Abstract

Background

   Sirtuin 5 (SIRT5) is a NAD^+-dependent lysine deacylase. The SIRT5
   deficiency mouse model shows that it is dispensable for metabolic
   homeostasis under normal conditions. However, the biological role of
   SIRT5 and acylation in pathological states such as obesity and type 2
   diabetes (T2D) remains elusive.

Methods

   The hepatic SIRT5-overexpressing ob/ob mouse model (ob/ob-SIRT5 OE) was
   established by CRISPR/Cas9 gene editing tool Protein malonylation and
   succinylation lysine sites were identified by immunoprecipitation
   coupled lipid chromatography - tandem mass spectrometry (LC-MS/MS)
   methods.

Findings

   The ob/ob-SIRT5 OE mice showed decreased malonylation and
   succinylation, improved cellular glycolysis, suppressed
   gluconeogenesis, enhanced fatty acid oxidation, and attenuated hepatic
   steatosis. A total of 955 malonylation sites on 434 proteins and 1377
   succinylation sites on 429 proteins were identified and quantitated.
   Bioinformatics analysis revealed that malonylation was the major SIRT5
   target in the glycolysis/gluconeogenesis pathway, whereas succinylation
   was the preferred SIRT5 target in the oxidative phosphorylation
   pathway.

Interpretation

   Hepatic overexpression of SIRT5 ameliorated the metabolic abnormalities
   of ob/ob mice, probably through demalonylating and desuccinylating
   proteins in the main metabolic pathways. SIRT5 and related acylation
   might be potential targets for metabolic disorders.

Fund

   National Key R&D Program of China, the National Natural Science
   Foundation of China, the Strategic Priority Research Programs (Category
   A) of the Chinese Academy of Sciences, the Interdisciplinary Medicine
   Seed Fund of Peking University and the National Laboratory of
   Biomacromolecules.

   Keywords: Sirtuins, SIRT5, Malonylation, Succinylation, Metabolism,
   Proteomics
     __________________________________________________________________

Research in context.

Evidence before the study

   Evidences of the biological role of SIRT5 were mostly from studies of
   SIRT5 deficiency mouse models. Yu et al. observed little phenotype of
   SIRT5 deficiency mice under the basal high-fat diet (HFD) challenged
   conditions. Interestingly, Sadhukhan et al. and Hershberger et al. have
   demonstrated that SIRT5 is required for a normal cardiac function under
   the condition of aging or cardiac pressure overloading. At the
   molecular level, SIRT5 has been suggested to regulate biological
   processes of ammonia detoxification, glycolysis, fatty acid oxidation,
   and ketone body production. The studies were all done based on the
   mouse model of SIRT5 deficiency. Although Ogura et al. established
   SIRT5 Tg mice, they also generated it in the non-ob/ob mice.

Added value of this study

   The current studies established SIRT5 overexpression mouse model on the
   basis of ob/ob mouse strain. The findings here provide evidences to
   evaluate the role of SIRT5 under pathological conditions.

Implications of all the available evidence

   Summarizing all the available evidences about biological function SIRT5
   from studies of both deficiency and overexpression, normal and obesity
   mouse models, it seems that SIRT5 do play an important role in key
   metabolic biological processes, but in a chronic and humble way. Thus,
   long term observation and high sensitive detection methods are
   suggested in experiments designed to further explore the role of SIRT5
   in details.

   Alt-text: Unlabelled Box

1. Introduction

   Sirtuins are a family of conserved nicotinamide adenine dinucleotide
   (NAD)-dependent deacylases that play important roles in various
   physiological and pathological processes [[47]1]. There are seven
   sirtuins (SIRT1-7) in mammalian cells. SIRT1 to SIRT3 exhibit the most
   deacetylation activity and target a variety of proteins involved in
   cellular metabolism and energy homeostasis [[48]2,[49]3]. SIRT4 has
   been reported as a deacylase, and plays an important role in leucine
   metabolism and insulin secretion [[50]4]. SIRT6 possesses many
   enzymatic activities including ADP-ribosyltransferase [[51]5],
   deacetylase [[52]6], and long-chain deacylase [[53]7,[54]8]. Similar to
   SIRT6, SIRT7 hydrolyzes long-chain fatty acyl modifications but in an
   rRNA-dependent way [[55]9]. In addition, SIRT7 is also a protein
   deacetylase and desuccinylase [[56]10,[57]11].

   Unlike the aforementioned sirtuins, SIRT5 is unique for its robust
   lysine-demalonylase, -desuccinylase, and -deglutarylase activities
   [[58][12], [59][13], [60][14]]. Proteomic studies have identified
   thousands of malonylation, succinylation, and glutarylation lysine
   sites in both wild-type and SIRT5 KO mice [[61][14], [62][15],
   [63][16], [64][17], [65][18]]. Analyses of these acylation datasets
   demonstrated the enrichment of proteins in various metabolic pathways
   including glycolysis/gluconeogenesis, fatty acid β-oxidation, and
   oxidative phosphorylation. Many proteins in these metabolic pathways
   are targeted by SIRT5. Indeed, deficiency of SIRT5 in cells has been
   shown to inhibit glycolysis, suppress fatty acid β-oxidation, and
   attenuate ketone body production through regulating levels of
   malonylation or succinylation on key enzymes such as GAPDH [[66]18],
   VLCAD [[67]17], and HMGCS2 [[68]16]. Despite the well-studied
   biological roles of SIRT5 at molecular and cellular levels, SIRT5 seems
   to be dispensable for metabolic homeostasis in normal conditions at the
   whole animal level [[69]19]. However, recent studies have revealed that
   SIRT5 protects mice from ischemia-reperfusion injury, cardiac
   dysfunction induced by aging and cardiac-pressure overload [[70][20],
   [71][21], [72][22]]. This protective role is mediated by
   desuccinylation of proteins in fatty acid oxidation and oxidative
   phosphorylation pathways. Considering its widespread expression and
   large number of substrates, SIRT5 can function in more pathological
   conditions, especially in chronic metabolic diseases such as obesity
   and type 2 diabetes (T2D).

   Interestingly, proteomic hypermalonylation in ob/ob mice has been
   observed by our group and confirmed by Bowman et al. [[73]23,[74]24].
   Consistent with this observation, malonyl-CoA, the group donor of
   malonylation, has been shown to elevate in the skeletal muscle and the
   liver of OLETF rats and in muscle biopsies of patients with obesity and
   T2D [[75]25,[76]26]. Because SIRT5 plays pivotal roles in metabolic
   regulation and is the sole enzyme that catalyzes the demalonylation
   reaction, we hypothesized that SIRT5 and its deacylase activity might
   play an important role in insulin resistance of obesity and T2D
   [[77]27]. To address this hypothesis, we used the T2D ob/ob mouse
   model. An alternative Sirt5 gene was introduced into the mouse genome
   and specifically overexpressed in liver. Hepatic overexpression of
   SIRT5 significantly reduced levels of protein malonylation and
   succinylation. A series of biochemical, cellular biological and
   whole-animal characterization analyses were carried out.

2. Materials and methods

2.1. Antibodies and reagents

   Anti-acetyllysine (PTM-105), anti-malonyllysine (PTM-901),
   anti-succinyllysine (PTM-401), anti-glutaryllysine (PTM-1151),
   anti-succinyllysine conjugated agarose beads (PTM-402), and
   anti-malonyllysine conjugated agarose beads (PTM-904) were purchased
   from PTM Biolabs. SIRT5 (Cat# 15122-1-AP, RRID:[78]AB_2188778), NDUFS4
   (15849-1-AP), COXIV(Cat# 11242-1-AP, RRID:[79]AB_2085278), and GAPDH
   (Cat# 60004-1-Ig, RRID:[80]AB_2107436) antibodies were purchased from
   ProteinTech. ATP5A (Cat# ab14748, RRID:[81]AB_301447), UQCRC2 (Cat#
   ab14745, RRID:[82]AB_2213640), SDHB (Cat# ab14714, RRID:[83]AB_301432)
   and COXII (Cat# ab110258, RRID:[84]AB_10887758) were from Abcam. TOM20
   (Cat# 612278, RRID:[85]AB_399595) was from BD Biosciences. pCPT-cAMP
   (C3912-10MG), glucose detection kit (GAGO20-1KT), lactate detection kit
   (MAK064-1KT), β-hydroxybutyrate assay kit (MAK041) and free fatty acid
   quantitation kit (MAK044) were from SigmaAldrich. Neutral lipid probe
   LipidTOX ([86]H34475) was from Invitrogen. Sequencing-grade trypsin
   (V5113) was purchased from Promega. A list of other chemicals used for
   lipid chromatography – tandem mass spectrometry (LC-MS/MS) can be found
   in a previous paper [[87]23].

2.2. Animals and samples

   The original Sirt5 conditional knock-in C57BL/6 mouse was generated by
   Biocytogen (Beijing, China). Briefly, the targeting vector was
   constructed by inserting Sirt5 coding sequence into the pCAGGS vector
   which contains a CAG promoter, WPRE 3` UTR and Ploy A sequences. A
   floxed “Stop” elements were then inserted between CAG promoter and
   Sirt5 CDS region (Fig. S1A). After cloning the 5′ and 3′ homologous
   arms of the Hipp 11 locus on chromosome 11 into the vector, the vector
   was linearized and micro-injected into the mouse fertilized eggs
   together with the mRNA of CRISPR/Cas9 system. Hipp 11 locus has been
   shown to be a safe, intergenic and transcriptionally active region of
   chromosome 11, as the Rosa 26 locus [[88]28]. Correct insertion was
   confirmed by Southern blot experiments (Fig. S1C). A total of 11
   founders from C57BL/6 background have been identified and used for
   further breeding. To specifically overexpress Sirt5 gene in liver of
   ob/ob mouse, a sequential crossing procedure was done among homozygous
   Sirt5 KI mouse, heterozygous Alb-cre mouse and heterozygous ob/+ mouse
   ([89]Fig. 1B). All the mice are under C57BL/6 background. Two strains
   of mice were obtained under the procedure. One was the Sirt5 hepatic
   overexpressing (ob/ob-SIRT5 OE) mouse containing at least one allele of
   alternative Sirt5 gene and Alb-cre. The other one was control ob/ob
   mouse (ob/ob-Control) containing only the floxed Sirt5 KI allele. Males
   of those two strains at the age of 5 to 7 months were used for the
   experiences in the current work. Mouse studies were approved by the
   Animal Experimentation Ethics Committee (Institute of Biophysics, CAS)
   and the National Health and Medical Research Council of China
   Guidelines on Animal Experimentation.

Fig. 1.

   [90]Fig. 1
   [91]Open in a new tab

   Generation of ob/ob-SIRT5 OE mice. See also Fig. S1.

   (a) Strategy of an alternative Sirt5 allele knock-in by CRISPR/Cas9
   method.

   (b) Schematic diagram for breeding strategy to generate ob/ob-SIRT5 OE
   (OE) and ob/ob-Control mice (Ctl).

   (c) Examples of genotyping results.

   (d and e). Representative Immunohistochemistry (IHC) (d) and Western
   blot detection (e) of liver tissues from Ctl and OE mice determined
   with anti-SIRT5 antibody.

   (f and g) Western blot detection with pan anti- malonyllysine (f) and
   anti- succinyllysine (g) antibodies in the same samples as that of “e”.

2.3. Genotyping of the Ctl and OE mouse

   Genotype of the mouse was under the standard PCR procedure. Briefly,
   mice tails were incubated with lysis buffer (100 mM Tris-Cl pH 8·5,
   300 mM NaCl, 0·5 mM EDTA, 0·5% SDS, 0·2 mg/mL Protease K and 0·5 mg/mL
   RNase A) overnight at 55 °C. DNA was extracted by
   phenol/chloroform/isoamyl alcohol buffer and washed twice by 70%
   ethanol and dissolved in TE buffer (10 mM Tris-Cl pH 8·0, 0·5 mM EDTA).
   Primers used for genotyping include Alb-creF:
   5-GAAGCAGAAGCTTAGGAAGATGG-3, Alb-creR: TTGGCCCCTTACCATAACTG;
   Leptin-muF: 5-TGACCTGGAGAATCTCT-3, Leptin-muR: 5-CCTGAAGACCCCTGGAA-3;
   SIRT5 KIF: 5-AGTACCAGACTGCCCTGA-3, SIRT5KIR: 5-CACTCCCACTGTCCTTTC-3.
   Genotype were repeated before experiments.

2.4. Primary hepatocyte isolation

   Mice were anesthetized and abdominal cavity was dissected. The liver
   was perfused with KRG buffer (120 mM NaCl, 20 mM NaHCO3, 20 mM glucose,
   5 mM HEPES pH 7·4, 5 mM KCl, 1 mM MgSO[4], 1 mM KH[2]PO[4]) with 0·5 mM
   EGTA through the visceral vena cava. Two minutes later, when the blood
   drains out, the solution is switched to KRG buffer with 25 mg/mL
   collagenase type 4. The liver is cut to release the hepatocytes. The
   hepatocytes are washed with ice-cold DMEM and the suspension passed
   through a 75 μm cell strainer, and spun at 50 g for 1 min. The wash
   step is repeated until the supernatant appear clear after spinning. The
   pellet was suspended with full culture medium and transferred to
   culture plates.

2.5. Glucose detection, intraperitoneal glucose tolerance test (ipGTT)

   Serum glucose levels were detected by the Accu-Chek Blood Glucose Meter
   (Roche). Random glucose was detected at 11 am with normal diet. Fasting
   glucose was detected at 9 am after overnight fasting. Glucose levels of
   culture medium were detected by the Glucose Assay Kit (Sigma-Aldrich)
   following the manufacturer's instructions. For ipGTT, mice fasted 16 h.
   The baseline glucose levels (t = 0) were recorded using blood from the
   tip of mouse tail. 30 min later, glucose (2 g glucose/kg body weight)
   were injected. The blood glucose level was measured at 15, 30, 60, 90,
   120, 180, and 240 min (t = 15, t = 30, t = 60, t = 90, t = 120,
   t = 180, and t = 240).

2.6. Body component detection and respiratory rate (RER) determination

   Body components including fat mass and lean mass were detected by
   EchoMRI following the manufacture's instruction. VO[2], VCO[2], and RER
   were determined by the TSE Phenomaster System with the standard
   procedures. Mice were put into the cage 1 day before detection.
   Concentration of O[2], CO[2], and RER were detected and calculated
   every 6 min for 2 days. Five male mice per group were used in this
   experiment.

2.7. Soxhlet extraction of liver lipids

   Whole livers were isolated and dried at 50 °C for 7 days. The dried
   livers were weighted and put into Soxhlet extraction apparatus with the
   solvent of diethyl ether. The extraction lasts for 4 h at the
   temperature of 35 °C. After extraction, livers were dried for an
   additional hour to get rid of the residual solvent. Then, livers were
   weighted again. Total lipids (mainly triglyceride, TG) in the liver
   were the weight difference before and after extraction.

2.8. Cellular fractionation, western blotting, coomassie blue staining, and
immunofluorescence

   Tissue subcellular fractionation was performed following standard
   protocols [[92]29]. Western blotting and Coomassie blue staining
   analysis were carried out following protocols described previously
   [[93]23]. Immunofluorescence and confocal microscopy was performed as
   previously described [[94]30].

2.9. Pathway enrichment analysis

   Biological pathway enrichment for malonylated and succinylated proteins
   was analyzed by DAVID Web Services (DAVID 6·8). All malonylated and
   succinylated proteins and SIRT5-regulated malonylation and
   succinylation proteins (Ctl/OE Ratio ≥ 2, p-value<0·05) were submitted.
   The whole mouse proteomic was used as background. The default
   parameters were used for our analysis. Fisher exact p-value was used to
   estimate the enrichment of pathways. All the enrichment pathways were
   summarized in Table S3. Enrichment of main metabolic pathways including
   “Fatty acid degradation”, “Glycolysis / Gluconeogenesis”, “TCA cycle”,
   and “Oxidative phosphorylation” were selected and are shown in [95]Fig.
   2J and K.

Fig. 2.

   [96]Fig. 2
   [97]Open in a new tab

   Identification, quantification, and analysis of malonylome and
   succinylome.

   (a) Diagram of strategy for proteomic identification of malonylome and
   succinylome.

   (b and c) Fold change distribution of Kmal and Ksucc peptides between
   Ctl and OE samples. Green and red dots indicates SIRT5-regulated Kmal
   and Ksucc peptides with the threshold of ratio (Ctl/OE) ≥2 and p-value
   <0·05.

   (d and e) Number of Kmal and Ksucc sites (d) and proteins (e). Green
   and red column indicates number of SIRT5-regulated Kmal and Ksucc sites
   or proteins.

   (f and g) Distribution of number of total (f) and SIRT5-regulated (g)
   Kmal sites on proteins. Typical proteins with large number of modified
   lysine sites were listed.

   (h and i) Distribution of number of total (h) and SIRT5-regulated (i)
   Ksucc sites on proteins. Typical proteins with large number of modified
   lysine sites were listed.

   (j and k) Pathway enrichment of Kmal and Ksucc proteins. Four main
   metabolic pathways were selected to represent. Black column: total Kmal
   or Ksucc proteins. Green: SIRT5-regulated Kmal proteins. Red:
   SIRT5-regulated Ksucc proteins.

2.10. Tissue lysate preparation, tryptic digestion and affinity enrichment of
acylated peptides

   Four mice per group were scarified by cervical dislocation. Tissues
   were homogenized in 1 mL lysis buffer (50 mM Tris-Cl pH 7·4, 150 mM
   NaCl, 0·5 mM EDTA, 1 mM dithiothreitol, 1% Triton X-100, 0·5% sodium
   deoxycholate, and 0·1% sodium dodecyl sulfate). The homogenized tissues
   were centrifuged at 12,000 rpm and the supernatant containing the
   tissue lysates were precipitated with acetone and resolved in 8 M urea
   with 50 mM Tris-HCl, pH 8·0. The samples were reduced by 10 mM
   dithiothreitol, alkylated by 10 mM iodoacetamide (45 min at RT in the
   dark), and digested with sequencing-grade modified trypsin (Promega)
   with a substrate ratio of 1:50 (w/w) overnight at 37 °C. After
   digestion, the samples were desalted and fractionated by eluting with
   different concentration of acetonitrile (from 5% to 80%). The
   fractionated peptides were dried and resolubilized in NETN buffer
   (50 mM Tris–Cl, pH 8·0, 0·5% NP-40, 100 mM NaCl, and 1 mM EDTA).
   Anti-malonyllysine antibody-conjugated resin was first added and
   incubated at 4 °C for 12 h with gentle shaking. Then the samples were
   centrifuged at the speed of 2000 g for 2 min. The supernatant was
   collected and anti-succinyllysine antibody-conjugated resin was added
   and incubated at 4 °C for 12 h with gentle shaking to precipitate
   succinyllysine peptides ([98]Fig. 4A). Both anti-malonyllysine and
   anti-succinyllysine antibody-conjugated beads were washed two times
   with NETN buffer, once with ETN buffer (50 mM Tris–Cl, pH 8·0, 100 mM
   NaCl, and 1 mM EDTA) and once with purified water. The bound peptides
   were eluted by 0·1% trifluoroacetic acid and dried using the SpeedVac.
   The dried peptides were dissolved in 0·1% FA and desalted with C18
   ZipTips.

Fig. 4.

   [99]Fig. 4
   [100]Open in a new tab

   Hepatic Overexpression of SIRT5 in ob/ob Mice Reduce Liver Triglyceride
   (TG) Content.

   (a) Represent image of primary liver cells from Ctl and OE mice.
   Neutral lipid was stained with probe-Lipid TOX Green.

   (b) Represent hematoxylin and eosin (H&E) staining of liver tissues
   from Ctl and OE mice. White dots indicate lipid droplets in cells.

   (c and d) Liver lipids content of Ctl and OE mice were measured by the
   Soxhlet extractor method. Lipids were extracted by diethyl ether. The
   weight loss after diethyl ether extraction represent weight of total
   lipids in liver (c). The percentage of lipids in liver was calculated
   as weight of lipids divide by weight of dried liver (d). N = 11–12.

   (e) Weight of liver tissues from Ctl and OE mice after dried for
   7 days. Results are shown as mean ± SE. N = 11–12; *: P < 0·05.

   (f) Body weight of Ctl and OE mice at the age of 6 month. N = 15–19;
   NS: non-significant. (g-j) Serum levels of TG (g), Cholesterol (h),
   β-hydroxybutyrate (i), and FFA (j) of mice. N = 14–18.

   (k-l) Oxygen consumption and respiratory rate (RER) were detected at
   different time points for 24 h. N = 8.

2.11. Protein identification and quantification by LC-MS/MS

   The nanoLC-MS/MS experiments were performed on a Q Exactive (Thermo
   Scientific) equipped with an Easy n-LC 1000 HPLC system (Thermo
   Scientific). The labeled peptides were loaded onto a 100 μm id×2 cm
   fused silica trap column packed in-house with reversed phase silica
   (Reprosil-Pur C18 AQ, 5 μm, Dr. Maisch GmbH) and then separated on an a
   75 μm id×20 cm C18 column packed with reversed phase silica
   (Reprosil-Pur C18 AQ, 3 μm, Dr. Maisch GmbH). The peptides bounded on
   the column were eluted with a 135-min linear gradient. The solvent A
   consisted of 0·1% formic acid in water solution and the solvent B
   consisted of 0·1% formic acid in acetonitrile solution. The segmented
   gradient was 4–8% B, 5 min; 8–22% B, 90 min; 22–32% B, 22 min; 32–95%
   B, 3 min; 95% B, 15 min at a flow rate of 300 nL/min.

   The MS analysis was performed with Q Exactive mass spectrometer (Thermo
   Scientific). With the data-dependent acquisition mode, the MS data were
   acquired at a high resolution 70,000 (m/z 200) across the mass range of
   300–1600 m/z. The target value was 3·00E + 06 with a maximum injection
   time of 60 ms. The top 20 precursor ions were selected from each MS
   full scan with isolation width of 2 m/z for fragmentation in the HCD
   collision cell with normalized collision energy of 30%. Subsequently,
   MS/MS spectra were acquired at resolution 17,500 at m/z 200. The target
   value was 5·00E + 04 with a maximum injection time of 80 ms. The
   dynamic exclusion time was 40s. For nano electrospray ion source
   setting, the spray voltage was 2.0 kV; no sheath gas flow; the heated
   capillary temperature was 320 °C.

   The raw data from Q Exactive were analyzed with Proteome Discovery
   version 2·2·0·388 using Mascot (version 2·5·1) and Sequest HT search
   engine for protein identification and Percolator for false discovery
   rate (FDR) analysis. The Uniprot mouse protein database (updated on
   03-2017) was used for searching the data from mouse sample. Some
   important searching parameters were set as following: trypsin was
   selected as enzyme and two missed cleavages were allowed for searching;
   the mass tolerance of precursor was set as 10 ppm and the product ions
   tolerance was 0·02 Da.; the cysteine carbamidomethylation were
   specified as fixed modifications; The methionine oxidation and Lys
   malonylation (+86·0004 Da) or Lys succinylation (+100·0160 Da) were
   chosen as variable modifications. FDR analysis was performed with
   Percolator and FDR <1% was set for protein identification. The peptides
   confidence was set as high for peptides filter. Protein label-free
   quantification was also performed on Proteome Discovery 2·2·0·388 using
   the areas of identified peptides. Only unique and razor peptides of
   proteins were selected for protein relative quantification. The
   normalization to the protein median of each sample was used to correct
   experimental bias and the normalization mode was selected as total
   peptide amount. Protein ratios were calculated as median of all
   possible pairwise ratios. Missing values were imputed using low
   abundance resampling model.

3. Results

3.1. Generation of ob/ob-SIRT5 OE mice

   An alternative SIRT5 coding sequence was inserted into the Hipp 11
   locus on chromosome 11 of wild-type C57BL/6 mice by the CRISPR/Cas9
   method ([101]Fig. 1a). The success of correct knock-in was confirmed
   and off-target insertion was excluded by Southern blot experiments
   (Fig. S1a–c). Hepatic SIRT5-overexpressing ob/ob mice (ob/ob-SIRT5 OE)
   and the control mice (ob/ob-Control) were generated by sequential
   crossing among the Sirt5 allele knock-in mice (Sirt5 KI^flox/flox),
   Alb-cre mice (Alb-cre^+/−) and heterozygous ob/ob mice (ob/+)
   ([102]Fig. 1b). Genotyping demonstrated that both ob/ob-SIRT5 OE (OE)
   and ob/ob-Control (Ctl) mice contained the alternative Sirt5 allele,
   but only OE mice contained the Alb-cre allele ([103]Fig. 1c).
   Consistent with the genotyping results, overexpression of SIRT5 in
   liver samples of OE mice was also confirmed by immunohistochemistry
   ([104]Fig. 1d) and Western blot experiments ([105]Fig. 1e). There was a
   great decrease of total protein malonylation and succinylation in liver
   tissues from OE mice compared to that of Ctl mice ([106]Fig. 1f and g),
   whereas the total protein level and β-actin were comparable (Fig. S2h).
   Additionally, overexpression of SIRT5 was not observed in other tissues
   including white adipose tissue, heart, muscle, and spleen (Fig. S2a–d).
   Although there was a slight increase of SIRT5 in kidney of OE mice,
   levels of malonylation and succinylation in kidney were not
   significantly affected (Fig. S2f and g).

   To detect the localization of overexpressed SIRT5 proteins, we
   performed the immunofluorescence experiments and found that signal of
   SIRT5 proteins merged well with that of mitochondrial biomarker protein
   TOM20 (Fig. S1d), suggesting the mitochondrial localization of SIRT5.
   Because previous work showed that endogenous SIRT5 proteins localized
   in both mitochondria and cytosol [[107]15], we then further determined
   the localization of SIRT5 by cell fractionation and found that
   overexpressed SIRT5 proteins were also localized in the cytosol (Fig.
   S1e). Altogether, we overexpressed SIRT5 in liver of ob/ob mice, which
   significantly reduced levels of malonylation and succinylation on
   multiple proteins.

3.2. Identification, quantification, and analysis of the malonylome and
succinylome in response to hepatic overexpression of SIRT5 in ob/ob mice

   To identify the malonylome and succinylome in response to hepatic
   overexpression of SIRT5 in ob/ob mice, we developed a workflow
   involving enrichment of malonylated and succinylated peptides coupled
   with identification by tandem mass spectrometry ([108]Fig. 2a). With
   this approach we identified and quantified 1100 unique peptides
   corresponding to 955 lysine sites on 434 proteins for malonylation and
   1790 unique peptides corresponding to 1377 lysine sites on 429 proteins
   for succinylation with a FDR of <1% (Table S1). The peptides were
   quantitated by the PD2·4 Label-Free Quantitation Program which extracts
   and measures ion intensity chromatograms of MS1 scan. For each unique
   peptide, abundance ratios (Ctl/OE) and p-values were calculated between
   the four Ctl and four OE samples based on the quantification results.
   Levels of malonylation were down-regulated on 329 out of 1100 peptides
   with ratio (Ctl/OE) increase ≥2-fold and p-value <0·05 ([109]Fig. 2b).
   This represented 306 out of 955 (32%) malonylated lysine (Kmal) sites
   and 200 out of 434 (46%) malonylated proteins ([110]Fig. 2d and e).
   Meanwhile, levels of succinylation were down-regulated on 146 out of
   1790 peptides by the same threshold ([111]Fig. 2c), which corresponded
   to 123 out of 1377 (9%) succinylated lysine (Ksucc) sites, and 95 out
   of 429 (22%) succinylated proteins ([112]Fig. 2d and e). Although the
   abundance of malonylation and succinylation changed greatly from Ctl to
   OE mice, the abundance of proteins was not significantly altered (Table
   S1), for 99% of proteins (3098/3145) changed less than one fold among
   the 3145 quantified proteins (Fig. S3a).

   To analyze the distribution of malonylation and succinylation at
   protein level, we calculated the number of Kmal and Ksucc sites on each
   protein ([113]Fig. 2f–i). The data showed that some proteins were
   intensively modified. For example, 35 unique peptides corresponding to
   20 Kmal sites on ALDH1L1 were identified ([114]Figs. 2f and S3b), and
   63 unique peptides representing 36 Ksucc sites on CPS1 were found
   ([115]Figs. 2h and S3c). There were also proteins highly regulated by
   SIRT5. For instance, all seven identified malonylation sites on ECHA
   were targeted by SIRT5 ([116]Figs. 2g and S3d). Interestingly,
   Sadhukhan et al., recently reported that SIRT5 could regulate the
   activity of ECHA through desuccinylation [[117]20]. They identified 28
   succinylated lysine sites on ECHA, which was regulated by SIRT5. Six
   out of the seven malonylated lysine sites in this study were overlapped
   with the 28 succinylated lysine sites. These results suggested that
   demalonylation might be another mechanism that SIRT5 regulates the
   activity of ECHA. In addition, we compared the SIRT5-regulated
   malonylation dataset identified using hepatic SIRT5-overexpressing mice
   with a previous dataset identified using SIRT5-deficient mice
   [[118]18]. We found that 32 lysine sites overlapped between 306
   SIRT5-targeted sites in our overexpression dataset and 183
   SIRT5-regulated sites in the deficiency dataset (Table S2, Fig. S3e and
   f). We propose that these 32 lysine sites across 25 proteins were the
   most potential SIRT5-mediated demalonylation substrate sites. We also
   compared the malonylome and succinylome and found limited overlap
   between the two types of acylation (Fig. S4a and b). Most of the highly
   malonylated proteins (≥10 Kmal sites per protein) were slightly
   succinylated, whereas most of the heavily succinylated proteins (≥10
   Ksucc sites per protein) were moderately malonylated (Fig. S4c).
   Interestingly, all the highly malonylated proteins were localized in
   the cytosol, whereas almost all the heavily succinylated proteins were
   in mitochondria (Fig. S4c). These results suggest that the malonylome
   is significantly distinct from the succinylome.

3.3. Hepatic overexpression of SIRT5 in ob/ob mice demalonylates proteins in
the glycolysis/gluconeogenesis pathway to improve glucose metabolism, and
desuccinylates proteins in the oxidative phosphorylation pathway to
facilitate fatty acid metabolism

   To investigate the potential pathways regulated by SIRT5, pathway
   enrichment analysis was done using DAVID Web Services (Table S3)
   [[119]31]. For those SIRT5-regulated proteins, there was little
   enrichment for malonylated proteins on “Oxidative phosphorylation”
   pathway ([120]Figs. 2j and [121]3g), and reduced enrichment for
   succinylated proteins on the “Glycolysis/Gluconeogenesis” pathway
   ([122]Figs. 2k and [123]3a).

Fig. 3.

   [124]Fig. 3
   [125]Open in a new tab

   Hepatic overexpression of SIRT5 in ob/ob mice demalonylates proteins in
   the Glycolysis / Gluconeogenesis pathway to improve glucose metabolism,
   and desuccinylates proteins in the oxidative phosphorylation pathway to
   facilitate fatty acid metabolism.

   (a) Proteins and acylation sites regulated by SIRT5 overexpression in
   glycolysis pathway. Red: Ksucc sites; Green: Kmal sites.

   (b) Levels of lactic acid in culture medium of primary hepatocytes 24 h
   after isolation. Data are represented as mean ± SEM. N = 3; **:
   P < 0·01.

   (c) Extracellular acidification rate (ECAR) of primary hepatocytes 24 h
   after isolation. Glucose were provide as the only carbon source. N = 3;
   *: P < 0·05.

   (d) Levels of glucose in culture medium of primary hepatocytes after
   6 h incubating with glucose-free medium containing lactate, pyruvate
   and glutamine. pCPT: pCPT-cAMP is an PKA activator which promote
   gluconeogenesis. N = 6.

   (e and f) Fold change (Ctl/OE) of Kmal sites (Green), Ksucc sites (Red)
   and expression levels of related proteins (Gray) in glycolysis (e) and
   oxidative phosphorylation pathways (f). Several represent sites and
   proteins were marked.

   (g) Proteins and acylation sites regulated by SIRT5 overexpression in
   oxidative phosphorylation pathway. Red: Ksucc sites; Green: Kmal sites.

   (h) Oxygen consumption rate (OCR) detection of primary hepatocytes from
   Ctl and OE mice. Palmitoyl-cartinine or Succinate were used as
   substrates. Rot: rotenone; AA: antimycin A. Error bars were calculated
   based on 10 technical repeats for each pair of Ctl and OE mice. A total
   of 3 pairs of Ctl and OE mice were detected. One pair of representative
   results was present here.

   (i) Electron microscope observation of liver tissues from Ctl and OE
   mice with magnification of 30,000×. MT: mitochondria.(j) Western blot
   detection of five represent proteins from five complex (I-V) in
   oxidative phosphorylation pathway.

   To determine whether the Glycolysis/Gluconeogenesis pathway was
   authentically affected by SIRT5 overexpression, lactic acid was
   detected in the medium of hepatocytes derived from Ctl and OE mice. We
   found that the level of lactic acid in OE hepatocytes were higher than
   that of Ctl hepatocytes ([126]Fig. 3b). To confirm the result,
   extracellular acidification (ECAR) value was examined when glucose was
   provided as the only source for energy metabolism. We found that ECAR
   value in OE hepatocytes was also higher than that of Ctl hepatocytes
   ([127]Fig. 3c). In addition, we evaluated gluconeogenesis flux through
   comparing glucose level in culture medium of hepatocytes from Ctl and
   OE mice. The glucose-free culture medium contains lactate, pyruvate,
   and glutamine. After 6 h of culturing, we found that the level of
   glucose in OE hepatocytes was lower than that of Ctl hepatocytes, which
   indicated the inhibition of gluconeogenesis in OE mice ([128]Fig. 3d).
   To further confirm this, gluconeogenesis was enhanced by pCPT-cAMP (a
   PKA activator). The glucose levels in OE hepatocytes was still lower
   than that of Ctl hepatocytes ([129]Fig. 3d). Interestingly, most of the
   SIRT5-regulated proteins in the glycolysis/gluconeogenesis pathway
   showed significant decrease of malonylation ([130]Fig. 3a and e),
   whereas the expression level of these proteins remained unchanged
   ([131]Fig. 3e, gray dot). Collectively, these results suggested that
   overexpression of SIRT5 enhanced glycolysis and inhibited
   gluconeogenesis in the hepatocytes of ob/ob mice, probably through
   demalonylating proteins in this pathway.

   To examine whether fatty acid oxidation was affected by overexpressing
   SIRT5 in liver, we measured the oxygen consumption rate (OCR) of
   primary hepatocytes from Ctl and OE mice with palmitoyl-carnitine used
   as the substrate. We observed an increase of OCR value in primary
   hepatocytes from OE mice ([132]Fig. 3h). Similarly, there was a trend
   of increased succinate oxidation in primary hepatocytes from OE mice,
   although the increase was not significant ([133]Fig. 3h).
   Interestingly, there was a significant decrease of succinylation of
   many proteins in the oxidative phosphorylation pathway ([134]Fig. 3f
   and g). In the meantime, expression levels of these proteins were
   relatively unchanged ([135]Fig. 3f, gray dot). In addition, proteins
   levels of complexes in the oxidative respiratory chain remained
   comparable between Ctl and OE samples ([136]Fig. 3j). The morphology of
   mitochondria from the two types of samples was also similar ([137]Fig.
   3i). The results indicated that hepatic overexpression of SIRT5
   improved mitochondrial fatty acid oxidation of hepatocytes from ob/ob
   mice by desuccinylating proteins.

3.4. Hepatic overexpression of SIRT5 in ob/ob mice reduces liver TG content

   When preparing primary hepatocytes, we observed reduced lipid droplets
   in OE hepatocytes compared to Ctl under the microscope. To confirm this
   observation, primary hepatocytes were stained with a lipid droplet
   probe and observed under a confocal microscope. We discovered that
   lipid droplets in hepatocytes from OE mice were fewer and smaller than
   those of Ctl mice ([138]Fig. 4a). In addition, hematoxylin and eosin
   staining of liver tissue also exhibited a trend of decreased lipid
   droplets in OE mice ([139]Fig. 4b). To confirm this observation, we
   isolated the whole liver of each mouse and extracted total lipids via
   the Soxhlet extractor method using the solvent of diethyl ether.
   Diethyl ether is a strong solvent for many types of lipids including
   TG, free fatty acid, sterols, and various phospholipid. However, TG is
   the main content of lipids in the cell, especially in the liver with
   steatosis. We found that weight of lipids dropped from an average of
   0·92 g (39% of the dehydrated liver) per liver in Ctl mice to 0·68 g
   (33% of the dehydrated liver) in OE mice ([140]Fig. 4c and d).
   Furthermore, the weight of dehydrated liver in Ctl mice was 0·3 g
   heavier (average of 2·3 vs 2·0 g) than that of OE mice ([141]Fig. 4e).
   Meanwhile, the total body weight and body components remained unchanged
   ([142]Figs. 4f, S5a and b). Interestingly, we observed slightly
   elevated level of serum TG in OE mice ([143]Fig. 4g), and similar
   levels of cholesterol, β-hydroxybutyrate, free fatty acid (FFA),
   high-density lipoprotein (HDL), and low-density lipoprotein (LDL)
   ([144]Fig. 4h–j, S5e, and 5f). In addition, indirect calorimetry
   experiments showed that oxygen consumption (VO[2]), respiratory rate
   (RER = VCO[2] /VO[2]), and activity counts remained unchanged
   ([145]Fig. 4k, l, and S5 h). These results indicated that hepatic
   overexpression of SIRT5 in ob/ob mice could reduce TG content in liver
   and attenuated hepatic steatosis.

   To investigate glucose metabolism at the whole animal level, we
   examined the level of blood glucose after hepatic overexpression of
   SIRT5 and discovered that levels of blood glucose under both fed and
   fasted conditions were comparable (Fig. S5c and d). To further explore
   the effect of SIRT5 on regulation of glucose metabolism on the whole
   animal, the mice were challenged with 2 g glucose/kg body weight
   injected intraperitoneally and blood glucose was measured at different
   time-points. Results demonstrated that there was little improvement of
   glucose tolerance in OE mice (Fig. S5g). The aforementioned results
   suggested that hepatic overexpression of SIRT5 has limited effect on
   the glucose homeostasis at the whole animal level of ob/ob mice.

4. Discussion

   In the current study, we generated the ob/ob-SIRT5 OE mice, and
   observed elevated lactate and ECAR in SIRT5 OE hepatocytes. Considering
   most of the enzymes in the glycolysis pathway can be regulated by
   SIRT5, the elevation of lactate and ECAR suggested an improved
   glycolysis flux. These results are consistent with previous studies by
   Nishida et al. who showed that SIRT5 regulate glycolysis by
   demalonylation of related enzymes [[146]18]. However, an alternative
   explanation for elevation of lactate and ECAR might be the decreased
   metabolism of glucose-derived pyruvate. Actually, we detected enhanced
   oxidation of fatty acids after SIRT5 overexpression. Other groups have
   reported the role of SIRT5 in regulating fatty acid oxidation
   [[147]17,[148]20]. SIRT5 induced acceleration of fatty acid oxidation
   pathway might compete with glucose-derived pyruvate oxidation for
   energy production in the mitochondria. In this case, the elevation of
   lactate and ECAR could be explained by the inhibition of pyruvate
   oxidation caused by enhancement of fatty acid oxidation. However, the
   two explanations are not mutually exclusive. Both mechanisms contribute
   to the elevation of lactate and ECAR in the OE hepatocytes.

   Although we observed increased glucose production in SIRT5 OE
   hepatocytes, the whole organismal blood glucose levels and glucose
   tolerance were comparable between Ctl and OE mice. The complicated
   regulation of glucose metabolism might explain this discrepancy. For
   example, blood glucose levels are determined by intestinal absorption,
   hepatic gluconeogenesis, and cellular consumption in muscles. The
   metabolic improvement in hepatocytes could be overwhelmed by other
   organs at the whole-animal level. Additionally, the effect of hormones
   in systemic circulation cannot be ignored. Local levels of insulin and
   glucagon were capable of reversing some beneficial effects of SIRT5
   overexpression.

   We also found a decreased hepatic TG content, as well as increased
   serum TG level in OE mice. However, other lipid metabolites in serum
   such as FFA, cholesterol, and β-hydroxybutyrate remain comparable in
   Ctl and OE mice. The ob/ob mice synthesize a huge amount of fatty acid
   de novo from the carbohydrate in the normal chow diet, and transform
   them into TG. The hepatic generated TG can either stay in the liver or
   secrete into serum and be store in adipose tissue. Thus, the decreased
   hepatic TG content in OE mice might results from decreased de novo
   synthesis of TG, increased lipolysis and fatty acid oxidation, and
   enhanced secretion into the circulation system. Our observation
   suggested that decreased hepatic TG content was caused by increased
   fatty acid oxidation and enhanced secretion. SIRT5 has been shown to
   regulate the fatty acid oxidation pathway by many groups. Whether SIRT5
   plays an important role in accelerating export of TG from liver to
   blood remains to be investigated. An interesting cue for this
   hypothesis is FABPL which has been indicated as an SIRT5
   regulated-protein in both SIRT5KO and SIRT5OE data (Fig. S3e and f).
   FABPL is one of the high-abundance proteins involved in lipid transport
   in the liver. SIRT5-regulated FABPL might play a role in stimulating
   hepatic TG secretion.

   In addition to the ob/ob-SIRT5 OE mice, we also generated non-ob/ob
   mice with SIRT5 overexpression (WT-SIRT5 OE). Although there was a
   significant decrease of malonylation and succinylation in the liver
   tissue (Fig. S6a), no apparent phenotypes were observed in WT-SIRT5 OE,
   which is consistent with previous studies of SIRT5 KO mice. In
   addition, few histopathological changes have been observed (Fig. S6d).
   Thus, we did not perform the proteomic studies on non-ob/ob mice in
   current study. We analyzed previous malonylome and succinylome data
   identified in the SIRT5 KO non-ob/ob system. There were 120
   SIRT5-regulated malonylated and 140 SIRT5-regulated succinylated
   proteins identified in the non-ob/ob system [[149]16,[150]18]. Compared
   with 200 SIRT5-regulated malonylated and 95 SIRT5-regulated
   succinylated proteins in the ob/ob system, we found 55 and 44
   overlapped malonylated and succinylated proteins, respectively (Fig.
   S6b and c). There were 145 malonylated and 51 succinylated proteins
   specific regulated in the ob/ob system.

   The comparison between the malonylome and succinylome at the level of
   lysine sites and proteins revealed that these two types of acylation
   were quite different. The proportion of overlapped proteins that
   contained at least one malonylation-lysine site and one
   succinylation-lysine site was <50%. Malonylation proteins localize
   equally in the cytosol and mitochondria, but succinylation proteins
   distribute mostly in the mitochondria. The localization pattern is
   similar to the distribution model of their group donor molecules
   malonyl-CoA and succinyl-CoA. Malonyl-CoA was first recognized as the
   intermediate of the fatty acid synthesis pathway and is produced mainly
   in cytosol by the enzymes acetyl-CoA carboxylases 1 and 2 (ACC1/2). A
   recent study discovered that malonyl-CoA can also be synthesized
   directly from malonate by ACSF3 in the mitochondria of mammalian cells
   [[151]24]. The wide distribution of malonyl-CoA in both the cytosol and
   mitochondria might explain the malonylated protein localization. In
   contrast, succinyl-CoA is either mainly synthesized from
   α-ketoglutarate in the TCA cycle or alternatively synthesized from
   propionyl-CoA, the β-oxidation products of odd-numbered fatty acids.
   Both of the synthesizing reactions are in the mitochondria. The
   exclusive mitochondrial distribution of succinyl-CoA is consistent with
   the localization of succinylation proteins.

   In the present study, we enriched Kmal and Ksucc peptides in the same
   samples by the sequential immunoprecipitation approach. The
   disadvantage of the present method is that the first round of
   enrichment by Kmal antibody might affect the second round of Ksucc
   enrichment. Although antibodies for each of the acylation are
   acylation-specific, there might be some cross-reaction of the two
   antibodies because of the structure similarity of the two types of
   acylation. We found that the first immunoprecipitation of malonylated
   peptides induced an approximate 2·5% decrease of the total succinylated
   peptides identified in the second round of enrichment. Thus, the
   immunoprecipitation method might slightly affect the comparison between
   malonylation and succinylation of the same proteins.

   Acylation, like malonylation and succinylation, can spontaneously
   occurs [[152]32], which further indicates that malonylation and
   succinylation are strongly associated with local concentrations of
   malonyl-CoA and succinyl-CoA. Metabolic intermediates can locally
   modify proteins, most of which are metabolic enzymes. The physiological
   function of these reactions has not yet been elucidated. One
   explanation is the regulatory role of acylation, which functions as a
   feedback mechanism to control metabolic flux. The other possibility is
   the “carbon stress” hypothesis, which is similar to “oxidative stress”
   [[153]33]. According to the “carbon stress” hypothesis, metabolic
   intermediates such as malonyl-CoA and succinyl-CoA spontaneously modify
   proteins and lower the global quality of proteins. In many chronic
   metabolic diseases such as T2D and cardiovascular diseases,
   concentrations of intracellular metabolic intermediates are often
   higher than in healthy individuals. The high concentrations of
   intermediates modify proteins and disturb their normal functions.
   Evidence from our current study shows that high levels of sirtuins
   improve the activity of metabolic pathways, and reverse some abnormal
   phenotypes probably by lowering the modification of key metabolic
   enzymes. Our study here provides an alternative method to recognize and
   understand metabolic abnormality of obesity and T2D which would benefit
   the therapy of these diseases.

   The following are the supplementary data related to this article.

   Supplementary material, including Figure S1 to Figure S6.
   [154]mmc1.docx^ (3.9MB, docx)
   Table S1

   Mass spectrometry details for identification and quantification of
   Kmal, Ksucc peptides and total proteins.
   [155]mmc2.xlsx^ (1.9MB, xlsx)
   Table S2

   Comparison of Kmal sites between SIRT5 overexpression dataset (present
   work) and SIRT5 deficiency dataset (previous work of Nishida. et al.).
   [156]mmc3.xlsx^ (18.3KB, xlsx)
   Table S3

   KEGG pathway analysis of Kmal and Ksucc proteins.
   [157]mmc4.xlsx^ (119KB, xlsx)

Acknowledgments