Trans‐repression of NFκB pathway mediated by PPARγ improves vascular endothelium insulin resistance

Abstract Previous study has shown that thiazolidinediones (TZDs) improved endothelium insulin resistance (IR) induced by high glucose concentration (HG)/hyperglycaemia through a PPARγ‐dependent‐NFκB trans‐repression mechanism. However, it is unclear, whether changes in PPARγ expression affect the endothelium IR and what the underlying mechanism is. In the present study, we aimed to address this issue. HG‐treated human umbilical vascular endothelial cells (HUVEC) were transfected by either PPARγ‐overexpressing (Ad‐PPARγ) or PPARγ‐shRNA‐containing (Ad‐PPARγ‐shRNA) adenoviral vectors. Likewise, the rats fed by high‐fat diet (HFD) were infected by intravenous administration of Ad‐PPARγ or Ad‐PPARγ‐shRNA. The levels of nitric oxide (NO), endothelin‐1 (ET‐1) and cytokines (TNFα, IL‐6, sICAM‐1 and sVCAM‐1) and the expression levels of PPARγ, eNOS, AKT, p‐AKT, IKKα/β and p‐IKKα/β and IκBα were examined; and the interaction between PPARγ and NFκB‐P65 as well as vascular function were evaluated. Our present results showed that overexpression of PPARγ notably increased the levels of NO, eNOS, p‐AKT and IκBα as well as the interaction of PPARγ and NFκB‐P65, and decreased the levels of ET‐1, p‐IKKα/β, TNFα, IL‐6, sICAM‐1 and sVCAM‐1. In contrast, down‐expression of PPARγ displayed the opposite effects. The results demonstrate that the overexpression of PPARγ improves while the down‐expression worsens the endothelium IR via a PPARγ‐mediated NFκB trans‐repression dependent manner. The findings suggest PPARγ is a potential therapeutic target for diabetic vascular complications.

non-canonical tissues, such as the endothelium, is less clear. Several studies support a role for IR in the development of premature cardiovascular AS independent of T2DM and obesity, 8,9 but the cause of IR is still poorly understood.
Peroxisome proliferator-activated receptor γ (PPARγ), a superfamily member of nuclear transcription factors, plays essential roles in gluco-lipid homoeostasis and adipogenesis and it is a molecular target of insulin-sensitizing drugs such as thiazolidinediones (TZDs). 10 As exogenous agonists of PPARγ, TZDs affect multiple pathophysiological processes such as lipid metabolism, vascular modifications and production of inflammatory mediators, which are involved in the development of diabetic cardiovascular complications. [11][12][13] For example, TZDs decrease plasminogen activator-1 and C-reactive protein levels 14 as well as reduce coronary hyperplasia after coronary stent implantation. 15 Insulin-sensitizing therapy with TZDs is considered as a promising intervention for patients with T2DM. 16 However, numerous clinical studies have shown that TZDs have many side effects including fluid retention, worsening heart failure and weight gain. 17 Therefore, reduction in side effects of TZDs especially negative effects on cardiovascular system is needed.
Although TZDs have anti-diabetic and anti-atherogenic effects by acting on macrophages and lymphocytes by trans-repressing nuclear factor kappaB-(NFκB-) dependent target genes, 18 the effects of TZDs on vascular endothelial cells and the underlying mechanisms remain poorly understood. Our recent data revealed that TZDs improved vascular endothelium IR induced by high concentration of glucose /hyperglycaemia (HG) through a PPARγ-dependent NFκB trans-repression mechanism. 19 However, it is unclear whether changes in PPARγ expression affect endothelium IR and what the underlying mechanism is. Therefore, in the present study, we sought to investigate the effects of alterations of PPARγ expression on endothelium IR and explore its underlying molecular mechanisms.

| Cell culture
Cultures of HUVEC and HEK293T were performed as described in our previous study. 20 Briefly, the cells were cultured in gelatincoated six-well plates and propagated in DMEM, which was supplemented with 10% FBS, 100 IU mL −1 penicillin and 0.1 mg mL −1 streptomycin. The cells were cultured at 37°C in a 95% O 2 -5% CO 2 humidified atmosphere.

| Establishment of endothelium IR model in vitro and in vivo
The methods for establishment of endothelium IR model in vitro and in vivo have previously been described in details. 19 Briefly, HUVEC with a 90% confluence was first pre-treated with a complete DMEM containing 33 m mol L −1 of glucose (HG) for 48 hours. After that, the DMEM was replaced by a fresh serum-free medium and then the HUVEC was further cultured for 4 hours. Subsequently, the cells were treated with 5 mIU L −1 insulin (final concentration) for 10 minutes. Finally, the supernatants were collected and the levels of nitrite and ET-1 were assayed; and the cells were used to detect the expression of AKT and p-AKT.
All animal procedures were approved by the Institutional Animal  At the end of 2 weeks after dietary manipulation, the HFD-fed rats were injected intraperitoneally (i.p.) with a low dose of STZ (35 mg kg −1 ) while the control rats were given the vehicle for STZ (ie citrate buffer, pH 4.4, 1 mL kg −1 , i.p.) respectively. Physical parameters including body weight, body length, body mass index (BMI) and fat coefficient were measured. Also, fasting plasma glucose (FPG) and serum insulin (FINS), triglyceride(TG), cholesterol (CH) as well as the homoeostatic model assay of IR (HOMA-IR) were tested. In addition, the serum levels of nitrite and ET-1 as well as the expression of AKT and p-AKT from aorta tissue were assayed before modelling (premodel) and after modelling (post-model). 19

3T3-L1 cells with post-transfection of different adenoviral vectors
The 90% confluent HEK293T cells were transfected with adenoviruses containing either wild-type full-length cDNA of PPARγ

| In vitro experimental protocols
The 90% confluent HUVEC was first pre-treated with a fresh complete DMEM containing HG for 48 hours and then further cultured for 4 hours with a fresh serum-free DMEM (IR). Next, the cells were randomly allocated to two batches. One batch of cells was transfected with adenoviruses containing either wild-type full-length cDNA of PPARγ (IR+PPARγ) or a cDNA-scramble of PPARγ (vehicle, IR+Veh); while the other was done with those containing either a shRNA of PPARγ (IR+shRNA) or a shRNA-scramble of PPARγ (vehicle, IR+Veh). The cells that were neither treated with HG nor transfected were considered as the Ctrl. After transfection for 24 hours, all the cells were washed with PBS twice and further cultured with the fresh serum-free DMEM for an additional 12 hours. At the end, the supernatants were used to test the levels of NO, ET-1 and cytokines (TNFα, IL-6, sICAM-1 and sVCAM-1) and the cells were used to measure the expression levels of PPARγ, eNOS, AKT, p-AKT, IKKα/β, p-IKKα/β and IκBα. Besides, the interaction between PPARγ and NFκB-P65 was evaluated by immunoprecipitation.

| In vivo experimental protocols
The rats with systemic and endothelium IR were first randomly divided into five groups (Six rats per group), ie IR, IR+Ad-PPARγ (IR+PPARγ), IR+Ad-PPARγ-shRNA (IR+shRNA) and their respective scrambles (IR+Veh). The rats were intravenously administered with Ad-PPARγ (IR+PPARγ group),Ad-PPARγ-shRNA (IR+shRNA), their vehicles (IR+Veh groups), normal saline (IR group). The six rats that were neither treated with HFD nor transfected were considered as the Ctrl group. After treatment for a week, serum levels of NO, ET-1 and other cytokines (TNFα, IL-6, sICAM-1 and sVCAM-1) were assayed, functional assessment of rat aorta was performed, and expression levels of PPARγ, eNOS, AKT, p-AKT, IKKα/β, p-IKKα/β and IκBα from aorta were determined by Western blots.

| Functional assessment of rat aorta
The functional assessment of rat aorta was carried out by a modification of the previously described method. 22 Briefly, all the rats were fasted for 12 hours and anaesthetized with ketamine (70 mg kg −1 , ip). Then, the rats were sacrificed by a cervical dislocation. The thoracic aorta was carefully isolated and an approximately 3 mm long aortic ring for each of the rat was prepared. The ring was isometrically mounted on a myograph (model 610M, DMT, Denmark).
The aortic ring was first equilibrated for 45 minutes under a resting tension of 0.5 g and then was pre-treated with 1 μ mol L −1 PE. The concentration-response curves to acetylcholine (Ach, 10 −10 -10 −5 M) or sodium nitroprusside (SNP, 10 −10 -10 −5 M) were performed after 40%-60% maximal contraction was induced with 1 μ mol L −1 PE (. The vasodilation at each concentration was measured and expressed as the percentage of force generated in response to PE.

| Western blots for protein expression levels
Western blots were performed according to the method described in our previous study with a minor modification. 23

| Measurement of nitrite
The levels of nitrite were determined by a nitrite assay kit according to the manufacturer's instruction. Since the NO is very unstable and rapidly converted into nitrite, nitrite concentrations were measured to indicate NO levels. Nitrite reacts with Griess reagents to produce a colour, which can be measured by a spectrophotometer at 550 nm. Briefly, a standard curve was prepared using a series of nitrite concentrations. 100 μL of the sample was added into a 96well microplate and then the Griess reagent I and II were in turn added and mixed. Next, the mixture was incubated at 37°C for 60 minutes. Finally, the OD was determined at 550 nm with a spectrophotometer (Bio-Rad Laboratories).

| Statistical analysis
All data were expressed as the mean ± SEM. Significance was tested with unpaired t test, one-way ANOVA and homogeneity test of variance wherever appropriate. P value <0.05 was considered to be statistically significant.  Figure 1A); while transfection of Ad-PPARγ-shRNA into 3T3-L1 decreased those by 29% (vs. Veh or Ctrl, Figure 1B). Moreover, constitutive PPARγ expression in the Ctrl (HUVEC in vitro and rat vascular endothelium in vivo) group was found, while that in IR or IR+Veh group decreased significantly compared with Ctrl group. However, due to transfection of the over-or down-expressional adenoviral vectors, the expression levels of PPARγ were notably up-regulated or down-regulated respectively (vs. IR+Veh, Figure 1C,F), indicating that the vectors were successfully constructed and the transfection met the experimental needs.

| Amelioration of vascular endothelium IR in vitro and in vivo by PPARγ
The extent of endothelium IR was evaluated by the levels of NO and ET-1 stimulated by insulin. 24 As shown in Figure 2

| PPARγ improves endothelium-dependent vasodilation in IR rats
As endothelial dysfunction occurs followed the vascular endothelial IR, we next examined the effects of PPARγ expression on endothelium-dependent vasodilation in vivo. As expected, the endotheliumdependent vasodilation induced by Ach in IR rats (IR group) was decreased by up to 70% compared with the Ctrl rats, whereas the endothelium-independent vasodilation caused by SNP failed to be affected at all in IR rats (IR group). Overexpression of PPARγ, nevertheless, almost restored the endothelium-dependent vasodilation damaged by HFD rather than the endothelium-independent vasodilation ( Figure 3A,B); down-expression of PPARγ, in contrast, worsened the endothelium-dependent rather than endothelium-independent vasodilation damaged by HFD ( Figure 3C,D).

| Physical interaction of PPARγ with P65
contributes to decreases of cytokines in HUVEC As shown in Figure 6A,B, after exposure of HUVEC to HG for 48 hours, both endogenous and exogenous interaction of PPARγ with P65 occurred. Since the interaction of PPARγ with P65 hinders the binding of heterodimer of P65 and p50 to the promoter of target genes,the levels of cytokines including TNFα, IL-6, sICAM-1 and sVCAM-1 were tested. As anticipated, the levels of TNFα, IL-6, sICAM-1 and sVCAM-1 in IR group were increased by 62%, 50%, 56% and 65% respectively (vs. Ctrl). However, overexpression of PPARγ opposed whereas the down-expression worsened the increases induced by HG ( Figure 6C-F).

| Effects of PPARγ expression on the levels of cytokines in the IR rats
The levels of serum cytokines including TNFα, IL-6, sICAM-1 and sVCAM-1 were tested in the IR rats. Consistent with those in vitro, the levels of TNFα, IL-6, sICAM-1 and sVCAM-1 in IR rats were F I G U R E 1 PPARγ expression in HEK293T, 3T3-L1 and vascular endothelial cells in vitro and in vivo. The 90% confluent HEK293T cells (A) and 3T3-L1 cells (B) were transfected with Ad-PPARγ and Ad-PPARγ-shRNA respectively. Besides, the 90% confluent HUVEC were pre-treated with freshly prepared complete DMEM containing HG for 48 h and then transfected with Ad-PPARγ (C), Ad-PPARγ-shRNA (D) and their respective scrambles (Vehicle) respectively. After transfection for 24 h, the cells were washed with PBS twice and further cultured with fresh serum-free DMEM for an additional 12 h. Finally, the cells were harvested and PPARγ expression levels were detected by Western blots. In addition, the aortal endothelia from rats tranfected with Ad-PPARγ (E) and Ad-PPARγ-shRNA containing adenoviral vectors (F) were used to examine the PPARγ expression levels by Western blots. Data are expressed as mean ± SEM of 4 in vitro and 6 in vivo independent experiments, respectively. *P < 0.05, **P < 0.01, vs. Ctrl; # P < 0.05, ## P < 0.01, vs. IR or IR+Veh, ns = no significance. Ctrl: normal control, IR: insulin resistance, Veh: vehicle, PPARγ: Ad-PPARγ, shRNA: Ad-PPARγ-shRNA, IR+Veh: IR+vehicle, IR+PPARγ: IR+Ad-PPARγ, IR+shRNA: IR+Ad-PPARγ-shRNA increased by 64%, 113%, 230% and 62% respectively (vs. Ctrl group). However, overexpression of PPARγ notably reduced while the down-expression further elevated the increased serum cytokines induced by HFD ( Figure 7A-D).

| DISCUSSION
PPARγ is a nuclear receptor that acts as a transcription factor upon activation, by regulating the transcription and expression of specific genes encoding proteins involved in insulin signalling and gluco-lipid metabolism. PPARγ is highly expressed in adipose tissue, skeletal muscle, liver, pancreatic β-cells, heart, colon, placenta and in the cells of vascular and immune systems 25 and plays critical roles in regulating insulin sensitivity, gluco-lipid metabolism and adipogenesis. 10,26,27 In the current study, we have shown overexpression of PPARγ nota- In the study, we have shown that PPARγ was highly expressed in 3T3-L1 cells, moderately expressed in HUVEC and HEK293T cells. F I G U R E 2 Amelioration of vascular endothelium IR in vitro and in vivo by PPARγ. The 90% confluent HUVEC were pretreated with freshly prepared complete DMEM containing HG for 48 h and then transfected with Ad-PPARγ, Ad-PPARγ-shRNA and their respective scrambles (Vehicle), respectively. After transfection for 24 h, the cells were washed with PBS twice and further cultured with fresh serum-free DMEM for additional 12 h. Subsequently, the cells were treated with insulin (5 mIU L −1 , final concentration) for 10 min. At the end, the supernatants were collected and used for the assay of the levels of nitrite (A) and ET-1(B); and the cells were used to detect the expression of AKT and p-AKT (C). Besides, the serum levels of nitrite (D) and ET-1(E) as well as the expression of AKT and p-AKT from aorta tissue (F) in Ad-PPARγcontaining rats and Ad-PPARγ-shRNA-containing rats were measured. Data are expressed as mean ± SEM of 4 in vitro and 6 in vivo independent experiments respectively. *P < 0.05, **P < 0.01, vs. Ctrl; #P < 0.05, ##P < 0.01, vs. IR or IR+Veh. Ctrl: normal control, IR: insulin resistance, IR+Veh: IR+vehicle, IR+PPARγ: IR+Ad-PPARγ, IR+shRNA: IR+Ad-PPARγ-shRNA the expression of PPARγ, indicating that construction and transfection of the viral vectors were successful.
Next we tested whether the changes of PPARγ expression levels affected endothelium IR and dysfunction. It is well known that vascular endothelium is not only a vascular barrier, but also an important endocrinal organ. 28 It secrets numerous vasoactive substances including NO and ET-1 to fine-tune normal vessel integrity and tension stimulated by a physiological dose of insulin. 29 F I G U R E 3 PPARγ improves endothelium-dependent vasodilation in the IR rats The rats in IR+PPARγ or IR+shRNA group were intravenously administered with Ad-PPARγ or Ad-PPARγ-shRNA respectively, while the rats in IR or IR + Veh group were intravenously given normal saline or a vehicle. The rats that were neither treated with HFD nor transfected were considered as a normal control group (Ctrl). Then, the rat aorta was used to assess the vasodilation function 1 week after treatment. Data are expressed as mean ± SEM of 6 rats. *P < 0.05, **P < 0.01, ***P < 0.001, vs. Ctrl; # P < 0.05, ## P < 0.01, vs. IR or IR+Veh, ns = no significance. Ach: acetylcholine, SNP: sodium nitroprusside. Ctrl: normal control, IR: insulin resistance, IR+Veh: IR+vehicle, IR+PPARγ: IR+Ad-PPARγ, IR+shRNA: IR+Ad-PPARγ-shRNA F I G U R E 4 Effects of PPARγ on eNOS expression in vitro and in vivo. The expression levels of eNOS were examined by Western blots when PPARγ were overexpressed in vitro (A) and in vivo (C), and down-expressed in vitro (B) and in vivo (D). The grouping and treatments in vitro and in vivo were the same as described in Figures 2 and 3, and data are expressed as mean ± SEM of 4 in vitro and 6 in vivo independent experiments, respectively. *P < 0.05, **P < 0.01, ***P < 0.001, vs. Ctrl; #P < 0.05, ##P < 0.01, vs. IR or IR+Veh. Ctrl: normal control, IR: insulin resistance, IR+Veh: IR+vehicle, IR+PPARγ: IR+Ad-PPARγ, IR+shRNA: IR+Ad-PPARγ-shRNA Production of NO and ET-1 induced by insulin is modulated by IRS-1/PI3K/AKT/NO and Raf/MAPK/ERK/ET-1 pathways. Under normal conditions, the two pathways are kept in a balance. However, due to long-term hyperglycaemia, hyperlipidemia and hyperinsulinemia, the balance is broken and the endothelium IR and dysfunction resultantly occurs, which results in cardio-cerebral vessel event cascades. 30,31 Accordingly, NO and ET-1 are often chosen as the markers of endothelium IR. 24 In the current study, we have demonstrated that exposure of HG or HFD noticeably decreased the NO levels but increased the ET-1 levels, resulting in the endothelium IR, and we further identified that endothelium IR and dysfunction which were induced by HG or HFD were mediated by an NFκB-independent pathway. Over-production of reactive oxygen species . The grouping and processing in vitro and in vivo were the same as described in Figures 2 and 3, and data are expressed as mean ± SEM of 4 in vitro and 6 in vivo independent experiments, respectively. *P < 0.05, **P < 0.01, vs. Ctrl; #P < 0.05, ##P < 0.01, vs. IR or IR+Veh. Ctrl: normal control, IR: insulin resistance, IR+Veh: IR+vehicle, IR+PPARγ: IR+Ad-PPARγ, IR+shRNA: IR+Ad-PPARγ-shRNA is composed of endothelium-dependent induced by Ach and endothelium-independent triggered by SNP. The former reflects the endothelium function while the latter does vascular smooth muscle function. 32 Our present data revealed that in the aorta from the IR rat, the endothelium-dependent vasodilation induced by Ach was severely damaged but the endothelium-independent vasodilation F I G U R E 6 Physical interaction of PPARγ with P65 contributes to decreases of cytokines in HUVEC. The cells were harvested to assay the association of PPARγ with NFκB-P65 by immunoprecipitation (A and B) and the supernatants were used to measure the levels of TNFα (C), IL-6 (D), sICAM-1(E), and sVCAM-1(F) by ELISA. The grouping and processing were the same as described in Figure 2 and data are expressed as mean ± SEM of 4 independent experiments. *P < 0.05, **P < 0.01, vs. Ctrl; #P < 0.05, vs. IR or IR+Veh. Ctrl: normal control, IR: insulin resistance, IR+Veh: IR+vehicle, IR+PPARγ: IR+Ad-PPARγ, IR+shRNA: IR+Ad-PPARγ-shRNA F I G U R E 7 Effects of PPARγ expression levels on the levels of cytokines in IR rats. The serum levels of TNFα (A), IL-6 (B), sICAM-1(C), and sVCAM-1(D) were measured by ELISA. The grouping and processing were the same as described in Figure 3 and data are expressed as mean ± SEM of 6 rats. *P < 0.05, **P < 0.01, ***P < 0.01, vs. Ctrl; #P < 0.05, ##P < 0.01, ###P < 0.001, vs. IR or IR+Veh. Ctrl: normal control, IR: insulin resistance, IR+Veh: IR+vehicle, IR+PPARγ: IR+Ad-PPARγ, IR+shRNA: IR+Ad-PPARγ-shRNA triggered by SNP was not at all, suggesting that in IR or early state of diabetes, dysfunction of endothelial cells occurs while the function of vascular smooth muscle cells remains intact. However, overexpression of PPARγ considerably improved the endotheliumdependent rather than the endothelium-independent vasodilation. In contrast, down-expression of PPARγ significantly exacerbated the endothelium-dependent vasodilation but had no effect on the endothelium-independent vasodilation. Taken together, these data suggest that the changes of PPARγ expression levels indeed influence function of endothelium rather than vascular smooth muscle.
As discussed above, overexpression of PPARγ improves endothelium IR and function, which is involved in the production and availability of NO. It is well known that eNOS catalyzes L-arginine to convert into NO in vascular endothelium under insulin stimulation.
NO diffuses into VSMC where it activates guanylate cyclase (GC).
Activated GC catalyzes GTP to form GMP which sequentially activates protein kinase G (PKG). PKG ultimately leads to a decrease of free cytoplasm calcium concentration, which further results in dephosphorylation of myosin light chain (MLC) and vasodilation. 33 In the current study, we have shown that HG or HFD reduced the expression levels of eNOS and PPARγ. Nevertheless, overexpression of PPARγ markedly reversed the reduction of eNOS expression levels induced by HG or HFD; conversely, down-expression of PPARγ significantly deteriorated the reduction, suggesting that eNOS-dependent NO production is mediated by PPARγ.
Various studies have confirmed that IR and AS are a chronic inflammation process. 34,35 Besides regulating the metabolism of gluco-lipids and adipogenisis, PPARγ has an anti-inflammatory effect. 36,37 A large number of studies have reported that PPARγ plays above-mentioned roles mainly via two mechanisms. [38][39][40] One is the trans-activation of PPARγ-dependent insulin pathway, and the other is the trans-repression of PPARγ-dependent NFκB pathway. The former is a canonical mechanism; that is, upon activation by endogenous or synthetic ligands, PPARγ heterodimerizes with retinoid X receptor (RXR). The PPARγ/RXR heterodimer undergoes conformational changes which alter co-activator/co-repressor dynamics and binds further to PPRE in the promoter region of the target genes. Thus, transcription initiation of the target genes takes place. 41,42 The latter is a non-canonical mechanism; namely, activated PPARγ/RXR heterodimer interacts physically with NFκB and impedes NFκB to bind with the promoter region of the target genes encoding inflammation factors such as TNFα, IL-6, sICAM-1 and sVCAM-1 and therefore plays anti-inflammation. 43,44 However, it is unclear whether the changes of PPARγ expression affect NFκB pathway. In the study, we have shown that HG or HFD indeed increased the levels of p-IKKα/β and cytokines including TNFα, IL-6, sICAM-1 and sVCAM-1 and decreased the IκBα levels. It might be due to the fact that activated IKKα/β, an upstream kinase of IκBα, phosphorylates IκBα at the sites of serine 32 and 36, leads to the IκBα degradation, and then gives rise to the translocation and activation of NFκB (P65/P50), and eventually causes the expression of inflammation genes (TNFα, IL-6, sICAM-1 and sVCAM-1, etc).
Nonetheless, overexpression of PPARγ markedly normalized the changes induced by HG or HFD and promoted the interaction between PPARγ and NFκB-P65. By contrary, down-expression of PPARγ significantly enhanced the changes induced by HG or HFD.
These data demonstrate that overexpression of PPARγ may repress NFκB trans-activation and improve the endothelium IR through a PPARγ-dependent NFκB trans-repression mechanism.
In conclusion, the present study confirmed that the changes of PPARγ expression affected endothelium IR. Overexpression of PPARγ improved endothelium IR while down-expression of PPARγ worsened endothelium IRvia a PPARγ-dependent NFκB transrepression pathway. Since loss of PPARγ function exists in the patients with T2DM and AS, 45 the current findings suggest PPARγ is a potential therapeutic target for diabetic vascular complications.

CONFLI CT OF INTEREST
The authors confirm that there are no conflicts of interest.