Regulation of TFPIα expression by miR-27a/b-3p in human endothelial cells under normal conditions and in response to androgens

The increased risk of cardiovascular events in older men is multifactorial, but the significant reduction of testosterone levels has been involved. As this hormone regulates the expression of TFPI by unknown mechanisms, we aimed to evaluate the role of miRNAs in the regulation of TFPIα expression under normal conditions and in response to androgens. In silico studies allowed the selection of 4 miRNAs as potential TFPIα regulators. Only miR-27a/b-3p significantly reduced TFPIα expression in two endothelial cell lines. Luciferase assays demonstrated a direct interaction between miR-27a/b-3p and TFPI 3′UTR. Ex vivo analysis of TFPI and miRNA levels in 74 HUVEC samples from healthy subjects, showed a significant and inverse correlation between TFPI and miR-27a-3p. Moreover, anticoagulant activity of TFPIα from cells supernatants decreased ~30% with miR-27a/b-3p and increased ~50% with anti-miR-27a/b-3p. Interestingly, treatment of EA.hy926 with a physiological dose of dihydrotestosterone (30 nM) significantly increased (~40%) TFPIα expression with a parallel decreased (~50%) of miR-27a/b-3p expression. In concordance, increased levels of miR-27a/b-3p normalized the up-regulation induced by testosterone. Our results suggest that testosterone is a hinge in miR-27/TFPIα regulation axis. Future studies are needed to investigate whether testosterone variations are involved in a miR-27/TFPIα dysregulation that could increase the cardiovascular risk.

Scientific RepoRts | 7:43500 | DOI: 10.1038/srep43500 Therefore, the purpose of this study was to determine whether TFPIα is regulated by miRNAs through a testosterone-dependent mechanism.
Results and Discussion miR-27a/b-3p directly bind and regulate the expression of TFPIα. We first sought to identify miR-NAs as novel and specific regulatory elements in TFPI 3′ UTR. Computational results from the different algorithms showed up 4 miRNA candidates: miR-27a/b-3p, miR-19b and miR-24 (Table 1).
We validated in silico results by transfecting EA.hy926 endothelial cells (ECs) with all the miRNA candidates. Only miR-27a/b-3p mimics produced a decrease in both TFPI mRNA and extracellular TFPIα levels ( Fig. 1A and B), as well as anti-miR-27a/b-3p produced an increase in TFPI mRNA and extracellular TFPIα levels ( Fig. 1C and D). These results were confirmed in another EC type ( Fig. 1E-H). To reveal whether the effect of miR-27a/b-3p on TFPIα was direct or indirect, we cloned a fragment of TFPI 3′ UTR (NM_006287) into a luciferase vector. As a negative control, we deleted the seed region for miR-27a/b-3p. MiR-27a/b-3p significantly reduced the luciferase activity, whereas no effect was observed with the mutated luciferase vector (Fig. 2). Thus, we here described for the first time that TFPIα is directly regulated by two miRNAs, miR-27a/b-3p in ECs and this post-transcriptional regulation of TFPIα might have consequences on the endothelial functions. Emerging evidences have already pointed toward miRNAs as relevant regulators of haemostatic proteins. For example, TF is regulated by miRNAs from the miR-17~92 cluster in monocytes 7 and by miR-223 in ECs 20 . On the other hand, protein S (PS), a cofactor for full length TFPIα 21 , is regulated by miR-494 22 . Thus, we can speculate that the regulation of TFPIα , TF, and PS may be exerted by a miRNA hub that modulate the TF/TFPI axis, as it has been described for other signaling pathways 20,23 . Indeed, in silico analyses showed that miR-27a/b-3p could also target TF and PS, although these interactions have to be further demonstrated in vitro.

miR-27a-3p inversely correlated with TFPI expression in HUVECs.
With the aim to test whether the TFPI:miR-27a/b-3p interaction has physiological consequences, we quantified these transcripts in 74 HUVEC samples (from healthy human donors). By using a correlation model, we found that TFPI mRNA and miR-27a-3p levels were inversely and significantly associated (Fig. 3A). However, we found a lack of agreement between the in vitro and ex vivo results for miR-27b-3p ( Fig. 3B) that may be justified because its expression was 18-fold lower than miR-27a-3p in HUVECs (Fig. 3C), which is in accordance with published results 24 . Indeed, miR-27a and miR-27b were not correlated (Fig. 3D). Therefore it is expected that, in physiological conditions, the lower concentration of miR-27b-3p in HUVECs may explain its lack of association with TFPI levels and potentially a minor role in regulating TFPIα in ECs. As a control, we evaluated the correlations of miR-19b and miR-24 with TFPI. As expected, no significant correlations were observed ( Fig. 3E and F), which strengthen the results obtained for miR-27a-3p. Accordingly, the quantification of miR-27a/b-3p might be of interest in pathological conditions where miRNA deficiency or over-expression may have an important impact 25 . miR-27a/b-3p levels modulate the anticoagulant effect of TFPIα. We next investigated the functional relevance of TFPIα regulation by miR-27a/b-3p. TFPIα anticoagulant activity was measured in the supernatant of HUVECs transfected with both miRNAs. Transfection with miR-27a/b-3p significantly decreased TFPIα anticoagulant activity, while anti-miR-27a/b-3p significantly increased TFPIα anticoagulant activity (Fig. 4). These latter observations were further confirmed in EA.hy926 cells transfected with miR-27a-3p ( Figure S1). This finding suggests that miR-27a/b-3p levels variations may have an impact on TFPIα functionality. Hypoxic conditions have been described previously to up-regulate miR-27a/b-3p expression 26 while infective and inflammatory processes down-regulate their expression 27,28 . Given the extensive cross-talk between the coagulation and inflammatory systems, TFPI included 29 , additional studies are certainly needed to reveal the role of miR-27a/b-3p in TFPIα regulation and the impact on their activity under pathological conditions. TFPIα regulation by testosterone is dependent of miRNA expresssion. It is described that testosterone increases TFPI mRNA and protein levels 11,12 . Nevertheless, the specific mechanisms of this regulation are largely unknown. In addition, testosterone may also regulate miRNA expression 16,19 . Thus, we wondered whether miR-27a/b-3p could be implicated in TFPI regulation by testosterone. We treated EA.hy926 cells with a physiological dose of the main testosterone metabolite, dihydrotestosterone (DHT, 30 nM) and we observed a significant increase of both TFPI mRNA (Fig. 5A) and protein after 24 h (Fig. 5B). Indeed, such doses of DHT decreased miR-27a/b-3p ( Fig. 5C and D). These results were also shown in HUVECs ( Figure S2). Key enzymes implicated  . Protein levels in cell culture supernatants were measured by densitometry after western blotting with specific anti-TFPI antibody in EA.hy926 (B) and in HUVECs (F). A representative blot is shown. The TFPI mRNA in EA.hy926 (C) and in HUVECs (G) and the respective protein levels (D,F) of cells transfected with miR-27a/b-3p inhibitors, and determined by the same methods and in the same conditions, is also shown. The 2 −ΔCt method was followed to calculate the relative abundance of mRNA compared with endogenous control expression of β -actin (ACTB) (Ct = Threshold Cycle; Δ Ct = Ct sample gene -Ct endogenous control). For miR-27b and miR-24 the same scrambled control was used. Representative blots are shown. All results are represented as mean ± SD from at least three experiments performed in triplicate (*p < 0.05; **p < 0.01; ***p < 0.001).
in the miRNA biogenesis: Drosha, Ago2, Exportin 5, and Dicer were not significantly affected by DHT treatment ( Fig. 5E). Therefore, regulation of miR27a/b-3p by testosterone seems to be specific and non-dependent of miRNA biogenesis machinery. Although it has been shown that testosterone increases the rate of pri-miR-27a maturation through the binding of the androgen receptor (AR) to the promoter of the gene in LNCaP prostate cancer cells 30 , our results showed that DHT provokes a decrease of miR-27a/b-3p in ECs. These results could be partially explained through differences of regulation pathways between both cell types that can certainly have an impact on the miR-27a/b-3p-testosterone regulation 31 . Future studies are needed to investigate whether testosterone variations are involved in a miR-27/TFPIα dysregulation that could increase the cardiovascular risk inherent to aging. However, an important limitation of forthcoming studies will be the accessibility to the adequate samples, e.g. ECs, that can be circumvented by the use of an appropriate animal model. Finally, combination of androgen treatment with miR-27a/b-3p transfection ( Fig. 6A and C) but not with miR-19b or miR-24 ( Figure S3) in EA.hy926 cells abolished TFPI up-regulation produced by 30 nM DHT as well as the anticoagulant activity, showing a relationship between testosterone, TFPI, and miR-27a/b-3p. In fact, when ECs transfected with anti-miR-27a/b-3p where treated with 15 nM DHT, we observed a significant increase of TFPI mRNA (Fig. 6B) and of the anticoagulant activity (Fig. 6C) further demonstrating the potential involvement of miR-27a/b-3p in the regulation of TFPI levels in ECs. The effect of miR-27a/b-3p mimics were also shown in HUVECs ( Figure S4). Thus, the assessment of therapeutic alternatives with miRNAs may be of interest. This model suggests that in ECs high levels of miR-27a/b-3p might reduce the protective role of testosterone against cardiovascular disease 14,15 . This situation might be of particular relevance in men with low testosterone levels in relation with age or pathologies such as hypogonadism. Indeed, we hypothesized that miR-27a/b-3p may be upregulated in the plasma of men with low concentration of testosterone and TFPI 13 in comparison with men with normal levels. Results showed no differences in miR-27a/b-3p expression that may be due to a substantial overlap of TFPI expression between the 2 groups (Fig. 7). Data from higher sample size will help to elucidate the role that miR-27a/b-3p may play in the testosterone mediated downregulation of TFPI in vivo, and thus whether Plasmids including TFPI WT 3′ UTR or TFPI mutant 3′ UTR in which the seven nucleotides forming the seed region of miR-27a/b-3p were deleted. (B) HTC-DK cells were transfected with either TFPI WT 3′ UTR or TFPI mutant 3′ UTR along with 100 nM miR-27a/b-3p precursor. A scrambled precursor was used as control. Luciferase activities were normalized to renilla activities. The 2 −ΔCt method was followed to calculate the relative abundance of mRNA compared with endogenous control expression of β -actin (ACTB) (Ct = Threshold Cycle; Δ Ct = Ct sample gene -Ct endogenous control). All results are represented as mean ± SD from at least three experiments performed in triplicate (*p < 0.05; **p < 0.01). it could be a contributing factor to the risk of thrombosis associated with low testosterone levels 14,15 . In an attempt to further demonstrate the regulation of TFPI by miR-27a/b-3p through testosterone, we compared the levels of miR-27a/b-3p and TFPI mRNA in HUVECs in both genders and found no differences ( Figure S5). This can be due because at delivery, even if testosterone levels are higher in boys than in girls these levels are 10 times lower than those found in adult individuals 32 . Thus, the regulatory role of testosterone in newborns may be probably negligible.
In conclusion, our results showed that miRNAs represent a novel mechanism by which TFPIα , a relevant haemostatic molecule, is modulated in ECs. Further in vivo studies must be performed to reveal the real transcendence of our results. We suggest that testosterone may be a hinge in miR-27a/b-3p-TFPIα regulation axis (Fig. 8). Given the high variability of miRNA expression levels among healthy population in different tissues, the pathophysiological consequences of this regulation have yet to be established i.e., risk of cardiovascular disease 33,34 . In addition, the relevance of our experimental data should be further determined in the context of a potential use of miRNA inhibitors as anti-thrombotic drug.

Cell line and tissue samples. Endothelial cell line EA.hy926 was obtained from the American Type Culture
Collection (Manassas, VA). ECs were maintained in phenol-red free DMEM supplemented with 2 mM glutamine, and 10% fetal bovine serum (Life Technologies, Madrid, Spain). Human colon cancer cell line HCT-116 deficient for Dicer (HCT-DK) was a kind gift from Dr. Renato Baserga (Thomas Jefferson University, PA). HCT-DK were cultured in McCoy's 5A (Sigma-Aldrich, Madrid, Spain) supplemented with 2 mM glutamine and 10% fetal bovine serum. Umbilical cord samples were obtained from healthy donors that gave informed consent (La Fe Hospital, Valencia, Spain) (n = 74). Studies with human HUVECs were approved by Local Ethics Committee from Hospital Universitario Morales Meseguer in Murcia and performed according to the declaration of Helsinki. Primary cultures of HUVECs were obtained by standard procedures. Briefly, HUVECs were obtained by collagenase digestion from umbilical cords from healthy donors, and were cultured at 37 °C in humidified air containing 5% CO 2 . The selection of ECs was performed by filtration. HUVECs were grown to confluence in T-75 flasks precoated with endothelial cell attachment factor (Sigma-Aldrich, Madrid, Spain), in medium M199 1X (+ ) Earle's, 2mM L-glutamine, 25 mM HEPES, L-aminoacids (Life Technologies, Madrid, Spain), supplemented with 20% fetal bovine serum (Life Technologies, Madrid, Spain), 1% endothelial cell growth factor (ECGF) (Sigma-Aldrich, Madrid, Spain), 1 mM sodium pyruvate, 50 U/mL penicillin, and 50 μ g/mL streptomycin sulphate (Life Technologies, Madrid, Spain). Confluent EC monolayers were harvested from the culture flasks with 0.25% trypsin/EDTA solution/0.02% PBS (Biochrom, Cambridge, UK). Cells to be used for total RNA isolation were stored at − 80 °C after adding 5 volumes of RNAlater (Life Technologies, Madrid, Spain). Although we had no direct confirmation that the isolated HUVECs were a pure endothelial culture, each culture was examined using a LEITZ DM-IRB inverted fluorescence research microscope (LEICA, Wetzlar, Germany) to ensure that the morphology was consistent with ECs. In all cases cells were cultured under conditions favoring EC growth (medium supplemented with ECGF) and showed the same growth pattern: endothelial-like morphology like as cobblestone-shape area, spindle-shaped cell with cytoplasm distributed homogeneously, without presence of hyperplasia or hill-and-valley morphology, typical for vascular smooth muscle cells culture.  After 48 h levels of TFPI mRNA were measured by qRT-PCR as well as TFPI anticoagulant activity (C). The 2 −ΔCt method was followed to calculate the relative abundance TFPI mRNA compared with endogenous control expression of ACTB (Ct = Threshold Cycle; Δ Ct = Ct sample gene -Ct endogenous control). All results are represented as mean ± SD from at least three experiments performed in triplicate (*p < 0.05; **p < 0.01; ***p < 0.001).
Scientific RepoRts | 7:43500 | DOI: 10.1038/srep43500 Master Mix, No AmpErase ® UNG (Life Technologies, Madrid, Spain). The 2 −ΔCt method was followed to calculate the relative abundance of miRNA and mRNA compared with endogenous control expression, U6 for cells miRNA and ACTB for mRNA. Plasma miRNA was normalized with geometric mean of miR-103-3p and miR-191 (Ct = Threshold Cycle; Δ Ct = Ct sample gene-Ct endogenous control).
Western blotting and antibodies. Cell supernatants were collected from 24-well plates and BCA assays (ThermoFisher Scientific, Madrid, Spain) were performed to calculate protein concentration of samples. Protein concentration variability between samples was low (4.8-5.5 μ g/μ L for EAhy.926 and 1.3-1.6 μ g/μ L for HUVEC). Normalization in the western blot was based on the amount of protein loaded. For this purpose, 100 μg (EAhy.926) and 30 μ g (HUVEC) of total protein was separated by SDS-PAGE in reducing conditions by electrophoresis in 8% polyacrylamide gels. Gels were transferred onto PVDF membranes (Amersham Hybond P 0.45, GE Healthcare, Barcelona Spain) and blocked 1 hour in 5% w/v non-fat drymilk. Antibody against human TPFI (#ADG72, Sekisui Diagnostics, Rüsselsheim, Germany) was used 1:5,000 dilution and incubated overnight at 4 °C. The secondary antibody labeled with peroxidase (GE Healthcare, Barcelona, Spain) was used 1:10,000. ECL  Prime Detection Kit and ImageQuant LAS 4000 Imager (GE Healthcare, Barcelona, Spain) were used for sample detection. Densitometric analyses were performed with ImageJ software (http://rsb.info.nih.gov/ij/). Linearity between the amount of sample and densitometry was assessed (r 2 = 0.95). Data were expressed as changes relative to cells transfected with SCR, taken as 100%.
Plasmid construction and deletion mutagenesis. PCR product (198 bp) containing a fragment of TFPI 3′ UTR (NM_006287) with miR-27a/b-3p binding site was cloned into pCR 2.1 vector (Life Technologies, Madrid, Spain) using the primers: (5′ GAGCTCCGTTATTTTTACCGTGTTTTG and 5′ ACGCGTCGTTTGAGTGGTTTTCAG). SacI and MluI were used to digest positive clones (New England Biolabs, Ipswich, MA). The insert was subcloned into luciferase reporter plasmid pMIR-REPORT (Life Technologies, Madrid, Spain) previously digested with these enzymes. A deletion mutant of pMIR-REPORT-TFPI without miR-27a/b-3p seed region-binding site was generated using the primers S: 5′ GAGCTCCGTTATTTTTACCGTGTTTTG and AS: 5′ ACGCGTCGTTTGAGTGGTTTTCAG with the QuikChange site-directed mutagenesis kit (Agilent Technologies, Santa Clara, CA). Insertion of TFPI 3′ UTR fragment and seed sequence deletion was confirmed by sequencing. PCR amplification was performed on plasmid DNA (Platinum Taq High Fidelity, Life Technologies, Madrid, Spain) and cleaned-up using ExoSAP-IT (Affymetrix, Santa Clara, CA). Sequences were performed using the BigDye Terminator Reaction Chemistry v3.1 on Applied Biosystems 3130 XL DNA analyzers (Life Technologies, Madrid, Spain). Alignments and sequence analysis were performed using Seqman Pro program (Lasergene version 7.1, DNASTAR, Madison, WI).
TFPIα functional assay. The inhibitory activity of TFPIα was measured by functional chromogenic assay on supernatants from transfected cells (EAhy.926 and HUVECs), where TF-FVIIa dependent FXa generation was quantified using Actichrome ® TFPI activity assay kit (Sekisui Diagnostics, Rüsselsheim, Germany) following manufacturer's protocol. Briefly, 20 μ L of culture supernatant (adjusted by BCA method to optimal concentration depending on cell line) were added into the wells of a micro-test plate. Then, 20 μ L of FVIIa/TF complex were added and incubated at 37 °C for 30 min. After that, 20 μ L of human FX were added into each well and incubated for 15 min. After adding 20 μ L of EDTA, the reaction was initiated by adding 20 μ L of spectrozyme FXa substrate. Absorbance was read at 405 nm in a spectrophotometer for 10 min and Vmax was calculated. Statistical analysis. Statistical analysis was performed with SPSS 21.0 (SPSS, Inc., Chicago, IL, USA) and GraphPad Prism 3.0 (GraphPad Software, Inc., La Jolla, CA, USA). For two groups comparisons, non-parametric test (U Mann-Whitney) was used. For multiple comparisons, one-way ANOVA on ranks with a post-hoc test (Bonferroni correction) were used. Association between continuous variables was analyzed by the Spearman's rank correlation coefficient. Normal distribution of each variable was analyzed using one sample Kolmogorov-Smirnov test with Lilliefors significance correction. Results were expressed as mean ± SD. All tests were two-sided, and statistical significance was defined as a p-value < 0.05 (or smaller for corrected p-values in multiple comparisons).