To date, more than one thousand plant miRNA genes have been annotated and some of them have been well characterized [3
]. However, the number of plant miRNAs appears not to be saturated and many other functional miRNAs in plant species remain to be investigated. Compared to annotated miRNAs from Arabidopsis and rice, very few miRNAs from cotton plants have been identified. Recently, several studies performed in silico
identification of miRNAs from G. hirsutum
]. Approximate 18 highly conserved miRNA families were detected and several less conserved miRNAs (or families) were found. When compared to the miRNAs predicted previously, most of them could be recovered by deep sequencing, and only a small portion of them (e.g
. miR391 and miR400) were not [17
]. These missing miRNAs might be attributed to the fact that the sequences (e.g
. EST or GSS) used for prediction were derived from tissues such as leaves or roots rather than ovules. Also, it was likely that false positive predictions were included. On the other hand, several miRNAs (or families) such as miR159, miR172, miR319, miR473, miR477, miR479, miR530, miR535, miR858 and miR894 were not successfully predicted, suggesting that these miRNAs may be tissue-specific in cotton ovules. MiR172 and miR390 have been recently cloned from cotton (G. hirsutum
) ovule using a traditional cloning approach [32
], both of which were also detected in this study.
The present study is the first to deep sequence the small RNA population of G. hirsutum ovules where cotton fiber cells initiate and develop. Millions of unique siRNA sequences ranging from 18~28 nt in length were detected. Analysis of the evolutionary conservation of these miRNAs revealed 111 individual conserved miRNAs belonging to 22 families. Together with the several G. hirsutum miRNAs existing in miRBase (Release 12.0, Sept, 2008, www.sanger.ac.uk/Software/Rfam/ftp.shtml), this study will bring the number of miRNAs in G. hirsutum up to 120.
The vast majority of conserved miRNAs from cotton ovules is not surprising. Most of the miRNAs identified in this study are conserved in Arabidopsis and only a few are conserved in other plant species. The phenomenon can be explained by the fact that cotton fibers (seed trichomes) and epidermal hairs (leaf trichomes) are phenotypically similar; both types of trichomes use a common mechanism, e.g
. that closely associated with the transcription factors for regulating trichome initiation and development [9
]. Notably, some highly conserved miRNA families such as miR156/157, miR167 and miR172 were sequenced more than ten thousands or even one hundred thousands times in a single library. These highly conserved miRNAs may represent a relationship between evolutionary conservation and expression abundance [24
]. On the opposite, some miRNA families that are less conserved or even species-specific have very low abundance. From an evolutionary view, these miRNAs play a role in establishing and maintaining phenotypic diversity between different groups of organisms and are involved in regulation of the lineage-specific pathways and functions [24
]. In addition to the conserved miRNAs, two putative miRNAs identified in this study do not have orthologs in Arabidopsis and other species (Table ). Since the non-conserved miRNAs usually express at a low level and in specific cell types (ovules), it is suggested that these cotton-specific miRNAs may have expanded after the divergence of the monocot and dicot plant lineages, supporting the presumption that a diverse set of miRNAs are evolving rapidly and independently with each species [34
Our comparative analysis of small RNA abundance between the wild-type and fl mutant indicated that the mutant contains an altered small RNA population, with a smaller proportion of total RNA reads. However, in fl mutants many small RNA families with 21-22 nt were enriched (Figure ). Differential miRNA abundance was also found between the wild-type and mutant. A surprising observation was that the majority of miRNA families in fl mutant had significantly higher abundance than in wild-type (Figure ). This result suggests that the mutant has a changed regulation of MIRNA expression during the fiber development. Further identification of the regulatory process and metabolic pathway in mutants will provide insights into the impaired miRNA biogenesis and abnormal trichome cell differentiation.
It is of interest that a large number of miRNAs from ovules potentially target transcription factors, which was consistent with our previous predictions [17
]. In Arabidopsis
, miR159 mediates cleavage of MYB101
transcripts, the two transcription factors that function as positive regulators of abscisic acid (ABA) response [35
]. Similarly, miR319 is complementary to a highly conserved motif in the coding region of the GAMYB-related clade of MYB
]. In this study, several members of putative MYB families were predicted to have binding sites for miR169, miR396, miR399 and miR858, suggesting the potential regulation of fiber development. MiR858 is of particular interest because it targets the GhMYB10
mRNA. Ectopic expression of GhMYB10
in transgenic tobacco plants causes abnormal cell shapes of leaf trichomes [38
Amongst the predicted targets of miR167 was Auxin Response Factors (ARF). These proteins are bound to the auxin response elements and regulate auxin-mediated transcriptional activation/repression. In vitro
-cultured cotton ovules, exogenous auxin is required to promote fiber cell development [26
]. Our data have demonstrated that in the mutant, miR167 was expressed more highly. In rice culture cells, miR167 was shown to cleave ARF8 mRNA. The abundance of miR167 was controlled by the level of auxin in growth medium. When cells were grown in auxin-free medium, miR167 level decreased [12
]. Similarly, a putative ARF8 transcript was predicted to be targeted by miR167 in cotton ovules. This suggests that auxin levels are possibly higher in the mutant ovules. We predicted miR160 to target an ARF10-like mRNA transcript, which expressed at higher levels in the mutant library, just like miR167. MiR393 targets putative Transport Inhibitor Response 1 (TIR1) transcripts in cotton. TIR1 is an auxin receptor involved in a mechanism leading to the Aux/IAA degradation [39
]. Inhibition of TIR1 by miR393 would down-regulate auxin signaling. MiR393 showed an expression level nearly 2 fold higher in mutants than in wild-type. Interestingly, most of the miRNAs involved in the auxin pathway were found to be up-regulated in the mutant.
In Arabidopsis, several miRNAs like miR399 and miR395 are induced by the nutrient deficiency [40
]. Under normal growth conditions, these miRNAs do not express. However, both miR399 and miR395 were moderately sequenced in this study, particularly in the wild-type ovules (Table ). This can be attributed to the advantage that deep sequencing can detect miRNAs at a very low level. Under phosphorus starvation, miR399 targets a ubiquitin-conjugating E2 enzyme, which in turn regulates Pi acquisition [41
]. In this study, deep sequencing identified 8 miR399 members and 7 unique genes were predicted as potential targets of miR399. More interestingly, all of these predicted targets appear not to be correlated with phosphate metabolism. This result provides a new clue to the multiple roles of miR399 that may play in diverse cell types or species. MiR399f/g were predicted to have complementarity to a putative MYB family transcription factor. Besides, miR399g targets four cotton vacuolar ATP synthase subunit B transcripts. In cotton fiber, elongation is driven by turgor pressure generated by vacuolar H+
-ATPase activity on tonoplasts [43
]. The process occurs synchronously with the increase in the rate of cell elongation, indicating that vacuolar H+
-ATPase may play a crucial role in cotton fiber development [44
Expression of miR398 was much lower in mutant than in wild-type. Previously, miR398 in Arabidopsis was identified to target gene coding Cu/Zn superoxide dismutase [45
]. Similar target was predicted for the miR398 from cotton ovules. Interestingly, specific cotton Cu/Zn superoxide dismutase have been recently detected in the secondary cell walls of developing cotton fibers and are suggested to be involved in cell wall growth [46
]. Whether miR398 regulates superoxide dismutase and cell wall growth would be an interesting topic to be investigated.