Second, we could detect a clear overrepresentation of uridine at the 59 start of major miRNA species, consistent with the known bias towards this nucleoside at the first position of miRNAs and the preference of mammalian Ago 2 for uridine or adenine in this position. While the modal miRNA tag length in cardiomyocytes was 22 nt, we observed common miRNA end heterogeneity, which was much more pronounced at the 39 end than at the 59 start site. As observed previously, heterogeneity at both termini was also less pronounced at the Drosha compared to the Dicer cleavage site, perhaps suggesting more a precise cleavage by Drosha, or indicating a cumulative effect of variability in successive processing steps. Additionally, it is clear that observable miRNA 39 ends in our dataset did not strictly follow 59 start choice, and thus much of the variability at the 39 end probably derives from post-processing addition and trimming events. This is likely further reflected in our analysis of miRNA duplexes as we find that a large minority do not conform to the 2 nt 39 overhang rule. Similar trends were seen in previous analyses, and the basis for some of the diversity of miRNA termini and length has been ascribed to structural features of the pre-miRNA. Additionally, GDC-0879 proteins associating with Drosha or the hairpin loop sequences are known to regulate individual hairpin processing, implying that many instances of end variability will be cell context-specific. Our data indicated a tendency towards symmetric structural Y-27632 dihydrochloride ROCK inhibitor distortions around and within in the Drosha cleavage region, while for Dicer, the immediate cleavage region was typically base-paired but immediately adjacent to either the terminal loop or other local distortions. Similar observations were made for C. elegans Drosha and Dicer sites. These structural features, while not strictly required, may contribute to miRNA processing accuracy, in addition to the inbuilt ��ruler�� functions of the Drosha and Dicer complexes. Importantly, in our dataset we observe high levels of variability for numerous individual miRNAs, which are likely to impact on their role in cardiac biology. Evidence in favor of this notion includes the observation that isomiR expression is regulated during development and variations at the 39 end can influence miRNA stability, loading into distinct Argonaute complexes, as well as potentially affecting target mRNA binding and the cellular location of miRNAs. There is also the strong expectation that 59 isomiRs will have distinct mRNA targeting properties, since changes to the miRNA 59 end will alter their seed region identity.