This is the first report that comprehensively analyzes the miRNA expression profile during antigen-specific T cell differentiation. Our results indicate that only a few miRNAs are dominantly expressed in CD8 T cells and that their expression is dynamically regulated during the differentiation process. Moreover, the level of miRNA expression appears to inversely correlate with the activation status of the cells. For example the level of miRNA expression was highest in the non-replicating and functionally quiescent naïve T cells, but was dramatically reduced in the actively dividing and functionally competent effector T cells and their expression once again increased when the antigen-stimulated T cells were converted into the relatively quiescent memory T cells. Theses results suggest that miRNAs may modulate the differentiation status of antigen-stimulated T cells.
Recent studies in conditional dicer-deficient mice and mice deficient for individual miRNAs have highlighted the importance of miRNAs in regulating T cell development, homeostasis and function
. However, a comprehensive study of miRNA profiles during antigen-specific T cell differentiation has not been reported. Our direct cloning results show that although over 90 known miRNAs were cloned from CD8 T cells, only 7 miRNAs were dominantly expressed and constituted ~60% of the total miRNA pool. This is also consistent with previous studies showing that some miRNAs are dominantly expressed in specific cell types and tissues 
. Since the miRNA-mediated repression of gene expression is dependent on the miRNA amounts as well as the target mRNA levels
, the dominantly expressed miRNAs may control the abundantly expressed genes in CD8 T cells. However, we are not suggesting that the other miRNAs expressed at low frequencies are functionally unimportant.
A recent study has shown that regulation of miRNA expression characterizes the stage-specific development of thymocytes
. We also found that the miRNA expression levels dynamically change between naïve, effector and memory T cell subsets. A striking global downregulation of miRNA was seen in effector T cells. This trend was evident in all 3 methods examined: direct cloning, microarray profiling as well as RT-PCR. According to total microarray hybridization signal, the total miRNA level is down regulated 2.98 fold in effector compared to naive T cells. Interestingly, the changes in miRNA expression we observed appear to inversely correlate with the degree of cellular activation. Antigen stimulation of naïve T cells results in a massive upregulation of gene expression that is needed for effector T cells to divide actively, gain functional competence and change migratory properties
, and effector T cell is the cell type where we found the least overall miRNA. Since the upregulation of gene expression occurs at both the protein and mRNA levels, while the miRNA-mediated effect is largely thought to be mediated by translational repression, it is possible that the abundantly expressed miRNAs may primarily control the key transcription factors involved in T cell activation. Massive downregulation of miRNA expression has also been reported in cancer cells 
. Although the global decrease in miRNA level has been hypothesized to reflect the dedifferentiated status of cancer cells 
, our results suggest that it may indeed reflect the quantum of protein synthesis needed for cellular activity because effector cells are more differentiated compared to naïve T cells. In contrast to the general trend for downregulation in effector T cells, a few miRNAs were higher in effector T cells compared to naïve T cells. Indeed several genes, including those associated with lymph node trafficking, expressed in naïve T cells are silenced in effector T cells
. Further functional studies are needed to test if miRNAs upregulated in effector T cells suppress the expression of naïve T cell-specific genes in effector cells.
Two recent studies have also profiled miRNA in CD8 T cells
and our results are consistent with both the studies. For example, cloning miRNAs from single positive thymic CD8 T cells, Nelson et al
also found that a few miRNAs prevailed in frequency. In fact, the highly expressed miRNAs were also similar to this study, in that 7 miRNAs appeared in both studies among the 10 highest expressed miRNAs, with miR-142-3p showing the highest frequency in both. However, compared to their study, miR-181a and miR-24 were lower in frequency and miR-29a and miR-142-5p were higher in our study. Landgraf et al 
profiling miRNA expression in human peripheral blood CD8 T cells also found miR-142 to be the highest expressed miRNA. Similarly, they also found high level expression of miR-15a family (including miR-15a, miR-15b and miR-16), miR-21, miR-98 family (including all the let-7), miR-150 and miR-26a.
Although many methods can be used to profile miRNA expression, direct small RNA cloning is the only method that provides actual visualization of mature miRNAs. In our study, direct cloning revealed that mature miRNAs exhibit an extensive degree of end polymorphism in mouse T cells. Although we can not rule out the possibility that some of the observed polymorphism may reflect artifacts of the cloning procedure, this appears to be unlikely because many variant sequences were cloned multiple times and a high frequency of 5′ end variation was only seen in miR-142, while the majority of miRNAs had identical 5′ end. Moreover, although not described in detail except for a study by Ruby et al in C elegans
, miRNA end polymorphism has also been observed by others using small RNA cloning approach
and also other methods like RNA-primed Array-based Klenow Extension (RAKE)
. Our data show that the ends of mature miRNAs are much more polymorphic than reported in C. elegans
. We found 2 or more main variations for most mature miRNAs in the mouse T cells while one dominant sequence characterizes most mature miRNAs in C. elegans
. Although not alluded to by the authors, a careful examination of northern blot results of some published reports also reveal multiple bands for one miRNA
. One convincing example in C. elegans
is miR-84. As shown in in the paper by Lau and coworkers
, miR-84 is the only miRNA (out of 15 that were tested) that shows 3 clearly visible bands in northern blots, indicating that miR-84 has 3 main variations. Similarly, an examination of the deep sequencing data depicted in Table S1
in the report by Ruby and coworkers 
also shows that miR-84 has 3 main variations. Thus, some miRNAs in C. elegans
also have more than 1 dominant variation, suggesting that multiple dominant sequences for one miRNA is a conserved phenomenon, although it happens to a lower extent in C. elegans
Clearly, the changes in the 5′ end seed sequence can affect target selection
. However, the biological significance of the 3′ end polymorphism we observed is not immediately apparent, although the extent of their occurrence suggests that it could be important in miRNA biology. For example, it may possibly affect the stability/subcellular localization and/or miRNA functional efficacy. It has been reported in plants that untemplated 3′ end extension appears to be a signal for mature miRNA degradation
. It is possible that it might be the same in mammals. Whatever the significance, our study shows that designation of one sequence for a give miRNA may not be accurate to describe all miRNA species in a cell. Moreover as we have shown, the 3′ end variation can affect the detection efficiency of real-time PCR assay employing stem-loop primer and thus should be taken into consideration when designing a detection method for miRNAs. The method RAKE that requires the exact match of the 3′ end also appears to be affected 
. Additionally, the real picture of mature miRNA sequence might help generate more accurate models in bioinformatic calculations such as target prediction.
It is intriguing how the end polymorphism is generated. Since a majority of mature miRNAs have identical 5′ ends, irrespective of whether they are derived from the 5′ or 3′ chains of the hairpin, it appears that both drosha and dicer can set a precise 5′ end. Thus, the majority of end polymorphisms is likely generated downstream of drosha/dicer processing. It is reasonable to predict that there are enzymes with end modification activity that could generate mature miRNA end polymorphism.
End polymorphism in miR-142 was very different compared to other miRNAs—both miR-142-5p and miR-142-3p were the only miRNAs that showed a high frequency of 5′ end changes. This polymorphism appears to result from a shift in the drosha/dicer processing, in that 1-3 nucleotide deletions at the 5′ end were associated with a corresponding number of templated nucleotide extensions at the 3′ end. It has recently been proposed that the distance from the stem-flanking ssRNA junction determines the drosha cleavage site and the cleavage occurs approximately 11bp from the junction. Dicer then measures from the terminal set by drosha for further processing
. Although most miRNA 5′ ends were intact, suggesting that drosha/dicer processing is precise, the changes observed in miR-142 suggest that a shift in drosha/dicer processing can sometimes occur at a relatively high frequency. Since 5′ end changes can alter target specificity
, this may also represent a novel mechanism to generate additional miRNAs. Further studies are needed to verify this hypothesis. Although how a shift in drosha/dicer processing occurs is not known, A to I editing may be involved. It has been reported that pri-miR-142 is under heavy A to I editing by ADAR. miR-142 expression level increased significantly after ADAR was mutated and this change appears to be miR-142-specific and not a result of a general effect on the level of miRNA in ADAR mutant mice
. Thus, it is possible that the A to I editing may change the stem loop structure or affect the ability of drosha-DGCR8 to cleave precisely. Interestingly, 2 (out of 34) instances of A to I editing could be seen in miR-142-5p, while it was barely detected in other miRNAs in this study ( and S4
In summary, our study documents the changes in miRNA expression during T cell differentiation and points to the extensive degree of polymorphism in mature miRNA ends. Further studies on the differentially expressed miRNAs should aid in the understanding of T cell differentiation process and may ultimately pave the way to manipulate the process of adaptive immunity.