In our study, we showed a reduced expression of protein SERCA2 in infarcted tissue when compared to corresponding remote myocardium of patients with MI. These results are in accordance with previously demonstrated decrease in SERCA2 mRNA and protein levels in ischemic-reperfused heart and animal models of myocardial ischemia . A decrease in content and activity of SERCA2 is also observed during ageing . To eliminate misinterpretation of western blot results due to different ages of our patients, we did not compare SERCA2 in remote myocardium and infarcted tissue of patients with MI to healthy adult hearts. Since simultaneous decrease in SERCA2 mRNA and protein levels is not always observed , and there remain gaps in understanding regulation of SERCA2 by its protein regulators , we presumed that SERCA2 might be regulated by miRNAs. MicroRNA microarray expression analysis revealed 43 differently expressed miRNAs in infarcted tissue when compared to corresponding remote myocardium, 10 of these showing up-regulation (speculatively being involved in SERCA2 regulation). According to miRNApath analysis of calcium signalling pathway, approx. half of differentially expressed miRNAs are related to SERCA2 but not to its regulatory proteins. Compared to our previous study based on microarray analysis , where the miRNA expression in infracted tissue was calculated relative to the miRNA expression in healthy human hearts (independent groups of samples), only half of miRNAs showed similar expression in the present study, suggesting specifically expressed miRNAs in infracted tissue as well as in remote myocardium compared to healthy adult hearts. This is in accordance to recent findings that some miRNAs in the non-infarcted area might also participate in the pathophysiology response to MI .
SERCA2 has two major isoforms, SERCA2a and SERCA2b, both of which are expressed in heart from early developmental stages, although it is believed that SERCA2a is the major isoform expressed in the heart . Using four different available prediction algorithms, 213 miRNAs were predicted for binding to 3'-UTR mRNA SERCA2; however, used algorithms did not distinguish between SERCA2 isoforms. Of predicted miRNAs, two times more corresponded to an isoform SERCA2b than to an isoform SERCA2a, and only approx. 10% were differentially expressed in infarcted tissue compared to remote myocardium. Finally, only one of predicted and differentially expressed was up-regulated (miR-574-3p).
miR-574-3p was also predicted to target ATP2A2 in our previous study, and has been elevated in infarcted compared to healthy adult hearts . We therefore used this and other up-regulated miRNAs to search for potential binding using algorithm RNA22. The interaction of miRNA:mRNA usually occurs via non-strict base pairing, and the most important should be matching to the 1–8 nucleotides from the 5´ end of miRNA (“seed” region). miRNA-mRNA binding may also occur in other ways . Some features of miRNA:mRNA binding have been implicated in bioinformatics and experimental approaches, but most of the target sites identified by available tools are based on seed region complimentarity, therefore might be biased against the one class of binding sites. Since the majority of a given mRNA sequence is highly structured and local RNA accessibility of the binding site may be a critical feature of miRNA target recognition , we determined the free-energy (ΔG) of the miRNA:mRNA binding. We further tested all 10 up-regulated miRNAs for free-energy of 70 nt of 5´ and 3´ of the predicted binding sites to test whether the predicted binding sites are located in a region of very high free energy, suggesting a locally accessible site. Complex RNA secondary structures may prevent miRNA/mRNA interactions and may have inhibitory effects on miRNA:mRNA interactions. Repression of mRNAs may be also increased by multiplicity (several binding sites in a transcript for a single miRNA) and cooperativity (several miRNAs bind to a single transcript) . In line with that, several miRNAs were predicted for binding to the same positions of 3´-UTR SERCA2 (7 predicted miRNAs binding positions for SERCA2a, and 9 predicted miRNAs binding positions for SERCA2b), and for the majority of the predicted miRNAs there was more than one putative binding site. Most of predicted miRNAs have at least one binding site with perfect complimentarity to seed region.
As previously described, some of the up-regulated and in silico validated miRNAs have been already related to cardiovascular diseases . In addition to miR-122, which was specifically expressed on present microarrays compared to our previous study  and is according to recent report down-regulated in plasma of patients with MI , miR-574 and miR-140 have not been yet described as involved in cardiovascular pathology. Other up-regulated and in silico validated miRNAs were miR-199a, four members of miR-320 family and miR-483. miR-199a was described as involved in the maintenance of cell size in cardiomyocytes , and as a master regulator of a hypoxia-triggered pathway . miR-320 was shown to be involved in the regulation of cardiac ischemia/reperpusion injury through targeting heat-shock protein 20: over-expression enhanced cardiomyocyte apoptosis, whereas knockdown was cytoprotective . miR-483 was described as in vitro regulator of angiogenesis through serum response factor .
Further analysis using miRBase revealed that 43 differentially expressed miRNAs belong to 25 miRNA Gene family, and that half of differentially expressed miRNAs is clustered. Further, the 4 clusters are fully represented among differentially expressed miRNAs. This is expected since many genetically clustered and co-transcribed miRNAs are often expressed at different levels . TAM tool was used to annotate 43 differentially expressed miRNAs; more than half of the defined functions are related to heart pathology and physiology, as well as to 21 disease outcomes related to heart pathology. Based on annotation we used nine miRNAs to validate microarray results, using two different qPCR technologies.
Most of the validated miRNAs have been already described as being involved in post-infarct remodelling . Using TaqMan based technology; the miR-1 and miR-133 were used to compare present study to our-previous research  and confirmed up-regulation of miR-1 in remote myocardium. However, to the best of our knowledge, this is the first report of miR-98 down-regulation in human MI, which has been shown to be involved in regulation of cardiac hypertrophy , and its role in MI was proposed also through inflammation, although the verification still lacks . Using Sybr Green the differential expression was shown for miR-21, which is deregulated in numerous cardiac diseases [24, 36–38], although the role of miR-21 in cardiac pathology is controversial at present. miR-21 is broadly expressed in multiple tissue, expression in the heart cells is developmental and age-dependent , there is difference in expression between infracted regions and border zone  as well as in cell type (cardiomyocytes, fibroblasts, etc.) . We also showed down-regulation of miR-126, which plays an important role in ischemic angiogenesis , miR-125a/b, which are involved (according to TAM) in myocardial remodelling after MI, although their target genes in cardiovascular pathology are not known at the present.
In addition, we showed that the expression analysis using TaqMan or Sybr Green gave similar results. We also showed that analysis of RNAlater and FFPE with TaqMan gives the same results, which was also partially examined in our previous study . In addition, analysis of RNAlater and FFPE with Sybr Green is comparable. However, some discrepancies can be seen between RNAlater and FFPE samples, and this can be due to sampling error. FFPE tissues are microscopically examined before cutting for subsequent RNA isolation, but this cannot be performed when using RNAlater stored samples. Further, normalization strategy in qPCR experiments (often used for validation of microarray results) is in addition to the recent study  also important for the outcome of miRNA expression analysis in the human MI. Based on our experience, the best RG would be RNU6B, especially when using Sybr Green. In accordance to our previous studies [11, 33], miR-26b is as good RG, but only when using TaqMan technology.