Podocyte Geranylgeranyl Transferase Type-I Is Essential for Maintenance of the Glomerular Filtration Barrier

A tightly regulated actin cytoskeleton attained through balanced activity of RhoGTPases is crucial to maintaining podocyte function. However, how RhoGTPases are regulated by geranylgeranylation, a post-translational modification, has been unexplored. The authors found that loss of the geranylgeranylation enzyme geranylgeranyl transferase type-I (GGTase-I) in podocytes led to progressive albuminuria and foot process effacement in podocyte-specific GGTase-I knockout mice. In cultured podocytes, the absence of geranylgeranylation resulted in altered activity of its downstream substrates Rac1, RhoA, Cdc42, and Rap1, leading to alterations of β1-integrins and actin cytoskeleton structural changes. These findings highlight the importance of geranylgeranylation in the dynamic management of RhoGTPases and Rap1 to control podocyte function, providing new knowledge about podocyte biology and glomerular filtration barrier function.

Abstract

Significance Statement

A tightly regulated actin cytoskeleton attained through balanced activity of RhoGTPases is crucial to maintaining podocyte function. However, how RhoGTPases are regulated by geranylgeranylation, a post-translational modification, has been unexplored. The authors found that loss of the geranylgeranylation enzyme geranylgeranyl transferase type-I (GGTase-I) in podocytes led to progressive albuminuria and foot process effacement in podocyte-specific GGTase-I knockout mice. In cultured podocytes, the absence of geranylgeranylation resulted in altered activity of its downstream substrates Rac1, RhoA, Cdc42, and Rap1, leading to alterations of β1-integrins and actin cytoskeleton structural changes. These findings highlight the importance of geranylgeranylation in the dynamic management of RhoGTPases and Rap1 to control podocyte function, providing new knowledge about podocyte biology and glomerular filtration barrier function.

Background

Impairment of the glomerular filtration barrier is in part attributed to podocyte foot process effacement (FPE), entailing disruption of the actin cytoskeleton and the slit diaphragm. Maintenance of the actin cytoskeleton, which contains a complex signaling network through its connections to slit diaphragm and focal adhesion proteins, is thus considered crucial to preserving podocyte structure and function. A dynamic yet tightly regulated cytoskeleton is attained through balanced activity of RhoGTPases. Most RhoGTPases are post-translationally modified by the enzyme geranylgeranyl transferase type-I (GGTase-I). Although geranylgeranylation has been shown to regulate activities of RhoGTPases and RasGTPase Rap1, its significance in podocytes is unknown.

Methods

We used immunofluorescence to localize GGTase-I, which was expressed mainly by podocytes in the glomeruli. To define geranylgeranylation's role in podocytes, we generated podocyte-specific GGTase-I knockout mice. We used transmission electron microscopy to evaluate FPE and measurements of urinary albumin excretion to analyze filtration barrier function. Geranylgeranylation's effects on RhoGTPases and Rap1 function were studied in vitro by knockdown or inhibition of GGTase-I. We used immunocytochemistry to study structural modifications of the actin cytoskeleton and β1 integrins.

Results

Depletion of GGTase-I in podocytes in vivo resulted in FPE and concomitant early-onset progressive albuminuria. A reduction of GGTase-I activity in cultured podocytes disrupted RhoGTPase balance by markedly increasing activity of RhoA, Rac1, and Cdc42 together with Rap1, resulting in dysregulation of the actin cytoskeleton and altered distribution of β1 integrins.

Conclusions

These findings indicate that geranylgeranylation is of crucial importance for the maintenance of the delicate equilibrium of RhoGTPases and Rap1 in podocytes and consequently for the maintenance of glomerular integrity and function.

Introduction

Podocytes are known for their complex structures with foot processes forming an intertwined pattern with slit diaphragms spanning the space between adjoining foot processes constituting the final part of the permselective glomerular barrier.¹ The actin cytoskeleton is crucial for maintaining these structures and contains a complex signaling network through its connections to slit diaphragm and focal adhesion proteins.²

Impairment of the filtration barrier function with loss of proteins into the urine is in part ascribed to podocyte foot process effacement (FPE), a pathological process that entails disruption of the actin cytoskeleton and the slit diaphragms resulting in shortened and flattened foot processes.³ Maintenance and regulation of the actin cytoskeleton is therefore deemed crucial for normal podocyte morphology and function. Crucial actin regulators, RhoGTPases, RhoA, Rac1, and Cdc42, are known to promote actin cytoskeleton reformation in distinct ways: RhoA through induction of stress fibers and focal adhesions, Rac1 through formation of lamellipodia, and Cdc42 through formation of filopodia.⁴ Various studies have shown how an unbalanced activity of RhoGTPases can cause nephrotic syndrome in mice^5–9 and that maintaining their balance is crucial to preserve glomerular barrier permselectivity.^10,11 Identification of loss-of-function mutations in RhoGTPase-controlling proteins has further supported the role of RhoGTPases in glomerular disease and nephrotic syndrome.^12,13

RhoGTPases are involved in many intracellular processes such as cell migration, progression of cell-cycle and cytoskeletal dynamics.¹⁴ Guanine nucleotide exchange factors (GEFs) control the activation of RhoGTPases, while GTPase-activating proteins (GAPs) regulate the inactivation of RhoGTPases. A third group of proteins, RhoGDIs (Rho GDP dissociation inhibitors), control the activity of RhoGTPases by inhibiting GTP hydrolysis and nucleotide exchange and by repositioning them away from their subcellular locations.^15,16

The lipid modification called prenylation (a covalent attachment of isoprene units) is also believed to regulate the subcellular localization and activity of small GTPases.^17–20 Prenylation is performed by four heterodimeric enzymes: farnesyltransferase (FTase), geranylgeranyl transferase type-I (GGTase-I), Rab geranylgeranyl transferase (type-II, targeting RabGTPases, important for vesicular trafficking), and geranylgeranyl transferase type-III (targeting ubiquitin ligase FBXL2).^21,22 FTase transfers 3 isoprene units (farnesyl), while GGTases transfer 4 units (geranylgeranyl) to the cysteine in a carboxyl terminus CaaX sequence.²³ FTase and GGTase-I share one subunit (Fnta) but differ in the second subunit (respectively, Fntb or Pggt1b). RhoGTPases RhoA, Rac1, and Cdc42 and the RasGTPase Rap1 are typical substrates of GGTase-I, while FTase targets lamins, RasGTPases, and other RhoGTPases.

Together with RhoGTPases, Rap1 has been shown to have a role in regulating podocyte actin cytoskeleton through participating in MAGI2 signaling and cellular adhesion (integrin activation and clustering).^24,25 Moreover, deactivation of Rap1 using modulation of GEFs and GAPs in podocytes has been shown to cause glomerulosclerosis.^24,26 Conversely, it is unknown how a chronically active Rap1 would affect the podocytes and the filtration barrier.

Strikingly, although geranylgeranylation is considered essential for their activity, the regulation of RhoGTPase and Rap1 activity in podocytes using this lipid modification has remained unexplored.²⁰ To address this lack of knowledge, we have sought to provide new insights into how geranylgeranylation of RhoGTPases and Rap1 affect podocyte function. We found that GGTase-I is crucial for maintaining podocyte morphology and filtration barrier function.

Materials and Methods

Reagents

The following reagents were used: GGTI-298 trifluoroacetate salt (Cayman Chemicals, Ann Arbor, MI); Simply Blue Safe stain (Thermo Fisher, Waltham, MA); Mouse Albumin-ELISA E99-134 (Bethyl Laboratories, Montgomery, TX); Creatinine Assay MAK080 (Sigma Aldrich, Stockholm, Sweden); FuGene 6 Transfection reagent (Promega, Madison, WI); active RhoGTP Pull-Down and detection kit; Rac1/Cdc42 #16118, RhoA #16116, Rap1 #16120 (Pierce, Thermo Scientific, Rockford, IL); and Qproteome Cell Compartment Kit #37502 (Qiagen, Hilden, Germany). The following antibodies were used: Rac1 (BD Biosciences, San Jose, CA), Cdc42 (Cell Signaling Technologies, Danvers, MA), RhoA (Cell Signaling Technologies), Pggt1b (Atlas Antibodies, Stockholm, Sweden), np-Rap1 (Santa Cruz Biotech, Santa Cruz, CA), Alexa488/594 Phalloidin (Invitrogen, Thermo Fisher Scientific, Waltham, MA), total β1 integrin (Merck Millipore, Burlington, MA), active β1 integrin (BD Biosciences, San Jose, CA), Synaptopodin conjugated to CoraLite 594 (Proteintech Rosemont, IL), Wilms tumor 1 conjugated to Alexa 594 (Santa Cruz), MEK 1/2, Histone, Vimentin (all from Cell Signaling Technology), Na/K-ATPase (Abcam, Cambridge, UK), and HRP-conjugated antibodies W4021, W4011, V8051 (Promega, Madison, WI).

Human Material

The experiments conducted with human material were performed in accordance with the Declaration of Helsinki. Ethical approval for collecting human biopsies was given by the Gothenburg Regional Ethical Board, and an informed consent was signed by the patient before the collection of kidney tissue. The material was obtained from tumor-associated nephrectomies performed at Sahlgrenska University Hospital. A portion was taken from the nonaffected cortex of the kidney and cryopreserved in optimal cutting temperature compound.

Murine Models

The study was performed after the approval from the Regional Laboratory Animal Ethics Committee (#67-2016). To obtain podocyte-specific depletion of geranylgeranyl transferase type-I, mice with the Pod-Cre transgene were used. In these mice, the expression of Cre-recombinase is controlled by the podocin gene, nphs2, specifically expressed in podocytes.²⁷ Cre-recombinase will cut flanking lox p sequences surrounding the gene of interest, leaving the animal floxed for that gene. For the geranylgeranyl transferase depletion experiment, male B6.Cg-Tg(nphs2-cre)295Lbh/J mice (Jackson laboratory, Bar Harbor, ME) were bred with mixed background female B6.CCgTm(Pggt1b^fl/fl)295Lbh/J mice.²⁸ The breeding gave rise to B6.C-CgTm(Pggt1b^fl/fl)295Lbh/J Tg(nhphs2-cre) mice (referred to as Pggt1b^fl/fl), as well as the littermate mice Pggt1b^fl/+ and Pggt1b^+/+. For farnesyltransferase experiments, the male B6.Cg-Tg(nphs2-cre)295Lbh/J mice (Jackson laboratory, Bar Harbor, ME) were bred with mixed background female B6.C-CgTm(Fntb^fl/fl)295Lbh/J mice, to obtain B6.C-CgTM(Fntb^fl/fl)295Lbh/J Tg(nhphs2-cre) mice, referred to as Fntb^fl/fl mice. Genotyping of mice was performed through extraction of DNA from clipped ear fragments followed by PCR and agarose gel analysis. The following primers were used for genotyping: Pod-Cre FW 5′-cacagctccaccaagacaca-3′, RV 5′-aggcaaattttggtgtacgg-3′, resulting in a band at 160 bp, Pggt1b FW 5′-cctgaatgcagatctgtgga-3′, RV 5′-cctatgaaagcagcacgaca-3′, resulting in a band at 370 bp for floxed genes and 260 bp for wild-type genes. For genotyping of Fntb floxed mice, the following primers were used: FW 5′-ggtggatggggaaattggg-3′ RV 5′-agcagccacctggagactta-3′, resulting in a band at 360 bp for the floxed gene and 270 for the wild-type gene. Il-2 was used as control, resulting in a band at 330 bp.

Metabolic Cages

After genotype assessment, the mice were singly housed in metabolic cages at 21–23°C and 55% humidity in Scantainer and Scanclime units (Scanbur A/S, Karlslunde, Denmark). Mice were kept in the metabolic cages for 24 hour and thereafter returned to their home cage. The 24-hour urine was collected, aliquoted and stored at −80°C until analysis was performed.

Urine Analysis

A first assessment of proteinuria was performed with gel electrophoresis of urine samples to screen for proteinuria. See Supplemental Material for method description.

Urine Albumin/Creatinine Ratio

For measurement of urine albumin content, an enzyme-linked immunosorbent assay (ELISA, Bethyl Laboratories) was used according to the manufacturer's protocol. Urinary creatinine was assessed with a colorimetric assay kit (Sigma Aldrich). The assay was performed according to the manufacturer's protocol. The urine albumin/creatinine ratio (ACR), was calculated as μg albumin/mg creatinine.

Tissue Harvest

Kidneys were collected from three mice per age group and genotyped (n=12). The mice were anesthetized by administrating 3% isoflurane, 1.5 l/min air, through a mask. A midline incision of the anterior abdominal wall and peritoneum was performed, and the kidneys were visualized. After removal of the kidneys, the renal capsules were removed. One of the kidneys was cut longitudinally, and the two pieces were submerged in Karnovsky buffer (2% paraformaldehyde, 2.5% glutaraldehyde, 0.01% sodium azide in 0.05 M sodium cacodylate buffer, pH 7.2) for further preparation for TEM analysis. The other kidney was cut in half at the renal pelvis, one-half was embedded in optimal cutting temperature compound and cryopreserved for immunohistochemical analysis, and the other half was submerged in RNA later for protein and RNA analysis. See Supplemental Material for tissue and glomerular preparation, qPCR, ddPCR, and immunohistochemistry.

Preparation for Transmission Electron Microscopy and Analysis

Tissues stored in Karnovsky buffer were prepared for transmission electron microscopy (TEM). Slices containing glomeruli (as verified with 500 nm tissue sections) were further postfixed with 1% w/v OsO₄ 1 w/v K₄Fe(CN)₆ in 0.1 M cacodylate buffer followed by en bloc staining with uranyl acetate 0.5% w/v. Specimens were dehydrated and embedded in epoxy resin. Ultrathin sections (70 nm) were obtained with a Reichert-Jung Ultracut E microtome (Leica Microsystems, Vienna, Austria) equipped with a diamond knife. They were contrasted with uranyl acetate 1% w/v for 5 minutes and lead citrate for 3 minutes before examination. The analysis was made using the TalosL120C (Thermo Fisher Scientific, Waltham, MA), with 120 kV working voltage, equipped with a 4kx4k CMOS Ceta Camera and TIA imaging software.

Podocyte Foot Process Analysis

Podocyte foot process analysis was performed on micrographs from Pggt1b^+/+ and Pggt1b^fl/fl mice at age 7 months. Three mice per group were analyzed, and a minimum of 3800 foot processes were counted per group. The number of podocyte foot processes over a measured length of glomerular basement membrane (GBM) was counted, and the length of the GBM of the capillary was measured using Fiji/ImageJ (National Institutes of Health, Bethesda, MD). The number of foot processes was divided by the length of the GBM to obtain the number of foot processes per μm GBM.

Cell Culture

Conditionally immortalized murine podocytes were grown under permissive conditions as has been described in previous protocols.²⁹ In short, cells were grown in RPMI 1640 (Gibco) with 10% fetal bovine serum (GE Healthcare Hyclone, Logan, UT) and 1% antibiotics (100 U/ml penicillin, 100 U/ml streptomycin, Lonza, Basel, Switzerland). The cell medium was supplemented with 20 U/ml of interferon-γ (Peprotech Nordic, Stockholm Sweden), and cells were propagated at 33°C. The cells were trypsinized (Trypsin-EDTA, Lonza), replated, and transferred to a 37°C incubator. At 37°C, the cells were kept in growth media without interferon-γ and allowed to differentiate for 14 days. Fourteen days post-thermal switch, the cells were fully differentiated podocytes and were used for further experiments.

HEK293 cells were used for transient transfection. The HEK293 cells were cultured in DMEM 4 g/l glucose (Lonza) with 10% FBS and 1% antibiotics (100 U/ml penicillin, 100 U/ml streptomycin, Lonza). During transfection, the cells were kept in antibiotic-free medium.

Plasmid Constructs, Transient Transfection, and Lentiviral Gene Silencing

The Pggt1b shRNA pLKO.1 vector TRCN0000345832 from Sigma (Sigma Aldrich) was used for construction of Pggt1b knockdown lentiviruses, and the scramble shRNA pLKO.1 vector, #1864 (Addgene, Watertown, MA), was used as control. For production of lentiviruses, transient transfection of HEK293 cells was performed using FuGene6 transfection reagent (Promega) according to the manufacturer's recommendations. The cells were transfected with the Pggt1b shRNA or control vector together with the packaging vector pCMV dr8.91 and envelope vector pCMV-VSV-G in a 3:2:1 ratio. 16 hours post-transfection, the media was changed on the cells. Virus-containing media was harvested 48 h post-transfection. Podocytes were infected with the viruses on day 7–9 post-thermal switch, using polybrene, 4 μg/ml (Sigma Aldrich), and the cells were used for further experiments 7 days postinfection.

Treatment with GGTI-298

To inhibit the activity of the geranylgeranyl transferase type-I, cells were treated with GGTI-298 trifluoroacetate (Cayman Chemical) in a final concentration of 10 μM. The compound was dissolved in DMSO for long-term storage. For experiments, the compound was further diluted in sterile 0.9% saline, before adding it to the cell media. The cells underwent treatment for 5 h and were then used for further analysis.

Immunofluorescence

Immunofluorescence staining was performed on cultured cells that had either undergone treatment with GGTI-298 trifluoroacetate or been infected with the Pggt1b silencing lentivirus. The cells were fixated in 2% PFA and 4% sucrose in PBS for 10 minutes. Thereafter followed permeabilization using 0.3% Triton-X 100 in PBS for 5 minutes at 4°C. The cells were washed in PBS and then stained with Alexa Fluor-488/594 Phalloidin (Invitrogen) for visualization of the actin cytoskeleton.

For staining of cells and frozen renal tissue sections with primary and secondary antibodies, 30 minutes of blocking was performed using 2% FBS, 2% BSA, 0.2% fish gelatin in PBS. The primary antibody, diluted in blocking buffer, was added to the cells/sections and left to incubate for 1 hour at room temperature or overnight at 4°C. The cells were then washed in PBS, and the secondary antibody, either Alexa 594 or 488 (Invitrogen), diluted in blocking buffer, was added to the cells. After 1-hour incubation at room temperature, the cells or sections were washed in PBS and mounted using ProLong Diamond Antifade mounting media (Thermo Fisher). For further antibody specifications and dilutions, please refer to the reagent section. Images were obtained using the Zeiss Confocal microscope imager Z2, LSM800 with an optical slice of 1–5 μm at a resolution of 1200 pixels per inch.

The quantification of stress fibers and active/total β1 integrins after Pggt1b KD in vitro was made using Fiji/ImageJ. The total fluorescence was calculated as the sum of the gray pixels in the whole image divided by the total number of pixels (mean gray value) to obtain the corrected total fluorescence. Mean values were used for calculations. Normalization was used to correct for different staining intensities. The analysis was conducted using three different sets of stainings for active β1 integrin (images: n=88 for Pggt1b KD, 71 for virus control) and total β1 integrin (n=74, 58) and 5 different sets of images for the stress fibers (phalloidin) quantification (n=162, 128). Each image contained a minimum of 3 to a maximum of 15 cells.

Rho/Ras-GTPase Activity Analysis

Activity of the RhoGTPases RhoA, Rac1, Cdc42, and the Ras-family GTPase Rap1 was analyzed using commercially available active GTPase pull-down and detection kits (Pierce, Thermo Scientific). The assays were performed according to the manufacturer's recommendation. Cultured podocytes underwent treatment with GGTI-298 trifluoroacetate or had the gene for Pggt1b silenced through lentiviral infection. The cells were harvested in lysis buffer from the kit supplemented with complete protease inhibitor (Roche, Sigma Aldrich). Lysates were then used for pull-down of the active form of each GTPase using GST-tagged protein binding domains of downstream effectors for each GTPase. A spin column with a glutathione resin was used for affinity pulldown. The GST fusion protein was thereafter eluted from the glutathione resin in 4× sample buffer (Laemmli-buffer, Bio-Rad Laboratories, Sundbyberg, Sweden) with 10% reducing agent (NuPAGE Sample reducing agent, Thermo Fisher). The GTPases were detected through immunoblotting using antibodies for each protein. For antibody dilutions, please refer to the reagent section.

Cell Fractionation

Cell fractionation was performed using the commercially available Qproteome Cell compartment kit (Qiagen, Hilden, Germany) and performed according to the manufacturer's protocol. In brief, Pggt1b was silenced through lentiviral infection in cultured murine podocytes. Cells were thereafter harvested using trypsin, and the number of cells was calculated using a cell counter. Approximately 5 million cells were used per replicate. Cell fractionation was thereafter performed by the addition of extraction buffers and repeated centrifugation according to the protocol enclosed in the kit, followed by an acetone precipitation step. Protein precipitates were thereafter dissolved in sample buffer (Laemmli buffer, Bio-Rad Laboratories, Sundbyberg, Sweden) and analyzed through immunoblotting using antibodies for the different RhoGTPases as well as specific markers for cellular fractions. For antibody dilutions, please refer to the reagent section.

Electrophoresis and Western Blot

SDS-PAGE was performed as recommended by the manufacturer following the V3 Western Workflow³⁰ (Bio-Rad Laboratories, Sundbyberg, Sweden). Samples were mixed with 4× Laemmli buffer (Bio-Rad Laboratories) and 10× reducing agent (NuPAGE Sample reducing agent, Thermo Fisher) and thereafter boiled at 95°C for 5 minutes. Samples were run on a Mini-Protean TGX Stain free Gel of either 10% or 4%–15% (Bio-Rad Laboratories) at 100 V. The gels were activated for 45 seconds on the ChemiDoc Touch system (Bio-Rad Laboratories), and the stain free gels were evaluated.

Transfer was performed using the Trans-Blot Turbo Transfer system (Bio-Rad Laboratories) and the 3-minute program for Mini-gels. The proteins were transferred to low-fluorescence polyvinylidene difluoride membranes (Bio-Rad Laboratories). Thereafter, the membranes were activated to verify the transfer and to obtain a total protein blot for normalization using the ChemiDoc Touch system. The membranes were placed in 5% dry milk (Bio-Rad Laboratories), in TBS-T (20 mM tris-hydroxymethyl aminomethane, 150 mM NaCl, and 1% tween 20, pH 8.0) blocking buffer for 1 hour at room temperature. The primary antibodies were then diluted in 5% milk blocking buffer and left to incubate for 1 hour or overnight with the membranes. The secondary HRP-conjugated antibodies, either anti-mouse, anti-rabbit, or anti-goat (Promega), were diluted in 5% milk blocking buffer, and the membranes were left to incubate with the secondary antibody for 1 hour at room temperature.

For visualization of the bands, Clarity Western ECL solution (Bio-Rad Laboratories) or SuperSignal West Femto Maximum Sensitivity Substrate (Thermo Fisher, Waltham, MA) was used. The membranes were left to soak for 5 minutes in development solution, and chemiluminescence was detected using the ChemiDoc Touch system. Normalization of visualized bands was performed against total lane protein according to instructions from the manufacturer, using the ImageLab software (Bio-Rad Laboratories). For dilutions of antibodies, please refer to the reagents section.

Statistical Analysis

Statistical calculations were made using GraphPad Prism version 9.2.0. After testing for normal distribution, differences between two groups were analyzed using either t tests or Mann-Whitney tests. For comparison between multiple groups, either one-way ANOVA or Kruskal-Wallis nonparametric tests (both with multiple comparison test and with Benjamini-Krieger-Yekuteli FDR correction) were used. Values of P <0.05 were considered significant. Mean±SEM is plotted per each group, unless stated otherwise.

Results

Generation of a Podocyte-Specific GGTase-I Knockout Mouse Model

Previous studies have shown that prenylation has a role in regulating mesangial cell proliferation.^31,32 Given the importance of prenylation for small RhoGTPases, we investigated the effects of depletion of GGTase-I in podocytes. We therefore generated a podocyte-specific (Pod-Cre) geranylgeranyl transferase β subunit (Pggt1b) floxed knockout mouse (Figure 2, A and B).²⁸

Figure 2.

Expression of geranylgeranyl transferase type I in human glomeruli. Human glomeruli were costained with GGTase-I, Synaptopodin, or WT-1 to illustrate GGTase-I localization. GGTase-I has a broad expression in the glomerulus and tubules, as in mice as seen in Figure 1C. GGTase-I is present in podocytes: Staining with WT-1 shows a clear colocalization in the podocyte cell body, with a perinuclear pattern.

Figure 1.

Podocyte-specific depletion of geranylgeranyl transferase type-I. (A) Illustration of the generation of podocyte-specific geranylgeranyl transferase deficient mice. Expression of Cre-recombinase was targeted to murine podocytes using the podocin promotor NPHS2 (Pod-Cre). Pod-Cre mice were bred with mice carrying floxed Pggt1b alleles to obtain Pod-Cre Pggt1b^fl/fl mice. (B) Genotype determination through PCR analysis of genomic DNA from ear clippings. Il-2 was used as control, resulting in a band at 330 bp. Pod-Cre expression was demonstrated by a band at 160 bp. Expression of wild-type Pggt1b resulted in a band at 260 bp, while the floxed gene resulted in a band at 370 bp. Heterozygous animals show both bands. (C) GGTase-I is broadly expressed in Pggt1b^+/+ mice glomerular and tubular cells. GGTase-I shows a perinuclear localization in podocytes (see magnified detail of the merged GGTase-I + WT-1 confocal image) and in other glomerular cells (see detail+DAPI). In Pggt1b^fl/fl, GGTase-I signal disappears only from the podocytes (see GGTase-I + WT-1 merged and detail image), but it is still present in other glomerular cells and in the tubules (detail+DAPI). (D) The expression of Pggt1b mRNA was reduced in Pggt1b^fl/fl mice compared with Pggt1b^+/+ controls only in isolated glomeruli and kidney tissues. **P <0.01, unpaired t test, n=3 per group. (D) The depleted Pggt1b mRNA expression is also evident in podocyte-enriched clones of floxed versus wild-type Pod-Cre animals after ddPCR analysis. This confirms the podocyte-specific Pggt1b gene exon 7 deletion. *P <0.05, Mann-Whitney test, n=3 (Pggt1b^+/+), 5 (Pggt1b^fl/fl).

Immunofluorescence staining of mouse kidney biopsies were performed in Pggt1b^fl/fl and Pggt1b^+/+ controls to evaluate GGTase-I expression (Figure 2C). GGTase-I was broadly expressed not only in glomeruli but also in tubules. GGTase-I was localized to the podocytes with a perinuclear expression pattern (see merged GGTase-I and WT-I image and its magnified detail) but also in other glomerular cells (see “detail+DAPI image”).

Decreased GGTase-I protein expression was observed in the biopsies from floxed mice, when comparing it with Pggt1b^+/+ controls (Figure 2C, lower panels). The depletion was only affecting the podocytes because GGTase-I was still expressed by other glomerular cells, as illustrated in the magnified detail of GGTase-I and WT-I and the detail+DAPI images.

Decreased mRNA expression of Pggt1b was confirmed using qPCR in isolated glomeruli and whole kidneys from Pggt1b^+/+ and Pggt1b^fl/fl mice. No decrease was observed in heart and liver tissues (Figure 2D). Pggt1b mRNA expression in podocyte-enriched clones derived from glomerular preparations was also investigated with ddPCR and showed a significant decrease in Pggt1b^fl/fl when compared with Pggt1b^+/+ controls (Figure 2D).

GGTase-I is Expressed in Human Podocytes

For translational purposes, we evaluated the presence and localization of GGTase-I in the human glomerulus. Immunofluorescence staining was performed, revealing a broad expression of GGTase-I in both the glomerular compartment and the tubular compartment (Figure 2). At the podocyte level, GGTase-I was found in the cell body, with a perinuclear localization, similar to what was observed in mice.

Depletion of Podocyte GGTase-I Causes Progressive Albuminuria in Mice

Pggt1b^fl/fl mice and their littermate controls were housed in metabolic cages for 24-hour urine collection and assessment of filtration barrier function through analysis of urinary albumin excretion. The mice were followed for approximately 7 months, and collection of urine was performed at regular intervals. A first assessment of albuminuria was performed immediately after collection with gel electrophoresis and Coomassie blue staining. In Pggt1b^fl/fl mice, the presence of a band at the molecular weight of albumin (67 kDa) indicated albuminuria (Supplemental Figure 1A). The band was faint or absent in urines from Pggt1b^fl/+ and Pggt1b^+/+ control mice. This was repeatedly observed throughout the study. No sex-related differences were observed during the study (data not shown).

For a more precise assessment, urinary albumin and creatinine were measured to obtain an ACR. ACR analysis showed that Pggt1b^fl/fl mice had a tendency toward increased albuminuria already after 1 month, compared with littermate controls (467.02±233.79 μg/mg versus 30.02±5.51 μg/mg, Figure 3A). At age 7 months, Pggt1b^fl/fl mice had developed significant albuminuria in comparison with their age-matched controls (1400.04±404.33 μg/mg versus 81.99±22.09 μg/mg, P <0.01, Figure 3A). Four of the Pggt1b^fl/fl mice and four controls were tested at least twice between ages 1 and 7 months and show once again albuminuria increasing over time (Figure 3B). Finally, Pggt1b^fl/+ mice showed no increase in albuminuria in comparison with Pggt1b^+/+ mice (Supplemental Figure 1B).

Figure 3.

Podocyte geranylgeranyl transferase type-I deficient mice develop albuminuria. (A) ACR shows increased albuminuria in Pggt1b^fl/fl mice at age 1 month compared with age-matched, Pggt1b^+/+ control mice. The same pattern is present in the 4-month-old floxed versus control mice and, finally, at age 7 months, there is a significant albuminuria in Pggt1b^fl/fl mice compared with control, assessed through the ACR. *P <0.05, **P <0.01, Kruskal Wallis test, n=13, 13, 7, 7, 13, and 13 mice per group, respectively. (B) ACR of Pggt1b^fl/fl and Pggt1b^+/+ tested 2 or 3 times between ages 1 and 7 months, showing progressive incremental albuminuria in the floxed animals and no change in the controls (n=4 mice per group).

Development of Albuminuria Occurs When Geranylgeranlyation but Not Farnesylation is Knocked down

Prenylation of small GTPases can be performed by either GGTase-I or FTase. The classic members of the RhoGTPase family (RhoA, Rac1, Cdc42, and RasGTPase Rap1) are all targeted by the GGTase-I.^23,33 Other RhoGTPases and RasGTPases are targeted by FTase.²³ To investigate whether podocyte-specific depletion of FTase had an effect on glomerular filtration barrier function, we generated podocyte-specific farnesyl depleted mice with the same Pod-Cre flox system used for Pggt1b (Supplemental Figure 2).

At approximately 4 months of age, the ACR analysis showed no significant albuminuria in the Fntb^fl/fl mice (96.7±16.04 μg/mg), in comparison with age-matched Pggt1b^+/+ mice (92.59±23.23 μg/mg). Pggt1b^fl/fl mice did however show a significant increase (Pggt1b^fl/fl 516.5±153.4 μg/mg) in albuminuria in comparison with both Fntb^fl/fl and Pggt1b^+/+ mice indicating that geranylgeranylation in podocytes has a major effect on glomerular barrier function while podocyte farnesylation does not (Figure 4).

Figure 4.

Albuminuria is specific to the depletion of geranylgeranyl transferase type-I. Depletion of Fntb in mice did not cause albuminuria, as compared with Pggt1b^+/+ and Pggt1b^fl/fl mice of the same age (as in Figure 3). Analysis was performed using albumin/creatinine ratio in mice of age 4 months. **P <0.01, one-way ANOVA, n=7, 7, and 8 mice per group, respectively.

Loss of GGTase-I in Podocytes Promotes Foot Process Effacement

Albuminuria is typically caused by pathological alterations in the glomerular filtration barrier.¹ Therefore, we assessed potential glomerular morphological changes underlying its development in Pggt1b^fl/fl mice with TEM. As seen in Figure 5A, the characteristic pattern of podocyte FPE could be observed in Pggt1b^fl/fl mice already at age 1 month. No effacement was seen in the Pggt1b^+/+ mice of the same age. With increasing age, a segmental FPE became more evident in the Pggt1b^fl/fl mice while the filtration barrier morphology of control mice remained unchanged (Figure 5B). To assess the morphological changes in the filtration barrier, the number of foot processes per µm of glomerular basement membrane was quantified. In 7-month-old Pggt1b^fl/fl mice, the number of foot processes per µm of glomerular basement membrane was significantly decreased in comparison with littermate controls (Pggt1b^fl/fl 1.926±0.037 versus Pggt1b^+/+ 2.566±0.040 foot processes/μm, P<0.001, Figure 5C). Light microscopy analysis of cryopreserved tissue sections revealed no macroscopical alterations in the glomerular or tubular morphology of Pggt1b^fl/fl mice in comparison with control mice (Supplemental Figure 3).

Figure 5.

Podocyte FPE in geranylgeranyl transferase type-I depleted mice. (A) Transmission electron microscopy was performed to assess morphological changes in the filtration barrier. FPE was observed in Pggt1b^fl/fl mice already at age 1 month. No effacement was seen in the control mice, Pggt1b^+/+, of the same age. (B) The FPE increased with time, as can be seen in the Pggt1b^fl/fl mice at age 7 months. Control mice showed no changes in their filtration barrier over time (n=12, 3 mice/genotype and age group). (C) Measurement of numbers of foot processes per μm of glomerular basement membrane was performed to assess the morphological changes in the glomerular filtration barrier. The analysis showed a significant decrease in foot processes per μm of glomerular basement membrane in Pggt1b^fl/fl mice in comparison with control Pggt1b^+/+ mice at age 7 months. ***P <0.001, unpaired t test, n=a minimum of 3800 foot processes were analyzed per each group, three mice per group.

These data show that geranylgeranylation is crucial for normal development of podocyte morphology and filtration barrier function.

GGTase-I Depleted Cells Show Increased GTPase Activity and GTPases Maintain Their Subcellular Localization

Podocyte FPE has been linked to rearrangement of the actin cytoskeleton because of a disturbed balance in RhoGTPase-activity.¹ Furthermore, it is believed that prenylation is essential for membrane anchoring and activation of RhoGTPAses.^17,34 To identify the underlying cause of FPE in our geranylgeranyl deficient mice, GTPase activity and subcellular localization was assessed through pull-down assays and cell fractionation analysis. Activity of GGTase-I in conditionally immortalized murine podocytes in culture was reduced either through knockdown of GGTase-I using lentiviral shRNA transduction or a 5-hour treatment with the GGTase-I inhibitor GGTI-298. To verify the efficiency in reduction of GGTase-I activity, inhibition of prenylation of GTPases was assessed through immunoblotting using an antibody specifically targeting nonprenylated Rap1 (Figure 6A). Cell fractionation analysis showed that the GTPases maintained their subcellular localization, that is, in the podocyte membrane and cytosol, in GGTase-I depleted cells (Figure 6B).

Figure 6.

GTPase activity is increased in geranylgeranyl transferase type-I depleted podocytes in vitro. (A) Activity of GGTase-I in conditionally immortalized murine podocytes in vitro was reduced through lentiviral shRNA knockdown of Pggt1b or a 5-hour treatment with the inhibitor GGTI-298. Reduction of prenylation after knockdown of Pggt1b and treatment with GGTI-298 was assessed using an antibody targeting nonprenylated Rap1. (B) Cell fractionation analysis of GGTase-I depleted cells shows that nonprenylated RhoGTPases maintain their subcellular localization and remain associated with the plasma membrane. C, cytosol; Cyt, cytoskeleton; M, membrane; N, nucleus. RhoGTPase activity pull-down assays showed that activity of the RhoGTPases RhoA (C), Rac1 (D), Cdc42 (E), and the RasGTPAse Rap1 (F) was increased both after lentiviral shRNA knockdown of Pggt1b and treatment with inhibitor GGTI-298 for 5 hour, compared with scrambled virus control or untreated cells. *P<0.05, **P<0.01, t test, n=3.

Reduction of GGTase-I activity through both inhibition and knockdown led to significantly increased activity of the RhoGTPases RhoA, Rac1, and Cdc42, as shown by increased GTP loading of the RhoGTPases (Figure 6, C and E). Additional proteins are known to be specifically prenylated by GGTase-I, such as RasGTPase Rap1.³³ Because Rap1 is an essential mediator of β₁ integrin activity and podocyte adhesion,²⁶ we investigated the consequences of GGTase-I depletion on the activity of Rap1. An increased activation of Rap1 was observed in both GGTase-I knockdown and inhibitor-treated podocytes (Figure 6F).

This shows the direct inhibitory effect that geranylgeranylation has on GTPase activity, both on RhoGTPases and on RasGTPase Rap1.

Depletion of GGTase-I in Murine Podocytes Alters Stress Fiber and β1 Integrin Morphology

To assess whether the altered RhoGTPase activity observed in the GGTase-I depleted podocytes had structural consequences for actin cytoskeleton formation, the podocytes were examined using confocal microscopy. Murine podocytes were depleted of GGTase-I as previously described, and the actin cytoskeleton was visualized using fluorescent phalloidin staining. The actin cytoskeleton of murine podocytes in vitro is usually structured in thin, parallel fibers spanning the width of the cell, similar to what is seen in the control cells. By contrast, we observed an altered distribution of stress fibers in the GGTase-I depleted podocytes. The main feature was increased thickness of individual actin fibers. Quantification of total fluorescence confirmed the increase in phalloidin staining (Figure 7A).

Figure 7.

Actin cytoskeleton formation and β1-integrin morphology are affected in geranylgeranyl transferase type I depleted podocytes. (A) Visualization of the actin cytoskeleton through immunofluorescence phalloidin staining revealed an altered distribution and increased thickness of actin fibers in the Pggt1b knockdown cells compared with scrambled control cells. Actin fibers ran in nonparallel directions in the Pggt1b knockdown cells, in comparison with the parallel actin fibers seen in control cells. The distribution of active (B) and total β1-integrin (C) was also affected in Pggt1b knockdown cells, with β1-integrin clusters seemingly larger in size, as shown by immunofluorescent staining. White arrows point at areas with different integrin cluster thickness. Calculations of the staining intensities are shown for all groups, confirming the increase in stress fiber thickness and integrin clustering. *P <0.05, ***P <0.001, Mann-Whitney test, at least 58 images per group, 3–15 cells per image.

The actin cytoskeleton is connected to the focal adhesion proteins, and one of the main regulators of podocyte adhesion is the α3β1 integrins.³⁵ Owing to the intrinsic interplay between RhoGTPases, stress fiber formation and focal adhesion assembly, we investigated how β1 integrins were affected by the depletion of GGTase-I. We observed that clusters of both total and active β1 integrins were thicker and larger in the depleted podocytes compared with control podocytes, similar to what was seen for phalloidin staining of the actin stress fibers. Active β1 integrins exhibited the largest variation, as shown by the staining intensity quantification (Figure 7, B and C).

These results underline how important protein geranylgeranylation is for podocyte morphology and consequent function: Its blockade causes multiple detrimental effects on actin fiber polarization, polymerization, and β1 integrin activity and quantity.

Discussion

Genetic data from patients with glomerular disease together with several reports from knockout mice and cultured podocytes have all pointed to the importance of tight control of RhoGTPase regulating proteins such as GEFs, GAPs, and GDIs in maintaining podocyte structure and filtration barrier function.

We discovered that the geranylgeranylation enzyme GGTase-I is expressed in the glomerulus, mainly in podocytes, both in humans and in mice. Podocyte-specific knock out of GGTase-I but not the farnesylation enzyme FTase in mice resulted in early onset of loss of podocyte foot processes and progressive proteinuria. The substrates of GGTase-I (RhoA, Rac1, Cdc42, and Rap1) were all activated as seen by increased GTP loading after GGTase-I inhibition in cultured podocytes. This resulted in pathophysiological alterations of podocyte actin cytoskeleton and β1 integrins. Our studies highlight the crucial role of geranylgeranylation in podocyte RhoGTPase and Rap1 regulation and its importance in podocyte morphology and filtration barrier function.

Previous studies have shown that prenylation of small GTPases is important for their intracellular localization, activity, and transformation.^17–20 Owing to the importance of small GTPases in various pathological processes, efforts have been made to find inhibitors of FTase and GGTase-I to inhibit the GTPase activity.^36–38 Trials of FTase inhibitor treatment revealed that GGTase-I is able to prenylate proteins otherwise normally targeted by FTase, suggesting that there is some target overlap between the two enzymes.^23,39 However, data indicate that there is specificity in some contexts: RhoA, Rac1, Cdc42, and Rap1 all seem to specifically be prenylated only by GGTase-I.²³ In light of this, our data regarding the absence of albuminuria in Fntb^fl/fl mice and the presence of albuminuria in Pggt1b^fl/fl mice, leads us to two conclusions: (1) The progressive albuminuria in our mice is specifically due to the absence of GGTase-I in podocytes; (2) the intracellular processes underlying albuminuria involve the classic RhoGTPases and Rap1 (geranylgeranylated), while the RhoGTPases which are farnesylated by FTase do not seem to contribute.²³

As previously mentioned, the reason for developing GGTase-I inhibitors was mainly to inactivate and consequently block processes regulated by RhoGTPases. It has been observed that depletion of GGTase-I in macrophages not only leads to increased GTP loading of Rac1 but also to development of erosive arthritis in mice.⁴⁰ Further studies on GGTase-I depletion in macrophages confirmed the increased GTP loading of RhoGTPases Rac1, RhoA, and Cdc42.^41,42 In agreement with these findings, we found that the podocyte-specific depletion of GGTase-I and even its inhibition with GGTI-298 led to an increased GTP loading of RhoA, Rac1, Cdc42, as well as Rap1. Furthermore, we found that the subcellular localization of RhoGTPases and Rap1 was not as reliant on prenylation as previously believed. Collectively, these results demonstrate that the regulatory role of prenylation on small GTPases is more complex than previously understood.

In TEM micrographs from GGTase-I depleted mice, we observed FPE, a process commonly known to be a result of imbalanced activity of RhoGTPases and an underlying cause of proteinuria.^1,3 Although our RhoGTPase activity data stem from in vitro experiments on conditionally immortalized murine podocytes, and not primary podocytes from Pggt1b^fl/fl mice, it is reasonable to believe that our in vitro results regarding RhoGTPase activity are transferable to the in vivo experiments. It has been shown that increased activity of either Rac1 or RhoA in podocytes leads to FPE and nephrotic syndrome in mice.^5–7,9 Interestingly, absence of Cdc42 proved to be affecting actin-nephrin binding causing slit diaphragm disruption resulting in congenital nephropathy in mice.^8,43

An intracellular process such as focal adhesion establishment is a good example of the importance of the regulation, in a timely and spatial manner, of the synergistic activities of RhoGTPases. Rac1 activity is needed for nascent adhesion formation, while simultaneous inhibition of RhoA activity occurs. As the nascent adhesion is transformed into a mature adhesion, Rac1 activity decreases while RhoA activity is increased.⁴⁴ Therefore, the uncontrolled increased activation of all three RhoGTPases, as observed in our study, can cause harm to the podocyte. This unrestrained priming could leave the podocyte in a stressed state unable to regulate the spatial and temporal control of RhoGTPase activity needed to maintain normal podocyte function. Both migration and immobility would be stimulated because three GTPases with contrasting functions are simultaneously active.¹⁰ Aberrant and uncontrolled RhoGTPase activation disrupts normal actin cytoskeleton formation, resulting in FPE and proteinuria, as observed in our Pggt1b^fl/fl mice.

In vitro, GGTase-I depletion affected integrin clustering. Rap1 is known to regulate cellular adhesion, where it activates and clusters integrins.²⁵ Rap1 was activated by the absence of prenylation; hence, the increased size of integrin clusters observed in Pggt1b depleted cells could be due to increased Rap1 activity. Like Cdc42, decreased activity of Rap1 has been connected to renal disease, through regulation of podocyte adhesion.²⁶ However, Rap1 also functions as an adapter protein for GEFs regulating Rac1 activity. Both Vav2 and Tiam1 have been shown to depend on Rap1 activity to promote cellular spreading.⁴⁵ Increased Rap1 activity could also lead to an increased activity of Rac1.

It is not completely understood what is causing the increased GTP loading of RhoGTPAses in the absence of GGTase-I. It has been suggested that the lack of prenylation of the RhoGTPases inhibits their ability to interact with RhoGDIs in the cytoplasm.⁴⁰ Previous studies have shown that knock out of RhoGDI-α in mice led to development of nephrotic syndrome,⁴⁶ probably because of the inability to control the GDP/GTP-bound state of RhoGTPases. Disruption of the interaction with RhoGDI could thereby be a plausible cause of RhoGTPase activity and proteinuria in our mice. Increased interaction between nonprenylated (np) RhoGTPases and IQGAP1, a protein that stabilizes GTP-bound RhoGTPases, has been suggested to partly explain the increased GTP loading of the small GTPases.⁴² IQGAP1 is known to interact with nephrin in podocyte dynamics, making it plausible that IQGAP1 could be involved in the intracellular processes in our study.⁴⁷ An increased interaction of np-Rac1 with Tiam1 was further suggested to explain the increased GTP-Rac1 after GGTase-I depletion.⁴² Activity of TIAM1 in podocytes is as yet uninvestigated, but the possibility of np-RhoGTPases having an increased affinity for other GEFs is a possible explanation for increased GTPase activity.

Nitrogen containing bisphosphonates (N-BPs) are used for treatment of osteoporosis and multiple myeloma and inhibit the production of farnesyl pyrophosphate, the precursor for geranylgeranyl pyrophosphate.⁴⁸ Acute kidney injury and focal segmental glomerulosclerosis have been reported as adverse events after treatment with N-BPs.^49–52 The exact mechanism underlying these adverse events is not yet known. Because N-BP treatment has been shown to decrease geranylation and increase RhoGTPase activity in osteoclasts and macrophages, speculations have focused on the potential effects of prenylation inhibition of RhoGTPases and the consequences for renal cells.^49,53,54 Our results support these speculations and provide a possible explanation for the pathological processes in N-BP–induced renal disease.

In this study, we show that RhoGTPases and Rap1 can be regulated in a previously undescribed way in podocytes. We show that geranylgeranylation is not a prerequisite for RhoGTPase activation. Furthermore, we suggest that inhibition of geranylgeranylation leads to the unbalanced activity of Rho and Rap1 GTPases in podocytes causing integrin and actin cytoskeleton dysregulation in vitro resulting in FPE and albuminuria in vivo.

In conclusion, our study shows that geranylgeranylation is crucial for podocyte function and maintenance of filtration barrier function.

Supplementary Material

jasn-34-641-s001.pdf ^{(3.4MB, pdf)}

Disclosures

M.O. Bergö reports Consultancy: Baxter and Pfizer; Honoraria: Baxter and Pfizer; and Speakers Bureau: Baxter and Pfizer. L. Buvall reports Employer: Gothenburg University, AstraZeneca; Ownership Interest: Oncorena Holding AB; and Patents or Royalties: AstraZeneca. J. Nyström reports Consultancy: AstraZeneca AB, and ANI Pharmaceuticals; Ownership Interest: Oncorena AB; Research Funding: AstraZeneca AB; Patents or Royalties: Oncorena AB; and Advisory or Leadership Role: Head of Institute at Institute of Neuroscience and Physiology, Oncorena board member, The Gelin research Fund board member, Royal Swedish Academy of Sciences Section Pharmacology and Physiology, The Swedish Research Council chair, The Swedish Physicians Society prize juror. All remaining authors have nothing to disclose.

Funding

We would like to thank the Swedish Medical Research Council 2014–6169 and 2019-01394, Åke Wiberg research foundation grant M16-0020 and M15-0089, the Inga-Britt and Arne Lundberg Research foundation grant LU2017-0026, the Swedish Kidney Foundation grant F2019-0045 for financially supporting this study.

ACKNOWLEDGMENTS

The authors would like to thank Dr. Peter Mundel for the immortalized murine podocytes; Dr. Martin E. Johansson for the help with pathology analysis; Dr. Murali K. Akula for providing the Pggt1b^fl/fl and Fntb^fl/fl parental mice; Dr. Maria Johansson for the metabolic cages; and Dr. Gerald F. DiBona for careful editing of the manuscript. Finally, we acknowledge the Centre for Cellular Imaging at the University of Gothenburg and the National Microscopy Infrastructure, NMI (VR-RFI 2019-00217) for providing assistance in microscopy.

Author Contributions

L Buvall, J. Nyström, L. Bergwall, R. Boi, and K. Ebefors were responsible for data curation, formal analysis, investigation, methodology, validation and visualization; L. Buvall and J. Nyström were responsible for funding acquisition and resources; and all authors wrote, reviewed and edited the manuscript.

Data Sharing Statement

All data generated or analyzed during this study are included in this article and its supplemental materials.

Supplemental Material

This article contains the following supplemental material online at http://links.lww.com/JSN/D624.

Supplemental methods.

Supplemental Figure 1. Pggt1b in vivo knock out: albuminuria detection and variation over time.

Supplemental Figure 2. Fntb in vivo knock out.

Supplemental Figure 3. Glomerular morphology in geranylgeranyl transferase type-I depleted mice.

Supplemental Figure 4. Full blots and relative total protein blots used for normalizations and calculations in Figure 6 and Supplemental Figure 1.