**4. Discussion**

Over 1900 miRNAs have been identified in humans and many of them have been reported to be involved in a variety of human diseases, biological functions, and signaling pathways, including the TGF-β/BMP route [23,24]. Among these, there are a growing number of miRNAs with high potential to be used as biomarkers in plasma and/or serum to clinically diagnose or provide accurate prognosis for survival in patients with cardiovascular diseases [47]. HHT is an autosomal dominant, genetic disorder in which patients develop hemorrhagic vascular lesions called telangiectases and AVMs. A few reports have previously searched for specific miRNAs in plasma [25,26] or peripheral mononuclear cells [27] from HHT patients. Recently, it has been reported that Drosha, a key enzyme in miRNA biogenesis, regulates vascular development and homeostasis via the TGFβ/BMP pathway, and rare missense mutations in the Drosha gene may predispose carriers to HHT [21,22]. These findings support the hypothesis that miRNAs play a functional role in HHT. However, the expression of miRNAs and their potential application as biomarkers in HHT remains mostly unexplored. In addition, how miRNAs might be involved in HHT development is not fully understood. Our goal was to measure several miRNAs in HHT patient-derived plasma as in previous studies [25,26], but with the novel strategy of selecting miRNAs potentially targeting the predominant HHT genes. We report here for the first time that plasma miRNA-370 levels are downregulated in HHT1, but not in HHT2, patients; while plasma miRNA-10a levels are increased in HHT2, but not in HHT1, patients. While further studies in a large

cohort of patients are warranted, this finding suggests that miRNA-370 and miRNA-10a could be used as biomarkers with potential application in HHT diagnosis [48]. Actually, this possibility would be of grea<sup>t</sup> interest because there is a need to diagnose HHT in those individuals who do not present with all of the typical symptoms, such as in asymptomatic children and young adults [4,5]. When the HHT gene mutation of the family is known, the genetic testing is the choice option in children and teenagers; the genetic results being definite for either the positive diagnosis or exclusion [10,49]. However, in the absence of a known family mutation, genetic determinations imply the sequencing of at least *ENG* and *ACVRL1* genes, a process that is time-consuming and expensive. Even so, in 10–15% of the subjects, the mutation is not identified. Therefore, it would be useful to have alternative tools to ideally allow an earlier, faster, cheaper, and easier HHT diagnosis. In this sense, several potential biomarkers, including miRNAs, have been described in HHT [48]. Thus, the specific downregulation of miRNA-370 in HHT1 and upregulation of miRNA-10a in HHT2 can be added to the reported dysregulated expression of miRNA-27a, miRNA-205 [25], miRNA-210 [26], miRNA-28-5p, and miRNA-361-3p [27] in different HHT subsets, as shown in Table 2. Unfortunately, due to the different experimental approach and material source (plasma or peripheral mononuclear cells (PMNCs)) used in each of these studies, no comparative conclusions could be drawn. Regarding miRNA-10a and miRNA-214, a previous report has suggested their dysregulated expression using miRNA array analysis in a pool of genotyped and nongenotyped HHT patients with pAVMs [26]. However, to the best of our knowledge, no conclusive, quantitative measurements of these miRNAs have been made so far. In this work, we find that miR10a expression is specifically upregulated in HHT2, whereas the levels of miRNA-214 were unaffected in either HHT1 or HHT2 patients. Nonetheless, a direct comparison between the results by Zhang et al. (2013) and this work is not feasible since the HHT population analyzed in each case is different and pAVMs are present in up to 50% of the HHT population with a predominant association with HHT1 patients [5,50]. Future investigations should be addressed to unify the source of the patient samples and its methodological processing in order to identify miRNAs that could serve as reliable biomarkers in HHT.


**Table 2.** miRNAs associated with HHT 1.

1 Total RNA from either plasma or peripheral mononuclear cells (PMNCs) was isolated, and levels of specific miRNAs were quantified, as indicated in the corresponding references. Increased (↑), decreased (↓), or unaffected (↔) levels of miRNAs are shown. HHT, hereditary hemorrhagic telangiectasia. ND, not determined.

A good biomarker is a defined characteristic that is measured as an indicator of normal biological processes, pathogenic processes, or responses to an exposure or intervention [51,52]. This broad definition encompasses molecular characteristics such as the dysregulated expression of miRNA-370 and miRNA-10a presented in this work. Once a potential biomarker is identified, the next step is to assure that it can be measured reliably and precisely [51,52]. In this regard, further validation studies using a larger cohort of HHT patients are needed, including disease stratification analysis to assess the possible correlation between the levels of miRNA-370 and miRNA-10a and the severity of symptoms. In addition, it would be interesting to measure the circulating subsets of miRNA-370 and miRNA-10a such as those in plasma exosomes, plasma free miRNAs, or mononuclear cells from HHT patients.

To understand the value of a biomarker, it is necessary to know the pathophysiological relationship between the biomarker and the relevant clinical endpoint [51]. That miRNAs have a functional impact

on HHT development is supported by recent next-generation sequencing studies of DROSHA, a key ribonuclease involved in miRNA biogenesis [21]. Thus, four rare variants of DROSHA, predicted to damage protein function (P32L, P100L, K226E, R279L), are present at a much higher frequency in HHT patients compared with healthy controls, suggesting that they may contribute to the HHT onset [22]. However, the function of miRNAs associated with HHT, such as miRNA-370 and miRNA-10a, remains to be elucidated. Both miRNA-370 and miRNA-10a regulate angiogenesis [36,38,53], a VEGF-dependent biological process involved in the pathophysiology and therapy of HHT [54,55], by targeting, at least, the VEGF receptors (VEGFR) KDR (kinase insert domain receptor) [38,39] and FLT1 (fms-related tyrosine kinase 1) [45], respectively. In addition, circulating VEGF levels are increased in HHT1 and HHT2 patients compared with the control population [48,56,57], and therapeutic antibodies to VEGF (bevacizumab) are currently used to alleviate HHT symptoms [55,58]. Interestingly, the VEGF/VEGFR pathway can be activated by endoglin or ALK1, thus promoting angiogenesis [59,60]. Therefore, by regulating VEGF-dependent angiogenesis, miRNA-370 and miRNA-10a may potentially impact the pathophysiology of HHT1 and HHT2. Of note, several experimental reports have already demonstrated the involvement of miRNA-370 in di fferent functions of endothelial cells, the target cells in HHT where endoglin is predominantly expressed [38,40]. Thus, miRNA-370 suppresses retinal capillary endothelial cell growth and apoptosis [38], inhibits the angiogenic activity of endothelial cells, and reduces microvessel density and sprouting in vivo, whereas a miRNA-370 inhibitor promotes endothelial sprout formation [40]. The antiangiogenic activity of miRNA-370 is fully compatible with the key functional role of its putative target endoglin in endothelial cells, as endoglin promotes angiogenesis in vivo and in vitro, at least, by stimulating proliferation and anti-apoptotic activity of endothelial cells [61]. Noteworthy, miRNA-370 and miRNA-10a are predicted to target *ENG* and *ALK1*, respectively, thus inhibiting its expression. While targeting of *ENG* by miRNA-370 in nonendothelial cells has been previously validated [34], to the best of our knowledge, the validation of *ALK1* targeting by miRNA-10a has not been reported. One limitation of our study is that the predicted targeting of *ENG* and *ALK1* by miRNA-370 and miRNA-10a, respectively, has not been experimentally validated in human endothelial cells. Whether the altered levels of circulating miRNA-370 or miRNA-10a in HHT1 or HHT2 patients contribute to the regulated expression and function of endoglin or ALK1 in the endothelium remains an interesting avenue of research in the future. Because the levels of miRNA-370 or miRNA-10a are specifically dysregulated in patients who present a deficient expression of *ENG* (HHT1) or *ALK1* (HHT2), respectively, a possible link between the expression levels of the HHT genes and those miRNAs can be postulated (Figure 3). Since haploinsu fficiency underlies the pathogenic mechanism in HHT1 and HHT2, it can be speculated that the dysregulated expression of miRNA-370 and miRNA-10a may a ffect HHT development by modulating the expression levels of *ENG* or *ALK1*. However, there is an opposite e ffect in the expression levels of miRNA-370 (upregulated) versus miRNA-10a (downregulated), suggesting a contrary e ffect on the expression levels of *ENG* or *ALK1*. Moreover, the expression levels of miRNA-214, which is predicted to target *ENG* and *ALK1*, are not significantly a ffected in our HHT population study. These apparent discrepancies are likely explained by the fact that the genes encoding miRNA-370 (MIR370; chromosome 14), miRNA-10a (MIR10A; chromosome 17), and miRNA-214 (MIR214; chromosome 1) are di fferentially regulated by distinct gene expression programs. Furthermore, miRNA-370 and miRNA-10a may target not only *ENG* or *ALK1*, but also the other genes, especially those relevant to vascular development, which may turn out be of greater relevance in the complex mechanism of lesion development in HHT.

67

**Figure 3.** Hypothetical model for the involvement of miRNA-370 and miRNA-10a in HHT. Top, in healthy subjects, basal levels of miRNA-370 and miRNA-10a are predicted to target endoglin, ALK1, as well as the VEGF/VEGFR pathway, thus regulating angiogenesis. Bottom, in HHT patients, heterozygous mutations lead to a deficient expression of endoglin (HHT1) or ALK1 (HHT2), associated with increased levels of VEGF and the dysregulated expression of miRNA-370 and miRNA-10a, resulting in abnormal angiogenesis. Treatment with antibodies to VEGF (bevacizumab) contributes to angiogenesis normalization. ALK1: activin receptor-like kinase 1; VEGF: vascular endothelial growth factor; VEGFR: VEGF receptors.
