Dipeptidyl Peptidase-4 Inhibitor Sitagliptin Attenuates Arterial Calcication By Downregulating Oxidative Stress-Induced Receptor for Advanced Glycation End Products in Low-Density Lipoprotein Receptor Knockout Mice

Background: Plasma advanced glycation end products (AGEs) activates the receptor for advanced glycation end products (RAGE) and the activation of RAGE is implicated to be the pathogenesis of type 2 diabetic mellitus patient vascular complications. Attenuating the activation of RAGE may exert a protective effect against the development of cardiovascular disease. Dipeptidyl peptidase-4 (DPP4) inhibitors are a new class of oral hypoglycemic agents for the treatment of type 2 diabetes mellitus. Whether sitagliptin, a DPP-4 inhibitor, has a benecial effect on vascular calcication remains undetermined. Methods: In the present study, we fed low-density lipoprotein receptor knockout (LDLR -/- ) mice a high fat diet to induce diabetic mellitus and studied the effect of orally administered sitagliptin on the high fat diet fed LDLR -/- mice aorta medial calcication, RAGE expression, oxidative stress, aorta calcium content. Tumor necrosis factor (TNF)-α combined with S100A12 was used to induce HASMC oxidative stress, activation of NADPH, up-regulation of the bone markers and RAGE expression, and cell calcium deposition. Effect of sitagliptin, siRNA for RAGE and apocynin on blunting TNF-α and S100A12 induced HASMC oxidative stress, calcication and NADPH activation were also investigated. Results: Sitagliptin attenuated the HFD-induced LDLR -/- mice hyperlipidemia, hyperglycemia, increase in serum TNF-α, aorta calcium deposition and the expression of RAGE in the medial layer of the aorta. TNF-α combined with S100A12 stimulated HASMC RAGE expression, calcium deposition, nicotinamide adenine dinucleotide phosphate (NADPH) oxidase (Nox) activation, and up-regulation of bone marker (bone morphogenetic protein-2, Msh homeobox 2, and runt ‐ related transcription factor 2) expression. Sitagliptin and apocynin (APO), an NADPH oxidase inhibitor, suppressed the TNF-α+S100A12 effects on the activation of NADPH oxidase and (NF)-κB and the resultant oxidative stress, up-regulation of RAGE and bone markers expression and calcium deposition. Our ndings suggest that sitagliptin imparts its protective effect by suppressing NADPH oxidase and NF-κB activation to blunt the up-regulation of RAGE expression.

Sitagliptin and apocynin (APO), an NADPH oxidase inhibitor, suppressed the TNF-α+S100A12 treatment effects on the activation of NADPH oxidase and Nuclear factor (NF)-κB and the resultant oxidative stress, up-regulation of RAGE and bone markers expression and calcium deposition. Our ndings suggest that sitagliptin imparts its protective effect by suppressing NADPH oxidase and NF-κB activation to blunt the up-regulation of RAGE expression.
Conclusion: Our ndings suggest that sitagliptin may suppress the initiation and progression of artery calci cation by inhibiting the activation of NADPH oxidase and NF-κB and the resultant up-regulation of expression of RAGE. Background Type 2 diabetes mellitus (DM2) patients have a higher risk of developing atherosclerosis, artery calci cation, morbidity and mortality rates than non-diabetic population [1,2]. The mechanism causing the DM2 patients to have a higher risk of vascular calci cation is not yet fully understood. Hyperglycemia and hyperlipidemia are suggested to be the major risk factors [3]. The pathogenesis of artery calci cation involves various types of cells and complicated regulated processes. Smooth muscle cells (SMCs), are believed to be involved in the pathogenesis of vascular calci cation [4]. Medial calci cation represents a concentric calci cation that is preceded by matrix vesicle-nucleated mineralization accompanied with calcium phosphate deposits in the arterial tunica media [5]. Multiple factors, such as in ammation, oxidative stress, adiposity, insulin resistance, and advanced glycation end-products, contribute to the induction and progression of diabetic medial calci cation [6][7][8]. Important transcription factors, such as Msh homeobox 2 (MSX2), Osterix, and runtrelated transcription factor 2 (RUNX2), are crucial in the programing of osteogenesis [9,10]. Currently, no therapy is available to reverse the vascular calci cation.
Understanding the association between the DM complications and HASMC osteotransformation could aid the development of anti-artery calci cation therapeutics.
In addition to elevated bone matrix protein (BMP) expression and calcium phosphate deposit in vascular SMCs (VSMCs), diabetic patients also have a higher level of plasma advanced glycation end products (AGEs) and receptors for AGEs (RAGEs) than the nondiabetic population [8,11]. AGEs activate RAGE and AGE/RAGE signaling enhances VSMC calci cation by activation of nicotinamide adenine dinucleotide phosphate (NADPH) oxidase (Nox) 1 pathway [12]. Recent studies suggested that DM2 patients has a higher level of circulatory S100A12, a RAGE agonist than the non-diabetic person [13]. This RAGE-S100A12 interaction is implicated to cause coronary artery diseases and vascular calci cation [14,15].
The RAGE is expressed in endothelial and smooth muscle cell (SMC), and the activation of RAGE augments in ammation by stimulating the release of in ammatory cytokines and generation of oxidative stress [16]. It was found that S100A12 transgenic apolipoprotein-E (ApoE) null mice on regular rodent chow developed much more server calci ed atherosclerotic plaques and medial calci cation than the wild type ApoE null animal [14]. The authors concluded that the RAGE/S100A12 signaling play a signi cant role in the diabetic artery atherosclerosis and calci cation. It is suggested that the activation of RAGE activates nuclear factor (NF)-κB which leads to a wide spectrum of pathological in ammation conditions and results in various diseases such as atherosclerosis [17]. The activation of NF-κB also upregulates RAGE expression and this positive feedback loop of between RAGE and NF-κB signaling pathway results in a perpetuation of in ammation state [18]. AGEs, amyloid beta peptide, DNA-binding protein high-mobility group box-1/amphoterin, and S100/calgranulins activate RAGE and NF-κB [18]. A previous study using transgenic human S100A12 CLB6/J mice to investigate the role of S100A12 on artery calci cation using a chronic kidney disease mouse model [19]. This study showed that the oxidative stress induced by ureteral obstruction played a critical role for the RAGE/S100A12 signaling to induce aorta calci cation.
Dipeptidyl peptidase-4 (DPP-4) is a multifunctional enzyme found in catalytically active soluble form in plasma and on the surface of most cell types [20]. DPP4 knockout mice study revealed that the absence of this enzyme improves glycemic control and reduces animal fat mass [21]. DPP4 degrades incretin hormones, such as type I glucagon-like peptide (GLP-1) that is widely known for its regulatory effect in glucose metabolism [22]. DPP4 inhibitors are a new class of oral hypoglycemic agents used for the treatment of type 2 diabetes mellitus without causing weight gain [23]. Sitagliptin and alogliptin improve endothelial function and impart an anti-in ammatory effects which is suggested to retard the progression of carotid atherosclerosis in DM2 patients [24,25]. Sitagliptin and alogliptin were reported to inhibit the progression of atherosclerosis in ApoE-de cient mice [26,27]. Recent animal studies suggest that DPP4 inhibitor may contribute its anti-atherosclerotic effects by reducing the reactive oxygen species (ROS) generation, preventing mitochondrial depolarization, improving endothelial functions, and reducing vascular in ammation [28]. Hyperglycemia and dyslipidemia are known risk factors associated with artery calci cation. Thus, DPP4 inhibitor treatment may blunt the development of atherosclerosis in diabetic patients. Gemigliptin was found to protect against vascular calci cation in a adenine induced chronic kidney disease rat model and phosphate induced VSMC calci cation [29]. It was reported that TNF-α play a crucial role in artery calci cation in diabetes LDLR −/− mice [7]. Our previously study showed that tumor necrosis factor (TNF-α) and vascular TNF receptor (TNFR) signaling lead to human artery smooth muscle cell (HASMC) calci cation and antioxidants blunt the TNF-α/TNFR signaling to retard the HASMC calci cation [30]. We hypothesized that the hyperglycemia and hyperlipidemia of the DM2 patients induce systematic oxidative and the resultant elevation of circulatory (TNF)-α and S100A2 triggers the vascular calci cation. Sitagliptin may improve the DM2 patient artery calci cation by either improve the patient hyperglycemia and hyperlipidemia conditions or by functioning as an antioxidant to bunt TNF-α/TNFR signaling. In this study, we used HFD to induce diabetic condition in LDLR −/− mice to study the effect of orally administered sitagliptin on the artery calci cation and RAGE expression. We use TNF-α + S100A12 treated cultured HASMCs as a model system to study how the systemic in ammation affect DM2 patient aorta calci cation. We also explored the possible mechanisms by which sitagliptin contributes its protective effects.

Animal study
Weaned male low-density lipoprotein receptor knockout (LDLR −/− ) mice (Jackson Labs #002207; C57BL/6J background) were fed with a normal diet (Picolab Rodent Diet 20 #5053: 5% fat, 21% protein, 3.3% sucrose, and 28% starch) for 2 months then the animals were randomized into 3 groups: 1) normal diet group, 2) high fat diet (HFD) Harlan Teklad, Diet TD88137 (21% milk fat (42% fat calories), 34% sucrose, and 0.15% cholesterol)) group and 3) HFD + sitaglptin group. Animals in group 2 are gavaged with 100 µL of distilled water daily. Animals in group 3 were given 100 mg − 1 Kg − 1 day − 1 of sitagliptin by gavaging various amount of sitagliptin solution (25 mg/mL) (BioVision Research Products, Milpitas, CA). All mice were kept in microisolator cages under a 12 h day/night cycle. The entire animal was given free access to chow and water. All animal study protocols complied with the Guide for the Institutional Animal Care and Use Committee of Taipei Veterans General Hospital (IACUC no.2020 − 265, Taipei, Taiwan) and the Guide for the Care and Use of Laboratory Animals of the US National Institutes of Health (8th edition, 2011). The LDLR −/− mice were sacri ced after 24 weeks of treatments by exsanguination under anesthesia (100 mg − 1 kg − 1 ketamine-HCl and 20 mg − 1 kg − 1 xylazine via IP injection) after 6 hours fasting. The animals were considered as adequately anesthetized when no attempt to withdraw the limb after pressure could be observed. The thoracic cavity was opened for blood and aorta (from heart to diaphragm) sample collections.

Histology and immunohistochemistry
Aorta samples were cut into 4 sections and processed for histological staining as described in our previous study [30]. Para n sections (5 µm) from the dissenting aorta were stained using various agents for semi-quanti cation of atherosclerotic lesion size and severity (hematoxylin and eosin (H&E) staining) and aortic calcium deposition (alizarin red S staining). Immunohistochemical (IHC) staining of RAGE and VSMC actin (SM α-actin) was performed as previously described [31].

Cell cultures and cell viability assay
HASMCs were purchased from Life Technology (Grand Island, NY. catalog number C0075C). The cells were grown and passaged as described previously [30]. Brie y, the HASMCs were grown in M231 medium containing SMC growth supplement and a 1% antibiotic-antimycotic mixture in an atmosphere of 95% air and 5% CO 2 at 37°C in plastic asks. At con uence, the cells were subcultured at a ratio of 1:3, and passages 3 through 8 were used. The cytotoxicity of S100A12 protein and sitagliptin on HASMC cell viability were measured with the 3-(4,5-dimethylthiazol-2-yl)-2, 5-diphenyl tetrazolium bromide (MTT) assay.
Quanti cation of aorta or cultured HASMC calcium deposit Cultured cell Ca content was determined using a BioChain Calcium Kit (BioChain, Hayward, CA, USA) as previously described [32]. Brie y, the 6 well plate cultured cells were counted and demineralized with 250 µl 0.6 N HCl for 12 h. A working reagent was prepared by mixing 75 µl reagent A and 75 µl reagent B and was equilibrated to room temperature before use. A volume of 50 µl diluted standards or samples were transferred to each well of a clear-bottom 96-well plate. Then, 200 µl working reagent was added, and the solution was mixed by light tapping. After incubation for 3 min at room temperature, absorbance was measured at 570-650 nm with a 96-well reader. The units of these results are ug/mL calcium. The other 6 well plate cultured cells counted and solubilized in 200 µl lysis solution containing 0.1 N NaOH and 0.1% sodium dodecyl sulfate (SDS) at room temperature for 5 min. Protein concentration was measured with a Bio-Rad DC Protein Assay Kit. The calcium content of the cultured cells normalized to the protein content was reported.
The aortic segments from experimental mice were extracted using 0.6 N HCl for 24 h and the calcium content of the extracts were determined using a BioChain Calcium Kit (BioChain, Hayward, CA, USA) as previously described [30]. The results were expressed as µg/mg of wet aortic tissues.
Nox activity assay and hydrogen peroxide determination Nox activity was determined with superoxide-dependent lucigenin chemiluminescence, as previously described [30]. Con uent HASMCs in 6-well plates were pretreated with various concentrations of antioxidant reagents followed by treated TNF-α + S100A12 with or without sitagliptin for 1 day. Cell membrane extract (40 µg) and 5 µM dark-adapted lucigenin were added to a 96-well luminometer plate and adjusted to a nal volume of 250 µl with oxidase assay buffer before 100 µM NADPH was added. Relative light units (RLUs) were measured with a luminometer (Dynatech ML2250, Dynatech Laboratories Inc., VA). Light emission was recorded every 3 minutes for total 30 minutes and expressed as mean RLUs/min. The ROS production in HASMCs was determined by uorometric assay using dichloro-dihydro-uorescein diacetate (DCFH-DA) as the probe. This method is based on the oxidation by H 2 O 2 of non uorescent DCFH-DA to uorescent 2',7'-dichloro uorescin. Con uent HASMCs in 24-well plates were pretreated with various concentrations of antioxidant reagents followed by treated TNF-α + S100A12 with or without sitagliptin for 1 day. The cells were washed with PBS, then 250 µL of serum-free M231 containing 10 µM DCFH-DA was added to the well for 30 minutes. The uorescence intensity (relative uorescence units) was measured at 485 nm excitation and 530 nm emission using a uorescence microplate reader after the plates are incubated for 45 min at 37 ℃. siRNA transfection siRNA oligonucleotides against RAGE were suspended in RNase-free water at a concentration of 10 µM.
Cells were seeded one day before transfection to ensure HASMC were 85-95% con uent on the day of transfection. For transfection, the regular cell culture medium was replaced with a serum-free medium without antibiotics. The cells were transfected with siRNA using oligofectamine at a ratio of 1 siRNA: 2 oligofectamine (µg:µl) at a nal concentration of 25-50 nM siRNA. The cells were incubated with the siRNA-oligofectamine complex for 5 h. Then, the serum-free medium was replaced with a normal medium (containing 10% FBS) without antibiotics, and the cells were incubated for 48 h before further analysis.

Statistical analyses
Data were expressed as means ± standard deviation (SD). Statistical evaluation was performed using Student's t-test or one-way analysis of variance, followed by Dunnett's test. A P value of < 0.05 was considered signi cant.

Results
Effect of HFD and Sitagliptin on HFD-induced LDLR −/− body weight gain, aortic calci cation and Atherosclerotic plaque formation.
We found that HFD LDLR −/− mice has a signi cant higher body weight than the regular chow fed ones (33.8 ± 1.9 VS. 41.8 ± 2.3 g). Sitagliptin attenuated HFD-induced body weight gain ( Fig. 1a; 41.8 ± 2.3 VS. 36.8 ± 2.5 g). HFD fed animals has a higher level of blood glucose, triglyceride (TG), cholesterol (CHL), low-density lipoproteins (LDL) and TNF-α than the regular chow fed animals (Table 1). These ndings suggest that HFD fed LDLR −/− mouse is a valid model as an obesity-associated type II diabetes animal model. Sitagliptin treatment moderately lowered the HFD fed animal blood glucose but drastically lowered the fasting triglyceride (TG) (539 ± 121 VS. 358 ± 128 mg/dL) and TNF-α level (53 ± 17 VS. 30 ± 7 pg/mL). Numerically, a moderate reduction in LDL was observed, but the level of reduction is not statistically signi cant. HFD-fed LDLR −/− mice had a remarkable increase in calci cation in the medial layer of the upper descending aorta as indicated by alizarin red staining and aorta calcium content analysis ( Fig. 1b and   1c) compared to the regular chow fed animals. HFD also caused the LDLR −/− mice to have a much higher serum DPP4 activity (Fig. 1d). Sitagliptin treatment signi cantly blunted the HFD induced calcium deposition in the media of the descending aorta (Fig. 1b), aorta acid extract Ca (Fig. 1c) and serum DPP4 activity (Fig. 1d). Our ndings showed that HFD causes the LDLR −/− to become obsess, hyperglycemic and hyperlipidemic which lead to oxidative stress and elevated TNF-α. (*your data showed that is what happened). The elevated TNF-α may induce a cascade of events leads to the initiation and propagation of artery atherosclerotic plaque formation and artery calci cation. Sitagliptin protected the HFD fed LDLR −/− mice from developing severe artery calci cation. Our ndings suggest a possibility that sitagliptin may protect the obese type 2 DM patients against the initiation and progression of atherosclerotic plaque formation and aorta calci cation for by blunting the cascade of events induced by the elevated TNF-α.
Effect of TNF-α and S100A12 on HASMC cell RAGE expression and calcium deposition It is known that obese type 2 DM patients has higher levels of TNF-α and S100A12 compared to nondiabetic subjects [34]. The S100A12/RAGE signaling is associated with the severity of coronary artery diseases and vascular calci cation of the type 2 diabetes mellitus patients [14,15]. We use human artery smooth muscle cells (HASMC) model to investigate 1) if TNF-α + S100A12 treatment affect HASMC calci cation and 2) if and how sitagliptin blunts the effect of TNF-α + S100A12 treatment. We found that TNF-α (10 ng/mL) treatment signi cantly increased the HASMC S100A12 expression (Fig. 2a). This nding suggests that TNF-α may induce a pathological condition for the HASMC to express S100A12. Under physiological condition, S100A12 is mainly expressed by the myeloid cells. However, under a pathological condition such as coronary artery plaque rupture, S100A12 is expressed in HASMC [35]. Incubating HASMC with S100A12 did not signi cantly increase the HASMC S100A12 expression (Fig. 2a). It is possible that the S100A12 may have activated the HASMC RAGE to trigger a cascade of events results in a small increase in S100A12 expression but the increase did not reach statistically signi cant level (Fig. 2a). Before we investigate effect of S100A12 on HASMC RAGE expression, we investigate the toxicity of S100A12 treatment. We found that S100A12, up to 200 ng/ml level, did not impair the HASMC cell vitality as indicated by cell proliferation rate (Fig. 2b). Incubating HASMC with increasing level of S100A12 produced a dose dependent increase in HASMC RAGE expression (Fig. 2c).
Our data suggest that the S100A12 in the medium activated the HASMC RAGE which then triggered a mechanism to up-regulate the HASMC RAGE expression. It was reported that RAGE signaling leads to activation of NF-κB and activation of NF-κB stimulates RAGE expression [18]. In the presence of TNF-α (10 ng/mL), S100A12 produced a much more pronounced dose response on HASMC RAGE expression (Fig. 2c). Figure 2d showed that the potency of S100A12 treatment on HASMC calcium deposits were much more pronounce in the presence of TNF-α. Previous study from this laboratory showed that incubating HASMC with TNF-α activate NF-κB as indicated by the increase in P65 in the HASMC nucleus [7]. Figure 2c and 2d suggest that RAGE/S100A12 and NF-κB forms an perpetual amplifying loop as suggested by literature [18]. It was proposed that activation of RAGE results in oxidative stress which lead to activation of NF-κB and the activated NF-κB stimulates more RAGE expression to form a perpetual amplifying loop between S100A12 [18].
Effect of sitagliptin, N-acetyl cysteine (NAC), small interfering RNA (siRNA) for TNF receptor 1 (TNFR-1) and Si RNA for RAGE on the TNF-α plus S100A12 treated HSMC RAGE expression and calcium It has been reported that that tenegliptin, a DPP4 enzyme inhibitor, have anti-atherosclerotic effects by reducing the production of reactive oxygen species (ROS) and in ammatory cytokines in adipocytes and aorta [28]. The authors suggested that there may be a not yet fully understood GLP-1 independent mechanism that also contributed to the DPP4 inhibitor anti-atherosclerotic effect. Our LDLR −/− mice study showed that Sitagliptin only produced a small reduction of the HFD fed LDLR −/− mice serum glucose and LDL level yet it is very effective in protecting the HFD fed LDLR −/− mice from developing aorta calci cation ( Fig. 1 and Table 1). Previous study from this lab showed that N-acetyl cysteine (NAC) and APO were very effective in reducing the serum nitrotyrosin and aorta calcium deposit of the HFD fed LDLR −/− mice [30]. It was proposed that the RAGE signaling was at least in part mediated by the oxidative stress from NADPH oxidase [18]. It is possible that Apo inhibits the NADPH oxidase activity and NAC neutralized the ROS produced by NADPH oxidase to exert their protective effect on HFD induced aorta calci cation observed in our previous study. We further speculated that sitagliptin may 1) function as an anti-oxidant to neutralize the oxidative stress induced by HFD and/or 2) it may inhibit the NADPH oxidase to alleviate the high fat diet induced oxidative stress to protect the HFD fed LDLR −/− mice from developing aorta calci cation. Since there is no commercial source of mouse S100A9/9 and AGE (advance glycated end produce), we used TNF-α + S100A12 treated HASMC as a model system to test our hypothesis. We incubated HASMC with various dose of sitagliptin and found that sitagliptin was not toxic to HASMC as indicated by the cell proliferation rate (Fig. 3a). As previous found, combined TNF-α + S100A12 treatment up-regulated HASMC RAGE expression and calci cation (Fig. 2c, 2d and 3b). Sitagliptin treatment appear to exert a does depend effect on suppressing the TNF-α + S100A12 induced calci cation (Fig. 3b) 50 µM of sitagliptin was as effective as 1000 µM of N-acetyl cysteine (NAC), in blunting the TNF-α + S100A12 induced calci cation (Fig. 3b). Since one molecule of antioxidant can only neutralize one molecule of ROS, thus, we suspect that sitagliptin may be functioning as a noncompetitive NADPH oxidase inhibitor or it prevent the activation of NADPH activation. We found that siRNA for RAGE signi cantly lowered the TNF-α + S100A12 induced RAGE (Fig. 3b) and calci cation without affecting TNF receptor (TNFR) expression (Fig. 3c). As with siRNA for RAGE, we found that sitagliptin treatments blunted the TNF-α + S100A12-induced RAGE expression but did not affect TNFR expression (Fig. 3c). The siRNA for TNFR-1 is very effective in blunting the TNF-α + S100A12 induced upregulation of the RAGE and TNFR-1 expression (Fig. 3b) and decreased calci cation (Fig. 3c). It is reported that TNFR1-induced activation of the classical NF-κB pathway [36], p65 shRNA downregulation RAGE protein accumulation [37] and Vascular remodeling and arterial calci cation are directly mediated by S100A12 (EN-RAGE) [19]. These nding suggests that the cross talk between RAGE/S100A12 signaling and NF-κB pathway stimulate HASMC RAGE expression which lead to cell calci cation. Sitagliptin may disrupt the cross talk between RAGE signaling and NF-κB either by inhibiting NADPH oxidase activity or preventing it from being activated to blunt the effect of TNF-a + S100A12 on upregulation of RAGE expression.
Sitagliptin blunts TNF-α + S100A12-induced HASMC oxidative stress, osteogenic markers expression, and calci cation Activated-RAGE promoted NADPH oxidase activity [38] which leading to generation of intracellular H2O2 in vascular SMCs, endothelial cells, and broblasts [39] Oxidative stress plays a crucial role in RAGE signaling to trigger artery calci cation [12]. It was suggested that RAGE pathway involved NADPH oxidase. We hypothesized that sitagliptin disrupted the cross talk between RAGE signaling and NF-κB either by inhibiting NADPH oxidase activity or preventing the activation of NADPH oxidase. If this this hypothesis is true, then, TNF-α + S100A12 treatment should induce oxidative stress and sitagliptin should effectively attenuate the TNF-α + S100A12 induced oxidative stress. Experimental data showed that TNFα + S100A12 treatment stimulated the hydrogen peroxide production in HASMC (Fig. 4a). Sitagliptin, NAC and APO all were found to attenuate the TNF-α + S100A12 treatment effect on HASMC hydrogen peroxide generation (Fig. 4a). As we found in previous experiment, 100 µM of Sitagliptin as effective as the 500 µM APO in reducing the TNF-α + S100A12-induced oxidative stress (Fig. 4a). These ndings support our hypothesis that sitagliptin is unlikely to reduce the TNF-α + S100A12 treated HASMC H 2 O 2 production by directly neutralizing the ROS produced by NADPH oxidase. We investigated 1) if TNF-α + S100A12 treatment stimulates HASMC NADPH oxidase and NF-κB activation 2) if Sitagliptin reduces the NADPH oxidase and NF-κB activation. We found that TNF-α + S100A12 increased HASMC NADPH oxidase activity, presumably, by promoting NADPH oxidase activation (Fig. 4b). One hundred micromolar (100 µM) of sitagliptin was as effective as 500 µM of apocynin (APO), a NADPH oxidase inhibitor, in suppressing the TNF-α + S100A12 treated HASMC NADPH oxidase activity (Fig. 4B). These results showed that TNF-α + S100A12 stimulate the HASMC NADPH oxidase activation and sitagliptin may either inhibit the NADPH oxidase activity or by blocking TNF-α + S100A12 from activating NADPH oxidase activation to attenuate the superoxide generation in HASMCs NADPH oxidase (Nox) consists of membrane-bound (gp91phox and p22phox) and cytosolic components (p47phox, p67phox, p40phox, and Rac proteins). Nox activation requires the translocation of the cytosolic component p47phox to the cell membrane. Figure 4b showed the HASMC membrane NADPH oxidase activity of the control HASMC (no treatment) or cells been treated with TNF-α + S100A12, TNF-α + S100A12 + Apo or TNF-α + S100A12 + sitagliptin for a day. Apo and sitagliptin were very effective in inhibiting the HASMC cell membrane NADPH oxidase activity (Fig. 4b). APO was found to blunt the TNF-α + S100A12 treatment induced migration of cytosolic P65 to the cell nucleus (Fig. 4c). As with APO, sitagliptin also blunted the TNF-α + S100A12-induced migration of cytosolic P65 and P47 (Fig. 4c and 4d). Our data indicated that both Apo and sitagliptin attenuated the TNF-α + S100A12 induced activation of NF-κB and the resultant NADPH oxidase activation. Thus, both Apo and sitagliptin should be capable of disrupting the literature proposed cross talk between RAGE signaling and NF-κB. Thus, we investigated if 1) TNF-α + S100A12 treatment promotes HASMC transgenic markers expression and calcium deposit, and 2) if sitagliptin blunted the TNF-α + S100A12 effect on HASMC osteogenic markers expression and cell calcium deposit. We found that TNF-α + S100a12 treatment signi cantly up-regulated the HASMC osteogenic markers expression and calcium deposit (Fig. 5a and 5b). Sitagliptin signi cantly blunted the effect of TNF-α + S100A12 treatment on HASMC osteogenic markers expression and calcium deposit (Fig. 5a and 5b). Both APO and Sitagliptin were all very effective in blunting the impact of TNF-α + S100A12 on RAGE expression and MSX-2 expression. (Fig. 5c). This nding suggests a possibility that a high level of antioxidant can block TNF-α without affecting RAGE expression.
If TNF-α up-regulate VSMC RAGE expression, then, the HFD fed LDLR −/− aorta should have a higher level of RAGE expression than the regular chow fed ones. In addition, orally administer sitagliptin should drastically reduce the HFD fed LDLR −/− mice aorta RAGE expression. We found that HFD feed LDLR −/− mice indeed have a drastically higher the level of the RAGE expression in the medial layer of the upper descending aorta as indicated by the IHC staining (Fig. 5d). Orally administered sitagliptin signi cantly blunted the HFD induced up-regulation of RAGE expression (Fig. 5d).

Discussion
Sitagliptin, a DPP-4 inhibitor, was the rst drug to be commercialized and approved for treatment of patients with type 2 diabetes mellitus (DM2) in Oct 2006 in USA and in 2007 by the European Medicines Agency at a dosage of 100 mg daily [40]. Since hyperglycemia and dyslipidemia are two known risk factors of arterial calci cation [41], sitagliptin, saxagliptin and alogliptin are commercially available DPP-4 inhibitors which, in addition to lowering DM2 patients' blood glucose level, may also protect patients from cardiovascular diseases [40]. It has been reported that the presence of S100A12, a known RAGE agonist, drastically accelerated the development of artery atherosclerosis and artery calci cation of regular chow fed Apo E -/-mice but S100A12 produced no effect on the Apo E +/+ mice [14]. Vascular smooth muscle cells (VSMC) isolated from the artery of the Apo E -/-S100A12 transgenic mice only develop elevated oxidative stress, expression of osteogenic marker expression and calci cation when the VSMC were grown in a "conditioned media (cell culture media containing macrophage and serum isolated from hyperlipidemic APE -/-mice) [14]. These ndings suggest that the ROS and/or in ammatory cytokines from the macrophages hyperlipidemic Apo E -/-activated may activate a mechanism which synergistically interacts with the RAGE/S100A12 signaling to stimulate artery atherosclerosis and calci cation. One µM of APO or 10 µM of DPI (diphenylene iodonium), two known NADPH oxidase inhibitor, effectively blunt the osteogenic marker expression and cell calci cation in the "conditioned media" grown S100A12 transgenic Apo E -/-VSMC [14].
The authors concluded that RAGE/S100A12 signal induced aorta calci cation is mediated by, at least in part, by NADPH oxidase. It was reported ureter ligation induced chronic kidney disease caused the S100A12 transgenic mice to develop oxidative stress and aorta calci cation. The shame operation did not cause the S100A12 transgenic mice to develop aorta calci cation. These ndings suggest a possibility that CKD induced a mechanism that synergistically interact with the RAGE/S100 signal to stimulate aorta calci cation. The ureter ligation produced no effect on the wild type mice [19].
Inclusion of hydrogen peroxide and phosphate in the medium stimulated the VSMC from the wild type or S100A12 mice aorta to accumulate Ca. The potency of H 2 O 2 + phosphate on promoting VSMC is doubled by the presence of VSMC S100A12 [19]. Transecting the wild type VSMC with si RNA for Nox 1 produced no effect on the H 2 O 2 treated cell calcium deposit, but siRNA for Nox 1 transfection signi cantly reduce S100A12 transgenic VSMC calci cation [19]. These ndings lead us to hypothesize that the ROS and/in ammatory cytokines in circulation, presumably from the macrophages and neutrophils activated by hyperlipidemia or hyperglycemia, activate VSMC NF-κB. The activated NF-κB causes VSMC to express osteogenic markers and up-regulate RAGE expression. The VSMC RAGE is activated by S100A or AGE to trigger the activation of NADPH oxidase. The ROS from the NADPH oxidase further activates NF-κB. This perpetual amplifying cross talk between RAGE signaling and NF-κB pathway leads to VSMC osteotransformation and artery calci cation (SFig 1). If this hypothesis is true, then, incubating HASMC with TNF-α and S100A12 should stimulate RAGE expression, NADPH oxidase and NF-κB activation, osteogenic markers expression and cell calcium deposit. Experimental data con rm our speculation (Figs. 3,4 and 5).
Sitagliptin was found to protect HFD fed LDLR −/− mice against artery calci cation (Table 1 and Fig. 1). Dobrian et al reported that sitagliptin supplement reduced HFD fed wild type mice reduced mice body weights, improved animal glucose tolerance and reduced the level of in ammation cytokines in the adipose and pancreatic islet [42]. Based on typical mice feed intake, we calculated that we used a sitagliptin dose about 4 times lower than what was used in Dobrian et al study [42] yet we found that yet we found our sitagliptin dose is su cient to lower the HFD fed LDLR −/− serum DPP4 activity and nitrotyrosin level (Fig. 1d and Table 1). Previous study from this laboratory showed that intraperitoneal injected APO and NAC are very effective in reducing the HFD fed LDLR −/− mice systemic oxidative stress and in ammation as indicated by serum nitrotyrosin and TNF-α level. They are also very effective in lowering aorta calci cation without drastically improving the hyperglycemia and hyperlipidemia condition [30]. Those ndings lead us to suspect there is a GLP-1 independent mechanism that, at least contributed in part, to the Apo and NAC protective effects against HFD induced aorta calci cation. We further speculate that this GLP-1 independent mechanism may be related to their ability to prevent the accumulation of ROS from NADPH oxidase in VSMC by neutralizing the ROS or inhibiting NADPH oxidase activity. As with APO and NAC, we found that sitagliptin only produced modest improvements in hyperglycemia and hyperlipidemia conditions yet it drastically improve the systemic in ammation and oxidative stress and aorta calci cation (Table 1, Figs. 1 and 6). We thus suspect that the protective effect of sitagliptin may, at least contributed in part, by blunting systemic ROS/in ammatory cytokines induced NF-κB activation or the cross talk between RAGE signaling and NF-κB pathway. We used a HASMC model to test our hypothesis. Sitagliptin and Apo were found to block the TNF-α + S100A12 induced upregulation of RAGE expression, NADPH oxidase and NF-κB activation, osteogenic markers expression and calcium deposit in a HSMC model (Figs. 3, 4, 5 and 6).
Based on literature and our data, we proposed that sitagliptin might attenuate RAGE/S100a12 signaling by inhibiting NADPH oxidase activation. The reduction of ROS production due to lowered NADPH oxidase activity lead to lowered NF-κB activation, thus, lowers stimulation of RAGE expression. As a result, sitagliptin blunts the perpetual cross talk between RAGE signaling and NF-κB pathway to lower the osteogenic markers expression and calcium deposit (Fig. 6). If TNF-α + S100A12 treated HASMC is a good model representing the pathological condition of the hyperlipidemia and hyperglycemic mice aorta, then, HFD fed LDLR −/− mice should have a higher level of RAGE expression than those on regular chow. Experimental data con rmed our speculation (Fig. 6d). Our nding of a correlation between aorta calci cation and RAGE expression agrees with literature data. Wang et al reported that the aorta valve of apoE -/-mice have signi cantly higher expression levels of aorta RAGE and osteogenic markers and calcium deposit than the wild type mice [43].
We invested if GLP-1 attenuated TNF-α-induced HASMC calcium deposit (SFig 1), We found that GLP-1 did not blunt the TNF-α-induced MSAMC calci cation. Our nding that Si RNA for TNF-R1 or RAGE blocked the TNF-α + S100A12 induced RAGE expression and siRNA for RAGE signi cantly blunted the TNF-α + S100A12 induced up-regulation of RAGE expression and calcium deposit (Fig. 3) lead us to conclude that GLP-1 does not blunt the cascade of events initiated by activation of TNFR or RAGE signaling. Since our animal study showed that sitagliptin drastically lowered the serum TNF-α level and aorta calci cation of the HFD fed LDLR −/− mice without dramatically improvement in blood glucose level or blood lipid chemistry (Table 1), we suspect that protective effect of sitagliptin on lowering the circulatory in ammatory may be mainly due to improvement in glucose control or due to GLP-1 exerting anti-in ammatory effect on the HASMC. Multiple studies have shown that DPP-4 inhibitor increase DM2 patients' circulatory GLP-1 level and lowered the circulatory in ammatory cytokines level [44]. It is not clear to us if the effect of sitagliptin on lowering serum circulatory in ammatory cytokines is due to its impact on serum GLP-1 level or because sitagliptin exert a direct impact on macrophage, monocytes and neutrophils in ammatory cytokine expression.

Conclusion
In conclusion, our HASMC experiment showed the protective effect of sitagliptin against aorta calci cation is contributed, at least in part, by 1) blocking the activation of HASMC membrane surface NADPH and 2) downregulation of RAGE expression and NF-κB activation (Fig. 6). Low-density lipoprotein receptor knockout: LDLR This study was supported, in part, by research grants from Taipei Veterans General Hospital (V101C-178). These funding agencies had no in uence on the study design, data collection or analysis, decision to publish, or preparation of the manuscript.

Abbreviations
Authors' contributions (1) CP and PH contributed to the conception and design of the study and acquisition of data, CY, and JS contributed to analysis and interpretation of data, (2) CP, CY and JS contributed to drafting the article or revising it critically for important intellectual content, (3) CP, PH, JW and SJ contributed to nal approval of the version to be submitted. All authors read and approved the nal version of the manuscript, and ensure it is the case.  Effect of the combination of TNF-α with S100A12 on induced RAGE accumulation and calcium deposition in HASMCs. (a) The accumulation of S100A12 by TNF-α but not recombinant S100A12 in HASMCs was assayed. (b) The cytotoxicity effect of S100A12 was determined by MTT assay. (c) The induction of RAGE by S100A12 in HASMCs was determined by Western blotting assay. The accumulation of RAGE by combination of S100A12 with TNF-α in HASMCs was assayed. S100A12 enhanced TNF-αinduced RAGE protein accumulation in HASMCs. (d) HASMCs were cultured in osteogenic differentiation medium treatment with S100A12 for 4 days in the presence or absence of TNF-α. Calcium deposition was induced dose dependently by TNF-α for 4 days. N = 6 for each set of experiments. *P < 0.05 compared with the control group. Sitagliptin attenuated TNF-α combination with S100A12-induced calcium deposition in HASMCs. (a) The cytotoxicity effect of sitagliptin was determined by MTT assay. (b) Sitagliptin and NAC attenuated the calcium deposition mediated by TNF-α combined with S100A12. The induction of RAGE protein accumulation by TNF-α combined with S100A12 in HASMCs was attenuated by sitagliptin. Western blotting assay showed the effects of the knockdown of TNFR1 and RAGE proteins by these siRNAs.
Compared with the TNF-α combined with S100A12-stimulated cells in the presence of scrambled siRNAs, any combination of TNFR1 or RAGE siRNAs dramatically abolished TNF-α-stimulated calci cation. *P < 0.05 compared with the control group, and #P < 0.05 compared with the TNF-α combined with S100A12 groups. N = 6 for each set of experiments. (c) Sitagliptin attenuated the calcium deposition mediated by TNF-α combined with S100A12. Suppressed RAGE accumulation in HASMCs and siRNAs against RAGEdownregulated calcium deposition (NAC: antioxidant agent, SC: scramble, TNF-R1: TNF-α receptor 1).

Figure 4
Sitagliptin attenuated TNF-α combination with S100A12-induced oxidative stress, NF-κB, and p47 activation in HASMCs. (a) HASMCs were cultured in osteogenic differentiation medium for 1 day in the presence or absence of TNF-α combined with S100A12 concomitantly with sitagliptin (50 µM), APO (apocynin: 500 µM; Nox inhibitor). Intracellular hydrogen peroxide generation was assessed by DCF-AM staining (b) Nox activity was evaluated by lucigenin chemiluminescence. (c) The expression level of NF-κB subunit p65 (nucleus and cytosol fraction) was assessed by Western blotting. hnRNP c1/c2 was used as a nucleus fraction loading control. (d) The expression level of Nox subunit p47 (membrane fraction) was assessed by Western blotting. Caveoli-1 was used as a membrane fraction loading control. *P < 0.05 compared with the control group, and #P < 0.05 compared with the TNF-α combined with S100A12 groups. N = 6 for each set of experiments.

Figure 5
Sitagliptin modulated ROS generation and the expression of RAGE. (a) HASMCs were cultured in osteogenic differentiation medium for 1 day in the presence or absence of TNF-α combined with S100A12 and concomitantly with sitagliptin. Sitagliptin blocked the induction of MSX-2, BMP2, and RUNX2 accumulation induced by TNF-α combined with S100A12. (b) Antioxidant agents (NAC and APO) attenuated the TNF-α combined with S100A12-induced bone marker MSX-2 and (c) RAGE accumulation in HASMCs (sitagliptin; NAC: ROS scavenger; APO: Nox inhibitor). (d) Immunostaining of aortic RAGE showed that sitagliptin could signi cantly decrease the stimulatory effects of HFD on RAGE. Arrowhead indicates positive RAGE position. *P < 0.05 compared with the control group, and #P < 0.05 compared with the TNF-α combined with S100A12 groups. N = 6 for each set of experiments.