Small RNA regulation of ovule development in the cotton plant, G. hirsutumL

Cotton (Gossypium spp.) ovule development is an interesting and unique developmental process because the differentiation and development of natural fiber, a seed epidermal trichome, occurs along with cottonseed embryogenic development [1]. Cotton is a very important natural textile fiber source, and cottonseed is a significant food source for humans and livestock [2]. Further, cotton fiber is not only an excellent single-celled model system to study cell elongation and cellulose biosynthesis in plants from a biological perspective, but also from a commercial perspective as the high cellulose content makes it an excellent biomass-material to produce ethanol-based biofuels [1, 3]. Thus, molecular genetic studies of the stage-by-stage process of ovule development is important for understanding molecular mechanisms of fiber development and the goal of effective manipulation of fiber characteristics.

Cotton fiber is derived from a single cell. Fiber growth involves four overlapping developmental stages: fiber initiation, elongation (primary wall synthesis), wall thickening (secondary wall synthesis), and desiccation (maturation) [4]. Lint fiber initiation is conventionally timed at or just before anthesis [1]. Fiber initiation is a synchronous process that usually ends at 2 days post anthesis (DPA), but may extend up to 5 DPA [5]. This is then followed by an elongation stage at 5–20 DPAs, biosynthesis of secondary wall at 21–40 DPA, and maturation at approximately 40–60 DPA [1, 6–8]. Fuzz fiber initiation and development occur after initiation of lint fiber development but this, too, is subject to variation from variety to variety [4, 8].

Recent advances in cotton genomics [3, 9] have led to the identification of the core genetic components contributing to the cotton fiber development and its molecular mechanisms [for reviews see [1, 8]]. Although these studies have elucidated many aspects of fiber development and revealed important candidate genes expressed during developmental phases [5, 7, 8, 10], many aspects of fiber cell differentiation remain unknown [11]. In particular, the potential role of small RNAs, including microRNAs (miRNAs) and endogenous silencing RNAs (esiRNAs), during ovule development remains to be determined [8]. Characterization of these small RNAs during the different stages of fiber development will contribute to identification of key molecular interactions that will, in turn, lead to better understand of the molecular mechanisms regulating cotton fiber development.

The universe of small RNAs, including both the 21 – 23 nt long miRNAs and the 24 nt long esiRNAs, has been expanding at an ever-increasing rate since their initial discovery in the early 1990s [12, 13]. Small RNAs regulate their targets via transcriptional or posttranscriptional suppression either by DNA or histone modifications (esiRNAs) or direct cleavage of mRNAs and translational repression (miRNAs and siRNAs) [14]. The ~21 nt size class also includes trans-acting siRNAs (ta-siRNAs) that, unlike other siRNAs, potentially silence messages that are different from the RNAs from which they have been processed, but share some level of sequence similarity. Ta-siRNA biogenesis is mediated by specific miRNAs that process 21 snt size ta-siRNAs from TAS genes through direct cleavage [15].

The impact of small RNAs on a wide array of cellular processes in both plants and animals has grown to the point where it is becoming more and more difficult to find cellular processes that are not impacted by them to some degree. In plants, small RNAs have been implicated in processes as diverse as flowering [16, 17] and overall cellular defense [18, 19]. However, it is in development, such as fiber development in cotton, where small RNAs appear to have a major role [20, 21]. For this reason, we have used a size-directed small RNA cloning strategy to isolate, clone, and sequence small RNAs expressed in eleven DPA periods of fiber development (0 – 10 DPA). We sequenced more than 6,500 clones and have identified nearly 2,500 candidate small RNAs. Among these candidates, we found 583 unique sequence signatures. Surprisingly, few of these candidates were identified as miRNAs in miRBase [22, 23]. The majority of the sequences we found are, rather, the 24 nt long signatures corresponding to esiRNAs like those found in Arabidopsis thaliana and other plants. In addition, only 6.5% of identified small RNA sequences (or 8% in more saturated portion) were observed in more than one DPA of ovule development. While so-called "deep sequencing" using next generation platforms [24] will undoubtedly increase multi-DPA representation of small RNAs, our initial observations suggest that the initiation and elongation stages of cotton fiber-development are at least partially regulated by specific sets of small RNAs. Finally, target predictions based on ovule-derived small RNA sequences indicate involvement in numerous important biological processes including processes involving previously reported fiber-associated proteins.

We have cloned and sequenced small RNAs derived from eleven DPA periods (0–10 DPA) of cotton fiber development. Our results support the potential importance of small RNAs in developing cotton ovules. Overall, we find that small RNA sequences are more diverse and abundant in early development periods (0–2 DPA) than in subsequent periods (3–10 DPA). This suggests that the genetic processes regulated by small RNAs in the initiation phases of ovule and fiber development are at least qualitatively different than those at later periods. Whether this says that those early genetic, physiological, and biochemical mechanisms are more complex in the initiation phase than they are in other stages like the elongation phase is unknown at this point. However, it remains that the small RNAs in our study were very diverse: 44% of them were represented by small RNAs sequenced only once and that the majority of these were found in the earliest DPAs of development. This is similar to the case of Arabidopsis, in which 65% of all unique small RNAs were sequenced only once [25]. Although the material and overall approach were different from what we did here, unique small RNAs represented ~38% of total reads in a genome-wide survey in Arabidopsis [25] while the unique small RNA sequences in cotton ovules represented ~23% in our study. In addition, the fact that only a small number of unique candidate small RNA sequences spanned two or more DPA periods suggests that small RNA regulation in each DPA period in cotton is different. Perhaps, small RNA regulation in each DPA is highly specific and that shifting controlling biological processes occurs quite rapidly between days in ovule development.

A surprising finding in our study is that, out of 583 candidate small RNA sequences from 0–10 DPA developing ovule tissues, only two plant miRNA families (miR172 and miR390) were confirmed in miRBase. There is evidence that miRNAs constitute a much smaller proportion of the small RNAs in plants than in animals [30]. Our data are clearly consistent with this view. However, in our data, there are many other 21- to 23-mer small RNA sequences that do not match any of the currently miRBase-annotated plant or other organism miRNAs. These are potential candidates for new cotton-specific miRNAs, which need to be further explored.

Replicated sequence differences observed in the mature sequence of both miR172 and miR-390 suggest the existence of multiple miR172 and miR-390 family members in cotton, and potentially, there are different miR172 and miR-390 members functioning at different DPA periods. The miRBase confirmed miRNAs putatively target proteins that may play an important role both in ovule embryonic and in fiber development. Proteins putatively targeted by miR172 and miR390 include MYB and zinc finger transcription factors, glycosyl transferease family proteins, action/hydrogen exchanger, translation initiation factors (eIF-4A), and myosin heavy chain proteins. These proteins have been reported to be associated with fiber development, including being important components in gene regulation, in cytoskeleton and cellulose synthesis, and in proton and cation transporting [5, 7, 8]. In addition, although experimental validations are needed, other putative target proteins of miR172 may also be involved in the fiber development process. For example, short chain dehydrogenase/reductase (SDR) which is targeted by miR172 at 0 DPA. SDR has cellulose and pectin containing cell wall oxireductase activity and is involved in ABA biosynthesis, in which ABA is considered to be an important phytochormone in fiber development [8]. Additional proteins targeted by miR172 that are likely candidate proteins affecting fiber development include phosphoenolpyruvate carboxylase [31], the glutamate receptor involved in dendritic cell growth [32], and YT521-B-like family proteins changing alternative splice site usage in concentration dependent manner [33].

In Arabidopsis, miR390 was reported to target TAS3 trans-acting siRNA (ta-siRNA) biogenesis through coupling with AGRONAUTE7 (AGO7) and regulating AUXIN RESPONSE FACTOR3 (ARF3). Identification of at least two miR390s, of which, one is expressed at 7 DPA, is good evidence supporting specific involvement of miR390 in fiber development as auxin response is one of the most important factors in fiber initiation and elongation in cotton [8]. In addition, since the majority of ta-siRNAs are 24 nt long and are hypothesized to be generated in miRNA-induced cascades [34, 35], it is possible that miRNAs that are expressed at different DPAs of ovule development are the headwaters of within-DPA ta-siRNA cascades that may be indicated here by the abundance of unique 24-mer RNAs in the various DPAs.

Although structurally characterized [25, 36], the function of miR-853 is not clear in the literature. The ovule-derived candidate miR853-like small RNA targets several unknown proteins in Arabidopsis and cotton gene index databases. Arabidopsis ath-miR853 matches several putative targets in cotton gene index database. Of those, both extensin-like proteins and RAS-related proteins are known to be involved in fiber development [8, 37, 38]. In addition, a palmitoyl – acyl carrier protein thioesterase, catalyzing the palmitic acid of the fatty acid family in plants [39], could be important. The role of fatty acids (FA) and very long chain fatty acids (VLCFA) in fiber development has been reported [40–43]. This suggests that miR853-like small RNAs play a role in fiber development of cotton, possibly as miRNAs, and requires further study.

Other small RNAs in both the total or in the abundant copy portion targeted many a priori fiber development-associated proteins that have been reported in previous studies. Ovule-derived candidate small RNAs, putatively targeted many transcription and translation factors, biosynthesis/metabolism (catabolism), hormone mediated signal transduction pathways, and hormone responsive proteins and factors of all key plant phytochormones such as IAA, ABA, GA, BR, ethylene and cytokinin, which are known to be the key factors associated with fiber development [1, 8]. Other important fiber-associated factors reported in the literature are involved in the transportation of proteins, carbohydrates, lipids, ions, and electrons [7, 44–47], in lipid and fatty acids biosynthesis/metabolism [40–43], in cytoskleton formation [1, 5, 7, 8, 10, 48], in peroxidase activity [49], in carbohydrate biosynthesis/metabolism [1, 45, 47, 50–53], and in DNA biogenesis (e.g. endoreduplication) [1, 8, 54]. Those factors were also found to be targeted by candidate small RNAs in cotton ovules in our study. It is noteworthy to mention that ovule-derived candidate small RNAs also target some of the recently highlighted proteins such as prohibitin and steroid sulfotransferase, MATE efflux protein, and transducin family proteins that are differentially expressed in fiber initials [5], and actin depolymerizing factors (ADF). Some of which, like GhADF2, are predominantly expressed in fiber tissue [55]. Others, like the dynamin family proteins, are differentially expressed in fibers cells at 15 DPA period of the superior quality chromosome substitution line CSB22sh [7]. Recently, MADS-box genes [56] and genes of vesicle coating and trafficking [57] were found to be associated with fiber development where these related proteins are targeted by ovule-derived small RNAs annotated in this study.

Although in silico target predictions have been shown a good tool to putatively annotate the small RNA functions [29], the experimental validation of the exact biological functions of small RNAs in plant cells is necessary. Transient expression systems using in vitro ovule culture [1] with these candidate small RNAs may provide a valuable tool for rapid validation of small RNAs and miRNAs. Cloning and characterization of small RNAs selectively from the fiber cells of developing ovules using a newly developed methodology [5] should efficiently facilitate the identification of fiber-specific small RNAs/miRNAs in cotton and differentiation of fiber-specific [5] versus ovule specific [58] small RNA signatures. In addition, characterization of small RNAs/miRNAs from fiber mutants such as naked seed (n₁), Ligon lintless (Li₁, L₂), pilose mutant (H₂), immature fiber mutant (im), and other fiber mutants with distinctive fiber development may help to identify key small RNAs/miRNAs affected by these variations and further elucidate the mechanisms of the fiber development process. Annotation of small RNA pools from the remaining ovule and fiber development stages (10–50 DPA) will also be important for the understanding of developmental processes such as the secondary wall deposition and maturation stages of fiber development [48]. Consequently, with the availability of complete cotton genome sequences [9] in a near future, mapping of these siRNAs throughout the cotton genome will facilitate studies of structural and functional processes, biogenesis, and evolution of these ovule-derived small RNAs/miRNAs in cotton. These all require further attention and efforts on comprehensive studying of small RNA world of complex fiber development process in cotton.

We have carried out an initial survey of small RNA species expressed during cotton ovule development from 0 DPA to 10 DPA. Our results provide initial evidence of extensive small RNA-mediated regulation of complex ovule and fiber development processes in cotton. The majority of small RNA sequence signatures observed corresponds to the 24 nt size characteristic of endogenous silencing RNAs (esiRNAs), and there is very little carry over between DPA periods. Where there is DPA-to-DPA carry over, those sequences that have been identified are plant miRNAs. The observation that there are considerably more different sequence signatures present in the earliest DPAs of ovule development (0 DPA to 2 DPA) than in later periods may indicate that there is a shift in the regulatory landscape after 2DPA with fewer small RNAs present but in higher numbers later on. The overall patterns of small RNA expression observed raise the possibility that this regulation is consistent with miRNA-initiated small RNA regulatory cascades potentially targeting a large number of previously known fiber-associated proteins as well as previously unknown targets. Confirmation of our results, in particular the three miRBase-confirmed plant miRNAs and a the large number of 24 nt esiRNAs putatively involved in fiber initiation and elongation stages of ovule development, by ongoing deep sequencing efforts will greatly facilitate understanding of the developmental mechanisms involved.