Vascular miR-181b controls tissue factor-dependent thrombogenicity and inflammation in type 2 diabetes

The study protocol was approved by the local ethics committee and was performed in accordance to the ethics principles in the Declaration of Helsinki. Prior to participation in the study, each patient gave a written informed consent. 46 patients with known diabetes mellitus type 2 admitted for poor glycemic control at the Diabetes Center NRW Bad Oeynhausen, Germany, were included in the study. Table 1 shows the patient characteristics.

Tissues and cells from age- and sex-matched 12–16 weeks old miR-181a/b-1−/− mice (B6.Mirc14^tm1.1Ankr as already described [29], termed “miR-181−/−” throughout this article) were kindly provided by Andreas Krueger, Institute for Molecular Medicine in Frankfurt, Germany). These mice carry a constitutive deletion of miR-181a1 and miR-181b1 and were housed under specific pathogen free conditions in the central animal facility at the Institute for Molecular Medicine in Frankfurt, Germany.

The protein plasma concentrations were assessed by using specific ELISA systems (Table 2) according to the manufacturer’s instructions. Fasting blood glucose levels in patients were measured by a standardized laboratory test, while further glucose levels during the day were obtained by a finger-prick test from capillary blood.

The measurement of TF activity was performed as described before [2]. The recombinant FVIIa was kindly provided by Novo Nordisk.

HMEC-1 and THP-1 were transfected with 200 nM of the oligonucleotides listed in Table 2 using the siRNA transfection reagent interferin (VWR) according to manufacturer’s protocol. 24 h post transfection cells were starved in MCBD 131 medium (Gibco) for 2 h and then stimulated with 10 ng/mL TNFα for 2 h for gene expression analysis and 6 h or 24 h for TF or VCAM1 protein expression, respectively. THP-1 cells were stimulated with LPS (10 µg/mL, Sigma) for 2 h for mRNA expression analysis and 6 h for assessment of TF activity. For siRNA-mediated knock down experiments THP-1 were transfected for 48 h.

For real-time PCR, total mRNA was isolated from cultured cells and murine tissue with peqGOLD Trifast (Peqlab) and for patient blood plasma using the mirVana Paris Kit (Life Technologies). Gene expression was determined using GoTaq^® Probe qPCR Master Mix (Promega) with FAM-tagged TaqMan^® gene expression assays (life Technologies) or via SYBR Green assays using SYBR^® Select Master Mix (applied Biosystems) with specific primers (Table 2). Relative gene expression was determined via the comparative C(t) (ΔΔCt) method with GAPDH or 18S ribosomal RNA as endogenous control for mRNA and U6 snRNA for miR.

Western blots were performed as described before [3]. Nuclear and cellular lysates were isolated using the NE-PER™ Nuclear and Cytoplasmic Extraction Reagents (Thermo Fisher Scientific) following the manufacturer’s instructions. Antibodies used are listed in Table 2. The western blots have been performed ≥ 3 times.

For protein staining aortas and spleen were embedded in Tissue-Tek and immunofluorescence was performed with specific antibodies and probes (Table 2).

TF activity of whole blood mouse MVs was assessed as previously described [4]. Briefly, whole blood was drawn by vena cava puncture and 10% citrate added. The cell counts for normalization of the TF activity to the monocyte count was done with a Vetscan automated cell counter. The whole blood was stimulated with 10 µg/mL LPS for 4 h to induce MV release and centrifuged 2 × at 3000g. The plasma was centrifuged again for 40 min, 4 °C, at 18000g to isolate MVs. MV TF activity was determined by adding 2 nM mouse recombinant FVIIa (kindly provided by H. Ostergaard, Novo Nordisk) and 50 nM FX (Haematologic Technologies) to resuspended MVs in HBS (10 mM Hepes, pH 7.4, 137 mM NaCl, 5.3 mM KCl, 1.5 mM CaCl2). FXa was measured by using the chromogenic substrate Spectrozyme FXa (Sekisui Diagnostics).

BMDMs were in vitro–differentiated as described before [5]. Briefly, isolated bone marrow cells were cultivated in DMEM supplemented with 20% L cell medium, 10% FCS, 1 mM l-glutamine, penicillin, and streptomycin. On day 7 macrophages were seeded at a density of 1 × 10⁶ per well. On the next day, macrophages were stimulated with 1 µg/mL LPS (Enzo Life Sciences) for 4 h. The cells were then lysed for RNA analysis, western blot or measurement of FXa generation as described above.

The statistical analyses were performed using the software GraphPad Prism 7. Correlation of patient characteristics were performed via Pearson correlation for normally distributed values or Spearman correlation in case of not normal distribution. For binary variables, a point-biserial correlation was performed. Differences between 2 groups were examined using a Mann–Whitney test or student’s t test (2-tailed). For comparisons of 1 parameter between more than two related groups the one-way ANOVA test with Dunn’s multiple comparison as post hoc test was used. For data sets with 4 groups and 2 variables a 2-way ANOVA test with Sidak’s multiple comparison post hoc test was performed. Data are represented as mean ± SEM. p-values < 0.05 are considered statistically significant.

In the healthy vasculature, circulating miRs epigenetically control pro-inflammatory transcription factors and their gene products to preserve vascular homeostasis [3, 30‐32]. However, in a diabetic environment this protective miR transcriptome is disrupted [33]. In particular, diabetes-related loss of TF-specific miRs, such as miR-126 or miR-223 contributes to vascular inflammation and a pro-thrombotic state [33, 34].

Here we show that miR-181b, a key regulator for endothelial function, beta cell function, peripheral insulin sensitivity, and NFκB signaling [22, 28, 35], participates in the control of vascular TF activity in diabetes. In poorly controlled diabetics, miR-181b was associated with reduced activation of the TF pathway, lower leukocyte count, and fibrinogen, which both constitute independent risk factors for cardiovascular mortality among patients with diabetes [36].

In contrast to markers of inflammation, miR-181b levels did not correlate with blood glucose or HbA1c in our cohort. Although we and Sun et al. [28] found that miR-181b expression in HMEC-1 and human umbilical vein endothelial cells (HUVEC) is reduced by high glucose concentrations, Yang et al. [37] reported activation of the miR-181a/b-specific transcription factor signal transducer and activator of transcription (STAT) 3 in HUVEC treated with sera from diabetics pointing to a more complex regulation of miR-181b expression in humans. This may explain why the miR-181b expression in our healthy controls did not differ from those in poorly controlled diabetes (Additional file 6: Figure S5A). In contrast, miR-126 that has been shown to be reduced in diabetes in the population-based Bruneck cohort, was significantly lower in our diabetes cohort compared to healthy controls confirming the quality of our measurements (Additional file 6: Figure S5B). We assume that diabetic conditions beyond glycemia impact miR-181b expression resulting in differential expression levels among diabetics. Although we cannot rule out an impact of co morbidities and anti-diabetic drugs on miR-181b expression in our cohort, it seems likely that chronic inflammation in an advanced state of the disease is a more relevant factor than blood glucose for miR-181b suppression. In accordance, we found co expression of miR-181b with miR-126 and miR-19a that are both reduced by inflammatory cytokines [20, 21].

Our data indicates that miR-181b restrains NFκB-depending TF expression in human and murine ECs via targeting of KPNA4. The direct contribution of vessel wall-derived TF to thrombosis has been a matter of debate [4]. Yet, through allosteric activation of FVIIa and/or FXa generation, TF promotes pro-inflammatory PAR signaling leading to sustained vascular dysfunction. Consequently, an EC-specific deletion of TF reduced circulating inflammatory cytokines [38]. Recently, clinical trials provided strong evidence that the TF/VIIa/Xa pathway contributes to vascular complications and pharmacological blockage of FXa at low doses prevented ischemic CVD endpoints [39]. However, inhibition of vascular TF and PARs are challenged by bleeding issues due to their role in hemostasis and platelet function, respectively [40]. Our study identifies miR-181b as a potent regulator for pathological TF-induced vascular inflammation. Further clinical research needs to assess the therapeutic potential of miR-181b in the setting of diabetes and CVD.

In the vasculature, hematopoietic cells represent the main pool of circulating TF and contribute to thrombosis [13, 26]. Here, we show that miR-181b reduces TF activity in human monocytes through control of PTEN.

The strong negative correlation of circulating miR-181b with plasma TF activity in our patient cohort suggests that the mostly endothelial-derived miR-181b impacts monocytic TF production in humans. Recent studies reported that ECs can modulate the phenotype of monocytes via release of endothelial MVs. Njock et al. [41] showed that miR-181b along with other anti-inflammatory miRs is enriched in EC-derived MVs. Treatment of recipient THP-1 with EC-MVs or with miR-181b mimics repressed LPS-induced inflammatory gene expression in monocytes. These data along with our study highlight that endothelial miR-181b exerts anti-thrombotic properties beyond the vasculature by also modulating the thrombotic phenotype of monocytic cells and support the hypothesis that the endothelium at least indirectly participates in the regulation of plasma TF activity.

In contrast to ECs, miR-181b did not alter NFκB nuclear translocation in human monocytes. Different host cell factors that preclude down regulation of KPNA4 may explain this observation. However, we are the first to show that miR-181b controls upstream NFκB activation in monocytes via targeting PTEN. The PI3K/AKT pathway was previously shown to negatively regulate LPS-induced TF transcription in human monocytes through downstream control of NFκB along with the TF-specific transcription factors activator protein (AP)-1 and early growth response (EGR)-1 [42]. Consequently, targeting of the PI3K/AKT negative regulator PTEN by miR-181b reduces LPS-induced TF expression and possibly other inflammatory factors.

In the murine system, miR-181b deficiency failed to alter monocyte-derived MV-TF activity but increased TF expression in spleen and BMDM.

In line with previous reports showing that miR-181b does not alter NFκB reporter activity of mouse peripheral blood mononuclear cell (PBMC)s [32], we did not observe changes in MV-associated monocytic TF activity or TAT levels. As suggested by Sun et al., differential expression of importin isoforms in mouse monocytes may be the underlying cause. However, our study shows that miR-181b reduces hematopoietic TF production in murine BMDM. Macrophages not only produce TF-positive thromboinflammatory MVs [14] but also account for TF accumulation in atherosclerotic plaques [43]. Reduced TF expression in macrophages may contribute to the observation that miR-181b overexpression in Apo E−/− mice protects from atherosclerosis [32].

miR expression in patient plasma samples was normalized to U6 which can be altered in some diseases. However, this has not been reported in patients with diabetes and our expression profiles in both cohorts are in accordance with previously published data [34]. Moreover, the miR-181−/− mouse model analyzed here carries a deletion of miR-181b-1 and the adjacent miR-181a-1 as well. Due to transcriptional co-regulation and high sequence similarities both miRs share largely the same functions. Accordingly, some of our findings might also be explained by loss of miR-181a.