https://orcid.org/0000-0003-3434-2394
Abstract
In diabetic neuropathy, there is activation of axonal and sensory neuronal degeneration pathways leading to distal axonopathy. The nicotinamide-adenine dinucleotide (NAD+)-dependent deacetylase enzyme, Sirtuin 1 (SIRT1), can prevent activation of these pathways and promote axonal regeneration. In this study, we tested whether increased expression of SIRT1 protein in sensory neurons prevents and reverses experimental diabetic neuropathy induced by a high fat diet (HFD). We generated a transgenic mouse that is inducible and overexpresses SIRT1 protein in neurons (nSIRT1OE Tg). Higher levels of SIRT1 protein were localized to cortical and hippocampal neuronal nuclei in the brain and in nuclei and cytoplasm of small to medium sized neurons in dorsal root ganglia. Wild-type and nSIRT1OE Tg mice were fed with either control diet (6.2% fat) or a HFD (36% fat) for 2 months. HFD-fed wild-type mice developed neuropathy as determined by abnormal motor and sensory nerve conduction velocity, mechanical allodynia, and loss of intraepidermal nerve fibres. In contrast, nSIRT1OE prevented a HFD-induced neuropathy despite the animals remaining hyperglycaemic. To test if nSIRT1OE would reverse HFD-induced neuropathy, nSIRT1OE was activated after mice developed peripheral neuropathy on a HFD. Two months after nSIRT1OE, we observed reversal of neuropathy and an increase in intraepidermal nerve fibre. Cultured adult dorsal root ganglion neurons from nSIRT1OE mice, maintained at high (30 mM) total glucose, showed higher basal and maximal respiratory capacity when compared to adult dorsal root ganglion neurons from wild-type mice. In dorsal root ganglion protein extracts from nSIRT1OE mice, the NAD+-consuming enzyme PARP1 was deactivated and the major deacetylated protein was identified to be an E3 protein ligase, NEDD4-1, a protein required for axonal growth, regeneration and proteostasis in neurodegenerative diseases. Our results indicate that nSIRT1OE prevents and reverses neuropathy. Increased mitochondrial respiratory capacity and NEDD4 activation was associated with increased axonal growth driven by neuronal overexpression of SIRT1. Therapies that regulate NAD+ and thereby target sirtuins may be beneficial in human diabetic sensory polyneuropathy.
Introduction
Two significant advances have been made in unlocking the underlying molecular mechanisms of axonal degeneration. The first advance is recognizing the importance of nicotinamide-adenine dinucleotide (NAD+) in regulating axonal maintenance, degeneration and regeneration (Wang and He, 2009; Di Stefano and Conforti, 2013; Gerdts et al., 2016; Sasaki, 2019). NAD(H) furnishes reducing equivalents to the mitochondrial electron transport chain to generate ATP. However, NAD+ also acts as a degradation substrate for enzymes such as sirtuins, poly (ADP-ribosyl) transferase 1 (PARP1) and cluster of differentiation 38 (CD38) (Mouchiroud et al., 2013; Yang and Sauve, 2016; Yoshino et al., 2018). Activation of proteins that deplete NAD+ levels (PARP1 and CD38), or regulate NAD+ metabolism [sterile alpha and TIR motif constraining 1 (SARM1)], or inhibit sirtuin 1 (SIRT1) [deleted in bladder cancer protein 1 (DBC1)], promote axonal degeneration (Osterloh et al., 2012; Geisler et al., 2016; Turkiew et al., 2017). In contrast, knockout of PARP1, CD38, SARM1 and DBC1 protect mice against high fat diet (HFD)-induced neuropathy or chemotherapy-induced neuropathy (Barbosa et al., 2007; Obrosova et al., 2008; Escande et al., 2015; Sasaki et al., 2016; Turkiew et al., 2017). Proteins that resynthesize NAD+ via the salvage pathway such as NMNAT1–3, protect against axonal degeneration (Araki et al., 2004; Press and Milbrandt, 2008; Sasaki et al., 2009; Yahata et al., 2009; Babetto et al., 2010; Gilley and Coleman, 2010; Di Stefano et al., 2017). Dietary supplementation with precursors of NAD+ protect against HFD-induced neuropathy (Trammell et al., 2016). In Wlds mutant mice, a model for neuropathy and Wallerian degeneration, the mutation resulting in a fusion protein [consisting of ubiquitination factor (Ube4b) plus NMNAT1] (Mack et al., 2001) extends the half-life of NMNAT1 and protects against nerve transectional degeneration and chemotherapy-induced peripheral neuropathy (Wang et al., 2002; Coleman and Freeman, 2010).
The second advance was that sirtuins play a key role in ageing, neurodegeneration and metabolic syndrome. Sirtuins, with homologies to the Silence Information Regulator 2 (SIR2) gene of the yeast Saccharomyces cerevisiae (Brachmann et al., 1995; Smith et al., 2000), are class III histone deacetylases but not structurally homologous to the other histone deacetylases. There are seven mammalian sirtuins, of which, SIRT1 overexpression and/or its activation by certain compounds, for example resveratrol, improved health and lifespan (reviewed in Imai and Guarente, 2014). SIRT1 is a deacetylase and its identified targets are transcription factors. Prominent amongst these transcription factors are p53 (Vaziri et al., 2001), FoxO family members (Brunet et al., 2004), NF-κB (Yeung et al., 2004), and PGC-1α (reviewed in Rodgers et al., 2008; Dominy et al., 2010). Deacetylation of these factors regulates cell death, survival, and energy metabolism. Sirtuins cleave NAD+ into nicotinamide and 1′-O-acetyl-ADP-ribose (Tanner et al., 2000) or 2′- and 3′-O-acetyl-ADP-ribose (Jackson et al., 2003) and thereby deacetylate lysine residues. SIRT1 activity requires NAD+ (Chalkiadaki and Guarente, 2012). In contrast to other NAD+ metabolizing enzymes, activation of SIRT1 by resveratrol protected mice against HFD-induced obesity and insulin resistance (Lagouge et al., 2006; Cantó et al., 2012; Rajman et al., 2018). This causes a decrease in PGC-1α acetylation and an increase in PGC-1α activity (Nemoto et al., 2005; Gerhart-Hines et al., 2007; Rodgers et al., 2008). No direct effect of neuronal SIRT1 overexpression on HFD-induced neuropathy has been tested.
The HFD induces a mild indolent form of neuropathy in mice that mimics the metabolic profile and neuropathy of metabolic syndrome and mild type 2 diabetes mellitus (Obrosova et al., 2007; Vincent et al., 2009; Guilford et al., 2011). In humans, this mild form of diabetes is associated with reduced plasma NAD+ (Imai and Kiess, 2009; Yoshino et al., 2011, 2018) and the resulting less severe neuropathy may be more amenable to treatment. As the metabolic and neuropathy phenotype are similar in the HFD mouse to human type 2 diabetes mellitus and metabolic syndrome, it is a useful model to test efficacy of potential therapies in human diabetic neuropathy. In this study we tested whether neuronal overexpression of SIRT1 protein could prevent HFD-induced neuropathy and more importantly, could it reverse HFD-induced neuropathy. We generated a neuron-specific and doxycycline (DOX)-regulated transgenic mouse. Using mice fed a HFD or control diet (CD), we determined if nSIRT1OE protected against and reversed HFD-induced neuropathy and explored the potential mechanisms of action including regulation of PARP1 and NEDD4-1. NEDD4-1 is the most abundant mammalian neuron E3 ubiquitin ligase and is involved in axonal outgrowth and regeneration (Liu et al., 2009; Di Antonio, 2010; Drinjakovic et al., 2010). We propose that nSIRT1OE promotes protein quality control (Tomita et al., 2015) and improves mitochondrial metabolism to support axonal regeneration.
Materials and methods
Generation and genotyping of nSIRT1OE transgenic mice
Animal experiments were performed in accordance with the Guide for the Care and Use of Laboratory Animals, approved by the University of Maryland Institutional Animal Care and Use Committee. We used a bidirectional tetracycline responsive element (TRE)-Tight-Bi vector to induce expression of two genes [mouse Sirt1 cDNA and mitochondrially localized enhanced yellow fluorescent protein cDNA (mito-eYFP)] simultaneously in a DOX-regulated manner. Mito-eYFP encodes a fusion protein of mitochondrial targeting sequence (subunit VIII of human cytochrome c oxidase) with eYFP. Mito-eYFP cDNA was cloned at multiple cloning site 1 of the plasmid pTRE-Tight-Bi (Chandrasekaran et al., 2006). Mouse Sirt1 cDNA with a haemagglutinin (HA) tag (pcruz HA SIRT1) was cloned at multiple cloning site 2 of the plasmid pTRE-Tight-Bi. An anti-HA antibody was used to determine Sirt1 transgene expression. The DNA containing Sirt1 cDNA, pTRE-Tight-Bi promoter and mito-eYFP, devoid of vector, was injected into fertilized B3:C57BL/6 hybrid eggs to create TRE-SIRT1/mito-eYFP transgenic (Tg) founders. The transgenic mouse was backcrossed with wild-type C57Bl6J mice for nine generations. Transgenic mice expressing the tetracycline-controlled transactivator protein (tTA) under regulatory control of the forebrain-specific calcium/calmodulin-dependent kinase IIα (CaMKIIα) promoter (CaMKIIα-tTA) were purchased from the Jackson Laboratory (Mayford et al., 1996). Mice that are positive for both CaMKIIα-tTA and TRE-SIRT1/mito-eYFP (double-positive or bigenic) were made by crossing TRE-SIRT1/mito-eYFP mice with CaMKIIα-tTA mice. Bigenic mice were identified by tail DNA PCR.
Immunofluorescence labelling of SIRT1 and mito-eYFP in brain and dorsal root ganglion sections
CamKIIα/mito-eYFP (nmito-eYFP, that expresses only mito-eYFP) and CamKIIα/SIRT1/mito-eYFP (nSIRT1OE, that expresses both mito-eYFP and mouse SIRT1 protein) mice were perfusion-fixed with 4% paraformaldehyde. The dorsal root ganglia (DRGs) were isolated and embedded in 10% gelatin. Brains and DRGs were transferred into 30% sucrose solution. The brains (20 μm coronal sections) and DRGs (40 μm sections) were cut on a freezing microtome. The cut sections were stored in cryoprotectant, washed with potassium phosphate-buffered saline (KPBS) followed by antigen retrieval in 0.01 M citrate buffer using a PELCO Biowave microwave tissue processor (5 min at 550 W). Sections were incubated with SIRT1 primary antibody (Millipore 07-131, 1:300) at 4°C for 40–48 h, washed with KPBS, and incubated in goat anti-rabbit (Alexa Fluor® 594; 1:600) secondary antibody. Following washes in KPBS, sections were counterstained with Hoechst 33342. Brain and DRG sections were imaged on a Keyence BZ-X800E fluorescence microscope using appropriate filters. The optical sectioning function was applied during image acquisition to eliminate fluorescence blurring to produce confocal-like images.
Control and high fat diet
Wild-type and nSIRT1OE Tg C57BL6 mice were fed with a control diet or HFD. The control diet (Harlan-Teklad #2018) contained 6.2% fat (18% calories from fat), 18.6% protein (24% calories from protein), and 44.2% carbohydrate (58% calories from carbohydrate). The HFD (Bio-Serv #F3282) contained 36% fat (60% calories from fat), 20.5% protein (15% calories from protein), and 37.5% carbohydrate (26% calories from carbohydrate). DOX was added to the control diet or HFD at a dose of 200 mg/kg and fed ad libitum.
Neuropathy measurements
Peripheral neuropathy was tested following the guidelines of the European diabetic neuropathy study group (Biessels et al., 2014). Mechanical allodynia was assessed using Somedic von Frey monofilaments (Russell et al., 2008; Biessels et al., 2014). Briefly, ordinal numbers >4 were applied gently on the fat part of both plantar heels until the filament started to bend and maintained for ∼2 s. The threshold was defined as the minimal bending force of the thinnest filament sensed by the mouse in an ascending and descending series. A withdrawal response is considered valid only if the hind paw is completely removed from the platform. Hargreaves test was used to test thermal nociception, which assesses small nerve fibre function (Chandrasekaran et al., 2015). Light from a halogen bulb lamp was delivered to the plantar surface of the mouse hind paw through the base of the glass panel to induce the heat stimuli. The time taken for the mouse to lift or lick its hind paw was recorded automatically. Three measurements were performed with intervals of 1–2 min.
Mice were anaesthetized with isoflurane, 4–5% for induction and 1–2% for maintenance (Choi et al., 2014; Chandrasekaran et al., 2015, 2017). During nerve conduction studies, the mice were provided with thermal support and isoflurane anaesthesia via facemask (1–2%). Nerve conduction velocity (NCV) studies were performed in the left hind limb and in the tail using platinum electrodes, placed adjacent to the nerve, using a 60–80 mA square wave stimulus for 0.1–0.3 ms to obtain near nerve recordings. The G1 (active) and G2 (indifferent) recording electrodes were separated by 10 mm. In the left hind limb, a distal latency was obtained to the dorsum of the paw and a proximal fibular/sciatic conduction velocity obtained. Tail NCVs were recorded over a 4-cm distance measured from the base of the tail. Orthodromic motor conduction velocities were obtained by recording at the tip of the tail and stimulating with the cathode proximal to the G1 recording electrode. Orthodromic sensory tail conduction velocities were obtained by placing G1 at the base of the tail and stimulating 4 cm distally. Sensory responses were averaged until the sensory nerve action potential response was stable. Tail and limb near nerve temperatures were maintained at 32–33°C. The onset latency and peak amplitude were measured.
Intraepidermal nerve fibre density (IENFD) was measured using PGP9.5 antibody staining in a blinded fashion, as previously described (Choi et al., 2014; Chandrasekaran et al., 2015, 2017). To delineate fibre crossing at the dermo-epidermal junction, slides were counterstained by dipping in eosin (Sigma-Aldrich Eosin Y solution HT110316). IENFD was calculated (as fibres/mm) by the number of complete baseline crossings of nerve fibres at the dermo-epidermal junction divided by the measured length of the epidermal surface using standardized validated methods (Lauria et al., 2005; England et al., 2009; Chandrasekaran et al., 2017).
Tissue collection
At experiment termination, blood was collected; glycosylated haemoglobin (GHb), plasma insulin, and lipid profiles were measured. The mice were euthanized, DRG and hind paw skin were collected (Russell et al., 2008; Choi et al., 2014). We used the whole DRG, which includes DRG neurons with surrounding satellite cells. L3 to L5 DRGs were homogenized in RIPA buffer with protease and phosphatase inhibitors in a tissue lyser (Qiagen Inc.) and extracts were centrifuged at 13 000g. Protein samples (20 μg) were analysed by western blot analysis (Choi et al., 2014; Chandrasekaran et al., 2017). The source and dilution of the various antibodies used in this study were: rabbit polyclonal anti-SIRT1 (Millipore 07-131, 1:1000), mouse monoclonal (m)Ab anti-DBC1 (Cell Signaling Technology #5857, 1:1000), rabbit mAb anti-PARP (Cell Signaling Technology #9532, 1:1000), goat polyclonal anti-CD38 (Santa Cruz Biotechnology, #SC7049S, 1:1000), rabbit polyclonal acetylated lysine antibody (Cell Signaling Technology, #9441, 1:1000), rabbit anti-seasonal polyclonal HA antibody (Origene #TA160089, 1:1000), rabbit polyclonal PGC-1α antibody (Novus Biologicals, #NBP1-04676, 1:1000), mouse mAb anti β-actin (Cell Signaling Technology, #3700, 1:1000) and rabbit mAb anti GAPDH (Cell Signaling Technology, #5174, 1:1000). The intensity was normalized to β-actin or GAPDH.
In-gel extraction and mass spectroscopy analysis of major acetylated proteins in dorsal root ganglion extract
Gel electrophoresis was carried out with all lanes loaded with DRG protein extract from wild-type (WT)+HFD samples. Two lanes were cut, transferred to PVDF membrane and subjected to western blot analysis with acetylated lysine antibody. The region of the immunoblot containing the major acetylated protein band (∼100 kDa) was identified and placed next to untransferred remaining gel. The corresponding region of the gel was excised and subjected to in-gel digestion with some modifications (Supplementary Table 1). Peptides were separated using a 2-h chromatographic gradient online with a data-dependent mass spectroscopy (MS)/MS duty cycle of the top 10 most abundant ions. Database search, peptide quantification and identification of acetylated lysine-containing peptides (Shevchenko et al., 2006) were performed using MaxQuant version 1.6.1.0 (Cox and Mann, 2008).
Adult mouse sensory neuron culture to measure mitochondrial respiration
DRG neurons from adult mice were cultured as described (Choi et al., 2014; Chandrasekaran et al., 2015). In brief, the well of the Seahorse plates was coated with 250 µl poly-l-lysine (PLL; 100 µg/ml in water) and laminin (200 µl of 2.5 µg/ml in PBS). DRGs were collected from adult wild-type or nSIRT1OE Tg mice, digested with papain and collagenase to dissociate DRGs neurons and cultured in SHTE medium (selenium 5.2 µg/ml, hydrocortisone 7.6 µg/ml, transferrin 10 µg/ml, oestradiol 5.4 µg/ml) as described (Choi et al., 2014). A specific amount of glucose was added to the SHTE medium (control = 5.5 mM, high glucose = 30 mM glucose). After 24 h of incubation, basal, oligomycin-sensitive, and uncoupled respiration were measured by sequential addition with pyruvate, oligomycin (1.5 µM) to inhibit ATP synthase, FCCP (0.75 µM) to enable maximal rates of oxygen consumption and then treated with a combination of rotenone (1 µM) and antimycin A (1 µM) to block mitochondrial respiration. Basal level of oxygen consumption, oligomycin-sensitive respiration, the maximal respiration capacity and the non-mitochondrial oxygen consumption were measured (Hill et al., 2009; Brand and Nicholls, 2011). Data are expressed as oxygen consumption per minute per 4000 cells.
Statistical analysis
Comparison of dependent variables was performed on transformed data using factorial ANOVA with a post hoc Tukey test to determine the significance among the groups. Individual comparisons were made using Student’s t-test, assuming unequal variances as previously described (Russell et al., 2008). The associations between the mitochondrial function and measures of neuropathy (NCV and mechanical allodynia) were evaluated using Spearman correlation statistics.
Data availability
The raw data that support the findings of this study are available from the corresponding author, upon reasonable request.
Results
Generation of the neuron-specific, inducible SIRT1 overexpressing (nSIRT1OE) transgenic mouse
A transgenic mouse (TREbi-mSIRT1OE/mito-eYFP) that expressed both mouse SIRT1 protein (mSIRT1) and mito-eYFP and regulated by DOX was generated. To direct the expression (mSIRT1/mito-eYFP) to neurons, the TREbi-mSIRT1OE/mito-eYFP mouse was mated with a neuron-specific promoter (CaMKIIα) driven tetracycline transactivator (CaMKIIα-tTA) mouse to generate the bigenic nSIRT1OE mouse. Previously generated DOX-regulated neuron-specific mito-eYFP (nmito-eYFP) bigenic mice that express only mito-eYFP were used as controls to study the effect of nSIRT1OE (Chandrasekaran et al., 2006). The neuron-specific promoter CaMKIIα-driven tetracycline transactivator is a tet-off system; transgene expression occurs on a DOX-free diet and is inhibited with a DOX diet. Phenotype, breeding pattern, and transmission of the transgenes was similar up to 15 generations.
Positive bigenic nSIRT1OE and nmito-eYFP mice maintained on a normal diet, were sacrificed at 3 months of age. Brain and DRGs were isolated and expression of SIRT1 protein was evaluated (Fig. 1 and Supplementary Figs 1–3). In the hippocampal brain sections (Fig. 1B and C), higher expression of SIRT1 protein (red fluorescence) was localized mainly to the neuronal nucleus and was distinct from mitochondria (green fluorescence). In DRGs, SIRT1 expression was observed in the nucleus and cytosol of small to medium sized neurons (Fig. 1D and E). Protein extracts from DRG extracts by western blot analysis showed a 2–3-fold increase in the expression of SIRT1 protein and activity (Fig. 1A and Supplementary Fig. 1). Feeding nSIRT1OE Tg mice with DOX (200 mg/kg) abolished the overexpression of the transgene SIRT1 protein. The nSIRT1OE Tg mice allowed us to test if nSIRT1OE would prevent and treat mice against HFD-induced peripheral neuropathy.
Expression and regulation of SIRT1 in brain and DRGs (n = 6). (A) SIRT1 protein expression was measured by western blot in protein extracts from DRGs using a polyclonal rabbit antibody that recognizes both the endogenous protein and the transgene. In contrast, anti-HA antibody recognizes only the SIRT1 transgene expression. GAPDH was used as loading control. nSIRT1 expression was induced by feeding the bigenic mice with normal diet or suppressed by feeding with DOX-containing diet. (B and C) Coronal hippocampal brain sections from nmito-eYFP (express eYFP in neuronal mitochondria) and nSIRTOE/mito-eYFP mice (express SIRT1 in neurons and eYFP in neuronal mitochondria) were incubated with primary rabbit polyclonal SIRT1 antibody. Sections were then incubated with goat anti-rabbit (Alexa Fluor® 594) secondary antibody, followed by Hoechst 33342 counterstain. Brain and DRG sections were imaged on a Keyence BZ-X800E fluorescence microscope using appropriate filters. Optical sectioning function was applied during image requisition to eliminate fluorescence blurring to produce a confocal-like image. (D and E) Sections of DRG from nmito-eYFP and nSIRTOE/mito-eYFP mice were immunostained with primary rabbit polyclonal SIRT1 antibody, then with goat anti-rabbit (Alexa Fluor® 594) secondary antibody, and finally with Hoechst 33342 dye. The images represent an overlay of all stains. Higher magnification is shown in the inserts. n = nucleus; m = mitochondrion; s = SIRT1. In DRG, the satellite glial cells that surround DRGs are intensely stained with Hoechst 33342 to mark the nuclei of satellite glial cells. In nSIRT1OE showed clear nuclear localization in hippocampal CA1 neurons and this localization was distinct from mito-eYFP fluorescence. In contrast, in DRG neurons, SIRT1 has a nuclear and cytoplasmic localization. Green = neuronal mitochondria; red = neuronal SIRT1; blue = neuronal nuclei. Scale bars = 50 µm. WT = wild-type.
nSIRT1OE prevents metabolic abnormalities in high fat diet fed mice
To determine if nSIRT1OE can prevent the metabolic changes observed in HFD-fed mice (Obrosova et al., 2007; Vincent et al., 2009; Guilford et al., 2011; Biessels et al., 2014), wild-type and nSIRT1OE Tg mice (n = 11 per group) were fed with either control diet (6.2% fat) or HFD (36% fat) for 2 months and diet-induced changes in body weight and blood parameters were measured (Table 1). The baseline measurements in control diet-fed mice showed no significant differences between wild-type and nSIRT1OE (Group 1 versus Group 3 in Table 1). The mean body weight, plasma glucose, H1bAc, insulin and lipid levels were significantly increased in WT+HFD mice compared to WT+CD mice. Neuronal overexpression of SIRT1 protein did not alter HFD-induced weight gain or increase in plasma blood glucose (P = 0.151) or H1bAc but did alter lipid and insulin levels (Group 2 versus Group 4 in Table 1). Triglyceride level was decreased (WT+HFD = 134 ± 15 mg/dl versus nSIRT1OE+HFD = 96 ± 7 mg/dl, P < 0.05), non-esterified fatty acids (NEFA) were decreased (WT+HFD = 0.6 ± 0.1 mM versus nSIRT1OE+HFD = 0.4 ± 0.03 mM, P < 0.05) and plasma insulin levels increased (WT+CD = 2.5 ± 0.2 ng/ml versus nSIRT1OE+CD = 6 ± 0.9 ng/ml, P < 0.001). Measurement of the intraperitoneal glucose tolerance test (n = 6) showed a significant increase in area under the curve (AUC) in WT+HFD mice and in nSIRT1+HFD mice compared to mice fed with a controlled diet, but there was no significant difference in AUC between WT+CD and nSIRT1OE+CD mice or between WT+HFD and nSIRT1OE+HFD mice (Table 1 and Supplementary Fig. 4).
Metabolic and neuropathy end points in control diet-fed wild-type and nSIRT1OE, HFD-fed wild-type and nSIRT1OE mice at 2 months
| Parameters | Wild-type | nSIRT1OE | Significance | |||||
|---|---|---|---|---|---|---|---|---|
| Control diet | HFD | Control diet | HFD | 1 versus 2 | 2 versus 4 | 1 versus 3 | 3 versus 4 | |
| (n = 11) | (n = 11) | (n = 11) | (n = 11) | |||||
| 1 | 2 | 3 | 4 | |||||
| Metabolic endpoints | ||||||||
| Body weight, g | 30 ± 3 | 41 ± 4 | 28 ± 3 | 39 ± 5 | <0.001 | NS | NS | <0.001 |
| Plasma glucose, mg/dl | 108 ± 20 | 200 ± 35 | 111 ± 9 | 178 ± 9 | <0.001 | NS | NS | <0.001 |
| HbA1c, % | 5.4 ± 1 | 6.7 ± 1.1 | 5 ± 0.6 | 6.8 ± 0.9 | 0.016 | NS | NS | 0.007 |
| Insulin, ng/ml | 2.5 ± 0.2 | 4 ± 0.5 | 6 ± 0.9 | 6.2 ± 1.1 | <0.001 | <0.001 | <0.001 | NS |
| Total cholesterol, mg/dl | 80 ± 4.2 | 151 ± 12 | 78 ± 6.7 | 153 ± 10 | <0.001 | NS | NS | <0.001 |
| Triglycerides, mg/dl | 41 ± 6.1 | 62 ± 2.3 | 39 ± 1.9 | 36 ± 1 | <0.001 | <0.001 | NS | NS |
| HDL cholesterol, mg/dl | 70 ± 4 | 136 ± 9 | 69 ± 5 | 143 ± 6 | <0.001 | NS | NS | <0.001 |
| Non-HDL cholesterol, mg/dl | 10 ± 3 | 15 ± 6 | 8 ± 2 | 10 ± 3 | NS | NS | NS | NS |
| NEFA, mM | 0.5 ± 0.06 | 0.6 ± 0.1 | 0.3 ± 0.03 | 0.4 ± 0.03 | NS | <0.05 | <0.05 | NS |
| GTT-AUC (n = 6), mg × min/dl × 103 | 3.6 ± 0.46 | 7.2 ± 0.72 | 3 ± 0.29 | 6.1 ± 1 | <0.001 | NS | NS | <0.001 |
| Neuropathy endpoints | ||||||||
| SMNCV, m/s | 45.2 ± 7.7 | 36.6 ± 8.2 | 47.5 ± 4.7 | 46.3 ± 7.1 | <0.001 | <0.001 | NS | NS |
| TML, ms | 1.3 ± 0.12 | 2.2 ± 0.19 | 1.3 ± 0.14 | 1.4 ± 0.12 | <0.001 | <0.001 | NS | NS |
| TSNCV, m/s | 34.9 ± 2.5 | 31.3 ± 3.6 | 34.5 ± 4.9 | 35.6 ± 1.4 | 0.009 | 0.003 | NS | NS |
| Von Frey, g | 1.1 ± 0.2 | 0.4 ± 0.2 | 1.2 ± 0.12 | 1.0 ± 0.3 | <0.001 | <0.001 | NS | NS |
| Hargreaves, s | 8.1 ± 1 | 12 ± 1.9 | 7.6 ± 1 | 8.8 ± 1.4 | <0.001 | <0.001 | NS | NS |
| Parameters | Wild-type | nSIRT1OE | Significance | |||||
|---|---|---|---|---|---|---|---|---|
| Control diet | HFD | Control diet | HFD | 1 versus 2 | 2 versus 4 | 1 versus 3 | 3 versus 4 | |
| (n = 11) | (n = 11) | (n = 11) | (n = 11) | |||||
| 1 | 2 | 3 | 4 | |||||
| Metabolic endpoints | ||||||||
| Body weight, g | 30 ± 3 | 41 ± 4 | 28 ± 3 | 39 ± 5 | <0.001 | NS | NS | <0.001 |
| Plasma glucose, mg/dl | 108 ± 20 | 200 ± 35 | 111 ± 9 | 178 ± 9 | <0.001 | NS | NS | <0.001 |
| HbA1c, % | 5.4 ± 1 | 6.7 ± 1.1 | 5 ± 0.6 | 6.8 ± 0.9 | 0.016 | NS | NS | 0.007 |
| Insulin, ng/ml | 2.5 ± 0.2 | 4 ± 0.5 | 6 ± 0.9 | 6.2 ± 1.1 | <0.001 | <0.001 | <0.001 | NS |
| Total cholesterol, mg/dl | 80 ± 4.2 | 151 ± 12 | 78 ± 6.7 | 153 ± 10 | <0.001 | NS | NS | <0.001 |
| Triglycerides, mg/dl | 41 ± 6.1 | 62 ± 2.3 | 39 ± 1.9 | 36 ± 1 | <0.001 | <0.001 | NS | NS |
| HDL cholesterol, mg/dl | 70 ± 4 | 136 ± 9 | 69 ± 5 | 143 ± 6 | <0.001 | NS | NS | <0.001 |
| Non-HDL cholesterol, mg/dl | 10 ± 3 | 15 ± 6 | 8 ± 2 | 10 ± 3 | NS | NS | NS | NS |
| NEFA, mM | 0.5 ± 0.06 | 0.6 ± 0.1 | 0.3 ± 0.03 | 0.4 ± 0.03 | NS | <0.05 | <0.05 | NS |
| GTT-AUC (n = 6), mg × min/dl × 103 | 3.6 ± 0.46 | 7.2 ± 0.72 | 3 ± 0.29 | 6.1 ± 1 | <0.001 | NS | NS | <0.001 |
| Neuropathy endpoints | ||||||||
| SMNCV, m/s | 45.2 ± 7.7 | 36.6 ± 8.2 | 47.5 ± 4.7 | 46.3 ± 7.1 | <0.001 | <0.001 | NS | NS |
| TML, ms | 1.3 ± 0.12 | 2.2 ± 0.19 | 1.3 ± 0.14 | 1.4 ± 0.12 | <0.001 | <0.001 | NS | NS |
| TSNCV, m/s | 34.9 ± 2.5 | 31.3 ± 3.6 | 34.5 ± 4.9 | 35.6 ± 1.4 | 0.009 | 0.003 | NS | NS |
| Von Frey, g | 1.1 ± 0.2 | 0.4 ± 0.2 | 1.2 ± 0.12 | 1.0 ± 0.3 | <0.001 | <0.001 | NS | NS |
| Hargreaves, s | 8.1 ± 1 | 12 ± 1.9 | 7.6 ± 1 | 8.8 ± 1.4 | <0.001 | <0.001 | NS | NS |
GTT-AUC = glucose tolerance test-area under the curve; HDL = high density lipoprotein; LDL = low density lipoprotein; NEFA = non-esterified fatty acids; NS = not significant; TML = tail motor latency; TSNCV = tail sensory nerve conduction velocity.
Metabolic and neuropathy end points in control diet-fed wild-type and nSIRT1OE, HFD-fed wild-type and nSIRT1OE mice at 2 months
| Parameters | Wild-type | nSIRT1OE | Significance | |||||
|---|---|---|---|---|---|---|---|---|
| Control diet | HFD | Control diet | HFD | 1 versus 2 | 2 versus 4 | 1 versus 3 | 3 versus 4 | |
| (n = 11) | (n = 11) | (n = 11) | (n = 11) | |||||
| 1 | 2 | 3 | 4 | |||||
| Metabolic endpoints | ||||||||
| Body weight, g | 30 ± 3 | 41 ± 4 | 28 ± 3 | 39 ± 5 | <0.001 | NS | NS | <0.001 |
| Plasma glucose, mg/dl | 108 ± 20 | 200 ± 35 | 111 ± 9 | 178 ± 9 | <0.001 | NS | NS | <0.001 |
| HbA1c, % | 5.4 ± 1 | 6.7 ± 1.1 | 5 ± 0.6 | 6.8 ± 0.9 | 0.016 | NS | NS | 0.007 |
| Insulin, ng/ml | 2.5 ± 0.2 | 4 ± 0.5 | 6 ± 0.9 | 6.2 ± 1.1 | <0.001 | <0.001 | <0.001 | NS |
| Total cholesterol, mg/dl | 80 ± 4.2 | 151 ± 12 | 78 ± 6.7 | 153 ± 10 | <0.001 | NS | NS | <0.001 |
| Triglycerides, mg/dl | 41 ± 6.1 | 62 ± 2.3 | 39 ± 1.9 | 36 ± 1 | <0.001 | <0.001 | NS | NS |
| HDL cholesterol, mg/dl | 70 ± 4 | 136 ± 9 | 69 ± 5 | 143 ± 6 | <0.001 | NS | NS | <0.001 |
| Non-HDL cholesterol, mg/dl | 10 ± 3 | 15 ± 6 | 8 ± 2 | 10 ± 3 | NS | NS | NS | NS |
| NEFA, mM | 0.5 ± 0.06 | 0.6 ± 0.1 | 0.3 ± 0.03 | 0.4 ± 0.03 | NS | <0.05 | <0.05 | NS |
| GTT-AUC (n = 6), mg × min/dl × 103 | 3.6 ± 0.46 | 7.2 ± 0.72 | 3 ± 0.29 | 6.1 ± 1 | <0.001 | NS | NS | <0.001 |
| Neuropathy endpoints | ||||||||
| SMNCV, m/s | 45.2 ± 7.7 | 36.6 ± 8.2 | 47.5 ± 4.7 | 46.3 ± 7.1 | <0.001 | <0.001 | NS | NS |
| TML, ms | 1.3 ± 0.12 | 2.2 ± 0.19 | 1.3 ± 0.14 | 1.4 ± 0.12 | <0.001 | <0.001 | NS | NS |
| TSNCV, m/s | 34.9 ± 2.5 | 31.3 ± 3.6 | 34.5 ± 4.9 | 35.6 ± 1.4 | 0.009 | 0.003 | NS | NS |
| Von Frey, g | 1.1 ± 0.2 | 0.4 ± 0.2 | 1.2 ± 0.12 | 1.0 ± 0.3 | <0.001 | <0.001 | NS | NS |
| Hargreaves, s | 8.1 ± 1 | 12 ± 1.9 | 7.6 ± 1 | 8.8 ± 1.4 | <0.001 | <0.001 | NS | NS |
| Parameters | Wild-type | nSIRT1OE | Significance | |||||
|---|---|---|---|---|---|---|---|---|
| Control diet | HFD | Control diet | HFD | 1 versus 2 | 2 versus 4 | 1 versus 3 | 3 versus 4 | |
| (n = 11) | (n = 11) | (n = 11) | (n = 11) | |||||
| 1 | 2 | 3 | 4 | |||||
| Metabolic endpoints | ||||||||
| Body weight, g | 30 ± 3 | 41 ± 4 | 28 ± 3 | 39 ± 5 | <0.001 | NS | NS | <0.001 |
| Plasma glucose, mg/dl | 108 ± 20 | 200 ± 35 | 111 ± 9 | 178 ± 9 | <0.001 | NS | NS | <0.001 |
| HbA1c, % | 5.4 ± 1 | 6.7 ± 1.1 | 5 ± 0.6 | 6.8 ± 0.9 | 0.016 | NS | NS | 0.007 |
| Insulin, ng/ml | 2.5 ± 0.2 | 4 ± 0.5 | 6 ± 0.9 | 6.2 ± 1.1 | <0.001 | <0.001 | <0.001 | NS |
| Total cholesterol, mg/dl | 80 ± 4.2 | 151 ± 12 | 78 ± 6.7 | 153 ± 10 | <0.001 | NS | NS | <0.001 |
| Triglycerides, mg/dl | 41 ± 6.1 | 62 ± 2.3 | 39 ± 1.9 | 36 ± 1 | <0.001 | <0.001 | NS | NS |
| HDL cholesterol, mg/dl | 70 ± 4 | 136 ± 9 | 69 ± 5 | 143 ± 6 | <0.001 | NS | NS | <0.001 |
| Non-HDL cholesterol, mg/dl | 10 ± 3 | 15 ± 6 | 8 ± 2 | 10 ± 3 | NS | NS | NS | NS |
| NEFA, mM | 0.5 ± 0.06 | 0.6 ± 0.1 | 0.3 ± 0.03 | 0.4 ± 0.03 | NS | <0.05 | <0.05 | NS |
| GTT-AUC (n = 6), mg × min/dl × 103 | 3.6 ± 0.46 | 7.2 ± 0.72 | 3 ± 0.29 | 6.1 ± 1 | <0.001 | NS | NS | <0.001 |
| Neuropathy endpoints | ||||||||
| SMNCV, m/s | 45.2 ± 7.7 | 36.6 ± 8.2 | 47.5 ± 4.7 | 46.3 ± 7.1 | <0.001 | <0.001 | NS | NS |
| TML, ms | 1.3 ± 0.12 | 2.2 ± 0.19 | 1.3 ± 0.14 | 1.4 ± 0.12 | <0.001 | <0.001 | NS | NS |
| TSNCV, m/s | 34.9 ± 2.5 | 31.3 ± 3.6 | 34.5 ± 4.9 | 35.6 ± 1.4 | 0.009 | 0.003 | NS | NS |
| Von Frey, g | 1.1 ± 0.2 | 0.4 ± 0.2 | 1.2 ± 0.12 | 1.0 ± 0.3 | <0.001 | <0.001 | NS | NS |
| Hargreaves, s | 8.1 ± 1 | 12 ± 1.9 | 7.6 ± 1 | 8.8 ± 1.4 | <0.001 | <0.001 | NS | NS |
GTT-AUC = glucose tolerance test-area under the curve; HDL = high density lipoprotein; LDL = low density lipoprotein; NEFA = non-esterified fatty acids; NS = not significant; TML = tail motor latency; TSNCV = tail sensory nerve conduction velocity.
nSIRT1OE prevents neuropathy induced by a high fat diet
We tested if nSIRT1OE could prevent peripheral neuropathy. After 8 weeks of HFD feeding (n = 11), wild-type mice showed a significant slowing of sciatic motor NCV (SMNCV) (Table 1) (from 45.2 ± 7.7 m/s in WT+CD to 36.6 ± 8.2 m/s in WT+HFD; P < 0.001), increased tail motor latency (from 1.3 ± 0.12 ms to 2.2 ± 0.19 ms; P < 0.001) and decreased tail sensory NCV (from 34.9 ± 2.5 m/s in WT+CD to 31.3 ± 3.6 m/s in WT+HFD; P < 0.001). These changes in NCV in HFD-fed mice were consistent with the development of peripheral neuropathy. After 8 weeks of HFD, there was a significant decrease in the Von Frey paw withdrawal threshold in HFD compared to control diet mice (WT+CD = 2.1 ± 0.2 g versus WT+HFD = 1.4 ± 0.2 g; P < 0.001) that remained the same up to 16 weeks, consistent with development of tactile allodynia. In contrast, the HFD-fed nSIRT1OE mice had preserved NCVs and the velocities were comparable to control diet-fed mice, for example SMNCV (P = 0.645, Table 1). These findings are consistent with protection against peripheral neuropathy by nSIRT1OE. Neuronal SIRT1OE mice had normal tactile allodynia at 8 weeks after initiation of HFD treatment compared to nSIRT1OE+CD or WT+CD (Table 1).
Eight weeks after feeding wild-type and nSIRT1OE mice with controlled diet or HFD, mice were euthanized and paw skins were examined for IENFD (Fig. 2A–E). Skin biopsies showed a significant decrease (from 34 ± 5 fibres/mm to 22 ± 6 fibres/mm; P < 0.001) in the IENFD from WT+HFD mice compared to WT+CD mice. On the other hand, in nSIRT1OE mice fed with control diet, there was a significant increase in IENFD (from 34 ± 5 fibres/mm in WT+CD to 52 ± 10 fibres/mm in nSIRT1OE+CD mice; P < 0.001). In nSIRT1OE mice fed with a HFD, there was a decrease in IENFD (41 ± 6 fibres/mm), but the IENFD count was still higher than in wild-type mice fed with either a control diet or HFD [nSIRT1OE+HFD = 41 ± 6 fibres/mm compared to WT+CD = 34 ± 5 fibres/mm and WT+HFD = 22 ± 6 fibres/mm (P < 0.001)].
nSIRT1OE prevents HFD-induced neuropathy (n = 6). IENFD in paw skin biopsy is preserved in nSIRT1OE HFD-fed mice. Images show 50-µm thick paw skin sections immunostained with anti-PGP9.5 antibody in wild-type (WT) and nSIRT1OE mice that were fed a control diet or HFD for 2 months. e = epidermis; d = dermis. In the insets, to delineate fibre crossing at the dermo-epidermal junction better, slides were counterstained by dipping in eosin (Sigma-Aldrich Eosin Y solution HT110316). ef = example of an epidermal fibre; df = example of a dermal fibre. Scale bars = 100 µm; insets = 50 µm. (E) Graphic representation of the IENFD. The large central black horizontal bar for each group indicates the mean. The smaller lower bar is the 25th percentile and the smaller upper bar is the 75th percentile. ***P < 0.001 WT+HFD compared to WT+CD; ++P < 0.01 nSIRT1OE+CD compared to WT+CD and ###P < 0.001 nSIRT1OE+HFD compared to WT+HFD.
nSIRT1OE reverses peripheral neuropathy induced by a high fat diet
To test if nSIRT1OE would reverse HFD-induced neuropathy, 3-month-old nSIRT1OE Tg mice were fed a control diet containing DOX (200 mg/kg) for 2 weeks to shut off nSIRT1OE. The mice were then fed either with DOX+CD or with DOX+HFD for an additional 2 months (Fig. 3A). SMNCV and Von Frey hind paw withdrawal latency (mechanical allodynia) were measured after 2 months (Fig. 3C and D). No significant changes in SMNCV and paw withdrawal latency were observed in mice fed with CD+DOX mice. However, in HFD+DOX fed nSIRT1OE (nSIRT1OE Off) mice, the SMNCV decreased from 32 ± 5 m/s to 24 ± 6 m/s (P < 0.001), intermediate measurements at 2 and 4 weeks showed consistent decreases in SMNCV. The Von Frey test also showed a significant worsening in mechanical allodynia after 2 months of feeding with HFD+DOX (from 2.06 ± 0.2 g to 1.44 ± 0.12 g; P < 0.001). After 2 months on HFD+DOX, a group of mice (n = 8) were fed only HFD without DOX (HFD–DOX) to induce nSIRT1OE (nSIRT1OE On). The rest of the group (n = 8) were fed HFD+DOX for an additional 2 months (Fig. 3A). Induction of nSIRT1OE without DOX in the diet improved both SMNCV (from 24 ± 6 m/s to 36 ± 6 m/s; P < 0.001) and Von Frey mechanical allodynia (1.44 ± 0.12 g to 2.2 ± 0.16 g; P < 0.001) (Fig. 3C and D). There was no significant decrease in SMNCV and mechanical allodynia in CD+DOX mice between 2 and 4 months. Measurement of blood glucose (Fig. 3B) showed that feeding HFD+DOX increased the random blood glucose levels in nSIRT1OE mice (from 120 ± 14 mg/dl to 210 ± 18 mg/dl; P < 0.001). In nSIRT1OE+HFD−DOX mice, the increase in random blood glucose level was lower (184 ± 15 mg/dl) but was not statistically significantly different compared to nSIRT1OE+HFD+DOX mice. On the other hand, triglycerides and NEFA showed a decrease in nSIRT1OE+HFD−DOX compared to nSIRT1OE+HFD+DOX mice (triglycerides: 69 ± 5 from 78 ± 3.1 mg/dl; NEFA: 0.49 ± 0.03 mM from 0.57 ± 0.04 mM) suggesting that nSIRT1OE could influence peripheral lipid metabolism in peripheral neuropathy.
nSIRT1 overexpression reverses neuropathy (n = 6). (A) Schematic representation of the reversal study. SIRT1 transgene expression was turned off by feeding the nSIRT1OE mice with DOX for 2 weeks before the start of the experiment. Zero-time refers to the time of onset of feeding the mice with a HFD. Group 1: nSIRT1OE Off + CD (4 months); Group 2: nSIRT1OE Off + HFD (4 months); Group 3: nSIRT1OE Off + HFD (same as Group 2 but up to 2 months only); Group 4: nSIRT1OE On + HFD (2–4 months). Group 4 are the same mice as in Group 3 with nSIRT1OE Off (0–2 months) but with nSIRT1OE On from 2 to 4 months. As the data were not statistically different for Groups 3 and 4 up to 2 months, the data has been combined in the graphs. (B) Weekly measurement of random blood glucose before and after turning on SIRT1OE. nSIRTOE did not significantly affect the blood glucose: when SIRT1 was turned on (nSIRT1 On + HFD, Group 4 at 4 months) compared to Group 2 at 4 months (nSIRT1 Off + HFD). However, there was a significant difference between Group 1 (SIRT1 Off + CD) at 4 months and Group 2 (SIRT1 Off + HFD) and Group 4 (SIRT1 On + HFD) (***P < 0.001). (C) Measurement of the sciatic-fibular motor nerve conduction velocity showing reversal in nSIRT1OE mice after turning on SIRT1 expression with the removal of DOX from the diet.***P < 0.001, Group 1 (SIRT1Off + CD) at 2 and 4 months and Group 2 (nSIRT1OE Off + HFD) at 2 months. +++P < 0.001, Group 2 (SIRT1 Off + HFD) at 4 months compared to Group 4 (SIRT1 On + HFD). (D) Paw withdrawal threshold measured using Von Frey filaments showing reversal in nSIRT1OE mice after turning on SIRT1 expression with the removal of DOX from the diet. ***P < 0.001, Group 1 (SIRT1 Off + CD) at 2 and 4 months compared with Group 2 (nSIRT1OE Off + HFD) at 2 months. +++P < 0.001, Group 2 (SIRT1 Off + HFD) at 4 months compared to Group 4 (SIRT1 On + HFD).
Quantification of IENFD showed that 2 months of feeding nSIRT1OE mice with DOX+HFD decreased the PGP9.5-positive IENFD counts from 36 ± 6 fibres/mm to 23 ± 9 fibres/mm (P < 0.001). Removal of DOX from the diet increased the counts to 37 ± 7 fibres/mm. There was no significant difference in IENFD counts in nSIRT1OE+CD+DOX mice compared to nSIRT1OE+CD mice suggesting that addition of DOX to the diet had minimal effect on peripheral neuropathy (Fig. 4A–F).
nSIRT1OE reverses HFD-induced loss of IENFD (n = 6). (A) Schematic representation of the reversal study representing the four groups (Groups 1–4). SIRT1 transgene expression was turned off by feeding the nSIRT1OE mice with DOX for 2 weeks before the start of the experiment. Zero-time refers to the time of onset of feeding the mice with a HFD. Intra-epidermal nerve fibre immunohistochemistry and fibre counts in paw skin biopsy from nSIRT1OE are reversed in HFD-fed mice by nSIRT1OE. Images show 50-µm thick paw skin sections immunostained with anti-PGP9.5 antibody (B, C, E and F). e = epidermis; d = dermis. In the insets, to delineate fibre crossing at the dermo-epidermal junction better, slides were counterstained by dipping in eosin (Sigma-Aldrich Eosin Y solution HT110316). ef = example of an epidermal fibre; df = example of a dermal fibre. Scale bars = 100 µm; insets = 50 µm. (D) Graph showing the combined IENFD for all groups. There is a reduction in IENFD in HFD-fed SIRT1 Off + HFD-fed mice (Groups 2 and 3) as compared with SIRT1 Off + CD-fed mice (Group 1) at 2 and 4 months. ***P < 0.001 nSIRT1 Off + HFD compared to nSIRT1 Off + CD at 4 months. ++P < 0.01 nSIRT1 Off + HFD compared to nSIRT1 Off + CD at 2 months. Neuronal overexpression of SIRT1 protein (nSIRT1OE) by the removal of DOX in HFD-fed mice after 2 months (Group 4) reversed the loss of epidermal fibres and IENFD counts at 4 months. ###P < 0.001 nSIRT1OE On + HFD at 4 months + HFD compared to nSIRT1 Off + HFD at 2 months.
nSIRT1OE maintains mitochondrial bioenergetic function
To determine if nSIRT1OE was able to regulate mitochondrial bioenergetic function, we assessed the cellular bioenergetics profile of sensory neurons derived from DRG of age-matched wild-type and nSIRT1OE mice. The oxygen consumption rate was measured in DRG neuronal culture prepared from 3-month-old adult mice, using the Seahorse Biosciences XF24 analyser (Fig. 5 and Table 2). The cultures were treated for 24 h in low (5.5 mM) and high (30 mM) d-glucose. Basal, oligomycin sensitive, and uncoupled respiration were measured by sequential addition with pyruvate and mitochondrial complex inhibitors, in order of injection, included pyruvate, oligomycin (1.5 µM) to inhibit ATP synthase, FCCP (0.75 µM) to enable maximal rates of oxygen consumption, and then treated with a combination of rotenone and antimycin A to block respiratory electron flux at complexes I and III. Maximal oxygen consumption rate induced by the uncoupler FCCP in cultured neurons grown in 30 mM glucose (103 ± 16 pmol O2/min) was significantly decreased (P < 0.01) compared with neurons grown in 5.5 mM glucose (146 ± 19 pmol O2/min), indicating impairment of maximal electron transport activity in the hyperglycaemic state. DRGs prepared from nSIRT1OE mice had a higher maximal oxygen consumption rate, 219 ± 28 pmol O2/min in 5.5 mM glucose (P = 0.017) and 154 ± 24 pmol O2/min in 30 mM glucose. Basal respiration was also higher in nSIRT1OE, 58 ± 3 pmol O2/min in wild-type versus 87 ± 4 pmol O2/min in nSIRT1OE at 5.5 mM glucose (P < 0.01) and 51 ± 2 pmol O2/min versus 83 ± 4 pmol O2/min at 30 mM glucose (P < 0.05). Calculation of the respiratory control ratio and spare respiratory capacity showed higher values in nSIRT1OE DRG neurons when compared with wild-type DRG neurons. The respiratory control ratio was 6 ± 0.6 in wild-type versus 10 ± 0.8 in nSIRT1OE neurons with 5.5 mM glucose, P < 0.05. The spare respiratory capacity was 88 ± 9 in wild-type with 5.5 mM and 52 ± 6 with 30 mM glucose compared to 132 ± 15 with 5.5 mM and 71 ± 8 with 30 mM glucose in nSIRT1OE DRG (P < 0.05). We interpret these results to suggest that although hyperglycaemia caused a significant decrease in maximal oxygen consumption rate, the mitochondria in DRG neurons from nSIRT1OE mice are better coupled and have higher spare respiratory capacity than wild-type DRG neurons.
Measurement of mitochondrial function in cultured neurons using the XF24 analyser
| Wild-type | nSIRT1OE | |||
|---|---|---|---|---|
| Glucose, mM | 5.5 | 30 | 5.5 | 30 |
| Basal respiration, pmol O2/min/4000 DRG | 58 ± 3 | 51 ± 2 | 87 ± 4++ | 83 ± 4* |
| Oligomycin sensitive | 17 ± 2 | 15 ± 3 | 26 ± 3 | 23 ±4 |
| Uncoupled respiration | 146 ± 19 | 103 ± 16 | 219 ± 28++ | 154 ± 24* |
| Spare reserve capacity | 88 ± 9 | 52 ± 6 | 132 ± 15+ | 71 ± 8* |
| Wild-type | nSIRT1OE | |||
|---|---|---|---|---|
| Glucose, mM | 5.5 | 30 | 5.5 | 30 |
| Basal respiration, pmol O2/min/4000 DRG | 58 ± 3 | 51 ± 2 | 87 ± 4++ | 83 ± 4* |
| Oligomycin sensitive | 17 ± 2 | 15 ± 3 | 26 ± 3 | 23 ±4 |
| Uncoupled respiration | 146 ± 19 | 103 ± 16 | 219 ± 28++ | 154 ± 24* |
| Spare reserve capacity | 88 ± 9 | 52 ± 6 | 132 ± 15+ | 71 ± 8* |
Oxygen consumption rate (OCR) was measured at basal level with the subsequent and sequential addition of oligomycin, FCCP and rotenone + antimycin A to DRG neurons cultured from adult wild-type and nSIRT1OE mice. Levels of OCR are normalized per 4000 cells. Basal, oligomycin sensitive and uncoupled OCR were measured as described in the ‘Materials and methods’ section. Spare respiratory capacity was calculated after subtracting the non-mitochondrial respiration. ++P < 0.01, +P < 0.05, comparison between nSIRT1OE DRG at 5 mM glucose compared to wild-type DRG at 5 mM glucose and *P < 0.05, comparison between nSIRT1OE DRG at 30 mM glucose compared to wild-type DRG at 30 mM glucose.
Measurement of mitochondrial function in cultured neurons using the XF24 analyser
| Wild-type | nSIRT1OE | |||
|---|---|---|---|---|
| Glucose, mM | 5.5 | 30 | 5.5 | 30 |
| Basal respiration, pmol O2/min/4000 DRG | 58 ± 3 | 51 ± 2 | 87 ± 4++ | 83 ± 4* |
| Oligomycin sensitive | 17 ± 2 | 15 ± 3 | 26 ± 3 | 23 ±4 |
| Uncoupled respiration | 146 ± 19 | 103 ± 16 | 219 ± 28++ | 154 ± 24* |
| Spare reserve capacity | 88 ± 9 | 52 ± 6 | 132 ± 15+ | 71 ± 8* |
| Wild-type | nSIRT1OE | |||
|---|---|---|---|---|
| Glucose, mM | 5.5 | 30 | 5.5 | 30 |
| Basal respiration, pmol O2/min/4000 DRG | 58 ± 3 | 51 ± 2 | 87 ± 4++ | 83 ± 4* |
| Oligomycin sensitive | 17 ± 2 | 15 ± 3 | 26 ± 3 | 23 ±4 |
| Uncoupled respiration | 146 ± 19 | 103 ± 16 | 219 ± 28++ | 154 ± 24* |
| Spare reserve capacity | 88 ± 9 | 52 ± 6 | 132 ± 15+ | 71 ± 8* |
Oxygen consumption rate (OCR) was measured at basal level with the subsequent and sequential addition of oligomycin, FCCP and rotenone + antimycin A to DRG neurons cultured from adult wild-type and nSIRT1OE mice. Levels of OCR are normalized per 4000 cells. Basal, oligomycin sensitive and uncoupled OCR were measured as described in the ‘Materials and methods’ section. Spare respiratory capacity was calculated after subtracting the non-mitochondrial respiration. ++P < 0.01, +P < 0.05, comparison between nSIRT1OE DRG at 5 mM glucose compared to wild-type DRG at 5 mM glucose and *P < 0.05, comparison between nSIRT1OE DRG at 30 mM glucose compared to wild-type DRG at 30 mM glucose.
Mitochondrial respiration (n = 6). Measurement of mitochondrial function in cultured DRG neurons using the XF24 analyser. Oxygen consumption rate was measured at basal level with the subsequent and sequential addition of oligomycin, FCCP and rotenone+antimycin A (AA) to DRG neurons cultured from (A) 3-month-old wild-type (WT) and nSIRT1OE mice. Levels of oxygen consumption rate were normalized per 4000 cells. DRG neurons were cultured for 24 h either in low (5.5 mM) or high (30 mM) glucose. Dotted line represents oxygen consumption rate measurements in DRG neurons cultures in 30 mM glucose, straight lines represent oxygen consumption rate measurements in DRG neurons cultures in 5 mM glucose. Red coloured lines are from nSIRT1OE DRG neurons and blue are from wild-type DRG neurons. From the oxygen consumption rate, basal respiration and maximal capacity respiration were calculated and are shown in B and C. The significance was calculated by ANOVA multiple comparison post hoc Tukey analysis. The statistical significance is indicated. The raw data are shown in Table 2.
nSIRT1OE regulates NAD+-dependent proteins involved in axonal growth and survival
Deacetylation of transcription factors, cofactors, and histones by SIRT1 has been shown to enhance mitochondrial metabolism (Gerhart-Hines et al., 2007; Rodgers et al., 2008). SIRT1 requires NAD+ as a cofactor to deacetylate proteins. However, there are other enzymes that also use NAD+ and they may limit NAD+ availability for SIRT1 to promote mitochondrial metabolism (Cantó et al., 2015). We speculated that SIRT1 might regulate the activity of other NAD+-consuming enzymes by decreasing the expression or activity of these proteins. Therefore, we measured the levels of NAD+-consuming proteins by western blot analysis in extracts of DRG neurons from wild-type and nSIRT1 mice. The results showed no significant changes in the levels of DBC1 and CD38 in WT+HFD compared to WT+CD, whereas there was a significant decrease in SIRT1 protein, PGC-1α protein, and an increase in the activated form of PARP1 in WT+HFD samples (Fig. 6A–E). The decrease in SIRT1 and PGC-1α protein in WT+HFD mice was normalized in nSIRT1OE+HFD samples and correspondingly there was a significant decrease in the activated PARP1 form of the protein (Fig. 6A–E).
NEDD4-1 deacetylation by nSIRT1OE (n = 6). Western blot analysis of NAD+-consuming enzymes and acetylated proteins in DRG protein extracts prepared from wild-type (WT) and nSIRT1OE mice fed either a control diet (CD) or HFD for 4 months. (A) Preparation of DRG protein extracts, blot analysis, the source and the dilution of the antibodies used are described in the ‘Materials and methods’ section. (B–E) Quantification of the intensity of the bands are shown. Significant decrease in SIRT1 protein, detected with the rabbit polyclonal anti-SIRT1 that recognized both endogenous and overexpressed SIRT1 protein (Millipore 07-131, 1:1000) (B), decrease in PGC-1α protein levels (C), increase in cleaved PARP1 protein (D) and increase in acetylated 118 kDa protein (E) were observed in WT+HFD samples, but not in nSIRT1OE DRG neurons with a control diet or HFD. ***P < 0.001 WT+HFD compared with nSIRT1OE+HFD, WT+CD, and nSIRT1OE+CD in B, C and E and WT+HFD compared with WT+CD, and nSIRT1OE + CD in D. ###P = 0.008 (D) for a significant decrease in cleaved PARP1 protein in nSIRT1OE+HFD mice when compared with WT+HFD.
To determine if nSIRT1OE alters deacetylation of cellular proteins, protein extracts from DRG neurons were subjected to western blot analysis with acetylated lysine antibody. Increased acetylation of several proteins of >100 kDa were noted in WT+HFD DRG extracts compared to WT+CD DRG extracts. The levels of acetylated proteins were decreased in DRG protein extracts from nSIRT1OE mice fed either a control diet or HFD, suggesting that SIRT1 deacetylates specific proteins (Fig. 6A and E). The major acetylated protein band (∼102 kDa) was cut and subjected to mass spectrometry analysis. The results showed that the protein that is maximally acetylated is E3 Ubiquitin Protein Ligase NEDD4-1, protein ID P46935, molecular weight: 102.71 kDa, and acetylated at positions 124 and 125. nSIRT1OE deacetylated this protein.
Discussion
The findings in this study indicate that improved mitochondrial respiratory capacity by SIRT1 and control of other NAD+-consuming pathways maintain cellular homeostasis to protect DRG neurons from HFD-induced peripheral neuropathy. Importantly, nSIRT1OE was able to prevent and reverse HFD-induced peripheral neuropathy. We propose that NEDD4-associated mitophagy and improved mitochondrial respiratory capacity act synergistically to increase axonal growth and repair (Supplementary Fig. 5). SIRT1 is central to this process. SIRT1 regulates mitochondrial function in the peripheral nerve through PGC1-alpha while also regulating mitophagy and axonal growth through NEDD4.
In cell culture studies, SIRT1 is found in the nucleus of most cell types (Michishita et al., 2005), in keeping with its activity of deacetylation of transcription factors. SIRT1 protein, however, appears to possess both nuclear localization signals and nuclear export signals, and could shuttle between the cytoplasm and the nucleus (Tanno et al., 2007; Sugino et al., 2010). Extranuclear localization, particularly in the mitochondrion, has been observed (Aquilano et al., 2013). Immunohistochemistry (Fig. 1) in nSIRT1OE showed clear nuclear localization in hippocampal CA1 neurons and this localization was distinct from mito-eYFP fluorescence. In contrast, in DRG neurons, SIRT1 has a nuclear and cytoplasmic localization.
Feeding with an HFD caused a significant increase in body weight, plasma glucose, cholesterol, triglycerides and NEFA in wild-type mice. The blood results showed similar increases in glucose levels between WT+HFD and nSIRT1OE+HFD mice, suggesting that nSIRT1OE does not reduce the glucotoxicity induced by an HFD. The major changes observed in nSIRT1OE+HFD compared to WT+HFD mice were that nSIRT1OE prevented the HFD-induced increases in triglycerides and NEFA (Table 1), suggesting that nSIRT1OE may promote lipid oxidation (Purushotham et al., 2009; Imamura et al., 2017). In addition, results from neuronal SIRT1OE or knockout mice suggest that altered neuronal insulin signalling is likely to be responsible for changes in blood insulin and triglyceride levels. The proposed hypothesis is that SIRT1 tonically inhibits neuronal insulin signalling. Deletion of SIRT1 decreases plasma insulin and enhances insulin sensitivity in the periphery, whereas nSIRT1OE increases plasma insulin and decreases insulin sensitivity in the periphery (Wu et al., 2011; Lu et al., 2013) thus reducing peripheral insulin signalling. In addition, nSIRT1 also regulates leptin sensitivity and neuronal leptin prevents HFD- and age-associated type 2 diabetic metabolic impairments by regulating glucose homeostasis, including hepatic glucose production (Sasaki et al., 2014). Interestingly, elevated insulin levels are also observed in the HFD mice but do not prevent neuropathy developing. A role for neuronal SIRT1 insulin signalling in adipose tissue lipolysis has also been proposed (Feige et al., 2008; Ramadori et al., 2010; Lou et al., 2017). Thus, increased neuronal SIRT1OE could influence peripheral lipid metabolism.
The current study provides novel evidence that nSIRT1OE prevented and reversed HFD-induced peripheral neuropathy. This observation is supported by the preservation and reversal of measures of neuropathy, including nerve conduction velocity, mechanical allodynia, thermal nociception, and IENFD in HFD-fed nSIRT1OE Tg mice (Table 1). Feeding wild-type mice with an HFD for 2 months decreased neuropathy endpoint measures and the decrease was significant compared to WT+CD mice (Table 1). Measurement of IENFD, a more sensitive measure of peripheral neuropathy, in particular ‘small-fibre neuropathy’, showed a significant increase in nSIRT1OE+CD mice compared to WT+CD (Fig. 2). Administration of resveratrol (an activator of SIRT1) in cultured adult rat DRG neurons was shown to increase neurite outgrowth compared to untreated cultures (Chowdhury et al., 2011; Roy Chowdhury et al., 2012). Moreover, SIRT1 activation confers neuroprotection in experimental neuritis (Shindler et al., 2007), where there may be axonal degeneration and nSIRT1OE prevented experimental allergic encephalomyelitis-induced demyelination (Nimmagadda et al., 2013). Expression of SIRT1 in dorsal horn in our transgenic mice could account for changes in diabetic pain pathways. These findings suggest a critical role for SIRT1 in axon development and maintenance. Diabetic peripheral axonal pathology includes retraction of terminal nerve endings and alterations in the peripheral terminals of sensory neurons located in the epidermis (Biessels et al., 2014). Recent studies on Wallerian axonal degeneration in cultured peripheral neurons provide a strong yet unknown link between NAD+ metabolism, axon degeneration and regeneration (Di Stefano and Conforti, 2013; Gerdts et al., 2016; Sasaki, 2019). Our results suggest that SIRT1 is uniquely suited to perform these three functions.
SIRT1 protein deacetylates members of the PGC-1α/ERR-alpha complex, which are essential metabolic regulatory transcription factors (Gerhart-Hines et al., 2007; Rodgers et al., 2008). PGC-1α activates nuclear DNA- and mtDNA-encoded transcription factors (NRF-2 and TFAM) causing mitochondrial biogenesis (Scarpulla, 2011). Our previous results showed that absence of PGC-1α (PGC-1α KO) exacerbates streptozotocin (STZ)-induced peripheral neuropathy (Choi et al., 2014). On the other hand, overexpression of human TFAM protected C57BL6 mice against STZ-induced neuropathy (Chandrasekaran et al., 2015). These results suggest that failure of the SIRT1-PGC-1α-TFAM signalling axis results in suppression of mitochondrial oxidative phosphorylation and development of peripheral neuropathy. Protecting this axis may prevent or reverse peripheral neuropathy.
The cellular bioenergetics profile of DRG neurons, prepared from adult wild-type and nSIRT1OE mice, was assessed by measuring the oxygen consumption rate in DRG neurons cultured for 24 h either in low (5.5 mM) or high (30 mM) glucose using the Seahorse Biosciences XF24 analyser. The lower oxygen consumption rate values compared to other studies (Chowdhury et al., 2011) could be due to absence of neurotrophic factors in our culture medium. However, the mitochondrial coupling and respiration were maintained. Both basal oxygen consumption rate and maximal rate, in the presence of uncoupler, were higher in nSIRT1OE DRG neurons exposed to low and high glucose. Exposure to high (30 mM glucose) blunted the maximal oxygen consumption rate and significantly lowered spare respiratory capacity in both wild-type and nSIRT1OE DRG neurons suggesting that the cells were energetically stressed, and that mitochondrial workload was increased. However, the capacity in nSIRT1OE neurons was still higher than wild-type DRG neurons exposed to low glucose. We interpret these results to suggest that nSIRT1OE, or SIRT1 activation, primes the mitochondria by increasing their reserve capacity to combat mitochondrial stress under diabetic conditions (Roy Chowdhury et al., 2012).
Studies reveal that NAD+ metabolism plays a central role in locally mediated axon destruction and regeneration pathways (Gerdts et al., 2016; Sasaki, 2019). To deacetylate proteins, SIRT1 protein requires NAD+. The other well-characterized enzymes that deplete cellular NAD+ are PARP1 and CD38. Western blot analysis of DRG protein extracts (Fig. 6) showed a significant decrease in SIRT1 protein in wild-type mice fed a HFD, an increase in 89 kDa cleaved and activated PARP1 enzyme, and no significant changes in DBC1 and CD38. A plausible explanation for the differential effect of SIRT1 on activation of other NAD+-consuming enzymes might be due to the Km (affinity) of these enzymes for NAD+. The Km of SIRT1 for NAD+ is 94–96 µM, the Km for PARP1 is 50–97 µM, whereas the Km for CD38 is 15–25 µM. Thus, CD38 has a higher affinity for NAD+ and is less likely to be affected by NAD+ depletion with a HFD. The estimated total intracellular content of NAD+ in mammals ranges from 200 to 500 µM (Schmidt et al., 2004; Bai et al., 2011). Feeding the mice with an HFD causes a decrease in cellular NAD+ (Trammell et al., 2016). We suggest that the physiological protective pathway might involve inhibiting enzymes that would further degrade NAD+ but at the same time use the limiting NAD+ to activate oxidative metabolism. This is consistent with the observation that there was a decrease in cleaved activated PARP1 in nSIRT1OE+HFD compared to WT+HFD. Absence of an effect on the levels of DBC1 and CD38 by nSIRT1OE also suggests that the decrease in NAD+ is not enough to affect DBC1 and CD38. SIRT1 negatively regulates PARP1 activity by deacetylation (Rajamohan et al., 2009). Furthermore, PARP1 and SIRT1 have opposing effects on p53 nuclear accumulation and activation following cytotoxic stress (Vaziri et al., 2001; Langley et al., 2002). SIRT1 also deacetylates DBC1 (Kim et al., 2008; Zhao et al., 2008). Thus, SIRT1 may regulate activities of other NAD+-consuming enzymes via deacetylation (Cantó et al., 2015). In the absence of nSIRT1OE, activation of these NAD+-consuming pathways during stress may decrease cellular NAD+, overwhelm SIRT1 and promote autophagy and cellular death. Results from other laboratories are consistent with the observation that deletion (knockout) or inhibition of axonal degenerating pathways involving SARM1, PARP1, CD38 and DBC1 preserve NAD+ for SIRT1 to mediate axonal regeneration (Barbosa et al., 2007; Obrosova et al., 2008; Osterloh et al., 2012; Chiang et al., 2015; Geisler et al., 2016; Turkiew et al., 2017; Walker et al., 2017). The present results also suggest that SIRT1 uses NAD+ to deacetylate proteins to maintain cellular homeostasis by regulating the protein quality control via a ubiquitin pathway.
Maintenance of cellular homeostasis, elimination of damaged proteins by degradation, is achieved by molecular chaperones and the ubiquitin-proteasome system. The NAD+-dependent deacetylase SIRT1 positively regulates cellular homeostasis in response to oxidative stress (Tomita et al., 2015). However, its contribution to protein quality control in diabetic conditions remains unexplored. In the DRG extracts from nSIRT1OE+HFD mice compared to WT+HFD mice, the major deacetylated protein was determined to be E3 protein ligase NEDD4-1. NEDD4-1 is the most abundant E3 ubiquitin ligase in mammalian neurons. NEDD4-1 is a crucial modulator of axonal and dendritic growth, required for proper formation and function of neuromuscular junctions (Drinjakovic et al., 2010; Schmeisser et al., 2013; Hsia et al., 2014). Deacetylation of NEDD4-1 promotes the activity of NEDD4-1 in protein degradation, avoiding a build-up of misfolded protein aggregates, for example, synuclein aggregation (Kim et al., 2016). In NEDD4-1 knockdown cells, abnormal mitochondria are observed, suggesting that activation of NEDD4 by nSIRT1OE-mediated deacetylation promotes mitophagy via ubiquitination of the protein sequestosome 1 that leads to degradation of abnormal mitochondria. (Liu et al., 2009; Di Antonio, 2010; Schmeisser et al., 2013). E3 ligase NEDD4 promotes axon branching by degrading phosphatase and tensin homologue (PTEN) (Di Antonio, 2010; Drinjakovic et al., 2010; Christie et al., 2012). These results suggest a new role for SIRT1 in protein quality control and axonal growth using an NAD+-dependent deacetylation-mediated pathway. A flow chart showing how nSIRT1OE could prevent and reverse HFD-induced peripheral neuropathy by deacetylating proteins that synergistically promote mitochondrial oxidative metabolism and axonal growth and regeneration is suggested (Supplementary Fig. 5).
Our studies suggest that nSIRT1OE enhances the SIRT1/PGC-1α axis in neurons and protects against neuropathy in mice fed an HFD that show glucose intolerance and hyperlipidaemia. The bioenergetics profile of DRG neurons from nSIRT1OE mice show a higher basal and maximal respiratory capacity with nSIRT1OE. This is consistent with the concept that SIRT1 primes neuronal mitochondria to overcome failure of bioenergetic function induced by an increased glucose load. SIRT1 is central to regulation of both mitochondrial function and axon growth and regeneration. For example, SIRT1 uses NAD+ to deacetylate proteins, prevent depletion of cellular NAD+, and maintain cellular homeostasis. This occurs by regulating protein quality control via the ubiquitin pathway. Importantly, the major protein deacetylated by SIRT1 was the E3 protein ligase NEDD4-1, which is critical in axonal growth. In addition to addressing a potential pathogenic pathway in diabetic neuropathy, the study suggests that medications, which prevent NAD+ depletion induced by diabetes mellitus and activate SIRT, may provide a therapeutic intervention for diabetic neuropathy. Human studies are limited, but a specific SIRT1 mutation was associated with type 1 diabetes mellitus in one patient study (Biason-Lauber et al., 2013) and low SIRT1 has been associated with insulin resistance in the offspring of those with type 2 diabetes mellitus (Rutanen et al., 2010). There are a growing number of clinical trials on sirtuin activators and NAD+-boosting drugs, to evaluate their role in protecting against cardiovascular and metabolic diseases (Berman et al., 2017; Öztürk et al., 2017; Martens et al., 2018; Singh et al., 2019).
Acknowledgements
We thank Dr Brian Hampton of the University of Maryland School of Medicine Protein Analysis Laboratory and staff for their contribution to this publication. We thank Gautam Srinivasan for his help with the figures. We thank Seth Crawford for his professional contribution in consolidating and formatting the figures.
Funding
Supported in part by the National Institute of Diabetes and Digestive and Kidney Diseases, National Institutes of Health 1R01DK107007-01A1, Office of Research Development, Department of Veterans Affairs (Biomedical and Laboratory Research Service and Rehabilitation Research and Development, 101RX001030), Diabetes Action Research and Education Foundation, University of Maryland, Baltimore, Institute for Clinical & Translational Research (ICTR) and the Baltimore GRECC (J.W.R.), VA BX000917 (T.K.), 1K08NS102468 - 01A1 (C-Y.H.) and the Atlantic Nutrition Obesity Research Center, grant P30 DK072488 from the National Institute of Diabetes and Digestive and Kidney Diseases, National Institutes of Health.
Competing interests
The authors report no competing interests.
References
Abbreviations
- CD =
control diet
- DOX =
doxycycline
- DRG =
dorsal root ganglion
- HFD =
high fat diet
- IENFD =
intraepidermal nerve fibre density
- (SM) NCV =
(sciatic motor) nerve conduction velocity
