Filopodia are essential for steroid release

Abstract

Steroid hormones, crucial for development and physiology, were traditionally believed to diffuse passively through membranes. However, recent evidence shows insect steroid ecdysone being secreted via regulated exocytosis, but the mechanisms ensuring successful hormone release into circulation remain unclear. Our study identifies specialized membrane protrusions, signaling filopodia, in the Drosophila prothoracic gland as essential for vesicle-mediated steroid release. Confocal imaging reveals that these actin- and tubulin-rich structures form a membrane-intertwined basal domain critical for secretion. Disrupting filopodia by interfering with basement membrane interactions-Perlecan or β-integrin-or filopodia-specific protein expression-α-actinin-significantly reduces ecdysone signaling by impairing its release, despite proper production in the gland. Additionally, filopodia dynamics, such as length and density, align with secretion timing and hormone circulating levels, suggesting their role in synchronizing release with physiological needs. The systematic presence of membrane protrusions in steroid-secreting glands across species prompts a comprehensive re-evaluation of steroid release mechanisms.

Fig. 1. The PG is a two-cell layer organ surrounded by a BM.

a, a’ The PG is organized into a dorsal and ventral layer. Dotted lines in the ventral layer image outline the trachea (a). 3D visualization of the ring gland (a’) composed of the PG (prothoracic gland), the corpora allata (CA), and the corpora cardiaca (CC). The aorta and trachea are indicated. b-b” Maximal projection of prothoracicotropic hormone (PTTH) staining (b) and two Z-sections revealing that PTTH neurons project their axons at the midline of the PG (arrowheads in b’ and b”). n ≥ 10 independent experiments (c) Synaptic PTTH boutons reveal PTTH in the presynapse and Torso in the postsynapse. n^experiment ≥ 3. d, d’ Expression of trol-GFP in the BM plane and in a lateral plane (d). The XZ section (d’) facilitates the observation of the trol-GFP dots. These dots (indicated by arrowheads) are CIVICs. n^experiment ≥ 10. e Z-sections reveal that the adherens junction proteins Arm and ECad, the septate junction markers Cora, Dlg1, as well as βPS, are uniformly distributed along the PG cell membranes. n^experiment ≥ 10. f Heatmap of Syt1-GFP in an XZ section of a PG cell. Flip-out Syt1 clones were generated by crossing the heat shock Flipase (hsFLP); Act > CD2>Gal4 line with UAS-Syt1-GFP. Following heat shock, CD2 is excised in random cells, enabling Syt1-GFP expression under Act-Gal4 control, here abbreviated as hsFLP; Act»Syt1-GFP. The quantification, shown below, is made across ROI boxes appearing in dotted lines, n^clones = 7 (see Methods). Data are presented as mean ± 95% CI and were subjected to the Friedman test followed by Dunn’s multiple comparison tests (ns not significant, *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001). Source data are provided as a Source Data file. White scale bars: 20 μm; yellow scale bar: 5 μm. g Schematic representation of a PG dorsal layer and XZ cut.

Fig. 2. PG cells extend actin- and tubulin-based filopodia.

a Flip-out clones randomly expressing the fluorescent marker RFP in PG cells extend membrane protrusions (see arrowheads in the inset). The image corresponds to a maximum projection of a Z-stack. b Flip-out clone expressing the membrane marker myr-GFP. Long filopodia are observed on the sub-BM side (arrowheads), while fewer and shorter filopodia are observed in a lateral plane. The YZ section reveals a long filopodium present in the sub-BM region (arrowhead). c Flip-out clone expressing the membrane marker PLC-GFP shows filopodia in the sub-BM region (arrowheads) but not in the lateral plane. d myr-GFP (left) or PLC-GFP (right) flip-out clones induced in the fat body. No filopodia are observed. e Two neighboring cells (RFP positive due to CoinFLP) express complementary GFP fragments (spGFP^1–10 or spGFP¹¹) using the GRASP technique. Reconstructed GFP signal is detected on the sub-BM side, revealing filopodia contact between the two neighboring cells (arrowheads). A GFP signal is also observed in the lateral plane, appearing as a thin stripe. f Flip-out clones expressing an RNAi against mys (RFP positive) in a PG stained for βPS (green). In the sub-BM plane, βPS-decorated endogenous filopodia appear as punctate signals resembling projections in two WT cells (RFP negative, arrowheads in the zoomed image). No filopodia are detected in the lateral plane. g Flip-out PLC-GFP clone expressing cofilin RNAi. In the sub-BM plane, long filopodia containing F-actin are observed (arrowheads). In contrast, no actin-based filopodia are detected in the lateral plane despite F-actin stabilization. h Flip-out clone marked by PLC-GFP and the actin reporter Lifeact-Ruby, which label filopodia (arrowheads). i Flip-out clone expressing RFP in a PG stained against αTub85E. Microtubules are present in the filopodia (arrowheads in the zoomed images). j ex vivo experiment showing filopodia tracking through myr-GFP expression. Blue, white, and red arrowheads indicate different filopodia over time. Pictures are representative (n = 276 filopodia across 15 PG). White scale bars: 20 μm; yellow scale bars: 5 μm. k Schematic representation of filopodia distribution in PG cells, showing the presence of F-actin, tubulin, and βPS.

Fig. 3. The role of the MVB pathway, integrins, and BM components in establishing asymmetric membrane identity and filopodia organization in PG cells.

a Silencing Rab11 results in a 190-h delay in metamorphosis onset compared to the control. Number of sample tubes: n^phm>xw1118 = 8, n^{RNAi-Rab11xw1118} = 18, n^{phm>RNAi-Rab11} = 10. b Silencing Vps25, Vps36, or shrb in PG cells resulted in a several-hour delay compared to the control. Number of sample tubes: n^{phm> xw1118} = 8, n^{RNAi-Vps25xw1118} = 10, n^{RNAi-Vps36xw1118} = 6, n^{RNAi-shrbxw1118} = 4, n^{phm>RNAi-Vps25} = 8, n^{phm>RNAi-Vps36} = 2, n^{phm>RNAi-shrb} = 2. phm > RNAi-Rab11 PG lacks microtubule-based (c) and βPS-based (d) filopodia in the sub-BM region. Filopodia are visible with the PLC-GFP membrane marker in ex vivo control PG cells (e, arrowheads) but disappear upon Rab11 downregulation (f). Comparison of control w¹¹¹⁸ PG (g) and phm > RNAi-Rab11 PG (h) showing F-actin (single section) and PTTH staining (maximum projection of a Z-stack) processed with a depth color code from a Fiji plugin. XZ cuts reveal PTTH axons positioned between the PG bilayer in the control (g, arrowheads), whereas upon recycling pathway disruption, PTTH axons are also mislocalized just below the BM (h, arrowheads). i Axon from a phm > RNAi-Rab11 PG at the sub-BM region. The PTTH staining overlaps with its receptor Torso, suggesting that the mislocalized synaptic buttons are functional. j In phm > RNAi-Rab11 PG at the sub-BM region, mislocalized PTTH-containing synaptic buttons exclude βPS expression. Insets 1 and 2 highlight the signal exclusion between PTTH and βPS expression. k Z sections of a trol^JO271#49 mutant PG reveal the loss of a bilayer organization. l αTub85E staining in trol^JO271#49 mutant PG shows short microtubules in the sub-BM region, while in a lateral plane, αTub85E staining appears normal. m Down-regulation of mys impedes microtubule extension. n Developmental timing of trol^JO271#49 mutant vs w¹¹¹⁸ individuals. The trol^JO271#49 mutant enters pupariation 24 h after the control line. n^w1118 = 10, n^trolJO271#49 = 6. White scale bars: 20 μm; yellow scale bars: 5 μm. Data are presented as means ± SEM in (a, b, n). Source data are provided as a Source Data file.

Fig. 4. PG filopodia contain the machinery involved in ecdysone secretion.

Zoomed-in images of a PLC-GFP-expressing clone showing that CSP (a) and Syx1A (b) localize in filopodia in the sub-BM area. Arrowheads indicate CSP and Syx1A puncta within the filopodia. c A Ypet-Atet-expressing clone showing Ypet-Atet at the membrane and in filopodia (arrowheads) at the sub-BM region. Zoomed-in images of Ypet-Atet-expressing clones showing CSP (d, arrowheads) and Syx1A (e, arrowheads) dots in filopodia. f A clone expressing Syt1-GFP where Syt1-GFP localized at the membrane and in filopodia (arrowheads). g ex vivo tracking of a Syt1-GFP vesicle (arrowhead) in a PG filopodium over time. Note that Ypet-Atet and Syt1-GFP were overexpressed for a short period of time (7 h) to achieve low protein expression levels. White scale bars: 20 μm; yellow scale bars: 5 μm. For each result shown here, n ≥ 3.

Fig. 5. Alteration of PG filopodia components leads to ecdysone secretion defects.

a Screening for α- and β-tubulin subunits involved in timing development by expressing RNAi under the control of phm-Gal4. b Screening for general actin-associated proteins involved in developmental timing. c Down-regulation of filopodia-associated proteins diaphanous, enabled, or α-actinin leads to a significant developmental delay or arrest. d-d” ex vivo PG confocal images of control, dia PG-downregulation, and Actn silencing showing its effect on filopodia (d, arrowheads). Quantifying filopodia length and density using PLC-GFP clonal analysis confirms that dia and Actn down-regulation significantly impair both filopodia length and density (d”). n^w1118 = 9 clones, n^{RNAi-dia33424} = 9, n^{RNAi-Actn7762} = 9 (d’). Compared to a control PG, dia or Actn PG silencing does not noticeably affect overall PG shape (d”). e 20E feeding rescues developmental delay caused by silencing dia or Actn expression in the PG. Data in (c) for RNAi-Actn⁷⁷⁶² and RNAi-dia³³⁴²⁴ are reused in (e) as controls without ecdysone feeding. The same phm>Dcr-2 x w¹¹¹⁸ control has been used in (a–c, e). f-f” Actn down-regulation affects ecdysone secretion, not production. For the total ecdysone measurement, the number of independent samples is the following: n^{phm>Dcr-2xw1118} = 3, n^{RNAi-Actnxw1118} = 3, n^{phm>Dcr-2xRNAi-Actn} = 2. For the % of circulating ecdysone, the number of independent samples is: n^{phm>Dcr-2xw1118} = 9, n^{RNAi-Actnxw1118} = 9, n^{phm>Dcr-2xRNAi-Actn} = 6. Dib expression is unaffected upon Actn silencing, as immunostaining shows (f’). Similarly, nvd and phm RNA levels remain unchanged in RNAi-Actn conditions, as assessed by qPCR (f”). Number of independent samples: n^{phm>Dcr-2xw1118} = 3, n^{RNAi-Actnxw1118} = 3, n^{phm>Dcr-2xRNAi-Actn} = 3. White scale bars: 20 μm; yellow scale bars: 5 μm. Data are presented as follows: means of the differences ± SEM (a–c, e), median ± IQR (d’), mean ± SEM (f, f”). Except for (e), data were subjected to the Kruskal-Wallis test, followed by Dunn’s multiple comparison tests. In (e), data were subjected to the two-tailed Mann–Whitney test. (ns not significant, *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001). Source data are provided as a Source Data file.

Fig. 6. Filopodia increase the PG exchange surface at the sub-BM domain.

a Representative images of an eL3 and lL3 PG cell expressing Syt1-GFP. Filopodia are pointed out with arrowheads. b–e Filopodia length increases significantly from eL3 to lL3 (b), while their diameter remains stable (c). The density of filopodia also increases significantly during the L3 stage (d), resulting in a 70% increase in the exchange surface of the sub-BM area (e). Number of clones analyzed: n^eL3 = 11, n^lL3 = 8. Scale bars: 20 μm. Data are presented as median of the means ± IQR in (b, c) and as median ± IQR in (d, e). Data were subjected to the two-tailed Mann-Whitney test (ns not significant, *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001). Source data are provided as a Source Data file.