Suppression of Gut Bacterial Translocation Ameliorates Vascular Calcification through Inhibiting Toll-Like Receptor 9-Mediated BMP-2 Expression

Aims Vascular calcification (VC) is a primary risk factor for cardiovascular mortality in chronic renal failure (CRF) patients; thus, effective therapeutic targets are urgently needed to be explored. Here, we identified the role of intestinal bacterial translocation in CRF-related VC. Methods and Results Antibiotic supplementation by oral gavage significantly suppressed intestinal bacterial translocation, CRF-related VC, and aortic osteogenic gene and Toll-like receptor (TLR) gene expression in CRF rats. Furthermore, TLR4 and TLR9 activation in vascular smooth muscle cells (VSMCs) aggravated inorganic phosphate- (Pi-) induced calcification. TLR9 inhibition, but not TLR4 inhibition, by both a pharmacological inhibitor and genetic methods could significantly reduce CRF rats' serum or CRF-induced VC. Interestingly, bone morphogenic protein-2 (BMP-2) levels were increased in the aorta and sera from CRF rats. Increased BMP-2 levels were also observed in VSMCs treated with TLR9 agonist, which was blocked by NF-κB inhibition. Both siRNA knockdown of BMP-2 and NF-κB inhibitor significantly blocked TLR9 agonist-induced VSMC calcification. Conclusions Gut bacterial translocation inhibited by oral antibiotic significantly reduces CRF-related VC through inhibition of TLR9/NF-κB/BMP-2 signaling.


Introduction
Vascular calcification (VC) is the major cardiovascular complication in chronic renal failure (CRF) patients. The risk factors for VC in CRF patients include renal function decline, disordered mineral metabolism, and systemic inflammation [1]. Systemic inflammation is a common feature of CRF patients and closely related to morbidity and cardiovascular events [2]. Recently, accumulating evidence has demonstrated that the gastrointestinal tract is a major instigator of systemic inflammation in CRF [3]. Studies illustrated that increased intestinal permeability due to intestinal barrier dysfunction induces intestinal bacterial translocation in both CRF patients and experimental CRF models [4][5][6]. The colon wall inflammation is along with destruction of the intestinal epithelial tight junction barrier, which leads to translocation of bacterial DNA and lipopolysaccharide (LPS) into bloodstream. Gut bacterial DNA and LPS can be detected in the serum of CRF animals and dialysis patients and correlate with severity of systemic inflammation, suggesting that intestinal bacterial translocation is an important cause of systemic inflammatory response in CRF [3,4].
Bacterial translocation has been discovered in multiple diseases such as chronic liver disease, kidney injury, and atherosclerosis, during which LPS-and bacterial DNAinduced immune responses are the leading cause of organ damage [7][8][9][10][11]. TLR4 and TLR9, as the receptors for bacterial LPS and bacterial DNA, respectively, are involved in the morbidity and development of these diseases. Activation of the LPS/TLR4 signal increases the generation of reactive oxygen species and inflammatory cytokines in arterial endothelial cells and VSMCs [12]. Monocyte/macrophages (M/ Ms), activated by LPS, promote osteoblastic differentiation and mineralization of calcifying vascular cells (CVCs) [13], while bacterial DNA/TLR9 signaling promotes atherosclerosis [14] and vascular endothelial injury [9]. These data indicates a relationship between TLR4/TLR9 signaling and inflammatory vascular disease.
However, whether bacterial components contribute to the inflammation and VC in CRF individuals remains unknown. In the present study, oral antibiotic were used to suppress intestinal bacteria and its product LPS and bacterial DNA in CRF rats. We demonstrate that antibiotic administration alleviates intestinal bacterial translocation and suppresses vascular calcification in adenine-induced CRF rats through inhibition of TLR9/NF-κB/BMP-2 signaling.

Methods
2.1. Animal Protocols for the Rat Model. All animals received humane care in compliance with the Institutional Authority for Laboratory Animal Care of Peking University which complies with the Guide for the Care and Use of Laboratory Animals published by the US National Institutes of Health (NIH Publication No. 85-23, revised 1996). The adenineinduced CRF rat model was established as described previously [15][16][17]. 8-week-old male Wistar rats were pair-fed with standard chow containing 1.2% calcium and 0.6% phosphorus for the control group or 0.75% adenine and 1.0% phosphorus for the CRF and CRF + Anti groups for 6 weeks. Polymycin B sulfate (150 mg/kg·day per rat; Aladdin, Shanghai, China) and neomycin sulfate (450 mg/kg·day per rat; Aladdin, Shanghai, China) were given intragastrically for 6 weeks. For euthanasia, sodium pentobarbital (45 mg/kg, intraperitoneal injection) was used to anaesthetize the rats, the blood was drawn from the rat inferior vena cava, and then the rats were decapitated; the aortas were excised, and efforts were made to minimize animal suffering.
Abdominal aortas were excised and fixed in 4% formaldehyde and sectioned for von Kossa staining. For measurement of calcium content, the abdominal aortas were dried and weighed, each sample was extracted with 65% (w/w) HNO 3 for 24 h at 200°C to dissolve the minerals, and the calcium content was measured by atomic absorption spectrometry at 422.7 nm (Jena, novAA 300, Jena, Germany). Results were normalized by dry tissue weight.

Animal
Protocols for the Mouse Model. TLR4 -/and TLR9 -/mice were purchased from Nanjing Biomedical Research Institute of Nanjing University (NBRI), and the genetic background is C57BL/6J. The adenine diet-induced CRF mouse model was applied as we described previously [18]. 20% casein was added to the chow diet and adenine diet to cover the taste and smell of adenine. The chow diet contains 0.6% phosphate, and the adenine diet contains 1.0% phosphate. 8-week-old male c57, TLR4 -/-, or TLR9 -/mice were randomly divided into chow diet (Ctrl) and adenine diet (CRF) groups. Mice were housed under constant temperature (23 + 1°C) with a 12 h light and 12 h dark cycle with free access to water and chow and sacrificed by cervical dislocation. After an 8-week diet program, the calcium content of abdominal aortas was measured.
2.3. Aortic Ring Calcification. Thoracic aortas were removed in a sterile manner from 8-week-old c57, TLR4 -/-, and TLR9 -/mice. After the adventitia and endothelium were carefully removed, the vessels were cut into~1 mm rings and placed in a high-Pi (3 mmol/L) or normal culture medium at 37°C in 5% CO 2 for 7 days. The medium was replaced every 3 days.

Statistical Analysis.
All data are presented as mean ± SEM unless otherwise stated. Data was analyzed by the use of GraphPad Prism software. Statistical analysis involved one-way ANOVA for multiple comparisons, then Tukey-Kramer post hoc testing, and Student's unpaired t-test for comparisons between two groups. P < 0 05 was considered as statistically significant.

Antibiotic Administration Inhibits Intestinal Bacterial
Translocation in CRF Rats. To investigate intestinal bacterial translocation and its harmful product on CRF-induced VC, polymyxin B sulfate and neomycin sulfate were administered orally. The lymph node tissue and spleen tissue homogenate were cultured on a blood agar plate for 36 h. The colony number of bacteria per gram tissue was significantly increased in the mesenteric lymph node of CRF rats and reduced to normal level in antibiotic treatment rats; there was no significant difference in bacteria number in the spleen tissue among the three groups (Figures 1(a) and 1(b)). The mesenteric lymph node/body weight and spleen/body weight ratios were increased in CRF rats compared with control rats, which were reversed by antibiotic treatment (Figures 1(c) and 1(d)). Similarly, antibiotic administration significantly reduced serum LPS and bacterial DNA levels in CRF rats (Figures 1(e) and 1(f)). As expected, serum TNFα increased in CRF rats which was reversed to control level after treatment with antibiotics ( Figure 1(g)). In addition, we found that TLR4 and TLR9, two receptors for LPS and bacterial DNA, respectively, were elevated in the aortic tissue from CRF rats, and they were decreased to normal levels with antibiotic treatment (Figure 1(h)). However, antibiotic administration did not improve adenine-induced renal failure and hyperphosphatemia except a little decrease in serum creatinine (Supplementary Figure 1).

Antibiotics Reduce Vascular Calcification in CRF Rats.
Calcium deposition in the abdominal aorta, as assessed by von Kossa staining and calcium content assay, was increased in CRF rats, and antibiotic administration significantly ameliorated the calcium deposition (Figures 2(a) and 2(b)). Antibiotic administration significantly reduced the mRNA levels of osteogenic genes Msx2 and Cbfα-1 and increased smooth muscle lineage markers Actin2 and SM22α compared with those in vehicle-administered CRF rats' aorta ( Figure 2(c)).
These observations were further confirmed by western blot that Cbfα-1 protein increased while SM22α diminished in CRF rat aorta and antibiotic administration reversed these changes ( Figure 2(d)).

Bacterial Components LPS and DNA Promote Pi-Induced
Calcification and Osteoblastic Differentiation in HASMCs. As TLR4 and TLR9 are expressed in HASMCs responsive to their ligands [11], we aimed to test the effect of TLR4 and TLR9 signaling activation on Pi-induced VSMC calcification. Human aortic smooth muscle cells (HASMCs) were treated with TLR9 ligand control (a ligand binding to TLR9 but cannot activate it) or TLR9 activator E. coli ssDNA (ssDNA) or TLR4 activator LPS at the indicated concentration (ng/mL), with or without Pi (3.0 mmol/L) for 24 h or 7 days. Alizarin red staining revealed that ssDNA and LPS augmented Pi-induced HASMC calcification (Figure 3(a)). TLR9 activation significantly increased calcium content in Pitreated HASMCs (Figure 3 3.4. Antibiotics Ameliorate Vascular Calcification in CRF Rats through Reducing TLR9 Signaling. Pi-treated HASMCs were preincubated with PBS, TLR4 inhibitor, or TLR9 inhibitor for 30 min and then treated with serum from control rats or CRF rats. CRF rat serum augmented calcium deposition in Pi-treated HASMCs compared with control rat serum, which was notably alleviated by TLR9 inhibition but not TLR4 inhibition (Figure 4(a)). However, both TLR4 and TLR9 inhibitors could significantly reduce CRF rat serum-induced inflammatory cytokine secretion in primary macrophages (Supplementary Figure 2). Furthermore, aortic rings from WT, TLR4 -/-, and TLR9 -/mice were treated with Pi plus serum from control rats or CRF rats. CRF rat serum significantly increased Pi-induced calcium deposition in WT and TLR4 -/aorta, but not in TLR9 -/aorta (Figure 4(b)).
Next, WT, TLR4 -/-, and TLR9 -/mice were fed with highcasein diet (Ctrl) or high-casein diet plus adenine (CRF) for 8 weeks to induce a CRF-related VC model as we previously reported [18]. All mice subjected to CRF had reduced body weight, but there was no significant difference among the three genotype mice (Figure 4(c)). Importantly, calcium  content assay and von Kossa staining demonstrated that TLR9 knockout obviously reduced CRF-related VC, while TLR4 knockout had no effect on CRF-related VC (Figures 4(d) and 4(e)).

TLR9 Activation Promotes BMP-2 Expression through
Activation of NF-κB Signaling. BMP-2 is a member of the TGF-β superfamily and involved in physiological ossification [20]. BMP-2 induces osteoblastic differentiation by upregulating Msx2 and Cbfα-1 and promotes Pi uptake by VSMCs. Real-time PCR and ELISA analyses showed that both aortic Bmp2 mRNA and serum BMP-2 protein levels were significantly increased in CRF rats, which were inhibited by antibiotic administration (Figures 5(a) and 5(b)). Pi increased Bmp2 expression in HASMCs, which was further elevated by treatment with ssDNA ( Figure 5(c)). Similarly, the BMP-2 protein level in HASMCs and the supernatant increase after treatment with Pi were further elevated by ssDNA (Figures 5(d) and 5(e)).
And consequently, Pi-and Pi + ssDNA-induced calcium deposition was reduced by PTN treatment (Figure 5(h)).

Discussion
Vascular calcification is the leading cause of cardiovascular events in CRF patients [22]. Systemic inflammation occurs in CRF patients, which is closely related to the morbidity of CRF and cardiovascular events [2]. Here, we identified the role of intestinal bacterial translocation in CRF-related inflammation and VC. Antibiotic supplementation significantly suppressed intestinal bacterial translocation, systemic inflammation, and VC in CRF rats. TLR4 and TLR9 activation in vascular smooth muscle cells (VSMCs) aggravated inorganic phosphate-(Pi-) induced calcification. TLR9 inhibition, but not TLR4 inhibition, reduced CRF-related VC. Furthermore, our data reveals that NF-κB/BMP-2 signaling contributes to TLR9 activation-induced VC.
which was positively correlated with systemic inflammation level in CRF rats, indicating an increased intestinal permeability-induced intestinal bacterial translocation after CRF [4,23]. Specifically, the evidences of elevated intestinal permeability in CRF patients or animal models include the following: (1) endotoxinemia and bacterial DNA in the intestinal wall, mesenteric lymph nodes, or plasma occur without clinical infection [23,24]; (2) the intestinal permeability to macromolecules such as polyethylene glycol significantly increased [25,26]; and (3) gastrointestinal inflammation, including esophagitis, gastritis, duodenitis, and colitis, was found in CRF patients [27]. Thus, increased intestinal permeability is the major cause of systemic inflammation in CRF.
Intestinal bacterial translocation and intestinal flora disorder are distinct concepts. The former is caused by increased intestinal permeability, leading bacteria and its products into the blood, and the latter is the change in the composition or proportion of intestinal flora by pathological factors. In fact, a variety of diseases are associated with intestinal bacterial translocation, including empyrosis, stroke, acute pancreatitis, and cirrhosis of liver. Intestinal bacterial translocation is the main reason for the systemic inflammatory response in these diseases, in which gramnegative bacterial lipopolysaccharide-(LPS-) activated inflammatory response and bacterial DNA caused nonspecific immune response accounting for organ injury such as liver and kidney failure [7][8][9][10]28]. TLR4 and TLR9 are specific receptors for LPS and bacterial DNA, respectively. Thus, we aimed to investigate the effect of inhibiting intestinal bacterial translocation and eliminating bacterial products LPS and DNA by antibiotics on CRF-related VC. Polymycin B sulfate and neomycin sulfate were administered by oral gavage to CRF rats. These two kinds of antibiotics are not absorbed in the intestine, maintaining an intestinal bacterial eliminated state, and significantly reduces the harmful substances produced by intestinal bacteria such as LPS and bacterial DNA [29][30][31][32][33]. 6-week antibiotic administration obviously inhibited intestinal bacterial translocation and CRF-related VC. It also suppressed TLR4 and TLR9 mRNA level in CRF rats' aorta, suggesting that bacterial DNA and LPS might be involved in antibioticreduced VC. The activation of the LPS/TLR4 signal triggers the ?reactive oxygen species and inflammatory cytokines (IL-8, IL-6, MCP-1, etc.) in arterial endothelial cells and VSMCs, activating vascular inflammation [12]. There is a significant decrease in the risk of atherosclerosis in patients with TLR4 mutation [34]. Su et al. [35] reported that oxidized lowdensity lipoprotein promoted BMP-2 expression in arterial endothelial cells in a TLR4-dependent manner. However, the role of TLR4 in VSMC calcification, especially in CRFrelated VC, is not clear. In this study, our findings reveal that activation of TLR4 by high LPS level in HASMCs promoted Pi-induced VC, while the TLR4 inhibitor failed to block CRF rat serum-augmented VC and TLR4 -/also cannot ameliorate CRF-related VC. Though the LPS level in CRF rats was increased compared with the control group, it was only about 1.5-2 ng/mL. We and others [36] identified that in ?cultured HASMCs, LPS promoted Pi-induced calcification when its concentration was up to 100 ng/mL, while Isabel et al. used 1000 ng/mL LPS to stimulate human aortic valve interstitial cells to promote BMP-2 expression and bone formation. Thus, in a CRF rat model, it is likely that the LPS concentration was not enough to stimulate VSMCs directly. As LPS-activated M/Ms promoted osteoblastic differentiation and mineralization of CVCs [13], LPS may promote VC through interaction with M/Ms releasing inflammatory cytokines in CRF rats; however, we could not exclude the possibility that low LPS/TLR4 signaling directly promotes VC under chronic stimulation in vivo.
Bacterial DNA/TLR9 signaling is another pathway of bacterial pathogenicity. Recent studies showed that TLR9 activation promotes atherosclerosis [14], vascular endothelial injury [9], macrophage lipid accumulation [37], and myocardial dysfunction [38], suggesting a close relationship between TLR9 signaling and inflammatory vascular diseases. Our findings provide a framework that bacterial DNA significantly increased in CRF rat plasma, activation of TLR9 in HASMCs promoted Pi-induced VC, and TLR9 inhibition with inhibitors or genetic knockout significantly ameliorated CRF-related inflammation and VC, indicating that bacterial DNA/TLR9 signaling plays a key role in CRF-related VC. However, mitochondrial DNA is also a ligand of TLR9; we could not exclude the possibility that mitochondrial DNA plays a role in CRF-related VC. Regardless of the ligands, TLR9 would be a novel target for prevention of CRF-related cardiovascular events.
Usually, TLR9 dimerized when combined with bacterial DNA (containing unmethylated CpG motif), recruiting MyD88. And then interleukin-1 receptor-associated kinase 1 and 4 (IRAK1 and IRAK4) were attracted to MyD88 and phosphorylated and formed a complex with tumor necrosis factor receptor-associated factor 6 (TRAF6), leading to activation of transforming growth factor-activated kinase 1 (TAK1) and finally activation of IκB kinase and NF-κB signaling [39]. NF-κB signaling plays a key role in inducing procalcifying factor BMP-2 expression, which is a member of the TGF-β superfamily, involved in physiological ossification and repair of bone [20]. BMP-2 promotes VSMC osteoblastic differentiation by increasing Msx2 and Cbfα-1 expression. BMP-2 levels in CRF rat serum and aortic tissue were higher than those in control rats. BMP-2 was induced by TLR9 activation in HASMCs while being blocked by the NF-κB inhibitor. Most importantly, TLR9 activation no longer promoted Pi-induced VC after BMP-2 knockdown, suggesting that TLR9 promoted VC through BMP-2 upregulation.
Uremic toxins play a major role in the pathogenesis of CKD-associated oxidative stress and inflammation [40]. Indoxyl sulfate (IS) and p-cresyl sulfate (PCS), which are protein-bound uremic toxins, increase significantly during kidney injury. The translocation of these toxins from the "leaky gut" into the bloodstream further promotes systemic inflammation, adverse cardiovascular outcomes, and CKD progression [3]. Several studies elucidated that serum uremic toxin levels have a direct relationship with aortic calcification in CKD patients [41,42]. However, we did not detect serum uremic toxin levels in the present study. We do not exclude the possibility that the antibiotics' effect on CKD-related vascular calcification may partly attribute to decreased uremic toxins levels. And we believe that this is a promising and interesting project that we might research.
The potential therapeutic role of prebiotics and probiotics is being actively studied recently. Vaziri et al. [43] reported that feeding uremic rats with amylose maize-resistant starch (a prebiotic) improved creatinine clearance and reduced kidney inflammation and fibrosis. Small trials in hemodialysis patients have demonstrated that oligofructose-inulin or resistant starch administration obviously reduced circulating indoxyl sulfate and p-cresyl sulfate levels and improved gut microbiome [44,45]. It is possible that probiotics have an inhibitive effect on vascular calcification in CKD patients, but further studies are needed to explore the role.

Conclusions
Together, our data demonstrates that antibiotics suppress CRF-related VC through clearance of bacterial pathogen and inhibition of TLR9/NF-κB/BMP-2 signaling. The TLR9 signaling pathway might be a novel target for clinical prevention and treatment of vascular calcification and inflammation in CRF patients.

Data Availability
The data used to support the findings of this study are available from the corresponding author upon request.

Conflicts of Interest
The authors declare that they have no conflicts of interest.