For the transgenic RNAi and overexpression studies, the UAS-Gal4 system was employed. Flies were kept at 25°C or 29°C on standard food. RNAi stocks were obtained from the Vienna Drosophila Resource Center (Vienna, Austria): control RNAi targeting orco (100825), rap1 RNAi (20761), rap1 RNAi (110757) and magi RNAi (41735). UAS-Rap1^V12 (constitutively active), UAS-Rap1^N17 (dominant-negative), and UAS-Rap1 (wild-type) were a gift from Ulrike Gaul (LMU, Munich, Germany). The overexpression control UAS RFP (30556) was obtained from the Bloomington Stock Center (Bloomington, United States). UAS Sns and Sns-Gal4 driver lines were provided from Tobias Huber (UKE, Hamburg, Germany). For the rescue experiments, driver lines were recombined with the specific RNAi fly lines.

The pENTR™⁄DTOPO® Kit (ThermoFisher) was used for cloning of h-Rap1B V12, h-Rap1B N17, and h-Rap1B WT into pENTR (human Rap1B DNA was provided from Andreas Püschel, WestfälischeWilhelms-Universität Münster, Germany). Cloning into the destination vector was performed using the Gateway technology (Invitrogen). UAS-h-Rap1B^V12, UAS-h-Rap1B^N17, and UAS-h-Rap1B (wild-type) transgenic fly lines were generated using ΦC31-mediated germline transformation using landing sites attP40 and attP2.

Wandering third instar larvae were dissected and the garland cell nephrocytes fixed in 4% paraformaldehyde (PFA) for 15 min. After washing in PBS-T (PBS + 0.1% Triton), nephrocytes were incubated in the primary antibody at 4° overnight, followed by washes in PBS-T. The secondary antibody incubation lasted for 2.5 h at room temperature. After washes with PBS-T, they were mounted in Mowiol. Anti-Sns [custom generated (Dlugos et al., 2019)], anti-Pyd (PYD2) and anti-integrin β (CF.6G11) (all from Developmental Studies Hybridoma Bank, United States) were used as primary antibodies. Anti-Sns antibodies were diluted 1:100, anti-Pyd antibodies 1:20, anti-integrin β antibodies 1:100 in PBS-T. Goat-anti-mouse Alexa⁴⁸⁸, goat-anti-rabbit Alexa⁵⁹⁴, and DAPI (all from Invitrogen) were used as secondary antibodies and diluted 1:1000 in 10% goat serum in PBS-T.

Tangential und surface section images of the nephrocytes were obtained using a confocal microscope (Leica SP8). 1.5× zoom was set for the tangential section and 4× zoom for the surface section. For imaging, the integrated module LIGHTNING was applied. Image processing was done by LasX and ImageJ Software.

Wandering third instar larvae were dissected, and garland cell nephrocytes were fixed in 2% glutaraldehyde in Sørensen buffer overnight. Next, nephrocytes were washed with Sørensen buffer, osmium tetroxide 1% in Sørensen buffer was applied for 1 h, and samples were again washed with Sørensen buffer. Samples were then dehydrated in an ascending alcohol series and infiltrated with epon using a series of mixtures of epon and the intermedium propylene oxide and pure epon. After embedding in epon and polymerization at 60°C for 36 h, samples were cut in ultra-thin slices of 60 nm and contrasted with uranyl acetate for 20 min and lead citrate for 90 s. Images were taken with a transmission electron microscope (Phillips CM10 equipped with TVIPS CAM F416).

Quantification of slit diaphragms visualized across 1 μm was performed manually (LasX Software, EM measure). The Mann–Whitney U test was used to determine the statistical significance between two interventions. Slit diaphragms were quantified in 14 nephrocytes from 20 different animals for each of three independent crosses per genotype. Slit diaphragms were counted perpendicularly across 5 µm cell membrane in five different areas of each cell. To detect integrin β localization upon downregulation of rap1, ROIs were selected at the cell cortex and the cytosol close to the nucleus. The integrin reorganization was then calculated as previously described (Bayraktar et al., 2020): ROI at cell cortex + background area/ROI at cytosol + background area. Slit diaphragms were counted per µm basement membrane for 10 nephrocytes, each of 20–30 animals employing transmission electron microscopy images. Basement membrane length was quantified using ImageJ.

Human immortalized podocytes were cultured as previously described (Dlugos et al., 2019). Stable podocyte cell lines allowing doxycycline-dependent expression of CD16-CD7-Nephrin Cytoplasmic Domain (NCD) or CD16-CD7-HA (pInducer-21-puro-CD16-CD7-NCD or pInducer-21-puro-CD16-CD7-HA) were generated by lentiviral gene transfer as previously described (Dlugos et al., 2019).

Activation of nephrin was performed as previously described (Dlugos et al., 2019). In short, either CD16-CD7-NCD or CD16-CD7-HA (as control) stable cell lines cultured on 10 cm dishes were cooled down to 4°C for 5 min, incubated with an anti-CD16 antibody (#555404, BD Pharming, 4 µl per ml medium) on ice for 30 min, and then further incubated with an anti-IgG antibody (#ab6708, Abcam, 3 µl per ml medium) for another 5 min at 37°C. The first incubation step allows the anti-CD16 antibody to bind to the CD16 domain of the chimeric proteins and leads to a first clustering of the proteins. The second incubation step allows the binding of the anti-IgG antibody to the bound anti-CD16 antibodies for an enhancement of clusters, which initiates recruitment of Src kinases and consecutive phosphorylation of the intracellular nephrin domain (Verma et al., 2003; Jones et al., 2006; Verma et al., 2006).

After nephrin activation by clustering, cells were directly cooled down on ice, media was removed, and cells were washed with ice-cold PBS. Cells were then lysed in 500 µl ice-cold IP buffer (20 mM Tris-HCl, pH 7.4; 20 mM NaCl; 1 mM EDTA; 50 mM NaF; 15 mM Na₄P₂O₇; 1% (v/v) Triton X-100) containing protease inhibitor (cOmplete™ protease inhibitor cocktail, Roche Diagnostics) and phosphatase inhibitors (phosphatase inhibitor cocktail 2 and phosphatase inhibitor cocktail 3, Sigma Aldrich) and manually scratched from the plates. Further mechanical lysis was achieved by dragging the lysates through a 26, G needle. Lysates were kept on ice for 30 min, whereby they were vortexed every 3 to 5 min and were then centrifuged for 15 min at 14.000 × g and 4°C. The supernatant was transferred into a new reaction tube containing Sepharose G beads (GE Healthcare), which were washed three times for 5 min rotating in 1 ml IP buffer. 5% of the lysate was kept aside as input. 1 µl of active Rap1-GTP antibody (#26912, NewEast Biosciences) was added to 30 µl of beads slurry and lysate. Immunoprecipitation was performed rotating at 4°C for 1 h. Afterward, beads were washed three times at 4°C, rotating with 1 ml IP buffer each time, containing protease and phosphatase inhibitors. Following the washing steps 2× Laemmli was added to the beads, and the samples were boiled at 95°C for 5 min. Immunoblotting was performed as previously described (Dlugos et al., 2019). Primary antibodies were diluted 1:1000 in 5% bovine serum albumin in TBST. The following antibodies were used: Rap1A/B (#2399S, Cell Signaling Technologies), nephrin (#BP5030, OriGene), p-nephrin (#ab80298, Abcam), HA (#11867423001, Roche), β-tubulin (#T8328, Sigma Aldrich).

We recently showed that nephrin activation results in the activation of integrin β1 in podocytes (Dlugos et al., 2019). C3G appears to play a role in this pathway (Dlugos et al., 2019). C3G is an activator of the small GTPase Rap1 (Radha et al., 2011). However, the role of Rap1 in nephrin signaling to focal adhesions in podocytes is unclear. To analyze the function of Rap1 in nephrin signaling to integrin β, we combined the use of two model systems: podocyte culture and the in vivo Drosophila nephrocyte model. The nephrocyte expresses a slit diaphragm-like structure based on sticks and stones (Sns, ortholog of nephrin) and Kin-of-irre (Kirre, ortholog of Neph1) (Weavers et al., 2009; Zhuang et al., 2009). Like in podocytes, the nephrocyte slit diaphragm functions as a size and charge-selective barrier (Zhang et al., 2013a). In analogy to podocytes, nephrocytes express the cytoskeletal adapters Cindr (ortholog of CD2AP), Myoblast city (Mbc), Nck ortholog), and Mec-2 (ortholog of podocin) (Weavers et al., 2009; Zhuang et al., 2009). Thus, this model is currently often employed as an in vivo model for analyzing the slit diaphragm in a genetically tractable system (Zhang et al., 2013a; Zhang et al., 2013b; Fu et al., 2017; Hermle et al., 2017; Hochapfel et al., 2017).

To test whether Rap1 is essential for targeting of integrin β, we downregulated endogenous rap1 exclusively in nephrocytes employing the UAS-Gal4 system with a nephrocyte-specific driver (sns-Gal4) and two independent dsRNA lines targeting rap1. The localization of myospheroid, the Drosophila ortholog of integrin β, was determined in nephrocytes of third instar larvae using confocal microscopy. In tangential sections, integrin β exclusively accumulated at the cell cortex in control nephrocytes while nephrocytes with knockdown of rap1 showed diffuse cytoplasmic staining of integrin β with some aggregates (Figure 1A). When imaging the surface of nephrocytes, integrin β presented in a typical fingerprint-like pattern in control nephrocytes (Figure 1B), whereas its localization was severely altered with diffuse targeting of integrin β in rap1 knockdown nephrocytes (Figure 1B). Integrin accumulation at the plasma membrane was quantified in tangential sections (Figure 1C) (Bayraktar et al., 2020). To confirm that integrin β mistargeting is a specific effect of rap1 downregulation, we performed rescue experiments of the rap1 knockdown by expressing human Rap1B, which is not targeted by the dsRNA specific to the Drosophila rap1 gene. Nephrocytes expressing a control element [UAS-RFP, called control OE (overexpression)] instead of human Rap1B on rap1 knockdown background were prepared, and immunofluorescence analysis was performed with antibodies specific for integrin β (genotype: sns-Gal4, UAS rap1RNAi ²⁰⁷⁶¹ /UAS RFP or sns-Gal4, UAS rap1RNAi ¹¹⁰⁷⁵⁷ /UAS RFP). Analysis of tangential (Figure 2A, Supplementary Figure S1A) and surface sections (Figure 2B, Supplementary Figure S1B) of nephrocytes confirmed that downregulation of rap1 resulted in diffuse targeting of integrin β. Upon knockdown of rap1 and concomitant expression of UAS-h-Rap1B, the phenotype induced by silencing of rap1 was partly rescued. In these animals, integrin β was localized correctly at the plasma membrane (Figure 2A, Supplementary Figure S1A) and was found in the fingerprint-like pattern in surface sections (Figure 2B, Supplementary Figure S1B) (genotypes: sns-Gal4, rap1RNAi ²⁰⁷⁶¹ /UAS h-Rap1B; UAS h-Rap1B/Tm6b or sns-Gal4, rap1RNAi ¹¹⁰⁷⁵⁷ /UAS h-Rap1B; UAS h-Rap1B/Tm6b). We quantified the number of foot processes per 1 µm of the basement membrane, which confirmed the qualitative results (Figure 2C, Supplementary Figure S1C).

The activation state of small GTPases is tightly regulated by activating and inactivating factors (Lagarrigue et al., 2016; Jaskiewicz et al., 2018). To deduce whether the activity state of Rap1 is important for correct targeting of integrin β, we expressed wild-type Drosophila Rap1, dominant-negative Rap1^N17 with amino acid substitution of serine to asparagine at position 17, constitutively active Rap1^V12 with amino acid substitution of glycine to valine at position 12 or RFP as a control in wild-type nephrocytes. Compared to the control (genotype: sns-Gal4/UAS RFP), overexpression of rap1 resulted in moderate mistargeting of integrin β, which was more severe when overexpressing either constitutively active or dominant-negative Rap1 (genotypes: sns-Gal4/UAS d-rap1 or sns-Gal4/UAS d-rap1N17 or sns-Gal4; d-rap1V12) (Figures 3A–C). To test whether ectopic expression of human Rap1B variants exerts a similar phenotype to their Drosophila orthologs, we expressed human h-Rap1B, h-Rap1B^V12, h-Rap1B^N17, or RFP as a control on the wild-typic background (genotypes: sns-Gal4/UAS RFP or sns-Gal4/UAS h-Rap1B; UAS h-Rap1B or sns-Gal4/UAS h-Rap1B ^N17 ; UAS h-Rap1B ^N17 or sns-Gal4/UAS h-Rap1B ^V12 ; UAS h-Rap1B ^V12). Tangential sections (Figure 4A) or surface sections (Figures 4B,C) showed that misexpression of all three h-Rap1B variants resulted in mistargeting of integrin β, demonstrating that the function of rap1 in focal adhesion integrity appears to be conserved.

In mice, Rap1 is essential for slit diaphragm integrity (Potla et al., 2014). In the Drosophila nephrocyte model, the role of Rap1 in slit diaphragm integrity has not yet been characterized. Knockdown of rap1 by two different dsRNA in nephrocytes in vivo employing sns-Gal4 resulted in incorrect targeting of the slit diaphragm proteins Sns and Pyd (the fly ortholog of Zonula occludens (ZO-1) in tangential and surface sections (Supplementary Figures S2A,B). Next, we analyzed whether the expression of h-Rap1B can rescue the rap1 knockdown phenotype of slit diaphragm protein mistargeting. This showed that expression of h-Rap1B could rescue the rap1 knockdown-related phenotype of Sns and Pyd mistargeting induced by two independent dsRNA (Supplementary Figures S3A,B, S4A,B). Thus, the Drosophila nephrocyte model recapitulates the phenotypes induced by Rap1 loss of function in mice. We then utilized the nephrocyte model to characterize the role of Rap1 for slit diaphragm integrity further. It is presently not clear whether a balanced Rap1 activity is crucial for slit diaphragm integrity in mammalian podocytes in vivo or Drosophila nephrocytes. In contrast to the overexpression control (RFP), nephrocyte-specific overexpression of rap1, rap1 ^V12 , and rap1 ^N17 on the wild-typic background resulted in mistargeting of slit diaphragm proteins, Sns and Pyd. Similar to targeting of integrin β to focal adhesions, targeting of Sns and Pyd to the slit diaphragm was disturbed by expressing wild-type rap1 or the constitutively active or dominant-negative forms of rap1 (Figures 5A,B). To analyze whether this correlates with the breakdown of slit diaphragms, we performed transmission electron microscopy (TEM) analysis. Nephrocyte-specific overexpression of either wild-type, dominant-negative or constitutively active rap1 led to a reduction of slit diaphragms per µm basement membrane (Figures 6A,B). In analogy, misexpression of wild-type human Rap1B, constitutively active h-Rap1B^V12 or dominant-negative h-Rap1B^N17 on wild-typic background resulted in misexpression of Sns and Pyd and breakdown of slit diaphragms (Supplementary Figures S5A,B, S6). Thus, slit diaphragm integrity appears to be regulated by Rap1 activity.

It is well known that Drosophila is excellently suited to genetically dissect complex signaling cascades (Simon et al., 1991; Fortini et al., 1992; Brunner et al., 1994). The ability to suppress a phenotype caused by overexpression of a given gene (gain-of-function phenotype) by suppressing another gene is evidence that the two genes interact (Simon et al., 1991; Fortini et al., 1992; Brunner et al., 1994). We employed this technique to test whether the nephrin ortholog sns genetically interacts with rap1. Flies overexpressing sns in nephrocytes were crossed with flies with nephrocyte-specific knockdown of rap1 (sns-Gal4, UAS-sns; UAS-d-rap1RNAi ¹¹⁰⁷⁵⁷ or sns-Gal4, UAS-sns; UAS-d-rap1 RNAi ²⁰⁷⁶¹ ) and compared to flies with overexpression of sns and a control UAS-dsRNA element (sns-Gal4, UAS-sns; UAS-control RNAi). While flies overexpressing sns in a background of a control UAS-dsRNA element showed mistargeting of integrin β in nephrocytes, flies that overexpress sns in the background of rap1 suppression showed a normal integrin β localization in nephrocytes (Figures 7A–C). Importantly, two independent UAS-dsRNA elements tested could rescue integrin β targeting, implying that Rap1 functions downstream of Sns in signaling to integrin β at focal adhesions. Downregulation of rap1 was also able to partly rescue Sns and Pyd targeting on an sns gain-of-function genetic background (Supplementary Figures S7A,B). Thus, Rap1 also functions downstream of Sns during signaling that regulates slit diaphragm morphology.

Slit diaphragm integrity and integrin targeting were also rescued when overexpressing d-rap1 on the sns knockdown background (sns-Gal4>>sns RNAi x UAS-d-rap1, control: sns-Gal4>>sns RNAi x UAS-control) (Figures 8A–C). This confirms that Rap1 fulfills functions downstream of Sns.

In mammalian podocytes, Rap1 activity is also regulated via the scaffold proteins, MAGI-1 and MAGI-2. Drosophila has only one MAGI ortholog. To test whether dMAGI is essential for slit diaphragm integrity in nephrocytes, we performed a knockdown of the MAGI ortholog by RNAi in nephrocytes (sns-Gal4). Immunofluorescence analysis of magi knockdown nephrocytes showed a slightly reduced number of slit diaphragms when stained for Sns and Pyd to visualize slit diaphragms (Supplementary Figure S8). Thus, MAGI is also necessary for an intact slit diaphragm in nephrocytes.

To confirm that nephrin activation results in signaling that leads to activation of Rap1 in mammalian podocytes, we employed a podocyte culture system where nephrin signal transduction can be activated (Figure 9A) (Jones et al., 2006; Verma et al., 2006; George et al., 2012; George et al., 2014). Cultured human podocytes that inducibly express a chimeric nephrin protein by doxycycline (dox) treatment consisting of a CD16 extracellular domain, a CD7 transmembrane domain, and the cytoplasmic domain of nephrin (CD16-CD7-NCD). Incubation with anti-CD16 antibody induces clustering of chimeric nephrin, initiating recruitment of Src kinases to the nephrin intracellular domain and consecutive phosphorylation on activating tyrosine residues known to bind to cytoskeletal adapter proteins such as Nck, Crk1/2, CrkL, phospho-inositol-3-kinase (pi3k), and other signaling intermediaries (Jones et al., 2006; Verma et al., 2006; George et al., 2012; George et al., 2014). As controls, podocytes inducibly expressing CD16-CD7-HA, where the cytoplasmic domain of nephrin is replaced by HA-tag, were used. Podocyte lines were treated with dox to induce the expression of either CD16-CD7-NCD or CD16-CD7-HA followed by incubation with antibodies specific for CD16 and secondary IgG to trigger clustering. Podocyte lysates were then subjected to immunoprecipitation employing an antibody specific for GTP-bound, active Rap1A/B followed by immunoblotting with an antibody recognizing total Rap1A/B. Furthermore, immunoblots were performed with antibodies specific for nephrin, phospho-nephrin (pNephrin), HA, total Rap1A/B, and β-tubulin to control loading. This showed that the amount of the active form of Rap1-GTP was increased following activation of nephrin (Figure 9B).