We have identified the nucleotide sequences and structural domains that are required for the function of mir-181a-1 and mir-181c through mutagenesis and domain-swapping analyses. We show that not only the nucleotides in the 5′ and 3′ ends of the stem but also those in the pre-miRNA loop are critical for mir-181a-1 activity. We find that mir-181a-1 and mir-181c have distinct activities in early T cell development, and the distinct activities of mir-181a-1 and mir-181c are controlled by their pre-miRNA loops (, ), indicating that miRNA genes encoding identical or nearly identical mature miRNAs can exert different biological activities determined by their unique loop nucleotides. Interestingly, the pre-miRNA loop sequences of mir-181a-1 and mir-181c are divergent, but each is evolutionarily conserved in multiple animal species (), suggesting that members of the same miRNA gene families may have evolved to achieve distinct specificities or degrees of activity via alterations in their pre-miRNA loop sequences. However, mir-181a-1/c mutants do not change the 5′ ends of mature miRNAs produced () and the levels of mature miRNAs produced from these genes have no consistent correlation with the activities of corresponding miRNA genes (). Clearly, our results demonstrate that pre-miRNA loop nucleotides have a key role in controlling miRNA gene function.
Phylogenetic comparison of pre-miR-181a-1 and pre-miR-181c loop sequences.
When interpreting the above findings, it is critical to draw a distinction between the activity of a miRNA gene and the activity of a mature miRNA, which are often used interchangeably in the literature. In this study, we have examined the activities of miRNA genes, which may be contributed by one or more of the RNAs (mature miRNAs and their precursors) made from these miRNA genes, not the activities of mature miRNAs alone. Although siRNA duplexes have been shown to function as miRNA surrogates to target gene repression when transfected into mammalian cells 
, mature miRNAs delivered into cultured cells by transfection are diluted quickly during cell expansion, prohibiting their use in long-term T cell development culture assays. Thus, we were unable to quantitatively measure whether transfected mature miRNAs might be functionally equivalent to RNAs produced from miRNA genes. Furthermore, given the complex small RNA sorting pathways in animal cells 
, we have not attempted to determine the efficiency of transfected miRNAs being incorporating into the pathways used by corresponding miRNA genes in vivo
It is also important to emphasize that limited conclusions can be drawn by correlating the changes in mature miRNA levels to the activities of corresponding mutant miRNA genes (, , ). We believe that such analyses can only establish correlations, but not causal relationships, between the levels of mature miRNA expression and the activity of miRNA genes. However, strong and positive correlations would support that pre-miRNA loop nucleotides control the activity of miRNA genes by influencing mature miRNA levels, whereas negative or inconsistent correlations would disfavor such idea. Most importantly, miRNA genes make at least three RNA species, including pri-miRNA, pre-miRNA, or mature miRNA, prohibiting us from definitively attributing miRNA gene functions to one of these RNAs in the T cell assays. Moreover, the complex regulatory steps in the miRNA biogenesis pathways and during early T cell development may also compromise the robustness of such correlation analyses. Finally, we cannot determine whether alterations in pre-miRNA loop sequences affect all or only selected cognate targets. Since multi-target regulation is required for the mir-181a-1
function in T cells and the ectopic expression of a single miR-181a insensitive target is sufficient to block the function of the mir-181a-1 
, it is possible that pre-miRNA loop nucleotides only contribute to the regulation of one or a few targets among those that are regulated by the mir-181a/c
or their mutants.
Nevertheless, based on the functional importance of various nucleotides in mir-181a-1
genes, we may postulate some mechanisms by which pre-miRNA loop nucleotides control the activities of miRNA genes (). Pre-miRNA loops might influence miRNA function by controlling the processing of pri-miRNAs into pre-miRNAs, the transport and sub-cellular localization of pre-miRNAs, the processing of pre-miRNAs into mature miRNAs, or the loading of mature miRNAs into RISCs 
. For example, animal cells are thought to have multiple RISCs with distinct Argonatue proteins (AGO1–4). If pre-miRNA loops play a role in guiding small RNAs into various RISCs, which might have different efficiencies in gene silencing 
, then mature miRNA levels would not necessarily correlate with the function of the miRNA genes. Moreover, functional miRNAs may be generated from pre-miRNA loops, since pre-miRNA loop-derived small RNAs have been found in miRNA cloning and deep sequencing analyses 
. Finally, recent studies have shown that LIN-28 may recognize the pre-let-7 loop nucleotides and block the processing of human pri-let-7 RNAs into mature miRNAs in embryonic stem cells, suggesting that pre-miRNA loops may be recognized by LIN-28 or LIN-28-like RNA binding factors that control mature miRNA biogenesis 
. However, these RNA binding factors would control the activity of miRNA genes by blocking mature miRNA biogenesis 
, which is inconsistent with the fact that the levels of mature miR-181a/c have no consistent correlations with the effects of corresponding mutations on the activities of the mir-181a-1/c
genes (, , ). Therefore, pre-miRNA loop nucleotides may not control mir-181a-1/c
activity through recognizing the LIN-28 or LIN-28-like RNA binding factors.
A “heat map” of the functionally important nucleotides in the pre-miR-181a-1 region according to mutagenesis analyses.
Since pre-miRNA loop nucleotides — not the levels of mature miRNAs — were more critically linked to the function of miRNA genes, our findings raised an important question: which of the RNA species synthesized from miRNA genes is the functional one(s)? Definitively addressing this question is vital to understanding of the mechanisms by which miRNA genes control gene expression. One possibility is that pri-miRNAs and/or pre-miRNAs may have functional roles in target gene regulation before they are further processed. Supporting this idea, our findings demonstrate that pri-miRNAs and pre-miRNAs contain not only the mature miRNA sequences that can pair with cognate target sites, but also the loop nucleotides that are important for the activity and functional specificity of miRNA genes. Moreover, pre-miRNA-like stem-loop structures have been shown to be a common module for intermolecular RNA:RNA interactions 
. Thus, based on these findings (–), we speculate that pre-miRNA loops may play roles in target gene regulation as a functional motif of the pri-miRNA and pre-miRNA RNAs. This model can explain why both pre-miRNA loop and “seed” nucleotides play key roles with respect to the function of corresponding wild-type and mutant mir-181a-1/c