Sirtuin 5 Regulates Proximal Tubule Fatty Acid Oxidation to Protect Against AKI

BASIC RESEARCH www.jasn.org

Sirtuin 5 Regulates Proximal Tubule Fatty Acid Oxidation to Protect against AKI

Takuto Chiba,1 Kevin D. Peasley ,1 Kasey R. Cargill,1 Katherine V. Maringer,1 Sivakama S. Bharathi,1 Elina Mukherjee,1 Yuxun Zhang ,1 Anja Holtz ,2 Nathan Basisty ,2 Shiva D. Yagobian,1 Birgit Schilling,2 Eric S. Goetzman,1 and Sunder Sims-Lucas1

1Department of Pediatrics, University of Pittsburgh School of Medicine, University of Pittsburgh Medical Center Children’s Hospital of Pittsburgh, University of Pittsburgh, Pittsburgh, Pennsylvania; and 2Buck Institute for Research on Aging, Novato, California

ABSTRACT Background The primary site of damage during AKI, proximal tubular epithelial cells, are highly metabolically active, relying on fatty acids to meet their energy demands. These cells are rich in mitochondria and peroxisomes, the two organelles that mediate fatty acid oxidation. Emerging evidence shows that both fatty acid pathways are regulated by reversible posttranslational modifications, particularly by lysine acylation. Sirtuin 5 (Sirt5), which localizes to both mitochondria and peroxisomes, reverses posttranslational lysine acylation on several enzymes involved in fatty acid oxidation. However, the role of the Sirt5 in regulating kidney energy metabolism has yet to be determined. Methods We subjected male Sirt5-deficient mice (either +/2 or 2/2) and wild-type controls, as well as isolated proximal tubule cells, to two different AKI models (ischemia-induced or cisplatin-induced AKI). We assessed kidney function and injury with standard techniques and measured fatty acid oxidation 14 14 by the catabolism of C-labeled palmitate to CO2. 2/2 Results Sirt5 was highly expressed in proximal tubular epithelial cells. At baseline, Sirt5 knockout (Sirt5 ) mice had modestly decreased mitochondrial function but significantly increased fatty acid oxidation, 2/2 which was localized to the peroxisome. Although no overt kidney phenotype was observed in Sirt5 2/2 mice, Sirt5 mice had significantly improved kidney function and less tissue damage compared with controls after either ischemia-induced or cisplatin-induced AKI. This coincided with higher peroxisomal 2/2 fatty acid oxidation compared with mitochondria fatty acid oxidation in the Sirt5 proximal tubular epithelial cells. Conclusions Our findings indicate that Sirt5 regulates the balance of mitochondrial versus peroxisomal fatty acid oxidation in proximal tubular epithelial cells to protect against injury in AKI. This novel mechanism might be leveraged for developing AKI therapies.

JASN 30: 2384–2398, 2019. doi: https://doi.org/10.1681/ASN.2019020163

Received February 18, 2019. Accepted August 29, 2019.

Proximal tubular epithelial cells (PTECs) are highly T.C., K.D.P., and K.R.C. contributed equally to this work. sensitive to damage after AKI largely because of its 1,2,3 Published online ahead of print. Publication date available at high metabolic rate. To meet energy demands, www.jasn.org. PTECs rely on fatty acid b-oxidation (FAO). Paral- lel FAO pathways exist in mitochondria and in per- Correspondence: Dr. Sunder Sims-Lucas or Dr. Eric S. Goetzman, Department of Pediatrics, University of Pittsburgh School of Medicine, oxisomes, both of which are notably abundant in UPMC Children’s Hospital of Pittsburgh, University of Pittsburgh, 4401 PTECs. Mitochondrial FAO directly links to the Penn Avenue, Pittsburgh, PA 15224. E-mail: [email protected] or mitochondrial electron transport chain and is an [email protected] oxygen-intensive source of energy. Concomitant Copyright © 2019 by the American Society of Nephrology

2384 ISSN : 1046-6673/3012-2384 JASN 30: 2384–2398, 2019 www.jasn.org BASIC RESEARCH with oxygen utilization is the formation of reactive oxygen Significance Statement species (ROS), which are thought to be particularly damaging during ischemic- and cisplatin-induced AKI.4,5 Limiting mito- Proximal tubular epithelial cells, a primary site of injury in AKI, are chondrial FAO limits ischemic injury in the heart6,7 and kidney.8 rich in mitochondria and peroxisomes, the two organelles that Sirtuin 5 Sirt5 In contrast to mitochondria, little is known about the role of mediate fatty acid oxidation. Deletion of ( )reverses posttranslational lysine acylation of several enzymes involved in peroxisomes in AKI. Although peroxisomes do not produce fatty acid oxidation. The authors demonstrate that mice lacking energy, they metabolize long-chain fatty acids to acetate and Sirt5 had significantly improved kidney function and less tissue other short-chain products. Because of their hydrophilicity, damage following either ischemia-induced or cisplatin-induced the metabolites can cross plasma membranes to either leave AKI compared with wild-type mice. These differences coincided the cell or enter mitochondria to be oxidized to completion. with higher peroxisomal fatty acid oxidation compared with mitochondria fatty acid oxidation in the Sirt5-deficient proximal tu- Peroxisomes have been observed to physically interact with bular epithelial cells. Their findings indicate that Sirt5 regulates the mitochondria and ablation of peroxisomal function has sec- balance of mitochondrial versus peroxisomal fatty acid oxidation in ondary effects on mitochondrial function.9 Peroxisomes may proximal tubular epithelial cells to protect against injury in AKI. serve as a sink for mitochondrially produced ROS because This novel mechanism has potential therapeutic implications for of the abundance of catalase and other ROS-neutralizing en- treating AKI. zymes in the peroxisome.10,11 Peroxisomes may also serve to 2 2 2 protect PTECs from the accumulation of toxic long-chain together to produce WT (Sirt5+/+), Sirt5+/ ,andSirt5 / mu- fatty acids.12 tants. Age-matched 10- to 14-week-old male mice were used Emerging evidence shows both FAO pathways are regulated throughout the study. The University of Pittsburgh Institu- by reversible post-translational modifications (PTMs), in par- tional Animal Care and Use Committee approved all experi- ticular by lysine acylation and the sirtuin deacylase enzymes ments (approval no. 16088935). that remove these PTMs.13,14 Mammalian sirtuins Sirt1–Sirt7 differ in their subcellular localization and substrate specificity. Ischemic and Cisplatin AKI Models in Mice The nuclear/cytosolic sirtuin Sirt1 has been shown to protect Ischemic AKI was induced by a renal ischemia-reperfusion against AKI by maintaining peroxisomes, upregulating cata- injury (IRI) model as previously described,23 with modifica- lase, and eliminating renal ROS.15 The mitochondrial sirtuin tions. Briefly, mice were anesthetized with inhalant 2% Sirt3 is also renoprotective through its role in improving mi- isoflurane. Core body temperature of the mice was moni- tochondrial dynamics.16 Although its role in regulating kidney tored with a rectal thermometer probe and was maintained FAO is not clear, Sirt3 promotes mitochondrial FAO in liver at 36.8–37.2°C throughout the procedures with a water-heating and heart.14,17 Sirt5, which is unique among the sirtuins in its circulation pump system (EZ-7150; BrainTree Scientific) and substrate preference for succinyllysine, malonyllysine, and an infrared heat lamp (Shat-R-Shield). Buprenorphine (Par glutaryllysine,18,19 also promotes mitochondrial FAO in liver Pharmaceutical) was administered for pain control (0.1 mg/kg and heart.13,20 Intriguingly, Sirt5 was recently shown to local- body wt, administered subcutaneously). With aseptic tech- ize to peroxisomes. In direct contrast to its effect on mito- niques, a dorsal incision was made to expose the kidney, and chondrial FAO, Sirt5 was observed to suppress peroxisomal renal ischemia for 22 minutes was induced by unilateral clamp- FAO in vitro and in rodent liver.21 However, the role of Sirt5 in ing of the left kidney pedicle with a nontraumatic microvascular 2/2 the kidney has not yet been studied. Here, we report that Sirt5 clamp (18055–04; Fine Science Tools). Renal reperfusion was mice are protected against ischemic and cisplatin AKI. visually verified. Delayed contralateral nephrectomy of the right Although mitochondrial function is moderately suppressed kidney was performed at day 6.23 Mice were euthanized at day 7 2 /2 in Sirt5 kidney, peroxisomal function is enhanced to harvest blood and the injured kidney. Serum was separated both before and after AKI. These results indicate that Sirt5 from the blood and analyzed by the Kansas State Veterinary Di- regulates the balance of mitochondrial versus peroxisomal agnostic Laboratory for levels of creatinine and BUN. FAO in PTECs to protect against PTEC injury in AKI. This To induce cisplatin AKI, mice were given a single dose of novel mechanism may be leveraged for developing future 20 mg/kg intraperitoneal cisplatin (Fresenius Kabi) or vehicle AKI therapies. control of normal saline as described previously.24 Mice were euthanized on day 3 to harvest blood and the kidneys.

METHODS Cultured PTECs Primary mouse PTECs were isolated from single-cell kidney 2 2 2 Animals suspensionof10- to 14-week-old maleSirt5 / , Sirt5+/ ,B6/129 2 2 Homozygous global Sirtuin 5 knockout mice (Sirt5 / )22 were strain WT, or 129 strain WT mice by Dynabeads FlowComp obtained from the Jackson Laboratory (B6;129-Sirt5tm1Fwa/J, Flexi Kit (Thermo Fisher Scientiﬁc) conjugated to lotus tetra- stock no. 012757). B6/129S F2 strain wild-type (WT) mice gonolobus lectin (LTL) (L-1320; Vector Laboratories). Passage were used as controls. To produce appropriate littermate con- 3–6 PTECs were used for experiments. Human kidney proxi- 2 trols Sirtuin 5 heterozygous mutant mice (Sirt5+/ ) were bred mal tubular epithelial cells (hPTECs) were obtained from

JASN 30: 2384–2398, 2019 Sirt5 Deletion Protects from AKI 2385 BASIC RESEARCH www.jasn.org

American Type Culture Collection (Manassas, VA). Both types PTM-401; PTM Biolabs), and anti-Sirt5 antibodies (rabbit, of cells were cultured as previously described,25 with 1:50, 8782; Cell Signaling Technology). Anti-GAPDH or modifications. anti–a/b-tubulin were used as loading controls at 1:100. Mouse primary PTECs were plated at 106 cells per well and, under serum-restricted conditions, exposed to 24 hours of Real-Time PCR normoxia or hypoxia (FiO2 1%) in the hypoxia chamber Real-time PCR analysis was performed as previously de- (Coy Laboratory) or treated with 24 hours of 20 mM cisplatin scribed,28 to determine mRNA level in whole kidneys. Com- or vehicle control of normal saline. plementary DNA was reverse-transcribed from 500 ng of total hPTECs were maintained in a 1:1 mix of DMEM (Thermo RNAwith SuperScript First-Strand Synthesis System (Thermo Fisher Scientific) and Ham F12 (Thermo Scientific) containing Fisher Scientific). Real-time PCR analysis was performed with 5 mM glucose and supplemented with 2 mM GlutaMAX gene specific primer oligos, SsoAdvanced SYBR Green Super- (Thermo Scientific), 13 insulin-transferrin-selenium mix (Bio-Rad), and CFX96 Touch Real-Time PCR Detection (Gibco), 36 ng/ml hydrocortisone, and 0.1% (v/v) penicillin/ System with C1000 Thermal Cycler (Bio-Rad). Cycling streptomycin at 37°C in 5% CO2. conditions were 95°C for 10 minutes, then 40 cycles of 95°C To knockdown Sirt5, the plasmid containing Sirt5 small for 15 seconds and 60°C for 1 minute. Rn18S was used for interfering RNA (siRNA; 16708; Ambion) and the control endogenous control. Expression was compared with Rn18S 2ΔΔ plasmid (scrambled, AM4611; Thermo Fisher Scientific) endogenous control and analyzed using the 2 Ct method. were reverse-transfected into hPTECs using INTERFERin Primer sequences used are shown in Supplemental Table 1. transfection reagent (409–10; Polyplus) according to the manufacturer’s protocol. hPTECs were used for experiments Tissue Section Analysis 48 hours after siRNA treatment. To inhibit acyl-CoA oxidase-1 Kidneys were fixed in 4% paraformaldehyde and embed in (ACOX1) activity, we used 500 nM 10,12-tricosadiynoic acid paraffin. These tissues were sectioned at 4 mm. The kidney (Sigma). sections were stained with hematoxylin and eosin (H&E) or Masson trichrome. H&E-stained slides were subject to histo- Combined Glucose–Oxygen Deprivation–Mediated logic evaluation, and 340 magnification images of renal cor- In Vitro Ischemic AKI tex and medulla were obtained with a Leica DM2500 optical hPTECs were subjected to an in vitro ischemic AKI model by microscope. Semiquantitative scoring (0–4) for tubular injury combined glucose–oxygen deprivation (CGOD) insult. For was performed in terms of tubular dilation, proteinaceous cast the CGOD procedure, hPTECs in a 24-well or 96-well plates formation, and loss of brush border. To quantify the Masson were exposed to 4–24 hours of normoxia or hypoxia (FiO2 trichrome–stained, collagen-rich area, a tiling imaging system 1%) with defined glucose-free buffer (115 mM NaCl, via TissueFAXS PLUS (TissueGnostics) were used to capture 1.0 mM NaH2PO4, 26.2 mM NaHCO3,5.4mMKCl, the entire kidney section with 320 magnification objectives. 1.8 mM CaCl2, and 0.8 mM MgSO4). After CGOD procedure, Masson trichrome–positive area was quantified with NIS Ele- culture medium was replaced with glucose-containing buffer ments software (Nikon). All tissue evaluation was performed (+30 mM glucose) and hPTECs moved to a regular incubator in a blinded fashion. for normoxic treatment in the next 24 hours to simulate “re- Immunostaining were performed as previously described29 perfusion.” The defined glucose-free buffer was used instead with paraffin-embedded tissues and with following primary of glucose-free cell culture media because other components antibodies or lectins: b-galactosidase (chicken, 1:200, ab9361; in medium, such as amino acids, can alter the response of cells Abcam), endomucin (rat, 1:200, sc-65495; Santa Cruz Bio- to hypoxia.26,27 technology), kidney injury molecule 1 (Kim-1; rat, 1:100, MAB1817; R&D Systems), thiazide-sensitive NaCl cotrans- LDH Release Assay porter (rabbit, 1:250, AB3553; Millipore), neutrophil gelati- Cell death was assessed by lactate dehydrogenase (LDH) efflux nase–associated lipocalin (NGAL; rat, 1:50, ab70287; Abcam), using the LDH release assay kit (Promega) in accordance with Peroxisomal Membrane Protein 70 (rabbit, 1:200, ab85550; the manufacturer’s protocol. Cytotoxicity was expressed as the Abcam), dolichos biflorus agglutinin (1:200, FL-1031; Vector ratio of the LDH release in the treated cell medium to that of Laboratories), or LTL (1:200, FL-1321; Vector Laboratories). the maximal LDH release. Palmitate Oxidation Assay Western Blots 14C-palmitate (PerkinElmer) was conjugated to BSA and used 14 14 Kidney lysates were lysed in radioimmunoprecipitation assay at 125 mMin200ml reactions as described. CO2 was cap- buffer (Thermo Fisher Scientific) and the homogenates plated tured on filter papers soaked in 1 M KOH and water-soluble 14 in triplicate to measure protein concentration using a Bradford C-labeled FAO products were separated by the method of assay kit (Bio-Rad). The following antibodies were used in Bligh and Dyer30 and subjected to scintillation counting. Per- this study: anti-Acox1 antibodies (1:1000, rabbit, ab59964; oxisomal FAO was defined as the rate of palmitate oxidation in Abcam), pan-succinyllysine antibodies (1:1000, rabbit, the presence of the irreversible mitochondrial FAO inhibitor

2386 JASN JASN 30: 2384–2398, 2019 www.jasn.org BASIC RESEARCH etomoxir (100 mM). Data were normalized to protein Skyline as described. See experimental details in Supplemental concentration. Appendix 1.32–34

Oroboros High-Resolution Respirometry Statistical Analyses Freshly prepared kidney homogenates were analyzed with an Data are presented as mean6SEM. Prism 7.0C software (GraphPad) Oroboros Oxygraph-2K using our previously published was used for statistical analysis. To determine whether sample method.31 Complex I respiration was defined as malate/ data has been drawn from a normally distributed population, pyruvate/glutamate–driven oxygen consumption in the pres- D’Agostino–Pearson omnibus test or Shapiro–Wilk test was per- ence of ADP, whereas Complex II respiration was defined as formed. For parametric data, ANOVAwith post hoc Turkey com- succinate-driven oxygen consumption. parison was used for multiple group comparison and t test was used to compare two different groups. For nonparametric Quantitative Mass Spectrometry data, Mann–Whitney U test was used. The threshold of Both 7-day renal IRI and contralateral control kidney tissues P,0.05 was set to consider data statistically significant. (n=3) were homogenized, trypsinized, and then succinylated peptides were enriched using the PTMScan Succinyl-Lysine Motif Kit (Cell Signaling Technologies). After reverse-phase RESULTS high-performance liquid chromatography/electrospray 2/2 ionization-tandem mass spectrometry (HPLC/ESI-MS/ Sirt5 Kidneys Exhibit Protein Lysine MS), succinyl enriched samples were analyzed by data- Hypersuccinylation independent acquisition (e.g., SWATH) on a TripleTOF 6600, Sirt5 is highly expressed in kidney35 but its role in kidney and site-specific changes in succinylation were quantified using function has not been examined. We confirmed that global

A B WT SIRT5-/- C D SIRT5 kDa 12341234 SIRT5 200 500 P=0.12 150 400 75 300 100 50 200 50 25 100 Relative Expression 0 Relative Expression 0 WT Sirt5-/- Anti-succinyllysine Kidney Tubule GAPDH

EF G H BUN Creatinine 300 3

200 2

mg/dL 100 mg/dL 1

200μm 0 0 WT Sirt5-/- WT Sirt5-/- WT Sirt5-/-

2 2 2 2 Figure 1. Sirt5 / kidneys exhibit protein lysine hypersuccinylation but normal morphology and function. (A) Real-time PCR of Sirt5 / kidney conﬁrms no detectable Sirt5 mRNA in mouse kidney (n=4/group). (B) Western blotting of kidney lysates (50 mg) with anti- 2 2 succinyllysine antibody was used to assess succinylation levels in mouse kidneys derived from WT or Sirt5 / mice (n=4/group). (C) The 2 2 Sirt5 / mutant allele contains a lacZ cassette that allows for the interrogation of expression using an anti–b-galactosidase (b-gal) antibody (Sirt5–b-gal, red). Abundant expression of Sirt5 is colocalized in the proximal tubular epithelial cells (LTL, green). Scale bar, 25 mm. (D) Real-time PCR of whole kidney compared with isolated LTL-positive PTECs suggest that Sirt5 expression is enriched in 2 2 PTECs (n=3/group). (E and F) H&E-stained kidneys of WT and Sirt5 / kidneys at baseline. No discernable differences were noted 2 2 between WT and Sirt5 / kidneys. Scale bar, =200 mm. (G and H) Serum levels of (G) BUN and (H) creatinine were not signiﬁcantly 2 2 different at baseline between WT and Sirt5 / mice (n=4/group). All data are presented as dot plots plus mean6SEM, t test.

JASN 30: 2384–2398, 2019 Sirt5 Deletion Protects from AKI 2387 BASIC RESEARCH www.jasn.org

2 2 2/2 Sirt5 knockout mice (Sirt5 / ) have no detectable Sirt5 Sirt5 Kidneys Are Protected against Ischemic mRNA in the kidney (Figure 1A). Similar to liver and AKI in Mice 2 2 heart,13,36 Sirt5 deletion in the kidney leads to accumulation To test the role of Sirt5 in ischemic AKI, Sirt5 / mice were of PTMs known as lysine succinylation, shown by pan anti- subjected to a renal IRI-mediated ischemic AKI model and succinyllysine Western blotting for whole kidney lysates were euthanized 7 days later. H&E-stained histopathology re- 2 2 (Figure 1B). The mutant allele in Sirt5 / mice bears a lacZ vealed decreased proteinaceous casts, less severe tubular dila- 2 2 reporter cassette.22 b-galactosidase immunostaining revealed tion, and reduced loss of brush border in Sirt5 / kidneys that Sirt5–b-galactosidase is colocalized in LTL-positive (Figure 2, A, D, and G). Kim-1, which localizes to injured PTECs (Figure 1C), thiazide-sensitive NaCl cotransporter– PTECs,37 wasmuchmoreprominentinWTkidneys(Figure2B positive distal tubular epithelial cells, and dolichos biﬂorus versus Figure 2E). Further, NGAL, another marker for kidney agglutinin–positive collecting ducts, but was absent from en- tubular injury,38 was observed in many WT but was virtually 2 2 domucin-positive microvasculature (Supplemental Figure 1). absent in Sirt5 / PTECs (Figure 2, C and C’ versus Figure 2, In WT mice, Sirt5 mRNA was expressed in the whole kidneys F and F’). Real-time PCR for Lcn2 (NGAL mRNA) conﬁrmed its 2 2 and was enriched in isolated primary PTECs (Figure 1D). At decreased mRNA level in Sirt5 / kidneys (Figure 2J). Finally, 2 2 2 2 baseline, Sirt5 / kidneys appear histologically (Figure 1, E kidney function was protected in Sirt5 / mice, as evidenced by and F) and functionally normal with unaltered serum creati- reduced serum creatinine and BUN (Figure 2, H and I). Inter- nine and BUN compared with WT (Figure 1F). estingly, similar results were observed by using the heterozygous

H&E Kim1/LTL/DAPI NGAL/LTL/DAPI NGAL/LTL/DAPI ABCC’

WT 100µm WT WT WT

D E F F’

Sirt5-/- Sirt5-/- Sirt5-/- Sirt5-/-

G Injury Scoring H Creatinine I BUN J Lcn2 (NGAL) **** * * 4 3 **** 300 250

3 200 2 200 150 2 mg/dL mg/dL 100 1 100 1 Relative mRNA 50

Semiquantitative Score 0 0 0 0 Control Sirt5-/- WT Sirt5-/- WT Sirt5-/- WT Sirt5-/-

2 2 Figure 2. Sirt5 / kidneys are protected against ischemic AKI at 7 days after injury. (A, D, and G) H&E-stained kidneys of (A) WT and 2 2 2 2 (D) Sirt5 / kidneys that were subjected to the ischemic AKI model. The Sirt5 / kidneys had decreased injury score as compared with WT kidneys, evidenced by proteinaceous casts in WT, dilated tubular structures, and vacuolization (arrowheads) that were decreased in 2 2 Sirt5 / kidneys (n=5–8/group). Scale bar, 100 mm. (B and E) Immunostaining for Kim-1 (red) in the PTECs (LTL, green) showed a large 2 2 increase in the amount of Kim-1 expression in the (B) WT compared with very low levels in the (E) Sirt5 / PTECs. Scale bar, 100 mm. (C and C’, inset) and (F and F’, inset) Immunostaining for NGAL (red) in PTECs (LTL) (arrowheads) was observed in many WT but were 2 2 virtually absent from Sirt5 / PTECs. Scale bar, 50 mm. (J) Real-time PCR for Lcn2 (NGAL mRNA) conﬁrms the decrease of Lcn2 in 2 2 2 2 Sirt5 / kidneys compared with WT (n=5–7). (H and I) Serum levels of (I) creatinine and (J) BUN are decreased in the Sirt5 / mice 2 2 compared with WT, suggesting protected kidney function in the Sirt5 / mice (n=11–14/group). All data are presented as dot plots plus mean6SEM. *P,0.05; ****P,0.001, t test.

2388 JASN JASN 30: 2384–2398, 2019 www.jasn.org BASIC RESEARCH

2 mutant (Sirt5+/ ) and their littermate WT controls, suggesting background of the C57Bl/6 and 129Sv strains, and 129Sv-strain that half Sirt5 gene dosage is sufficient to protect against ischemic mice have been shown to exhibit resistance to ischemic AKI.39 AKI (Supplemental Figure 2). We also evaluated post-AKI fibrosis To determine if this is the case in cultured mouse PTECs, we 2 2 with the samples of day 7 after renal IRI. Although mRNA level of subjected PTECs from WT B6/129, WT 129, and Sirt5 / mice 2 2 fibrosis markers (Col1a1 and Tgf-b1) were reduced in Sirt5 / to 24 hours of 1% hypoxia. The results confirm that WT 129 is kidneys, little collagen-rich area was observed both in WT and more protective than WT B6/129, evidenced by reduction of 2 2 2 2 Sirt5 / kidneys (1.5%–2.3%, Supplemental Figure 3). Havcr1 and Lcn2 mRNA; and that Sirt5 / is more protective PTECs are sensitive to ischemic and cisplatin AKI, and are than WT 129, shown by reduction of Havcr-1, IL-18, and Lcn2 reliant upon mitochondria for energy metabolism; Sirt5 is mRNA (Supplemental Figure 4). We further modeled ischemic known to regulate mitochondrial function and is enriched AKI in vitro with primary hPTECs using a CGOD protocol.40 in PTECs (Figure 1). Therefore, we hypothesized that the pro- In this protocol, glucose and oxygen are deprived for the first 2 2 tection against ischemic AKI in Sirt5 / was occurring 24 hours and then added back to “reperfuse” for an additional within PTECs. To test this, primary PTECs were isolated 24 hours. siRNA knocks down Sirt5 levels with an efficiency of 2 2 from Sirt5 / and B6/129-strain WT mice and exposed to approximately 80%, when compared with a scrambled siRNA 24 hours of 1% hypoxic insult in vitro. After the hypoxia ex- control (Figure 3C). At the end of the 24 hours of CGOD and posure, expression of the tubular injury markers Havcr1 the subsequent 24-hour “reperfusion” period, hPTECs with (Kim-1 mRNA), Lcn2,andIL-18 was significantly reduced Sirt5 knockdown demonstrated protection as evidenced by re- 2 2 2 2 in Sirt5 / PTECs (Figure 3A). Sirt5 / mice are a mixed duced LDH efflux (Figure 3D).

A Hypoxia B Cisplatin Primary mouse PTEC Primary mouse PTEC

Lcn2 (NGAL) 1.5 * ** *** 1.5 **** 1.0 1.0

0.5 WT (B6/129) 0.5 -/-

Relative mRNA Sirt5 Relative mRNA

0.0 0.0 Havcr1 IL-18 Lcn2 WT Sirt5 -/- (Kim-1) (NGAL) (B6/129)

CDIn vitro ischemic-AKI via CGOD Human PTEC

Scr Ctrl Sirt5 80 Scrambled - Baseline siRNA siRNA Sirt5 KD - Baseline Scrambled - Injured Sirt5 60 Sirt5 KD - Injured Tubulin * + Glucose 40 + Oxygen %LDH Efflux 20 *

0 0122436 48 Time (h)

Figure 3. Loss-of-function of Sirt5 in proximal tubular epithelial cells is protective against injury in vitro. (A) After 24 hours of 2 2 1% hypoxia, primary PTECs derived from Sirt5 / kidneys exhibited reduced mRNA levels of injury markers Havcr1 (Kim-1 mRNA), Lcn2 (NGAL mRNA), and IL-18, compared with B6/129-strain WT PTECs (n=3/group). (B) After 24 hours, cisplatin-treated primary 2 2 mouse PTECs from Sirt5 / kidneys exhibited reduced mRNA levels of Lcn2, compared with WT PTECs (n=3/group). (C) Western blotting conﬁrms siRNA knockdown (KD) of Sirt5 in primary hPTECs. (D) hPTECs are subjected to CGOD-mediated in vitro ischemic AKI model; Sirt5 KD hPTECs showed a marked decrease in LDH release (n=3/group). All data are presented as dot plots plus mean6SEM. *P,0.05; **P,0.01; ***P,0.001; ****P,0.001, t test.

JASN 30: 2384–2398, 2019 Sirt5 Deletion Protects from AKI 2389 BASIC RESEARCH www.jasn.org

A B C Injury Scoring 4 *

-/- 0 100µm Sirt5 Semiquantitative Score WT Control Sirt5-/-

DECreatinine BUN 250 * 6 *** 200 4 150 100 mg/dL mg/dL 2 50 0 0 WT Sirt5-/- WT Sirt5-/-

FGWeight loss * 25 ** 250 WT 20 200 Sirt5-/- 15 150 10 100 * 5 50 % Weight Loss Relative mRNA 0 0 WT Sirt5-/- Havcr1 (Kim1) Lcn2 (NGAL)

NGAL/LTL/DAPI NGAL/LTL/DAPI NGAL/LTL/DAPI H H’ I

WT WT Sirt5-/-

2 2 Figure 4. Sirt5 / kidneys are protected against cisplatin-induced AKI at 3 days after injury. (A–C) H&E-stained kidneys of (A) WT and 2 2 2 2 (B) Sirt5 / kidneys that were subjected to the cisplatin AKI model. The Sirt5 / kidneys had decreased injury score as compared with WT kidneys, evidenced by proteinaceous casts in WT (arrowheads), dilated tubular structures, and vacuolization (arrows) that were 2 2 decreased in Sirt5 / kidneys (n=3–8/group, Mann–Whitney U test). Scale bar, 100 mm. (D and E) Serum levels of (D) creatinine and (E) 2 2 2 2 BUN are decreased in the Sirt5 / mice compared with WT, suggesting protected kidney function in the Sirt5 / mice (n=8–12/group, 2 2 Mann–Whitney U test). (F) Sirt5 / mice showed reduced weight loss than WT controls after the cisplatin injury (n=9–13/group, t test). 2 2 (G) Real-time PCR for Havcr1 (Kim-1 mRNA) and Lcn2 (NGAL mRNA) conﬁrm the reduction of Havcr1 and Lcn2 mRNA in Sirt5 / 2 2 kidneys compared with WT (n=4–8/group, Mann–Whitney U test). (H–I) WT and Sirt5 / kidneys were immunostained for NGAL (red) and PTECs (LTL, green). WT kidneys (H and H’) had increased NGAL positive (arrows, red) in LTL-positive PTECs compared with 2 2 Sirt5 / kidneys (I). Scale bar, 50 mm. All data are presented as dot plots plus mean6SEM. *P,0.05; **P,0.01; ****P,0.001.

2/2 Sirt5 Mice Are Protected against Cisplatin-Induced intraperitoneally) of the nephrotoxin cisplatin and were AKI euthanized 3 days later. Cisplatin speciﬁcally targets PTECs. Todetermine if the protection afforded by Sirt5 loss-of-function H&E-stained kidney tissues demonstrated less severe tubular 2 2 2 2 extends to other forms of AKI, Sirt5 / mice were chal- injury in cisplatin-treated Sirt5 / mice compared with WT lenged with a high dose (20 mg/kg body wt, administered mice (Figure 4, A–C). Kidney function was also protected in

2390 JASN JASN 30: 2384–2398, 2019 www.jasn.org BASIC RESEARCH

2 2 Sirt5 / mice as evidenced by reduced serum creatinine and in uninjured kidney and 16 in injured kidney, with 15 of these BUN (Figure 4, D and E). During the 72 hours after cisplatin proteins shared across injured and uninjured tissues. The spe- treatment, WT mice lost 14.5% body wt compared with just cific lysine residues targeted by Sirt5 and the fold changes 2 2 2 2 8.4% among Sirt5 / mice (Figure 4F), suggestive of a blunted (Sirt5 / :WT) are detailed in Figure 6. There are five key en- 2 2 response to cisplatin in the Sirt5 / mice. Real-time PCR anal- zymatic steps to mitochondrial FAO, including activation to ysis for Havcr1 and Lcn2 confirmed decreased mRNA level of the metabolically active acyl-CoA derivative and four subse- 2 2 the tubular injury markers in Sirt5 / kidney (Figure 4G). quent enzymatic steps for acyl-CoA chain-shortening. There Consistent with this, many WT PTECs were colocalized with are multiple enzymes at each of the five steps with different 2 2 NGAL, but this was not observed in Sirt5 / PTECs (Figure 4, acyl-CoA chain-length specificities, i.e., long chain, medium H and I). Renoprotection was also confirmed in the PTECs as chain, and short chain. Sirt5-targeted FAO proteins are spread evidenced by the strong expression of Lcn2 in WT primary across all five steps of the pathway. Interestingly, among the PTECs treated with 20 mM cisplatin in vitro as compared succinylated FAO peptides that were quantified in both the 2 2 with Sirt5 / PTECs (Figure 3B). Finally, cultured primary uninjured and injured kidneys, the degree of hypersuccinylation 2 2 2 PTECs derived from heterozygous Sirt5+/ mice reduced (Sirt5 / :WTratios) was always higher in the uninjured kidneys mRNA level of Lcn2 after 24 hours of cisplatin treatment (Figures 5D and 6). when compared with the PTECs derived from their littermate 2/2 WT controls (Supplemental Figure 5). Total FAO Is Increased in Sirt5 Kidneys Both at Baseline and after Ischemic AKI Mitochondrial FAO Enzymes Are Hypersuccinylated in Our lysine succinylome analysis implicated FAO as a key path- 2/2 2 2 Sirt5 Kidney way to confer renoprotection in Sirt5 / kidneys. This path- 2 2 Sirt5 is a protein deacylase with a high affinity for succinylly- way is known to be suppressed in Sirt5 / liver and heart.13,20 sine, followed by glutaryllysine and malonyllysine.18,19 The To explore the effects of Sirt5 loss-of-function on kidney FAO, 14 14 lysine succinylome has not been characterized in kidney. we followed the catabolism of C-labeled palmitate to CO2 Therefore, we used quantitative mass spectrometry to measure in fresh kidney lysates prepared from uninjured kidneys. Con- 2 2 site-level lysine succinylation in Sirt5 / and WT kidneys after trary to expectation, baseline FAO rates were significantly el- 2 2 22 minutes of renal IRI and compared uninjured contralateral evated in Sirt5 / kidneys (Figure 7A). This was not because kidneys (Supplemental Figure 6). Altogether, a total of 1299 of a greater number of mitochondria, as quantification of the unique succinylation sites were identified in kidney (Supple- Tomm20-positive mitochondria (Figure 7B) and the mito- mental Table 2A). Not all sites were present in all groups. In chondrial-to-nuclear DNA ratio (Figure 7C) indicated similar WT mice, there were 489 unique succinylated peptides in un- mitochondrial abundance across genotypes. Consistently, injured contralateral kidneys and 731 in the kidneys that had baseline FAO rates were trending to be increased (P=0.10) in 2 2 been subjected to 22 minutes of renal IRI 7 days previously cultured Sirt5 / PTECs (Supplemental Figure 7). We also 2 2 (Supplemental Table 2, B and C). In Sirt5 / mice the number analyzed respiratory chain function, previously shown to be 2 2 of unique succinylated peptides in contralateral and injured impaired in Sirt5 / liver mitochondria,31 in kidney lysates kidneys was 1009 and 905, respectively (Supplemental Table 2, using an Oroboros high-resolution respirometer. As in liver, 2 2 2 2 D and E). Thus, as expected, Sirt5 / kidneys had more over- uninjured Sirt5 / kidney mitochondria demonstrated a sig- all succinylated peptides than WT. However, although renal nificant decrease in respiration on the Complex II substrate IRI induced an increase in lysine succinylation in WT, the succinate (Figure 7, D and E). 2 2 2 2 opposite was seen in Sirt5 / kidneys. We next repeated the same set of experiments on Sirt5 / The small sample size (n=3/group) limited statistical power. and WT kidney tissue isolated day 7 after renal IRI. FAO rates, Therefore, statistical comparisons of site-level succinylation overall, were an order of magnitude lower in kidneys subjected 2 2 were limited to uninjured WT versus uninjured Sirt5 / , to 22 minutes of renal IRI compared with baseline (compare 2 2 and injured WT versus injured Sirt5 / kidneys. Not all suc- Figure 7F to Figure 7A). A more severe 30 minutes of renal IRI cinylated lysine residues are targets for Sirt5 deacylation13;we reduced FAO rates even further (Figure 7G), suggesting the defined Sirt5 target sites as those with statistically greater suc- pathway is highly sensitive to renal IRI. After either 22 or 2 2 2 2 cinylation levels in Sirt5 / kidneys compared with WT. In 30 minutes of renal IRI, Sirt5 / kidneys still displayed sig- uninjured kidneys, 191 Sirt5 target sites were identified, nificantly higher FAO than WT (Figure 7, F and G). Kidneys spread across 96 different proteins, whereas in injured kid- subjected to 22 minutes of renal IRI were also analyzed for neys, 201 sites were found across 96 proteins (Supplemental mitochondrial content and mitochondrial respiration. There Table 3, A and B). Overall, .95% of the succinylated kidney was a nonsignificant trend toward increased mitochondrial 2 2 proteins targeted by Sirt5 were mitochondrial proteins. abundance in Sirt5 / kidneys after renal IRI (P=0.06; Figure Reactome pathway analysis was performed on the lists of 6I), supported visually by immunostaining for Tomm20 Sirt5 target proteins in injured and uninjured kidney. The top (Figure 7H). Oroboros mitochondrial respiration measures pathway hit was mitochondrial FAO (Figure 5, B and C). A in kidney lysates, normalized to total protein, showed equal total of 17 FAO-related proteins were found to be Sirt5 targets mitochondrial respiratory capacity across genotypes.

JASN 30: 2384–2398, 2019 Sirt5 Deletion Protects from AKI 2391 BASIC RESEARCH www.jasn.org

AB Hypersuccinylated Pathways in Sirt5-/- Control Kidney SIRT5-Targeted Proteins Mitochondrial β-Oxidation Control Protein Localization TCA Cycle Fatty Acid Metabolism

18 78 18 Pyruvate Metabolism Mitochondrial Protein Import Branched-Chain Amino Acid Catabolism Cristae Formation Injured Electron transport Formation of ATP

0 51015 -Log (P value)

C Hypersuccinylated Pathways D in Injured Sirt5-/- Kidney FAO Succinylation β Mitochondrial -Oxidation 2000 TCA Cycle 1500 1000 Pyruvate Metabolism 500 Protein Localization 200 Cristae Formation 150 Fatty Acid Metabolism 100

Formation of ATP Control KO/WT Electron Transport 50

Mitochondrial Biogenesis 0 Mitochondrial Protein Import 0 50 100 150 200 Injured KO/WT 051015 -Log (P value)

2 2 Figure 5. Sirt5 / kidneys have increased succinylation of mitochondrial proteins. (A) Venn diagram: 96 hypersuccinylated proteins were quantiﬁed in contralateral uninjured kidneys and 96 in 22-minute renal IRI kidneys; 78 out of 96 were common between groups. (B and C) Reactome pathway analysis was performed using the lists of hypersuccinylated proteins; shown are the top ten hits for contralateral uninjured (A) and 22-minute renal IRI kidneys (B). (D) The degree of hypersuccinylation at speciﬁc lysine residues on FAO proteins tended to be higher in contralateral control (uninjured) kidneys than in 22-minute IRI kidneys; the graph depicts the 2 2 fold change in succinylation (Sirt5 / :WT) in control (uninjured) kidneys (y axis) plotted against the ratios for the same lysine residues in injured kidneys (x axis). The blue line represents a 1:1 relationship, i.e., if succinylation were equal on FAO proteins across control and post-IRI. It can be seen that most points lie above this line, indicating higher relative succinylation in the contralateral controls. n=3/group.

2/2 2 2 Increased FAO in Sirt5 Kidneys Is Localized to Recently, peroxisomal gain-of-function was reported in Sirt5 / Peroxisomes cell lines and mouse liver.13,21,31,43 We therefore interrogated 2 2 There are two parallel pathways of FAO, one in mitochondria Sirt5 / kidneys for peroxisomal FAO at baseline and after that directly transmits reducing equivalents into the electron renal IRI. The catabolism of 14C-palmitate was followed to 14 transport chain and acetyl-CoA into the TCA cycle, and one in CO2 and water-soluble (short-chain) products in kidney ho- peroxisomes that partially chain-shortens long-chain fatty acids mogenates in the presence of etomoxir, an irreversible inhibitor and releases short-chain products into the cytosol. These short- of mitochondrial FAO. Etomoxir-resistant FAO is ascribed to 2 2 chain products can either be taken up by mitochondria and peroxisomes.8 Sirt5 / kidney homogenates at baseline metabolized to completion or released from the cell. PTECs con- displayed a signiﬁcantly higher rate of peroxisomal FAO than tain an abundance of peroxisomes and increased peroxisomal WT kidney homogenates (Figure 8A). A similar result was ob- function has been previously linked to renoprotection.41,42 tained using freshly isolated primary mouse PTECs (Figure 8B)

2392 JASN JASN 30: 2384–2398, 2019 www.jasn.org BASIC RESEARCH

Medium-chain FA Long-chain Acyl-CoA

Sir5–/–/WT Sir5–/–/WT Carnitine Palmitoyl- CONTRA- transferase-1 PROTEIN SITE LATERAL INJURED MITOCHONDRIAL MEMBRANE Sir5–/–/WT Sir5–/–/WT ACSM1 534 43.6*** 16.0** ACSM2 526 84.8*** 51.7** CONTRA- Carnitine PROTEIN SITE LATERAL INJURED 533 207.6*** 59.5*** Acyl-CoA Palmitoyl- 540 189.3** 13.3*** Synthetase CPT2 544 5.2* 544 16.5*** 16.2*** transferase-2 558 31.5*** O 570 1826.1*** 191.7***

RCH2 CH2 C SCoA

ACADM 179 0.4* ETFA 225 200.8* 86.1** Electron FAD Acyl-CoA 212 10.8* ETFB 248 16.7** Transfer Dehydrogenase ACADV 483 24.6** 13.4** 250 32.0** Flavoprotein FADH2 ACD10 1056 535.5* 22.9*

DECR 260 6.1* O ECH1 316 58.2* 19.7** ECI1 255 7.6* 3.4* RCH CH C SCoA 275 22.3* 9.2** ECI2 159 9.2* 3.1* ECHA 60 36.3** 11.0** Enoyl-CoA 214 11.8* 7.1** ECHM 41 3.1* Hydratase 289 14.8** 43 5.0* 295 10.8** 10.5** 101 65.0** 11.7** 350 24.2** 282 107.2** 71.5** OH O 351 18.2** 284 146.2** 353 25.1* 11.1** 406 32.4** RCH CH2 C SCoA 55.5* 411 41.2** 11.7** + 413 11.7** HCDH 81 757.1* 194.8** NAD 3-Hydroxyacyl-CoA 414 167.6* 57.0* 87 21.3** Dehydrogenase 415 57.0* 89 31.9* NADH 644 31.2** 185 0.47* 728 4.2* 241 0.44* O O 311 114.21** 312 38.8** R C CH2 C SCoA

ECHB 278 6.2*** 3.5* THIM 209 18.2* 10.4* HSCoA 3-Ketoacyl-CoA 189 7.5** Thiolase 214 2.8*

* q <0.05 | ** q <0.01 | *** q <0.001 O O

3 CCH SCoA + R C SCoA

2 2 Figure 6. Mitochondrial FAO enzymes are hypersuccinylated in Sirt5 / kidney. Schematic showing the steps of mitochondrial FAO and the relative enrichment in lysine succinylation on FAO enzymes as determined by mass spectrometry. For each protein, the lysine 2 2 residues with statistically significant differences in succinylation are listed and the relevant fold-increases are provided (ratio of Sirt5 / :WT 2 2 signal) provided. Two statistical comparisons were made: uninjured contralateral control (WT versus Sirt5 / , n=3) and 22-minute IRI (WT 2 2 versus Sirt5 / , n=3/group). and a similar trend was observed using cultured hPTECs Protein 70–positive peroxisomes in baseline and postrenal IRI with Sirt5 siRNA knockdown (Supplemental Figure 8). tissues (Figure 8, E and F). To test whether inhibition of After a severe 30 minutes of renal IRI, peroxisomal FAO peroxisomal FAO abrogated the beneficial effect of Sirt5 inhi- 2 2 was seven-fold higher in Sirt5 / kidney homogenates bition, we used Sirt5 siRNA in cultured hPTECs 610,12- (Figure 8C). The 30-minute renal IRI homogenates were tricosadiynoic acid, an irreversible peroxisomal FAO inhibitor, Western-blotted for ACOX1, the rate-limiting enzyme of during in vitro ischemic AKI (Figure 8G). The combined peroxisomal FAO. The abundance of ACOX1 was significantly Sirt5/peroxisomal FAO inhibition removed the protective 2 2 2 2 higher in Sirt5 / kidney (Figure 8D). Sirt5 / kidneys were role of Sirt5 knockdown confirming the role of peroxisomal confirmed to have greater abundance of Peroxisomal Membrane FAO switching in driving the PTECs protection. These data

JASN 30: 2384–2398, 2019 Sirt5 Deletion Protects from AKI 2393 BASIC RESEARCH www.jasn.org

A Baseline FAO B C Baseline mtDNA 60 * Baseline - Tomm20/LTL/DAPI 3

40 2

20 1 mtDNA/nDNA pmoles/mg/hr

0 WT Sirt5-/- WT Sirt5-/- 0 WT Sirt5-/-

D High-Resolution Respirometry E Baseline Respiration F 22-Min IRI FAO G 30-Min IRI FAO 1.0 Succ 1.2 4 0.5 WT Rot * * 0.4 * Sirt5-/- 3 0.8 0.3 0.5 2 0.2 ADP Glut 0.4 pmoles/ug/hr Pyr/Mal 1 pmoles/ug/hr

flux per mg protein 0.1 flux per mg protein 2

Basal 2 O

0.0 O 0.0 0 0.0 -/- -/- -/- -/- 0 1020 30 WT Sirt5 WT Sirt5 WT Sirt5 WT Sirt5 Time [min] Complex I Complex II

H I 22-Min IRI mtDNA J Respiration 22-min IRI 22-Min IRI - Tomm20/LTL/DAPI 5 1.2

4 P=0.055 0.8 3

2 0.4

mtDNA/nDNA 1 flux per mg protein 2 O WT -/- 0 0.0 Sirt5 WT Sirt5-/- WT Sirt5-/- WT Sirt5-/- Complex I Complex II

2 2 2 2 Figure 7. FAO is enhanced in Sirt5 / kidneys. (A) Total FAO is increased at baseline in whole kidney homogenates of Sirt5 / mice compared with WT, as measured by 14C-palmitate oxidation. (n=3/group). (B) Immunostaining reveals equivalent numbers of Tomm20-positive 2 2 mitochondria (red) in WT and Sirt5 / PTECs (LTL, green) at baseline. Scale bar, 50 mm (C). Mitochondrial DNA (mtDNA) measured by 2 2 real-time PCR was found to be equivalent between control and Sirt5 / kidneys (n=3–4/group). (D and E) Baseline respirometry re- 2 2 vealed that Complex II speciﬁcally and not Complex I of the electron transport chain was reduced in Sirt5 / kidneys compared with 2 2 WT (n=6–7/group). (F and G) After a moderate (22 minutes) (F) or severe (30 minutes) (G) renal IRI, Sirt5 / kidneys displayed an 2 2 increased rate of FAO (n=3–4/group). (H) After renal IRI, mitochondria (Tomm20, red) appear to be present in both WT and Sirt5 / 2 2 proximal tubules (LTL, green). Scale bar, 50 mm. (I) Mitochondrial DNA trended to be increased in Sirt5 / kidneys compared with WT. (n=3/group). (J) No difference was observed in Complex I or Complex II utilization of the electron transport chain after renal IRI (n=3/group). All data are presented as dot plots plus mean6SEM. *P,0.05, t test.

2 2 confirm that Sirt5 / kidneys undergo an FAO switch from that the beneficial nature of Sirt5 loss-of-function stems from mitochondria to peroxisomes. the unique combination of only a minor loss of mitochondrial function combined with a significant peroxisomal gain-of- function. Deletion of Sirt5 in kidney caused only a modest DISCUSSION reduction in mitochondrial function at baseline, i.e.,approximately 10% loss of succinate-induced respiration and a Current study interrogated the role of Sirt5 during AKI suppression of mitochondrial FAO, and after injury, mito- and found that loss-of-function of Sirt5 was renoprotective chondrial respiratory chain function was the same across (Figure 9). This finding is opposite to those observed for Sirt1 genotypes. This could be because of the clear accumulation and Sirt3, for which deletion exacerbates AKI.16,44 We propose of protein succinylation in WT mitochondria after injury

2394 JASN JASN 30: 2384–2398, 2019 www.jasn.org BASIC RESEARCH

Baseline Baseline Tubule 30-Min IRI ADBC -/- Peroxisomal FAO Peroxisomal FAO Peroxisomal FAO WT Sirt5 15 8 0.5 ACOX1

0.4 * * 6 * GAPDH 10 0.3 4 ACOX1 0.2 5 1.5

pmoles/ug/hr pmoles/ug/hr 2 pmoles/ug/hr 0.1 * 0 0 0.0 1.0 WT Sirt5-/- WT Sirt5-/- WT Sirt5-/-

E Baseline - Pmp70/LTL/DAPI 0.5 ACOX1/GAPDH

0.0 WT Sirt5-/-

In vitro ischemic-AKI G via CGOD Human PTEC n.s. WT Sirt5-/- **

F 22-Min IRI - Pmp70/LTL/DAPI 80 **** ***

20 % Cytotoxicity 0 Sirt5 siRNA --++ TDYA -+-+ WT Sirt5-/-

Figure 8. Increased FAO by Sirt5 loss-of-function is localized to peroxisomes. (A–C) Etomoxir-insensitive 14C-palmitate oxidation, which represents peroxisomal FAO, was measured in kidney homogenates at baseline (n=3/group) (A), in freshly isolated PTECs (n=3/group) (B), or after 30-minute renal IRI (n=3/group) (C), using t test. (D) Western blot analysis showed an upregulation of the rate- 2 2 limiting peroxisomal FAO enzyme ACOX1 in Sirt5 / kidney homogenates, using Mann–Whitney U test. (E and F) PMP70 (red) is a peroxisome marker, LTL (green) is a marker for PTECs and DAPI (blue). (E) shows baseline and (F) shows postrenal IRI; (n=4/group). At baseline (E) there appear to be more peroxisomes present in the proximal tubules (arrowheads) and this persists after injury. (F) Scale bar, 50 mm. (G) hPTECs was treated 6Sirt5 siRNA and 610,12-tricosadiynoic acid (TDYA; ACOX1 inhibitor) during CGOD-mediated in vitro ischemic AKI and was evaluated its effect by LDH release (cell death marker). TDYA treatment abrogates the protective effect by Sirt5 siRNA (n=3/group, Tukey comparison). All data are presented as dot plots plus mean6SEM. *P,0.05; **P,0.01; ***P,0.001.

(731 succinylated peptides versus 489 at baseline), such that renoprotection in mice treated with etomoxir, in transgenic more than 50 peptides exhibited significantly higher succiny- mice overexpressing Sirt1 in the kidney, and in animals 2 2 lation in WT kidney compared with Sirt5 / kidney after treated with fibrates, a class of drugs that promotes perox- renal IRI (Supplemental Table 3C). Protein succinylation in isomal proliferation.46 mitochondria is determined by the concentration of succinyl- PTECs are the major site of injury during several forms of CoA, which chemically reacts with lysine residues on the sur- AKI, and are rivaled only by the liver in terms of peroxisome face of proteins, and the countering activity of Sirt5.45 abundance.12 As previously reported for liver,21 we observed a We speculate that WT kidneys rely more upon mitochondria peroxisomal gain-of-function, at least for the FAO pathway, in 2 2 2 2 during AKI than Sirt5 / kidneys, causing accumulation of Sirt5 / kidneys. Peroxisomes are also known to play an im- 2 2 succinyl-CoA and therefore greater protein succinylation, ef- portantroleinthebrain.47 Interestingly, Sirt5 / mice are fectively closing the gap in mitochondrial function between protected from ischemic stroke,48 whereas in the heart, Sirt5 2 2 WT and Sirt5 / kidneys. Peroxisomal function, in contrast, deletion exacerbates ischemic injury.49 The heart contains 2 2 remains enhanced in Sirt5 / kidney after injury. Increased only “microperoxisomes,” which are not thought to contrib- peroxisomal function has previously been linked to ute significantly to FAO.50 We speculate that the capacity for

JASN 30: 2384–2398, 2019 Sirt5 Deletion Protects from AKI 2395 BASIC RESEARCH www.jasn.org

Fatty Acids humans because of a switch away from mitochondrial FAO toward peroxisomal FAO. Our ﬁndings uncover an attractive and novel candidate pathway modulated by Sirt5 for the treatment of AKI.

Fatty Acids ACKNOWLEDGMENTS

Chiba, Peasley, and Cargill designed and performed experiments, pro- S S vided intellectual input, and wrote the manuscript. Maringer, Bharathi, FAO S S FAO Mukherjee, Zhang, and Yagobian performed experiments and provided S intellectual input. Holtz and Schilling coordinated sample preparation Mitochondria Peroxisome for mass spectrometric analysis followed by data acquisition. Holtz, Basisty, and Schilling processed and interpreted mass spectrometry data and generated data results and graphical displays. Goetzman and Sims- Acetate Lucas designed and performed experiments, edited the manuscript, Secretion Medium-chain FA provided intellectual input, and oversaw the project. All authors approved the ﬁnal version of this manuscript. FAO “Switching”: We would like to give special thanks to Dr. David R. Emlet for PTEC cultures, to the Center for Biologic Imaging for training Oxygen consumption and instrumentation of the tiling imaging, and to Michele Mulkeen Oxidative stress at the Histology Core Laboratory for histology training, all at the University of Pittsburgh. Injury

2 2 Figure 9. Proposed model by which FAO switching in Sirt5 / DISCLOSURES kidney mediates protection against AKI. Sirt5 loss-of-function blocks mitochondrial FAO via hypersuccinylation and thus re- Dr. Sims-Lucas reports grants from the National Institute of Diabetes and duced function of FAO key enzymes in the PTECs (see Figure 6). Digestive and Kidney Diseases, during the conduct of the study. Metabolic adaptation to blocked mitochondrial FAO in PTECs involves compensatory FAO in the peroxisome, thereby reducing oxygen requirements, reducing oxidative stress, and leading to FUNDING protection against injury. S, hypersuccinylation. This work was supported by the National Institute of Diabetes and Digestive and Kidney Diseases R01 DK090242 (Dr. Sims-Lucas) and R56 DK121758-01 peroxisomal FAO might explain why the kidney and brain, but (Dr. Goetzman). The authors acknowledge support from the National Insti- not the heart, are privy to the protective effects of Sirt5 de- tutes of Health shared instrumentation grant for the TripleTOF system at fi the Buck Institute (1S10 OD016281 to Dr. Bradford Gibson). Dr. Basisty ciency. Also, defective FAO in PTECs is responsible for in- was supported by a postdoctoral fellowship from the Glenn Foundation terstitial fibrosis.51 We showed that although mRNA for two for Medical Research. Dr. Chiba was supported by the American Heart Asso- 2 2 fibrosis markers are decreased in Sirt5 / kidneys, no discern- ciation postdoctoral fellowship (17POST33670685). able difference was observed in tissue across genotypes. Thus, future studies will evaluate whether Sirt5 loss-of-function leads to fibrosis at later time points (28–42 days post AKI) SUPPLEMENTAL MATERIAL with more severe ischemic insult, known to drive fibrosis.52 Therapeutically, because of the deleterious effect on the This article contains the following supplemental material heart, inhibition of Sirt5 would need to be localized to the online at http://jasn.asnjournals.org/lookup/suppl/doi:10.1681/ kidney. Recent advances have been made in targeting drug ASN.2019020163/-/DCSupplemental. delivery to the kidney.53,54 Although our data in the Sirt5 het- Supplemental Appendix 1. Supplemental materials and methods. erozygous mice is therapeutically promising as a half gene Supplemental Figure 1. Localization of Sirt5 in the kidney. 2 dosage is sufficient to elicit protection and would likely Supplemental Figure 2. Heterozygous deletion of Sirt5 (Sirt5+/ ) not lead to off-target effects. Further, because of the unique kidneys are protective against ischemic AKI at 7 days after injury. 2 2 substrate specificity of Sirt5, several Sirt5-specific inhibitors Supplemental Figure 3. Sirt5 / kidneys may reduce interstitial have been developed that demonstrate in vitro efficacy.55,56 fibrosis 7 days after ischemic AKI. 2 2 In conclusion, our results demonstrated that Sirt5 loss-of- Supplemental Figure 4. Sirt5 / proximal tubules is more protective function ameliorated ischemic and cisplatin AKI in mice and than B6/129- or 129-strain WT cells against hypoxic insult.

2396 JASN JASN 30: 2384–2398, 2019 www.jasn.org BASIC RESEARCH

Supplemental Figure 5. PTECs derived from heterozygous Sirt5 16. Morigi M, Perico L, Rota C, Longaretti L, Conti S, Rottoli D, et al.: Sirtuin 2 knockout kidneys (Sirt5+/ ) are protective against cisplatin injury. 3–dependent mitochondrial dynamic improvements protect against – Supplemental Figure 6. Quantitative mass spectrometry workflow acute kidney injury. JClinInvest125: 715 726, 2015 17. Sack MN: Emerging characterization of the role of SIRT3-mediated summary. mitochondrial protein deacetylation in the heart. Am J Physiol Heart Supplemental Figure 7. FAO is enhanced in primary mouse PTECs Circ Physiol 301: H2191–H2197, 2011 2 2 derived from Sirt5 / mouse. 18. Tan M, Peng C, Anderson KA, Chhoy P, Xie Z, Dai L, et al.: Lysine Supplemental Figure 8. Trend in increased FAO by Sirt5 knock- glutarylation is a protein posttranslational modification regulated by – down is localized to peroxisomes. SIRT5. Cell Metab 19: 605 617, 2014 19. Du J, Zhou Y, Su X, Yu JJ, Khan S, Jiang H, et al.: Sirt5 is a NAD- Supplemental Table 1. Primer sequences for real-time PCR dependent protein lysine demalonylase and desuccinylase. Science analysis. 334: 806–809, 2011 Supplemental Table 2. Relative quantification for succinylation 20. Park J, Chen Y, Tishkoff DX, Peng C, Tan M, Dai L, et al.: SIRT5-mediated sites identified in mouse kidneys. lysine desuccinylation impacts diverse metabolic pathways. Mol Cell 50: – Supplemental Table 3. Changes in succinylation sites identified in 919 930, 2013 21. Chen XF, Tian MX, Sun RQ, Zhang ML, Zhou LS, Jin L, et al.: SIRT5 in- mouse kidneys. hibits peroxisomal ACOX1 to prevent oxidative damage and is down- regulated in liver cancer. EMBO Rep 19: e45124, 2018 22. Lombard DB, Alt FW, Cheng HL, Bunkenborg J, Streeper RS, Mostoslavsky R, et al.: Mammalian Sir2 homolog SIRT3 regulates global REFERENCES mitochondrial lysine acetylation. Mol Cell Biol 27: 8807–8814, 2007 23. Skrypnyk NI, Harris RC, de Caestecker MP: Ischemia-reperfusion model 1. Susantitaphong P, Cruz DN, Cerda J, Abulfaraj M, Alqahtani F, of acute kidney injury and post injury fibrosis in mice. J Vis Exp 78: Koulouridis I, et al.; Acute Kidney Injury Advisory Group of the Ameri- 50495, 2013 can Society of Nephrology: World incidence of AKI: A meta-analysis. 24. Xu Y, Ma H, Shao J, Wu J, Zhou L, Zhang Z, et al.: A role for tubu- Clin J Am Soc Nephrol 8: 1482–1493, 2013 lar necroptosis in cisplatin-induced AKI. J Am Soc Nephrol 26: 2. Bonventre JV, Weinberg JM: Recent advances in the pathophysiology 2647–2658, 2015 of ischemic acute renal failure. JAmSocNephrol14: 2199–2210, 2003 25. Emlet DR, Pastor-Soler N, Marciszyn A, Wen X, Gomez H, Humphries 3. Chevalier RL: The proximal tubule is the primary target of injury and WH 4th, et al.: Insulin-like growth factor binding protein 7 and tissue progression of kidney disease: Role of the glomerulotubular junction. inhibitor of metalloproteinases-2: Differential expression and secre- Am J Physiol Renal Physiol 311: F145–F161, 2016 tion in human kidney tubule cells. Am J Physiol Renal Physiol 312: 4. Malek M, Nematbakhsh M: Renal ischemia/reperfusion injury; from F284–F296, 2017 pathophysiology to treatment. J Renal Inj Prev 4: 20–27, 2015 26. Xie LP, Zheng XY, Qin J, Tong YY: Amino acids protects against renal 5. Miller RP, Tadagavadi RK, Ramesh G, Reeves WB: Mechanisms of cis- ischemia-reperfusion injury and attenuates renal endothelin-1 disorder platin nephrotoxicity. Toxins (Basel) 2: 2490–2518, 2010 in rats. Chin J Traumato 7: 87–90, 2004 6. Lopaschuk GD, Ussher JR, Folmes CD, Jaswal JS, Stanley WC: 27. Weinberg JM: The effect of amino acids on ischemic and toxic injury to Myocardial fatty acid metabolism in health and disease. Physiol Rev 90: the kidney. Semin Nephrol 10: 491–500, 1990 207–258, 2010 28. Di Giovanni V, Walker KA, Bushnell D, Schaefer C, Sims-Lucas S, Puri P, 7. Fragasso G, Palloshi A, Puccetti P, Silipigni C, Rossodivita A, Pala M, et al.: Fibroblast growth factor receptor-Frs2a signaling is critical for et al.: A randomized clinical trial of trimetazidine, a partial free fatty acid nephron progenitors. Dev Biol 400: 82–93, 2015 oxidation inhibitor, in patients with heart failure. JAmCollCardiol48: 29. Mukherjee E, Maringer K, Papke E, Bushnell D, Schaefer C, Kramann R, 992–998, 2006 et al.: Endothelial marker-expressing stromal cells are critical for kidney 8. Portilla D, Dai G, Peters JM, Gonzalez FJ, Crew MD, Proia AD: Etomoxir- formation. Am J Physiol Renal Physiol 313: F611–F620, 2017 induced PPARalpha-modulated enzymes protect during acute renal 30. Bligh EG, Dyer WJ: A rapid method of total lipid extraction and puri- failure. Am J Physiol Renal Physiol 278: F667–F675, 2000 fication. Can J Biochem Physiol 37: 911–917, 1959 9. Peeters A, Shinde AB, Dirkx R, Smet J, De Bock K, Espeel M,et al.: 31. Zhang Y, Bharathi SS, Rardin MJ, Lu J, Maringer KV, Sims-Lucas S, et al.: Mitochondria in peroxisome-deficient hepatocytes exhibit impaired Lysine desuccinylase SIRT5 binds to cardiolipin and regulates the respiration, depleted DNA, and PGC-1a independent proliferation. electron transport chain. J Biol Chem 292: 10239–10249, 2017 Biochim Biophys Acta 1853: 285–298, 2015 32. Gillet LC, Navarro P, Tate S, Röst H, Selevsek N, Reiter L, et al.: Targeted 10. Fransen M, Lismont C: Peroxisomes and cellular oxidant/antioxidant data extraction of the MS/MS spectra generated by data-independent balance: Protein redox modifications and impact on inter-organelle acquisition: A new concept for consistent and accurate proteome communication. Subcell Biochem 89: 435–461, 2018 analysis. MolCellProteomics11: O111.016717, 2012 11. Fransen M, Lismont C, Walton P: The peroxisome-mitochondria con- 33. Meyer JG, D’Souza AK, Sorensen DJ, Rardin MJ, Wolfe AJ, Gibson BW, nection: How and why? Int J Mol Sci 18:E1126,2017 et al.: Quantification of lysine acetylation and succinylation stoichi- 12. Vasko R: Peroxisomes and kidney injury. Antioxid Redox Signal 25: ometry in proteins using mass spectrometric data-independent ac- 217–231, 2016 quisitions (SWATH). J Am Soc Mass Spectrom 27: 1758–1771, 2016 13. Rardin MJ, He W, Nishida Y, Newman JC, Carrico C, Danielson SR, 34. Schilling B, Rardin MJ, MacLean BX, Zawadzka AM, Frewen BE, et al.: SIRT5 regulates the mitochondrial lysine succinylome and met- Cusack MP, et al.: Platform-independent and label-free quantitation of abolic networks. Cell Metab 18: 920–933, 2013 proteomic data using MS1 extracted ion chromatograms in skyline: 14. Hirschey MD, Shimazu T, Goetzman E, Jing E, Schwer B, Lombard DB, Application to protein acetylation and phosphorylation. Mol Cell et al.: SIRT3 regulates mitochondrial fatty-acid oxidation by reversible Proteomics 11: 202–214, 2012 enzyme deacetylation. Nature 464: 121–125, 2010 35. Nakagawa T, Lomb DJ, Haigis MC, Guarente L: SIRT5 Deacetylates 15. Hasegawa K, Wakino S, Yoshioka K, Tatematsu S, Hara Y, Minakuchi H, carbamoyl phosphate synthetase 1 and regulates the urea cycle. Cell et al.: Kidney-specific overexpression of Sirt1 protects against acute 137: 560–570, 2009 kidney injury by retaining peroxisome function. JBiolChem285: 36. Sadhukhan S, Liu X, Ryu D, Nelson OD, Stupinski JA, Li Z, et al.: Me- 13045 –13056, 2010 tabolomics-assisted proteomics identifies succinylation and SIRT5 as

JASN 30: 2384–2398, 2019 Sirt5 Deletion Protects from AKI 2397 BASIC RESEARCH www.jasn.org

important regulators of cardiac function. Proc Natl Acad Sci U S A 113: 47. Berger J, Dorninger F, Forss-Petter S, Kunze M: Peroxisomes in 4320–4325, 2016 brain development and function. Biochim Biophys Acta 1863: 37. Ichimura T, Hung CC, Yang SA, Stevens JL, Bonventre JV: Kidney injury 934–955, 2016 molecule-1: A tissue and urinary biomarker for nephrotoxicant-induced 48. Diaz-Canestro C, Merlini M, Bonetti NR, Liberale L, Wüst P, Briand- renal injury. Am J Physiol Renal Physiol 286: F552–F563, 2004 Schumacher S, et al.: Sirtuin 5 as a novel target to blunt blood-brain 38. Bolignano D, Donato V, Coppolino G, Campo S, Buemi A, Lacquaniti A, barrier damage induced by cerebral ischemia/reperfusion injury. Int J et al.: Neutrophil gelatinase-associated lipocalin (NGAL) as a marker of Cardiol 260: 148–155, 2018 kidney damage. AmJKidneyDis52: 595–605, 2008 49. Hershberger KA, Abraham DM, Martin AS, Mao L, Liu J, Gu H, et al.: 39. Lu X, Li N, Shushakova N, Schmitt R, Menne J, Susnik N, et al.: C57BL/6 Sirtuin 5 is required for mouse survival in response to cardiac pressure and 129/Sv mice: Genetic difference to renal ischemia-reperfusion. overload. J Biol Chem 292: 19767–19781, 2017 JNephrol25: 738–743, 2012 50. Harrison EH, Walusimbi-Kisitu M: Properties and subcellular localiza- 40. Haun SE, Murphy EJ, Bates CM, Horrocks LA: Extracellular calcium is a tion of myocardial fatty acyl-coenzyme A oxidase. Am J Physiol 255: mediator of astroglial injury during combined glucose-oxygen depri- H441–H445, 1988 vation. Brain Res 593: 45–50, 1992 51. Kang HM, Ahn SH, Choi P, Ko YA, Han SH, Chinga F, et al.: Defective 41. Gulati S, Singh AK, Irazu C, Orak J, Rajagopalan PR, Fitts CT, et al.: fatty acid oxidation in renal tubular epithelial cells has a key role in Ischemia-reperfusion injury: Biochemical alterations in peroxisomes of kidney fibrosis development. Nat Med 21: 37–46, 2015 rat kidney. Arch Biochem Biophys 295: 90–100, 1992 52. Kramann R, Tanaka M, Humphreys BD: Fluorescence microangiography 42. Negishi K, Noiri E, Sugaya T, Li S, Megyesi J, Nagothu K, et al.: A role of for quantitative assessment of peritubular capillary changes after AKI in liver fatty acid-binding protein in cisplatin-induced acute renal failure. mice. J Am Soc Nephrol 25: 1924–1931, 2014 Kidney Int 72: 348–358, 2007 53. Williams RM, Shah J, Tian HS, Chen X, Geissmann F, Jaimes EA, et al.: 43. Guedouari H, Daigle T, Scorrano L, Hebert-Chatelain E: Sirtuin 5 pro- Selective nanoparticle targeting of the renal tubules. Hypertension 71: tects mitochondria from fragmentation and degradation during star- 87–94, 2018 vation. Biochim Biophys Acta Mol Cell Res 1864: 169–176, 2017 54. Hull TD, Agarwal A, Hoyt K: New ultrasound techniques promise 44. Gao R, Chen J, Hu Y, Li Z, Wang S, Shetty S, et al.: Sirt1 deletion leads to further advances in AKI and CKD. J Am Soc Nephrol 28: 3452– enhanced inflammation and aggravates endotoxin-induced acute 3460, 2017 kidney injury. PLoS One 9: e98909, 2014 55. Kalbas D, Liebscher S, Nowak T, Meleshin M, Pannek M, Popp C, et al.: 45. Wagner GR, Bhatt DP, O’Connell TM, Thompson JW, Dubois LG, Backos Potent and selective inhibitors of human sirtuin 5. JMedChem61: DS, et al.: A class of reactive acyl-CoA species reveals the non-enzymatic 2460–2471, 2018 origins of protein acylation. Cell Metab 25: 823–837.e8, 2017 56. Rajabi N, Auth M, Troelsen KR, Pannek M, Bhatt DP, Fontenas M, et al.: 46. Schoonjans K, Staels B, Auwerx J: Role of the peroxisome proliferator- Mechanism-based inhibitors of the human sirtuin 5 deacylase: Struc- activated receptor (PPAR) in mediating the effects of fibrates and fatty ture-activity relationship, biostructural, and kinetic insight. Angew acidsongeneexpression.J Lipid Res 37: 907–925, 1996 Chem Int Ed Eng 56: 14836–14841, 2017

2398 JASN JASN 30: 2384–2398, 2019