Rabphilin involvement in filtration and molecular uptake in Drosophila nephrocytes suggests a similar role in human podocytes

The rabphilin 3A (RPH3A) gene encodes a Rab small GTPase family effector protein that has been implicated in vesicle docking/fusion reactions and the regulation of exocytosis and endocytosis processes in the central nervous system (Burns et al., 1998). RPH3A and Rab3A, a small G protein family member, complex with each other and are restrictedly expressed in neurons and podocytes, in which they are found around vesicles contained in the foot-processes (Rastaldi et al., 2003). RPH3A expression is altered in mouse and human proteinuric disease, suggesting a role for this protein in glomerulopathies (Rastaldi et al., 2003). Moreover, several human proteinuric diseases show enhanced RAB3A expression (Rastaldi et al., 2003). Other data that support the importance of RPH3A in human proteinuric diseases are combined genomic and metabolomic analyses that have previously associated increased urinary albumin excretion (UAE) in the general population (Marrachelli et al., 2014; Hwang et al., 2007) with an RPH3A gene polymorphism. This increase was independent of known factors that can influence UAE, such as hypertension and diabetes (Marrachelli et al., 2014). Importantly, the structural and functional impact of altered RPH3A expression on podocytes is still unknown.

Drosophila provides a suitable experimental model as it combines podocyte and renal proximal tubule functions in the same cell, the nephrocyte (Fu et al., 2017a). This makes Drosophila nephrocytes interesting for in vivo studies that can be difficult to perform in mammals due to a lack of accessibility. These nephrocytes that form the Drosophila excretory system (Denholm and Skaer, 2009; Helmstädter and Simons, 2017) are the cells involved in the removal of waste products from the haemolymph and Malpighian tubules. The function of Malpighian tubules, regarded as analogous to the renal tubular system, is to form urine for the excretion of toxic substances (Helmstädter and Simons, 2017; Helmstädter et al., 2017). There are two distinct nephrocyte populations: the garland cell nephrocytes around the oesophagus (binucleate) and the pericardial nephrocytes (mononucleate) situated along the heart. Both types of nephrocytes are functionally, structurally and molecularly similar to human podocytes (Helmstädter and Simons, 2017; Helmstädter et al., 2017; Na and Cagan, 2013; Fu et al., 2017b; Weavers et al., 2009; Hochapfel et al., 2017; Hermle et al., 2017; Zhuang et al., 2009).

As in mammalian podocytes, the nephrocyte basement membrane carries out charge-selective filtration (Hermle et al., 2017). The second filtration barrier is the nephrocyte diaphragm (ND) formed by Kirre and Stick and stones (Sns), which are the Drosophila orthologues of the mammalian slit diaphragm (SD) proteins Neph1 and nephrin, respectively. The nephrocyte membrane forms invaginations called labyrinthine channels, in which most endocytic uptake occurs (Helmstädter and Simons, 2017; Zhuang et al., 2009; Tutor et al., 2014). Kirre and Sns seal these invaginations to form 40-nm-wide slit pores, which are the entry to the labyrinthine channels (Zhuang et al., 2009). Cubilin (Cubn) and Amnionless are located in the innermost part of the labyrinthine channels, in which these proteins have an important role in endocytosis (Zhang et al., 2013a,b). Another requirement for the differentiation and maintenance of nephrocytes is the expression of the transcription factor Krüppel-like factor 15 (Klf15) (Ivy et al., 2015; Mallipattu et al., 2012). In Klf15 mutants, both garland cells and pericardial nephrocytes are absent, although the Klf15 mutant flies exhibit normal lifespan (Ivy et al., 2015).

In this study, we report the expression of Rph, the Drosophila orthologue of RPH3A, in the pericardial nephrocytes colocalizing with molecular markers of both endocytic and exocytic pathways. Nephrocyte-targeted interference of Rph expression blocked the endocytosis of different sizes of dextrans, impaired toxin removal and reduced the expression of essential nephrocyte genes, such as Cubn and sns. Importantly, Rph silencing disrupted the ultrastructure of the nephrocyte, abrogated the labyrinthine channels, reduced the number of endosomes and ultimately resulted in Drosophila nephrocyte loss. Our results indicate that Rph is necessary for the endocytic pathway and Rph loss of function has a strong impact on the structure, function and differentiation of nephrocytes. The role of Rph in the maintenance of the structure and function of nephrocytes opens a window to explore the contribution of RPH3A gene polymorphisms to the risk of developing chronic kidney disease.

In order to use Drosophila as a model to study the potential role of Rph in nephrocytes, we first investigated the expression and localization of this protein in pericardial nephrocytes of adult 7-day-old female flies. To label the nephrocytes, we used the UAS-Gal4 binary system to express the GFP marker under the control of the Hand-Gal4 driver (Sellin et al., 2006), which promotes expression in the cardiomyoblasts and pericardial cells starting at embryonic stage 12. Semi-intact heart preparations of Hand-Gal4>UAS-GFP flies allowed the direct visualization of pericardial nephrocytes under a fluorescence confocal microscope. To detect Rph, we took advantage of the high degree of evolutionary conservation of this protein (Fukuda et al., 2004) and performed immunofluorescence with a commercial rabbit polyclonal anti-Rph3A antibody generated against a synthetic peptide (amino acids 1-18) corresponding to rat Rph3A. In control flies, the immunofluorescence showed punctate localization of the signal in pericardial nephrocytes (Hand-Gal4>UAS-GFP; Fig. 1A-A‴). The signal decreased when the same immunofluorescence was performed in nephrocytes expressing a Rph RNAi construct (Perkins et al., 2015), demonstrating the specificity of the antibody (Hand-Gal4>UAS-GFP UAS-IR-Rph, abbreviated IR-Rph; Fig. 1B-B‴, Fig. S1A). Moreover, quantification of the Rph transcripts by RT-qPCR from manually isolated adult Drosophila hearts, which are enriched in nephrocytes, confirmed a decrease of up to 50% of the normal expression of this gene in the Rph RNAi sample compared with the control (Fig. S1B). Following the same immunofluorescence assay, we demonstrated that Rph is also expressed in nephrocytes from control larvae, and its signal is decreased in nephrocytes from Rph knockdown larvae (Fig. S2). To validate the RNAi Rph effect, we studied Rph expression in different interference lines. We observed that RNAi lines from different sources exhibit different levels of Rph signal (Fig. S3A-C″,D). In the three lines analyzed, Rph signal was reduced but the strongest effect was detected with line 25950 from Bloomington Drosophila Stock Center (BDSC), which was used as reference IR-Rph in the following experiments (Fig. 1, Fig. S1A,B).

To assess the role of Rph in Drosophila nephrocytes, we counted the number of functional pericardial nephrocytes (GFP⁺ cells with the nuclei intact) and measured cell volume in both control and IR-Rph flies, as these two parameters have previously been demonstrated to be informative about cell fate and activity, respectively (Ivy et al., 2015; Fu et al., 2017c). The volume dispersion index, calculated as SD volume/ median in 1-week-adult flies, was robustly increased in IR-Rph flies (Fig. 2A), thus reflecting an abnormal variability in the mean volume of Rph knockdown nephrocytes (Fig. 2C,D). Furthermore, the average number of functional nephrocytes was significantly reduced in IR-Rph flies compared with control flies (Fig. 2B, Fig. S3E). To assess whether the loss of nephrocytes in the IR-Rph flies was anticipated by defective differentiation during development, we studied the expression of the Drosophila orthologue of KLF15 (Klf15), which is critical for the development and differentiation of Drosophila nephrocytes, and can be used as a terminal differentiation marker (Ivy et al., 2015; Mallipattu et al., 2012, 2017). Immunofluorescence using a polyclonal antibody against a synthetic peptide of human KLF15 (amino acids 45-94), showed Klf15 expression in the nuclei of control nephrocytes (Fig. 2E-E″), whereas the signal was absent in most of the nephrocytes of the IR-Rph 1-week-old flies (Fig. 2F-F″, Fig. S4A). In contrast, Klf15 expression was normal in IR-Rph nephrocytes of 1-day-old flies (Fig. 2G-H″), indicating that the expression of this transcription factor is lost with time in IR-Rph nephrocytes. The specificity of this KLF15 antibody was tested in control and retinoic acid (RA)-supplemented nephrocytes (Fig. S5). Consistent with the well-known upregulation of Klf15 expression by RA, we observed that the levels of Klf15 in control flies fed with RA were robustly increased compared with the nephrocytes of control flies fed with standard food (Fig. S4B, Fig. S5).

Importantly, the GFP signal pattern in the 1-week-old nephrocytes was lighter compared with control nephrocytes at the same age (Fig. 2E,F, Fig. S3F). These results show that decreased levels of Rph in the pericardial nephrocytes promotes dedifferentiated nephrocytes, which are ultimately lost. Therefore, Rph is necessary for sustained nephrocyte cell fate.

It is well known that Rph participates in vesicular trafficking in other cell types, such as neurons (Burns et al., 1998). It has been shown that Rph can interact with Rab3a and Rab27a in vitro and can make complexes with these Rabs in rat adrenal medulla PC12 cells (Fukuda et al., 2004). Moreover, previous studies in mammalian models found Rph3a in complex with Rab3a in the kidney, in which they specifically localize in podocytes (Rastaldi et al., 2003). Given that the main effects of Rph3a are mediated by these Rab proteins in neuronal and endocrine systems, we performed double immunofluorescence with antibodies against Rph and Rab3 or Rab27 in pericardial nephrocytes (Fig. 3). Image analysis of the signals revealed that 58.8% of the Rph signal overlapped with Rab3, whereas 80% of the Rab3 signal coincided with Rph (Fig. 3A-A⁗,D). In the case of the colocalization with Rab27, we observed that almost all the Rph signal colocalized with the Rab27 protein (more than 96% of colocalization percentage), whereas only 13.9% of the Rab27 signal was overlapped with Rph (Fig. 3B-B⁗,E). These data support that the protein interaction of mammalian Rph in vesicular transport is conserved by its Drosophila orthologue in nephrocytes.

To study the role of Rph in the endocytic pathway, we performed a double immunofluorescence assay against Rph and Hepatocyte growth factor-regulated tyrosine kinase substrate, Hrs, (Fig. 3C-C⁗) in pericardial nephrocytes. Hrs is an essential protein that has been implicated in cell signaling and intracellular membrane trafficking (Raiborg and Stenmark, 2002; Raiborg et al., 2001). Several reports have demonstrated a role for Hrs in endocytic sorting of ubiquitinated membrane protein (Raiborg and Stenmark, 2002). According to the colocalization coefficients, 100% of the Rph signal colocalized with Hrs and 98.9% of Hrs signal colocalized with Rph (Fig. 3C-C⁗,F). These results suggested a role for Rph in the pericardial nephrocyte endocytic pathway and vesicular trafficking.

To test the functional relevance of lack of Rph in intracellular membrane trafficking, we performed an immunofluorescence assay against Rab3, Rab27 and Hrs in Rph RNAi knockdown nephrocytes.

We observed that the Hrs signal decreased ∼40% compared to control nephrocytes (Fig. S6), Rab3 expression decreased robustly (up to 70%; Fig. 3G-H″″, Fig. S7), and Rab27 signal remained unchanged (Fig. S6). This result indicates that the loss of Rph has a strong impact in Rab3⁺ vesicles and suggests an alteration in endocytic trafficking.

Nephrocyte scavenger activity of toxic molecules, such as ingested silver nitrate (AgNO₃), is mediated by endocytosis, and disruption of this pathway increases the mortality of larvae fed with AgNO₃ (Das et al., 2008). To test whether reduced Rph expression affects this crucial function of nephrocytes, we obtained the survival curves of adult and larval Rph RNAi knockdown flies fed with an AgNO₃-supplemented diet. In adults, we observed a reduction in lifespan in the Rph RNAi knockdown flies (Fig. S3G) that was not worsened by AgNO₃ (Fig. 4A). When larvae were fed with increasing concentrations of AgNO₃, pupation percentage dropped in IR-Rph larvae relative to wild-type controls (Fig. 4B, Fig. S3H). These findings indicate that reduced levels of Rph in Drosophila sensitize larvae to the toxicity of AgNO₃, which supports the notion that endocytosis is altered in Rph RNAi knockdown flies.

The insect nephrocyte combines filtration with protein reabsorption, using evolutionarily conserved genes and subcellular structures, which support its usefulness as a simplified model for both podocytes and the renal proximal tubule function (Fu et al., 2017a). Previous studies have shown that Drosophila nephrocytes can uptake fluorescent proteins from haemolymph (Weavers et al., 2009), which can be used as a reliable functional readout for nephrocytes in vivo. Circulating dextran molecules are filtrated from the haemolymph, endocytosed, and stored by nephrocytes, but the efficiency of this process is dependent on the size of the particles. In normal conditions, dextran molecules below 70 kDa easily cross the ND and are endocytosed in the labyrinthine channels by the Cubn-Amnionless system (Hermle et al., 2017; Zhang et al., 2013b). Defects in endocytosis might inhibit the uptake of proteins regardless of their size, and defects in the ND structure might facilitate the access of larger molecules to the labyrinthine channels (Hermle et al., 2017). Therefore, the accumulation of fluorescent dextran in the nephrocytes reflects both filtration efficiency and endocytosis. By adding fluorescently labelled dextrans of 10 kDa (Fig. 4E-E″,F-F″) and 70 kDa (Fig. 4G-G″,H-H″) to the dissected nephrocytes of adult flies, we observed that even the smallest sizes of dextran could not accumulate in the Rph RNAi knockdown nephrocytes (Fig. 4C,D). Thus, reduced expression of Rph alters molecular uptake by nephrocytes.

Drosophila Cubn and Amnionless are specifically expressed in nephrocytes and function as co-receptors for protein and molecular uptake, similar to their roles in the mammalian proximal tubule (Hermle et al., 2017; Zhang et al., 2013b) and podocytes (Prabakaran et al., 2012; Gianesello et al., 2017). To test the hypothesis that small size molecules were not endocytosed owing to an alteration in the Cubn-Amnionless system, we performed a quantitative determination of Cubn and Amnionless expression. Quantification of the transcripts of these two genes showed a reduction of Cubn expression of ∼50% of normal levels in the dissected Rph RNAi knockdown nephrocytes, whereas Amnionless levels were normal (Fig. 5A,B). These data suggest a relevant role for Rph in protein reabsorption through the maintenance of Cubn expression.

The expression of sns and kirre transcripts, which encode proteins that form the slit pores, was measured in control and IR-Rph nephrocytes. Although there were no changes in the expression of kirre (Fig. 5D), sns expression was 20% lower in the IR-Rph flies compared to controls (Fig. 5C). To independently assess the reduction of sns expression, we used a sns-Gal4 line to drive expression of the GFP protein as a reporter of sns expression. Semi-intact heart preparations of sns-Gal4>UAS-GFP UAS-IR-Rph flies were directly visualized under the fluorescence microscope (Fig. 5E,F), and homogenates were quantified for GFP fluorescence levels (Fig. 5G). We observed a decrease in the fluorescence signal in the Rph RNAi knockdown nephrocytes, thus supporting a role for Rph in the regulation of sns expression.

The filtration slit is a key functional structure in nephrocytes and disruption of the genes that form it leads to nephrocyte death (Zhuang et al., 2009). Transmission electron microscopy (TEM) was therefore used to examine the ultrastructure of pericardial nephrocytes in control and IR-Rph flies (Fig. 6). In comparison to control nephrocytes (Fig. 6A,A′), the size and number of NDs were not altered in the nephrocytes of the IR-Rph flies (Fig. 6B-D). However, the labyrinthine channels were absent in the nephrocytes of these flies and the basement membrane was altered in some regions. In addition, the number of endocytic vesicles and vacuoles was significantly reduced (Fig. 6E, Fig. S8), whereas there were no changes in the number of lysosomes (Fig. 6E). In agreement with our results regarding dextran uptake impairment and AgNO₃ sensitization, the changes in nephrocyte ultrastructure support disruption of the endocytic trafficking in the Rph RNAi knockdown nephrocytes.

RA is a metabolite of vitamin A, which is acquired directly through the diet as it cannot be synthesized by the body. It is involved in different processes such as cellular differentiation, proliferation, apoptosis regulation and inflammation inhibition (Niederreither and Dollé, 2008). It is known that RA also plays a role in podocyte differentiation via KLF15 regulation (Mallipattu et al., 2012; Niederreither and Dollé, 2008; Guo et al., 2018; Mallipattu and He, 2015; Sharma et al., 2014). The restoration of podocyte differentiation markers by RA has allowed this metabolite to be considered as a treatment for renal diseases (Mallipattu and He, 2015). Non-functional pericardial nephrocytes are characterized by a loss of Sns functionality, including filtration and reabsorption defects. These non-functional nephrocytes undergo the process of dedifferentiation, and their nuclei are fragmented and their GFP signal is reduced (Fig. 1B-B″). The expression of the IR-Rph construct causes a loss of function in the nephrocytes and, consequently, a decrease in their number and a decrease in fly survival. In this study, we administered different concentrations of RA to IR-Rph flies. We observed that the flies treated with RA presented a significantly higher number of functional nephrocytes than the IR-Rph flies fed with the standard nutritive media (Fig. 7A). Additionally, these RA-treated flies showed significantly longer survival curves (Fig. 7B). These results suggest that RA could improve functional defects in IR-Rph flies.

In order to determine whether this functional restoration is due to dedifferentiation inhibition, as described in podocytes via the regulation of KLF15 (Mallipattu et al., 2016, 2012), We carried out an immunoassay for these proteins. In IR-Rph flies fed with nutritive medium containing RA at 10 μM (Fig. 7D-D‴), the Klf15 signal was clearly more intense than in IR-Rph flies fed with standard nutritive medium (Fig. 7C-C‴), whereas in both cases Rph levels in the nephrocytes were reduced compared with control (healthy) nephrocytes (compare nephrocytes from Fig. 7C″,D″ with the nephrocyte from Fig. 1A″) and RA had no effect on Rph levels in IR-Rph flies fed with the compound (Fig. 7C″,D″). These results indicate that the increment in the functional nephrocyte number and the lifespan of the flies could be a result of Klf15 expression regulation.

We assessed whether RA was capable of rescuing reabsorption or endocytosis by administering 10 kDa and 70 kDa of dextran to IR-Rph flies fed with either standard nutritive medium (control) or supplemented with RA at 10 μM (Fig. 8). Flies treated with RA captured the 10-kDa dextran molecules significantly more efficiently than the non-treated group (Fig. 8A,A′,C,C′). The 70-kDa dextran molecules were also captured (Fig. 8B,B′,D,D′), however not as efficiently as in the control nephrocytes shown in Fig. 6. These findings suggest that RA partially rescues the filtration and endocytic processes in Drosophila nephrocytes.

Accordingly, when RA was administered, sns and Cubn expression both increased to normal levels in IR-Rph flies (Fig. 8E,G), with no changes in Kirre and Amnionless gene expression (Fig. 8F,H). These findings suggest that the compound rescues the filtration and endocytic functions mediated by the Cubn/Amnionless complex, which were decreased in IR-Rph flies. This restoration could be due to the upregulation of Klf15, which controls sns expression, as proven in previous studies (Mallipattu et al., 2012; Guo et al., 2018). Additionally, as shown in the TEM images, the nephrocytes of RA-treated flies recovered the labyrinthine channel structures (Fig. 8I,J), which were absent in non-treated IR-Rph nephrocytes (Fig. 6). These results suggest that the restoration of Klf15 expression by RA is sufficient to partially rescue the filtration and endocytosis defects shown by IR-Rph flies.

Rph3A expression is known to be altered in proteinuric diseases, such as in human and mouse models (Rastaldi et al., 2003). In addition, previous studies have suggested that a polymorphism in the RPH3A gene increases the risk of microalbuminuria (Marrachelli et al., 2014). However, little is known about the role of this protein in podocytes. Taking advantage of the high degree of conservation of the filtration structures between human podocytes and Drosophila nephrocytes, and the myriad of genetic tools available in flies, we studied the role of Rph in this model system and the consequences of silencing its expression.

In this study, we show that Rph, a protein that is well known as a Rabs effector, is expressed in Drosophila pericardial nephrocytes, and has a relevant function in haemolymph clearance, the maintenance of essential structures and nephrocyte cell fate. However, it is not clear whether the dysfunction of one of these aspects is a cause or consequence of the dysfunction of the others. As Fig. S9 illustrates, the loss of Rph function had an impact on basement membrane thickness and the maintenance of labyrinthine channels. Moreover, Rab3⁺ vesicles were greatly reduced in Rph RNAi knockdown nephrocytes, indicating a vesicular trafficking reduction and suggesting a blockage of Rab3-mediated endocytosis. Consequently, reduced levels of Rph in Drosophila nephrocytes sensitized larvae to the toxicity of AgNO₃. Of note, we did not observe toxicity in adult flies, which supports the hypothesis that nephrocytes are the primary system for mitigating xenotoxin insults in larvae, but other cells can compensate for their loss in adults (Ivy et al., 2015). Finally, the interruption of the endocytosis pathway suggests a role for Rph in protein reabsorption, presumably by the modulation of Cubilin expression, which was reduced.

Several studies show that the endocytic pathway plays an important role in the development, maintenance, and damage of the podocyte and might lead to alterations in cell morphology (Hochapfel et al., 2017; Bechtel et al., 2013; Chen et al., 2013). In connection with this observation, our data support a role for Rph in the maintenance of nephrocytes, which was assessed by the reduced expression of terminal marker Klf15. It is known that reduced levels of KLF15 in podocytes promote alteration of the filtration and endocytic pathways, which ultimately causes the loss of cell fate (Mallipattu et al., 2012). In our study, Klf15 levels were reduced by the interference of Rph expression but either those levels were not low enough to produce changes in Amnionless expression or we could not detect Amnionless reduction with RT-qPCR. In this regard, it should be taken into account that the dissected tissue sample used for RNA extraction to perform RT-qPCR did not exclusively contain nephrocytes. It was contaminated with remnants of cardiac tissue and body wall muscles. This experimental limitation might have concealed small differences in gene expression levels. Of note, we were able to detect an impact of Klf15 levels on Cubilin expression, which is part of the Cubn-Amnionless complex.

A reduction in the levels of Rph also affects the expression of genes directly involved in the nephrocyte structure. Sns and Kirre (human Nephrin and NEPH1 orthologues) interact through their extracellular domains to form the ND and are essential for its formation. The Cubn-Amnionless system is in charge of the endocytosis of small size proteins across the labyrinthine channels (Hermle et al., 2017; Zhang et al., 2013b). Both systems can be altered by a reduction in the gene expression or mislocalization of any of their components.

The absence of sns expression led to a dramatic reduction of ND number (Zhuang et al., 2009). In our case, TEM images did not show any effect in ND number, probably because the reduction in sns expression of IR-Rph nephrocytes was only 20%. Accordingly, recent studies have documented the significance of cytoplasmic Sns in regulating actin organization (Muraleedharan et al., 2016), which could be direct or a result of altered cortical actin when ND proteins are lost or reduced. The same study highlights the role of the reciprocal regulation between cytoskeletal components and ND proteins as essential for size and charge-dependent filtration.

The RA-treated nephrocytes showed sns and Cubn expression levels similar to control nephrocytes (Fig. 8E,G), presented labyrinthine channels (Fig. 8I) and were able to carry out the filtration function (Fig. 8C′) and partial endocytosis (Fig. 8D′). These restorations in IR-Rph nephrocytes are probably due to the upregulation of Klf15 expression through the administration of RA and demonstrate that forced maintenance of the nephrocyte differentiation state rescues Rph silencing phenotypes.

Our data highlight the relevance of Rph in nephrocyte structure and function. Specifically, knockdown of Rph seems to impair nephrocyte performance at two levels: filtration by basement membrane disruption and subsequent cytoskeletal modifications leading to important structural alterations (absence of labyrinthine channels) and nephrocyte loss, and protein uptake by influencing Cubn expression and vesicular trafficking (Fig. S9). This suggests a dual role for Rph3A in the mammalian excretory system and could explain why a polymorphism in human RPH3A leads to microalbuminuria (Marrachelli et al., 2014). Importantly, the Rph3A expression pattern in mammals is not well described, and its presence in different tissues, either podocyte or proximal tubule, might influence its activity. The role of Rph in the maintenance of the structure and function of nephrocytes opens a window to explore the contribution of RPH3A gene polymorphisms to the risk of developing chronic kidney disease.

Hand-Gal4, UAS-IR-Rph (BDSC, 25950), UAS-IR-bcd, UAS-GFP and yw were obtained from BDSC (Indiana University), sns-Gal4 UAS-GFP was obtained from M Ruiz-Gómez (Centro de Biología Molecular Severo Ochoa, Spanish National Research Council and Autonomous University of Madrid, Spain). Other UAS-IR-Rph lines used were line 1, 524338 (construct ID: 7330), and line 2, 109337 (construct ID: 107492), which were obtained from the Vienna Drosophila Resource Center. The recombinant line Hand-Gal4 UAS-GFP was generated during this study to mark adult pericardial nephrocytes. All crosses were maintained at 25°C on standard nutritive medium.

More than 50 newly hatched males per genotype were collected, placed in tubes containing standard nutritive medium or medium supplemented with different concentrations of AgNO₃ (0.005% or 0.01% AgNO₃) and kept at 25°C (Fig. 4A) or 29°C (Fig. S3G). The number of deceased flies was scored on a daily basis, and flies were transferred to fresh medium every 2 to 3 days. Survival curves were obtained using the Kaplan–Meier method, and statistical curve comparisons were carried out according to the log-rank (Mantel–Cox) test (α=0.05).

Twenty first instar larvae (L1) from each genotype were transferred to vials supplemented with different concentrations of AgNO₃ and maintained at 25°C. The number of larvae that reached pupal stage was scored to determine the toxicity. Results were analyzed by two-tailed unpaired Student's t-test (α=0.05), applying Welch's correction whenever necessary. The experiment was performed in triplicate.

One-day-old Hand-GFP>UAS-IR-Rph females were fed with standard media supplemented with different concentrations of RA diluted in food (1 μM, 10 μM and 50 μM) for 1 week; for the survival analyses, the flies were fed with this medium until their death, and for the Klf15 analysis, flies were fed with 10 μM RA from the larval stage to 1-week-old adults. RA-treated flies were maintained at 25°C for all tests, except for the lifespan analysis in which they were maintained at 29°C. All the experiments were performed in triplicate.

Adult hearts with pericardial nephrocytes from 7-day-old females were dissected in artificial Drosophila haemolymph according to Selma-Soriano et al. (2018a), fixed with 4% paraformaldehyde (PFA) in PBS for 20 min and permeabilized by washing three times with PBS containing 0.3% Triton X-100 (PBS-T) for 10 min. Then, hearts were blocked in PBS-T containing 0.5% bovine serum albumin for 30 min at room temperature and incubated with the corresponding primary antibody (1:100) overnight at 4°C. Primary antibodies used were anti-Rph (Abcam, ab3338; with 50% of homology), anti-Hrs (DSHB, AB 2722114), anti-Rab3 and anti-Rab27 (BD Bioscience, 610379 and 558532, respectively) and anti-Klf15 (Ivy et al., 2015) (Abcam, ab22851; with 59% of homology). After three washes with PBS-T, the secondary antibodies (1:200), AlexaFluor-647 donkey anti-rabbit (Life Technologies, A31573), anti-mouse biotinylated (Sigma-Aldrich, B7264) and anti-rabbit biotinylated (Sigma-Aldrich, B8895) were incubated for 2 h at room temperature. Hearts with pericardial nephrocytes were then incubated with ABC solution (ABC kit, VECTASTAIN) for 30 min at room temperature, followed by washes and incubation with streptavidin-Texas Red (Vector Laboratories, 1:500). All images were taken using an LSM 800 confocal microscope (Zeiss) and were processed using ZEN software.

ZEN software was used to quantify Rph, Klf15 and GFP signals from immunostaining images. The area of the nephrocyte signal was selected and the intensity of the pixels was scored. For the analysis, at least three different biological samples were used. Results were analyzed using a two-tailed unpaired Student's t-test (α=0.05), applying Welch's correction whenever necessary.

Homogenates from five 1-week-old females, per triplicate, were quantified using a fluorescence microplate reader (Tecan). Results were analyzed using a two-tailed unpaired Student's t-test (α=0.05), applying Welch's correction whenever necessary.

For the analysis of the number and volume of nephrocytes, adult female fly hearts were dissected in 1× PBS. Confocal images were obtained with a FLUOVIEW FV1000 confocal microscope (Olympus) using a 10× or 20× air objective. Three-dimensional structures were reconstructed from confocal stacks using Imaris 7.1 software (Bitplane). For quantification, confocal imaging settings were invariant within each experiment. For volume quantification, at least three nephrocytes from three different flies were analyzed.

Dissected abdomens from 7-day-old females were processed as described previously by Selma-Soriano et al. (2018b). Briefly, abdomens were fixed in 1% PFA and 2% glutaraldehyde in 0.2 M phosphate buffer for 30 min and postfixed with 1% osmium tetroxide for 2 h. After dehydration, abdomens were embedded in epoxy resin. Semi-thin (1.5 μm) and ultra-thin sections were obtained using an ultramicrotome (Ultracut E, Reichert-Jung and Leica). Samples were analyzed using a JEOL JEM-1010 TEM at an operating voltage of 80 kV.

Quantification of TEM images was carried out by counting the number of SDs in the plasma membrane of pericardial nephrocytes at 8000× magnification. The pore size was measured at three different levels (innermost, middle and uppermost pore site) at 30,000× magnification. The number of endosomes and lysosomes was obtained from 800× magnification images of sections containing entire cells. The endosomes are the white vesicles near the membrane (Fig. S8) and the lysosomes are the dark vesicles located at the centre of the nephrocyte. Images from at least four different nephrocytes of the same genotype were scored. Results were analyzed using a two-tailed unpaired Student's t-test (α=0.05).

Adult female fly hearts were dissected in artificial Drosophila haemolymph and were incubated with 0.1 mg/ml 10-kDa dextran (AlexaFluor568, Invitrogen) or with 0.1 mg/ml 70-kDa dextran (Thermo Fisher Scientific, D1830). After 15 min of incubation, the hearts were washed three times with PBS. Pericardial nephrocytes were then immediately mounted for confocal imaging. Zen Software was used to measure the volume and signal of fluorescent dextran for the dextran quantification. At least three nephrocytes from three different flies were analyzed. Results were analyzed using a two-tailed unpaired Student's t-test (α=0.05).

For RT-qPCR, RNA was first isolated using TRIzol Reagent (Invitrogen) from 100 dissected adult hearts (pooled). RNA purity and concentration were determined using a NanoDrop 1000 (Thermo Scientific). Total RNA (1 μg) was reverse transcribed using Superscript II Reverse Transcriptase (Invitrogen). SYBR-Green-based real-time qPCR was performed using a QuantStudio 5 Real-Time PCR System (Applied Biosystems). Rp49 and Tub48B were used as endogenous references. In all cases, relative expression to the endogenous genes and the control group was obtained by the 2^−ΔΔCt method. Pairs of samples were compared using a two-tailed unpaired Student's t-test (α=0.05), applying Welch's correction whenever necessary.

We thank the Microscopy Facility of the Servei Central de Suport a la Investigació Experimental at the University of Valencia for their technical support with the TEM.