This study provides the first profiling of miRNAs in patients with non-vavular PAF and HCs. We found that in the LAA miR-155, miR-146b-5p and miR-19b are aberrantly up-regulated in patients with AF compared with HCs. A group of miRNAs regulate genes encoding cardiac ion channel proteins and other relevant genes, and several studies have reported the direct involvement of miRNAs in controlling cardiac excitability and arrhythmogenesis in certain pathological conditions [12]. Some such miRNAs have been shown to be directly involved in AF, and some are considered to have the potential to regulate AF based on their target genes. However, to our knowledge, earlier investigations verified the regulation of miRNAs in the atrial tissue from patients with valvular AF and other co-morbidities in non-AF patients (such as coronary heart disease). Therefore, from those studies it would not possible to deduce whether pre-existing heart diseases already influenced the expression of miRNAs.

Previous studies have reported that expression of miRNAs such as miR-1, miR-328 and miR-499 are altered in valvular AF patients, but it did not significantly change in our study [13‐15]. This variation in the findings, compared with other studies, may be attributed to the differences in the tissues that were sampled (LAA in the current study versus left atrial tissue in the other previous studies) and the heterogeneity of human myocardial samples. AF is a multifaceted electrical phenotype involving complex factors beyond electrophysiology and can occur in a variety of pathological settings. The different types of AF have distinct underlying mechanisms, and different miRNAs might be involved. The results of Cooley and Xiao et al. showed that the presence of valvular heart disease in AF influenced miRNA expression patterns in both left and right atria [16, 17].

Morphological and electrophysiological differences have been demonstrated between theright atria and left atria, which at least in part, may reflect different mechanisms involved in AF between the right atria and left atria. Thus, it is not surprising that AF-associated miRNAs of the right atria may differ from those of the left atria. A recent study compared the potential differences of AF-associated miRNAs in the RA and LA from rheumatic mitral valve disease patients. They have found the different distributions of AF-associated miRNAs in the right atrial appendage and LAA from rheumatic mitral valve disease patients. This may reflect different miRNA mechanisms in AF between the RA and LA. 26 But for the non-valvular AF, AF substrate in the left atria is related to AF initiation and maintenance, which plays an important role to initiate and perpetuate AF. And the potential difference of AF-associated miRNAs between right atria and left atria are still unknown due to lack of the RA and LA sample availability.

In our study miR-155, miR-146b-5p, and miR-19b were significantly up-regulated in non-valvular PAF patients. Most important, we also found significantly higher levels of expression of miR-155 and miR-146b-5p were independently associated with LAD, AF duration and hsCRP levels. However, only miR-155 and miR-146b expression level, LA dimension and AF duration were found to significantly predict AF recurrence in multivariate analysis.

Less invasive AF ablation procedures have been developed due to increasing knowledge about the pathophysiology of AF and due to the development of ablation devices, which replace the original ‘cut-and-sew’ technique for scarification [26, 27]. Wolf et al. developed an off-pump procedure, in which the pulmonary veins (PVs) are isolated, ganglionic plexus are ablated, and the LAA is amputated through a bilateral thoracotomy [26]. The LAA is derived from the primordial left atrium, which is formed mainly by the adsorption of the primordial PVs and their branches. This similar embryological origin suggests that the LAA may initiate AF like the PVs. This could explain the potential and active role of LAA as an underestimated source of AF. In our study, miRNAs profiling in LAA provided very compelling evidence, which illustrated the potential of miRNAs as a new mechanism for non-valvular PAF.

Interestingly, miR-155, which is reported to be involved in cardiovascular diseases, showed the largest increases in our AF patients compared with HCs. miR-155 might directly influence arrhythmia outcomes by targeting critical ion channel gene expression, including CACNA1C and KCNA4, which encode the cardiac L-type Ca²⁺ channel (I_CaL) α1c and the initial component of the transient outward potassium current (I_to1), respectively. And our results showed that the expression of the CACNA1C was significantly lower in the AF group than the HC group, not only for the protein level, but also the mRNA level. Moreover, miR-155 up-regulation may also play a role in inflammation [21]. In fact, emphasis has been given to miR-155 because it represents a typical multifunctional miRNA [28]. Even though the mechanism that regulates miRNA function and expression has not been completely understood, the information presently available (i.e. over expression in AF patients) allows us to recognize miR-155 as a gene of potentially paramount clinical importance in AF diagnosis and treatment. It is reasonable to speculate that evaluation of mRNA-155 in tissues or biological fluids might be utilized as a biochemical parameter for AF detection and prognosis.

MiR-146, is also thought to be a mediator of inflammation along with miR-155 [29]. The expression of miR-146 is up-regulated by inflammatory factors such as interleukin 1 and tumor necrosis factor [22]. miR-146 regulates a number of targets, which are mostly involved in Toll-like receptor pathways that bring about a cytokine response as part of the innate immune system [30]. Recently, miR-146b-5p was reported to be up-regulated in AF patients, which is consistent with our findings [18]. Potential targets of miR-146b-5p include TGIF1, MMP16 and TIMP4 that are reported to be regulated in cardiomyocyte fibrosis, and these might be expected to contribute to AF by promoting fibrosis, an established AF substrate [31]. This possibility remains to be investigated. Our results showed that the expression of the TIMP-4 was significantly lower in the LAA of the AF group than in that of the HC group, not only for the protein level, but also the mRNA level.

MiR-19b also takes part in inflammatory responses enhancing or repressing pro-inflammatory mediators’ expression [23]. It positively regulates Toll-like receptor signaling with DICER-1 deletion and miRNA depletion. The miR-19b is an important protagonist in this phenomenon, regulating positively the NF-kB activity [23]. The miRNA depletion inhibits cytokines production by NF-kB. Also, it is directly involved in the modulation of several NF-kB signaling negative regulators expression, indicating the importance of miR-19b on NF-kB signaling. This miRNA has not been extensively studied, but our data suggests a possible role of this miRNA in the pathology of AF, which is yet to be identified.

One of the limitations of the present study is that at this point we are unable to provide molecular insights for the cause of miRNA dysregulation in non-valvular PAF. The levels of miRNAs may be affected by multiple factors such as the change in expression in the tissue, the release of the miRNAs by the cells and the stability of miRNAs. Secondly, the exact targets and pathways by which alterations in miRNAs cause AF remain elusive. Although our analysis has indicated some potential genes and pathways that are involved in AF, it is difficult to estimate the actual false-positive rate of the overall target prediction. A better understanding of the biological significance of these broad, but often subtle, changes in miRNA expression that are found in AF patients could be achieved with the development of novel experimental models, in which the levels of one or more miRNAs could be precisely manipulated.