Mitochondrial Fission Triggered by Hyperglycemia Is Mediated by ROCK1 Activation in Podocytes and Endothelial Cells

Albuminuria Is Markedly Decreased in Mice Lacking ROCK1 in Several Experimental Models of Diabetes

We generated two diabetic mice models with targeted deletion of ROCK1 to assess the specific role of ROCK1 in DN. First, we used ROCK1-null (ROCK1^−/−) and Lepr^db/+ mice to generate diabetic Lepr^db/db ROCK1^−/− mice (hereafter referred to as db/db:ROCK1^−/−). Western blotting confirmed absence of ROCK1 protein in the kidney, heart and spleen (Figure 1A). Four groups of mice were investigated: db/+:ROCK1^+/+, db/+:ROCK1^−/−, diabetic db/db:ROCK1^+/+, and diabetic db/db:ROCK1^−/− mice. Basic parameters, including systolic blood pressure, body weight, kidney weight, plasma glucose, and urinary albumin excretion were determined for each animal. Comparison of the non-diabetic ROCK1 knockout mice with their control wild-type littermates showed that the knockout animals did not differ significantly from wild type in any of these parameters, which is in general agreement with previous reports (Chang et al., 2006; Haudek et al., 2009; Zhang et al., 2006). Double mutant diabetic mice (db/db:ROCK1^−/−) were similar to diabetic db/db:ROCK1^+/+ mice for blood pressure, body weight, kidney weight/body weight, and serum glucose levels (Figures 1B, 1C, 1D, 1E, 1F, and 1G), However, a marked reduction in the urinary albumin excretion was observed in the diabetic db/db:ROCK1^−/− mice compared with that in db/db:ROCK1^+/+ mice at 16-weeks of age (346±106 vs. 227±106 μg/mg, p<0.05) (Figure 1H). Consistent with the biochemical changes in the urine, Periodic Acid-Schiff (PAS) staining also revealed a significant reduction in mesangial matrix expansion in db/db:ROCK1^−/− mice compared with that in db/db:ROCK1^+/+ diabetic mice by 24 weeks of age (Figure 1I). Furthermore, the glomerular basement membrane thickness was reduced (Figure 1J), and the loss of podocytes was prevented in db/db: ROCK1^−/− mice (Figure 1K).

To evaluate the consequences of targeted deletion of ROCK1 in a different model of DN, ROCK1^+/+ and ROCK1^−/− mice were made diabetic with streptozotocin (STZ). Since we used a high dose of STZ to induce proteinuria (Chow et al., 2006), diabetic mice were monitored for 16–24 weeks after STZ injections to avoid potential acute kidney injury associated with STZ administration in the first 6–8 weeks. Our findings indicate that diabetic ROCK1^−/− and ROCK1^+/+ mice exhibited similar weights, blood pressure, and blood glucose levels 16 weeks post STZ injection, but knockout of ROCK1 prevented the marked increase in the diabetes-induced kidney weight/body weight ratio (Figures 2A, 2B, 2C, and 2D). Importantly, the increase in the urinary albumin excretion and enhanced mesangial matrix expansion observed in diabetic control mice after 16 weeks of hyperglycemia were prevented in diabetic ROCK1^−/− mice (Figure 2E, 2F). Furthermore, consistent with a renoprotective effect of ROCK1^−/−, the glomerular basement membrane thickness and podocyte numbers were preserved in diabetic ROCK1^−/− mice (Figures 2G, 2H). Taken together, these results demonstrate a critical role of ROCK1 activation in DN, and suggest that targeted deletion of ROCK1 is renoprotective in both type 1 and type 2 models of diabetes.

Inducible Cre-Recombinase Activation of Podocyte-Specific cA-ROCK1 Bitransgenic Mice Causes Albuminuria

To understand the role of ROCK1 overexpression in the development of albuminuria, we generated a conditionally inducible knock-in of constitutively active (cA) ROCK1 by targeting cA-ROCK1 into the ROSA 26 locus through homologous recombination. The strategy for generating cA-ROCK1 knock-in mouse is shown in Figure 3A. We generated a targeting vector carrying the cA-human ROCK1 mutant, and a transcriptional stop cassette flanked by LoxP recombination sites upstream of the cA-ROCK1 coding region to maintain the allele silent until the introduction of Cre recombinase. Two transgenic mouse lines that showed germ line transmission of the transgene were selected and maintained for further studies. Offspring genotypes from heterozygous mating followed Mendelian frequencies. To investigate the role of ROCK1 specifically in podocytes, we crossed Rosa^cA-ROCK1 transgenic mice (Tg^{cA-ROCK1flox/+}) with an inducible podocyte-specific Cre transgenic line (Tg^{podocin-iCreERT2/+}) to generate single-copy cA-ROCK-1 overexpression in podocytes (Figure 3B). The podocyte-specific Cre line of transgenic mice which expresses iCreER^T2 under the control of the podocin promoter was previously generated in our laboratory (Wang et al., 2010). The use of iCreER^T2 allowed us to control expression of floxed cA-ROCK1 in the presence of tamoxifen in a temporal-specific manner in podocytes. We generated mice that were heterozygous for the floxed cA-ROCK1 and carried the podocin-Cre transgene (Tg^{cA-ROCK1flox/+};Tg^{podocin-iCre/+}), and then activated podocin-Cre by administrating tamoxifen to bitransgenic mice at a dose of 2 mg/day/mouse for a total of 10 injections by 8 weeks of age (Wang et al., 2010). The Cre-activated bitransgenic mice (hereafter referred to as podocin Cre cA-ROCK1) were indistinguishable in size from their control littermate mice (Tg^{cA-ROCK1flox/+} or Tg^{podocin-iCre/+}, hereafter referred to as control mice) (Figure 3C). We assessed ROCK1 activation in the glomeruli of 15-week-old mice by isolating glomeruli from podocin Cre cA-ROCK1 mice 5 weeks after induction with tamoxifen (Figures S1-A, S1-B, and S1-C). ROCK1 activation assay revealed a ~3.5 fold increase of ROCK1 activation in the kidney glomeruli (Figure 3D), whereas it was unchanged in other tissues (Figure S1-D). Analysis of the inducible podocin Cre cA-ROCK1 mice showed significant increase in albuminuria (from 29±2 to 77±9 and 75±6 μg/mg, p<0.05) following 2 and 5 weeks of tamoxifen injection (Figure 3E), while no significant changes were observed on their blood pressure, body weight, blood glucose, and serum creatinine levels compared with their control littermates mice at 24 weeks of age (Figure 3F). Histologically, five weeks after induction, glomeruli from podocin Cre cA-ROCK1 mice exhibited increased mesangial matrix accumulation (Figure 3G). However, enhanced mesangial matrix was not observed consistently in all glomeruli. The main histological change was effacement of foot processes in podocytes in podocin Cre cA-ROCK1 mice on electron microscopy (Figure 3H). Furthermore, podocyte numbers were significantly reduced in podocin Cre cA-ROCK1 mice (Figure 3I). Taken together, these data demonstrate that podocyte-specific knock-in of cA-ROCK1 promotes albuminuria, podocytes effacement, and matrix expansion.

Deletion of ROCK1 Prevents Glomerular Apoptosis and Mitochondrial ROS Production in Vivo

We next explored the precise mechanisms for ROCK1-induced albuminuria. ROCK1 activation can induce apoptosis, and growing evidence has shown that mitochondrial ROS (mtROS) and apoptosis are critical in the pathogenesis of microvascular complications of diabetes, including DN (Coughlan et al., 2007; de Cavanagh et al., 2008; Kiritoshi et al., 2003; Nishikawa et al., 2000). Therefore, we tested the effect of ROCK1 on mtROS production in the glomeruli of control and STZ-induced diabetic mice. To specifically assess mitochondrial superoxide production, we performed Electron Paramagnetic Resonance (EPR) spectroscopy, and used 1-hydroxy-3-methoxycarbonyl-2,2,5,5-tetramethyl-pyrrolidine (CMH) as the spin trap (Piskernik et al., 2008; Pospisilik et al., 2007). We observed ~ 8-fold increase in the mtROS generation from the isolated kidney glomeruli of diabetic wild type mice, whereas this increase was prevented in diabetic ROCK1^−/− mice (Figure 4A, left panel). The number of apoptotic cells and the caspase-3 activity within the glomeruli of diabetic ROCK^−/− mice were also significantly reduced compared to that in wild type diabetic mice (Figure 4A, middle and right panels). Conversely, mtROS, frequency of apoptosis, and caspase-3 activity were all significantly increased in podocin Cre cA-ROCK1 mice (Figure 4B).

To elucidate the mechanisms by which deletion of ROCK1 can prevent cellular apoptosis, we carried out cell fractionation experiments to determine the effect of ROCK1 on the pro-apoptotic factors Bax and cytochrome c. Bax activation leads to its translocation to the outer mitochondrial membrane, where it can permeabilize the mitochondrial membrane by forming pores, allowing release of cytochrome c. We assessed Bax protein levels in the mitochondrial fractions, and found that ROCK1 deficiency led to significant decreases in the diabetic mice when compared with wild-type diabetic mice (Figure 4C). In contrast, mitochondrial levels of cytochrome c were significantly increased in diabetic ROCK1^−/− mice (Figure 4C, right panel). Collectively, these data indicate that ROCK1 may be involved in increased mitochondrial superoxide production, and diabetic-induced cellular apoptosis in the kidney glomeruli.

ROCK1 Mediates Diabetic-Induced Mitochondrial Fission in Vivo

Recent published work indicates that mitochondrial fission couples mitochondrial ROS with cell apoptosis in hyperglycemic conditions (Yu et al., 2006). However, two important questions remain to be addressed. First, it is unclear whether ROCK1 mediates hyperglycemia-induced mitochondrial fission, and second, it remains unknown whether ROCK1 plays a role in the pathogenesis of DN. Thus, to uncover the effect of the diabetic milieu and ROCK1 on mitochondrial dynamics in the kidneys in vivo, we examined the mitochondrial morphology in the kidney glomeruli isolated from STZ-induced diabetic ROCK1^−/− mice. In wild type (wt) diabetic mice, electron microscopy revealed increased mitochondrial fragmentation as evidenced by a significant increase in small, punctate mitochondria in podocytes compared to those in non-diabetic control mice, which exhibited long filamentous mitochondria (Figure 4D). However, the increased mitochondrial fragmentation in podocytes was prevented in diabetic ROCK1^−/− mice. To quantify changes in mitochondrial fragmentation in podocytes, we determined the aspect ratio (AR) of mitochondria in each group (Brooks et al., 2007; Yu et al., 2006). Wild type diabetic mice exhibit ~50% reduction of the AR ratio in podocytes compared with that of non-diabetic control mice (Figure 4E). In contrast, the AR ratio in podocytes from diabetic ROCK1^−/− was markedly improved and statistically significantly different from that in diabetic mice (p<0.01). To further assess mitochondrial fragmentation in diabetic podocytes, we performed 3D reconstruction of mitochondria from serial sections of EM micrographs. The mitochondria in wt diabetic mice were fragmented into short spheres. In contrast, podocytes from diabetic ROCK1^−/− exhibited a typical tubular network of mitochondria by 3D EM (Figure 4F). Similar results were obtained in db/db:ROCK1^−/− mice (Figures S2-A, and S2-B).

We next examined mitochondrial morphology in podocytes from podocin Cre cA-ROCK1 mice. We observed a significant increase in mitochondrial fission in these mice as suggested by a marked decrease in AR ratio (Figures 4G, and 4H). Furthermore, mitochondrial Bax (mtBax) levels were increased, whereas levels of mitochondrial cytochrome c were markedly reduced in the kidney glomeruli of podocin Cre cA-ROCK1 mice (Figures 4I).

Since Dynamin-related protein (Drp) 1 is a key component of the mitochondrial fission machinery (Jagasia et al., 2005), we tested whether the effect of ROCK1 on mitochondrial fission involved Drp1 activation. Mitochondrial recruitment of Drp1 appears to be required for its action on mitochondrial fission (Chang and Blackstone, 2010), therefore, we examined the possibility that ROCK1 mediates mitochondrial translocation of Drp1 in the diabetic milieu. We observed a marked increase in Drp1 in mitochondria-enriched fractions (mtDrp1) of the kidney glomeruli of wild-type diabetic mice (Figure 5A, left panel, and Figure S2-C). The increase in mitochondrial Drp1 levels was abolished in ROCK1^−/− diabetic mice. Conversely, mtDrp1 was significantly increased in the kidney glomeruli of podocin Cre cA-ROCK1 mice (Figure 5A, right panel, and Figure S2-C), indicating that ROCK1 plays an important role in remodeling of mitochondrial morphology in podocytes, possibly by regulating Drp1 recruitment to the mitochondria in vivo.

ROCK1 Mediates Hyperglycemic-Induced mtROS and Apoptosis in Podocytes and Kidney Microvascular Endothelial Cells (mECs) in Vitro

We next studied the effects of ROCK1 on mtROS, cellular apoptosis, and mitochondrial fission in cultured cells. We focused on podocytes and endothelial cells because of strong co-localization of ROCK1 with markers of podocytes and endothelial cells in the glomeruli (Figure S3-A). We simultaneously measured the effect of ROCK1 on mitochondrial superoxide production and apoptotic markers by flow cytometry in cultured podocytes and kidney microvascular endothelial cells (mECs) as previously described (Mukhopadhyay et al., 2007). The data in Figure 5B represent the distribution of non-apoptotic, apoptotic, and dead cells by Annexin V-FITC and Sytox Red staining and histogram analysis of mean intensity of oxidized MitoSOX Red in podocytes. Cells were separated into three groups: non-apoptotic cells (green), early apoptotic cells (purple), and dead cells (orange). The total fluorescence intensity of oxidized mitoSOX red measured by flow cytometry is depicted in light blue. High glucose stimulation (HG, 25mM for 24–36 hrs) increased the number of apoptotic cells ~4-fold (Figure 5C), and induced ~2.5-fold increase in the mtROS (Figure 5D) as measured by MitoSOX Red fluorescence intensity compared to that in normal glucose (NG, 5mM) in cultured podocytes. Next, we knocked down expression of ROCK1 with a specific ROCK1 siRNA (50nM, Figure S3-B) or used fasudil (10μmol/L) to assess the role of endogenous ROCK1. Knockdown of ROCK1 with siRNA or pharmacological inhibition of ROCK1 with fasudil significantly decreased the number of apoptotic cells (10±0.05 and 11%±0.8 vs 35±2%; p<0.05) and the production of mtROS (733±70 and 801±57 vs. 1459±73 mean fluorescence intensity, p<0.05) (Figures 5B, 5C, and 5D). These data suggest a critical role for ROCK1 activation in HG-induced mtROS production and cell apoptosis. Similar data were obtained in mECs (Figures S3-C, S4-A and S4-B).

To better understand the role of ROCK1 in HG-induced cellular apoptosis and its downstream signaling, we first examined the temporal profile of HG-induced ROCK1 activation in cultured podocytes and mECs. ROCK1 activity was significantly increased after 6 hrs with HG-treatment (25 mM) in both podocytes (Figure 5E) and mECs (Figure S5-A), suggesting that ROCK1 activation is an early response to HG stimulation. Mitochondrial fractionation experiments indicated that mtBax increased and cytochrome c decreased in cells upon their exposure to HG (Figures 5F, and 5G). However, ROCK1 knockdown prevented the effect of HG on mtBax and cytochrome c. Furthermore, knockdown of ROCK1 also reversed HG-induced caspase-3 activity in podocytes (Figure 5H), suggesting a critical role for ROCK1 in HG-induced mitochondrial apoptotic pathway, consistent with our in vivo studies. Similar results were obtained in mECs (Figures S5-B, S5-C, and S5-D). Conversely, cA-ROCK1 transfected-podocytes exhibited increased mtBax and caspase-3 activity and lower levels of cytochrome c in mitochondria (Figures 5I, and 5J).

ROCK1 Mediates Hyperglycemia-Induced Drp1 Recruitment to the Mitochondria

We next examined whether ROCK1 recruitment to the mitochondria was necessary for its effect on mitochondrial fission and increased ROS generation. We stimulated podocytes with HG as previously described and performed double immunofluorescence staining for ROCK1 and MitoTracker, a mitochondrial marker. ROCK1 did not co-localize to mitochondria in response to HG (25 mM for 18 hrs) and its protein expression was not detectable in the mitochondrial fractions from HG-stimulated podocytes in vitro or from kidney glomeruli obtained from STZ-induced diabetic mice in vivo (Figures S6-A, and S6-B). These observations suggest that the effect of ROCK1 on mtROS and apoptosis is not dependent on the recruitment of ROCK1 to the mitochondria.

Since Drp1 translocation to the mitochondria is a key event in mitochondrial fission (Jagasia et al., 2005; James and Martinou, 2008; Labrousse et al., 1999), we next tested the possibility that ROCK1 might exert its effect on mitochondrial fission by promoting Drp1 translocation to the mitochondria. We initially examined the effect of ROCK1 on subcellular distribution of Drp1. We observed that upon stimulation with HG, mitochondria became short and fragmented, whereas podocytes transfected with specific siROCK1 or siDrp1 maintained their elongated mitochondrial morphology in hyperglycemic conditions (Figures 6A, and 6B). We further observed that ROCK1 knockdown markedly reduced HG-induced Drp1 recruitment to the mitochondria in podocytes (Figures 6C, and S7-A) and in mECs (Figures S7-B, and S7-C), suggesting that ROCK1 may modulate mitochondrial fission through Drp1 translocation to the mitochondria. To confirm the effect of ROCK1 on mitochondrial dynamics, we next examined the effect of ROCK1 overexpression on mitochondrial fission. Whereas podocytes transfected with cA-ROCK1 exhibited more fragmented mitochondria with a marked increase in Drp1 recruitment to the mitochondria under normal glucose (NG, 5mM) conditions, this fragmented phenotype was reversed when Drp1 was knocked down by a lentiviral Drp1 short hairpin RNA (shRNA) (Figures 6D, 6E, 6-F, and S7-D). Notably, simultaneous analysis of measured superoxide production and apoptosis revealed that podocytes co-transfected with cA-ROCK1 and Drp1 shRNA had lower levels of mtROS and exhibited less cellular apoptosis (Figure 6G). Taken together, these data indicate that ROCK1 may promote mitochondrial fission, mtROS production, and cellular apoptosis through Drp1 translocation to the mitochondria.

ROCK1 Mediates Drp1 Recruitment to the Mitochondria by Phosphorylating Drp1 at Serine⁶⁰⁰ Residue

Because virtually nothing was known about the effect of ROCK1 on regulating subcellular localization of Drp1, we sought to define the regulatory mechanisms by which ROCK1 activated Drp1. We initially questioned whether ROCK1 can interact with Drp1 in response to HG stimulation in intact cells. Transfected mECs expressing Flag-Drp1, ROCK1, or co-expressing both plasmids were stimulated with HG (25 mM). Immunoprecipitation of cell lysates with an anti-Flag antibody, followed by Western blot analysis revealed that ROCK1 and Drp1 were co-immunoprecipitated when cells were treated with HG for 5 hrs (Figure 7A), indicating that ROCK1 and Drp1 are closely associated under hyperglycemic conditions.

The ability of ROCK1 to associate with Drp1, in response to hyperglycemic stimulation, led us to examine whether Drp1 might be a direct substrate of ROCK1. We tested the possibility that ROCK1, a serine/threonine kinase, can induce phosphorylation of Drp1 at a serine residue and whether serine phosphorylation of Drp1 could have mediated the effect of ROCK1 on Drp1 recruitment to the mitochondria. In support of our hypothesis, previous published work has shown that a series of kinases can influence subcellular localization of Drp1 by phosphorylating two main conserved serine residues of Drp1 (Chang and Blackstone, 2007b, 2010). Based on the consensus sequence motif of ROCK substrates (R/K-X-X-S/T or R/K-X-S/T where R is arginine; K, lysine; S, serine; and T; threonine) (Kang et al., 2007), we identified Ser⁶⁰⁰ in the mouse Drp1 isoform b as a potential phosphorylation site for ROCK1 (Figure 7B). This serine phosphorylation site is highly conserved among species and is located at the GTPase effector domain (GED) of Drp1 (Figure 7B). Using phospho-specific Drp1 antibodies, we observed that HG stimulation (25 mM for 36 hrs) increased phosphorylation of Drp1 at Ser⁶⁰⁰ (corresponding to Ser⁶³⁷ in human Drp1 isoform 1) (Figure 7C). ROCK1 knockdown, either by using dominant negative ROCK1 (dn ROCK1) or siROCK1, attenuated Drp1 phosphorylation at Ser⁶⁰⁰, but not at Ser⁵⁷⁹, another regulatory phosphoserine site of Drp1 (corresponding to Ser⁶¹⁶ in human Drp1 isoform-1) (Figure 7C).

To determine whether Ser⁶⁰⁰ was the precise site for phosphorylation by ROCK1 in vitro, we performed an in vitro kinase assay using bacterially expressed Drp1 as a substrate for purified ROCK1. ROCK1 protein was labeled with [γ-³²P] ATP in vitro, and Drp1 phosphorylation was assessed in wild type Drp1 (wt Drp1) or a Drp1 mutant protein (S600A) where Ser⁶⁰⁰ was replaced with alanine. As shown in Figure 7D, we observed that ROCK1 was able to phosphorylate wt Drp1 protein, but not Drp1 S600A protein in vitro. These data indicate that ROCK1 specifically phosphorylates Drp1 at the Ser⁶⁰⁰ residue.

Having identified the site of ROCK1-induced Drp1 phosphorylation, we next assessed the contribution of Ser⁶⁰⁰ phosphorylation on Drp1 recruitment to the mitochondria. Two recent publications have provided seemingly contradictory data on the functional consequences of Drp1 phosphorylation at Ser⁶⁰⁰. Whereas one study reported that phosphorylation of Drp1 at Ser^637/656 (human/rat) (corresponding to Ser⁶⁰⁰ in mouse Drp1 isoform b) decreased Drp1 recruitment to the mitochondria (Chang and Blackstone, 2007a), a more recent report demonstrated that phosphorylation of the same site in human Drp1 isoform 3 by the Ca²⁺/calmodulin-dependent protein kinase Iα (CaMK1α) enhanced Drp1 recruitment to the mitochondria and increased mitochondrial Drp1 abundance (Han et al., 2008). We tested the effect of phosphorylation of Drp1 at Ser⁶⁰⁰ on its subcellular localization by transfecting podocytes with Flag-Drp1 (wt) or Flag-Drp1 S600A. We found that HG (25 mM for 36 hrs) induced a significant increase in Drp1 recruitment to the mitochondria, that was reduced in cells transfected with Drp1 S600A (Figure 7E), suggesting that ROCK1-mediated phosphorylation of Drp1 at Ser⁶⁰⁰ site promotes Drp1 translocation to the mitochondria. Furthermore, to assess the effect of Drp1 phosphorylation at Ser⁶⁰⁰ on mitochondrial fission, we knocked down endogenous Drp1 in podocytes using a lentiviral Drp1 shRNA, and analyzed the mitochondrial morphology in cells transfected with either an empty vector, wild type (wt) Drp1, or S600A mutant of Drp1. Whereas HG-induced cells transfected with Drp1 shRNA had long mitochondria, cells transfected with wt Drp1 exhibited short punctate mitochondria (Figures 7F, and S7-E). Conversely, cells transfected with Drp1 S600A exhibited a significant shift from the fragmented phenotype to elongated mitochondria (Figure 7F, right panel), suggesting a direct effect of Drp1 phosphorylation at Ser⁶⁰⁰ on mitochondrial fission. Consistent with these results, Drp1 knockdown in HG-treated cells expressing Drp1 S600A decreased cellular apoptosis compared with cells expressing wt Drp1 (Figure 7G). Importantly, podocytes transfected with a phosphomimetic version of Drp1 (Drp1 S600D) showed a marked increase in mitochondrial fragmentation (Figure 7H). These gain-of-function experiments support the previously describe loss-of-function experiments, and indicate a critical role of ROCK1 in mitochondrial fission by promoting phosphorylation of Drp1 at Ser⁶⁰⁰.

Animals

All animal studies were carried out with approval of the Institutional Animal Care and Use Committee (IACUC) of Baylor College of Medicine. Generation and characterization of ROCK1^−/− mice have been previously described (Zhang et al., 2006). Since db/db mice are infertile, to generate db/db:ROCK1^−/− double mutant mice, FVB db/+ (Jackson Laboratory Bar Harbor, ME) mice were initially mated with FVB ROCK^−/− mice, and then db/+:ROCK1^+/− were crossed with ROCK1^−/− mice, and finally db/+:ROCK1^−/− mice were intercrossed to generate db/db:ROCK1^−/− double mutant mice. To create an inducible constitutively active (cA) ROCK1 knock-in mouse, a cA mutant of ROCK1 (human Δ3 cA-ROCK1) was used to generate the cA-ROCK1 knock-in targeting vector (Fujisawa et al., 1996). The ubiquitin C promoter was used to drive cA-ROCK1 cDNA expression only when the loxP-stop-loxP is removed. A Neo cassette driven by a PGK promoter was used to facilitate selection of positive ES cell clones. FRT sequences were recognized by FLP recombinase. We generated chimaeras by blastocyst injection of ES cells. After germline transmission, the mice were crossed to C57BL/6J mice expressing FLP recombinase to remove the neo cassette.

Plasmid DNA and siRNA

Mouse Drp1 S600A and S600D mutations were achieved by using PCR-based mutagenesis with the following primers respectively: CCAGTTG CAAGAAAGCTAGCTGCCCGAGAACAGCGAGAT (Forward) and AAAACTGGAT GCCCGAGAACAG (Reverse). C-terminal mouse Drp1 (isoform b, 490–699) was amplified by PCR from the cDNA clone (Open Biosystems) using the following primer set: GTGAATTCGAGCTAGCGTATATCAACACAAAACACCCC (forward) and AGG GATCCTCACCAAAGATGAGTCTCTCGGATTTCAGC (reverse), and subcloned between EcoRI and BamHI sites of pGTK3XF vector. siRNAs for ROCK1, Drp1, Bax and scramble siRNA controls were purchased from Thermo Scientific (Chicago, IL). Transfections were performed using Lipofectamine 2000^™ (Invitrogen, Carlsbad, CA) according to manufacturer’s protocols.

Generation of Drp1-knockdown stable cells

Mouse Drp1 shRNA lentiviral particle and the scramble shRNA lentiviral particle were purchased from Santa Cruz Biotechnology. These viral particles were used to infect undifferentiated podocytes with 8 μg/ml polybrene (Sigma) and were subsequently selected with 2 μg/ml puromycin (Sigma) in the presence of 20 U/ml IFN-γ at 33°C.

In vitro kinase assay

Recombinant GST-Drp1 proteins were purified using glutathione sepharose (GE Healthcare) according to the manufacturer’s protocol, and digested with PreScission protease (GE Healthcare) to remove the GST-tag. 1 μg of Drp1 wild-type or S600A mutant was incubated with 0.1 μg purified active ROCK1 (GST fusion of amino acid 17–535 of human ROCK1, Abcam) in a 30 μl reaction buffer [25 mM Tris (pH 7.5), 10 mM MgCl₂, 0.5 mM EGTA, 0.5 mM Na₃VO₄, 5 mM β-glycerophosphate, 2.5 mM DTT, 0.01% Triton X-100, 100 μM ATP and 10 μCi [γ-³²P]-ATP. After 30min incubation at 37°C, reactions were stopped by adding SDS sample buffer to 1X and separated by 12% SDS-PAGE.

Flow cytometry

Detection of apoptosis and mitochondrial superoxide production was performed as previously described (Mukhopadhyay et al., 2007). Samples were analyzed using BD FACSCantoII flow cytometer (BD Biosciences, CA) at the BCM Cell Sorting Facility.

Mitochondrial fragmentation

The Aspect ratio was calculated as previously described (Brooks et al., 2009; Brooks et al., 2007; Yu et al., 2006).

Statistical analysis

The data are summarized as the mean ± SEM. Statistical significance was assessed by performing analysis of variance (ANOVA) followed by the Tukey-Kramer post hoc analysis for multiple comparisons using an α value of 0.05 in GraphPad Prism software (San Diego, CA).