**1. Introduction**

Hereditary hemorrhagic telangiectasia (HHT) is an autosomal dominant, vascular disorder with a prevalence of approximately 1 in 8000 people worldwide [1,2]. HHT is characterized by the presence of telangiectases in skin and mucocutaneous tissue and arteriovenous malformations (AVMs), direct connections between arteries and veins without capillary beds, in internal organs [1]. Over 90% of all HHT patients present chronic epistaxis due to nasal telangiectases, whereas chronic bleeding from gastrointestinal telangiectases develops in at least 20% of patients. The presence of pulmonary, hepatic, and cerebral AVMs in HHT patients suggests that they are at risk of life-threatening hemorrhage and complications involving shunting, leading to stroke, high-output cardiac failure, or brain abscess [3–5].

HHT is genetically heterogeneous with heterozygous mutations in at least three known genes including endoglin (*ENG*) causing HHT1 [6], activin receptor-like kinase 1 (*ACVRL1* or *ALK1*), causing HHT2 [7], and mothers against decapentaplegic homolog 4 (*MADH4* or *SMAD4*) causing familial juvenile polyposis associated with HHT (JP/HT) [8]. In addition, mutations in the *GDF2* gene, encoding bone morphogenetic protein 9 (BMP9), were described as the cause of an HHT-like syndrome [9]. *ENG* and *ACVRL1* are the predominant genes whose mutations account for over 85% of all HHT patients [10,11] and are considered potential therapeutic targets [12]. The phenotypes generated by either *ENG* or *ACVRL1* mutations (HHT1 or HHT2, respectively) generally display distinct clinical manifestations, but an overlap of their clinical symptoms commonly occurs. Interestingly, all the genes mutated in HHT encode proteins that are involved in the signaling pathway of transforming growth factor beta (TGF-β)/bone morphogenetic protein (BMP). Thus, the auxiliary receptor endoglin associates with the signaling serine/threonine receptor ALK1 and both proteins are able to bind ligands like BMP9, leading to the phosphorylation and nuclear translocation of a Smad protein complex that includes Smad4 [12–14]. Because endothelial cells are functional targets of circulating BMP9, and predominantly express endoglin and ALK1, these are widely accepted as the target cells in HHT.

A deficient expression of the HHT genes has been postulated to underlie the molecular basis of HHT pathogenesis. Mono-allelic loss of expression leading to haploinsu fficiency of the respective HHT proteins has been shown to dysregulate TGF-β/BMP signaling in endothelial cells, negatively impacting in cell proliferation, migration, and recruitment during vascular remodeling and angiogenesis [11,12]. In addition, in mice, homozygous knockdown of *ENG* or *ACVRL1* leads to HHT phenotypes [15–17]. More recently, using next-generation sequencing, Marchuk et al. [18] have been able to demonstrate the presence of low-frequency somatic mutations in telangiectases of HHT1 and HHT2 patients, suggesting that the bi-allelic loss of *ENG* or *ACVRL1* may also contribute to the development of vascular lesions.

MicroRNAs (miRNAs) are small noncoding RNAs (about 22 nucleotides in length) able to post-transcriptionally regulate gene expression by RNA interference [19,20]. They act as guides in complementary miRNA–mRNA complexes that recruit enzymatic species to silence mRNAs by mRNA cleavage (perfect complementarity) or ribosome destabilization (nonperfect complementarity). In this process, Drosha is a ribonuclease that plays a key role for miRNA biogenesis and regulates the TGF-β/BMP pathway by interacting with the Smad protein family which, in turn, modulates gene expression via miRNAs [21]. Interestingly, mice lacking Drosha in the vascular endothelium develop a vascular phenotype resembling HHT including dilated and disorganized vasculature, AVMs, and hemorrhages [22]. In this regard, the emerging role of miRNAs in human diseases [23,24] has prompted a few studies searching for specific miRNAs in HHT patients [25–27]. However, the expression of miRNAs associated with HHT remains mostly unexplored. Our aim in this work was to identify dysregulated miRNAs in plasma samples from HHT patients through quantitative polymerase chain reaction compared with controls. We find i) a downregulation in HHT1, but not in HHT2, patients of miRNA-370, which is predicted to target *ENG*; and ii) an upregulation in HHT2, but not in HHT1, patients of miRNA-10a, whose putative target is *ALK1*. These results sugges<sup>t</sup> that dysregulated plasma levels of miRNA-370 or miRNA-10a are potential biomarkers that could help to identify undiagnosed HHT1 or HHT2 patients, respectively. In addition, they also may contribute to a better understanding of the complex biological processes associated with the development of HHT.

#### **2. Material and Methods**

#### *2.1. Algorithms for miRNA-target Predictions*

Computational prediction of miRNA targets was carried out using three target prediction programs which use di fferent algorithms: (i) microRNA (www.microRNA.org) [28]; (ii) Target Scan (http://www.targetscan.org/vert\_72/) [29]; and (iii) MicroCosm (https://tools4mirs.org/software/mirna\_ databases/microcosm-targets/) [30]. The mirSVR score was used as a regression method for predicting likelihood of target mRNA downregulation from sequence and structure features in microRNA/mRNA predicted target sites.
