Shear-sensitive microRNA-34a modulates flow-dependent regulation of endothelial inflammation

Vascular endothelial cells, which constitute the inner layer of the vascular wall, are continuously exposed to hemodynamic shear stress, which plays a major role in maintaining endothelial cell homeostasis and influences the development of atherosclerosis (Davies, 2009). Different flow patterns in the vasculature elicit unique biological responses in endothelial cells (Nigro et al., 2011). Endothelial cells that experience low and oscillating shear stress (OSS) are proinflammatory, largely because of the decreased expression of anti-inflammatory molecules and the increased expression of proinflammatory molecules. By contrast, endothelial cells that are exposed to physiological high shear stress (HSS) are anti-inflammatory and express atheroprotective molecules.

microRNAs (miRNAs) are a class of highly conserved, small noncoding RNAs that regulate gene expression at the post-transcriptional level (Bartel, 2004). Although many studies have described the roles of miRNAs in the modulation of endothelial response to shear stress (Fang and Davies, 2012; Fang et al., 2010; Hergenreider et al., 2012; Loyer et al., 2014; Schober et al., 2014; Wu et al., 2011), the mechanisms remain incompletely understood. miR-34a, which is a highly expressed miRNA in endothelial cells, reportedly plays a crucial role in the modulation of endothelial cell senescence (Ito et al., 2010). Interestingly, miR-34a is reportedly upregulated in human atherosclerotic plaques (Raitoharju et al., 2011) and endothelial progenitor cells (EPCs) that have been obtained from individuals with coronary artery disease (Tabuchi et al., 2012), which indicates that this elevated miRNA might contribute to endothelial dysfunction. Because local flow patterns, which control the distribution of atherosclerotic lesions, are the major physiological and pathophysiological stimuli that induce or suppress gene expression in endothelial cells, including miRNAs, we hypothesize that shear stress influences miR-34a expression; thus, miR-34a might be involved in the shear-stress regulation of endothelial function.

In the present study, we report that mechanosensitive miR-34a modulates the flow-dependent regulation of endothelial inflammation, partially through sirtuin 1 (SIRT1) and its downstream nuclear factor κB (NF-κB) signaling pathway.

After 24-h exposure to HSS (15 dyn/cm²), endothelial cells were elongated and aligned in the direction of the flow (Fig. 1A). Conversely, OSS (±5 dyn/cm²; 1 Hz) did not induce any detectable elongation or alignment of the cells (Fig. 1A). These results were consistent with previous findings in bovine aortic endothelial cells (Davies et al., 1986). As shown in Fig. 1B, the regulation of inflammation-associated genes by HSS or OSS appeared to be consistent with previous findings (Chen et al., 2010; Chiu et al., 2004; Nagel et al., 1994; Wu et al., 2011), increasing our confidence in the current data. Furthermore, the mRNA expression levels of several other atheroprotective or proatherogenic genes were determined (see supplementary material Fig. S1A). These findings further support previous claims regarding the atheroprotective effects of HSS and the atheroprone effects of OSS (Brooks et al., 2002; Chen et al., 2001; Ohura et al., 2003; Warabi et al., 2007).

Because miRNAs have been demonstrated previously to be involved in the flow-mediated regulation of gene expression in endothelial cells (Fang and Davies, 2012; Fang et al., 2010; Hergenreider et al., 2012; Schober et al., 2014; Wu et al., 2011), we measured the expression levels of several miRNAs that have an established role in vascular biology by using quantitative real-time polymerase chain reaction (qRT-PCR) analyses after 24-h of shear stress treatment. The endothelial-enriched miRNA miR-126 was only slightly, but not significantly, regulated by HSS and OSS (Fig. 1C), which is consistent with previous reports (Ni et al., 2011; Schober et al., 2014). The expression of the shear-responsive miRNA miR-126* (a minor miRNA sequence for miR-126), which has been reported to promote endothelial proliferation and limit atherosclerosis (Schober et al., 2014), was also significantly upregulated by HSS, whereas it was decreased by OSS (Fig. 1C). Consistent with a previous report (Hergenreider et al., 2012), the expression of both miR-143 and miR-145 was significantly induced by HSS, but the expression remained unchanged following OSS treatment (Fig. 1C). Furthermore, we determined that miR-21, which has been reported to be induced by OSS and to modulate flow-induced endothelial inflammation (Zhou et al., 2011), was significantly upregulated by OSS and, at a lower level, by HSS (Fig. 1C). Moreover, the expression level of miR-221, which has been demonstrated previously to indirectly decrease eNOS (also known as NOS3) expression (Suárez et al., 2007), was downregulated by HSS and OSS; the expression of miR-34a, which regulates endothelial cell senescence (Ito et al., 2010), was also significantly downregulated by HSS, whereas it was upregulated by OSS (Fig. 1C).

Among the shear-responsive miRNAs investigated, miR-34a is reportedly upregulated in human atherosclerotic plaques, but its role in the modulation of the endothelial response to HSS and OSS remains unclear; therefore, we further investigated the regulation and function of miR-34a in detail. Human umbilical vein endothelial cells (HUVECs) were exposed to HSS for various time periods. As determined by using qRT-PCR, miR-34a was significantly decreased as early as 6 h after HSS treatment, and this decrease was sustained for 24 h (Fig. 1D; supplementary material Fig. S1B,C). Conversely, OSS induced a significant increase in miR-34a expression at all time points assessed (Fig. 1D). We next measured the accumulation of the primary miR-34 gene transcript (pri-miR-34) in endothelial cells that had been exposed to HSS or OSS. As shown in Fig. 1E, pri-miR-34 expression was significantly decreased as early as 6 h after HSS treatment, and this decrease was sustained for 24 h. By contrast, OSS induced a significant increase in pri-miR-34a expression after 24 h of treatment (Fig. 1E). These results indicate that miR-34a is upregulated by atheroprone OSS but downregulated by atheroprotective HSS. However, the role of miR-34a in the mediation of the endothelial response to shear stress remains unknown.

Because inflammation plays a key role in atherosclerosis and the initial stage of atherosclerosis is characterized by the recruitment of leukocytes to activated endothelial cells, we examined the potential role of miR-34a in the regulation of endothelial cell inflammation. Significantly increased expression of miR-34a in cells that had been transfected with a miR-34a mimic was confirmed by using qRT-PCR analysis (see supplementary material Fig. S2A). As shown in Fig. 2A, transfection of miR-34a increased the protein level of the inflammatory biomarkers VCAM-1 and ICAM-1. By contrast, miR-34a transfection did not alter the protein level of IκBα (also known as NFKBIA), indicating that the upregulation of inflammatory biomarkers is IκBα independent. Because the pronounced expression of cell adhesion molecules on endothelial cells results in leukocyte recruitment and vascular inflammation (Ross, 1993), we performed an in vitro adhesion assay to determine whether miR-34a plays a role in this process. As expected, compared with control RNA-transfected cells, miR-34a transfectants exhibited a >4-fold increase in adherent monocytes (Fig. 2B). To confirm the results from the gain-of-function study, we performed a loss-of-function analysis using a miR-34a inhibitor (anti-34a). A significantly decreased expression of miR-34a in cells that had been transfected with anti-34a was confirmed through qRT-PCR analyses (see supplementary material Fig. S2B). Transfection of anti-34a dramatically decreased basal VCAM-1 and ICAM-1 expression levels, whereas IκBα protein levels remained unchanged (Fig. 2C). These findings, together with the results from the gain-of-function studies, indicate a role for miR-34a in the regulation of endothelial cell inflammation.

Atheroprotective flow exerts an anti-inflammatory effect in endothelial cells, largely because of the downregulation of proinflammatory molecules (Nigro et al., 2011). In this study, we found that HSS downregulated miR-34a expression, and the inhibition of endogenous miR-34a expression also suppressed the expression of proinflammatory molecules. Thus, we examined whether miR-34a downregulation is required for the anti-inflammatory effects of HSS. As shown in Fig. 3A, miR-34a overexpression significantly attenuated HSS-mediated suppression of VCAM-1 protein expression on endothelial cells. Moreover, the shear-induced ICAM-1 expression was increased through miR-34a (Fig. 3A). Although the introduction of miR-34a was not able to significantly abrogate the HSS-mediated suppression of MCP-1 (also known as CCL2) mRNA expression in endothelial cells (see supplementary material Fig. S3), it did result in slightly increased MCP-1 expression. These results suggest that the HSS-induced miR-34a downregulation, at least partially, contributed to HSS-mediated suppression of endothelial cell inflammation.

Because miR-34a was upregulated by OSS, we also determined whether this miRNA plays a role in the mediation of OSS-induced endothelial inflammation. As shown in Fig. 3B, the OSS induction of endothelial cell VCAM-1, ICAM-1 and MCP-1 was suppressed by transfection of anti-34a. The application of OSS to endothelial cells for 12 h induced a significant increase in their adhesiveness to THP-1 cells, which was also attenuated by anti-34a, thereby confirming the proinflammatory role of miR-34a in OSS-induced endothelial inflammation (Fig. 3C).

To elucidate the mechanism by which miR-34a promotes endothelial cell inflammation, we searched for negative regulators of endothelial cell inflammation among the predicted miR-34 targets. KLF4 represents an attractive candidate because it is also dramatically induced by shear stress and exhibits an anti-inflammatory function (Hamik et al., 2007). However, luciferase assays indicated that KLF4 was not a direct target of miR-34a (data not shown), suggesting that we can exclude the involvement of this mediator.

Among the negative regulators of endothelial cell inflammation, SIRT1 has been reported to be a direct target of miR-34 (Ito et al., 2010). Because HSS downregulated miR-34a expression and upregulated SIRT1 expression, we examined whether miR-34a was involved in HSS-regulated SIRT1 expression. Transfection of anti-34a dramatically upregulated SIRT1 protein levels, whereas the mRNA levels remained unaffected (Fig. 4A,B). To examine whether miR-34a overexpression could attenuate HSS-induced SIRT1 expression, HUVECs were transfected with a miR-34a mimic and exposed to HSS for 24 h. As shown in Fig. 4C,D, miR-34a transfection attenuated HSS-mediated SIRT1 induction at both the mRNA and the protein levels. Because the elevation of SIRT1 is tightly correlated with increased mitochondrial biogenesis (Csiszar et al., 2009) and HSS-increased mitochondrial biogenesis is reportedly mediated by SIRT1 (Chen et al., 2010), we examined whether miR-34a is involved in HSS-mediated increases in mitochondrial mass. As shown in Fig. 4E, HSS increased mitochondrial mass, which is consistent with results from a previous study (Chen et al., 2010). Importantly, miR-34a transfection attenuated HSS-induced mitochondrial biogenesis, which is consistent with SIRT1 regulation through miR-34a (Fig. 4E). These results suggest that the induction of SIRT1 by HSS is mediated by miR-34a.

To explore the role of SIRT1 in miR-34a-mediated endothelial cell inflammation, we transfected endothelial cells with a pCDNA3.1-SIRT1 expression vector that encodes SIRT1, which resulted in increased SIRT1 protein expression that was similar to that observed in HSS-exposed endothelial cells (Fig. 5). We found that SIRT1 overexpression could partially prevent miR-34a-induced VCAM-1 and ICAM-1 expression (Fig. 5). Interestingly, in comparison with control RNA, SIRT1 overexpression did not completely prevent the induction of VCAM-1 and ICAM-1 expression through miR-34a, suggesting that miR-34a might promote endothelial cell inflammation by synergistically affecting multiple targets.

SIRT1 can negatively affect NF-κB activity by directly interacting with the RelA/p65 subunit of NF-κB and regulating its deacetylation (Stein et al., 2010). Hence, we examined whether miR-34a can increase RelA/p65 acetylation. Following transfection of miR-34a, the subsequent western blot analysis demonstrated increased RelA/p65 acetylation, whereas the total RelA/p65 protein levels remained unchanged (Fig. 6A), which suggests that NF-κB signaling might be involved in miR-34a-mediated endothelial cell inflammation. This possibility was further examined using a small interfering (si)RNA for RelA/p65, which resulted in a significant reduction in RelA/p65 protein levels (Fig. 6B). As expected, in comparison with transfection of control siRNA, RelA/p65 knockdown dramatically reduced miR-34a-induced VCAM-1 and ICAM-1 expression (Fig. 6B). Because OSS significantly upregulated the expression of miR-34a, we questioned whether OSS could increase the acetylation level of p65. As shown in Fig. 6C, the p65 acetylation levels increased as early as 6 h after OSS treatment, and this increase was sustained for 12 h. The protein expression levels of the proinflammatory biomarkers VCAM-1 and ICAM-1 were also significantly upregulated by OSS (Fig. 6C). In addition, OSS-induced endothelial cell p65 acetylation was suppressed by treatment with the miR-34a inhibitor (see supplementary material Fig. S4). Collectively, these results suggest that NF-κB signaling is involved in miR-34a-mediated endothelial cell inflammation.

Here, we identify miR-34a as a mechanosensitive miRNA that is downregulated by physiological HSS and that is upregulated by atheroprone OSS in endothelial cells. We demonstrate that HSS-induced miR-34a downregulation contributes to HSS-mediated suppression of endothelial cell inflammation. We also demonstrate that miR-34a plays a role in the mediation of the OSS-induced proinflammatory response in endothelial cells. Mechanistically, miR-34a downregulates anti-inflammatory SIRT1 expression, which subsequently activates the downstream NF-κB signaling and stimulates endothelial inflammation.

Endothelial activation is the primary event in the development of atherosclerosis (Weber and Noels, 2011). Activation of the NF-κB pathway promotes the nuclear translocation of NF-κB, which results in the increased expression of adhesion molecules, such as VCAM-1 and ICAM-1, and the promotion of monocyte recruitment to the endothelium (Collins et al., 1995). The inflammatory status of the endothelium is governed by local flow patterns. We have demonstrated that enhanced external counterpulsation (EECP) inhibits intimal hyperplasia and the development of atherosclerosis in pigs on a high-cholesterol diet by increasing the arterial wall shear stress (Zhang et al., 2007). EECP exerts a retarding effect on atherosclerosis through the downregulation of proinflammatory gene expression (Zhang et al., 2010). However, the mechanisms that underlie the anti-inflammatory effects of HSS remain incompletely understood. Recently, the involvement of miRNAs in physiological shear-stress-induced suppression of endothelial cell inflammation has been demonstrated (Loyer et al., 2014). The present study provides additional insights into the mechanisms that underlie endothelial protective effects by demonstrating that miR-34a is involved in the anti-inflammatory actions of HSS. In this work, we described the proinflammatory effect of miR-34a. This result provides a potential explanation for the previous observation that miR-34a expression is elevated in human atherosclerotic lesions (Raitoharju et al., 2011); however, the potential role of miR-34a in the regulation of atherosclerosis remains to be established.

In normal human coronary arteries, the time-averaged wall shear stress was previously determined to be ∼16 dyn/cm² (Joshi et al., 2004; Stone et al., 2003; Wentzel et al., 2003). By contrast, the shear stress overlying plaques was >50 dyn/cm² (Gijsen et al., 2008; Stone et al., 2003). These findings raise the question – could this extreme increase in shear stress inhibit the development of atherosclerotic plaques? Our data showed that endothelial cells that experience extremely high shear stress might promote inflammation because of the decreased expression of atheroprotective genes and the increased expression of proinflammatory genes (see supplementary material Fig. S1A,D). Subsequent investigation showed that 30-dyn/cm² shear stress did not affect the expression of miR-143 or miR-34a (see supplementary material Fig. S1E). These findings suggested that an increase in shear stress restrains the atheroprotective effects of physiological HSS; however, the underlying mechanisms remain to be explored.

Our data demonstrated that both HSS and OSS induced the expression of ICAM-1, which might be explained by the finding that a shear-stress-response element (SSRE) exists within the promoter of the ICAM-1 gene (Degitz et al., 1991). However, the expression level of ICAM-1 that was induced by OSS was substantially higher compared with the expression of that induced by HSS. A reasonable explanation might be the activation of the NF-κB signal pathway by OSS (Mohan et al., 1997). However, the significance of the ICAM-1 that is induced by physiological HSS remains unexplored.

SIRT1 is an important modulator of cardiovascular functions, such as senescence (Ota et al., 2007), inflammation (Stein et al., 2010) and mitochondrial biogenesis (Csiszar et al., 2009). A previous study has demonstrated that HSS causes a marked increase in SIRT1 expression (Chen et al., 2010). However, the mechanisms that underlie the regulation of SIRT1 by shear stress are not fully understood. Our results demonstrated that HSS-mediated SIRT1 induction is facilitated by miR-34a downregulation, which provides one potential explanation for the finding that the SIRT1 protein is more highly expressed in mouse thoracic aorta that has been exposed to atheroprotective HSS compared with the aortic arch, which is under atheroprone flow (Chen et al., 2010).

The underlying mechanisms responsible for the regulation of miR-34a expression by different flow patterns remain unknown. Our data demonstrated that pri-miR-34 was downregulated by HSS but upregulated by OSS, suggesting that shear stress regulates the miR-34 gene at the transcriptional level. Previous studies have indicated that p53 directly activates miR-34a expression (Bommer et al., 2007). p53 mRNA levels in human aortic endothelial cells decrease after 24 h of exposure to steady laminar flow (Brooks et al., 2002). By contrast, the expression of the p53 protein was elevated in endothelial cells at sites of disturbed flow (Warboys et al., 2014). SIRT1 inhibits p53 activity by decreasing the acetylation of p53 at Lys382 (Langley et al., 2002). Thus, one potential mechanism involved in the flow-dependent regulation of miR-34a expression is the modulation of p53 expression and activity by HSS and OSS. In this study, we also examined the mRNA expression of Drosha and Dicer, which are involved in post-transcriptional miRNA regulation (Kim et al., 2009). Interestingly, we determined that the expression of both genes was significantly decreased as early as 6 h after HSS treatment, which was sustained for 12 h (with the exception of Dicer) (see supplementary material Fig. S1F,G). By contrast, we observed a significant increase in the expression of Dicer mRNA after 6 h of OSS treatment (see supplementary material Fig. S1G). These findings indicate that other potential mechanisms of shear-induced miR-34a downregulation occur by decreasing the levels of processing by Drosha and Dicer.

Our results on the HSS-induced inhibition of miR-34a expression are consistent with the findings of a recent study by Mai et al., who reported that miR-34a expression is downregulated by laminar shear stress at 12 dyn/cm² (Mai et al., 2013). Other groups have also reported an increase in miR-34a expression in response to unidirectional shear stress (Cheng et al., 2014; Weber et al., 2010). Although the reason underlying the difference between our results and that of others remains uncertain, it might be, at least, explained in part by the difference in cells types, as well as the cell culture and shearing systems used.

In conclusion, we report that miR-34a plays an important role in the modulation of the flow-dependent inflammatory response. Our findings provide new insights into the links between hemodynamic forces, miRNAs and endothelial inflammation.

HUVECs were purchased from ScienCell (Carlsbad, CA) and cultured in complete endothelial cell medium (ScienCell, Carlsbad, CA) supplemented with 5% fetal bovine serum (FBS) and 1% endothelial cell growth supplement (ECGS), and were used between passages four and six. THP-1 cells were purchased from Shanghai Institutes for Biological Sciences (Shanghai, China) and maintained in RPMI 1640 (Gibco, Invitrogen, Carlsbad, CA) supplemented with 10% FBS.

The coding sequence of SIRT1 (NM-012238.4) was cloned into pCDNA3.1 (Invitrogen, Carlsbad, CA) to generate the pCDNA3.1-SIRT1 expression vector. Briefly, the entire SIRT1 coding sequence was amplified from HUVEC cDNA by using PCR with the PrimeSTAR enzyme (TaKaRa, Dalian, China). Primers were designed to contain an overhang complementary to the BamHI and XhoI restriction sites in the pCDNA3.1+ vector sequence. These overhangs were subsequently used to insert the PCR product into the pCDNA3.1+ vector by using FastDigest restriction enzymes and T4 DNA Ligase (Fermentas, MD). The successful construction of the recombinant plasmid was confirmed by sequencing the inserted DNA segment.

Primer sequences are provided in supplementary material Table S1.

HUVECs at 40–60% confluence were transfected with miRNAs or siRNA using Lipofectamine^TM RNAiMAX reagent (Invitrogen, Carlsbad, CA) or Lipofectamine 2000 reagent (Invitrogen, Carlsbad, CA) according to the manufacturer's instructions. Endogenous miR-34a expression was inhibited by transfecting them with 50 nmol/l miR-34a inhibitor (micrOFF^TM hsa-miR-34a-5p Inhibitor, RiboBio, Guangzhou, China) or miRNA inhibitor negative control (micrOFF^TM Inhibitor Negative Control no. 24, RiboBio, Guangzhou, China). To overexpress miR-34a in HUVECs, cells were transfected with 10 nmol/l hsa-miR-34a-5p mimic (micrON^TM hsa-miR-34a-5p mimic, RiboBio, Guangzhou, China) or negative control mimic (micrON^TM Mimic Negative Control no. 24, RiboBio, Guangzhou, China). Endogenous p65 expression was knocked down by transfecting cells with 10 nmol/l of siRNA against p65 (RiboBio, Guangzhou, China) or negative control siRNA (siR-Ribo^TM Negative Control, Guangzhou, China).

To overexpress SIRT1, HUVECs at 80–90% confluence were transfected with the pCDNA3.1-GFP vector (which was kindly provided by Shi-Rong Pan, Division of Cardiology, First Affiliated Hospital of Sun Yat-sen University, China), or pCDNA3.1-SIRT1 vector (which encodes the entire coding sequence of SIRT1 but lacks the 3′-UTR) using Lipofectamine 2000 reagent (Invitrogen, Carlsbad, CA), according to the manufacturer's instructions with minor modifications. Briefly, for each transfected well (12-well format), DNA-Lipofectamine® 2000 complexes were prepared as follows, 1 µg of DNA was diluted into 400 µl of M199 medium without serum and gently mixed, followed by the addition of 1.6 µl of Lipofectamine® 2000 reagent directly to the diluted DNA. The mixture was gently mixed and incubated for 20 min at room temperature. The growth medium was removed from the cells, and 400 µl of the DNA–Lipofectamine® 2000 complexes was directly added to each well containing cells, which were gently mixed by rocking the plate back and forth. The cells were incubated at 37°C in a CO₂ incubator for 1 h. The transfection complexes were removed from the cells and replaced with 1 ml of complete growth medium. The cells were then incubated for an additional 48 h post-transfection before the assessment of transgene expression.

To expose a large amount of cells to flow for western blot analysis, we used a parallel-plate flow chamber (GlycoTech, Gaithersburg, MD) as previously described (Reinhart-King et al., 2008). Briefly, a glass slide that had been seeded with a confluent monolayer of HUVECs was mounted onto the bottom of the chamber (1 cm in width, 5.5 cm in length and 0.01 cm in height). A laminar HSS of 15 dyn/cm² was imposed on endothelial cells by perfusing culture medium through the channel between the endothelial-cell-containing glass slide and an acrylic plate in the flow chamber. For RNA extraction, and miRNA and cDNA qRT-PCR analyses, confluent HUVEC monolayers were exposed to HSS (15 dyn/cm²) or OSS (±5 dyn/cm²; 1 Hz) for the indicated times using 0.4-mm Luer µ-slides I (Ibidi, Martinsried, Germany; channel length, 50 mm; channel width, 5 mm; and channel height, 0.4 mm) as previously described (Hergenreider et al., 2012). The flow systems were maintained at 37°C in a humidified 5% CO₂ tissue culture incubator.

Selected miRNA expression was quantified using TaqMan microRNA assay kits (Life Technologies, Carlsbad, CA) as previously described (Fang and Davies, 2012). RNU6B, U6 or RNU44 were used as references for normalization. cDNA was quantified using LightCycler® 480 SYBR Green I master mix (Roche, Mannheim, Germany) and normalized to human glyceraldehyde-3-phosphate dehydrogenase (GAPDH) expression (Fang and Davies, 2012). The PCR primers for the genes of interest are listed in supplementary material Table S1.

Total protein was extracted from endothelial cells using radioimmunoprecipitation (RIPA) buffer (Cell Signaling Technology [CST], Danvers, MA) according to the manufacturer's protocol. Western blot analysis was performed as previously described with minor modifications (Magenta et al., 2011). Proteins were loaded into gels, subjected to SDS-PAGE electrophoresis and transferred to PVDF membranes. The membranes were incubated with primary antibodies against VCAM-1 (CST, 1∶1000), ICAM-1 (CST, 1∶1000), IκBα (CST, 1∶2000), SIRT1 (CST, 1∶2000), acetylated p65 (Lys310) (CST, 1∶1000), p65 (CST, 1∶2000), α-tubulin (Proteintech Group, 1∶4000) or β-actin (CST, 1∶4000) with gentle agitation overnight at 4°C. The membranes were washed three times for 5 min each with TBST (containing 0.1% Tween-20) and incubated with horseradish-peroxidase (HRP)-conjugated anti-mouse-IgG (Proteintech Group, 1∶20,000) or HRP-conjugated anti-rabbit-IgG (Proteintech Group, 1∶20,000) followed by chemiluminescence detection using Pierce^TM ECL western blotting substrate according to the manufacturer's protocol. The membranes were subsequently analyzed using ‘Quantity One’ software (Bio-Rad, Hercules, CA).

MitoTracker Green FM (Invitrogen, Carlsbad, CA) was used to stain mitochondria in HUVECs, as previously described with minor modifications (Chen et al., 2010). Cells that had been grown on coverslips or on 0.4-mm Luer µ-slides I were stained with 50 nmol/l MitoTracker for 30 min at 37°C in a humidified 5% CO₂ tissue culture incubator. After rinsing three times with M199 basic medium, cells were subjected to fluorescence detection using a fluorescence microscope (Zeiss, Oberkochen, Germany).

A leukocyte adhesion assay was used to determine endothelial-cell–monocyte interactions using the Endothelial Cell Adhesion assay kit (Millipore, Billerica, MA), according to the manufacturer's instructions with minor modifications. Briefly, THP-1 cells were labeled with Calcein AM and were added to each well containing an endothelial monolayer. After a 1-h incubation, non-adherent cells were carefully removed, and the cells in each well were gently washed three times using basic endothelial cell medium. Adherent THP-1 cells were visualized by using an inverted fluorescence microscope (Zeiss, Oberkochen, Germany). Two objective fields were randomly selected to count adherent THP-1 cells in each well, and four independent experiments were performed.

The data are presented as the means±s.e.m. from at least three independent experiments. Comparisons between two groups were performed using Student's t-test. One-way ANOVA followed by Bonferroni's post-hoc test or Dunnett's test was performed for multiple comparisons. All statistical calculations were performed using GraphPad Prism 5. A level of P<0.05 was considered significant.

We thank Xuanhong Zhang, Yinfen Wang and Daya Yang for their excellent technical assistance.