Reduced Sirtuin1 signalling exacerbates diabetic mice hindlimb ischaemia injury and inhibits the protective effect of a liver X receptor agonist

Abstract Diabetes mellitus causes endothelial dysfunction, which further exacerbates peripheral arterial disease (PAD). Improving endothelial function via reducing endothelial oxidative stress (OS) may be a promising therapy for diabetic PAD. Activation of liver X receptor (LXR) inhibits excessive OS and provides protective effects on endothelial cells in diabetic individuals. Therefore, we investigated the effects of LXR agonist treatment on diabetic PAD with a focus on modulating endothelial OS. We used a streptozotocin‐induced diabetes mouse model combined with a hindlimb ischaemia (HLI) injury to mimic diabetic PAD, which was followed by LXR agonist treatment. In our study, the LXR agonist T0901317 protected against HLI injury in diabetic mice by attenuating endothelial OS and stimulating angiogenesis. However, a deficiency in endothelial Sirtuin1 (SIRT1) largely inhibited the therapeutic effects of T0901317. Furthermore, we found that the underlying therapeutic mechanisms of T0901317 were related to SIRT1 and non‐SIRT1 signalling, and the isoform LXRβ was involved in LXR agonist‐elicited SIRT1 regulation. In conclusion, LXR agonist treatment protected against HLI injury in diabetic mice via mitigating endothelial OS and stimulating cellular viability and angiogenesis by LXRβ, which elicited both SIRT1‐mediated and non‐SIRT1‐mediated signalling pathways. Therefore, LXR agonist treatment may be a promising therapeutic strategy for diabetic PAD.

synthase and apoptosis. 4,5 Therefore, a therapeutic method focusing on alleviating endothelial OS and apoptosis should be considered.
As a member of the nuclear receptor superfamily, liver X receptors (LXRs) play pivotal roles in cardiovascular disease and diabetic complications. LXR agonists or genetic treatment can protect against atherosclerosis, myocardial ischaemia/reperfusion injury, myocardial hypertrophy and diabetic cardiomyopathy via repressing cellular inflammation, apoptosis and OS damage. [6][7][8][9] In addition, a previous study also showed that LXR agonist treatment inhibits high glucose (HG)-induced endothelial OS and senescence, with an additional atheroprotective effect in diabetes. 10 Hence, we hypothesized that LXR agonist treatment might inhibit endothelial OS and apoptosis, further promoting angiogenesis and protecting against diabetic PAD. To examine this hypothesis, we explored a mouse model of hindlimb ischaemia injury (HLI) with streptozotocin (STZ)-induced DM, followed by treatment with T0901317, a non-selective LXR agonist used in our previous study, 11 to characterize the effects of LXR agonist treatment on diabetic PAD with a focus on endothelial OS and apoptosis.
Silent information regulator 1 (Sirtuin1, SIRT1) is an NAD+dependent deacetylase that exerts its regulatory effects on both the nucleus and cytoplasm of endothelial cells (ECs). 12 A previous study revealed that endothelial SIRT1 ablation exacerbated hypoxic injury and impaired angiogenesis. 13 In contrast, ECs were rescued from hypoxic exposure through SIRT1 up-regulation. 14 Significantly, SIRT1 is essential for healthy vasculature, as endothelial SIRT1 deficiency leads to increased OS, inflammation and senescence. 15 Furthermore, a previous study showed that SIRT1 also deacetylates and activates LXR, 16 and the SIRT1-LXR axis contributes to atheroprotection by reducing inflammation. 17 Interestingly, our previous research demonstrated that LXR agonist treatment activated SIRT1, deacetylating its downstream signals and protecting myocardial cells via inhibiting OS and apoptosis during sepsis-induced myocardial injury. 11 However, the interplay between endothelial SIRT1 and LXR in response to diabetic PAD is still unclear. To elucidate this, we used endothelial-specific SIRT1 knockout mice treated with T0901317 to investigate the interaction between SIRT1 and LXR and evaluate the potential effects of LXR agonist treatment on diabetic PAD.

| Cell culture and treatment
Human umbilical vein endothelial cells (HUVECs) were obtained and cultured using the same method as our previous study. 4 HUVECs were cultured at a concentration of 5.5 mmol/L glucose for the normal treatment or 33.3 mmol/L, which represented the HG treatment. SIRT1, LXRα and LXRβ small interfering RNAs (siRNAs) were purchased from Santa Cruz and were transfected at a concentration of 100 nmol/L. After the indicated treatment, HUVECs were exposed to hypoxia (95% N 2 /5% CO 2 )/ serum deprivation (H/SD) treatment for 6 hours.

| Serial laser Doppler perfusion imaging of hindlimb
Laser Doppler perfusion imaging (LDPI, PeriScan-PIM3; Perimed) was performed to evaluate blood perfusion recovery of ischaemic hindlimbs. The perfusion ratio (PR, ratio of average LDPI index of ischaemic to non-ischaemic) was utilized to qualify the perfusion recovery. From an animal ethics point of view, mice with autoamputation were killed immediately, and PR was defined as 0.0 on postoperative day (POD) 21.

| Vascular casting mould assay
Gastrocnemius tissue was harvested for casting neovasculature on POD 21. Distal vessels were cast and sputter-coated with gold.
Then, the mould was fixed by conducting resin and photographed using scanning electron microscopy (S-4800, Hitachi).

| Histological staining
Gastrocnemius tissue was fixed with 4% paraformaldehyde and sectioned on POD 7 or POD 21. Morphological changes in gastrocnemius tissues were shown by haematoxylin-eosin (H&E) staining. Immunofluorescent staining was performed to detect the expression of α-SMA and CD31 for assessing angiogenesis. Slices were incubated with primary antibodies against α-SMA (Abcam, ab32575) and CD31 (Abcam, ab24590); then, they were stained with the respective fluorescent secondary antibodies.

| Measurement of reactive oxygen species
The levels of reactive oxygen species (ROS) in tissue and in cells were detected by the dihydroethidium (DHE, Beyotime, S0063) as described in a previous study. 4 Gastrocnemius tissue or HUVECs were labelled with a DHE probe (5 μmol/L) for 20 minutes in a dark incubator at 37℃. Fluorescence microscope was used to observe sections at 535 nm excitation.
Moreover, to detect the mean fluorescence density (autofluorescence modification), cells (1x10 5 ) were resuspended and labelled with DHE (5 μmol/L). Flow cytometry (Becton Dickinson Biosciences) was used to measure the fluorescence.

| Measurement of intracellular NOx
Intracellular NOx production (NO and its oxidized forms) was detected using a commercial nitrate/nitrite assay kit (Beyotime, S00233). The indicated samples were prepared in accordance with the manufacturer's instructions. A luminometer was utilized to detect the absorbance at 540 nm. NOx contents were expressed as nmol/10 5 cultured cells.

| Detection of cytokines and OSrelated indicators
Gastrocnemius tissue was collected and homogenized to evaluate angiogenesis and OS-related indicators. The expression of vascular endothelial growth factor (VEGF) and basic fibroblast growth factor (bFGF) was evaluated using commercial ELISA kits (R&D Systems) in accordance with the manufacturer's instructions. OS-related indicators: glutathione (GSH), malondialdehyde (MDA), catalase (CAT) and superoxide dismutase (SOD) in tissues were evaluated using the commercial testing kits, respectively (all purchased from Beyotime), following the manufacturer's instructions.
MDA and GSH were expressed as contents (nm/mg protein), and SOD and CAT were expressed as enzymatic activity units (U/mg protein). A microplate reader (Thermo Scientific) was used to measure the results.

| Tube formation assay
Matrigel (BD Biosciences) was pre-cooled and placed in 96-well dishes for 40 minutes. HUVECs were plated in each well at a concentration of 3 × 10 4 /100 μL. Tubes were observed with an inverted phase contrast microscope (Nikon) after 6 hours of incubation. Tube lengths were calculated with ImageJ software using an angiogenesis analyser (ImageJ News. 2012).

| Cell migration assay
HUVECs were resuspended at a concentration of 5 × 10 4 /100 μL and then placed in a migration chamber, which was immersed in a 24-well dish filled with 500 μL of medium. After 24-hours incubation, migrated cells were fixed with 4% paraformaldehyde and dyed with 0.1% crystal violet. Positive cells were counted in four random fields for each well.

| Caspase-3 activity assay
Gastrocnemius tissue was homogenized to assess caspase-3 activity using a caspase-3 activity kit (Beyotime, C1116), according to the manufacturer's instruction. The actual OD405 was calculated by subtracting the blank control without pNA, and the results are presented as μM/h/g.

| Real-time quantitative PCR
HUVECs were lysed using TRIzol reagent to extract total RNA. Reverse transcription was performed using an Omniscript RT Kit (Qiagen), followed by amplification of the cDNA with Fast SYBR Green Master Mix (TAKARA Biotechnology) on an ABI 7900HT System. The sequences of the forward and reverse primers were as follows: SIRT1 (forward) 5′-GCCAGAGTCCAAGTTTAGAAGA-3′, 5′-TTCAGCTCAGGGATGACCTT-3′. The relative SIRT1 mRNA transcript levels were calculated by normalizing them to GAPDH and expressed the data as a relative ratio.

| Western blotting
Gastrocnemius tissue and HUVECs were lysed to extract proteins.
Protein lysates (50 μg) were loaded and separated on the SDS-PAGE gel. Then, proteins were transferred onto nitrocellulose membranes.
Membranes were incubated with the corresponding primary anti-

| Statistical analysis
Results were presented as the mean ± SEM. Prism 6.0 (GraphPad Software) was used for analyses. Two groups were compared utilizing Student's two-tailed unpaired t test. One-way factor analysis of variance analysis was used for multi-group comparisons, followed by Dunnett's post hoc test. P values < .05 were regarded as statistically significant.

| Endothelial-specific SIRT1 deletion exacerbated hindlimb ischaemia injury in diabetic mice and inhibited the LXR-mediated protective effects
First, the aorta ventralis of Tie2-Cre-SIRT1 Loxp+/+ mouse was sectioned to evaluate endothelial SIRT1 expression through immunofluorescent staining. As shown in Figure 1A, endothelial SIRT1 was deleted specifically, as evidenced by the presence of only CD31 red fluorescence. Furthermore, mouse aortic ECs were isolated through repeating a differential adhesion procedure and were cultured in accordance with a previous report. 18 Endothelial SIRT1 expression was detected by Western blotting, and the results are shown in Figure 1B-C. The expression of endothelial SIRT1 decreased markedly in Tie2-Cre-SIRT1 Loxp+/+ mice compared with that of wild-type mice.
To establish the dose of the LXR agonist T0901317 for the following study, diabetic mice were treated with T0901317 at doses of 10 mg, 20 mg, 30 mg and 50 mg/kg for 21 days after HLI injury.
As shown in Figure 1D, PR was restored with increasing doses of T0901317. The 30 mg/kg dose had a better therapeutic effect than any other dose; this effect was weakened at a dose of 50 mg/ kg. Therefore, a dose of 30 mg/kg was used in our subsequent study.
Three months after STZ injection, mouse blood glucose levels increased significantly compared to that of normal mice ( Figure 1E).
T0901317 did not decrease the levels of blood glucose, with or without endothelial-specific SIRT1 deletion. STZ injection also dramatically reduced mouse body weight and plasma insulin content, with or without endothelial-specific SIRT1 deletion ( Figure 1F-G).
Endothelial-specific SIRT1 deletion inhibited hindlimb blood As shown in Figure 1J

| Endothelial-specific SIRT1 deletion restrained ischaemia-induced angiogenesis in diabetic mice and inhibited the pro-angiogenetic effect of LXR agonist treatment
α-SMA and CD31 co-staining were performed to evaluate angiogenesis and to determine the density of mature blood vessels. Endothelial-specific SIRT1 deletion further inhibited ischaemia-induced angiogenesis in diabetic mice, as evidenced by the decreased density of α-SMA, CD31 and α-SMA/CD31 in the  Figure 2G-H).

| Endothelial-specific SIRT1 deletion exacerbated ischaemia-induced endothelial apoptosis in diabetic mice and weakened the LXR-mediated anti-apoptotic effect
CD31 staining and TUNEL assays were performed to evaluate endothelial apoptosis, which was characterized as double-positive staining of POD 7. As shown in Figure 3A

| Endothelial-specific SIRT1 deletion exacerbated ischaemia-induced endothelial cell OS in diabetic mice and weakened the LXR-mediated antioxidative effect
As shown in Figure 4A

| SIRT1 silencing increased apoptosis, decreased tube formation and migration of HUVECs in response to H/SD + HG, and it weakened the protective effects of T0901317
To Moreover, the levels of phosphorylated-p53 (Thr55 and Ser46) were down-regulated by LXR agonist treatment, regardless of whether SIRT1 was silenced ( Figure 6N-P). Additionally, the level of hydroxylated hypoxia-inducible factor-1α (hydroxylated HIF-1α) was still down-regulated and the level of HIF-1α was up-regulated in the H/SD + HG+LXR + si SIRT1 group (vs H/SD + HG+si SIRT1 group, Figure 5Q-R). Furthermore, we assessed OS-related proteins to determine the mechanism by which LXR agonist treatment elicited antioxidative effects ( Figure 6E). Forkhead box transcription factor O1 (FoxO1), a target of SIRT1 that is related to OS, was also involved in our investigation. LXR agonist treatment down-regulated the level of Ac-FoxO1 compared to that of the H/SD + HG group; however, this trend was reversed by SIRT1 silencing (Figure 6F-G). Additionally, LXR agonist treatment decreased 3-NT and NADPH oxidase 4 (NOX4), which is a constitutive NADPH oxidase and generates intracellular superoxide; further, LXR agonist treatment increased SOD-2 expression compared to that of the H/SD + HG group ( Figure 6H-J). These LXR-elicited trends were reversed by SIRT1 silencing; however, the trends were still present compared to those of the H/SD + HG+si SIRT1 group ( Figure 6H-J).

| The isoform LXRβ was involved in LXR agonistelicited SIRT1 regulation
Since LXRα and LXRβ are distributed in specific organs and tissues, we explored which isoform was involved in ECs. As shown in conditions. Therefore, the isoform LXRβ, not LXRα, may be involved in LXR agonist-elicited SIRT1 regulation.  38,39 In fact, either acetylation or phosphorylation only led to partial p53 activation, and both modifications were required for complete p53 activation. 40 Hence, various LXR agonist-induced post-translational modifications of p53 may contribute to partly inhibiting OS and apoptosis in SIRT1 deletion conditions. Furthermore, the level of hydroxylated HIF-1α, which is related to hypoxia-induced angiogenesis by regulating the degradation of HIF-1α, 41,42 was still inactivated by LXR agonist treatment when SIRT1 was silenced, which may be a compensatory mechanism for angiogenesis in SIRT1 deficiency.

| D ISCUSS I ON
Although previous studies reported a link between SIRT1 deacetylation and activation of LXR, SIRT1 deletion compromised the normal responses to the LXR agonist and eliminated the antioxidative stress and anti-apoptotic effects of LXR agonist F I G U R E 7 The isoform LXRβ was involved in LXR agonist-elicited SIRT1 regulation. A, Representative blots of LXRα and LXRβ. Lane 2 and lane 4 were duplicates of lanes 1 and 3, respectively. B-C, Western blotting was used to analyse the expression of LXRα and LXRβ (n = 5). D-F, The effects of LXRα and LXRβ siRNA intervention were assessed by Western blotting (n = 3). G, Representative blots of SIRT1, FoxO1, Ac-FoxO1, p53 and Ac-p53 in each group. H-I, The levels of SIRT1 mRNA and protein in response to T0901317 under LXRα or LXRβ silencing conditions (n = 5). J-K, Western blotting was used to analyse the expression of Ac-FoxO1 and Ac-p53 (n = 5). *P < .05 between the indicated groups treatment. 11,16 Therefore, we explored whether blocking LXR could hinder LXR agonist-elicited SIRT1 regulation. Since LXRs consist of two different and highly homologous isoforms that are distributed in specific organs and tissues, 43 LXRα and LXRβ, we explored which isoform was involved in EC. In our experience, HG exposure decreased LXRβ but not LXRα expression under H/SD conditions. Furthermore, silencing LXRβ but not LXRα compromised LXR agonist-elicited SIRT1 up-regulation at both the mRNA and protein levels. Based on the above results, we determined that the isoform LXRβ might contribute to LXR agonist-elicited SIRT1 regulation. Several previous reports have demonstrated that activation of LXRα, which is specifically distributed in the myocardium, protects against cardiovascular disease by maintaining glucose homeostasis and mitigating myocardial apoptosis and OS. 8,9 Unlike LXRα, LXRβ is more ubiquitously expressed and highly related to endothelial functions. Activation of LXRβ exerts a protective effect on ECs after HG exposure by inhibiting senescence and OS with an additional mechanism for vascular protection. 10 In addition, LXR activation leads to LXRβ-and ERα-dependent processes, facilitating EC migration and preserving endothelial integrity by stimulating endothelial NO production. 44 These results further indicated that the endothelial LXRβ isoform may contribute to LXR agonist-elicited beneficial effects.
In conclusion, our current study revealed a protective role of LXR agonist treatment in diabetic mice with HLI via promoting endothelial viability, mitigating OS and enhancing angiogenesis. Then, we found that endothelial SIRT1 played a crucial role in LXR agonist-elicited beneficial effects. In vitro studies in HUVECs revealed that the underlying therapeutic mechanisms of LXR agonist treatment were related to SIRT1 and non-SIRT1 signalling, and the isoform LXRβ was involved in LXR agonist-elicited SIRT1 regulation.
Our findings further elucidate the interaction between LXR and SIRT1 and may provide positive evidence for further clinical trials to assess the potential therapeutic effect of LXR agonists in diabetic patients with vascular lesions.

CO N FLI C T O F I NTE R E S T
The authors declare that they have no conflict of interest.

DATA AVA I L A B I L I T Y S TAT E M E N T
The data that support the findings of this study are available from the corresponding author upon reasonable request.