Direct Thrombin Inhibitors Prevent Left Atrial Remodeling Associated With Heart Failure in Rats

Summary The present study tested the hypothesis that thrombin participates in formation of left atrial remodeling and that direct oral anticoagulants, such as direct thrombin inhibitors (DTIs), can prevent its progression. In a rat model of heart failure associated with left atrial dilation, we found that chronic treatment with DTIs reduces the atrial remodeling and the duration of atrial fibrillation (AF) episodes induced by burst pacing by inhibiting myocardial hypertrophy and fibrosis. In addition to the prevention of thromboembolism complicating AF, DTIs may be of interest to slow down the progression of the arrhythmogenic substrate.


SUMMARY
The present study tested the hypothesis that thrombin participates in formation of left atrial remodeling and that direct oral anticoagulants, such as direct thrombin inhibitors (DTIs), can prevent its progression. In a rat model of heart failure associated with left atrial dilation, we found that chronic treatment with DTIs reduces the atrial remodeling and the duration of atrial fibrillation (AF) episodes induced by burst pacing by inhibiting myocardial hypertrophy and fibrosis. In addition to the prevention of thromboembolism complicating AF, DTIs During AF, thrombus formation is promoted by blood stasis in poorly contractile atria together with a hypercoagulable state, as indicated by high circulating levels of fibrinolytic degradation products, plasminogen activator inhibitor (PAI)-1, and thrombinantithrombin complex (2). For all of these reasons, anticoagulation is a central therapeutic target for most AF patients. Anticoagulation can be achieved via vitamin K antagonists or, more recently, with direct thrombin inhibitors (DTIs) or direct factor Xa inhibitors, referred to as non-vitamin K antagonist oral anticoagulants (3).
Thrombin is the central protease of the coagulation cascade. It converts soluble plasma fibrinogen into insoluble clot-forming fibrin polymers, and activates several positive feedback steps to amplify its own generation (4). In addition, thrombin has pleiotropic cellular effects through the cleavage of proteaseactivated receptor (PAR)-1, including hemostasis, inflammation, cellular growth, and proliferation (4)(5)(6). For instance, PAR-1 promotes hypertrophy of neonatal rat cardiomyocytes and deoxyribonucleic acid synthesis in fibroblasts (5,7). In mice, PAR-1 overexpression induces eccentric hypertrophy and dilated cardiomyopathy, whereas PAR-1 deficiency is associated with reduced left ventricle dilation after myocardial infarction (MI) (8).
Several hormones, peptides, or pathways are recognized to be involved in atrial remodeling, including the renin angiotensin system (9), but little is known about the role of thrombin. In vitro, this protein induces alterations of the electric and mechanical properties of rabbit left atrial strips, which are prevented by the DTI dabigatran and a PAR-1 antagonist (10). The present in vivo study was undertaken to test the hypothesis that thrombin participates in left atrial remodeling and AF substrate formation, known to be promoted by heart failure (11)(12)(13), and that DTI can slow their progression.
It was conducted using a rat model of heart failure secondary to an extensive MI, which is associated with left atrial remodeling and AF susceptibility (14,15). We found that DTIs and PAR-1 antagonists prevent atrial remodeling and reduce AF susceptibility.   Figure 1A). Remarkably, treatment with the vitamin K antagonist, warfarin, did not significantly reduce left atrial dilation. Furthermore, dabigatran etexilate slightly reduced the left ventricular dilation without improving the systolic function evaluated by fractional shortening and ejection fraction ( Figure 1B).

METHODS
The inhibitory effect of dabigatran on left atria size was confirmed by macroscopic examination at sacrifice of the animals ( Figure 1C).
Similar effects on atrial dilation were observed with another DTI, the nonprodrug S35972, which has anticoagulant and antithrombotic activities comparable to dabigatran etexilate between 1 to 20 mg/kg in rats (16,17). In addition, because this drug is highly soluble, it could be subcutaneously delivered via osmotic minipumps, which is a less stressful and more reliable route of delivery than oral gavage. Thus, S35972 was used in the rest of the study. At 4 weeks post-MI, DTI S35972 dose-dependently reduced the left atrial area, which increased only 1.2-fold compared with sham animals at the optimal dose of 6 mg/kg/day, without any effect on left ventricle (Supplemental Figure 2A).
As PAR-1 is the main thrombin receptor in the heart (5) These results indicate that the DTIs inhibit and prevent the progression of the atrial remodeling associated with heart failure in rats.

DTIs REDUCE ATRIAL DILATION INDEPENDENTLY
OF THEIR ANTICOAGULANT PROPERTIES. We next investigated the relationship between antiremodeling and anticoagulant properties of DTIs. First, we determined their availability in rat plasma by means of mass spectrometry. In rats with MI treated with 25 mg/kg/day dabigatran etexilate or 6 mg/kg/day S35972, the plasma concentrations ranged from 28 to 72 ng/ml at peak for dabigatran and 22 to 85 ng/ml for S35972, which corresponds to the low therapeutic levels of DTI detected in humans (18). The inhibition of atrial dilation was maximal at DTI concentrations ranging from 30 to 100 nmol/l as previously reported for the inhibition of thrombin-mediated PAR-1 cleavage by dabigatran (19), but it decreased with high plasma concentrations ( Figure 3A). At the Jumeau et al.   Masson trichrome staining of atrial cross sections from vehicle-treated rats with MI revealed an overall hypertrophy of the myocardium, with enlarged myocytes and interstitial fibrosis as indicated by arrows in Figure 4A. In atria from rats with MI treated with DTI for 8 weeks, myocyte hypertrophy was markedly reduced and fibrosis was barely detectable. Staining of subendocardial and interstitial collagens with picrosirius red confirmed an increase in interstitial fibrosis in vehicle-treated rats with MI and a significant decrease in DTI-treated rats with MI ( Figure 4B). We then analyzed the effect of DTIs on messenger ribonucleic acid (mRNA) expression using rat gene-specific primers (Supplemental Table 1). First, we studied mRNA expression of connective tissue growth factor (CTGF), which is involved in the remodeling of extracellular matrix (20) and is up-regulated in AF patients (21). Post-MI treatments for 1 and 2 months prevented the up-regulation of CTGF in left atria, and treatment initiated 1 month after the MI tended to decrease mRNAs ( Figure 4C). We also analyzed the expression of PAI-1, which has both thrombogenic and fibrogenic activities (2,20). Post-MI treatment for 2 months abolished the induction of PAI-1 in the left atria ( Figure 4D), but it had no effect on both PAI-1 and CTGF mRNAs in the left ventricle ( Figure 4E). These

Direct Thrombin Inhibitor and Atrial Dilation
A U G U S T 2 0 1 6 : 3 2 8 -3 9 heavy chain (MHC) and natriuretic peptides (22). At 8 weeks post-MI, b-MHC transcripts significantly increased, whereas the a isoform (a-MHC) did not change in dilated atria of vehicle-treated rats ( Figure 5B). In addition, the mRNA level of brain natriuretic peptide (BNP) was increased, but not that of atrial natriuretic peptide (ANP). In sharp contrast, there was no change in b-MHC and BNP in atria from rats with MI treated with either the DTI S35972 or the PAR-1 antagonist F16618 in comparison with sham animals (Figures 5B and 5C). Of note, S35972 also down-regulated the expression of these markers in the ventricles of rats with MI (Supplemental Figure 3). We After 7 days of stimulation with 10 nmol/l thrombin, b-MHC but not a-MHC mRNAs were up-regulated, and this induction was prevented by F16618 ( Figure 6A).

Moreover, thrombin enhanced the expression of BNP
and ANP transcripts and the release of soluble BNP through PAR-1 (Figures 6B and 6C). It also increased PAI-1 mRNAs ( Figure 6D). Among the various PAR-1 effectors (23), the Rho-associated protein kinase (ROCK) is known to be involved in thrombin-induced endothelial permeability (26) and cardiomyocyte growth (27). Addition of the ROCK inhibitor Y27632 together with thrombin down-regulated BNP, ANP, and PAI-1 and reduced BNP secretion ( Figures 6B   to 6D). ROCK is a downstream effector of RhoA, which also activates the signal transducer and activator of transcription 3 (Stat3) (23). As expected, Stat3 is phosphorylated upon thrombin exposure, and both PAR-1 antagonist and ROCK inhibitor prevented this activation ( Figure 6E). Altogether, these results indicate that thrombin/PAR-1 induce atrial hypertrophic phenotypes through the Rho cascade.

DISCUSSION
The major finding of the present study is that chronic During heart failure, thrombin activates the RhoA Immunoblots of PAR-1 and nuclear factor of activated T cells 3 (NFATc3) and densitometric analysis. n ¼ 5 to 14 for each group; exact p values in blue are calculated versus sham rats and those in red versus vehicle-treated rats with MI. Abbreviations as in Figures 1 and 3.
Jumeau et al. Furthermore, we showed that 10 nmol/l thrombin, a concentration reached during the initiation phase of coagulation cascade (30), was sufficient to induce hypertrophic genes in atrial myocardium in vitro.
Interestingly, maximal inhibition of PAR-1 cleavage is obtained in vitro with 30 to 100 nmol/l dabigatran and 10 nmol/l thrombin, whereas higher DTI concentrations have activating effects (19). It is known that PAR-1 is activated at subnanomolar thrombin concentrations (31). Altogether, these data suggest distinct dose-dependency between the anticoagulant and antiremodeling effects of the DTIs. Indeed, they could inhibit the cellular actions of thrombin via PAR-1 before activation of coagulation cascade and formation of intra-atrial thrombus. In dilated and Direct Thrombin Inhibitor and Atrial Dilation fibrillating human atria characterized by a high risk of thrombus, it is likely that such a low level of thrombin is reached. In this line, DTI plasma concentrations, which inhibit atrial dilation in rats (28 to 75 ng/ml at peak and 22 to 85 ng/ml at steady-state) correspond to the low therapeutic levels in humans.
For example, DTI plasma levels, which have been measured in the RE-LY (Randomized Evaluation of Long-Term Anticoagulation Therapy) trial, ranged from 2 to 1,000 ng/ml at peak and 1 to 809 ng/ml at steady-state for the high dose of 150 mg dabigatran etexilate twice daily (18).
STUDY LIMITATIONS. The relevance of our experimental data to the human AF cannot be firmly established. They were obtained in a model of atrial remodeling, which develops with HF but not with spontaneous atrial arrhythmias. In some samples of human right atria from patients with paroxysmal AF, dilated atria, and normal ejection fraction (Supplemental Table 2), we analyzed mRNA expression of PAI-1 and hypertrophic genes using human gene-specific primers (Supplemental Table 3). PAI-1 and b-MHC/a-MHC ratio were increased, whereas the anticoagulant thrombomodulin was decreased (Supplemental Figure 4). This suggests a potential thrombin accumulation in dilated atria, because thrombomodulin accelerates its degradation (32). This is also consistent with a previous study on human atria that reported PAI-1 induction by the coagulation factor Xa in combination with rapid pacing of human atrial slice (33). Clinical studies on DTI have mainly evaluated the frequencies of stroke and major bleeding, and as specified for the large RE-LY trial, the case report forms were not prospectively designed to collect echocardiogram details (34). The