In order to build a normal animal, development of individual tissues and organs needs to be tightly coordinated with the developmental progression of the whole organism . This process is even more critical in insect species, including B. mori, that undergo well-defined developmental stages (metamorphosis). To gain insight into post-transcriptional gene regulation in silkworm development, determining a near-complete set of miRNAs and their expression patterns is essential.
Novel miRNAs in silkworm are preferentially expressed in the pupal stage
Recent studies in Drosophilid species led to the suggestion that most novel miRNAs originated from non-miRNA sequences and only a small fraction (4%) is expected to be retained during evolution . Only 2.5% of the surviving miRNAs are expressed at moderate levels . The observation that most of the newly identified silkworm-specific miRNAs were expressed as abundantly as some of the conserved miRNAs implies that these novel miRNAs evolved, survived, and possibly integrated into the silkworm post-transcriptional regulatory networks. Interestingly, most of these miRNAs showed highest expression in the pupae stage, suggesting a complex gene regulation involving species-specific miRNAs in the development of adult structures.
We have predicted over 1100 genes as targets for the newly identified silkworm-specific miRNAs and their stars using hitsensor software  (Additional file 1). Cloning of several novel miRNAs specifically from pupal stage suggests an important role for these miRNAs in larval-pupal-adult metamorphosis. Consistent with this suggestion, many predicted targets for novel miRNAs are genes essential for insect development, molting and metamorphosis, which are regulated by two hormones (ecdysone and its active metabolite 20-hydroxyecdysone (20E) and the sesquiterpenoid juvenile hormone (JH) . A novel miRNA, bmo-mir-2763 is predicted to target diapause hormone receptor-4. Bombyx diapause hormone receptor (BmDHR) is expressed in the prothoracic gland (PG), the organ, which synthesizes and releases the insect molting hormones, ecdysteroid . JH synthesis and degradation are two routes that control JH levels in the insect haemolymph (Sheng et al., 2008). Juvenile hormone acid O-methyltransferase (JHAMT) is a key enzyme involved in JH synthesis and is a predicted target for miRNA, bmo-miR-2998. The expression profile of JHAMT mRNA in silkworm and Drosophila melanogaster suggest that the suppression of JHAMT transcript is critical for the induction of larval-pupal metamorphosis . Juvenile hormone esterase (JHE) hydrolyses the JH, thus regulates the levels of JH . JHE is a predicted target for bmo-mir-2766 in silkworm. Thus, the genes encoding enzymes implicated in JH biosynthesis and degradation are predicted targets for novel miRNAs in silkworm. These target predictions suggest that JH levels are regulated by two novel miRNAs that are specifically and abundantly expressed in pupal stage.
A predicted target for novel miRNA (bmo-mir-2763) is nuclear receptor GRF (germ cell nuclear factor (GCNF)-related factor). Increasing evidence suggests that GCNF is important for both female and male reproduction . The developmental program of spermatogenesis is regulated by several transcription factors, one of which appears to be GCNF . GCNF is a sequence-specific repressor of transcription, which binds as a homodimer or an oligomer to its cognate response elements . Identification of GCNF as a target for silkworm-specific miRNA specifically expressed in pupal stage suggests a role for this miRNA in development of male and female reproduction.
During the terminal stages of differentiation of the silkworm ovary, the follicular cells that surround the oocyte produce a large number of related polypeptides that are involved in the formation of the eggshell or chorion . MicroRNA, bmo-mir-2763 is predicted to target a BmGATA beta isoform 3, which regulates the expression of a class of chorion genes expressed during the late stages of choriogenesis . Consistent with the detection of bmo-mir-2763 in the pupal stage, GATA-beta 3 in silkworm has been detected in pupae, but none of the larval tissues , suggesting that this process is restricted to pupal stage and novel miRNA (bmo-mir-2763) has a role in the process of choriogenesis. Another miRNA, bmo-mir-2733e-1 is predicted to target NADPH cytochrome P450 reductase, a component of the microsomal P450 electron transport system. It plays an essential role in the transfer of reducing equivalents from NADPH to various P450 molecules and said to be involved on in the ecdysone 20-hydroxylation during the insect's embryonic development 
The predicted target for bmo-miR-2998* is silk gland factor-1 (SGF-1). SGF-1 is a member of the fork head/HNF-3 family, which regulates transcription of tissue-specific genes . It plays a role in organogenesis processes such as those of the gut, silk glands, and nervous systems, act as a region-specific homeotic gene . It was also speculated that SGF-1 protein may be initially required for the development of silk glands and subsequently utilized in the control of genes coding for silk proteins. The miRNA, bmo-miR-2998* can potentially target Argonaute-2. Argonaute (Ago) 2 is the catalytic engine of RNA interference, but little is known concerning the regulation of Ago2 by miRNAs or miRNA* in animals, whereas in plants miR168 can target AGO-1, the major Argonaute involved in target mRNA degradation. Recent studies indicate that Ago2 is required for a variety of developmental processes that occur in a tissue-specific manner . These predictions indicate that several of the new miRNAs and miRNA* are likely to play potentially important roles in metamorphosis by regulating genes that participate in these processes.
Differential regulation of miRNAs during silkworm development implies a role for miRNAs in diverse aspects of B. mori physiology and development
Evolutionarily, miR-1 is highly conserved and involved in muscle development and physiology [65, 66]. The sole miR-1 gene in Drosophila is transcriptionally up-regulated by Twist and Mef2 during myogenesis . In addition to miR-1, miR-133 and miR-206 are designated as muscle-specific miRNAs because of their abundance in diverse muscle tissues [38, 67]. In B. mori, miR-1 level is very high, miR-133 is low and miR-206 is extremely low, which suggests that miR-1 likely plays a major role in muscle development and physiology in insects.
Holometabolous insects including silkworm undergoes apoptosis during development from larvae to pupae to moth. The spinning larval stage is a stage of transition to pupae during which gut, salivary glands, the silk gland and larval-specific muscles undergo programmed cell death and histolysis . Several miRNAs known to play a role in apoptosis are expressed in a stage-dependent manner in silkworm. miR-2 and miR-13 families possess identical "seed" sequences, and both show almost similar expression profiles in the silkworm. The miR-2 family is known to target proapoptotic genes (reaper, grim and sickle) in Drosophila . Preferential expression of miR-2 and miR-13 families in spinning larvae and moth suggests that these miRNAs may promote apoptosis in these two stages. Another miRNA, miR-34 in C. elegans, plays a role in apoptotic and non-apoptotic cell death . In B. mori, miR-34 is preferentially and abundantly expressed in pupae relative to the other three stages (Table 3), and this expression profile is similar to what was reported for C. elegans . Thus, different miRNAs may play similar roles but in a developmental stage-specific manner in B. mori.
Wg/Wnt signaling is highly regulated and inappropriate activation or inhibition of the pathway results in developmental defects and diseases [71, 72]. miR-8 is a negative regulator of Wnt signaling in Drosophila . Abundant miR-8 during the silkworm spinning-larval stage suggests a role in Wnt signaling. The seed sequences of miR-8 and miR-429 are identical, indicative of a potential in targeting similar sequences (messenger RNA targets). However, in the silkworm, the two miRNAs vary greatly in abundance: miR-8 level was high whereas miR-429 level was extremely low. Perhaps, miR-8 plays a predominant role in Wnt signaling in B. mori.
In Drosophila, miR-14 and miR-278 are implicated in fat/energy metabolism . miR-278 knockout flies are lean suggesting an excessive utilization of lipid stores. The similarity between the expression profiles of miR-278 in silkworm (Table 3) and Drosophila  suggests that miR-278 could play a role in regulating energy metabolism in silkworm. miR-14 is also preferentially expressed in pupae of silkworm, which suggests that miR-14 could also play a role in fat metabolism. On the basis of the roles of these two Drosphila 's miRNA, it is tempting to speculate that the increased levels of miR-14 and miR-278 observed in the pupal stage could prevent an excessive utilization of fat. This hypothesis would also be consistent with studies in lepidopteran and other insect classes that have shown a preferential utilization of glycogen stores during pupation (Arrese and Soulages, Annual Rev. Entomology, in press). By preventing the utilization of fat, needed for reproduction during the adult stages, these miRNAs would promote the utilization of glycogen stores.
Of all four developmental stages we analyzed, the spinning-larval stage is important - silk is synthesized and a cocoon is built during this period. Strikingly, the overall miRNA abundance is almost doubled from feeding to spinning larval stages, which suggests a role for miRNAs in silk synthesis, besides their importance in developmental transition. The increase in miRNA levels is mostly quantitative and limited to conserved miRNAs. The levels of novel miRNAs in spinning larvae were very low. Similarly, miRNA levels are also sharply elevated in moths relative to the pupae. Again the increase could not be attributed to qualitative differences. Many conserved miRNAs along with a few species-specific miRNAs become more abundant and may regulate reproduction or aging.
The most abundantly expressed miRNAs in B. mori are also highly conserved miRNAs e.g, miR-1, miR-8, miR-10, let-7, miR-263a, miR276a, and miR-306 and were expressed in all four stages, albeit their expression levels vary across different developmental periods. These abundantly expressed miRNAs accounted for most of the differences in miRNA abundance in different developmental stages of the silkworm. Variation in the level of a specific miRNA across developmental stages indicates stage-specific function, whereas a similar expression over a range of developmental stages points to a possible role in basal and/or constitutive processes. This analysis also revealed that many miRNAs are similarly regulated during development, suggestive of possible co-regulation of these miRNAs during silkworm development (Figure 4a and 4b). The changes in miRNA expression profile could imply a fine-tuning or tight regulation of their targets in a spatial- and temporal-specific manner. Taken together, conserved miRNAs along with the species-specific miRNAs, in part, orchestrate the developmental progression of B. mori.
miR* levels for several miRNAs are dynamically regulated during development
The specific accumulation of miR* sequences for miR-8, miR-10, miR-37, miR-281 and miR-965 was particularly striking in the silkworm (Table 3 and Figure 5). miR-10* had a higher frequency than miR-10 in Drososphila [23, 74] and locust . These observations suggest that miR-10* tends to accumulate at a higher level. This implies a potential functional role for miR-10*, and possibly other miR* that tends to accumulate at significant levels. Indeed, Okamura et al.  recently demonstrated a regulatory role for miR-276*, which can potentially target a gene and negatively regulate its expression. For several miRNAs, we found that the accumulation of miR* did not correspond with the miR levels in different stages of B. mori. miR* accumulation may also be developmentally regulated in the silkworm, which suggests a more complex regulation of miR and miR* sequences than was previously proposed. The observation that several miR*s are more abundant than their corresponding miRNAs indicates that silkworm miR*s could have some endogenous targets. Because of sufficient complementarity between miR and miR* sequences, the accumulation of miR* could potentially regulate the miRNA precursor itself, as was shown for few cases in Arabidopsis  and rice (Li and Sunkar, unpublished data).
Comparision between miRNA array and sequencing-based profiling in the silkworm
Recently, miRNA microarray was used to profile their expression during silkworm metamorphosis [33, 36]. The expression patterns of several miRNAs differed between the two previous reports and this study [33, 36]. For instance, using miRNA array analysis Zhang et al.  reported that miR-9c was highly expressed in larvae, although it was also present in eggs and pupae. This observation prompted the authors to suggest that miR-9c is involved in regulation of metamorphosis from larvae to pupae. However, our sequencing-based analysis of miR-9c abundance sharply contrasts with that in the previous report (Table 3 Figure 4a). First, miR-9 family represented by three members, miR-9, miR-9b and miR-9c in silkworm and these members differ by one nucleotide. Thus, it is difficult to determine which of these members is being detected from a hybridization-based approach. Second, miR-9c expression is extremely low: we recovered one read from the pupae and two reads from the adults, and none from the feeding or spinning larvae (Table 3). Third, we recovered abundant miR-9 and miR-9b reads from all four libraries, and both exhibited a dynamic regulation (Figure 4). Overall, miR-9b level was lower than that of miR-9 in any given stage. miR-9 family had the highest expression in moths and the lowest in larvae. In another miRNA array-based study, He et al.,  demonstrated that let-7b and let-7d were expressed only in larval and pupal stages but not in the adult stage. Since let-7 family is represented by eight members in the silkworm, each of which differed by one nucleotide, which makes distinguishing difficult by the hybridization-based approach. On the basis of read counts, only let-7a and let-7j are highly abundant, and the remaining six members are low (Table 3). In contrast to what was reported , we recovered reads for both let-7b and let-7d from the adult library (albeit in low numbers) but none from the larval or pupal libraries. Another miRNA, miR-277, was reported to be specifically expressed in the moth stage . We detected miR-277 in all four stages. The level was highest in pupae and estimated to be threefold greater than that in the moths. Similarly, the expression profiles of several other miRNAs, as deduced from the array analysis, did not match our analysis. Perhaps, the array-based approach [33, 36] needs further validation and we believe that the sequencing-based study is more reliable. Results from miRNA array analysis are likely affected by false signals, cross-hybridizations or background noise.