In silico identification and characterization of conserved miRNAs and their target genes in sweet potato (Ipomoea batatas L.) Expressed Sequence Tags (ESTs)

Abstract

The endogenous small non-coding micro RNAs (miRNAs), which are typically ~21–24 nt nucleotides, play a crucial role in regulating the intrinsic normal growth of cells and development of the plants as well as in maintaining the integrity of genomes. These small non-coding RNAs function as the universal specificity factors in post-transcriptional gene silencing. Discovering miRNAs, identifying their targets, and further inferring miRNA functions is a routine process to understand normal biological processes of miRNAs and their roles in the development of plants. Comparative genomics based approach using expressed sequence tags (EST) and genome survey sequences (GSS) offer a cost-effective platform for identification and characterization of miRNAs and their target genes in plants. Despite the fact that sweet potato (Ipomoea batatas L.) is an important staple food source for poor small farmers throughout the world, the role of miRNA in various developmental processes remains largely unknown. In this paper, we report the computational identification of miRNAs and their target genes in sweet potato from their ESTs. Using comparative genomics-based approach, 8 potential miRNA candidates belonging to miR168, miR2911, and miR156 families were identified from 23 406 ESTs in sweet potato. A total of 42 target genes were predicted and their probable functions were illustrated. Most of the newly identified miRNAs target transcription factors as well as genes involved in plant growth and development, signal transduction, metabolism, defense, and stress response. The identification of miRNAs and their targets is expected to accelerate the pace of miRNA discovery, leading to an improved understanding of the role of miRNA in development and physiology of sweet potato, as well as stress response.

Introduction

The small, non-coding, and endogenously expressed microRNAs (miRNAs) are about ~21–24 nucleotides-long RNA molecules encoded not only in the genomes of plants and animals but also in viruses. Single-stranded precursors RNA molecules give rise to these endogenous miRNAs, which are subsequently processed to form stem-loop structures.^1,2 In plants, precursor miRNAs are cut into small, double-stranded RNAs with the help of dicer-like-enzyme (DCL1).³ After formation of the mature miRNA, they are incorporated into an RNA-induced silencing (RISC) complex that interact with the complementary sites of the target gene transcripts, which in turn aid in the downregulation of the target gene expression either by cleavage or by repressing the translation process on the basis of the degree of complementarity of miRNA within its target.^1,4-6

The miRNAs shows a high degree of complementarity (near perfect) with their targets in plants. But in contrast, miRNAs usually display partial complementarities to their targets in animal.^7-9 Although miRNAs share similarities in general, the pre-miRNAs are presumed to have larger and more variable stem-loop structures in case of plants. In most of the cases the mature plant miRNAs identify a single target site within the coding region, pair their target sites with near-perfect complementarity, and guide to cleave the target mRNA.¹⁰ In plants, the complementarity between miRNA and its targets is very high. Earlier studies conducted by Rhoades et al.¹¹ and German et al.¹² have shown that the cleavage of the target mRNA occurs in the middle of the mRNA-miRNA duplex, i.e., between the 10th and 11th nucleotide in the 5′ end of the miRNA. A miRNA may be involved in the regulation of multiple gene expression, while it is also possible that multiple miRNAs control a single gene. Therefore, the identification of the target mRNAs is essential for the functional analysis of the miRNAs.

It is interesting to note from many recent studies that many miRNAs are evolutionarily conserved in both animal and plant kingdoms. This sequence conservation of miRNAs has facilitated the prediction and study of miRNAs in non-model plants whose complete genome is yet to be decoded. Till date, several high-throughput techniques have been used for experimental determination of miRNAs. Direct cloning and deep sequencing has been used predominantly apart from the homology-based comparative-genome analysis methods using expressed sequence tag (ESTs) and genomes survey sequences (GSS), available in the public domain. The known mature miRNAs are evolutionarily conserved across the plant kingdom, ranging from mosses and ferns to higher flowering plants. This feature has enhanced the process of identification and characterization of conserved miRNAs through comparative genomics-based approach in other plant species.¹³ Furthermore, identification of conserved miRNAs from ESTs has some advantages over other methods.¹⁴ With the aid of high-throughput bioinformatics tools, miRNAs can be predicted efficiently with a high degree of accuracy, whereas they cannot be detected easily through direct cloning because of their low expression level and/or spatiotemporal expression pattern.¹⁵ As of now, comparative genomics-based approach has been used in the identification and characterization of miRNAs at genome-wide scale in various plant species including soybean,¹⁴ strawberry,¹⁵ cotton,¹⁶ wheat,¹⁷ potato,¹⁸ apple,^19,20 Solanaceae,²¹ switchgrass,²² citrus,²³ field mustard,²⁴ and Chinese cabbage,²⁵ etc. According to the latest release of miRBase (Release 20, June 2013), a sum of 7398 miRNAs from 72 diverse plant species belonging to the kingdom Viridiplantae have been identified. The tuber-bearing and economically important dicotyledonous plant sweet potato belongs to the family Convolvulaceae. This crop is usually grown in temperate, tropical, and subtropical areas across the globe.²⁶ The storage roots of sweet potato plant hold maximum amount of starch, which is used as staple food, animal feed, and also as raw material for alcohol production. Its stems and foliage are also used as forage.²⁷ Sweet potato is also rich in vitamin A, B, and C, iron, phosphorous, and calcium. In recent years, several methods have been employed in the improvement of sweet potato crop but breeding is constrained by the complexity of the genetics of this out-crossing hexaploid crop (2n = 6x = 96) and lack of genomic resources.²⁸ Notwithstanding the knowledge generated in recent years on the economically important sweet potato plant, the comparative genomics-based analysis of miRNA that regulate various regulatory pathways is still missing.

As of now, 23 406 EST entries of sweet potato have been reported in the dbEST of NCBI (April 1, 2013) and are quite useful for identification of potential new miRNAs. The conserved nature of these known plant mature miRNAs offers a novel way for identification of new orthologs through comparative genomics in other species. We used an EST-based comparative-genomics approach for the identification and characterization of conserved miRNAs and their putative target genes. Here, we performed a computational analysis and a homolog search, using a total of 3227 unique known miRNAs from Viridiplantae, against 11–658 assembled ESTs of sweet potato. After a series of filtration processes, only 8 miRNAs folding in 75 pre-miRNAs, belonging to 3 miRNA families, were identified in sweet potato. All the pre-miRNAs of sweet potato forms stem loop structures with stable minimal folding free energy (MFE). In addition, the target genes of the newly identified miRNAs of sweet potato were predicted, and the potential functions were elucidated.

Results

Mining of sweet potato ESTs for novel miRNAs

Earlier studies conducted by various researchers have shown that mature miRNAs in plants within the plant kingdom are evolutionary conserved.^29-31 This conserved nature of the plant miRNAs has greatly improved the identification of conserved miRNAs using EST and/or GSS sequences through comparative genomics. A total of 5939 plant miRNAs belonging to kingdom Viridiplantae were downloaded from miRBase. After removal of duplicate miRNAs from the known miRNA data set, only 3227 non-redundant reference miRNAs belonging to 847 different miRNA families were used to identify conserved miRNAs from Ipomoea batatas ESTs. A total of 23–406 ESTs of sweet potato were assembled by using in-house EST processing pipeline ESMP which includes the assembly program CAP3. Preliminary assembly resulted in 3050 contigs and 8608 singletons. The non-redundant 3227 mature miRNA of Viridiplantae were taken as reference and searched against the assembled ESTs for obtaining potential miRNAs by using locally installed BLAST. The contigs or singletons which closely matched (with maximum 2 mismatches without any gap) with the reference miRNA were chosen manually. The BLAST search predicted a total of 85 candidates (12 sequences from contig and 73 sequences of singleton) from the assembled EST showing homology with Viridiplantae mature miRNA. The predicted candidate miRNA sequences which closely matched with the previously known mature miRNAs were subjected to BLASTX for removal of protein coding sequence. The remaining 65 sequences were assessed for secondary structures by nucleic acid folding prediction software MFOLD. The probable false pre-miRNAs sequences were removed manually. Finally, based on the stringent criteria mentioned in the materials and methods section, the potential miRNAs of sweet potato were predicted. Only 8 (7 from singleton and 1 from contig) new miRNAs of sweet potato, which showed high degree of sequence identity with the known Viridiplantae miRNA, satisfied our criteria for miRNA encoding sequences as shown in Table 1. The predicted secondary structures of newly identified pre-miRNAs in sweet potato are shown in Figure 1A (A–D) and Figure 1B (E–H). It is evident from Table 1 that the length of the mature miRNAs in sweet potato ranges between 19–22 nucleotides. These potential miRNA candidates of sweet potato represent 3 different families: miR168, miR2911, and miR156. The 2 miRNA156 sub-families, miR156c and miR156j, were found to be located at the 5′ end of the miRNA precursors, while the 4 miRNA subfamilies, miR168, miR168b, miR168a, and miR2911, were found at the 3′ end of pre-miRNA. Interestingly, all the newly identified miRNAs in sweet potato were located on the plus strand. Minimal free folding energy (MFE) determines the stability of the perfect or near-perfect secondary stem loop hairpin structure of pre-miRNAs where a very low MFE value signifies a more stable secondary structure. The predicted potential miRNAs in sweet potato shows higher negative minimal free folding energies (MFEs), (ΔG Kcal/mol) was ranging from –28.10 to –67.10 Kcal/mol. In addition to minimal free folding energy, the minimal free folding energy index (MFEI) is also considered as one of the decisive factors to discriminate miRNAs from other types of coding and non-coding RNAs. The newly identified sweet potato microRNAs showed MFEI ranging from 0.77–0.95, as depicted in Table 1. Furthermore, only 1 contig sequence, i.e., contig56, was found homologous with 1 mature miRNA sequences (miR2911) and 2 singleton sequences, i.e., gi-76295387 and gi-114786556 were found to be homologous with 7 different mature miRNA sequences. The A+U content of predicted miRNAs in sweet potato falls in the eligibility range of > = 30% to < = 70%, as shown in Table 1. As evident from Table 2, the length of the identified pre-miRNAs in sweet potato is within the 76 to 146 nt long, with an average value of 103 nt. In the pre-miRNA of sweet potato, the distribution of G, C, A, and U are different, where the G (33.08%) and C (26.25%) nucleotides are predominant as compared with the other 2 nucleotides U (23.3%) and A (16.05%). The average GC content within the pre-miRNAs was reported to be 29.68%. The statistical summary of newly identiϕιed miRNA from sweet potato ESTs has been shown in Table 2. The frequency of miRNAs in the sweet potato EST collection was found to be approximately 0.068% (8 out of 23 406 ESTs).

Table 1. List of the potential miRNAs identified from ESTs of sweet potato (Ipomoea batatas L.).

Sweet potato	Mature miRNA sequence (MS)	miRNA Source	Strand	Loc	NM (nt)	LM (nt)	LP (nt)	A+U (%)	MFEs	MFEIs
Iba-miR1	UCGCUUGGUG CAGGUCGGG	> gi|76295387	+/+	3′	0	19	76	44.73	28.10	0.77
Iba-miR2	UCGCUUGGUG CAGGUCGGGA	> gi|76295387	+/+	3′	0	20	81	45.67	31.20	0.81
Iba-miR3	GGCCGGGGGA CGGACUGGGA	> Contig56	+/+	3′	0	20	97	24.74	67.10	0.95
Iba-miR4	CCCGCCUUGC AUCAACUGAA	> gi|76295387	+/+	3′	1	20	99	37.37	49.60	0.87
Iba-miR5	UCGCUUGGUG CAGGUCGGGAA	> gi|76295387	+/+	3′	0	21	83	45.78	32.00	0.82
Iba-miR6	CCCGCCUUGC AUCAACUGAA U	> gi|76295387	+/+	3′	0	21	101	38.61	50.70	0.89
Iba-miR7	UUGAGAGAAG AGAGAGAGCA C	> gi|114786556	+/+	5′	1	21	145	39.31	57.65	0.92
Iba-miR8	GUUGAGAGAA GAGAGAGAGC AC	> gi|114786556	+/+	5′	2	22	146	39.04	59.85

Characterization of the novel identified I. batatas miRNAs. The novel identified I. batatas miRNAs were characterized in terms of LP, precursor miRNA length; MFE, minimal folding free energies; MFEI, minimal folding free energy index; AMFE, Adjusted minimal folding energies; MS, mature sequence; NM, number of mismatches; LM, mature sequence length; Loc, Location of miRNA; miRNA source, accession number of singletons and contigs.

0.90

Figure 1A.

Newly identified potential miRNAs in sweet potato (Ipomoea batatas) along with mature and precursor sequences and their predicted stem loop structures. The matured miRNA portion is highlighted in black bar. The structures were generated using MFOLD program. The actual size of the precursors may be slightly shorter or longer than the one shown in the figures. (A) Iba-miR1 (homolog of gma-miR168b). (B) Iba-miR2 (homolog of cme-miR168). (C) Iba-miR3 (homolog of han-miR2911). (D) Iba-miR4 (homolog of zma-miR168b-3p).

Figure 1B.

Newly identified potential miRNAs in sweet potato (Ipomoea batatas) along with mature and precursor sequences and their predicted stem loop structures. The matured miRNA portion is highlighted in black bar. The structures were generated using MFOLD program. The actual size of the precursors may be slightly shorter or longer than the one shown in the figures. (E) Iba-miR5 (homolog of cca-miR168a). (F) Iba-miR6 (homolog of aly-miR168a-3p). (G) Iba-miR7 (homolog of ahy-miR156c). (H) Iba-miR8 (homolog of cme-miR156j).

Table 2. Some of the features of newly identified miRNAs in sweet potato.

Sweet potato miRNA	miRNA family	Homologous miRNA	A%	C%	G%	U%	A/U Ratio	G/C Ratio
Iba-miR1	miR168b	gma-miR168b	15.78	19.73	35.52	28.94	0.54	1.8
Iba-miR2	miR168	cme-miR168	16.04	20.98	33.33	29.62	0.54	1.58
Iba-miR3	miR2911	han-miR2911	10.30	34.02	41.23	14.43	0.71	1.21
Iba-miR4	miR168b	zma-miR168b-3p	15.15	30.30	32.32	22.22	0.68	1.06
Iba-miR5	miR168a	cca-miR168a	16.86	21.68	32.53	28.91	0.58	1.5
Iba-miR6	miR168a	aly-miR168a-3p	15.84	29.70	31.68	22.72	0.69	1.06
Iba-miR7	miR156c	ahy-miR156c	19.31	26.89	33.79	20	0.96	1.25
Iba-miR8	miR156j	cme-miR156j	19.17	26.71	34.24	19.86	0.96	1.28

Phylogenetic analysis of sweet potato miRNAs

The precursor sequences of the newly predicted miRNAs along with other members of the same family were aligned in ClustalW. The results from multiple sequence alignment showed high degree of sequence similarity among themselves. For instance, the precursor sequence of pre-miRNA Iba-miR1, Iba-miR2, and Iba-miR5 belonging to family miR168 share more than 90% sequence similarity with the other members of miR168. Based on the multiple sequence alignment consisting of newly identified pre-miRNAs of sweet potato and closely related members from the same family, the evolutionary relationship was established using the MEGA software package. As evident from the phylogenetic tree, the rate of evolution in miRNAs within sweet potato varies quite considerably and differs from other species. The diverse miRNA members within the same family are often distantly related (evident from the maximum likelihood tree, Figure 2). The phylogenetic analysis of sweet potato miRNAs with a bootstrap of 1000 revealed that the phylogram contained 3 major clades (Fig. 2), which represent similarities and divergence of miRNAs within the members of the same family. As evident from Figure 2, Iba-miR1, Iba-miR2, and Iba-miR5, along with the other members from miR168 family, formed the largest cluster (cluster-I). Although Iba-miR4 and Iba-miR6 belong to same miR168 family, they are clustered with miR168 members from Zea mays, Medicago truncatula, Populus trichocarpa, and Arabidopsis forming the second cluster (cluster-II). But Iba-miR3 along with members of miR2911 and Iba-miR4, Iba-miR6 with members of miR168 formed the third cluster (cluster-III), which was further subdivided into 2 sub-clusters (cluster IIIa and sub-cluster IIIb). The sub-cluster IIIa is formed of the members of Iba-miR3 along with its closest relative of family miR2911 whereas family members miR168 along with Iba-miR4 and Iba-miR6 formed sub-cluster IIIb.

Figure 2. Phylogeny analysis of newly identified pre miRNAs sequences along with their closely related miRNA families using Maximum Likelihood method. The precursor sequences representing the newly identified miRNAs of sweet potato are highlighted with a ∆.

Potential target genes for newly predicted miRNAs

The psRNATarget server was employed against the Arabidopsis thaliana DFCI Gene Index (AGI) and sweet potato ESTs to identify and understand the biological functions of the putative targets of potential miRNAs in sweet potato. While searching for potential candidate genes in psRNATarget server, the expectation value was adjusted to 0.5 to minimise false positive prediction. Based on their perfect or nearly perfect complementarity with their target sequences in Arabidopsis transcriptome, a sum of 42 potential targets genes were identified for the 3 predicted miRNA families in sweet potato as shown in Table 3. The potential sweet potato miRNA targets belong to a great many gene families that play various roles during physiological processes. Most of the newly identified miRNAs target more than one regulatory gene. Only one miRNA (Iba-miR1) belonging to family miR168b didn’t show any complementarity with the Arabidopsis model plant. In addition, most of the targets predicted in this study are involved in regulation of plant growth, development, signal transduction, post-transcriptional gene silencing, protection of plants against fungal pathogens, and in environmental stress conditions. The miRNA family miR156 showed the highest 26 independent target genes followed by family miR168 and miR2919 with 16 and 3 targets, respectively. A total of 22 predicted miRNA targets in Ipomoea batatas are transcription factors, including squamosa promoter-binding protein, HXXXD-type acyl-transferase, acyl carrier protein (ACP), RPM1-interacting protein 4 (RIN4), WPP domain-interacting protein 3, GATA, and DNA-damage-repair/toleration protein (DRT112). Most of the transcription factors targeted by sweet potato miRNA are involved in plant growth, development, and defense. Furthermore, sweet potato miRNAs not only targeted transcription factors but also various genes involved in various important biological processes including signal transduction, protein ubiquitination, metabolism, detoxification, and stress response. The functions of newly identified miRNA targeting genes of sweet potato are represented in Figure 3.

Table 3. Lists of the potential targets genes and their predicted functions for 5 newly identified miRNAs in sweet potato (Ipomoea batatas L.).

Ipomoea batatas miRNA	Target gene accession	Target protein	Function	GO Term annotation
								Molecular Function	Biological process
Iba-miR2	NP031662	Glutathione transferase	Stress response	GO:0006950 response to stress	Glutathione transferase activity
	TC366556	Glycosyl transferase family 8 protein-like	Transferring glycosyl groups	GO:0016740 transferase activity	Transferase activity
Iba-miR3	TC368699	Cyclin-dependent kinase D-3	Serine/threonine-protein kinase	GO:0004672 protein kinase activity GO:0005524 ATP binding GO:0006468 protein phosphorylation	Cyclin-dependent protein kinase activity
	DR751623	GATA transcription factor 24	Transcription factor	GO:0006351 transcription	Transcription regulation
	EL299566	DNA-damage-repair/toleration protein DRT112	Electron Transport Chain	GO:0009055 electron carrier activity GO:0000271 polysaccharide biosynthetic process	Electron transport Pathway
Iba-miR4	TC388327	Calmodulin-like protein 5	Calcium ion binding	GO:0005509 calcium ion binding	Potential calcium sensor
	TC359023	Ubiquitin activating enzyme 2	Protein ubiquitination	Not assigned	Ubiquitination pathway
Iba-miR5	NP031662	Glutathione S-transferase	Detoxification and Stress response	GO:0006950 response to stress GO:0009636 response to toxic substance	Glutathione transferase activity
	TC358800	Protein argonaute 1	RNA-mediated post-transcriptional gene silencing	GO:0003676 nucleic acid binding	Plant defense
	TC362362	Histone-lysine N-methyltransferase	histone-lysine N-methyltransferase activity	GO:0005515 protein binding	Methyltransferase
Iba-miR6	TC375720	RPM1-interacting protein 4 (RIN4) family protein	defense protein-related	Not known	Defense response
	NP227453	WPP domain-interacting protein 3	Plant development	GO:0006913 nucleocytoplasmic transport GO:0006997 nucleus organization GO:0048527 lateral root development	Plant growth and development
	EG490326	Auxin efflux carrier component 4	Auxin signaling	GO:0055085 transmembrane transport	Auxin mediated signaling pathway
	TC359023	Ubiquitin-activating enzyme E1 2	Protein modification and protein ubiquitination	GO:0006464 cellular protein modification process GO:0008641 small protein activating enzyme activity	Ubiquitination conjugation pathway
	TC388327	Calmodulin-like protein 5	Calcium ion binding	GO:0005509 calcium ion binding	Potential calcium sensor
	TC366700	GPI-anchored protein	Fertilization	Not assigned	Signaling pathway required for fertilization.
Iba-miR7	TC359888 TC383307 TC370310 TC383824 TC385243 TC391425 TC368286 TC382474 TC368029 TC370663 TC367576 TC388074	Squamosa promoter-binding-like protein 2	Transcription factor	GO:0003677 DNA binding	Plant growth and Development
Iba-miR8	TC359888, TC383307, TC370310, TC382474, TC370663, TC367576, TC388074, NP1659796, TC365809	Squamosa promoter-binding-like protein	Transcription factor	GO:0003677 DNA binding	Plant growth and Development
	NP1661966	HXXXD-type acyl-transferase-like protein	Transferring acyl groups other than amino-acyl groups	GO:0016740 transferase activity	Transferase activity
	NP174084	Hypersensitivity related-like protein	Transferring acyl groups other than amino-acyl groups	GO:0016747 transferase activity	Transferase activity
	NP173992	Squamosa promoter-binding-like protein 6	Transcription factor	GO:0003677 DNA binding	Transcription Transcription regulation
	TC364023	Protein kinase domain-containing protein	Kinase Serine/threonine-protein kinase	GO:0004672 protein kinase activity GO:0005524 ATP binding GO:0006468 protein phosphorylation	ATP binding
	TC369351	Acyl carrier protein (ACP) gene A2	Salicylic acid (SA) signaling pathway	GO:0006631 fatty acid metabolic process GO:0045300 acyl-[acyl-carrier-protein] desaturase activity GO:0055114 oxidation-reduction process	Fatty acid biosynthesis Fatty acid metabolism Lipid biosynthesis Lipid metabolism Plant defense

Figure 3. Statistics showing the functions of newly identified miRNA target genes of Ipomoea batatas L.

In addition, the ESTs targeted by miRNA identified in sweet potato were annotated using NCBI BlastX program to compare their respective functions. In most cases the annotated protein targets were transcription factors involved in signal transduction and different metabolic pathways. Among the annotated protein targets of sweet potato miRNA, aspartic protease in guard cell 1-like protein, secologanin synthase-like, diguanylate phosphodiesterase metal dependent hydrolase domain protein, and methionine synthase family protein are the most abundant proteins (data not shown). Finally an attempt was made to find common miRNA target genes from Arabidopsis and ESTs of sweet potato and only one miRNA, i.e., miR-Iba7, which targets the squamosa promoter-binding-like protein was found to be common in both the cases.

Functional annotation and pathway analysis of miRNA target genes

GO term analysis was performed on the predicted miRNA target gene set to better understand the probable function of newly identified miRNA in sweet potato, and the results are presented in Table 3. Functional annotation using GO term analysis is an important method in discovering the regulatory network of miRNA as it involves 3 components, viz. molecular function, biological process and cellular component. We have identified that 42 target genes are involved in 39 diverse molecular functions, of which 12 are involved in 28 biological processes, and 9 are involved in 24 cellular component functions. Furthermore the biological process identified through GO term annotation revealed that 3 miRNA families may be involved in 23 different biological processes including defense response, protein ubiquitination, signal transduction, stress response, lipid metabolism, plant growth, and development as shown in Table 3. Pathway analysis of the miRNA target genes in sweet potato was performed by KEGG, where a total of 26 metabolism networks were found to be targeted; including auxin mediated signaling pathway, ubiquitination conjugation pathway, fatty acid biosynthesis pathway, lipid biosynthesis pathway, electron transport pathway, and ethylene signaling pathway. Many miRNA families were found to be targeting the same pathway in sweet potato. The workflow of the identification and characterization of potential miRNAs, and target genes in sweet potato (Ipomoea batatas L.) using ESTs has been shown (Fig. 4).

Figure 4. Workflow of the identification and characterization of potential miRNAs, and target genes in sweet potato (Ipomoea batatas L.) using ESTs.

Discussion

In this present work, we performed a comparative genomics study for discovery of conserved miRNAs in sweet potato, an economically important plant belonging to family Convolvulaceae using known ESTs. Out of 23 406 ESTs in sweet potato, only 8 conserved miRNA candidates were identified that represents 3 different families of miRNAs, viz. miR168, miR2911, and miR156. More than 50% (5 out of 8) predicted sweet potato conserved miRNAs, belonged to the miR168 family whereas 25% (2 out of 8) belonged to the miR156 family. In the post-transcriptional gene silencing pathway, miR168 plays a pivotal role in controlling their own expression as well as those of other by targeting specific proteins.³² In Arabidopsis, miR168 regulates the function of all miRNAs by targeting AGO1 (ARGONAUTE1) which represent the catalytic subunit of the RNA-Induced silencing complex (accountable for post-transcriptional gene silencing expression).³³ The miR156 family, which is comprised of 2 members, is the most predominant miRNA family in Arabidopsis and is involved in various growth and developmental processes.³⁴ The number of predicted miRNAs in sweet potato (0.068% of ESTs) was much higher than that of previously reported miRNAs in different plant species, i.e., purple false brome (0.05%), soybean (0.0175%), switch grass (0.0277%), and in other plant species (0.010%).^13,14,22,35 In sweet potato, the newly identified pre-miRNA sequences range from 76 nt to 146 nt in length with an average value of 103 nt. This result is consistent with the previous reports in various plant species.^{13,14,16,22,36-40} In addition, the average GC content (60.58%) of the pre-miRNAs is within the acceptable range.¹⁴ Mature miRNAs prefer Uracil (U) be the first nucleotide.^35,38,39 Our finding is in agreement with these previous results, in that 50% of predSicted miRNA sequences (4 of the 8 mature miRNAs) have Uracil at the first position. Furthermore, majority of the potential sweet potato mature miRNAs (75%) have lengths of 20 and 21nt, which is in agreement with the previous reports on conserved miRNAs predicted in soybean, maize, switch grass, and Chinese cabbage.^14,22,26,41

The stability of a RNA secondary structure generally depends upon the minimal folding free energy (MFE). Usually, the lower the MFE, the more stable the RNA molecule. The precursor miRNA sequences have shown a lower MFE (−ΔG) as compared with other types of RNAs.⁴⁰ In sweet potato, the average MFE of the precursor sequences was 47.025 kcal/mol. It is also an issue for determining the stability of a RNA using MFE because different RNA strands contains a different number of nucleotides. So as to ensure the sensitivity and correctness in measuring the stability of RNA strands, the adjusted minimal folding free energy (AMFE) strategy was employed. Zhang et al.¹³ proposed that MFEI value of ~0.85 of the precursor miRNA can be used as a criterion to distinguish miRNA from other RNA types. In sweet potato, the newly identified miRNA possess MFEI value that ranges between 0.77 and 0.95, with an average of 0.86, which signifies that the newly identified precursors in sweet potato are likely to be true miRNA. In plants, the miRNAs are transcribed from either the sense or the anti-sense strand.⁴² Although it was first reported in animals, recent studies have confirmed this to be true in various plants including B. rapa, B. oleracea, and potato also.^43-45 In our study, however, miRNA pairs were found to be transcribing from sense strand only.

Earlier studies have shown that miRNAs are found as clusters in the genome of many plants, signifying the fact that they are transcribed from a single transcriptional unit, which in turn produce multiple mature miRNAs.⁴⁶ Our study also showed that singleton EST (gi-76295387) contains a cluster of 2 miRNA representing the miR168 family. In plants, miRNA clusters are reported in few families including miR169, miR172, miR156, miR399, miR395, and miR1219. Furthermore, these miRNA clusters are reported be conserved in various plant species.^14,47,48

The functional importance of the predicted micro-RNAs can be better understood by gaining information about their targets. Since most of the plant miRNAs shows perfect or near-perfect complementarity with their targets,⁴⁹ the identification of potential miRNA targets can be achieved by aligning the newly identified miRNA sequences with those of the target mRNAs. In sweet potato, 45 potential targets for 3 newly identified miRNA families were predicted. Earlier studies have shown that in plants, miRNAs mostly target transcription factors associated with growth and development.^22,47 The SQUAMOSA promoter-binding protein-like (SPL) gene family, which is the target of Iba-miR7, encode plant-specific transcription factors, which plays vital role in various biological processes including flower and fruit development, plant phase transition, gibberellins signaling, response to copper and fungal toxins, and sporogenesis. Mostly the SPL genes are post-transcriptionally regulated by miR156 in Arabidopsis, where a gene called AtSPL9 positively regulates the expression of another microRNA miR172. This regulatory pathway miRNA-SPL-miRNA (miR156-AtSPL9-miR172) plays a significant role during juvenile to adult leaf development, and the miR156-SPLs feedback interaction persists all through the Arabidopsis development and seems to be conserved in other plants.⁵⁰ Our results from the present study showed that the SPL proteins are targeted by miR156 in Ipomoea batatas. Earlier studies conducted by Kim et al.⁵¹ reported that the SPL proteins were targeted by miR156 in model plants including Arabidopsis and rice. In addition, GATA transcription factor 24 was targeted by sweet potato miR2911, which has been shown to regulate light response and to be related to the cold stress in Arabidopsis.⁵² The miR168a of sweet potato targets calmodulin-like protein 5 which is involved in the calcium-dependent regulation during plant response to endogenous and exogenous stimuli. The calmodulin-binding proteins have already been shown to be targeted by miRNAs in Capsicum annuum. The miR168a targets the stress modulation and detoxifying protein Glutathione S-transferase, which was reported to be downregulated in wheat leaves affected by drought.^53,54 In addition, miR168a also targets auxin efflux carrier component 4, which plays an important role during seed development, especially in auxin signaling. However, the sweet potato miR168b is not predicted to target any gene from Arabidopsis Gene Index by the psRNATarget server.

Analysis from GO term annotation showed that newly identified miRNAs of sweet potato targets important transcription factors involved in the regulation of development and growth in plants. Furthermore, sweet potato miRNAs target genes involved in metabolism (acyl carrier protein gene A2, Cyclin-dependent kinase D-3), signal transduction (auxin efflux carrier component 4), plant defense (endochitinase, RPM1-interacting protein 4 RIN4, protein argonaute1) and stress response (Glutathione S-transferase). In plants miRNAs act as negative regulator by either mediating the cleavage of target mRNAs or by repressing their translation. Cleavage of the target mRNA appears to be the vital and major form of gene regulation by miRNAs in plants.⁵⁵ But in our case results of psRNATarget showed that 77% of miRNA targets are repressed by translation inhibition. More such studies are necessary to validate the newly identified miRNA targets in sweet potato. To sum up, results from the present study indicates that few new sweet potato miRNAs could be identified through the comparative genomics based approach, which provides a powerful tool in discovery of potential miRNAs in plants devoid of their genome sequences.

The identification of miRNAs and their target genes provides tremendous potential in exploration of their possible role in plant growth and development. This in turn can contribute in the development of novel transgenic technology to improve the crop yield potential, which will help to feed the ever-growing population. Moreover, validation of the predicted miRNA in sweet potato is under way in our laboratory to confirm the accuracy of the predicted miRNAs through bioinformatics approach. The experimental approach to understand expression profile of the newly identified miRNA in sweet potato may provide a new dimension of regulatory network of miRNAs during stress.

Materials and Methods

Sequence database and reference miRNA data set

A total of 5939 known plant miRNAs were downloaded from the miRBase⁵⁶ repository (http://www.mirbase.org/cgi-bin/browse.pl) for conserved miRNA search in Ipomoea batatas. The duplicates in the downloaded reference miRNA set were removed by performing multiple sequence alignment using ClustalW⁵⁷ to avoid redundancy. Finally only 2327 unique (non-redundant) mature miRNA sequences were considered for conserved miRNA prediction. Publicly available EST sequences of sweet potato were downloaded from dbEST of NCBI (23 406 as of March 1, 2013). These EST sequences were then used to predict the miRNAs target gene candidates.

Computational resources

The widely used sequence assembly program CAP3 software (http://seq.cs.iastate.edu/cap3.html) incorporated in our in-house pipeline ESMP⁵⁸ was used to assemble the primitive EST sequences. The alignment tool BLAST⁵⁹ version 2.2.27 (September 2012) was used to identify the potentially conserved miRNAs and was downloaded from the NCBI website (ftp://ftp.ncbi.nlm.nih.gov/blast/executables/blast+/LATEST/). Mfold version 3.6 was obtained from (http://mfold.rna.albany.edu/?q=mfold/download-mfold/).⁶⁰ The target genes for the putative miRNAs of sweet potato were identified by using psRNA Target server (http://plantgrn.noble.org/psRNATarget/).⁶¹ In addition to psRNA Target server, the UEA sRNA toolkit was also employed for prediction of the sweet potato miRNA target genes by incorporating Arabidopsis gene index database and ESTs of sweet potato.

Identification of the conversed miRNAs in sweet potato ESTs

The conserved miRNAs in I. batatas were identified using the comparative genome-based approach. First of all the redundancies in the EST sequences were removed by performing sequence assembly using locally installed CAP3 program with default parameters. Then the contigs and singletons generated after assembly were used for BLAST search against mature plant miRNAs. The non-redundant 3227 mature miRNA of Viridiplantae were taken as reference and searched against the assembled ESTs for obtaining the potential miRNAs by using locally installed BLAST version 2.27 by adjusting the parameter settings as follows: an expect value cut-off of 10; the window size 7; a low-complexity sequence filter; number of descriptions and alignments were 1000; and automatically adjusted parameters for short input sequences to improve the veracity of outputs. All BLAST results were saved and used for further analysis. The contigs or singletons which closely matched (n/n, n-1/n, n-2/n nucleotide matches, where n represents the length of the known miRNA of Viridiplantae) with the reference miRNA were chosen manually. These candidate miRNA sequences which closely matched (no more than 2 mismatches) with the previously known mature miRNAs were used to search against NR protein database using BLASTX, in order to remove the protein-coding sequences. To select non-coding singleton or contig, we considered the E value less than e-5 along with the identity percent less than 25. Finally after removal of protein coding sequences from the candidate miRNAs, the remaining precursor sequences of potential miRNA homologs were assessed for secondary structure using the Zuker folding algorithm by MFOLD software. The following parameters were used for secondary structure prediction in MFOLD: linear RNA sequence; folding temperature fixed at 37°C; ionic conditions of 1M NaCl without divalent ions; percent sub-optimality number of 5; maximum interior/bulge loop size of 30; energy dot plot was turned on and the other parameters were set as default.

Finally, potential miRNAs were identified based on the following criteria: (1) The Substitutions between contig/singleton sequences and reference miRNA sequences not being greater than 2; (2) The minimum length of the pre-miRNA was not less than 45 nt; (3) The Pre-miRNA folded into the perfect stem-loop hairpin secondary structure; (4) A maximum mismatch of 6 between the miRNA/miRNA* duplex; (5) The miRNA-miRNA* duplex did not have loops; (6) The negative MFE (Minimal folding free energy) had the lower value and MFEI (Minimal folding free energy index) values of the predicted secondary structures were ~0.85 to distinguish miRNA from other types of non-coding RNAs. The MFE denotes the negative folding free energy (∆G) of the predicted secondary structure representing the pre-miRNAs whereas MFEI was calculated by employing the following equation: MEFI = [(MFE/length of the precursor miRNA sequence) × 100]/ (G + C) %; (7) The miRNA-miRNA* duplex did not contain any gap; (8) No more than 2 consecutive mismatches showed in the miRNA-miRNA* duplex. Finally, possible false sequences of pre-miRNAs were checked via manual inspection.

Prediction of potential targets of sweet potato miRNAs

Complementarity and target-site accessibility are the 2 key factors in the plant regulatory small RNA (miRNA) target recognition mechanism.^62-64 It has also been seen that most of the known plant miRNAs binds to the protein coding region of their mRNA targets with high degree sequence complementarity later degrade the target mRNA by a mechanism similar to RNA interface (RNAi).^65-67 To understand the biological functions of the newly identified sweet potato miRNAs, we have searched for putative target genes using the psRNATarget program against the Arabidopsis thaliana DFCI Gene Index (AGI) Release 15.38 and ESTs of sweet potato with the following parameters: maximum exception of 0.5 (for lower false positive prediction), length of complementarity score: 20, target accessibility-allowed maximum energy to un-pair the target site (UPE): 25, flanking length around the target accessibility analysis: 17 bp upstream and 13 bp in downstream and range of central mismatch leading to translation inhibition: 9–11 nt. Further, the following criteria were set for identification of target genes: range of central mismatch for translation inhibition: 9–11 nt, a maximum exception value of 0.5, maximum mismatch at complementary site ≤ 3 without any gaps and the maximum target sites of 2. In addition to psRNA target server, the plant target prediction tool available at UEA srNA Tool Kit was used for target prediction by following the guidelines of Allen et al.⁴⁹ and Schwab et al.⁶⁸

Phylogenetic Analysis of newly identified miRNAs

Standalone BLAST tool was employed for homolog search of the predicted miRNAs against all the known miRNAs families by allowing a maximum mismatch of 3 and e-value of < 0.001. The precursor sequences of the homologous miRNAs were identified and collected from miRBase. The collected precursor miRNAs sequence along with the precursors of newly identified miRNAs were aligned in ClustalW and subsequently used for phylogenetic analysis in MEGA version 5.2.⁶⁹ Then the maximum likelihood tree was constructed based on Jukes-Cantor model with 1000 bootstrap values to illustrate the evolutionary relationships among all the miRNA members of the family.⁷⁰

Functional annotation and pathway analysis of miRNA target genes

The function of the miRNA target genes were predicted by performing BLASTX search against non-redundant database of NCBI with an e-value of 1e-30. Similarly the functional assignments of the target genes were annotated using KOGnitor tool (http://www.ncbi.nlm.nih.gov/COG/grace/kognitor.html) and Gene ontology (GO) (http://www.geneontology.org/) database assigned biological process and cellular components to the miRNA target genes. Finally the metabolic pathways and their networks regulated by the potential miRNAs were searched through KEGG (Kyoto Encyclopedia of genes and Genomes) (http://www.genome.jp/kegg/pathway.html).

10.4161/psb.26543