Inhibition of insulin/IGF‐1 receptor signaling protects from mitochondria‐mediated kidney failure

Podocyte‐specific Phb2 knockout mice develop albuminuria and die prematurely

To understand the contribution of mitochondrial dysfunction to kidney disease, we deleted the Phb2 gene specifically in podocytes (Phb2^pko) by mating a conditional Phb2^fl/fl mouse line (Merkwirth et al, 2008) to podocyte‐specific Cre mice (NPHS2.cre) (Moeller et al, 2003). Mice of all genotypes were born following Mendelian rules (Supplementary Fig S1A). At birth, Phb2^pko mutants appeared as healthy as wild‐type or heterozygous controls and did not show signs of glomerular dysfunction. The absence of PHB2 resulted in progressive and marked proteinuria reaching an albumin‐to‐creatinine ratio of 470 mg/mg at day 28 (Fig 1A), growth retardation, and massive loss of body weight in comparison with their wild‐type (Phb2^fl/fl) and heterozygous littermates (Phb2^het) (Fig 1B and C). Phb2^pko mutant animals developed renal failure as indicated by a rise in serum creatinine and urea (Fig 1D and E). All animals died prematurely within 31–37 days postpartum (Fig 1F).

Phb2^pko mice develop glomerulosclerosis

Histological analyses confirmed a normal renal histology at 14 days after birth but revealed progressive glomerulosclerosis at later stages, a phenotype closely resembling human glomerular disease (Fig 2A). At 28 days, Phb2^pko mice presented with marked deposition of protein casts in the tubular system, collapsing glomerular capillaries and global glomerulosclerosis.

It is well established that in podocyte disease, the elaborate cytoarchitecture of podocyte foot processes is altered, leading to loss of the typical slit diaphragm structure. Slit diaphragms form a membrane‐like cell‐to‐cell contact that contains a protein complex of podocin and associated proteins (Brinkkoetter et al, 2013). Consistent with glomerulosclerosis being the result of podocyte disease, immunohistochemistry stainings for podocin showed alterations at 21 days and a dispersed and reduced staining at 28 days (Fig 2B). At 21 days, we observed podocyte effacement, that is, severe changes in foot process organization in all analyzed glomeruli of Phb2^pko mice (Fig 2C). Podocyte effacement results in proteinuria, progressive renal damage with glomerular scarring, and eventual loss of renal function, necessitating renal replacement therapy in humans. At 28 days, the podocyte foot process structure was completely lost. As PHB2 is required for maintaining the structure of mitochondrial cristae, we assessed mitochondrial architecture (Fig 2D and E). At 14 days, both mitochondrial cristae structures and overall morphology were unaffected. However, 1 week later, lamellar cristae structures were disorganized or completely lost in mitochondria of Phb2^pko podocytes, which is in line with previous reports of mitochondrial changes after deletion of Phb2 in neurons (Merkwirth et al, 2012). At 28 days, we were unable to analyze mitochondrial appearance due to progressive sclerosis and the overall loss of glomerular architecture.

As loss of PHB2 might render podocytes susceptible to apoptosis (Merkwirth et al, 2008, 2012; Baris et al, 2011; Wu & Wu, 2012), we studied whether the development of albuminuria was due to increased levels of apoptotic cell death resulting in reduced podocyte cell number. At 21 days, staining for cleaved caspase‐3 was completely negative (Supplementary Fig S2A). In addition, we quantified podocyte cell number by immunohistochemistry staining for WT‐1 in a blinded manner. Again, there were no differences in the overall podocyte cell number between Phb2^fl/fl control and Phb2^pko animals (Supplementary Fig S2B and C). In summary, Phb2^pko mice did not show signs of increased apoptosis or cell loss at this early time point.

The phenotype of Phb2^pko mice is not developmental

Although all relevant phenotypic changes were observed after day 14, at a time glomerulogenesis is completed, a developmental phenotype resulting from PHB2 loss could still not be excluded. We therefore tested whether the induction of PHB2 loss in adulthood might cause a similar phenotype. Phb2^fl/fl mice were mated to podocin‐iCreER(T2) mice that allow expression of Cre recombinase activity upon tamoxifen induction (Wang et al, 2010). The resulting podocin‐iCreER(T2) Phb2^pko animals (tPod–Phb2^pko) received a tamoxifen‐enriched diet and developed progressive glomerulosclerosis (Fig 3A and B) and massive albuminuria (Fig 3C) as observed in non‐inducible Phb2^pko mice. Thus, the phenotype of Phb2^pko mice does not appear to be the result of altered developmental programs.

Podocyte‐specific knockout of the insulin receptor and IGF‐1 receptor prolongs survival of Phb2^pko mice

Studies in yeast and the nematode Caenorhabditis elegans provided an intriguing link of mitochondrial PHB deficiency and altered insulin signaling (Artal‐Sanz & Tavernarakis, 2009; Schleit et al, 2013). In worms, loss of PHB2 decreased survival but prolonged lifespan of long‐lived daf‐2 mutants (Artal‐Sanz & Tavernarakis, 2009). In yeast, Phb2 deficiency activated the mitochondrial unfolded protein response (mtUPR), which was associated with reduced lifespan. This effect could be inhibited by dietary restriction reducing the mtUPR (Schleit et al, 2013). Given the complex, yet ill‐defined interactions of Phb2 deficiency and insulin signaling in model organisms, we tested whether inhibition of insulin and/or IGF‐1 receptor signaling specifically in podocytes modulated latency or severity of the disease phenotype of Phb2^pko animals. Surprisingly, genetic deletion of the insulin receptor (Insr) in podocytes but not of the IGF‐1 receptor (Igf1r) attenuated renal disease of Phb2^pko mice (Fig 4A and B). Phb2^pko mice lacking a functional insulin receptor in podocytes (Insr^pko) showed a highly significant survival advantage (Fig 4C). Generation of triple‐knockout animals with a genetic deletion of both insulin receptor and IGF‐1 receptor in addition to PHB2 (Phb2^pko/Insr^pko/Igf1r^pko) significantly alleviated renal disease, enhanced survival (Fig 4D and E), improved kidney function, and delayed the onset of kidney failure (Fig 5A).

To exclude the possibility that the Cre recombinase insufficiently recombines three genes and thereby alleviates the Phb2‐related phenotype in the triple‐knockout animals, we isolated genomic DNA from glomeruli of Phb2^pko and Phb2^pko/Insr^pko/Igf1r^pko mice and performed quantitative PCR experiments specific for the deleted Phb2 gene. These studies did not detect any differences in Phb2 gene deletion efficiency between the single‐ and the triple‐knockout mice clearly excluding the possibility of altered recombination efficiency as a basis for the survival advantage (Fig 5B). In line with this, Phb2^pko/Insr^pko/Igf1r^pko mice showed the same mitochondrial structural changes as seen in Phb2^pko mice (Fig 5C). Moreover, double‐knockout as well as triple‐knockout mice developed massive albuminuria (Supplementary Fig S3A and B).

Insr^pko/Igf1r^pko showed normal survival for the time analyzed (until 8 weeks after birth) and did not develop proteinuria or renal disease (Supplementary Fig S4A and B) in contrast to previous observations demonstrating renal disease at later time points in life (Welsh et al, 2010).

Loss of PHB2 leads to changes of mitochondrial morphology

To better understand the beneficial effect of podocyte‐specific insulin signaling deficiency for renal disease and the role of PHB2 on metabolic signaling in podocytes, we next generated a Phb2 knockdown podocyte cell culture model. As Phb2 deficiency leads to a reduced cellular proliferation rate impeding the generation of conventional stable cell lines (Merkwirth et al, 2008), we utilized a doxycycline‐inducible promoter to express short hairpin RNAs directed against Phb2 mRNA in mouse podocytes (Shankland et al, 2007). After doxycycline treatment, decreased mRNA expression of Phb2 but not Phb1 was detected (Supplementary Fig S5A). Western blot analysis revealed an interdependence of protein stability of PHB1 and PHB2 and showed reduced protein levels for both PHB1 and PHB2 in Phb2 knockdown podocytes consistent with previous findings (Artal‐Sanz et al, 2003; Merkwirth et al, 2008) (Supplementary Fig S5B). Knockdown of Phb2 in podocytes resulted in a disrupted reticular mitochondrial network and the accumulation of fragmented mitochondria (Fig 6A and B). The mitochondrial morphology was further analyzed by using morphometric image analysis of the ratio of discrete mitochondrial number to total mitochondrial area as a quantitative measure of mitochondrial length/interconnectivity (Losón et al, 2013). These data showed a significant difference between control and Phb2 knockdown podocytes (Fig 6C). Moreover, we detected less branches and smaller mitochondrial sizes after loss of PHB2 (Fig 6D and E). However, these structural abnormalities were not accompanied by an impaired oxidative phosphorylation (OXPHOS) system. We did not detect any differences in oxygen consumption rates between Phb2‐deficient and control podocytes using complex II substrates (succinate and glycerol‐3‐phosphate) (Fig 6F). Similar results were obtained when complex I substrates (pyruvate, glutamate, and malate) were used (unpublished data). Furthermore, there was no evidence for a compensatory upregulation of mitochondrial number as we did not detect differences in total mitochondrial mass by using both, quantification of mtDNA by qPCR (Fig 6G) and FACS analysis of podocytes stained with the mitochondrial marker MitoTracker (Fig 6H and I). Additionally, we did not detect any change in reactive oxygen species (ROS) in Phb2‐deficient podocytes (Fig 6H and I). These findings are in accordance with previous publications showing an unaltered mitochondrial respiratory function after loss of PHB2 in MEFs (Merkwirth et al, 2008). These results indicated that rather a cellular signaling program than a mitochondrial respiratory function defect may account for the severe podocyte dysfunction and cell loss.

Inhibition of mTOR signaling increases survival of Phb2^pko mice

Hyperactive mTOR in podocytes contributes to kidney disease, and genetic reduction of podocyte‐specific mTOR complex 1 (mTORC1) in diabetic animals suppresses the development of diabetic nephropathy (Gödel et al, 2011; Inoki et al, 2011). Therefore, we next investigated the activity of mTOR signaling in the podocyte cell culture model. Western blot analysis of Phb2‐deficient podocytes revealed hyperphosphorylation of S6 ribosomal protein (S6RP), a component of the 40S ribosomal subunit, indicative of a hyperactive mTOR signaling pathway (Fig 7A) (Magnuson et al, 2012). Hyperphosphorylation of S6RP was also observed in Phb2‐deficient glomeruli in vivo (Fig 7B). These data indicated that hyperactivity of mTOR signaling may contribute to the detrimental effects of loss of PHB2 in podocytes. And in fact, hyperphosphorylation was partially reversed in the protected Phb2^pko/Insr^pko/Igf1r^pko triple knockouts, where phosphorylated S6 ribosomal protein was mainly found in non‐podocyte cells (Supplementary Fig S6A) and treatment with a dual insulin receptor/IGF‐1 receptor inhibitor (BMS 536924, Tocris) decreased phosphorylation of S6RP in Phb2‐deficient podocytes in vitro to the level of control podocytes (Supplementary Fig S6B). These data raise the possibility that mitochondrial dysfunction resulting from Phb2 deficiency causes hyperactive insulin/mTOR signaling in podocytes. This hypothesis was further supported by experiments in C. elegans. Heat‐shock treatment of worms results in the nuclear translocation of DAF‐16 independent of insulin signaling (Supplementary Fig S7A). Cytoplasmic redistribution, a process known to be controlled by insulin receptor (DAF‐2) signaling, was used as an indicator of insulin receptor (DAF‐2) signaling in phb‐2‐proficient and phb‐2‐deficient worms (Supplementary Fig S7B–D). Consistent with hyperactive insulin signaling in phb‐2‐deficient animals, loss of PHB‐2 accelerated recovery of the DAF‐2‐dependent cytoplasmic localization of DAF‐16 after heat shock (Supplementary Fig S7A–D), further supporting a role for insulin receptor‐mediated mTOR hyperactivity resulting from the mitochondrial defects.

Taken together, these data suggested the intriguing possibility that the severe kidney disorder in Phb2^pko mice may be amenable to a therapeutic intervention with mTOR inhibitors. Treatment with the mTOR inhibitor rapamycin abrogated phosphorylation of S6RP in the cell culture model (Fig 7A). Daily treatment with rapamycin (3 mg/kg) by intraperitoneal injection beginning at week 2 alleviated the kidney disease and significantly enhanced survival of Phb2^pko animals (Fig 7C and D) and reversed the hyperphosphorylation of S6RP (Supplementary Fig S8A), suggesting that at least some of the beneficial effects of insulin and IGF‐1 receptor deficiency of podocytes are mediated through preventing mTOR hyperactivity. Interestingly, we did not observe major changes in proteinuria levels which may indicate additional mTOR‐independent effects accounting for the proteinuric phenotype (Supplementary Fig S8B).

Since active mTOR inhibits autophagy in podocytes (Inoki & Huber, 2012), we also quantified accumulation of the autophagosome marker LC3 in podocytes in vitro. Although a trend to less autophagy was seen in Phb2‐deficient podocytes compared to control podocytes (Supplementary Fig S9A), this effect was not statistically significant. Even inhibition of autophagosomal degradation by adding chloroquine to the cells did not result in any significant differences in the accumulation of autophagosomes (Supplementary Fig S9A). This is in accordance with our electron microscopy studies where we did not observe a reduced amount of autophagosomes in Phb2^pko mice, suggesting that the mTOR effect may be independent of influencing the level of autophagy.

Antibodies

The antibodies used in this study are listed in Table 1.

Mouse models

Mice in which exons 3 and 4 of the Phb2 gene are flanked by two loxP sequences (Phb2 flox/flox mice; Merkwirth et al, 2008) were mated to either NPHS2.Cre mice (Moeller et al, 2003) or tamoxifen‐inducible podocin‐iCreER(T2) (Wang et al, 2010) to generate podocyte‐specific Phb2 knockout mice Phb2^fl/fl;NPHS2.Cre (Phb2^pko) or tamoxifen‐inducible podocyte‐specific Phb2 knockout mice Phb2^fl/fl; podocin‐iCreER(T2) [podocin‐iCreER(T2) Phb2^pko]. We included mice from both genders carrying the conventional NPHS2.cre transgene while only 8‐week‐old male animals were exposed to a tamoxifen‐enriched diet (400 mg/kg; TD.55125; Harlan Laboratories, Indianapolis, IN, USA) ad libitum for 6 weeks to exclude any potential side effects in female animals. Based on a daily intake of 5 g of food per mouse, this corresponded to an oral dose of 2 mg/day tamoxifen per mouse.

To generate double‐ and triple‐knockout mice, first, Phb2 flox/flox mice were mated to Igf1r flox/flox and/or Insr flox/flox mice. Second, these mice were mated to podocyte‐specific Cre mice (NPHS2.cre mice) (Moeller et al, 2003) to obtain mice with a podocyte‐specific deficiency of either Phb2 and Igf1r (Phb2^pko/Igf1r^pko mice) or Phb2 and Insr (Phb2^pko/Insr^pko mice) or Phb2 and Insr and Igf1r (Phb2^pko/Insr^pko/Igf1r^pko mice).

All animals were backcrossed onto the C57BL/6 background for at least 10 generations. Mice were housed according to the standardized specific pathogen‐free conditions in the University of Cologne animal facility. The Animal Care Committee of the University of Cologne reviewed and approved the experimental protocol. The mice were sacrificed once they lost 15–20% of their maximal body weight following federal animal care regulations. Urine was analyzed with Coomassie gel stains, ELISA (mouse albumin ELISA kit; Bethyl Labs, Montgomery, TX, USA), and urine creatinine assay (Cayman Chemical, Ann Arbor, MI, USA). Blood samples were collected in the fifth week of life for quantification of serum urea and creatinine by the Institute for Clinical Chemistry of the University Hospital of Cologne, Germany. Renal tissue was embedded in OCT compound (Sakura, Torrance, CA, USA) and frozen at −70°C or fixed in 4% neutral buffered formalin for immunostaining.

For rapamycin injections, the protocol published by Zeng et al (2008) was used with some modifications. Animals received daily i.p. injections of either rapamycin solution (LC Laboratories, Woburn, MA, USA) or vehicle solution [5% (v/v) ethanol, 5% (v/v) Tween^® 80, 5% (v/v) PEG400] starting 2 weeks after birth until they were sacrificed because of weight loss or sickness. Phb2^pko and control mice were injected with either 3 μg rapamycin solution per g body weight or the same volume of vehicle solution (DMSO). Rapamycin solution was always freshly prepared prior to injection. Animals were randomly assigned to different treatment groups independently of genotyping results.

The sample size for the animal studies was based on ANOVA calculations according to the variable‐criteria sequential stopping rule (SSR). All analyses and quantifications were performed in a blinded fashion, and no animals were excluded from the evaluation or the statistics.

Immunohistochemistry

Indirect immunoperoxidase staining was performed on formalin‐fixed tissue. Briefly, 4‐μm‐thin tissue sections were deparaffinized in Xylene (VWR, Darmstadt, Germany) and rehydrated in graded ethanol. Endogenous peroxidase activity was blocked with 3% hydrogen peroxidase (Fischar, Saarbruecken, Germany). Sections were incubated overnight at 4°C with primary antibodies diluted in 1% BSA/TBS. The sections were washed repeatedly in TBS before incubation with biotinylated mouse anti‐rabbit secondary antibody (Jackson Immunoresearch, West Grove, PA, USA) diluted in 1% BSA/TBS for 1 h at room temperature. The ABC kit (Vector, Burlingame, CA, USA) was used for signal amplification, and 3,3′‐diaminobenzamidine (Sigma‐Aldrich, St Louis, MO, USA) was used as a chromogen. Slides were counterstained with hematoxylin (Sigma‐Aldrich), dehydrated, and covered with Histomount (National Diagnostics, Atlanta, GA, USA). Periodic acid Schiff (PAS) staining was performed for the assessment of glomerulosclerosis. Images were acquired with an Axiovert 200 M microscope/EC Plan‐Neofluar ×40/1.3 oil immersion or C‐Apo ×63/1.20 water immersion objective equipped with a charge‐coupled‐device camera (all from Carl Zeiss MicroImaging GmbH, Jena, Germany). Images were further processed using ImageJ/Fiji software version 1.46 (NIH, Bethesda, MD, USA).

Immunofluorescence

Fresh cut frozen sections were fixed with 4% PFA before overnight incubation at 4°C with anti‐phospho‐S6 ribosomal protein (Ser235/236) from Cell Signaling, Danvers, MA, diluted in 5% NDS/PBS+Triton X‐100. The sections were washed repeatedly with PBS before incubation with fluorescently labeled anti‐rabbit secondary antibody (Jackson Immunoresearch) diluted in PBS for 1 h at room temperature. Anti‐Podocin from Sigma‐Aldrich was directly labeled with a fluorescent dye according to the manufacturer's instructions (Zenon^® Rabbit IgG Labeling Kit; Life Technologies, Carlsbad, CA, USA). Sections were mounted with ProLong^® Gold antifade reagent (Life Technologies).

Cells were fixed with 4% PFA before overnight incubation at 4°C with the primary antibody diluted in 5% NDS/PBS+Triton X‐100. The sections were washed repeatedly with PBS before incubation with the fluorescently labeled secondary antibody (Jackson Immunoresearch) diluted in PBS for 1 h at room temperature. Mounting was done with ProLong^® Gold antifade reagent (Life Technologies). Images were acquired with an Axiovert 200 M microscope/EC Plan‐Neofluar ×40/1.3 oil immersion or C‐Apo ×63/1.20 water immersion objective equipped with a charge‐coupled‐device camera (all from Carl Zeiss MicroImaging GmbH) or an LSM 710/Axiobserver Z1 confocal microscope ×63/1.4 oil immersion objective operated by ZEN 2009 software. Images were further processed and analyzed using ImageJ/Fiji software version 1.46 (NIH; Schindelin et al, 2012). For morphometric analysis, images were processed using median and Mexican hat filtering (http://rsb.info.nih.gov/ij/plugins/mexican-hat/index.html). Mitochondria were then thresholded and binarized for subsequent analysis. Branching on skeletonized binary images was measured using the Analyse Skeleton plugin (Arganda‐Carreras et al, 2010).

Electron microscopy

Mice were perfused with electron microscopy fixation buffer (4% paraformaldehyde and 2% glutaraldehyde in 0.1 M sodium cacodylate, pH 7.4) and the kidneys post‐fixed in the same buffer for two weeks at 4°C. Samples were osmicated with 1% OsO4 in 0.1 M cacodylate and dehydrated in a graduated ethanol series. Infiltration with Epon and flat embedding was performed according to standard procedures. Thin (30 nm) cross sections were taken on an Ultracut UCT ultramicrotome (Reichert, Heidelberg, Germany). The sections were stained with 1% aqueous uranylic acetate and lead citrate and examined with a Zeiss EM 902 electron microscope (LEO, Oberkochen, Germany). Images were further processed with Adobe Photoshop CS4 version 11.0.0.0. (Adobe Systems, San Jose, CA, USA).

Cell culture

Conditionally immortalized podocytes were generated as previously described (Shankland et al, 2007) and cultured in RPMI media supplemented with 10% FBS and IFNγ. Mycoplasm contamination was excluded by PCR testing. Short hairpin RNAs (shRNAs) against PHB2 were designed based on the prediction of a publicly available prediction program (RNAi Designer; Invitrogen, Carlsbad, CA, USA). shRNAs were selected for efficient knockdown by using a Dual‐luciferase^® reporter assay (Promega, Madison, WI, USA). Podocytes containing either a doxycycline‐inducible scrambled shRNA (5′‐TGCTGAAATGTACTGCGCGTGGAGACGTTTTGGCCACTGACTGACGTCTCCACGCAGTACATTT‐3′) or two Phb2 shRNAs (5′‐TGCTGAGCTAAGTCCTTCAAGTTCTGGTTTTGGCCACTGACTGACCAGAACTTAGGACTTAGCT‐3′ and 5′‐TGCTGTAACAATGGACGGCAGCACTCGTTTTGGCCACTGACTGACGAGTGCTGGTCCATTGTTA‐3′) were generated by means of the T‐Rex^™ System by Invitrogen (Invitrogen). If not indicated otherwise, differentiation of podocytes was induced by culturing the cells at 37°C on Primaria plastic plates (BD Biosciences, San Jose, CA, USA) in the absence of IFNγ. After 10 days of differentiation, 2 μg/ml doxycycline was added to the medium and exchanged every 24 h. Experiments were performed after 96 h of doxycycline treatment. In our inhibition experiments, cells were treated either for 2 h with 10 ng/ml rapamycin (LC Laboratories) or for 24 h with 10 μM BMS 53692 (Tocris).

Autophagy experiments were carried out with undifferentiated podocytes at 33°C after treatment with 2 μg/ml for 96 h. Undifferentiated podocytes were used for this experiment because LC3 levels were below the detection limit in cultured differentiated podocytes. In this experiment, cells were treated with 10 ng/ml rapamycin for 24 h and 25 μM chloroquine for 2 h.

Statistical analysis

All results are expressed as means ± SEM. Statistical significance was evaluated using GraphPad Prism version 4.00c for Macintosh and OASIS (Online Application for the Survival Analysis of Lifespan Assays) (Yang et al, 2011). Unpaired Student's t‐test was applied in most experiments. A Log rank test (Bewick et al, 2004) was performed for survival curves, and a chi‐square test was applied for the subcellular localization of DAF‐16::GFP. A P‐value < 0.05 was considered significant.

For more detailed Materials and Methods, see Supplementary Information.