BASIC RESEARCH www.jasn.org

ADCK4 Deficiency Destabilizes the Coenzyme Q Complex, Which Is Rescued by 2,4-Dihydroxybenzoic Acid Treatment

Eugen Widmeier ,1,2 Seyoung Yu,3,4 Anish Nag,5 Youn Wook Chung ,6 Makiko Nakayama,1 Lucía Fernández-del-Río,5 Hannah Hugo,1 David Schapiro,1 Florian Buerger,1 Won-Il Choi,1 Martin Helmstädter,2 Jae-woo Kim,4,7 Ji-Hwan Ryu ,4,6 Min Goo Lee,3,4 Catherine F. Clarke,5 Friedhelm Hildebrandt,1 and Heon Yung Gee 3,4

Due to the number of contributing authors, the affiliations are listed at the end of this article.

ABSTRACT Background Mutations in ADCK4 (aarF domain containing kinase 4) generally manifest as steroid-resistant nephrotic syndrome and induce coenzyme Q10 (CoQ10)deficiency. However, the molecular mechanisms underlying steroid-resistant nephrotic syndrome resulting from ADCK4 mutations are not well under- stood, largely because the function of ADCK4 remains unknown. BASIC RESEARCH Methods To elucidate the ADCK4’s function in podocytes, we generated a podocyte-specific, Adck4-knockout mouse model and a human podocyte cell line featuring knockout of ADCK4. These knockout mice and podo- cytes were then treated with 2,4-dihydroxybenzoic acid (2,4-diHB), a CoQ10 precursor analogue, or with a vehicle only. We also performed proteomic mass spectrometry analysis to further elucidate ADCK4’sfunction. Results Absence of Adck4 in mouse podocytes caused FSGS and albuminuria, recapitulating features of nephrotic syndrome caused by ADCK4 mutations. In vitro studies revealed that ADCK4-knockout podo- cytes had significantly reduced CoQ10 concentration, respiratory chain activity, and mitochondrial poten- tial, and subsequently displayed an increase in the number of dysmorphic mitochondria. However, treatment of 3-month-old knockout mice or ADCK4-knockout cells with 2,4-diHB prevented the develop- ment of renal dysfunction and reversed mitochondrial dysfunction in podocytes. Moreover, ADCK4 inter- acted with mitochondrial proteins such as COQ5, as well as cytoplasmic proteins such as myosin and heat shock proteins. Thus, ADCK4 knockout decreased the COQ complex level, but overexpression of ADCK4 in ADCK4-knockout podocytes transfected with wild-type ADCK4 rescued the COQ5 level.

Conclusions Our study shows that ADCK4 is required for CoQ10 biosynthesis and mitochondrial function in podocytes, and suggests that ADCK4 in podocytes stabilizes proteins in complex Q in podocytes. Our study also suggests a potential treatment strategy for nephrotic syndrome resulting from ADCK4 mutations.

JASN 31: 1191–1211, 2020. doi: https://doi.org/10.1681/ASN.2019070756

Coenzyme Q (CoQ, ubiquinone)—a lipophilic com- Received July 30, 2019. Accepted February 22, 2020. ponent located in the inner mitochondrial mem- E.W. and S.Y. contributed equally to this work. brane, Golgi apparatus, and cell membranes—plays Published online ahead of print. Publication date available at 1 a pivotal role in oxidative phosphorylation. CoQ www.jasn.org. shuttles electrons from complexes I and II to complex Correspondence: Prof. Heon Yung Gee, Department of Pharma- 2 III in the mitochondrial respiratory chain. It also has cology, Yonsei University College of Medicine, Avison BioMedical a critical function in antioxidant defense because of Research Center Rm#225, 50-1 Yonsei-ro, Seodaemun-gu, Seoul 3 03722, Republic of Korea, or Dr. Friedhelm Hildebrandt, Boston its redox potential. The CoQ biosynthesis pathway Children’s Hospital, EN561, 300 Longwood Avenue, Boston, MA has been extensively studied in Saccharomyces cere- 02115. E-mail: [email protected] or friedhelm.hildebrandt@childrens. visiae.4 A minimum of 12 proteins, encoded by the harvard.edu Coq genes, form a complex that serves to stabilize Copyright © 2020 by the American Society of Nephrology

JASN 31: 1191–1211, 2020 ISSN : 1046-6673/3106-1191 1191 BASIC RESEARCH www.jasn.org each other and is involved in coenzyme synthesis.5 Based on Significance Statement protein homology, approximately 15 homologous COQ genes have been identified in humans.1 ADCK4 mutations generally manifest as steroid-resistant nephrotic fi Primary CoQ deficiencies due to mutations in ubiquinone syndrome, and cause coenzyme Q10 (CoQ10)de ciency. However, ’ biosynthetic genes (COQ2, COQ4, COQ6, COQ7, COQ9, ADCK4 s function remains obscure. Using mouse and cell models, the authors demonstrated that podocyte-specific Adck4 deletion in PDSS1, PDSS2, aarF domain containing kinase 3 [ADCK3], mice significantly reduced survival and caused severe FSGS, effects – and ADCK4) have been identified.6 12 Clinical manifestations that were prevented by treatment with 2,4-dihydroxybenzoic acid of CoQ10 deficiency vary depending on the genes involved, (2,4-diHB), a CoQ10 precursor analogue. ADCK4-knockout podo- fi and mutations in the same gene can result in diverse pheno- cytes exhibited a signi cantly reduced CoQ10 level and defects in types depending on the mutated allele.13 COQ2,11 COQ6,10 mitochondrial function that were rescued by 2,4-diHB treatment, thus these phenotypes were attributed to decreased CoQ levels. 9 12 10 PDSS2, and ADCK4 have also been implicated in steroid- The authors also found that ADCK4 interacted with mitochondrial resistant nephrotic syndrome (SRNS). Although no effec- proteins, including COQ5, and that ADCK4 knockout decreased tive therapy has been described for SRNS, supplementation of COQ complex levels. These findings reveal that ADCK4 is required for CoQ biosynthesis and mitochondrial function in podocytes, CoQ10 in cases of SRNS resulting from CoQ deficiency acts to 10 alleviate the associated clinical symptoms.10,12,14 This is par- and suggests a treatment strategy for nephrotic syndrome caused by ADCK4 mutations. tially true for ADCK4-related glomerulopathy and several cases have been reported accordingly.12,14 Recently, 2,4- dihydroxybenzoic acid (2,4-diHB) has been shown to amelio- METHODS rate disparate phenotypes in yeast, mouse, and Caenorhabditis elegans models harboring a mutation in the respective Mouse Breeding and Maintenance COQ7, Mclk1,orclk-1 genes.15–17 In these studies, 2,4- The animal experimental protocols were reviewed and ap- diHB was shown to act as an alternate or “bypass” ring proved by the Institutional Animal Care and Use Committee precursor, able to restore endogenous CoQ biosynthesis. In at the University of Michigan (#08619), Boston Children’s addition, 2,4-diHB prevents renal disease of podocyte- Hospital (#13-01-2283), and Yonsei University College of 2 2 specific Coq6 / mice.18 Medicine (#2015-0179). All mice were handled in accordance ADCK3 (also known as COQ8A) and ADCK4 are two with the Guidelines for the Care and Use of Laboratory Ani- mammalian orthologs of yeast Coq8p/Abc1, which belong mals. Mice were housed under pathogen-free conditions with to the microbial UbiB family; they appear to result from a light period from 7:00 AM to 7:00 PM and had ad libitum access gene duplication in vertebrates.6,19 UbiB and Coq8p are to water and irradiated rodent chow (catalog #0006972; required for CoQ biosynthesis in prokaryotes and yeast, re- LabDiet, St. Louis, MO). Targeted Adck4tm1a(EUCOMM)Hmgu spectively, and are speculated to activate an unknown mono- (Adck4tm1a) embryonic stem cells were obtained from oxygenase in the CoQ biosynthesis pathway.19 Coq8p, EUCOMM and injected into the blastocysts of mice. Chimeric ADCK3, and ADCK4 are present on the matrix side of the mice were bred with C57BL/6J mice to establish germline 1 1 inner mitochondrial membrane.20–22 Coq8p is essential for transmission. Nphs2.Cre (stock #008205) and Pgk1.Flpo the organization of high molecular mass Coq polypeptide (#011065) mice were obtained from Jackson Laboratory. complexes and for phosphorylated forms of the Coq3, Mouse lines were bred onto the C57BL/6J genetic background. Coq5, and Coq7 polypeptides that are involved in methyl- Genotyping was performed by PCR using the following pri- ation and hydroxylation steps in CoQ biosynthesis.21,23 Sim- mers: #1, GGATAGGGGGCTGGAGAGATG; #2, GCCCGC ilarly, it has been shown that ADCK3 interacts with CoQ CTCCCTGTATCTTAG; #3, TCGGAGAGGAAAGGACTG biosynthesis enzymes in a protein complex (complex Q).24 GAG; #4, CCCTTTCCCTTGAGTTCACAGC; and #5, TGG Moreover, ADCK3 lacks protein kinase activity in the trans CCTCAAACTCATGAAAATACTCC. Mice were maintained form; exhibits ATPase activity; and has highly conserved, un- in mixed sex and were randomly assigned to the different ex- orthodox protein kinase–like domains, including the KxGQ perimental groups. For mouse studies, experimental results motif, present in UbiB and eukaryotic COQ8 homologs, in- were validated over multiple litters, across several generations cluding ADCK4.25 of the mouse colony with n.6. Data collection of urine, whole However, it is not clear whether ADCK4 functions in a blood, and plasma analysis data were blinded to genotype so manner similar to that of ADCK3. Mutations in the ADCK4 that the operator did not know the genotype during perform- (also known as COQ8B) gene generally manifest as ing the measurements. Histologic and ultrastructural analysis adolescence-onset SRNS, sometimes accompanied with med- was not blinded to genotype but all samples were processed ullary nephrocalcinosis or extrarenal symptoms, including using the same protocol. seizures.12,26 The molecular mechanisms underlying SRNS resulting from ADCK4 mutations are not well understood, Supplementation of 2,4-diHB to the Mice in Drinking largely because the function of ADCK4 is unclear. Therefore, Water in this study, we investigated the function of ADCK4 using 2,4-diHB at a concentration of 25 mM was administered to mouse and cell models. the mice via drinking water and changed twice a week.

1192 JASN JASN 31: 1191–1211, 2020 www.jasn.org BASIC RESEARCH

The treatment was started at 3 months of age and continued eosin, Periodic acid–Schiff, Masson trichrome, SFOG, and up to 18 months of age. Jones silver stain following the standard protocols for histo- logic examination. Urine Analysis Urine was collected by housing the mice overnight (12 hours) Ultrastructural Analysis in metabolic cages. All samples were immediately frozen and The kidney tissues and cells were fixed in 2.5% glutaraldehyde, stored at 280°C. The samples were thawed on ice before urine 1.25% paraformaldehyde, and 0.03% picric acid in 0.1 M so- albumin and creatinine measurements. Urinary albumin was dium cacodylate buffer (pH 7.4) overnight at 4°C. They were measured using the Albumin Blue Fluorescent Assay Kit (Ac- then washed with 0.1 M phosphate buffer, postfixed with 1% tive Motif), as per the manufacturer’s instructions. Urine cre- osmium tetroxide dissolved in 0.1 M PBS for 2 hours, dehy- atinine was measured using the liquid chromatography with drated in ascending gradual series (50%‒100%) of ethanol, tandem mass spectrometry (LC-MS/MS) method as described and propylene oxide was used for infiltration. Samples were previously.27 Proteinuria was expressed as milligram of albu- embedded using the Poly/Bed 812 Kit (Polysciences) accord- min per milligram of creatinine. ing to manufacturer’s instructions. After pure fresh resin embedding and polymerization in a 65°C oven (TD-700; Whole Blood and Plasma Analysis DOSAKA, Kyoto, Japan) for 24 hours, sections of approxi- Blood was collected from mice via the facial vein bleeding mately 200–250 nm thickness were cut and stained with tolu- method and collected in citrate tubes. The blood sample was idine blue for light microscopy. Sections of 70 nm thickness subsequently analyzed using the Vetscan VS2 Chemistry An- were double stained with 6% uranyl acetate (22400; EMS) for alyzer, as per the manufacturer’s instructions. Plasma samples 20 minutes and lead citrate (Fisher) for 10 minutes for con- obtained by centrifugation of whole blood were immediately trast staining. The sections were cut using a Leica EM UC-7 frozen at 280°C. Plasma creatinine was measured using the with a diamond knife (Diatome) and transferred onto copper LC-MS/MS method as described previously.27 and nickel grids. All the sections were observed by transmis- sion electron microscopy (TEM; JEM-1011, JEOL, and Zeiss Immunoblotting and Immunofluorescence Staining 912) at an acceleration voltage of 80 kV. These experiments were performed as described previously.28 Anti-podocin (P0372) and anti-FLAG M2 (F3165; Sigma- Plasmids, Cell Culture, Transfection, and Lentivirus Aldrich); anti-nidogen (NBP1-97701; Novus); anti-nephrin Transduction (GP-N2; Progen); anti-synaptopodin (PA5-56997; Thermo- Single guide RNAs (sgRNAs) targeting human ADCK4 Fisher); anti-COQ3 (28051-1-AP), anti-COQ5 (17453-1- (sgRNA1, GCTGCACAATCCGCTCGGCAT; sgRNA2, GTA AP), and anti-COQ9 (14874-1-AP; Proteintech, Rosemont, AGGTCTGCACAATCCGCT; and sgRNA3, GACCTTATG IL); anti–a smooth muscle actin (anti-aSMA; #19245), TACAGTTCGAG) were cloned into BsmBI-digested lenti- anti–p-mammalian target of rapamycin (anti–p-mTOR; CRISPR v2 (plasmid #52961; Addgene). ADCK4 cDNA was #5536P), anti-mTOR (#2983P), anti–p-p38 (#4511S), cloned into the p3xFLAG CMV26 (C-terminal) vector (Sigma- anti–p-extracellular signal–regulated kinase (anti–p-ERK; Aldrich). Bacterial alkaline phosphatase (BAP) cDNA cloned #4377S), anti-pJNK (#9251S), anti-p38 (#9212S), and anti-LC3 into the p3xFLAG CMV7 vector was digested using Kpn1 and (#4108S; Cell Signaling Technology, Danvers, MA); anti-collagen EcoR1 restriction enzymes (New England BioLabs) and ligated IV (ab6586), anti-COXIV (ab33985), and anti-actin (ab49900; into the p3xFLAG CMV24 vector. Abcam, Cambridge, UK); and ADCK4 (LS-C119206; LSBio, Immortalized human podocytes29 were maintained in Seattle, WA) were purchased from the indicated commercial sour- RPMI1GlutaMAX-I (Gibco) supplemented with 10% FBS, ces. Alexa Fluor 488 Phalloidin and secondary antibodies were penicillin (50 IU/ml)/streptomycin (50 mg/ml), and insulin- purchased from Invitrogen. Fluorescent images were obtained transferrin-selenium-X. Human proximal tubule (HK-2) and using an SP53 or SP8-U-FLIM laser scanning microscopes HEK293 cells were maintained in DMEM supplemented with (Leica) using excitation wavelengths of 405, 488, and 594 nm 10% FBS and 1% penicillin/streptomycin. Plasmids were and 1003/633 oil immersion objectives (HCX PL APO CS transfected into podocytes or HEK293 cells using Lipofecta- 1003/1.44 OIL and HC PL APO CS2 633/1.40 OIL) or an mine 2000 (Invitrogen). HEK293 cells stably expressing LSM 700 microscope (Carl Zeiss) using excitation wavelengths p3xFLAG-ADCK4 or BAP were selected and maintained of 405, 488, 555, and 639 nm and a 403 water immersion with 1 mg/ml G418. objective (C-Apochromat, 1.2 numerical aperture, Zeiss). Im- To establish ADCK4 knockout (KO) cells, lentiCRISPR v2, ages were processed and analyzed using Leica AF, Zeiss LSM pMD2.G, and psPAX2 were transfected into Lenti-X software, ImageJ, and Adobe Photoshop CS6 software. 298T cells (Clontech). Supernatants containing lentivirus were collected 48 hours after transfection and passed through Histologic Analysis a0.2mM filter. Cultured podocytes and HK-2 cells were trans- The kidney tissues were fixed in 4% paraformaldehyde, sec- duced with lentivirus, selected, and maintained with 4 mg/ml tioned (5 mm thickness), and stained with hematoxylin and puromycin.

JASN 31: 1191–1211, 2020 ADCK4 Stabilizes Coenzyme Q Complex 1193 rsiepoenrain proteinuria gressive P nml e ru) otdln ipasteosto ea alr.Nt htoc hoi ea alr nus rnr lui excretion albumin urinary ensues, failure renal chronic once that Note failure. SRNS. renal of in onset observed the as displays reduced line Dotted group). per animals 1194 1. Figure einsria eido 1 asadhzr ai f1.2cmae ihta fltemt otos(o-ak[Mantel (log-rank controls littermate of that with compared 17.52 of ratio hazard and days 316 of period survival median a 5 AI RESEARCH BASIC CD A .01 aadrto[o ak) B rnr lui-raiiertosra nlssa niae gsadgntpsrvae pro- revealed genotypes and ages indicated at analysis serial ratio albumin-creatinine Urinary (B) rank]). [log ratio hazard 0.0001; PAS H&E Percent survival (%) Control (10mo) 100 10 20 30 40 50 60 70 80 90 Nphs2.Cre 0 JASN 01 213 12 11 10 9 8 Nphs2.Cre Control (n 1 www.jasn.org ;Adck4 Nphs2.Cre =17 + ;Adck4 ) Adck4 flox/flox Nphs2.Cre P (10 mo) auswr acltduiga unpaired an using calculated were values flox/flox 1 p =0.0001 Month ;Adck4 flox/flox iedvlpdFG.(A) FSGS. developed mice (n + =15 ; 41 61 18 17 16 15 14 flox/flox )

Masson’s Trichrome Jone’s Silver uatmc rdhxgn,btnti itraecnrl bakdaod ( diamond) (black controls littermate in not but hexagon), (red mice mutant Control (10mo) Nphs2.Cre B Albumin/creatinin ratio (mg/mg) 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 0 5 Adck4 Nphs2.Cre Control (n 357910111213234567 1 t sn ns ns ns (10 mo) et * test: ;Adck4 flox/flox P =17 + flox/flox , * *** *** ; .5 ** 0.05, ) **** * **** **** uatmc xiie eue iesa with span life reduced exhibited mice mutant Nphs2.Cre 8 Glomerular sclerosis (%) **** **** **** **** P 100 10 50 60 70 80 90 , 0 2 4 6 8 Month 0.01, Control (n f7 f9 1of85 3of93 1 of74 ** + ;Adck4 ***P , =3 flox/flox *** 0.001, ) 41 61 18 17 16 15 14 ** ** ** ** (n **** JASN 105 of Nphs2.Cre Adck4 =15 108 ****P ) 31: 101 of flox/flox 104 , 1191 + .01 each 0.0001; ; (n 73 of 76 – – =4 o]test, Cox] n 21 2020 1211, 5 ) 74 of 79 15 – 17 www.jasn.org BASIC RESEARCH

Cell Viability Assay linearly from 7.2 pmol to 400 pmol (to obtain a typical stan- Cell viability assays were performed using the Cell Counting dard curve for CoQ quantification in the cell pellets). The sam- Kit-8 (Dong-In Biotech). Cell suspensions (100 ml; 13105/ml) ples were vortexed in 2 ml of methanol for 30 seconds, followed with culture medium were added to a 96-well plate and incu- by addition of 2 ml petroleum ether. After vortexing for an bated for 24 hours in a carbon dioxide incubator. The medium additional 30 seconds, the organic upper layer was transferred was replaced with phenol-free fresh medium with or without to a new tube. Another 2 ml of petroleum ether was added to 30 mM AA and/or 500 mM 2,4-diHB for 15 hours. Four repli- the original methanol layer and samples were vortexed again cate wells were included for each condition. CCK-8 reagent for 30 seconds. The organic phase was removed and the com- (10 ml) was added to each well and incubated for 1 hour; OD bined organic phase was dried under a stream of nitrogen gas. of the sample at 450 nm was measured. The samples were resuspended in 200 ml of ethanol containing 1 mg/ml benzoquinone to oxidize all of the lipids. HPLC- Isolation of Glomeruli MS/MS analysis of the samples was performed using a 4000 Mice were anesthetized with isoflurane. The kidneys were per- QTRAP linear MS/MS spectrometer from Applied Biosystems fused with Dynabeads (M450-epoxy, 4.5 mm; Invitrogen). Dy- (Foster City, CA). Applied Biosystem software, Analyst version nabeads were washed with ice-cold PBS containing 0.1% BSA 1.4.2, was used for data acquisition and processing. Chromato- before use, according to the manufacturer’s instructions. graphic separation was achieved through a reverse phase Luna Then, 20 ml of Dynabeads in HBSS at a concentration of 5 mM PFP(2) column (Phenomenex) with a mobile phase 83107 beads/ml was injected into the left cardiac ventricle. comprising 90% solvent A (95:5 mixture of methanol/isopro- Following perfusion, the kidneys were removed, minced, and panol containing 2.5 mM ammonium formate) and 10% sol- digested with 1 mg/ml collagenase A and 0.1 mg/ml deoxyri- vent B (isopropanol containing 2.5 mM ammonium formate) bonuclease type I at 37°C for 30 minutes with gentle agitation. at a constant flow rate of 1 ml/min. All samples were analyzed The digested tissue was then gently passed in series through in multiple-reaction-monitoring mode. The precursor and 180-,90-,and75-mm sieves, followed by intermittent rinsing product ion transitions monitored were as follows: mass/ with ice-cold sterile HBSS. To eliminate adherence of the glo- charge (m/z), 795.6/197.08 for CoQ9; m/z, 812.6/197.08 for meruli, each sieve was rinsed with HBSS containing 1% BSA CoQ9 with ammonium adduct; m/z, 863.6/197.08 for CoQ10; before use. The mixture passed through the sieves was then m/z, 880.6/197.08 for CoQ10 with ammonium adduct; m/z, 3 centrifuged at 200 g for 5 minutes. The supernatant was 919.7/253.1 for dipropoxy-CoQ10; and m/z, 936.7/253.1 for discarded and the pellet was dissolved in 2 ml of ice-cold HBSS dipropoxy-CoQ10 with ammonium adduct. The method and then transferred into a 1.7 ml tube. Glomeruli that con- used to quantify the signals corresponding to CoQ9 and 31 tained Dynabeads were isolated using a magnetic particle con- CoQ10 was as described. centrator and washed at least three times with ice-cold sterile HBSS. The entire procedure was performed on ice, except for Mitochondrial Respiratory Enzyme Activity the collagenase digestion, which was performed at 37°C. Measurement Cell lysates (15–50 mg) were diluted in phosphate buffer Cellular Lipid Extraction and CoQ Quantification via (50 mM monopotassium phosphate, pH 7.5) and subjected HPLC-MS/MS to spectrophotometric analysis for isolated respiratory Cells (approximately 0.1 g) and isolated glomeruli were chain complex activities at 37°C using a spectrophotometer thawed on ice and resuspended in 1.5 ml of PBS (0.14 M so- (PerkinElmer). Complex II activity was measured at 600 nm 2 2 dium chloride, 12.0 mM monosodium phosphate, and («519.1 mmol 1cm 1) after the addition of 20 mM succinate, 8.1 mM disodium phosphate; pH 7.4), followed by homoge- 80 mM dichlorophenolindophenol, 300 mM potassium cya- nization using a polytron (PT 2500E; Kinematica) for 1 minute nide, and 50 mM decylubiquinone. Complex II activity was at 10,000 rpm on ice. Lipid extracts were prepared as previ- defined as the flux difference with or without 10 mM malonate. ously described with minor modifications.30 Briefly, dipropoxy- Complex II and III activity was also determined at 550 nm 21 21 CoQ10 was used as the internal standard and was added at a («518.5 mmol cm ) in the presence of 10 mM succinate, constant volume to all of the cell pellets as well as to a set of five 50 mM cytochrome c, and 300 mM potassium cyanide. Com- CoQ9 and CoQ10 standards of known concentrations ranging plex II-III activity was defined as the flux difference before and

data point represents the mean value of technical duplicates; the error bars represent SEM. (C) Kidney serial sections and represen- 1 tative images of 10-month-old mice. The Nphs2.Cre ;Adck4flox/flox mutant mice exhibited severe FSGS (arrows) with severe interstitial fibrosis and tubular atrophy (arrow heads). In contrast, wild-type littermate control mice displayed normal histologic kidney mor- phology. Scale bars, upper row 500 mm, middle row 100 mm, and lower row 20 mm. (D) Glomerular sclerosis by analyzing Masson trichrome staining. n53–4 mice at 10 months of age in each group; each graph bar indicates a single animal and .70 glomeruli were counted per an animal; P values were calculated using an unpaired t test; ****P,0.0001. H&E, hematoxylin and eosin; PAS, Periodic acid–Schiff.

JASN 31: 1191–1211, 2020 ADCK4 Stabilizes Coenzyme Q Complex 1195 1196 2. Figure ooi gen n h aeetmmrn akrndgn(e) omlepeso atr fpdcnwsosre n10- in observed was podocin of pattern expression normal A (red). nidogen marker mice. membrane control basement littermate month-old the and (green) podocin AI RESEARCH BASIC C

Nphs2.Cre + flox/flox Nphs2.Cre ;Adck4 A JASN Control (10 month) (10 month) Nphs2.Cre+;Adck4 flox/flox Control (10 month) (10 month) B 1 www.jasn.org ;Adck4 Nphs2.Cre+; flox/flox Podocin 890x Adck4 Control flox/flox Nphs2.Cre Podocin ieehbtdgoeuoah.()Immuno (A) glomerulopathy. exhibited mice 1 ;Adck4 2900x Nidogen 3D surface plot flox/flox iesoe otyoitdpdcnsann arw) pern nyon only appearing (arrows), staining podocin omitted mostly showed mice 4800x Mean (FIU) 55 10 15 20 25 30 35 40 45 50 fl eg detail merge 0 5 oecnesann o h ltdahamprotein diaphragm slit the for staining uorescence Control D Filtration slit frequency per micron GBM 0 1 2 3 4 5 *** Adck4 Nphs2.Cre Control 30000x flox/flox **** Adck4 Nphs2.Cre JASN + ; 31: flox/flox 1191 + ; – 21 2020 1211, www.jasn.org BASIC RESEARCH after the addition of 10 mM thenoyltrifluoroacetone. All chem- contains all assembled proteins, with peptide sequences, pep- icals were obtained from Sigma-Aldrich. tide spectrum match counts, and the iTRAQ tag–based quan- tification ratio. The iTRAQ tag–based quantification was used Determination of Reactive Oxygen Species to determine the relative abundance of proteins identified in Intracellular reactive oxygen species (ROS) production was the iTRAQ data set. One-way ANOVA was used to identify measured using the CellROX Deep Red reagent (Invitrogen) proteins differentially expressed between control and following the manufacturer’s instructions. Control and ADCK4 KO podocytes, corrected using the Bonferroni mul- ADCK4 KO podocytes were seeded at 13104 cells/well in a tiple comparison test. Quantitative ratios were log2 normal- 96-well plate and incubated for 24 hours. The CellROX re- ized for final quantitative testing. The relative abundance of agent was added to the medium at a final concentration of the proteins was varied 1.5-fold between control and ADCK4 5 mM, and the plate was then incubated for 30 minutes. After KO podocytes (n54) or between AA-treated control and AA- washing twice with HBSS containing calcium and magnesium, treated ADCK4 KO podocytes (n52). The raw data files of the the cells were treated with hydrogen peroxide (H2O2;100or iTRAQ data set were submitted to the ProteomeXchange Con- 300 mM), tert-butyl hydroperoxide (tBHP; 30 or 100 mM), or AA sortium via the PRIDE partner repository6 with the data set (30, 100, or 500 mM) in HBSS. The fluorescence at 640/665 nm identifier PXD016725. was measured using a Cytation 5 cell imaging multimode reader (Biotek, Winooski, VT). Mitochondrial ROS generation Identification of ADCK4 Interactors was assessed using the MitoSOX Red reagent (Invitrogen), a red Proteins (75 mg) from HEK293 cells stably expressing mitochondrial superoxide indicator. The MitoSOX reagent was p3XFLAG-ADCK4 or -BAP were incubated with 80 mlof added at final concentration of 5 mM, and the cells were in- FLAG M2 agarose beads (Sigma-Aldrich) for 48 hours at 4°C cubated for 15 minutes and then washed twice with HBSS. in an orbital shaker. The agarose beads were washed four Subsequently, the cells were treated with 500 mMH2O2, times with lysis buffer to restrict nonspecific binding. Subse- 100 mM tBHP, 30 mM AA, or 100 mM AA. The fluorescence quently, 200 ml of elution buffer containing 150 ng/ml 3xFLAG at 510/580 nm was measured using Cytation 5. peptide was added and the samples were incubated overnight. The eluates were analyzed by immunoblotting, Coomassie Isobaric Tag Labeling for Relative and Absolute blue staining, and silver staining. The eluates were digested Quantification and subjected to nanoflow LC-MS/MS analysis. Peptides were Isobaric tag labeling for relative and absolute quantification separated on a C18 precolumn (75 mm32 cm, nanoViper, (iTRAQ) was performed by Poochon Scientificasdescribed Acclaim PepMap100; Thermo Fisher Scientific) and analytic previously.32 Proteins (100 mg) were extracted from control C18 column (75 mm350 cm, PepMap RSLC; Thermo Fisher and ADCK4 KO podocytes, digested with trypsin, and labeled Scientific). Peptides were analyzed using an LC-MS/MS sys- using the 8-plex iTRAQ Labeling Kit (AB Sciex). Fraction- tem consisting of Easy nLC 1000 (Thermo Fisher Scientific) ation of the iTRAQ–multiplex-labeled peptide mixture was and an Orbitrap Fusion Lumos mass spectrometer (Thermo carried out using an Agilent AdvanceBio column (2.7 mm, Fisher Scientific) equipped with a nanoelectrospray source. 2.13250 mm) on an Agilent UHPLC 1290f system (Agilent, MS/MS spectra were analyzed against the Uniprot human Santa Clara, CA). The HPLC-MS/MS analysis was performed database using the following software analysis protocol. The using a Thermo ScientificQ-Exactivehybridquadrupole- reversed sequences of all proteins were appended into the da- orbitrap mass spectrometer and a Dionex UltiMate 3000 tabase for the calculation of false discovery rate. ProLucid was RSLCnano system (Thermo Fisher Scientific, San Jose, CA). used to identify the peptides, with a precursor mass error of MS raw data files were searched against human protein se- 5 ppm and a fragment ion mass error of 200 ppm.33 The out- quence databases obtained from the National Center for Bio- put data files were filtered and sorted to compose the protein technology Information website using Proteome Discoverer list using DTASelect (The Scripps Research Institute, La Jolla, 1.4 software (Thermo Fisher Scientific) based on the SE- CA), with two and more peptide assignments for a protein QUEST and percolator algorithms. The false discovery rate identification and a false positive rate of ,0.01.34 The MS pro- was set to 1%. The resulting Proteome Discoverer report teomics data were deposited to the ProteomeXchange

a few capillary loops (arrowhead). (B) Quantification of antibody staining in (A) demonstrated a significantly decreased expression of 1 podocin in glomeruli of Nphs2.Cre ;Adck4flox/flox mice compared with glomeruli of control mice. (C) TEM representative images of 1 mice at the age of 10 months. Nphs2.Cre ;Adck4flox/flox mice revealed severe podocyte foot process effacement (arrows) and an increased amount of dysmorphic mitochondria (*). Glomerular basement membrane (GBM) is highlighted by a dotted line. Scale bars, 10 mm left panel, 2 mm middle and lower right panels, 1 mm upper right panel, and 250 nm high magnification panel. (D) 1 Nphs2.Cre ;Adck4flox/flox mutant mice showed significant loss of filtration slits per micron of glomerular basement membrane. n53–4 mice in each group, two to three glomeruli per animal were analyzed. P values were calculated using an unpaired t test; ***P,0.001, ****P,0.0001, error bars represent SD. FIU, fluorescence intensity units.

JASN 31: 1191–1211, 2020 ADCK4 Stabilizes Coenzyme Q Complex 1197 BASIC RESEARCH www.jasn.org

A B 80 ****ns ***** ** *** ns **** ** ** *** *** 75 100 70 65 90 60 80 55 50 70 p = 0.797 45 60 40 35 50 30 25 40 20 30 15 survival (%) Percent 20 Albumin/creatinine ratio (mg/mg) ratio Albumin/creatinine 10 Control + 25 mM 2,4-diHB (n=11) 5 + flox/flox 10 Nphs2.Cre ;Adck4 + 25 mM 2,4-diHB (n=9) 0 0 3456789101112131415161718 8 9 101112131415161718 Month Month Control + 25 mM 2,4-diHB (n=11) Nphs2.Cre+;Adck4flox/flox + 25mM 2,4-diHB (n=9)

C Nphs2.Cre+; Nphs2.Cre+; D Adck4flox/flox Adck4flox/flox 100 ns + 25 mM 2,4-diHB + 25 mM 2,4-diHB Control (18 mo) (18 mo) Control (18 mo) (18 mo) 80

60

40

20 Glomerular sclerosis (%)

0 7 of 69 15 of 115 15 of 99 13 of 111 15 of 117 16 of 81

H&E Control + 25 mM 2,4-diHB (n=3) + flox/flox

Jone’s Silver Jone’s Nphs2.Cre ;Adck4 + 25mM 2,4-diHB (n=3) PAS Masson’s Trichrome Masson’s

1 Figure 3. Treatment of Nphs2.Cre ;Adck4flox/flox mutant mice with 2,4-diHB prevented FSGS progression, resulting in normal sur- vival rate. (A) Urinary albumin-creatinine ratio serial analysis at indicated ages and genotypes (n59–11 animals per group). 1 Nphs2.Cre ;Adck4flox/flox mutant mice treated with 2,4-diHB (green square) were protected from developing severe, progressive proteinuria, although proteinuria was significantly increased compared with that in healthy treated littermate controls (black circle). Green arrow indicates the start of treatment. Each data point represents the mean value of technical duplicates; the error bars repre- 1 sent SEM. (B) Nphs2.Cre ;Adck4flox/flox mutant mice treated with 2,4-diHB presented similar survival rate as that of healthy treated littermate controls (log-rank [Mantel–Cox] test, P50.797). (C) Kidney serial sections and representative images of 18-month-old mice.

1198 JASN JASN 31: 1191–1211, 2020 www.jasn.org BASIC RESEARCH

Consortium via the PRIDE partner repository with the data set mice compared with littermate controls. Necropsy of 10-month- D identifier PXD016147. old Adck4 Podocyte mice revealed pale and significantly small kid- neys compared with those in littermate controls (Supplemental Gene Ontology Analysis Figure 1C), indicating that podocyte-specificdeletionofAdck4 D ADCK4 interactors were analyzed using the Database for An- causes structural and functional kidney defects in Adck4 Podocyte D notation, Visualization, and Integrated Discovery (DAVID) mice. To examine the renal function of Adck4 Podocyte mice, we for functional annotation. The functional annotation tool in performed serial urine and plasma analyses for 18 consecutive D the online version of DAVID (version 6.8) was run (http:// months (Supplemental Figure 1D). Adck4 Podocyte mice dis- david.abcc.ncifcrf.gov/) using the default parameters and played the first significant decrease in plasma albumin level Gene Ontology (GO) categories representing molecular func- at 5 months of age (Supplemental Figure 1E) and the increase tion, cellular component, and biologic process were separately in albumin-creatinine ratio (18.81-fold, P50.0005) remained analyzed for enrichment. A P value of ,0.05 was considered significant throughout the study period compared with those significant. in littermate controls (Figure 1B). The increase over time in D albuminuria was the maximum, up to 31.2-fold, in Adck4 Podocyte Statistical Analyses mice compared with that in littermate controls (Figure 1B). D Statistical analyses were performed using Graph Pad Prism The onset of kidney function decline in Adck4 Podocyte mice 7 software. The results are presented as mean6SE or SD for was associated with a significant increase in plasma creatinine the indicated number of experiments. Statistical analysis of and plasma BUN levels at 7 months of age, progressing to CKD, continuous data was performed with the two-tailed t test or followed by renal failure, and consequently death (Figure 1A, multiple comparison analysis, as appropriate. Specifictests Supplemental Figure 1, F–H). Histologic analysis of kidneys D performed in the experiments are indicated in the figure leg- from Adck4 Podocyte mice at 10 months of age demonstrated ends. The results with P,0.05 were considered statistically severe global and FSGS with extensive interstitial fibrosis and significant. tubular atrophy (Figure 1C). To characterize the glomerular D phenotype of Adck4 Podocyte mice, we quantified the number D of sclerotic glomeruli in Adck4 Podocyte mice at 10 months of age D RESULTS and found that Adck4 Podocyte mice had a significantly increased number of sclerotic glomeruli (mean 96.03%) compared with Podocyte-Specific Adck4 KO Mice Developed that in littermate controls (Figure 1D). To characterize the mo- D Progressive Proteinuria, Severe FSGS, and Increased lecular abnormalities in the glomeruli of Adck4 Podocyte mice, Adult Mortality we analyzed the expression pattern of the slit diaphragm pro- To evaluate the role of Adck4 in kidney function, we generated teins podocin (Figure 2A) and nephrin (Supplemental a transgenic Adck4 (Adck4tm1a) mouse line, using embryonic Figure 4A), basement membrane marker nidogen (Figure 2A, stem cells obtained from EUCOMM (Supplemental Figure 1A). Supplemental Figure 4B), and primary process marker synap- Efficient targeting of the Adck4 gene was confirmed by geno- topodin (Supplemental Figure 4B) in the kidneys of 10-month- typing (Supplemental Figure 1B). Whole body loss of Adck4 in old mice. Staining of podocyte markers was significantly D Adck4tm1a mice proved lethal, which is consistent with the re- reduced in the glomeruli of Adck4 Podocyte mice compared port of the International Mouse Phenotyping Consor- with that of the control glomeruli (Figure 2B, Supplemental tium (www.mousephenotype.org). To circumvent Adck4tm1a Figure 4, D and F), demonstrating that Adck4 function is embryonic lethality, we generated podocyte-specific Adck4 required for podocyte maintenance and homeostasis. In addi- 1 fl fl KO mice Adck4tm1d or Nphs2.Cre ;Adck4 ox/ ox (hereafter re- tion, we analyzed the expression of the fibrotic markers colla- D 1 ferred to as Adck4 Podocyte) by crossing the Nphs2-Cre mouse gen IV and aSMA (Supplemental Figure 4, A and C) in the fl fl D with the Adck4 ox/ ox mouse in which two loxP sites surround kidneys of Adck4 Podocyte mice. Indeed, the kidneys of D D exons 5 and 6 in the Adck4 gene. Although young Adck4 Podocyte Adck4 Podocyte mice presented significantly increased expres- mice appeared grossly normal, including their kidneys sion of collagen IV and aSMA (Supplemental Figure 4, E and (Supplemental Figure 2), increased morbidity (hunched posture G) in the glomeruli, characteristic of glomerular fibrosis. To D and seedy fur) (Supplemental Figure 1C), a significantly in- study the structural changes in the glomeruli of Adck4 Podocyte creased mortality (Figure 1A), and weight loss (Supplemental mice at the ultrastructural level, we performed TEM using D D Figure 3) were observed in older (.9monthsold)Adck4 Podocyte the kidney of 10-month-old Adck4 Podocyte mice. The results

1 The wild-type littermate control mice and Nphs2.Cre ;Adck4flox/flox mutant mice treated with 2,4-diHB displayed normal histologic kidney morphology. Scale bars, upper row 500 mm, middle row 100 mm, and lower row 20 mm. (D) Glomerular sclerosis by analyzing Masson trichrome staining. n53–4 mice at 18 months of age in each group; each graph bar indicates a single animal and .69 glomeruli were counted per an animal. P values were calculated using an unpaired t test; *P,0.05, **P,0.01, ***P,0.001. H&E, hematoxylin and eosin; PAS, Periodic acid–Schiff.

JASN 31: 1191–1211, 2020 ADCK4 Stabilizes Coenzyme Q Complex 1199 BASIC RESEARCH www.jasn.org

A Control Nphs2.Cre+;Adck4flox/flox

Control + 25 mM 2,4-diHB Nphs2.Cre+;Adck4flox/flox + 25 mM 2,4-diHB

B Nephrin 3D surface plot C SMA 3D surface plot Nphs2.Cre+; Nphs2.Cre+; Nphs2.Cre+; Control Adck4flox/flox Nphs2.Cre+; Control Adck4flox/flox Control Adck4flox/flox + 25 mM 2,4-diHB + 25 mM 2,4-diHB Control Adck4flox/flox + 25 mM 2,4-diHB + 25 mM 2,4-diHB

25 ns 15 ** ns ns **** * 20 *** **** 10 15

10 Mean (FIU)

Mean (FIU) 5 5

0 0 Control Control + 25mM 2,4-diHB Control Control + 25mM 2,4-diHB + flox/flox + flox/flox Nphs2.Cre ;Adck4 Nphs2.Cre ;Adck4 + 25mM 2,4-diHB Nphs2.Cre+;Adck4flox/flox Nphs2.Cre+;Adck4flox/flox+ 25mM 2,4-diHB

D 890x 2900x 4800x E 5 ****

4 Control (18 month)

+ 25 mM 2,4-diHB 3

2

per micron of GBM 1 flox/flox Filtration slit frequency Filtration

0 ; Adck4 + Nphs2.Cre+;”

(18 month) Control + flox/flox 25 mM 2,4-diHB Adck4 + + 25 mM 2,4-diHB 25 mM 2,4-diHB Nphs2 . Cre

1 Figure 4. Treatment with 2,4-diHB prevented loss of podocytes in Nphs2.Cre ;Adck4flox/flox mice. (A) Immunofluorescence staining of 18-month-old mice for the slit diaphragm protein nephrin (green) and the glomerular fibrosis marker aSMA (red). Scale bars, 20 mm. (B 1 and C) Quantification of antibody staining in (A). The (B) nephrin and (C) aSMA expression of Nphs2.Cre ;Adck4flox/flox mice treated with 2,4-diHB was comparable with that of treated littermate control mice. n53 images in each group were analyzed, P values were

1200 JASN JASN 31: 1191–1211, 2020 www.jasn.org BASIC RESEARCH

D revealed the abnormal structure of glomeruli, severe foot pro- the ultrastructural level in the glomeruli of Adck4 Podocyte mice cess effacement, and disturbed podocyte morphology in (Figure 4D) by preserving normal slit morphology. However, the D Adck4 Podocyte mice (Figure 2C). The number of filtration-slit filtration-slit frequency was found to decrease compared with units per micrometer of basement membrane was significantly that of littermate controls (Figure 4E). In summary, the treat- D D reduced in Adck4 Podocyte mice compared with that in wild-type ment of Adck4 Podocyte mice with 2,4-diHB significantly D mice (Figure 2D). In addition, the podocytes of Adck4 Podocyte prevented the development of FSGS and foot process efface- mice appeared to contain abnormal mitochondria character- ment, while maintaining normal renal function in treated ized by hyperproliferation and increased size as mice aged mice at 18 months of age. Therefore, our findings demonstrate (Supplemental Figure 5). Overall, the glomerular phenotype that 2,4-diHB is effective in protecting against renal disease D of podocyte-specific Adck4 KO mice recapitulates the pathol- progression and in improving survival in Adck4 Podocyte mice. ogy of FSGS in humans resulting from ADCK4 mutations. Loss of ADCK4 Caused Mitochondrial Defects in Treatment with 2,4-diHB Prevented the Development Podocytes of Renal Pathology in Podocyte-Specific Adck4 KO Mice To investigate the function of ADCK4 at the cellular level, we Given that albuminuria started at around 4 months of age and generated ADCK4 KO cells of human podocytes and HK-2 renal structural abnormalities and functional decline manifes- cells, which originated from human proximal tubule epithelial D ted relatively late in Adck4 Podocyte mice, we decided to initiate cells. Each cell line was subjected to deletion of exon 6 of the the treatment with 2,4-diHB at a concentration of 25 mM in ADCK4 gene using the CRISPR/Cas9 system, and the absence drinking water when the mice were 3 months old, to prevent of ADCK4 expression was confirmed by immunoblotting the disease onset and to mitigate disease progression. Supple- (Supplemental Figure 10, A–D). KO of ADCK4 did not affect mentation of 2,4-diHB did not have any effect on the survival the viability of both cell lines (Supplemental Figure 10E). We rate, albumin-creatinine ratio, and kidney function and his- have previously shown that the level of CoQ10 decreased in tology of the control mice (Figure 3, Supplemental Figure 6). fibroblasts and lymphoblasts derived from patients with D Adck4 Podocyte mice treated with 2,4-diHB showed a normal ADCK4 mutations,12 but not in podocytes. Therefore, in survival rate despite maintaining proteinuria compared with this study, we verified this finding using the established KO that of healthy treated littermate controls (Figure 3, A and B) cells and found that ADCK4 KO resulted in decreased CoQ9 and a significantly improved survival rate (P50.0078) com- in both cultured podocytes and HK-2 cells compared with DPodocyte pared with that of nontreated Adck4 mice, which dis- that in control cells (Figure 5A). However, CoQ10 was reduced played an increased mortality rate progressing to ESKD with a only in cultured podocytes, but not in HK-2 cells (Figure 5B). median survival period of 316 days and hazard ratio of 9.36 The basal CoQ10 level in podocytes was threefold higher (Supplemental Figure 6B). The mortality rate reduction in than that in HK-2 cells (Figure 5B). In addition, we measured D Adck4 Podocyte mice treated with 2,4-diHB was associated the CoQ content of the glomeruli isolated from control and D with significantly improved plasma albumin level and renal Adck4 Podocyte mice and found a decreasing trend for both DPodocyte function (Supplemental Figure 6, B–E). This revealed normal CoQ9 and CoQ10 in Adck4 mice; however, this trend glomerular histology and a significantly reduced rate of sclerotic was not statistically significant (Supplemental Figure 11). Be- glomeruli (mean 14.76%) up to 18 months of age in treated cause CoQ shuttles electrons from complexes I and II to com- D Adck4 Podocyte mice (Figure 3, C and D, Supplemental Figure 7, plex III in the mitochondrial respiratory chain,2 the activity A and B). The improvement in functional, histologic, and ultra- of complex II-III is dependent on the CoQ10 level in the mi- D structural findings in Adck4 Podocyte mice treated with 2,4-diHB tochondria. Therefore, we measured the activity of complexes was also associated with significantly improved expression of II and II-III and found that the activity of both was signifi- nephrin (Figure 4, A and B) and synaptopodin (Supplemental cantly reduced in ADCK4 KO podocytes (Figure 5C) com- D Figures 8 and 9). Moreover, Adck4 Podocyte mice treated with pared with that in control cells, but not in ADCK4 KO HK-2 2,4-diHB showed significantly reduced expression of the fibrotic cells (Figure 5D). Decreased complex II-III activity observed marker aSMA (Figure 4C). Treatment with 2,4-diHB served to in ADCK4 podocytes was partially rescued by the addition of maintain the normal podocyte morphology and configuration at 2,4-diHB to culture media (Figure 5E). The reduced form

calculated using ordinary one-way ANOVA/Tukey multiple comparisons test: *P,0.05, **P,0.01, ***P,0.001, ****P,0.0001, error 1 bars represent SD. (D) Representative TEM images of mice at the age of 18 months. Nphs2.Cre ;Adck4flox/flox mice treated with 2,4- diHB displayed mild foot process morphology changes with infrequent regions of effacement (arrows). The mitochondrial morphology 1 remained normal. Scale bars, 10 mm left panel and 2 mm for middle and right panels. (E) Nphs2.Cre ;Adck4flox/flox mice treated with 2,4-diHB at the age of 18 months displayed reduced frequency of filtration slits per micrometer of glomerular basement membrane (GBM) compared with littermate controls under treatment; however, they still had significant remaining filtration slits. n53–4micein each group, two to three glomeruli per animal were analyzed; P values were calculated using an unpaired t test; *P,0.05, ****P,0.0001, error bars represent SD. FIU, fluorescence intensity units.

JASN 31: 1191–1211, 2020 ADCK4 Stabilizes Coenzyme Q Complex 1201 BASIC RESEARCH www.jasn.org

A Coenzyme Q9 B Coenzyme Q10 60 5000

4000 40 3000

2000

pmol/mg protein * pmol/mg protein 20

9 *

* 10 * * 1000 CoQ CoQ 0 0

Control Control Control Control

ADCK4 KO#1ADCK4 KO#2 ADCK4 KO#1ADCK4 KO#2 ADCK4 KO#1ADCK4 KO#2 ADCK4 KO#1ADCK4 KO#2 Podocytes HK-2 Podocytes HK-2

C Cultured podocytes D HK-2 150 150

100 100 * * ** * ** ** 50 50 * Enzyme Activity * Enzyme Activity (normalized to control) (normalized to control)

0 0

Control Control Control Control

ADCK4 ADCK4KO#1 KO#2 ADCK4 ADCK4KO#1 KO#2 ADCK4ADCK4 KO#1 KO#2 ADCK4 ADCK4KO#1 KO#2 Control + TTFA Control + TTFA

Control + Antimycin A Control + Antimycin A Complex II Complex II-III Complex II Complex II-III

E Cultured podocytes F Cultured podocytes 150 150 ** ** ** 100 100 ** **

50 50 Complex II - III Enzyme activity (normalized to control) Normalized to control (%) 0 0 Control ++- - - - Control + +--++-- ADCK4 KO#1 --++-- ADCK4 KO#1 - -++--++ ADCK4 KO#2 ----++ 30 μM AA - - - -++++ 500 μM -+-+-+ 500 μM -+-+-+-+ 2,4-diHB 2,4-diHB

Figure 5. Coenzyme Q was deficient in ADCK4 KO podocytes. (A and B) Coenzyme Q contents of cultured podocytes and HK-2 cells.

The CoQ9 level was decreased in (A) both cultured podocytes and HK-2 cells and the CoQ10 level was severely deficient in (B) ADCK4 KO podocytes. (C and D) Respiratory chain complex II and succinate-cytochrome c reductase (complex II-III) enzyme activities were measured in (C) podocytes and (D) HK-2 cells. Complex II-III activities were decreased only in podocytes, whereas complex II activities were affected in both cell lines. (E) Succinate-cytochrome c reductase (complex II-III) enzyme activities were measured in podocytes.

1202 JASN JASN 31: 1191–1211, 2020 www.jasn.org BASIC RESEARCH of CoQ (QH2) plays a role as a potent lipid-soluble antioxi- of mitochondria by TEM. Although the mitochondria of con- dant, scavenging free radicals and preventing lipid peroxida- trol podocytes were less affected by AA, AA-treated ADCK4 KO tive damage.1 Although ADCK4 KO in itself did not affect the podocytes showed more severe mitochondrial defects such as viability of cultured podocytes (Supplemental Figure 10E), we swollen and shortened cristae and fewer inner membranes examined cell viability upon AA treatment because CoQ- (Figure 6, A and B). These results indicate that ADCK4 KO deficient yeast mutants were found to be more sensitive to confers susceptibility to cellular stress, such as autoxidation polyunsaturated fatty acids, such as AA, which are prone to of polyunsaturated fatty acids. To examine functional defects autoxidation and break down into toxic products.35 AA of mitochondria in ADCK4 KO cells, we measured the mito- treatment reduced cell viability in both the control and chondrial membrane potential using JC-10 dye, which is con- ADCK4 KO podocytes, and ADCK4 KO podocytes were rel- centrated in the mitochondrial matrix based on membrane atively more affected (Figure 5F). Decreased cell viability by polarization.36 The JC-10 fluorescence intensity was signifi- AA treatment was rescued by supplementation of 2,4-diHB cantly reduced in the ADCK4 KO cells (Figure 6C). This finding (Figure 5F). Overall, these findings suggested that the loss of was confirmed by another mitochondrial membrane potential ADCK4 caused CoQ deficiency and that podocytes were assay using the potentiometric probe tetramethylrhodamine more susceptible than HK-2 cells. methyl ester–based fluorimetric assay. Tetramethylrhodamine To understand the molecular changes induced by ADCK4 methyl ester fluorescence intensity was significantly reduced KO, we performed proteomic analysis and quantified changes in ADCK4 KO cells; however, supplementation of 2,4-diHB in protein abundance by MS-based proteomics using iTRAQ32 partially restored the lowered mitochondrial membrane po- in podocytes with or without AA treatment. Proteomic char- tential (Figure 6D). In contrast, mitochondrial membrane acterization of the control and ADCK4 KO podocytes revealed potential was not different between the control and KO in .2500 proteins, 421 (16%) of which were mitochondrial HK-2 cells (Supplemental Figure 12, B and C). In addition, proteins. By GO analysis, using the DAVID functional anno- we examined the proteins of the mitochondrial oxidative tation tool, differentially expressed proteins in the control phosphorylation system and found that UQCRC2 of complex and ADCK4 KO cells were annotated based on biologic pro- III was significantly decreased in ADCK4 KO podocytes com- cess. The results indicated that proteins related to cellular de- pared with control podocytes, whereas proteins in other com- fense response were upregulated and those associated with plexes were not different (Supplemental Figure 13, A and B). cytokine production pathway were downregulated in However, in glomerular lysates, there was no difference in ADCK4 KO podocytes compared with those in the control oxidative-phosphorylation-system complexes between con- D cells (Supplemental Figure 10F). In an injury situation, with trol and Adck4 Podocyte mice (Supplemental Figure 13, C and the use of AA treatment, coenzyme metabolism–related pro- D). Therefore, the loss of ADCK4 caused morphologic and teins and intermediate filament–related proteins were down- functional defects of mitochondria in cultured podocytes by dis- regulated, whereas DNA regulation proteins were upregulated rupting CoQ10 biosynthesis. in ADCK4 KO podocytes (Supplemental Figure 10G). CoQ acts as an antioxidant against ROS37;therefore, we investigated ROS production in ADCK4 KO podocytes. Disrupted Mitochondrial Morphology and Mitochondrial superoxide, assessed by the MitoSOX reagent, Mitochondrial Membrane Potential Were Observed in increased in ADCK4 podocytes compared with control podo- ADCK4 KO Podocytes cytes in the basal state, and the difference became more sig- Abnormal proliferation of polymorphous mitochondria in the nificant upon stimulation with 500 mMH2O2,100mMtBHP, cytoplasm of podocytes is one of the characteristic ultrastruc- 30 mM AA, or 100 mM AA (Figure 6E). In addition, cellular 11 tural findings of CoQ10-related diseases. Similarly, patients ROS production, measured using the CellROX reagent, also with ADCK4 mutations showed mitochondrial abnormalities increased in ADCK4 podocytes compared with control podo- 26 in podocytes and proximal tubules. We examined the ultra- cytes in the basal state, and treatment with H2O2,tBHP,orAA structure of mitochondria in control and ADCK4 KO cells by increased ROS production overall. However, no significant TEM. The formation of cristae was disrupted and the shape of difference was observed between control and ADCK4 KO po- mitochondria was disintegrated in ADCK4 KO podocytes docytes (Supplemental Figure 14). (Figure 6, A and B), whereas ADCK4 KO HK-2 cells showed Because we previously reported that ADCK4 knockdown normal features of the mitochondria (Supplemental Figure 12A). reduced podocyte migration,12 in this study we examined the We also observed the effect of AA treatment on the ultrastructure cytoskeleton of ADCK4 KO cells. Actin phalloidin staining

Decreased activities in ADCK4 KO podocytes were partially restored by the addition of 500 mM 2,4-diHB. (F) Cell viability was measured using the Cell Count Kit-8 assay. ADCK4 KO podocytes exhibited susceptibility to 30 mM AA treatment. Decreased cell viability was reversed by the addition of 500 mM 2,4-diHB. P values were calculated using an unpaired t test; *P,0.05, **P,0.005; error bars represent mean6SD. TTFA, thenoyltrifluoroacetone.

JASN 31: 1191–1211, 2020 ADCK4 Stabilizes Coenzyme Q Complex 1203 BASIC RESEARCH www.jasn.org

ABCultured podocytes 15 5000x 8000x 30000x 100 M AA 200 n.s. ** ) 2 150 10

100

Control 5 50 Number of cristae / section (10 m

Number of mitochondria 0 0 Cont ADCK4 Cont ADCK4 -rol KO#1 -rol KO#1

KO#1 150 10 ADCK4 n.s. ** )

2 8 100 6 4 50 2 KO#2 ADCK4 / section (10 m

Number of cristae 0

Number of mitochondria 0 -2 Cont ADCK4 Cont ADCK4 -rol KO#1 -rol KO#1 AA treatment

C JC-10 D TMRM 150 150 ** * ** 100 * * 100

50 50 Relative TMRM Relative fluoroscence (%) 590 / 520 nm ratio

0 0 Control+- - Control +----+ ADCK4 KO#1 -+- ADCK4 KO#1 -+--+- ADCK4 KO#2 --+ ADCK4 KO#2 --++-- 500 M 2,4-diHB -+-+-+

E 500 *** *** 400 *** *** * 300 ** ** ** ** ** *** 200 * * * *** ** *** (510/580 nm ratio) 100 Fluorescence mean intensity

0 Control +-- + -- + +---- + +---- + +---- +-- + -- ADCK4 KO#1--- +-++ + ----++----++----++---- ADCK4 KO#2-++-- +- + -++--- -++--- -++------ 2,4-diHB 500 M H2O2 100 M tBHP 30 M AA 100 M AA 2,4-diHB 2,4-diHB 2,4-diHB 2,4-diHB

Figure 6. ADCK4 KO podocytes showed mitochondrial defects. (A) TEM of podocytes showing mitochondrial morphology. Black and white boxed areas are enlarged mitochondria in ADCK4 KO podocytes showing abnormal fission and disrupted cristae (black arrows). Mitochondria after AA treatment were severely disrupted in ADCK4 KO podocytes. Scale bars, 0.5 mminfirst column, 0.1 mm in second column, and 0.05 mm in third and fourth columns. (B) Morphometric analyses showed that ADCK4 KO podocytes had similar number of

1204 JASN JASN 31: 1191–1211, 2020 www.jasn.org BASIC RESEARCH

A 100 M AA + B * 500 M 2,4-diHB 100 M AA 500 M 2,4-diHB 150 n.s. * 100 n.s. Control 50 Relative Area (%) Relative

0

AA AA ADCK4KO 2,4diHB 2,4diHB

AA + 2,4-diHB AA + 2,4-diHB Control ADCK4 KO#1

Figure 7. ADCK4 KO resulted in decreased cellular area. (A) Immunofluorescence of COXIV and phalloidin staining in podocytes. COXIV marks mitochondria, whereas phalloidin marks actin cytoskeleton. Staining intensity of COXIV or phalloidin was not different between control and ADCK4 KO podocytes. Scale bars, 50 mm. (B) Quantification of cellular area. Phalloidin-stained area was shrunk in ADCK4 KO podocytes, and it became more prominent upon AA treatment. Decreased cellular area was not reversed by the addition of 2,4-diHB. A total of 100 cells from three independent experiments; P values were calculated using an unpaired t test; *P,0.05, **P,0.005; the error bars represent mean6SD. revealed shrunk cellular area in ADCK4 KO podocytes (Figure 7), protein eluates were analyzed using a liquid chromatograph whereas ADCK4 KO HK-2 cells showed a surface area similar coupled to a high-resolution mass spectrometer (LC-MS/MS). to that of control HK-2 cells (Supplemental Figure 12D). The In total, 612 proteins were identified as interactors of ADCK4. mitochondrial marker COX IV was not affected by ADCK4 Among them, the cytoplasmic proteins, including myosin KO in podocytes or HK-2 cells (Figure 7, Supplemental (MYH10, MYH11, MYO1B, and MYO1C), filamin (FLNC), Figure 12D). Interestingly, shrunk cellular area in ADCK4 and kinase proteins (STK24, STK25, STK38, and ROCK1) KO podocytes was not restored by 2,4-diHB supplementa- were detected. In addition, the mitochondrial proteins, in- tion, suggesting that this cellular phenotype is not related to cluding ATP synthase subunit (ATP5L), cytochrome c oxidase decreased CoQ10 level and that ADCK4 might have other subunit (COX6A1 and UQCRQ), and COQ5 were also iden- cellular functions in addition to its role in the CoQ biosyn- tified as interactors (Figure 8A). GO analysis of mitochondrial thesis pathway. AA treatment significantly reduced cellular interactors showed that these proteins are involved in trans- area in both the control and ADCK4 KO podocytes (Figure 7). ferase activity, oxidoreductase activity, nucleotide binding, Shrunk cellular area by AA treatment was not rescued by 2,4- and ATPase activity (Figure 8B). Because COQ5, functioning diHB, further confirming that this phenotype is not related to as a C-methyltransferase in the CoQ biosynthesis pathway,38 CoQ10 deficiency (Figure 7). was identified as an interactor of ADCK4, we confirmed the interaction in podocytes by coimmunoprecipitation ADCK4 Interacted with and Stabilized COQ Proteins (Supplemental Figure 16A) and examined COQ proteins in We performed proteomic analysis to understand the function ADCK4 KO podocytes. Although mRNAs of several COQ of ADCK4 via the identification of its interactome. We gener- genes were not affected by ADCK4 KO (Supplemental ated HEK293 cells that stably overexpressed C-terminal, Figure 16B), the protein levels of COQ3, COQ5, and COQ9 FLAG-tagged BAP (BAP-3xFLAG) and ADCK4 (ADCK4- significantly decreased in ADCK4 KO podocytes compared 3xFLAG) (Supplemental Figure 15A). We confirmed that with control podocytes, indicating these complex Q proteins ADCK4-3xFLAG mostly localized to the mitochondria were destabilized in the absence of ADCK4 (Figure 8C). The (Supplemental Figure 15B). Following affinity purification us- decrease in the level of these proteins was further confirmed in D ing anti-FLAG beads (Supplemental Figure 15, C and D), glomerular lysates of Adck4 Podocyte mice (Supplemental

mitochondria, but cristae number was decreased compared with control cells. AA treatment further decreased mitochondrial volume density in ADCK4 podocytes. (C and D) Mitochondrial membrane potential (DC) was measured using (C) JC-10 and (D) tetrame- thylrhodamine methyl ester (TMRM). ADCK4 KO podocytes showed reduced DC compared with that of control podocytes. Reduced DC of ADCK4 KO podocytes was partially rescued by the addition of 500 mM 2,4-diHB. (E) The mitochondrial superoxide level measured using the MitoSOX reagent increased in ADCK4 KO cells. ADCK4 KO cells exhibited higher sensitivity to treatment with

H2O2, tBHP, and especially AA. Addition of 2,4-diHB decreased mitochondrial superoxide levels in both control and ADCK4 KO podocytes. P values in (B–E) were calculated using an unpaired t test; *P,0.05, **P,0.005, ***P,0.0001; the error bars represent mean6SD.

JASN 31: 1191–1211, 2020 ADCK4 Stabilizes Coenzyme Q Complex 1205 BASIC RESEARCH www.jasn.org

A Cytoplasmic Mitochondrial B

ADCK4 ADCK4 STK38 ATIC Mitochondrial Proteins MYH11 ATP5L Oxidoreductase Transferase ZYX COX6A1 activity activity SPIN1 NDUFA4 MTA3 SRC NDUDFA4NDUF4A COQ5 ATIC STK25 PITRM1 COX6A1 ALDH3A2 ATIC CDS2 ROCK1 COMTD1 UQCRQ COMTD1 TLE3 ALDH3A2 STK24 GK3P Nucleotide ATPase MYH10 COQ5 binding activity MYO1C DHX57 PC GMPPB DHX57 ATP5L FLNC AKR7A2 SRC ATAD3A IDE ATP5MG MYO1B PC GK3P ATP5MF UBR4 NEFH TES MRPL37 CAND2 UQCRQ

-4.50 0.00 4.50

Log2(ADCK4/BAP)

Cultured C Cultured D podocytes podocytes Control + - - - Control + - - Control ADCK4 KO#1 ADCK4 KO -+++ 150 ADCK4 KO#1 -+- 150 ADCK4 KO#2 ADCK4 clone --+ - ADCK4 KO#2 --+ 2,4-diHB ---+ kDa Blot kDa Blot: 100 * 100 52 COQ3 ** ADCK4 ** 52 (Short COQ5 level 50 ** 52 ** exposure) COQ5 ** 50 ** ADCK4

*** (normalized to actin, %)

COQ protein 52 (Long 38 0 COQ9 exposure)

(normalized to actin, %) Control + - - - 31 52 COQ5 52 0 ADCK4 KO - + + + ACTIN 52 WT - - + - 38 ACTIN COQ3 COQ5 COQ9 2,4-diHB - - - + 38

E Cultured podocytes F Cultured 150 podocytes Control + ------ADCK4 KO#1 -+++++++ + * Control ++ - - ** ADCK4 KO#1 --++ 100 100 M AA ++-- ADCK4 ** * ** kDa Blot * ** 52 kDa MOCK MOCK Wild type R178W D250G R320W H400Nfs*11 Q452* E483* Blot: p-p38 50 38 52 COQ5 COQ5 level 52

(normalized to actin, %) p-ERK 52 0 38 ADCK4 38 52 p-JNK E483* Q452* MOCK

D250G 38 Control

52 R178W R320W

ACTIN Wild type 52

H400Nfs*11 38 p38 ADCK4 KO #1 podocytes

Figure 8. ADCK4 interacted with COQ5 and stabilized complex Q in podocytes. (A) ADCK4-interacting proteins isolated form HEK293 cells overexpressing ADCK4-3XFLAG. Both cytoplasmic and mitochondrial proteins were detected as interactors by nanoflow LC-MS/ MS. (B) GO analysis showed that ADCK4-interacting mitochondrial proteins are associated with transferase activity, oxidoreductase activity, nucleotide binding, and ATPase activity. (C) Immunoblot showed that proteins in complex Q, COQ3, COQ5, and COQ9 were significantly reduced in ADCK4 KO podocytes. (D) Decreased COQ5 protein level was rescued by heterologous ADCK4 expression and 2,4-diHB treatment in ADCK4 KO podocytes. (E) Effect of ADCK4 mutations on COQ5 rescue in ADCK4 KO podocytes. COQ5 was rescued to a lower extent by truncated ADCK4 mutant proteins, whereas ADCK4 mutant proteins harboring missense variations were not different

1206 JASN JASN 31: 1191–1211, 2020 www.jasn.org BASIC RESEARCH

COOH COOH COOH COOH OH OH OH OH

CH3 HO CH H CO CH COQ3 COQ??? 3 3 3 COQ2 COQ6 COQ5 COQ7 COQ3

RRHO RR R R H3CO H CO H CO R H3CO 3 3 H3CO OH OH OH OH OH OH OH OH DMQ 4HB [ COQ4, ADCK3(COQ8A), ADCK4(COQ8B), COQ9] DMeQ CoQ10H2

OO OO O + – PDSS1 O 2H , 2e - P P H PDSS2 - P P + P O O - O (n - 1) - -O -O - O O O - O - O O O n O O O O decaprenyl PP DMAPP IPP H3CO CH3

H CO R COOH COOH 3 HO HO O

CoQ10 COQ2 COQ6, COQ3, COQ5, COQ???, [ COQ4, ADCK3(COQ8A), ADCK4(COQ8B), COQ9] R OH OH 2,4-diHB

COOH decaprenyl PP COOH

COQ2 COQ???, COQ5, COQ7, COQ3, [ COQ4, ADCK3(COQ8A), ADCK4(COQ8B), COQ9] OCH R 3 OCH3 OH OH Vanillic acid

Figure 9. Proposed pathway for the synthesis of CoQ10 via alternative ring precursors. 4-Hydroxybenzoic acid (4-HB) is the canonic aromatic ring precursor used in CoQ10 biosynthesis. The alternative ring precursors 2,4-diHB and vanillic acid also serve as substrates for COQ2. After attachment of the decaprenyl tail, the other COQ polypeptides that participate in the canonic biosynthetic pathway serve to modify the ring so that CoQ10 may be produced. This allows bypassing of certain steps. Some of the late-stage CoQ intermediates are needed to stabilize the Q complex (CoQ synthome). Cells with deficiencies in certain steps of the pathway show stabilization of certain COQ polypeptides in the presence of the alternate ring precursors, which can bypass the deficient steps. 4HB, 4-hydroxybenzoic acid; DMQ, demethoxy-coenzyme Q; DMeQ, demethyl-coenzyme Q; decaprenyl PP, decarprenyl pyrophosphate; DMAPP, dimethylallyl pyrophosphate; IPP, isopentenyl pyrophosphate; 2,4-diHB, 2,4-dihydroxybenzoic acid.

Figure 16C). Decreased COQ5 was restored not only by trans- DISCUSSION fection of wild-type ADCK4, but also by 2,4-diHB (Figure 8D). In addition, we examined the effect of ADCK4 In this study, we demonstrated that podocyte-specific deletion mutations, which were previously identified in individuals of Adck4 in mice resulted in proteinuria and foot process ef- with nephrotic syndrome.12 COQ5 was also rescued by facement, recapitulating the features of nephrotic syndrome ADCK4 mutant proteins; however, the extent of the rescue caused by ADCK4 mutations. These defects were efficiently was less than that by wild-type ADCK4 (Figure 8E). In addi- ameliorated by treatment with 2,4-diHB, an unnatural pre- tion, because STK38 is a negative regulator of MAPKKK1/2 cursor analogue of the CoQ biosynthesis pathway. ADCK4 KO kinase signaling,39 we examined the MAPK pathway by West- podocytes exhibited reduced activity of complex II and III, ern blotting and found that phosphorylated ERK1/2 was sig- mitochondrial defects, and sensitivity to AA, which resulted nificantly increased in ADCK4 KO podocytes (Figure 8F). from CoQ deficiency. ADCK4 mutations cause adolescence- Moreover, we observed considerably enhanced MAPK signaling onset nephrotic syndrome, which often progresses to ESKD in including p-p38, p-ERK1/2, and p-JNKunder lipid peroxidation the second decade of life.14 The late onset of renal disease is injury induced by AA (Figure 8F). Because Coq2 silencing re- differentiated from nephropathy resulting from mutations in sulted in increased autophagy and mitophagy in Drosophila WT1, NPHS1,orNPHS2, which generally manifests in the first D nephrocytes,40 we examined mTOR and LC3 in ADCK4 KO year of life. Similarly, in this study, Adck4 Podocyte mice exhibi- podocytes; however, we could not observe increased LC3-II ex- ted renal disease, beginning at approximately 4 months, and pression (Supplemental Figure 17). Taken together, ADCK4 progressed to ESKD by 12 months. ADCK4-associated contributes to stabilizing the Q complex, elucidating CoQ de- glomerulopathy can be partially treated by CoQ10 supplemen- ficiency in the absence of ADCK4, and other signaling path- tation; however, it is not always successful.12,14,41 Failure to ways were also affected upon ADCK4 KO in podocytes. respond to CoQ supplementation can be attributed to several from wild-type ADCK4. (F) Immunoblot analysis of the MAPK pathway. p-p38 and p-pERK were increased in ADCK4 KO podocytes and p-p38 was further induced by AA. Unphosphorylated p38 was used as the loading control. Densitometry analyses in (C–E) are representative of at least three independent experiments and band intensities were normalized to that of b-actin. *P,0.05, **P,0.005; t test.

JASN 31: 1191–1211, 2020 ADCK4 Stabilizes Coenzyme Q Complex 1207 BASIC RESEARCH www.jasn.org possible causes, one of which is related to the progression of the cells with AA, one of the polyunsaturated fatty acids, to renal disease to an irreversible stage. Therefore, early genetic induce lipid peroxidation stress and found that the MAPK diagnosis is necessary to recognize ADCK4 mutations. In this pathway signaling was activated in ADCK4 KO podocytes. regard, because the onset of ADCK4-associated glomerulop- The MAPK signaling pathway is essential in regulating several athy occurs relatively late in life, if properly diagnosed, its cellular processes including inflammation, cell stress response, therapy can be initiated before the disease becomes fulminant. and cell proliferation.46 In addition, AA treatment more sig- Another reason for the failure of CoQ supplementation might nificantly reduced cell viability in ADCK4 podocytes than in be the poor oral availability of CoQ, which makes its thera- control podocytes and the reduced viability was rescued by the peutic efficacy variable and limited. In this study, we demon- supplementation of 2,4-diHB. The reduced form of CoQ10 strated that 2,4-diHB efficiently ameliorated proteinuria and may act to scavenge lipid peroxyl radicals and function as an D prevented FSGS in Adck4 Podocyte mice. Treatment with 2,4- antioxidant, thereby preventing the initiation of lipid perox- diHB has been shown to bypass defects in the penultimate step idation, because it has been reported to eliminate perferryl of CoQ biosynthesis, which is mediated by Coq7(Clk1) hy- radicals.1,47 In this regard, ADCK4 KO may confer hypersen- droxylase (Figure 9).16,17 Further, 2,4-diHB has also been sitivity to lipid peroxidation stress. This suggests that cellular showntobemoreeffectiveintamoxifen-inducibleMclk1/Coq7 stress may be necessary to develop renal diseases in addition to KO mice than CoQ.15 It seems more readily absorbed than loss of ADCK4, partially explaining the relative late onset of CoQ and is safe, because it has been used as a food flavor nephrotic syndrome resulting from ADCK4 mutations. modifier due to its sweet taste. Therefore, translational studies Recent studies have revealed that ADCK3 lacks canonic are required to investigate whether 2,4-diHB can be effective protein kinase activity in the trans form; instead, it binds to 25,48 in individuals with ADCK4 mutations. lipid CoQ10 intermediates and exhibits ATPase activity. In 2,4-diHB is expected to be beneficial for enzymatic defi- this study, GO analysis also revealed that interactors of ciency in the CoQ biosynthetic pathway because it bypasses ADCK4, especially in the mitochondria, are significantly as- the defect in COQ7.1,15 It is, therefore, interesting that 2,4- sociated with oxidoreductase activity, which can be related to DPodocyte diHB rescued disease phenotypes in Adck4 mice be- the antioxidant property of CoQ10. Furthermore, proteomic cause ADCK4, although an uncharacterized mitochondrial analysis revealed the ATP-binding proteins as ADCK4 inter- protein, is not an enzyme directly involved in the CoQ bio- actors, suggesting that ADCK4 may have the ATPase activity, synthetic pathway. This finding suggests that ADCK4 supports like ADCK3. Yet, the precise role of ADCK4 is not clear and an enzymatic component in CoQ biosynthesis. Via proteomic further studies are required to verify whether ADCK4 has AT- analysis, we found that ADCK4 interacts with COQ5 and the Pase or kinase activity toward an undiscovered substrate. expression of proteins in complex Q—namely, COQ3, COQ5, In conclusion, our study suggests that ADCK4 in podocytes and COQ9—significantly decreased in ADCK4 KO podocytes. stabilizes proteins in complex Q in podocytes, and thereby It has been previously suggested that a physical and functional contributes to CoQ synthesis and plays a role in maintaining interaction between ADCK3 and COQ5 is important.24,25 the cytoskeleton structure. Cellular defects and renal pheno- In this study, phalloidin staining showed that the cytoskeleton types by ADCK4 deficiency were mostly rescued by 2,4-diHB of only podocytes was defective, but not that of HK-2 cells. This supplementation, an unnatural precursor analogue of CoQ10, observation suggests that mitochondrial dynamics may also play demonstrating the important role of ADCK4 in the CoQ important roles in maintaining the shape and function of podo- biosynthesis pathway. Our study provides insights into the cytes because podocytes, like neurons,42 require a proper supply functions of ADCK4 in CoQ biosynthesis and pathogenesis and high amount of energy to maintain their foot processes. In of nephrotic syndrome. addition, because the foot process has rich microfilaments, these interactions may also participate in podocyte homeostasis.43,44 Moreover, these cytoskeletal defects in ADCK4 KO podocytes were also consistent with the iTRAQ analysis data, which eluci- ACKNOWLEDGMENTS dated that cytoskeleton-related proteins were downregulated in ADCK4 KO podocytes compared with those in control podo- We thank Dr. Jin Young Kim and Gina Yoon (Korea Basic Science cytes. In contrast to other cellular defects, the shrunken cytoskel- Institute, Ochang, Division of Biomedical Omics Research) for as- eton of ADCK4 KO podocytes was not rescued by 2,4-diHB, sistance with the nanoflow LC-MS/MS analysis. We also thank Maria suggesting that this cellular phenotype may not be related to Ericsson, Louis Trakimas, Elizabeth Benecchi, and Peg Coughlin CoQ10 deficiency. This may also explain why CoQ10 supplemen- from the Electron Microscope Core Facility, Harvard Medical School, tation is not always effective. In addition, it has recently been for their excellent TEM services, and Charlotte Meyer and Evelyn shown that podocytes predominantly rely on glycolysis and that Flynn for her outstanding technical assistance. mitochondrial defects are not sufficient to cause FSGS,45 suggest- We also acknowledge the support of the University of Alabama at ing the pathogenic mechanism of ADCK4 mutationsiscomplex. Birmingham/University of California at San Diego O’Brien Core CoQ10 is well known for its antioxidant activity, thereby Center for AKI Research for the LC-MS/MS analysis (NIH 1P30 DK protecting cells from oxidative stress.3 In this study, we treated 079337) in this study. We thank Yonsei Advanced Imaging Center and

1208 JASN JASN 31: 1191–1211, 2020 www.jasn.org BASIC RESEARCH

Life Imaging Center in the Center for Biological Systems Analysis Supplemental Figure 9. Treatment with 2,4-diHB prevented loss of 1 fl fl (ZBSA) of the Albert Ludwig University of Freiburg for assistance synaptopodin expression in Nphs2.Cre ;Adck4 ox/ ox mice. with confocal microscopes. We thank MID (Medical Illustration & Supplemental Figure 10. Generation of ADCK4 knockout cells Design) for providing support with the medical illustrations. and characterization of ADCK4 deficient podocyte using iTRAQ. Dr. Choi, Dr. Helmstädter, Dr. Hugo, Dr. Lee, Dr. Nakayama, Supplemental Figure 11. Coenzyme Q contents of Nphs2.Cre1;- fl fl Dr. Schapiro, Dr. Widmeier, Dr. Buerger, and Dr. Yu carried out the Adck4 ox/ ox mice. animal experiments. Dr. Choi, Dr. Chung, Dr. Clarke, Dr. Fernández- Supplemental Figure 12. ADCK4 knockout (KO) did not cause del-Río,Dr.Hugo,Dr.Kim,Dr.Lee,Dr.Nag,Dr.Nakayama,Dr.Ryu, mitochondrial defects in HK-2 cells. Dr. Schapiro, Dr. Widmeier, and Dr. Yu carried out the cell experiments. Supplemental Figure 13. The effect of loss of ADCK4 on OXPHOS Dr.Gee,Dr.Hildebrandt,Dr.Widmeier,andDr.Yuconceivedanddi- complexes. rected the study. Dr. Widmeier and Dr. Yu wrote the paper with help Supplemental Figure 14. Quantitation and statistical analysis of from Dr. Hildebrandt and Dr. Gee. The manuscript was critically oxidative stress based on staining with CellROX oxidative stress reviewed by all authors. regents. Supplemental Figure 15. Purification of ADCK4-binding proteins for LC/MS-MS. DISCLOSURES Supplemental Figure 16. ADCK4 stabilized COQ proteins at the protein level. Dr. Hildebrandt is a cofounder of Goldfinch-Bio. The authors have Supplemental Figure 17. AA did not induce autophagic marker in nothing to disclose. ADCK4 KO podocytes.

FUNDING REFERENCES

This study was supported by National Institutes of Health grant DK076683 (to Dr. Hildebrandt) and National Science Foundation grant MCB-1330803 1. Stefely JA, Pagliarini DJ: Biochemistry of mitochondrial coenzyme Q (to Dr. Clarke). Dr. Hildebrandt is the William E. Harmon Professor. Dr. Gee biosynthesis. Trends Biochem Sci 42: 824–843, 2017 was supported by the Chung-Am (TJ Park) Science Fellowship and the Re- 2. Mitchell P: Protonmotive redox mechanism of the cytochrome b-c1 search Program through the National Research Foundation of Korea funded complex in the respiratory chain: Protonmotive ubiquinone cycle. by the Korean Government (Ministry of Science and ICT, South Korea), grant FEBS Lett 56: 1–6, 1975 2018R1A5A2025079. Dr. Widmeier was supported by the Leopoldina Fellow- 3. Mugoni V, Postel R, Catanzaro V, De Luca E, Turco E, Digilio G, et al.: ship Program, Deutsche Akademie der Naturforscher Leopoldina – Nationale Ubiad1 is an antioxidant enzyme that regulates eNOS activity by Akademie der Wissenschaften (German National Academy of Sciences Leo- CoQ10 synthesis. Cell 152: 504–518, 2013 poldina) grant LPDS 2015-07. 4. Awad AM, Bradley MC, Fernández-Del-Río L, Nag A, Tsui HS, Clarke CF: Coenzyme Q10 deficiencies: pathways in yeast and humans. Essays Biochem 62: 361–376, 2018 SUPPLEMENTAL MATERIAL 5. Tran UC, Clarke CF: Endogenous synthesis of coenzyme Q in eukary- otes. Mitochondrion 7[Suppl]: S62–S71, 2007 6. Lagier-Tourenne C, Tazir M, López LC, Quinzii CM, Assoum M, Drouot This article contains the following supplemental material online at N, et al.: ADCK3, an ancestral kinase, is mutated in a form of recessive fi http://jasn.asnjournals.org/lookup/suppl/doi:10.1681/ASN.2019070756/-/ ataxia associated with coenzyme Q10 de ciency. Am J Hum Genet 82: 661–672, 2008 DCSupplemental. 7. Mollet J, Giurgea I, Schlemmer D, Dallner G, Chretien D, Delahodde A, Supplemental Figure 1. Generation and validation of the et al.: Prenyldiphosphate synthase, subunit 1 (PDSS1) and OH- 1 flox/flox Nphs2.Cre ;Adck4 mouse model. benzoate polyprenyltransferase (COQ2) mutations in ubiquinone de- Supplemental Figure 2. Renal histology of 2 month-old ficiency and oxidative phosphorylation disorders. JClinInvest117: 1 fl fl – Nphs2.Cre ;Adck4 ox/ ox mice. 765 772, 2007 8. Quinzii CM, Kattah AG, Naini A, Akman HO, Mootha VK, DiMauro S, Supplemental Figure 3. Body weight of non-treated and treated et al.: Coenzyme Q deficiency and cerebellar ataxia associated with an 1 flox/flox Nphs2.Cre ;Adck4 mice. aprataxin mutation. Neurology 64: 539–541, 2005 Supplemental Figure 4. The expression of podocyte markers was 9. López LC, Schuelke M, Quinzii CM, Kanki T, Rodenburg RJT, Naini A, 1 fl fl decreased in Nphs2.Cre ;Adck4 ox/ ox mice. et al.: Leigh syndrome with nephropathy and CoQ10 deficiency due to Supplemental Figure 5. Deletion of Adck4 led to increased mito- decaprenyl diphosphate synthase subunit 2 (PDSS2) mutations. Am J Hum Genet 79: 1125–1129, 2006 chondria aspect ratio as mice age. 10. Heeringa SF, Chernin G, Chaki M, Zhou W, Sloan AJ, Ji Z, et al.: COQ6 Supplemental Figure 6. Treatment with 2,4-diHB protected from mutations in human patients produce nephrotic syndrome with sen- 1 flox/flox the renal failure of Nphs2.Cre ;Adck4 mice. sorineural deafness. J Clin Invest 121: 2013–2024, 2011 Supplemental Figure 7. Treatment with 2,4-diHB prevented focal 11. Diomedi-Camassei F, Di Giandomenico S, Santorelli FM, Caridi G, 1 fl fl segmental glomerular sclerosis in Nphs2.Cre ;Adck4 ox/ ox mice. Piemonte F, Montini G, et al.: COQ2 nephropathy: A newly described inherited mitochondriopathy with primary renal involvement. JAm Supplemental Figure 8. Treatment with 2,4-diHB prevented loss of Soc Nephrol 18: 2773–2780, 2007 fi expression pattern of podocyte markers and glomerular brosis in 12. Ashraf S, Gee HY, Woerner S, Xie LX, Vega-Warner V, Lovric S, et al.: 1 flox/flox Nphs2.Cre ;Adck4 mice. ADCK4 mutations promote steroid-resistant nephrotic syndrome

JASN 31: 1191–1211, 2020 ADCK4 Stabilizes Coenzyme Q Complex 1209 BASIC RESEARCH www.jasn.org

through CoQ10 biosynthesis disruption. JClinInvest123: 5179–5189, 31. Gasser DL, Winkler CA, Peng M, An P, McKenzie LM, Kirk GD, 2013 et al.: Focal segmental glomerulosclerosis is associated with a PDSS2 13. Acosta MJ, Vazquez Fonseca L, Desbats MA, Cerqua C, Zordan R, haplotype and, independently, with a decreased content of coenzyme Q10. Trevisson E, et al.: Coenzyme Q biosynthesis in health and disease. Am J Physiol Renal Physiol 305: F1228–F1238, 2013 Biochim Biophys Acta 1857: 1079–1085, 2016 32. Chung YW, Lagranha C, Chen Y, Sun J, Tong G, Hockman SC, et al.: 14. Korkmaz E, Lipska-Zie˛tkiewicz BS, Boyer O, Gribouval O, Fourrage C, Targeted disruption of PDE3B, but not PDE3A, protects murine heart Tabatabaei M, et al.; PodoNet Consortium: ADCK4-Associated glo- from ischemia/reperfusion injury. Proc Natl Acad Sci U S A 112: merulopathy causes adolescence-onset FSGS. J Am Soc Nephrol 27: E2253–E2262, 2015 63–68, 2016 33. Xu T, Park SK, Venable JD, Wohlschlegel JA, Diedrich JK, Cociorva D, 15. Wang Y, Oxer D, Hekimi S: Mitochondrial function and lifespan of et al.: ProLuCID: An improved SEQUEST-like algorithm with enhanced mice with controlled ubiquinone biosynthesis. Nat Commun 6: 6393, sensitivity and specificity. J Proteomics 129: 16–24, 2015 2015 34. Tabb DL, McDonald WH, Yates JR 3rd: DTASelect and Contrast: Tools 16. Xie LX, Ozeir M, Tang JY, Chen JY, Jaquinod SK, Fontecave M, et al.: for assembling and comparing protein identifications from shotgun Overexpression of the Coq8 kinase in Saccharomyces cerevisiae coq proteomics. J Proteome Res 1: 21–26, 2002 null mutants allows for accumulation of diagnostic intermediates of the 35. Do TQ, Schultz JR, Clarke CF: Enhanced sensitivity of ubiquinone- coenzyme Q6 biosynthetic pathway. J Biol Chem 287: 23571–23581, deficient mutants of Saccharomyces cerevisiae to products of autoxi- 2012 dized polyunsaturated fatty acids. Proc Natl Acad Sci U S A 93: 17. Liu JL, Yee C, Wang Y, Hekimi S: A single biochemical activity under- 7534–7539, 1996 lies the pleiotropy of the aging-related protein CLK-1. Sci Rep 7: 859, 36. Li J, Yu L, Gu X, Ma Y, Pasqualini R, Arap W, et al.: Tissue plasminogen 2017 activator regulates Purkinje neuron development and survival. Proc 18. Widmeier E, Airik M, Hugo H, Schapiro D, Wedel J, Ghosh CC, et al.: Natl Acad Sci U S A 110: E2410–E2419, 2013 Treatment with 2,4-Dihydroxybenzoic acid prevents FSGS progression 37. Genova ML, Pich MM, Biondi A, Bernacchia A, Falasca A, Bovina C, and renal fibrosis in podocyte-specific Coq6 knockout mice. JAmSoc et al.: Mitochondrial production of oxygen radical species and the role Nephrol 30: 393–405, 2019 of Coenzyme Q as an antioxidant. Exp Biol Med (Maywood) 228: 19. Kannan N, Taylor SS, Zhai Y, Venter JC, Manning G: Structural and 506–513, 2003 functional diversity of the microbial kinome. PLoS Biol 5: e17, 2007 38. Nguyen TPT, Casarin A, Desbats MA, Doimo M, Trevisson E, Santos- 20. Pagliarini DJ, Calvo SE, Chang B, Sheth SA, Vafai SB, Ong S-E, et al.: A Ocaña C, et al.: Molecular characterization of the human mitochondrial protein compendium elucidates complex I disease bi- COQ5 C-methyltransferase in coenzyme Q10 biosynthesis. Biochim ology. Cell 134: 112–123, 2008 Biophys Acta 1841: 1628–1638, 2014 21. Tauche A, Krause-Buchholz U, Rödel G: Ubiquinone biosynthesis in 39. Enomoto A, Kido N, Ito M, Morita A, Matsumoto Y, Takamatsu N, et al.: Saccharomyces cerevisiae: The molecular organization of O-methylase Negative regulation of MEKK1/2 signaling by serine-threonine kinase Coq3p depends on Abc1p/Coq8p. FEMS Yeast Res 8: 1263–1275, 38 (STK38). Oncogene 27: 1930–1938, 2008 2008 40. Zhu JY, Fu Y, Richman A, Zhao Z, Ray PE, Han Z: A personalized model of 22. Vazquez Fonseca L, Doimo M, Calderan C, Desbats MA, Acosta MJ, COQ2 nephropathy rescued by the wild-type COQ2 allele or dietary co-

Cerqua C, et al.: Mutations in COQ8B (ADCK4) found in patients with enzyme Q10 supplementation. J Am Soc Nephrol 28: 2607–2617, 2017 steroid-resistant nephrotic syndrome alter COQ8B function. Hum 41. Feng C, Wang Q, Wang J, Liu F, Shen H, Fu H, et al.: Coenzyme Q10 Mutat 39: 406–414, 2018 supplementation therapy for 2 children with proteinuria renal disease 23. He CH, Xie LX, Allan CM, Tran UC, Clarke CF: Coenzyme Q supple- and ADCK4 mutation: Case reports and literature review. Medicine mentation or over-expression of the yeast Coq8 putative kinase stabi- (Baltimore) 96:e8880,2017 lizes multi-subunit Coq polypeptide complexes in yeast coq null 42. Imasawa T, Rossignol R: Podocyte energy metabolism and glomerular mutants. Biochim Biophys Acta 1841: 630–644, 2014 diseases. Int J Biochem Cell Biol 45: 2109–2118, 2013 24. Floyd BJ, Wilkerson EM, Veling MT, Minogue CE, Xia C, Beebe ET, 43. Greka A, Mundel P: Cell biology and pathology of podocytes. Annu Rev et al.: Mitochondrial protein interaction mapping identifies regulators Physiol 74: 299–323, 2012 of respiratory chain function. Mol Cell 63: 621–632, 2016 44. Suleiman HY, Roth R, Jain S, Heuser JE, Shaw AS, Miner JH: Injury- 25. Stefely JA, Licitra F, Laredj L, Reidenbach AG, Kemmerer ZA, induced actin cytoskeleton reorganization in podocytes revealed by Grangeray A, et al.: Cerebellar ataxia and coenzyme Q deficiency super-resolution microscopy. JCI Insight 2: 94137, 2017 through loss of unorthodox kinase activity. Mol Cell 63: 608–620, 2016 45. Brinkkoetter PT, Bork T, Salou S, Liang W, Mizi A, Ozel C, et al.: An- 26. Park E, Kang HG, Choi YH, Lee KB, Moon KC, Jeong HJ, et al.: Focal aerobic glycolysis maintains the glomerular filtration barrier in- segmental glomerulosclerosis and medullary nephrocalcinosis in chil- dependent of mitochondrial metabolism and dynamics. Cell Rep 27: dren with ADCK4 mutations. Pediatr Nephrol 32: 1547–1554, 2017 1551–1566.e5, 2019 27. Young S, Struys E, Wood T: Quantification of creatine and guanidi- 46. Cowan KJ, Storey KB: Mitogen-activated protein kinases: New signal- noacetate using GC-MS and LC-MS/MS for the detection of cerebral ing pathways functioning in cellular responses to environmental stress. creatine deficiency syndromes. Curr Protoc Hum Genet 54: JExpBiol206: 1107–1115, 2003 17.3.1–17.3.18, 2007 47. Ernster L, Forsmark-Andrée P: Ubiquinol: An endogenous antioxidant 28. Hinkes B, Wiggins RC, Gbadegesin R, Vlangos CN, Seelow D, in aerobic organisms. Clin Investig 71[Suppl]: S60–S65, 1993 Nürnberg G, et al.: Positional cloning uncovers mutations in PLCE1 48. Reidenbach AG, Kemmerer ZA, Aydin D, Jochem A, McDevitt MT, responsible for a nephrotic syndrome variant that may be reversible. Hutchins PD, et al.: Conserved lipid and small-molecule modulation of Nat Genet 38: 1397–1405, 2006 COQ8 reveals regulation of the Ancient kinase-like UbiB family. Cell 29. Saleem MA, O’HareMJ,ReiserJ,CowardRJ,InwardCD,FarrenT,etal.:A Chem Biol 25: 154–165.e11, 2018 conditionally immortalized human podocyte cell line demonstrating nephrin and podocin expression. J Am Soc Nephrol 13: 630–638, 2002 30. Fernández-Del-Río L, Nag A, Gutiérrez Casado E, Ariza J, Awad AM, Joseph AI, et al.: Kaempferol increases levels of coenzyme Q in kidney See related editorial, “Mitochondria Matter: A Critical Role of ADCK4 in Stabilizing cells and serves as a biosynthetic ring precursor. Free Radic Biol Med the CoQ Complex in Podocytes in Steroid-Resistant Nephrotic Syndrome,” on pages 110: 176–187, 2017 1167–1169.

1210 JASN JASN 31: 1191–1211, 2020 www.jasn.org BASIC RESEARCH

AFFILIATIONS

1Division of Nephrology, Department of Medicine, Boston Children’s Hospital, Harvard Medical School, Boston, Massachusetts 2Renal Division, Department of Medicine, Medical Center – University of Freiburg, Faculty of Medicine, University of Freiburg, Freiburg, Germany 3Departments of Pharmacology, Yonsei University College of Medicine, Seoul, Korea 4Brain Korea 21 PLUS Project for Medical Science, Yonsei University College of Medicine, Seoul, Korea 5Department of Chemistry and Biochemistry, Molecular Biology Institute, University of California, Los Angeles, Los Angeles, California 6Severance Biomedical Science Institute, Yonsei University College of Medicine, Seoul, Korea 7Biochemistry and Molecular Biology, Yonsei University College of Medicine, Seoul, Korea

JASN 31: 1191–1211, 2020 ADCK4 Stabilizes Coenzyme Q Complex 1211 Supplemental Table of Contents: “ADCK4 deficiency destabilizes the coenzyme Q complex, which is rescued by 2,4-dihydroxybenzoic acid treatment”

Supplement Contents: • Supplemental Figure 1. Generation and validation of the Nphs2.Cre+;Adck4flox/flox mouse model • Supplemental Figure 2. Renal histology of 2 month-old Nphs2.Cre+;Adck4flox/flox mice • Supplemental Figure 3. Body weight of non-treated and treated Nphs2.Cre+;Adck4flox/flox mice • Supplemental Figure 4. The expression of podocyte markers was decreased in Nphs2.Cre+;Adck4flox/flox mice • Supplemental Figure 5. Deletion of Adck4 led to increased mitochondria aspect ratio as mice age • Supplemental Figure 6. Treatment with 2,4-diHB protected from the renal failure of Nphs2.Cre+;Adck4flox/flox mice • Supplemental Figure 7. Treatment with 2,4-diHB prevented focal segmental glomerular sclerosis in Nphs2.Cre+;Adck4flox/flox mice • Supplemental Figure 8. Treatment with 2,4-diHB prevented loss of expression pattern of podocyte markers and glomerular fibrosis in Nphs2.Cre+;Adck4flox/flox mice • Supplemental Figure 9. Treatment with 2,4-diHB prevented loss of synaptopodin expression in Nphs2.Cre+;Adck4flox/flox mice • Supplemental Figure 10. Generation of ADCK4 knockout cells and characterization of ADCK4 deficient podocyte using iTRAQ • Supplemental Figure 11. Coenzyme Q contents of Nphs2.Cre+;Adck4flox/flox mice • Supplemental Figure 12. ADCK4 knockout (KO) did not cause mitochondrial defects in HK-2 cells • Supplemental Figure 13. The effect of loss of ADCK4 on OXPHOS complexes • Supplemental Figure 14. Quantitation and statistical analysis of oxidative stress based on staining with CellROX oxidative Stress Regents • Supplemental Figure 15. Purification of ADCK4-binding proteins for LC/MS-MS • Supplemental Figure 16. ADCK4 stabilized COQ proteins at the protein level • Supplemental Figure 17. Arachidonic acid did not induce autophagic marker in ADCK4 KO podocytes A B C

D

ns ns ns ns ns nsns ns ns ns ns EFGH6 ***** *** ** ** ** ** ** * *** *** *** 2.5 *** ** ** ***** * *** 35 ns nsns nsns ns **** ns ns nsns * * ns

5 30 2.0 25 4 1.5 20 3

1.0 15 2 10

1 0.5 5 Plasma creatinine (mg/dl)

0 0.0 0 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 Month Month Month + flox/flox Control (n=17) Nphs2.Cre ;Adck4 (n=15) Control (n=17) Nphs2.Cre+;Adck4flox/flox (n=15) Control (n=17) Nphs2.Cre+;Adck4flox/flox (n=15) Supplemental Figure 1. Generation and validation of the Nphs2.Cre+;Adck4flox/flox mouse model. (A) Schematic diagram of the strategy for generating the conditional Adck4 allele. Genotyping primer locations are indicated with numbers #1, #2, #3, #4 and #5. (B) PCR-based genotyping of wild-type (1) and Adck4tm1a mice (2). Adck4tm1c allele is 1763 bp product on the presence of the lacZ cassette. Adck4tm1c allele using in the upper panel generates a 294 bp product from wild type allele (1) and 507 bp product from LoxP-containing conditional knockout locus (3). Nphs2.Cre+ specific PCR-product in the lower panel. (C) Nphs2.Cre+;Adck4flox/flox mutant mouse in the upper panel showed progressive reduction of physical condition compared to littermates controls. The Nphs2.Cre+;Adck4flox/flox mutant mice in the lower panel showed significant reduction of kidney size compared to littermates controls (mice at age 10 months. N=3 wild type and n=7 Nphs2.Cre+;Adck4flox/flox animals). IF, immunofluorescence; TEM, transmission electron microscopy. (D) Experimental Protocol. Albumin/creatinine ratio analysis started at 2 months of age. Serial urine and plasma analyses were performed for 18 consecutive months. Supplementation of 2,4-diHB to the mice in drinking water were started at 3 months of age. The kidney tissues were obtained from 2-month-old to 18- month-old mice and used for experiments. IF, (E-H) Nphs2.Cre+;Adck4flox/flox mice (red hexagon) developed severe hypoalbuminemia, subsequent renal failure and severe hyperphosphatemia in adulthood but not in littermates controls (black diamond). Dotted line displays onset of severe proteinuria (D) or onset of renal failure (E-G) respectively. (D-G, n=15-17 animals in each group). p-values were calculated using an unpaired t-test; ns=not significant, * p<0.05,**p<0.01, **** p<0.0001, error bars represent SEM. Supplemental Figure 2. Renal histology of 2 month-old Nphs2.Cre+;Adck4flox/flox mice. (A and B) Kidney serial sections representative images at indicated genotypes were stained accordingly to indicated conditions for histological analysis. At age of 2 months Nphs2.Cre+;Adck4flox/flox mutant mice and littermates control mice show a normal histological kidney morphology. (Scale bars: upper rows 200 μm, and lower row 40 μm). (C and D) Transmission electron microscopy representative images at indicated genotypes were performed for an analysis of a glomerular ultrastructure. At age of 2 months Nphs2.Cre+;Adck4flox/flox mutant mice and littermates control mice show a normal glomerular ultrastructure, podocytes and podocytes foot processes morphology as well as a normal morphology of podocytes mitochondria (white arrow heads). (Scale bars: upper rows 10 μmand 2 µm; lower rows 500 nm and 250 nm). ns ns ns ns ABC35 35

30 30

ns ns ns ns 25 25 Weight (g)

20 20

15 15 5 month 5 month

Supplemental Figure 3. Body weight of non-treated and treated Nphs2.Cre+;Adck4flox/flox mice. (A-C) Body weight analysis at indicated genotypes and at ages (A) of 3 months, (B) of 5 months and (C) at 8 months were analyzed to determine the physical condition over the time. The Nphs2.Cre+;Adck4flox/flox mice show normal developmental stage at ages of 3 and 5 months, however males are starting significant body weight loss upon the onset of renal failure at age of 8 months compared to littermates controls. Whereas Nphs2.Cre+;Adck4flox/flox mice under treatment with 2,4-diHB display consistently normal physical condition compared to littermates controls under treatment. (n=3-9 animals in each group. p-values were calculated using an ordinary one-way multiple comparisons ANOVA; ns=not significant, ** p<0.01, error bars represent mean +/- SD). ADE

BF

CG

Supplemental Figure 4. The expression of podocyte markers was decreased in Nphs2.Cre+;Adck4flox/flox mice. (A) Immunofluorescence staining for the slit diaphragm protein nephrin (green) and for the glomerular fibrosis marker collagen IV (red). Unlike normal expression pattern of nephrin and collagen IV in control mice, Nphs2.Cre+;Adck4flox/flox mice showed omitted nephrin staining, whereas collagen IV staining (arrow heads) was focally increased. (B) Immunofluorescence staining for the podocyte foot processes marker synaptopodin (green) and the basement membrane marker nidogen (red). Nphs2.Cre+;Adck4flox/flox mice showed mostly omitted synaptopodin staining (arrows) appearing only on a few capillary loops (arrow head). (C) Immunofluorescence staining of frozen kidney sections for the slit diaphragm protein nephrin (green) and for the glomerular fibrosis marker smooth muscle actin (αSMA) (red). Nphs2.Cre+;Adck4flox/flox mice showed omitted nephrin staining, whereas αSMA staining (arrow heads) was focally increased. (D-G) Respective antibody staining intensities were analyzed by a 3D surface plot (left panels) and quantified in right panels. FIU=fluorescence intensity units, n=2-3 images in each group were analysed, p-values were calculated using an unpaired t-test; ** p<0.01 error bars represent SD. iohnraapc ai in ratio aspect Mitochondria auswr acltduiga nardtts;n=o infcn,**** significant, ns=not the t-test; regarding unpaired an animals p using calculated (B) were old values month 10 µm and per mitochondria (A) n ratio: old aspect mitochondria months podocyte 2 of of (A) age. Deletion mice as ratio 5. aspect mitochondria Figure Supplemental infcn nraecmae oltemt otosa ieaged. mice as controls littermate to compared increase significant 000 ro asrpeetSD. represent bars error <0.0001 and B A (B) Mitochondria n/ m2 Mitochondria n/ m2 uniaieaayi ftrnmsineeto micrographs electron ttransmission of analysis Quantitative Nphs2.Cre + ;Adck4 Adck4 flox/flox e oincreased to led ooye showed podocytes 2 cytoplasm. p - A 100 90 80 70 p = 0.0078 60 50 40 30 20 Nphs2.Cre+;Adck4flox/flox (n=15) 10 Nphs2.Cre+;Adck4flox/flox + 25 mM 2,4-diHB (n=9) 0 891011 12 13 14 15 16 17 18 Month

B 6 ns nsns ns ns ns ns ns C

5

4

3

2

1 Plasma creatinine (mg/dl) 0 11 12 13 14 15 16 17 18 Month Control + 25 mM 2,4-diHB (n=9) Nphs2.Cre+;Adck4flox/flox + 25 mM 2,4-diHB (n=8)

DEnsnsns ns ns ns ns 12 ns nsns ns ns ns ns * 90 **

80 10

70 8 60

50 6 40 4 30

20 2 10

0 0 11 12 13 14 15 16 17 18 11 12 13 14 15 16 17 18 Month Month Control + 25 mM 2,4-diHB (n=9) Control + 25 mM 2,4-diHB (n=9) + flox/flox Nphs2.Cre+;Adck4flox/flox + 25mM 2.4-diHB (n=8) Nphs2.Cre ;Adck4 + 25 mM 2,4-diHB (n=8)

Supplemental Figure 6. Treatment with 2,4-diHB protected from the renal failure of Nphs2.Cre+;Adck4flox/flox mice. (A) Nphs2.Cre+;Adck4flox/flox mice exhibited reduced life span with a median survival of 316 days compared to Nphs2.Cre+;Adck4flox/flox mice under treatment with 2,4-diHB. (n=9-15 mice at age 18 months in each group. Log-rank [Mantel-Cox] test, p=0.0078;). (B-E) Treatment of Nphs2.Cre+;Adck4flox/flox mice with 2,4-diHB (green) protected from developing severe hypoalbuminemia, renal function decline, and hyperphosphatemia compared to treated littermate controls (black circle). (n=8-9 animals in each group). p-values were calculated using an unpaired t-test; ns=not significant, * p<0.05, error bars represent SEM. Supplemental Figure 7. Treatment with 2,4-diHB prevented focal segmental glomerular sclerosis in Nphs2.Cre+;Adck4flox/flox mice. (A and B) Kidney serial sections representative images at indicated genotypes were stained accordingly to indicated conditions for histological analysis. Unlike a normal histological kidney morphology seen in Nphs2.Cre+;Adck4flox/flox mice treated with 2,4-diHB at 12.5 months of age, Nphs2.Cre+;Adck4flox/flox mice at age of 12.5 months showed severe focal segmental glomerular sclerosis (arrows) with severe interstitial fibrosis and tubular atrophy (arrow head). (Scale bars: upper rows 200 μm, and lower row 40 μm). (C and D) Transmission electron microscopy representative images at indicated genotypes were performed for an analysis of a glomerular ultrastructure. Unlike a normal glomerular ultrastructure, podocytes and podocytes foot processes morphology as well as a normal morphology of podocytes mitochondria (white arrow heads) seen in Nphs2.Cre+;Adck4flox/flox mice treated with 2,4-diHB at 12.5 months of age, Nphs2.Cre+;Adck4flox/flox mutant mice at the same age showed severe podocytes foot processes effacement (arrows) and a severely impaired podocytes mitochondrial morphology (black arrow heads). (Scale bars: upper rows 10 μm and 2 µm; lower rows 500 nm and 250 nm). Supplemental Figure 8. Treatment with 2,4-diHB prevented loss of expression pattern of podocyte markers and glomerular fibrosis in Nphs2.Cre+;Adck4flox/flox mice . Immunofluorescence staining of frozen kidney sections in the upper panels for the podocyte foot processes marker synaptopodin (green), the basement membrane marker nidogen (red) and in the lower panels for the slit diaphragm protein nephrin (green) and for the glomerular fibrosis marker smooth muscle actin (αSMA) (red). (A and B) Unlike a normal expression pattern of synaptopodin, nephrin and αSMA seen in Nphs2.Cre+;Adck4flox/flox mice treated with 2,4-diHB at 12.5 months of age, Nphs2.Cre+;Adck4flox/flox mice at the same age showed mostly omitted synaptopodin staining appearing only on a few capillary loops (arrow heads) and omitted nephrin staining, whereas αSMA staining was focally increased (arrows). (C) A 3D surface plot depicting synaptopodin antibody staining intensities, is shown on the left panel. Quantification of antibody staining is shown on the right panel. (D) A 3D surface plot depicting nephrin antibody staining intensities, is shown on the left panel. Quantification of antibody staining is shown on the right panel. (E) A 3D surface plot depicting αSMA antibody staining intensities, is shown on the left panel. Quantification of antibody staining is shown on the right panel. (Scale bars: 20 µm, FIU=fluorescence intensity units, n=3 images in each group were analysed, p- values were calculated using an unpaired t-test: * p<0.05,**p<0.01, *** p<0.001, error bars represent SD). Supplemental Figure 9. Treatment with 2,4-diHB prevented loss of synaptopodin expression in Nphs2.Cre+;Adck4flox/flox mice. Immunofluorescence staining of frozen kidney sections for the podocyte foot processes marker synaptopodin (green) and the basement membrane marker nidogen (red). (A) The representative images at 18 months-old mice. Unlike a normal expression pattern of synaptopodin seen in wild type littermate control and wild type treated littermate control mice, Nphs2.Cre+;Adck4flox/flox mice treated with 2,4-diHB showed globally reduced expression pattern of synaptopodin, although Nphs2.Cre+;Adck4flox/flox showed mostly omitted synaptopodin staining appearing only on a few capillary loops (arrow heads). (B) Synaptopodin antibody staining intensities are shown in the upper panel and 3D surface plot depicting the respective antibody staining intensities are shown in the lower panel. (C) Quantification of synaptopodin antibody staining according to indicated genotypes. (Scale bars: 20 µm, FIU=fluorescence intensity units, n=3imagesineachgroupwereanalysed, p-values were calculated using ordinary one-way ANOVA/Tukey’s multiple comparisons test: ns=not significant, * p<0.05,**p<0.01, **** p<0.0001, error bars represent SD). AB

E CD

FG

Supplemental Figure 10. Generation of ADCK4 knockout (KO) cells and characterization of ADCK4 deficient podocyte using iTRAQ. (A) Three sgRNAs targeting the exon 6 of ADCK4. The each target DNA regions are indicated below pink-colored, green-colored, and blue-colored sequences, respectively. (B-C) Sanger sequencing of the exon 6 of ADCK4 confirmed genome editing in podocytes (B) and HK-2 cells originated from proximal tubule (C). (D) Western blots of ADCK4 expression in the CRISPR/Cas9-edited KO cell lines.

(E) Cell viability were measured using cell count kit-8 (CCK) assay. H2O2 was treated at 300μM for 16 hr in ADCK4 deficient-podocyte (left) and HK-2 cells (right). (p-values were calculated using an unpaired t-test; n.s.=not significant, error bars represent SD). (F) Heatmap of significantly increased or decreased proteins in ADCK4 knockout (KO) podocytes compared to control podocytes. Gene Ontology (GO) analysis shows that proteins related to defense response are increased in ADCK4 KO podocytes, whereas proteins associated with organic cycle compound metabolism are decreased in ADCK4 KO podocytes. (G) Heatmap of significantly increased and decreased proteins in ADCK4 KO podocytes compared to control podocytes upon arachidonic acid (AA) treatment.GO analysis shows that AA treatment results in decrease of proteins related to coenzyme metabolism and intermediated filament-based process, in contrast, proteins related to DNA packaging, protein methylation, and cellular component biogenesis are increased in ADCK4 KO podocytes compared to control podocytes. ABCoQ10 levels 800

600

400

200 p = 0.2479

0 Cont Nphs2.Cre+; Cont Nphs2.Cre+; -rol Adck4flox/flox -rol Adck4flox/flox

Supplemental Figure 11. Coenzyme Q contents of Nphs2.Cre+;Adck4flox/flox mice. (A, B) Coenzyme Q contents were measured using glomeruli isolated from 10 month-old control and flox/flox Nphs2.Cre+;Adck4 mice. The CoQ9 (A) and CoQ10 (B) level tended to decrease in. mice, but it was not statistically significant. p-values were calculated using an unpaired t- test and error bars represent SD. Supplemental Figure 12. ADCK4 knockout (KO) did not cause mitochondrial defects in HK-2 cells. (A) Transmission electron microscopy (TEM) of podocytes showing mitochondrial morphology. Each boxed areas are enlarged in right panel respectively. TEM shows normal mitochondrial morphology with regular shape of cristae even in ADCK4 KO HK-2 cells. (Black arrow in right panel) (Scale bars:, 0.5 μm in 5000x panel, 0.1 μm in 8000x panel, 0.05 μm in 30000x panel) (B-C) HK-2 were loaded with JC-10 (B) and TMRM (C) to evaluate mitochondrial membrane potential (ΔΨ). ADCK4 KO HK-2 cells showed normal ΔΨ compared to that of controls. (D) Immunofluorescence of COXIV (cytochrome oxidase subunit IV) and Phalloidin staining in HK-2. Phalloidin stained area displayed no differences between controls and ADCK4 KO HK-2. (Scale bar, 50 µm) (p-values were calculated using an unpaired t-test; n.s.=not significant) A

B Control ADCK4 KO#1 ADCK4 KO#2

I II III IV V

C D 150

100 Nphs2.Cre+;Adck4flox/+ Nphs2.Cre+;Adck4flox/flox 50

0 IIIIIIV

Supplemental Figure 13. The effect of loss of ADCK4 on OXPHOS complexes (A) The antibodies against proteins representing the mitochondrial oxidative phosphorylation system (OXPHOS) complexes were used to examine the expression of mitochondrial complex subunits in control and ADCK4 KO podocytes. (B) Quantification of mitochondrial complex subunits from (A) normalized by beta-actin. The results indicate that the UQCRC2 representing complex III was decreased in ADCK4 KO cells. p-values were calculated using an unpaired t-test; ** p<0.01. (C) Immunoblots of OXPHOS complexes in glomeruli of Nphs2.Cre+;Adck4flox/flox and Nphs2.Cre+;Adck4flox/+ mice. (D) Quantification of mitochondrial complex subunits from (C) normalized by beta-actin. OXPHOS complexes in glomeruli of Nphs2.Cre+;Adck4flox/flox were not significantly reduced compared to that of Nphs2.Cre+;Adck4flox/+ mice. Control ADCK4 KO#1 ADCK4 KO#2

Control+-- +-- +-- +-- +-- +-- +-- ADCK4 KO#1 - + - - + - - + - - + - - + - - + - - + - ADCK4 KO#2 - - + - - + - - + - - + - - + - - + - - + 100 M 300 M 30 M 100 M 30 M 100 M

H2O2 H2O2 tBHP tBHP AA AA

Supplemental Figure 14. Quantitation and statistical analysis of oxidative stress based on staining with CellROX oxidative Stress Regents. Podocyte cells were plated in a 96-well plate. The cells were stained with 5uM of CellROX deep red reagent for 30 minutes at 33C and washed for with HBSS. The cells were then treated with or without each chemicals for 30 min and analyzed fluorescence mean intensity with cytation 5. The results indicate that although basal level of ROS production were as twice as much higher in ADCK4 KO cells, treatment of

H2O2, tBHP and AA does not have impact on ROS production between control and ADCK4 KO cells. Supplemental Figure 15. Purification of ADCK4-binding proteins for LC/MS-MS. (A) HEK293 cells were transfected with 3xFLAG-tagged ADCK4 or BAP vectors and confirmed by immunoblot analysis. (B) HEK293 stably expressing 3xFLAG tagged ADCK4, BAP vectors were confirmed by immunofluorescence using COXIV (red), FLAG (green) antibodies. (Scale bar, 50 µm) (C) BAP and ADCK4 were detected from the elutes by immunoblot analysis. Cell lysates were immunoprecipitated with FLAG M2 antibody conjugated agarose beads and eluted by competing with 3xFLAG peptides. Each step was confirmed by immunoblot. (D) The elutes were resolved by SDS-PAGE and prepared for coomassie blue staining and silver staining. A

B

Control ADCK4 KO#1 ADCK4 KO#2

C

Nphs2.Cre+;Adck4flox/+ Nphs2.Cre+;Adck4flox/flox Normalized to actinNormalized (%)

Supplemental Figure 16. ADCK4 stabilized COQ proteins at the protein level. (A) The interaction between ADCK4 and COQ5 was confirmed by co-immunoprecipitation in cultured podocytes. 3xFLAG-tagged ADCK4 or BAP was transfected into podocoytes and immunoprecipitated with anti-FLAG antibody. Endogenous COQ5 was detected by immunoblotting. (B) Quantitative real-time PCR indicated that mRNAs of COQ genes were not changed by ADCK4 knockout. Error bars indicate SD of at least three experiments and p-values were calculated using an unpaired t-test; *** p<0.001. (C) Expression of COQ proteins in glomeruli of control and Nphs2.Cre+;Adck4flox/flox. Quantification of mitochondrial complex subunits normalized by beta-actin shows that COQ3, COQ5, and COQ9 level were decreased in glomeruli of Nphs2.Cre+;Adck4flox/flox. A B Cultured 1 μM Rapamycin 100 μM AA podocytes Control + + - - ADCK4 KO#1 - - + + 100 μM AA-+-+ kDa Blot Control p-mTOR 225

mTOR 225

38 Actin ADCK4 KO#1 ADCK4 17 LC3 A/B 12

Actin 38

Supplemental Figure 17. Arachidonic acid (AA) did not induce autophagic marker in ADCK4 KO podocytes. (A) Representative fluorescent images of podocytes transiently transfected with tfLC3. Autophagy was not induced in both control and ADCK4 KO podocytes. 100 µM AA treatment to control and ADCK4 KO podocytes transfected with tfLC3 did not represent increased mRFP puncta (white arrows) which indicated autolysosome, whereas 1 µM rapamycin induced mRFP puncta in both control and ADCK4 KO podocytes (Scale bar, 50 µm). (B) Western blot analysis of the phospho-mTOR, mTOR, autophagic markers LC3 signals in the total protein extracts from control and ADCK4 KO podocytes upon AA treatment. Expression levels were normalized to the internal control, ACTIN. LC3-II expression of ADCK4 KO podocytes treated with AA were not increased.