To investigate the effect of superfluous fat consumption on nephrocyte function, newly eclosed flies were fed a normal fat diet (NFD) or a HFD for 7 days. Then, the flies were subjected to fluorescent dye uptake assays (Weavers et al., 2009; Zhao et al., 2025) to determine pericardial nephrocyte function. HFD leads to enlarged crop (Liao et al., 2021; Zhao et al., 2023a), which was used as an indication of food consistency. Compared to the NFD-fed flies, the HFD-fed flies showed reduced FITC-conjugated albumin (FITC-albumin, 66 kD) fluorescence (Figure 1A and B). Likewise, HFD feeding significantly reduced 10 kD dextran uptake by the nephrocytes (Figure 1C and D). In line with HFD-caused lipid droplet accumulation in the larval nephrocytes (Lubojemska et al., 2021), we observed increased nephrocyte lipid droplets in the adults fed with HFD (Figure 1—figure supplement 1). Thus, confirming the HFD model in Drosophila for studying kidney disease (Lubojemska et al., 2021) and demonstrating that, in our hands, the consumption of superfluous fat caused nephrocyte uptake dysfunction.

The slit diaphragm (SD) is the fundamental filtration structure of nephrocytes. Thus, given the HFD-induced dysfunctional uptake, we next tested whether the SD is affected by HFD. Therefore, we looked at the distribution of polychaetoid (Pyd), the fly homolog of human tight junction protein 1 (TJP1, alias ZO-1) and a key component of the SD filtration unit (van de Leemput et al., 2022). Immunostaining with anti-Pyd antibody showed predominant membrane localization of Pyd in nephrocytes from NFD-fed flies. At the cortical surface of NFD nephrocytes, Pyd showed the fingerprint-like pattern characteristic of the SD (Figure 2A). However, in the HFD-fed flies, the SD fingerprint-like pattern appeared irregular and Pyd showed an uneven distribution with spots of increased signal intensity (Figure 2A). These might correlate to the trails of Pyd seemingly hanging from the cortical surface inside the HFD nephrocytes, along with cytosolic accumulation of Pyd protein (Figure 2A and B). Next, we used Sns-mRuby3, in which mRuby3 was tagged at the C-terminal of endogenous Sns (Delaney et al., 2024; Zhao et al., 2025), to verify the HFD effect on SD structure. We observed cytoplasm retention of Sns-mRuby3 in the nephrocytes of HFD-fed flies (Figure 2—figure supplement 1). These findings show that a HFD leads to structural disruption of the SD filtration unit in nephrocytes, that likely accounts for the nephrocyte functional deficit observed earlier.

To study the nephrocyte SD and cytoplasmic regions in more detail, we performed transmission electron microscopy (TEM). The nephrocytes of NFD-fed flies showed regularly arranged lacuna channels along the cortical surface (Figure 2C). The distance between the lacuna channels was significantly increased in nephrocytes from HFD-fed flies (Figure 2C and D). Nephrocytes from both NFD and HFD fed flies showed big subcellular vacuoles (Figure 2E). However, the vacuoles in NFD nephrocytes frequently contained electron-dense structures, whereas most vacuoles in HFD nephrocytes were clear (Figure 2E and F). Altogether, these findings demonstrate that HFD causes significant changes to the nephrocyte and its SD, both at the cortical and subcortical levels.

Activity of the Janus kinase/signal transducer and activator of transcription (JAK-STAT) pathway has been linked to diabetic kidney disease (Berthier et al., 2009); likewise, it is activated systemically in Drosophila in response to a chronic lipid-rich diet (Woodcock et al., 2015). Key components of the JAK-STAT pathway are conserved in flies, including Signal-transducer and activator of transcription protein at 92E (Stat92E), hopscotch (hop), domeless (dome), and Suppressor of cytokine signaling at 36E (Socs36E) (Figure 3A and B). Therefore, we tested whether the JAK-STAT pathway was activated in the nephrocytes of HFD-fed flies. We used a fluorescent reporter 10xStat92E-GFP of JAK-STAT pathway activity (Ekas et al., 2006). Low levels of Stat92E-GFP fluorescence were detected in the nephrocytes of flies on a regular diet (NFD). However, in nephrocytes from HFD-fed flies, the fluorescence was significantly increased (Figure 3C and D). These findings indicate that the consumption of superfluous fat indeed activates the JAK-STAT pathway in nephrocytes.

Since JAK-STAT signaling has been shown to play an important role during development (Brown et al., 2001; Liu et al., 2009), we wanted to isolate its requirement for mature nephrocyte filtration function. To do this, we used a dominant gain-of-function allele of hop (Tumorous-lethal, hop.Tum); a single amino acid change (Gly341Glu) whose expression leads to JAK-STAT pathway activation (Harrison et al., 1995). We expressed UAS-hop.Tum specifically in mature nephrocytes using a temperature-sensitive Dot-Gal4 driver (Dot-Gal4; tub-Gal80ts, referred to as Dot-Gal4ts), known as the TARGET system (McGuire et al., 2004), to turn on expression in adult flies specifically as they are switched to a different environmental temperature (Figure 4A). Overexpression of hop.Tum significantly reduced nephrocyte absorption ability in the adults, as shown with reduced FITC-albumin and 10 kD dextran in Dot-Gal4ts >UAS hop.Tum nephrocytes compared with control nephrocytes (Figure 4B–E).

To validate these results and remove between-fly variability, both biological and technical, we also tested the effect of JAK-STAT pathway activation in tissue mosaic clones using Flp-out (Delaney et al., 2024; Duan et al., 2020; Figure 4F). For this technique, first instar larvae are exposed to heat shock to activate the Hsp70P transcription factor, which induces Flippase (Flp). Flp activity, in turn, removes the stop cassette, thereby driving Gal4 expression. Gal4 drives the expression of UAS-GFP and UAS-hop.tum within the same cell. Thus, hop.Tum is only expressed in GFP-labelled nephrocyte clones; whereas, GFP-negative nephrocytes are indicative of no Gal4 being expressed and serve as internal control cells (Figure 4F). In line with the TARGET assays, UAS-hop.Tum overexpression significantly reduced nephrocyte absorption ability in adult flies (Figure 4G and H). These data support a role for the JAK-STAT pathway in nephrocyte function.

Next, we looked at additional JAK-STAT pathway components in the context of nephrocyte function. Suppressor of cytokine signaling at 36E (Socs36E) is transcriptionally regulated by Signal-transducer and activator of transcription 92E (Stat92E) and functions as a negative regulator of JAK-STAT signaling (Callus and Mathey-Prevot, 2002; Karsten et al., 2002; Figure 3B), and it is expressed in the nephrocytes (Figure 3A). Silencing Socs36E specifically in nephrocytes (Dot-Gal4 >Socs36E-RNAi) significantly reduced nephrocyte uptake function as evident in significantly decreased FITC-albumin (Figure 5A and B) and 10 kD dextran (Figure 5C and D) fluorescence compared to the control nephrocytes.

In Drosophila, the JAK-STAT pathway ligand unpaired (Upd) family is encoded by upd1, upd2, and upd3. These ligands bind to Domeless (Dome), a single-transmembrane receptor, to activate JAK-STAT signaling (Figure 3B). Of these, Upd2 is functionally equivalent to human Leptin and is highly expressed in the fat body, the fly’s functional equivalent of vertebrate adipose tissue (Hombría et al., 2005; Rajan et al., 2017; Rajan and Perrimon, 2012). In Drosophila, HFD has been shown to upregulate Upd2 expression and secretion (Rajan et al., 2017; Rajan and Perrimon, 2012). In our model, overexpression of Upd2, specifically in the fat body (ppl-Gal4 driver; ppl-Gal4>upd2 GFP), significantly reduced 10 kD dextran uptake in the nephrocytes compared to the control (ppl-Gal4/+) (Figure 5E and F), indicating reduced nephrocyte function. Since our upd2 over-expression line contains upd2-GFP, we could not use the FITC-albumin uptake assay, as both signals occupy the same imaging channel. Upd2-GFP signal was found in the nephrocytes in ppl-Gal4>upd2 GFP flies (Figure 5—figure supplement 1A), indicating an inter-tissue communication of Upd2-GFP. Notably, Upd2-GFP overexpression in the fat body caused cytoplasm retention of Sns-mRuby3 (Figure 5G and H) and compromised the cortical fingerprint pattern (Figure 5—figure supplement 1B).

Like the Janus kinase Hop, the negative regulator Socs36E and the ligand adipokine Upd2 are additional JAK-STAT pathway components that are required for nephrocyte function.

To determine that JAK-STAT forms a direct link between HFD and nephrocyte dysfunction, we knocked down Stat92E (Stat92E-IR; Recasens-Alvarez et al., 2017) in the nephrocytes to disrupt the JAK-STAT signaling pathway (Figure 3B). Deficiency for Stat92E did not affect the uptake function (FITC-albumin or 10 kD dextran) of adult nephrocytes in flies fed a NFD (Figure 6A–D). However, under HFD conditions, in which nephrocytes from control flies showed significantly reduced uptake, Stat92E-deficient nephrocytes showed FITC-albumin and 10 kD dextran fluorescence restored to NFD levels (Figure 6A–D). We used an independent Stat92E-RNAi line in the functional assays and obtained similar results, showing that dextran fluorescence was restored to NFD levels in Stat92E-deficient nephrocytes under HFD conditions (Figure 6—figure supplement 1). Interestingly, under HFD conditions, the cytoplasm retention of Sns-mRuby3 was restored to NFD levels in Stat92E-deficient nephrocytes (Figure 6—figure supplement 2). These findings demonstrate that Stat92E, and by extension the JAK-STAT pathway, is required for the nephrocyte dysfunction induced by HFD.

Figure 6—figure supplement 2.

(A) Representative confocal images of nephrocytes from female adults that were fed on a regular diet (normal fat diet, NFD) or high-fat diet (NFD supplemented with 14% coconut oil, HFD) for 7 days. Genotype: Control (Dot-Gal4-Gal4, sns-mRuby3/+); Stat92E depletion (Dot-Gal4-Gal4, sns-mRuby3/+; UAS-Stat92E-IR_#2). Sns-mRuby3 is shown in red. Scale bar: 5 µm. (B) Box plot shows the quantitation of Sns-mRuby3 protein distribution (cytoplasmic vs membrane) based on images in (A); middle line depicts the median and whiskers show min to max. Statistical analysis was performed with a two-way ANOVA corrected with Tukey; ****p<0.0001; ns, not significant; n=10 flies.

Methotrexate suppresses STAT activation; it inhibits the phosphorylation of JAK which is necessary for JAK-STAT pathway activation and was shown to do so without affecting other phosphorylation-dependent pathways (Thomas et al., 2015). In our model system, methotrexate treatment (10 µM; 60 min) reduced levels of Stat92E (10xStat92E-GFP) in the nephrocytes (Figure 7—figure supplement 1). Stat92E is a key component of the pathway; its reduction is indicative of decreased JAK-STAT signaling. Next, we treated flies on NFD or HFD with 10 μM methotrexate (60 min incubation; ex vivo) to study the effect of pharmacological JAK-STAT inhibition on nephrocyte function. In nephrocytes from control flies on a regular diet (NFD), the methotrexate treatment had no effect on uptake of FITC-albumin or 10 kD dextran (Figure 7A–D). However, in flies on a HFD, treatment with methotrexate led to significantly increased nephrocyte FITC-albumin and 10 kD dextran uptake, restoring levels to those observed in control fly NFD nephrocytes (Figure 7A–D). We asked whether methotrexate treatment has an effect on Sns-mRuby3 subcellular distribution. In line with our prediction, under HFD conditions, pharmacological JAK-STAT inhibition restored Sns-mRuby3 distribution to NFD levels (Figure 7—figure supplement 2). Thus, like the genetic intervention (Stat92E inhibition; Figure 6), pharmacological intervention with the JAK-STAT inhibitor methotrexate restored nephrocyte function caused by HFD. These findings support the notion that HFD-induced nephrocyte dysfunction is mediated by the JAK-STAT signaling pathway.

Figure 7—figure supplement 2.

(A) Representative confocal images of nephrocytes from control Drosophila (sns-mRuby3; 7-day-old females) fed a regular diet (normal fat diet, NFD) or high-fat diet (NFD supplemented with 14% coconut oil, HFD), with or without methotrexate (10 μM; ex vivo 60 min) treatment. Sns-mRuby3 is in red. Scale bar: 5 μm. (B) Box plot shows the quantitation of Sns-mRuby3 protein distribution (cytoplasmic vs membrane) based on images in (A); middle line depicts the median and whiskers show minimum to maximum. Statistical analysis was performed by two-way ANOVA with Sidak correction; ****p<0.0001; ns, not significant; n=12 flies (7-day-old females).

Fly lines were reared on a normal fat diet (NFD; Nutri-Fly German formula; Genesee Scientific, San Diego, CA) or a high-fat diet (HFD; NFD supplemented with 14% coconut oil), under standard conditions (25°C, 60% humidity, 12 hr:12 hr dark:light cycle), unless otherwise stated.

Drosophila stocks w¹¹¹⁸ (BDSC_3605), Dot-Gal4 (BDSC_67608), ppl-Gal4 (BDSC_58768), tub-Gal80ts (BDSC_7017), 10XStat92E-GFP (BDSC_26198), UAS-Stat92E-RNAi_#2 (BDSC_33637), and UAS-Socs36E-RNAi (BDSC_35036) were obtained from the Bloomington Drosophila Stock Center (BDSC). UAS-Stat92E-RNAi (VDRC_106980) was obtained from the Vienna Drosophila Resource Center (VDRC). UAS-hop.Tum was kindly provided by Prof. Norbert Perrimon (Harvard Medical School, Boston, MA; Howard Hughes Medical Institute, Boston, MA). UAS-upd2:GFP has been previously described (Hombría et al., 2005) and was kindly provided by Prof. Ylva Engström (Stockholm University, Stockholm, Sweden). The Flp-out line hs-Flp¹²²; UAS-Flp^JD1/CyO, Act-GFP^JMR1; Act5C>CD2>Gal4^S, UAS-mCD8-GFP^LL6/TM6b was generated previously (Zhao et al., 2015).

Nephrocyte functional assays were performed ex vivo at room temperature, following a previously described method (Wen et al., 2020) with minor changes. Drosophila females were dissected in Schneider’s Drosophila Medium (Thermo Fisher Scientific, MA), then incubated in a 10 kD Texas Red-dextran solution (0.05 mg/mL; D1828, Thermo Fisher Scientific, MA) in Schneider’s Drosophila Medium (Thermo Fisher Scientific, MA) for 20 min, or a FITC-albumin solution (10 mM; Sigma, A9771) in Schneider’s Drosophila Medium (Thermo Fisher Scientific, MA) for 5 min. The specimens were washed in artificial hemolymph twice, followed by fixation in 4% paraformaldehyde (PFA) for 60 min. Then the fixed specimens were washed thrice for 5 min in 1x phosphate buffered saline (1xPBS; pH 7.4) and mounted using Vectashield mounting medium (H-1000, Vector Laboratories, CA). FITC-albumin and 10 kD dextran specimens were imaged using a ZEISS LSM900 confocal microscope with ZEISS Zen acquisition software (blue edition; version 3.0) using a 20× Plan-Apochromat 0.8 N.A. air objective (ZEISS, Oberkochen, Germany). For quantitative comparison of fluorescence intensities, settings for the control condition were chosen to avoid oversaturation (using Range Indicator in ZEN blue; limiting the observed red dots to avoid oversaturation), then applied across the images for all samples/conditions within the assay. The fluorescence intensity of FITC-albumin and 10 kD dextran nephrocytes was determined using Fiji software (Image J; Schneider et al., 2012, version 2.9.0/1.53t; National Institutes of Health, Bethesda). For quantitation, for each genotype, the relative fluorescence intensity of 30 nephrocytes from 6 female flies (five nephrocytes/fly) was analyzed.

Immunostaining was performed as previously reported (Zhao et al., 2022) with minor changes. Flies were briefly rinsed in 95% ethanol and dissected in 1xPBS at room temperature. Specimens were incubated in primary antibodies overnight at 4°C. The incubations in secondary antibodies were performed either overnight at 4°C or for 2 hr at room temperature. The following antibodies were used: chicken anti-GFP (1:1,000; ab13970, RRID:AB_300798, Abcam, Cambridge, UK), mouse monoclonal anti-Pyd (1:100; RRID:AB_2618043, Developmental Studies Hybridoma Bank, IA), goat anti-mouse Alexa Fluor 488 (1:500; A11029, RRID:AB_2534088, Invitrogen, Eugene, OR), and goat anti-chicken Alexa Fluor 488 (1:500; A11039, AB_2534096, Invitrogen, Eugene, OR). DAPI (0.5 mg/ml in PBST (0.2% Triton X-100 in 1x PBS); D1306, Thermo Fisher Scientific, MA) was used to visualize the nuclei. The nephrocytes were imaged using a ZEISS LSM900 confocal microscope (under Airyscan mode for Pyd images) with ZEISS ZEN acquisition software (blue edition; version 3.0) and a 63x Plan-Apochromat 1.4 N.A. oil objective (ZEISS, Oberkochen, Germany). For quantitative comparison of fluorescence intensities, settings for the control condition were chosen to avoid oversaturation (using Range Indicator in ZEN blue; limiting the observed red dots to avoid oversaturation), then applied across the images for all samples/conditions within the assay. Image J (Schneider et al., 2012) was used for image processing (version 2.9.0/1.53t; National Institutes of Health, Bethesda, MD). Typically, six flies were imaged per condition, and representative images for each are displayed in the figures.

TEM was performed using standardized procedures. In brief, female adults (7 days old) were dissected in Schneider’s Drosophila Medium (Thermo Fisher Scientific, MA). The heart tube and the attached nephrocytes were dissected, removed, and fixed using Sorensen phosphate buffer (2% PFA, 2.5% EM grade glutaraldehyde, 2 mM CaCl₂, 0.1 M NaOH; provided by the Electron Microscopy Core Imaging Facility at the Center for Innovative Biomedical Resources (CIBR), University of Maryland School of Medicine, MD). The processed specimens were imaged using a Philips CM100 TEM, carried out at the Electron Microscopy Core Imaging Facility at the Center for Innovative Biomedical Resources (CIBR) (University of Maryland School of Medicine, MD). The LC-LC distances were measured using the Straight Line tool in Fiji (Image J Schneider et al., 2012, version 2.9.0/1.53t; National Institutes of Health, Bethesda, MD) to connect two adjacent lacuna channels (LCs); a straight line was drawn from the middle of the first LC to the middle of the sixth LC, covering five LC-LC intervals on a TEM image, then the measure function was used to obtain the distance value. In total, 60 LC-LC distances were measured per condition in 10 nephrocytes from six 7-day-old female flies. The presence of a dark electron-dense structure in the vacuoles was manually determined and counted in images obtained from 12 nephrocytes for NFD and 29 nephrocytes from HFD from six 7-day-old female flies.

Flp-out clone (Struhl and Basler, 1993) induction was performed as previously described (Duan et al., 2020). In brief, female virgins of hs-Flp¹²²; UAS-Flp^JD1/CyO, Act-GFP^JMR1; Act5C>CD2>Gal4^S, UAS-mCD8-GFP^LL6/TM6b (Zhao et al., 2015) were crossed with UAS-hop.Tum males. The embryos were collected for 24 hr. First instar larvae (24 hr after embryo collection) received a 10 min heat shock in a 37°C water bath to induce the mosaic clones. After the heat shock, the larvae were maintained at 25°C. One-day-old female adults were subjected to the 10 kD dextran functional assay (described above). The GFP-positive nephrocyte clones and their neighboring nephrocytes were analyzed.

Methotrexate (06563, Sigma-Aldrich, MO) was dissolved in DMSO to make a 10 mM stock solution. Flies were dissected as described for the nephrocyte functional assays. The dorsal cuticle (with nephrocytes) was transferred to methotrexate solution (10 μM) in Schneider’s Drosophila Medium (Thermo Fisher Scientific, MA) and incubated at room temperature for 60 min. The samples incubated in Schneider’s Medium supplemented with DMSO vehicle were used as a control. The specimens were rinsed with Schneider’s Drosophila Medium (Thermo Fisher Scientific, MA), then subjected to a functional assay or fixed in 4% PFA for immunochemistry.

Newly eclosed w¹¹¹⁸ flies were fed either a high-fat diet (HFD) or a control diet (CD) for 7 days before staining for lipid droplets using Nile Red. For Nile Red staining, nephrocytes were dissected in 1× PBS and fixed in 4% paraformaldehyde for 1 hr at room temperature. The samples were then washed three times with 1× PBS for 5 min each, followed by incubation with 0.1 μg/mL Nile Red (HY-D0718, MedChemExpress) for 10 min at room temperature. Afterward, the samples were rinsed three more times with 1x PBS for 5 min each.

Image J (Schneider et al., 2012) (version .9.0/1.53t; National Institutes of Health, Bethesda, MD) was used to process the raw data of confocal images and to measure the relative fluorescence intensity. The data sets were tested for normality using the Shapiro-Wilk test and plotted using GraphPad Prism 9 software (version 9.5.1). Normally distributed data were analyzed by the two-tailed Student’s t-test, or by two-way ANOVA with Sidak correction. P<0.05 was considered significant. The figures were arranged using Adobe Illustrator software (version 2022 26.2.1). Box plots show the median (center line), interquartile range (25th–75th percentiles; box), and whiskers extending to the minimum and maximum values.