment using specific RNAi lines. In addition, we expressed these RNAis together with ihog-RFP to visualize cytonemes. The aberrant extension of cytonemes observed by knocking down these proteins in the Hh-receiving cells resulted in reduced Ptc and Ci expression domains (Figure 1F), revealing a role of the A compartment cytonemes in regulating the Hh gradient. Thus, cytonemes control both Hh export from P compartment and Hh distribution and reception in the A compartment.

Cytoneme-mediated long distant communication between Hh-producing and Hh-responding cells is visualized when Ihog-RFP marked cytonemes from the A and P compartments meet at the A/P boundary in the basal part of the epithelium (Figure 1G, arrow; Video 3). Anterior compartment cells overexpressing Ihog consistently extend basal cytonemes, which invade the P compartment and heavily accumulate Hh (Figure 1H, arrow; Video 4), leadingto an extended Hh gradient (Figure 1—figure supplement 3). Taken together, these results indicate that Hh-receiving cells generate dynamic cytonemes directed towards Hh-producing cells, and that these A cytonemes can influence the graded distribution of Hh.

Video 3

Download asset

Video 4

Download asset

To further study the involvement of cytonemes in Hh reception we blocked endocytosis specifically in Hh-receiving cells by expressing shi^K44A, which encodes a dominant negative mutant of Dynamin (aka Shibire, Shi). Expression of this mutant protein blocks the rapid internalization of the Hh-Ptc complex in signal-receiving cells. This approach enables the visualization of ligand and receptor at the site of signal reception (Callejo et al., 2011; Torroja et al., 2004). In this mutant background, Hh co-localizes with Ptc at the basal membrane of Hh-receiving cells, revealing the presence of filopodia-like structures oriented in the A-P axis (Figure 2A, arrowheads). Extracellular staining of Ptc and Hh confirms this basal localization of Hh-Ptc complex for signal reception (Figure 2A’, arrow). Furthermore, cytonemes from Hh-producing cells (labelled with Ihog-RFP) invade the A compartment to contact the basal membrane of the Hh reception region (labelled with Ptc) in A cells (Figure 2B arrowheads and 2B’ arrow). We can also detect the receptor Ptc localized in A cytonemes when its internalization is impeded in the Hh-receiving cells (Figure 2C and C’, arrow) or co-localizing with P compartment Ihog-RFP labelled cytonemes (Figure 2—figure supplement 1). These results indicate that A compartment cells receive the Hh signal through their cytonemes, which also contain Ptc.

Figure 2 with 1 supplement see all

Download asset Open asset

We then expressed Hh-CD2 protein, which cannot be released from the plasma membrane and is therefore unable to form a concentration gradient (Strigini and Cohen, 1997). As expected Hh-CD2 localizes to basal cytonemes (Figure 3—figure supplement 1A), since this form of Hh is permanently anchored to the plasma membranes. Therefore, the Hh-CD2 protein expression allowed us to visualize interactions between cytonemes emanating from the A compartment (expressing Ihog-RFP) and cytonemes from the P compartment (expressing Hh-CD2) along their lengths (Figure 3A, arrowheads; Figure 3—figure supplement 1B). Hh-receiving cytonemes accumulate the co-receptor Ihog at contact sites with Hh-CD2-presenting cytonemes (Figure 3—figure supplement 1C and C’, compare C with D). Noticeably, the Hh receptor Ptc is also localized in these contact sites between cytonemes from A and P compartment cells (Figure 3B, arrows; Figure 3—figure supplement 1C’ and E). These results suggest that the Hh receptor complex binds Hh-CD2 protein at the basal membrane of Hh signal-receiving cytonemes, while blockage in Hh release and internalization results in Ptc also being retained in the cytoneme membrane (Figure 3—figure supplement 1E, compare E with F). Taken together, these data indicate that Hh reception is mediated by contacts between Hh-sending and Hh-receiving cytonemes.

Figure 3 with 1 supplement see all

Download asset Open asset

To further explore the mechanism of contact between A and P cytonemes we adapted the GRASP technique (GFP Reconstitution Across Synaptic Partner), which was originally developed to image membrane contacts at neuronal synapses (Feinberg et al., 2008; Gordon and Scott, 2009). We used UAS.CD4-GFP^1-10 with the complementary CD4-GFP¹¹ fragment regulated by LexAop (LexAop.CD4-GFP¹¹). Membrane contacts between Hh-sending and Hh-receiving cells were imaged by expressing the complementary GFP fragments separately in each compartment using the required LexA and Gal4 drivers. As expected, GFP fluorescence was localized to the A/P border when complementary fragments were expressed at the same time in each compartment. The pattern of GFP fluorescence varies along the apico-basal axis, as both A and P cells extend cytonemes along the basal surface of the epithelium (Figure 4A,B,B’; Video 5). Cytoneme membrane contacts occur on both sides of the A/P compartment border, although predominantly on the A side (Figure 4C). Furthermore, periodic annular structures are present along the length of overlapping cytoneme (arrows in Figure 4D and E), while the GRASP signal also indicates that cytonemes overexpressing Ihog-RFP are indeed in contact with neighbouring cell membranes, presumably receiving cytonemes (Figure 4E, arrows). Importantly, both Hh and Ptc co-localize in these basal membrane contacts (detected by GRASP signal) upon blockage of their uptake in a shi^K44A mutant background (Figure 4F, arrowheads). Therefore, we propose that these contacts are sites where Hh-Ptc interaction takes place all along the length of A and P cytonemes for signal reception (see diagram Figure 4G).

Figure 4

Download asset Open asset

Video 5

Download asset

It has been previously described that clones of ptc^−/− and smo^−/− located at the A/P compartment border are unable to internalize Hh, but do not block Hh signal transmission to more anterior wild type cells (Chen and Struhl, 1996). Our cytoneme transport hypothesis predicts that Hh could be transported across mutant ptc^−/− and smo^−/− clones in cytonemes that act as conveyer belts for long distance transport. Accordingly, we observed that Hh-producing cells still spread cytonemes when Hh reception was prevented in the A compartment by knocking down either Ptc or Smo, or both co-receptors Ihog and Brother of Ihog (Boi) at the same time (Figure 5—figure supplement 1). In addition, cytonemes from both compartments were still formed upon the elimination of Hh (Figure 5—figure supplement 2A–F) using the hh thermosensitive allele (hh^ts2) (Ma et al., 1993). No apparent changes in Hh-receiving cytonemes were also observed when Hh reception was prevented in ptc^−/− MARCM clones expressing Ihog-YFP (Figure 5—figure supplement 2G). Therefore, the presence of the network of cytonemes is independent of Hh signaling.

We then asked if cytonemes could facilitate the Hh transport across a ptc mutant territory. Hh signalling is fully activated in ptc^−/− mutant clones due to the lack of normal Ptc repression upon Smo (Chen and Struhl, 1996). We show here that Hh protein can be detected in the wild type cells anterior to the ptc^−/− mutant clone (Figure 5A). Accordingly, with previous experiments (Figure 5—figure supplement 1B), cytonemes emanating from Hh-producing cells orient normally towards the A compartment cells across a ptc mutant clone induced adjacent to the A/P border in the wing disc (Figure 5B). Consistently, abdominal histoblasts of the P compartment produce cytonemes that cross a ptc^−/− clone in their way towards the anterior cells (Figure 5C, Video 6). These cytonemes have similar dynamics (maximum extent, lifetime and elongation and retraction velocities) when crossing wild type or ptc^−/− mutant territories (Figure 5C and C’, Figure 5—figure supplement 3). Therefore, cytoneme dynamics is independent of the presence of the receptor Ptc.

Figure 5 with 3 supplements see all

Download asset Open asset

Video 6

Download asset

In addition, GRASP signal indicates that cytonemes from the Hh-receiving cells reach the Hh-sending cytonemes close to the A/P compartment border (Figure 5D, region 1 arrowhead). So that, GRASP signal is detected along these A cytonemes that cross the clone and interact with the P compartment cytonemes (Figure 5E, arrows). However, in subapical regions of the epithelium, where cytonemes are too short to reach the P compartment, GRASP fluorescence is not detected (Figure 5D, asterisk). It is also known that when a ptc^−/− clone is too large, wild type anterior cells are not able to respond to Hh (Chen and Struhl, 1996). Noticeably, in this situation we observe that cytonemes extending from the A cells do not reach the P cells and no GRASP fluorescence is detected (Figure 5D, region 2 arrowhead).

In smo^−/− clones originating in the A compartment, and monitored by low levels of Ci expression, the A/P compartment border is re-established in the position where Hh signalling is activated in cells anterior to the clone (Blair and Ralston, 1997; Rodriguez and Basler, 1997). As a result, A compartment smo^−/− cells integrate into the P compartment and maintain a straight boundary at the nearest anterior cells where Hh is then received. The Hh visualized within smo^−/− clones is restricted to the most basal region of epithelial cells, demonstrating again that Hh travels mostly basally (Figure 6A, arrowhead) (Bilioni et al., 2013; Bischoff et al., 2013; Callejo et al., 2011). In addition, cytonemes emanating from cells anterior to the clone are loaded with Hh (Figure 6B, arrowhead), and P cytonemes cross a smo^−/− clone reaching the A/P compartment border (Figure 6C, arrow). Accordingly, GRASP fluorescence is detected where the compartment border is re-established anterior to a smo^−/− clone (Figure 6D, arrow).

Figure 6

Download asset Open asset

All together these results indicate that Hh can be transported and received by cytonemes from P and A compartment, respectively, to cross ptc^−/− and smo^−/− mutant territories and re-establish the morphogen gradient.

Heparan sulphate proteoglycans (HSPGs) are key players in cell-matrix interaction, cell-cell signalling, and long-range Hh function (Lin, 2004; Tabata and Takei, 2004). We have previously demonstrated that P compartment cytonemes do not normally extend along mutant HSPG clones located at the A/P compartment border (Bischoff et al., 2013). Here we analyse the potential role of HSPG glypicans, in cytoneme function. The Drosophila glypicans Dally and Dlp are required for both Hh presentation and reception (Ayers et al., 2010; Bilioni et al., 2013; Han et al., 2004a; Yan et al., 2010). We observe that extension of Ihog-RFP expressing cytonemes is significantly reduced when glypicans levels are lowered by knocking down Dally or Dlp, in the opposite compartment (Figure 7A–C, arrows; Figure 7—figure supplement 1A–D, arrows). In agreement with this observation, cytonemes are rarely detected crossing through dally and dlp double mutant clones (Figure 7D, arrowhead). In addition, cytonemes emanating from the A compartment cells accumulate Dally and Dlp present in the neighbouring P compartment cells (Figure 7E, arrow; Bilioni et al., 2013). This effect is observed by expressing Ihog-RFP and eliminating Dlp in the same compartment; Dlp of one compartment accumulate in labelled cytonemes emanating from other one (Figure 7F, arrow). This dependence on the glypicans present in the neighbouring cells is also observed for the P compartment cytonemes (Figure 7—figure supplement 1E and F, arrows). In summary, all these results demonstrate the requirement of glypicans for cytoneme spreading and their specific role in the cytoneme-mediated interplay between Hh-producing and Hh-receiving cells.

Figure 7 with 1 supplement see all

Download asset Open asset

We then analysed the interaction between confronting cytonemes at the A/P compartment border when Ihog is overexpressed in one compartment and either Dally or Dlp overexpressed in the other one. Remarkably, we observe an accumulation of Ihog and glypicans along the membranes of the Hh-sending and Hh-receiving interacting cytonemes (Figure 8A–D, arrowheads). Moreover, these contact sites present well-defined annular structures containing both Ihog and the glypicans Dally and Dlp (Figure 8A’ and B’, arrows) similar to the ones visualized by GRASP (Figure 4D and E, arrows). These results show again that Hh-sending and Hh-receiving cytonemes touch-contact along their lengths and, importantly, reveal a role for an Ihog/glypican interaction in trans during this cell-cell interface.

Figure 8

Download asset Open asset

In a previous study we described that the P compartment cytonemes were critical for the Hh gradient establishment acting as conduits for Hh transport (Bischoff et al., 2013). Here we have extended this work and showed that the cytonemes emanating from signal-receiving cells in the A compartment also have a critical role in Hh gradient formation. Therefore, both A and P compartment cytonemes weigh in the formation of the Hh gradient. The two sets of cytonemes make contact through the basal side of the epithelium and show similar values of average maximum extent, lifetime, and elongation and retraction velocities. A and P cytonemes contact predominantly on the anterior side of the compartment boundary. Furthermore, receiving cytonemes include components of the Hh reception complex (Ptc, Ihog, Dlp and Dally), supporting their signalling role; consistently, we find that manipulating their actin dynamics modifies Hh graded signalling.

As implied in our earlier study, signal transmission might be dependent on cytoneme stability and the frequency with which A and P cytonemes make contact. More Hh would be delivered to A cells close to the A/P compartment border than to those further from the signal source. In agreement with this hypothesis, GRASP fluorescence analysis suggests that A cytonemes from cells close to the A/P boundary establish contact primarily with the cytonemes from adjacent P compartment cells, where Hh would also be delivered by further P cells cytonemes. The sum of these spatial interactions would then result in high levels of Hh signalling close to the A/P border. The longest cytonemes from cells further from the A/P compartment border would interconnect to deliver the reduced levels of Hh signalling in more distant A cells.

We analysed the behaviour of confronting A and P cytonemes with mutated or overexpressed Hh signalling components. Interestingly, the expression in the P compartment of membrane-tethered Hh (Hh-CD2), which results in a non graded signalling response (Strigini and Cohen, 1997), alters the receiving cytonemes cargo by hindering Ptc internalization. Despite that Hh receptor Ptc is usually rapidly internalized and not visualized at the plasma membrane, both Ptc and Ihog accumulate in A compartment cytonemes at contact sites with Hh-CD2 expressing P compartment cytonemes. Thus, this form of Hh attached to membranes plausibly affects the reception process, potentially explaining the absence of a signalling gradient and further confirming that Hh release from the presenting membranes is necessary. The mechanism by which Hh is liberated from plasma membrane anchoring lipids in Drosophila remains unknown. However, in strong support of a Hh shedding process, proteolytic removal of both lipidated peptides releases active Sonic hedgehog (Shh) from the surface of cultured cells (Ohlig et al., 2011).

On the other hand, Ihog trans interaction with glypicans in cytoneme-mediated cell-cell contacts has a critical role for cytoneme dynamics. The spreading capacity of Ihog overexpressing cytonemes is dependent on the glypicans presence in the membranes of neighbouring cells. Cytonemes protruding from one compartment aberrantly extend over the other compartment when is mutant for dally and dlp. This is in agreement with our previous results showing that the Tout-velu (Ttv) and Brother of Ttv (Btv) proteins (needed for the HSPGs synthesis) are required to stabilize Ihog cytonemes (Bischoff et al., 2013). In addition, Dally or Dlp overexpressing cytonemes from one compartment seem to be stabilized by Ihog overexpressing cytonemes from the other one, with which they associate along their length by trans interaction between Ihog and glypicans. Thus, cytonemes from one compartment can contribute to the stability of the cytonemes from the other one with which they contact.

However, to elucidate the specific mechanism of this altered cytoneme behaviour (impairment in elongation, stability or retraction processes) further in vivo analysis is required.

Taken together, these results establish that dynamic interactions between A and P compartment cytonemes play an important role in the formation of the Hh gradient. In conjunction components for both reception and presentation of Hh critically influence the performance of this cytoneme interaction.

The membrane-bound GRASP fluorescence shows that A and P compartment cytonemes overlap in a region spanning the A/P compartment boundary. Within this region, direct ‘kissing’ contacts appear along the membranes of interacting A and P cytonemes to connect the two sets of cytonemes. These contact points are associated with annular structures, which may represent more stable links and are reminiscent of synaptic buttons. Similar ‘synaptic contacts’ have been previously described between air sac primordium (ASP) and wing disc cells for Dpp morphogen transference (Roy et al., 2014). ASP cells receive Dpp by emitting long cytonemes that contact with the target cell of the wing disc by their tips. This process was visualized by the same membrane-bound GRASP. Although Dpp uses a cytoneme-mediated contact-dependent mechanism for signalling at distance, the reception takes place where Dpp-receiving cytoneme tip contacts with the Dpp source cell body, in contrast with the cytoneme-cytoneme contacts here described for Hh. Nevertheless, the ‘synaptic contacts’ described for Dpp reception in ASP might be conceptually comparable with the ‘kissing rings’ distributed along the length of overlapping Hh-sending and Hh-receiving cytonemes.

We have previously demonstrated that Hh moves along cytonemes in multivesicular bodies (MVB) to be secreted as extracellular vesicles known as exosomes (Gradilla et al., 2014), which raises intriguing questions about the mechanism of interaction between exosomal Hh and its receptor Ptc in the A cytonemes. Interestingly, Ptc and Hh can be detected co-localizing with cytoneme contact sites, suggesting that they could represent specific sites for Hh reception and internalization. These data strengthen our proposal of cytoneme-mediated Hh morphogen distribution and suggest that Hh containing exosomes could be received at intimate contacts between P and A cytonemes.

Finally, we have shown that cytonemes are able to cross ptc^−/− and smo^−/− clones, enabling the Hh transport from P to A cells for Hh signalling reestablishment in wild type cells located anterior to the clones. In agreement, it is known that the Hh signal can cross small mutant clones of ptc and smo to reach the adjacent wild type territory. Similar behaviour, has been described for ptc^−/− or smo^−/− clones induced in the A abdominal histoblast nests and attributed to different cell polarity and cell affinity (Struhl et al., 1997). Therefore, Hh transport and reception by cytoneme-mediated cell-cell contacts can explain the Hh signalling in wild type cells located anterior to ptc^−/− and smo^−/− clones.

In summary, the data strengthen our proposal of cytoneme-mediated Hh morphogen distribution and suggest that Hh containing exosomes could be received at intimate contacts between P and A cytonemes. The described intimate interaction between these two sets of cytonemes at membrane contact sites, where ligand and receptor complex colocalize, may facilitate morphogen reception.

A description of mutations, insertions and transgenes is available at Fly Base (http://flybase.org). The following mutants and transgenic strains were used: tub.Gal80^ts, hs-Flp122 (Bloomington Drosophila Stock Centre (BDSC), Indiana, USA; http://flystocks.bio.indiana.edu), hh^ts2 (Ma et al., 1993), shi^ts1 (Grigliatti et al., 1973), dally³² (Franch-Marro et al., 2005), dlp²⁰ (Franch-Marro et al., 2005), smo³ (Nüsslein-Volhard and Wieschaus, 1980) and ptc¹⁶ (Nakano et al., 1989).

The following Gal4 and LexA drivers were used for ectopic expression experiments using the Gal4/UAS (Brand and Perrimon, 1993) and LexA/LexAop (Yagi et al., 2010) systems: hh.Gal4 (Tanimoto et al., 2000), ptc.Gal4 (Hinz et al., 1994), dpp.LHG (Yagi et al., 2010) (referred as dpp.LexA), hh.LexA (generated at CBMSO department of development and differentiation), en.Gal4 (expressed exclusively in the P compartment was a gift from Christian Dahman).

The pUAS-transgene and LexAop-transgene strains were: UAS.hh-GFP (Torroja et al., 2004), UAS.dlp-GFP (Han et al., 2004b), UAS.dally-GFP (Eugster et al., 2007), UAS.shi^K44A (Moline et al., 1999), UAS.ihog-YFP and UAS.ihog-RFP (Callejo et al., 2011), UAS.GMA-GFP (Bloor and Kiehart, 2001), UAS.CD4-tdTomato (Han et al., 2011), UAS.CD4-GFP^1-10 and LexAop-CD4¹¹ (a gift from Kristin Scott), UAS.hh-CD2 (Strigini and Cohen, 1997), UAS.lifeactin-GFP (BDSC 35544), UAS.scar-RNAi (BDSC 36121), UAS.dia-RNAi (BDSC 33424), UAS.Cpa-RNAi (Vienna Drosophila RNAi Center (VDRC), Vienna, Austria, v16731), UAS.cher-RNAi (BDSC 35755), UAS.ptc-RNAi (BDSC 28795), UAS.smo-RNAi (BDSC 27037), UAS.ihog-RNAi (VDRC 102602), UAS.boi-RNAi (VDRC 108265), UAS.dally-RNAi (VDRC 14136), UAS.dlp-RNAi (VDRC 106578).

The LexAop.ihog-RFP construct used ihog-RFP from the pTWR vector (Callejo et al., 2011), which was introduced into the pLOTattB plasmid (Lai and Lee, 2006) carrying the lexA operator (LexAop). Transgenic strains were recovered using standard protocols.

Transient expression of transgenic constructs used the tub-Gal80^ts; Gal4 and LexA systems with fly crosses maintained at 18°C, with inactivation of the Gal80^ts repressor for 16–40 hr at restrictive temperature (29°C) before dissection. The transgene actin<CD2<Gal4 (Pignoni and Zipursky, 1997) was used to generate random ectopic clones of the UAS lines. Larvae of the corresponding genotypes were incubated at 37°C for 10 min to induce hs-Flp-mediated recombinant clones.

Mutant clones were generated by hs-Flp-mediated mitotic recombination. Larvae were incubated at 37°C for 45 min at 48–72 hr after egg laying (AEL). The genotypes were:

Figure 5A: y w hs-Flp122; FRT42D ptc¹⁶ / FRT42D arm.lacZ.

Figure 5B: y w hs-Flp122; FRT42D ptc¹⁶ / FRT42D arm.lacZ; UAS.ihog-RFP/ tub.Gal80^ts, hh.Gal4.

Figure 5C: y w hs-Flp122; FRT42D ptc¹⁶ / FRT42D ubi.RFPnls; UAS.GMA-GFP / hh.Gal4.

Figure 5D,E: y w hs-Flp122; FRT42D ptc¹⁶, UAS.CD4-GFP^1-10, LexAop.CD4-GFP¹¹/ arm.lacZ FRT42D; dpp.LexA, tub.Gal80^ts / LexAop.ihog-RFP, hh.Gal4.

Figure 6A: y w hs-Flp122; smo³ FRT40A / arm.lacZ FRT40A.

Figure 6B: y w hs-Flp122; smo³ FRT40A, ptc.Gal4 / arm.lacZ FRT40A; UAS.ihog-RFP / tub.Gal80^ts.

Figure 6C: y w hs-Flp122; smo³ FRT40A / arm.lacZ FRT40A; UAS.ihog-RFP / tub.Gal80^ts, hh.Gal4.

Figure 6D: y w hs-Flp122; smo³ FRT40A / arm.lacZ FRT40A, UAS.CD4-GFP^1-10, LexAop.CD4-GFP¹¹; dpp.LexA, tub.Gal80^ts / hh.Gal4.

Figure 7D: y w hs-Flp122; ptc.Gal4, tub.Gal80^ts / UAS.ihogRFP; dally³² dlp²⁰ FRT2A / ubi GFP FRT2A.

Video 6: y w hs-Flp122; FRT42D ptc¹⁶ / FRT42D ubi.RFPnls; UAS.GMA-GFP / hh.Gal4.

Immunostaining was performed according to standard protocols (Capdevila and Guerrero, 1994). Imaginal discs from third instar larvae were fixed in 4% paraformaldehyde (PF) in PBS for 20 min at room temperature (RT) and permeabilized in PBS 0,1% Triton (PBT) before incubating with PBT 1% BSA for blocking (1 hr at RT) and primary antibody incubations (overnight at 4°C). Incubation with fluorescent secondary antibodies (1/400 ThermoFischer) was performed for 1 hr at RT and then washing and mounting in mounting media (Vectashield). Primary antibodies were used at the following dilutions: rabbit polyclonal anti-Hh (α-Hh) (Bilioni et al., 2013), 1:500; mouse monoclonal α-Ptc (Capdevila and Guerrero, 1994), 1:150; rat monoclonal α-Ci (a gift from B. Holmgren), 1:20, mouse monoclonal α-Dlp (Lum et al., 2003), 1:30; mouse monoclonal α-Smo (Hybridome bank), 1:30; rabbit polyclonal α-βGal (from Jackson laboratories), 1/1000.

The protocol for the extracellular labelling using α-Hh and α-Ptc antibodies is described in (Torroja et al., 2004). Imaginal discs from third instar larvae were dissected on ice, transferred immediately to ice-cold M3 medium containing α-Hh (1:30 dilution) and α-Ptc (1:10 dilution) antibodies and incubated at 4°C for 1 hr. The incubation with the primary antibody under these 'in vivo' conditions, without detergents prior to fixation, prevented antibody penetration of cells. Imaginal discs were then washed in ice-cold PBS, fixed in PBS 4% PF at 4°C, washed in PBT and incubated with secondary fluorescent antibody as above.

Laser scanning confocal microscope (LSM710 Zeiss) was used for confocal fluorescence imaging of imaginal discs. ImageJ software (National Institutes of Health) was used for image processing and for image analysis.

Gradient lengths were determined as described in Bischoff et al. (2013). To quantify the maximum extent of cytonemes, we measured the ten longest protrusions in the wing pouch, from the A/P border to the tip, using the line tool in ImageJ. Then, the average length of the ten longest cytonemes was used for the quantitative analysis.

Imaging of pupal abdominal histoblasts was done using a chamber as described in Seijo-Barandiarán et al. (2015). Hh signalling filopodia from histoblasts of dorsal abdominal segment A2 were filmed using 40x magnification taking Z-stacks of around 30 μm of thickness with a step size of 1 μm every 2 min (Video 1 and Video 2) or 30 s (Video 6 and Video 7) using a LSM710 confocal microscope. All movies were analysed with Fiji and displayed at a rate of 7 frames per second. All imaged pupae developed into pharate adults and hatched normally.

Video 7

Download asset

Signalling filopodia were tracked using MTrackJ plugin of ImageJ (https://imagescience.org/meijering/software/mtrackj/). We took base (track#1) and tip (track#2) point (x, y) coordinates in each frame for each signalling filopodium (cluster) and coloured them by cluster, showing filopodia in different colours. Signalling filopodia were analysed in a 30 min time-window in all movies. We discarded filopodia not clearly visible within the 30 min movies due to tissue movement or surrounding dying larval epithelial cells (as shown in Video 6). The tracking was done in a region of 49 μm x 76 μm for experiments expressing GMA in Hh-producing (N = 4 pupae) or Hh-receiving (N = 4 pupae) cells, and for experiments co-expressing GMA and Ihog in Hh-producing (N = 4 pupae) or Hh-receiving (N = 4 pupae) cells. Tracking was restricted to a region of 79 μm x 116 μm for the ptc^−/− mutant clone experiment (N = 1 pupa), where filopodia from the ptc^−/− mutant territory and the wild-type territory was defined by the area references shown in Video 7, and used for statistical analysis.

Using the (x, y) coordinates of base and tip points of each filopodium given by the MTrackJ we calculated the elongation velocity, retraction velocity and lifetime as we describe below in ‘Filopodia dynamics models description’. We also calculated filopodium extent through this formula:

Filopodium extent (E): $E=\sqrt{{\left(x_{tip}-x_{base}\right)}^2+{\left(y_{tip}-y_{base}\right)}^2}$

We have distinguished two populations of filopodial dynamic behaviours (Figure 1—figure supplement 1):

We used R (http://www.R-project.org) to perform the Shapiro-Wilk normality test and pooled data of the studied variables was compared between different genotypes using the Mann-Whitney-Wilcoxon test of homogeneity of variances.

For statistical analysis of Emax comparing Hh producing and receiving filopodia dynamics we used pooled data comparing 104 filopodia of 4 ptc.Gal4>UAS.GMA-GFP pupae, 100 filopodia of 4 hh.Gal4>UAS.GMA-GFP pupae, 47 filopodia of 4 ptc.Gal4>UAS.GMA-GFP, UAS.ihog-RFP pupae and 75 filopodia of 4 hh.Gal4>UAS.GMA-GFP, UAS.ihog-RFP pupae. Statistical analysis of 72, 79, 28 and 30 filopodia for Ve and 94, 82, 36 and 31 for Vr was done from ptc.Gal4>UAS.GMA-GFP; hh.Gal4>UAS.GMA-GFP; ptc.Gal4>UAS.GMA-GFP, UAS.ihog-RFP; and hh.Gal4>UAS.GMA-GFP, UAS.ihog-RFP pupae, respectively; because not all filopodia had an elongation or retraction phase during the recording period.

For statistical analysis of Emax of Hh producing filopodia dynamics either crossing ptc^−/− mutant or wild-type territories we compared 23 filopodia crossing ptc^−/− mutant clone territory with 29 filopodia crossing wild-type territory of 1 pupa. For statistical analysis of Ve and Vr we used that data of 17 filopodia crossing wild-type territory and 14 filopodia crossing ptc^−/− mutant territory in an elongation and a retraction phase.

The funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication.

We are grateful to Mar Casado in our department for generating the LexA Hh insertion, to Eva Caminero for injecting the embryos to produce the transgenic fly lines and to the confocal microscopy facilities of the Centro de Biología Molecular ‘Severo Ochoa’ for skillful technical assistance. We are also grateful to the members of IG lab for discussions during the development of this work, to Marcus Bischoff for help with live imaging and to David Gubb, Pedro Ripoll, Ana-Citlali Gradilla and Adrián Aguirre-Tamaral for comments on the manuscript. We thank to Kristin Scott for the fly lines to do the GRASP experiments, to Xinhua Lin for the dlp-GFP and dally-GFP fly strains, to Christian Dahman for the en.Gal4 expressed exclusively in the P compartment, and the BDSC and VDRC for stocks. Work was supported by grants BFU2014-59438-P from the Spanish Ministry of Economy and Competitiveness (MINECO), and by a personal and by institutional grants from the Fundación Areces. ISB were supported by FPI fellowship from the Spanish Ministry of Economy and Competitiveness (MINECO). L. González-Méndez was supported by a contract of BFU2014-59438-P grant from the Spanish MINECO).

1. Received: December 7, 2016
2. Accepted: July 17, 2017
3. Version of Record published: August 21, 2017 (version 1)

© 2017, González-Méndez et al.

This article is distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use and redistribution provided that the original author and source are credited.