Sculpting the maturation, softening and ethylene pathway: The influences of microRNAs on tomato fruits

High-throughput sequencing of short RNAs of three fruit ripening stages

Two distinguishing features of the small RNAs libraries are the populations and distributions. The composition of small RNAs often reflects roles of different categories of small RNAs in a particular tissue or species or associated biogenetic machines. Since most of the small RNAs with known functions are 20-24 nt long, we only investigated the distribution of small RNAs of 18-25 nt in the three libraries (Figure 1; Figure 2). In the three ripening stages, the 24-nt size class was the overall most abundant class of sRNAs (Figure 2), which was consistent with that of Arabidopsis [36] and rice [37], but different from that of wheat, Chinese yew and grapevine [38–40]. The higher percentage of 24-nt small RNAs in tomato fruit may reflect the complexity of the tomato genome because 24-nt siRNAs are known as predominant heterochromatin-associated sRNAs [41]. The 21-24nt sRNAs exhibited a peak at breaker stage, and showed a sharp reduction at the red ripening stage, suggesting potential roles in ripening process.

Identification of known and novel miRNAs in tomato fruit

Conserved miRNAs were found in many plant species and have important functions in plant development and stress response [8]. To identify the conserved miRNAs in tomato, sRNA sequences obtained by deep sequencing were compared with the currently known mature plant miRNAs in miRBase [42]. After Blastn searches and further sequence analysis, a total of 103 conserved miRNAs (Figure 3), belonging to 24 miRNA families were identified and all of which can be consulted in two important database (http://www.mirbase.org/; http://ted.bti.cornell.edu/cgi-bin/TFGD/sRNA/miRNA.cgi).

In addition to the conserved miRNAs, there are known miRNAs that are not-conserved, but are found in only one or a few plant species such as Arabidopsis or Physcomitrella patens [8, 23]. Ten members belong to 9 non-conserved miRNAs (miR158, miR161, miR173, miR393, miR398, miR403, miR414, miR858, miR894) were present in our data sets at a low abundance, with the exception of miR403 that were represented by a relative high number (Table 2). All of the non-conserved miRNAs were validated by the construction of the small RNA cDNA library method (Figure 4) [43].

MiRNA family	References	Relative number of reads found in tomato
		Mature green	Breaker stage	Red ripe
Conserved miRNAs
miR156	Zhang J et al, 2008; Yin Z et al, 2008; Moxon S et al, 2008.	115.99	31.95	15.14
miR157	Zhang J et al, 2008; Yin Z et al, 2008; Moxon S et al, 2008.	349.64	154.19	137.34
miR159	Pilcher et al, 2007; Zhang J et al, 2008; Moxon S et al, 2008.	59.89	48.06	82.55
miR160	Pilcher et al, 2007; Yin Z et al, 2008; Moxon S et al, 2008.	46.06	30.28	1.44
miR162	Pilcher et al l, 2007; Yin Z et al, 2008; Moxon S et al, 2008.	878.08	929.27	167.99
miR164	Pilcher et al, 2007; Moxon S et al, 2008	996.58	496.45	188.53
miR165	Moxon S et al, 2008	38.94	24.73	3.60
miR166	Pilcher et al, 2007; Moxon S et al, 2008	2750.64	2154.69	523.78
miR167	Zhang J et al, 2008; Yin Z et al, 2008; Moxon S et al, 2008.	265.06	209.19	8.29
miR168	Pilcher et al, 2007; Yin Z et al, 2008; Moxon S et al, 2008.	4967.40	3130.92	62.36
miR169	Zhang J et al, 2008; Yin Z et al, 2008; Moxon S et al, 2008.	0.84	0	0
miR171	Pilcher et al, 2007; Zhang J et al, 2008; Moxon S et al, 2008.	19.68	53.61	2.16
miR172	Zhang J et al, 2008; Yin Z et al, 2008; Moxon S et al, 2008; Itaya et al, 2008.	2993.09	2733.37	798.11
miR319	Zhang J et al, 2008; Yin Z et al, 2008; Moxon S et al, 2008.	0.42	0	0
miR390	Moxon S et al, 2008; Itaya et al, 2008.	97.98	46.67	2.16
miR394	Moxon S et al, 2008.	6.70	4.44	2.16
miR395	Zhang J G et al 2008; Moxon S et al, 2008.	0.41	0	0
miR396	Moxon S et al, 2008.	1300.58	2175.81	334.89
miR397	Moxon S et al, 2008.	0.42	0	0
miR399	Zhang J et al, 2008; Yin Z et al, 2008; Moxon S et al, 2008.	44.38	17.50	1.08
miR408	Pilcher et al, 2007, Moxon S et al, 2008	0.42	0.28	0
miR472	Itaya et al, 2008; Moxon S et al, 2008.	19.26	12.78	0.36
miR482	Pilcher et al, 2007.	1.67	2.51	0
miR827	http://ted.bti.cornell.edu/cgi-bin/TFGD/sRNA/miRNA.cgi	2.09	6.39	14.78
Non-conserved miRNAs
miR158		2.93	1.67	0
miR161		0.42	0	0
miR173		0.41	0	0
miR393		0.42	0	0
miR398		0.40	0	0
miR403		210.20	138.91	27.04
miR414		0.42	0	0
miR828	Moxon S et al, 2008.	0	0	0
miR858	Moxon S et al, 2008.	5.19	12.31	1.22
miR894	Moxon S et al, 2008.	10.05	5.00	0.36

A number of criterions are used for evaluating whether a small RNA is a genuine miRNA, such as formation of a stable hairpin structure, starting with a 5' uridine, lower minimal free energies (MFEs) for hairpin structure of its precursors, and detection of miRNA*s [44, 45]. By using these rules, we identified 11 novel miRNAs belonging to 7 families (named miRZ1-miRZ7, Figure 5, Figure 6, additional file 1, 2). Except for miRZ6, most of the novel miRNAs were in low copies. The possible reason is that the encoding genes are not broadly conserved and expressed at much lower levels [46].

Prediction of the targets of identified miRNAs in tomato fruit

Target prediction for miRNAs is straightforward because it is assumed that most of them match their targets with almost perfect complementarity [9, 47]. The putative target genes for all identified conserved and non-conserved miRNAs were searched by using the web-based computer psRNA Target Server (http://plantgrn.noble.org/psRNATarget/) which identifies putative targets regulated at post-transcriptional or translational levels.

Most conserved miRNA targets that are conserved across several plant species, including Arabidopsis [48], rice [49], grape [50], poplar [51] and wheat [52]. And most miRNA families have multiple target sites, suggesting that these miRNAs are functionally divergent. In our study, 152 targets of conserved and non-conserved miRNAs were predicted (Figure 7). The majority of such targets are various transcriptional factors including SBP (miR156), MYB (miR159, miR319, miR172), NAM (miR164), and MADS-Box (miR396) that regulate plant development [53] or phytohormone signal transduction [54]. Other conserved miRNAs targets include F-box protein (miR394, miR414), ATP sulfurylase (miR395), Pectate Lyase (miR482), endo-1, 4-beta- glucanase (miR396), Laccase (miR397), all of which are involved in regulation of metabolic processes. Several targets of the miRNAs are AGO protein (miR168, miR403) which regulate their own biosynthesis pathway. In addition, targets of the conserved miRNAs include disease resistance proteins (miR472, miR482) which are related to pathogen resistance [33, 34]. Furthermore, ACS and EIN2, which are involved in ethylene biosynthesis and signal transduction [17], are putative targets of miR159 and miR828, respectively.

The targets of the novel miRNAs identified in our library were predicted, most of which were unknown (Additional file 1). MiRZ1 was predicted to be related to disease response, MiRZ6 was predicted to be involved in virus infection response. Interestingly, a target of miRZ7 is beta-galactosidase which is an important enzyme affects fruit softening. Another member's target of miRZ7 family is starch synthase which participates in starch biosynthesis, including transient starch [55].

Expression profiles of the miRNAs involved in tomato fruit ripening

It has been shown that high throughput sequencing provides an alternative way to gain the genome-wide transcript profiles of miRNAs abundance [56] and allows us to determine the abundance of various miRNA families and even to distinguish different members of a given family of one organism. To elucidate the potential roles of miRNAs in tomato fruit ripening process, we profiled the expression levels of known and novel miRNAs. We computed the normalised counts of known miRNA sequences, plotted them across the three ripening stages and validated some results by stem-loop RT-PCR (Figure 8).

The abundance of most known miRNAs decreased during fruit ripening, and some were found only in the mature green stage (Table 2). A few known miRNAs were expressed at high levels in fruit, such as miR157, miR162, miR164, miR166, miR168, miR172, miR396. On the other hand, other miRNAs, such as miR161, miR173, miR393, miR397, miR398 and miR414, were expressed at relatively lower levels and can only be detected in mature green stage, Particularly, the expression levels of miR159 decreased at the breaker stage and increased at red ripening stage compared with that of the mature green stage which reveals its probable specific roles in fruit ripening or ethylene pathway, as one of the predicted targets is 1-aminocyclopropane -1-carboxylate synthase which is a crucial enzyme in ethylene biosynthesis [17]. Expression of another miRNA, miR396, increased from mature green to breaker stage, then dropped sharply in the red stage. Interestingly, targets of miR396 include MADS-Box protein and endo-1, 4-beta-glucanase, both of which are involved in fruit ripening and softening process [57]. Moreover, 3 miRNAs (miR394, miR414, and miR482) predicted to have targets related to fruit ripening and softening and ethylene response were validated (Figure 8).

Colorless non-ripening (CNR), a member of the squamosa-promoter binding protein (SBP) family that was shown to be involved in fruit ripening [58], is targeted by miR156 [31, 39]. The expression levels of miR156 also decreased during fruit ripening (Table 2 and Figure 8). Two miRNAs, miR828 and miR1917, whose targets are ethylene-insensitive 2 (EIN2) and serine/threonine protein kinase (CTR1) which are vital regulators in ethylene signal transduction[17, 59], reported previously [26] and not shown in our sequencing results, showed reduced expression during fruit ripening (Figure 8). A novel miRNA (miRZ7) expression levels were also analyzed which was predicted to relate to fruit softening.

Overwhelming the majority of the stem-loop RT-PCR results was in agreement with sequencing results. Only two miRNAs (miR414, miRZ7) were exception which could be detected by stem-loop RT-PCR but were not detected or barely detectable by sequencing. This could be due to cross amplification of other highly homologous small RNA members or biases occurred during library generation or sequencing in some samples.

Expression profiles of the miRNAs responses to exogenous ethylene in tomato fruit

Ethylene is a regulator of various physiological and morphological responses, including seedling development, leaf and flower senescence, induction of fruit ripening, floral sex determination, resistance to pathogen infection, and adaptation to stress conditions [60–63]. Tomato has proven to be a highly successful model system for fruit development and ripening. And the role of ethylene during fruit ripening has been most thoroughly studied in tomato fruit [15, 16]. In order to find out the relationships between ethylene treatment and miRNAs expression, exogenous ethylene and 1-methylcycloproene were used to treat the mature green fruits. The expression levels of a number of known and novel miRNAs were tested by stem-loop RT-PCR. Three miRNAs (miR394, miR414, miR1917) were down regulated; five miRNAs (miR156, miR159, miR396, miR482, miRZ7) were up regulated, while one miRNA (miR828) whose target is EIN2 was not affected, by exogenous ethylene treatment (Figure 9).

High-throughput sequencing of short RNAs in tomato

High-throughput sequencing has been used to study miRNAs at the whole genome level in several model plant species, including Arabidopsis [23], rice [25], and wheat [64]. The successful application of high-throughput sequencing technology to systemically identify plant miRNAs in recent years has greatly advanced our knowledge on the functions of miRNAs in plants.

The composition of small RNAs often reflects roles of different kinds of small RNAs in a specific tissue or species. Several plant species, such as Pinus cordata, Populus balsamifera and grapevine were shown to contain substantially more 21-nt than 24-nt sRNAs [37, 39, 65]. However, we observed an unusually high level of 24-nt sRNAs compared with the 21-nt class (Figure 1; Figure 2), as in several other plant miRNAs studies [23, 36, 37, 66]. The number of 24-nt small RNAs was almost three times that of the 21-nt class, such a high percentage of 24-nt small RNAs may reflect the complexity of the tomato genome because 24-nt siRNAs are known to be involved in heterochromatin modification, especially for genomes with high content of repetitive sequences [67, 68].

Plant MIR genes are possible to arise from gene duplication events which then evolved by random mutations into short, imperfectly paired hairpins [46, 69]. Those non-conserved miRNAs are believed to be evolutionarily recent and generally represented by single copy MIR genes [8]. Recently, many non-conserved miRNAs were reported in several species [36, 37, 64, 66]. Some of these non-conserved miRNAs are not found in other species/families and several non-conserved miRNAs are found in different phylogenetic families. We also found nine non-conserved miRNAs in our database, although most of them were present at a very low level. However, the low number of sequence reads does not rule out that they could be expressed at a high level in specific cells. Few non-conserved miRNAs were expressed at a relatively high level, suggesting that they may represent an intermediate status between the deeply conserved and the less conserved miRNAs.

New species-specific miRNAs are considered to be young miRNAs that have evolved recently, and are often expressed at a lower level than conserved miRNAs, as was reported for Arabidopsis and wheat [23, 36, 38]. This observation is true for many new tomato miRNAs identified here. However, few new miRNAs were expressed at a high level in a tissue-specific manner (miRZ6). In some cases we observed inconsistency between the level of miRNAs identified by Solexa sequencing and experimental verification. It is possible that during library generation or sequencing some bias could occur for certain sequences in some samples.

In contrast to the 454 pyrosequencing sequencing in tomato previously [39], more reads were obtained in each of our three libraries, and miRNAs at different fruit ripening stages were additionally analyzed in our study. In our database, most conserved miRNAs have been detected, including 33 conserved and non-conserved miRNA families, which cover all the published miRNAs in the two databases for tomato miRNAs (http://www.mirbase.org/, http://ted.bti.cornell.edu/cgi-bin/TFGD/sRNA/miRNA.cgi).

Dalmay and colleagues studied the correlation between the short RNAs and fleshy fruit development using deep sequencing technology [70]. However, mostly sRNAs were analyzed, which is different from our miRNAs study. But they reported several miRNAs, and validated one novel miRNA related to the glutamate accumulation which contributes to the tomato fruit taste.

Targets of known and novel miRNAs in tomato fruit

Target prediction is necessary for assessing miRNAs' putative functions. Currently, the most efficient tool available for this is the bioinformatics approach facilitated by the high degree of homology between miRNA and its target sequences in plants [71]. Our analysis reveals that most of the predicted targets in tomato have a conserved function among a variety of plant species [72]. Consistent with previous reports, the largest of these targets in tomato fruit were plant-specific transcription factors, such as AP2, NAC, SBP and the ARF family [53]. The second largest targets encoded a range of different proteins implicated in a variety of metabolic processes, such as ATP sulfurylase, Pectate Lyase, endo-1, 4-beta-glucanase, Laccase. In addition, functions of several targets are largely unknown.

Although many conserved miRNA targets were predicted, only a limited number of miRNA targets were identified experimentally [31, 39]. Several miRNAs slice their targets and can be validated by RACE. Alternatively, other miRNAs inhibit target gene expression through translational arrest, such as miR156 and miR172, which have been shown to regulate their target genes (SBPs and AP2) predominantly by inhibiting their translation [73–76]. Indeed, genetic and biochemical evidence is accumulated that plant miRNA-guided silencing has a widespread translational inhibitory component [2, 77].

The novel tomato miRNAs target different genes with a wide variety of predicted functions. It may be worthy to note that those miRZ6 targets six members of disease resistance protein (CC-NBS-LRR class), which are likely to be involved in host-pathogen interactions. Surprisingly, one target of miRZ7 is beta-galactosidase which is a crucial enzyme for fruit softening, which is beneficial for clarify the cooperating roles in freshly fruit ripening and senescence. The target of another member in miRZ7 family is starch synthase which participates in regulating starch biosynthesis (Additional file 1) [55].

Fruit ripening and softening related miRNAs

Functional analysis has been carried out only for role of a few tomato miRNAs morphological development [28–30]. Conserved and non-conserved miRNAs regulate genes involved in fleshy fruit development were reported previously [39]. A target gene of miR156 belong to squamosa-promoter binding protein (SBP) family called CNR was validated, which plays pivotal roles in fruit ripening [57]. Over-expression of miR156 can cause the fruit red colour slightly lighter than the wild type, which was thought that the expression of miR156 and CNR overlaps only partially and the function of the miRNA is to suppress CNR expression in specific cell types. CNR mRNA is up-regulated in fruit at the breaker stage [57], coincidently, our sequence results showed a negative correlation as expected. Two important targets of miR396 are endo-1, 4-beta-glucanase and MADS-box protein, both of which are important regulators in fruit ripening and softening [78, 79]. Our sequencing results showed that the abundance of miR396 peaked at the breaker stage and decreased sharply in red ripening stage suggesting its possible function in fruit ripening. A target of miR482 is pectate lyase which is an important enzyme in fruit softening [80, 81]. Meanwhile, a novel miRNA, whose target is beta-galactosidase, which is also a key softening related enzyme [82, 83]. The expression levels of the two miRNAs were also coinciding with the assumption including the softening ripening stage (Figure 8).

Intriguingly, an in-depth study of the tomato transcriptome and proteome unraveled the regulation mechanisms during fruit ripening stages. The transcriptional regulation is mostly responsible for global reductions in plastid gene expression prior to or at the onset of fruit formation whereas post-transcriptional events become predominant at the breaker stage [84, 85]. These post-transcriptional events did not lead to significant reductions in mRNA levels, although mRNA stability was reported to be modestly affected for some genes [84]. miRNAs and their targets have been shown to regulate in translational inhibition way in a fairly large fraction [2].

Ethylene biosynthesis and signal transduction related miRNAs in tomato

Recent researchers have witnessed the great progress in the area of identification of interaction between miRNAs pathways and phytohormone responses, which improves our understanding of mechanism of plant development controlled by miRNAs and hormone action to a large extent. One of the most important phytohormone is ethylene which plays vital roles during all stages of the plant life cycle, functioning through seed germination to ripening and various abiotic stress conditions [86]. MiR159 and miR394 were reported to be associated with ethylene in rice [87]. Most targets of miR159 are MYB, and a novel target not related to MYB in tomato recently [30]. Another important target of miR159 is 1-aminocyclopropane-1-carboxylate synthase which plays vital roles in ethylene biosynthesis [17]. As a typical climacteric plant, abundance ethylene was synthesized at the breaker stage approximately, and the expression level of miR159 was suppressed which match our results well. MiR394 and miR414 also target F-box family proteins which are important participants in the signal transduction pathways of different plant hormones [88, 89]. The stability of the ethylene signaling regulators EIN2 and EIN3 are modulated by the F-box proteins ETP1/2and F-box proteins EBF1/2 respectively. Ethylene Insensitive 2 (EIN2), a target of miR828, is the key positive regulator of ethylene signal transduction and other hormones such as abscisic acid, auxin, cytokinin and jasmonate and thus may represent a point of crosstalk between multiple hormone signaling pathways [90–92]. CTR1 is a negative regulator of ethylene response that likely interacts directly with receptor molecules to form a signaling complex [93, 94]. As with the ethylene receptors, all tissues evaluated express CTR1 genes and their mRNAs are differentially accumulated depending on tissue. LeCTR1 induction is associated with tissues at stages of development associated with increased ethylene, including fruit ripening and is a validated target of miR1917 [95]. Exogenous ethylene can influence the miRNAs expression patterns, most of which shown negative correlation to their targets well except for miR828, was not affected by exogenous ethylene treatment (Figure 9). The most likely reason was that miR828 affected this target slightly and we cannot catch the discrepancy.

RNA analysis

Tomato samples from three ripening stages (mature green, breaker, and red-ripening stage) of Solanum lycopersicum 'Ailsa Craig' were used to prepare for the high throughout sequencing. Total RNAs were extracted using TRIzol reagents. For each sample, the 18-30nt small RNAs were ligated with 5'- and 3'- RNA adapter by T4 RNA ligase after being purified by polyacrylamide gel electrophoresis. The RNAs were subsequently transcribed to single-stranded cDNAs using Superscript II reverse transcriptase. Thereafter the cDNAs were used as templates for double-stranded cDNA synthesis by PCR amplification using primers that anneal to adapters. The purified cDNAs were sequenced on an Illumina-Solexa 1 G Genetic Analyzer (BGI).

Bioinformatics analysis

The adapter sequences of the Solexa sequencing results were removed. And sequences larger than 30nt and smaller than 18nt were discarded. All high quality sequences were considered as significant and further analyzed. Small RNA sequences were mapped to tomato genome (ftp://ftp.sgn.cornell.edu/tomato_genome/ITAG_devel_release/). rRNAs, tRNAs, snRNAs and snoRNAs were removed from the matched sequences through BLASTn search using NCBI Genebank database (http://www.ncbi.nlm.nih.gov/blast/Blast.cgi/). Mismatches were not allowed in the above two approaches. The unique sequences left were aligned with known miRNAs from miRBase 10.0 (http://www.mirbase.org/) [42]. The potential novel miRNAs were analyzed using mireap. Parameters were set based on a previous plant miRNAs study [45]. The court number was normalized as transcript per million (TPM). Target predictions were performed using the psRNATarget (http://plantgrn.noble.org/psRNATarget/) [64].

Exogenous ethylene and 1-Methylcycloproene treatment

Solanum Lycopersicum (Ailsa Craig) plants were grown in soil under standard greenhouse conditions. Mature green fruits were harvested and divided into three groups, the first group was treated with 50 μl/L of ethylene for 6 h, the second group was treated with 0.5 μl/L 1- methylcyclopropene (1-MCP) for 24-h, and the third group was used as a control. After treatments, samples were frozen immediately in liquid nitrogen and stored at -80°C until RNA extraction [96].

MicroRNAs Detection

Small RNA samples from above fruit samples were isolated using the miRNA isolation kit (Bioteke) according to the manufacturer's instructions. Small RNA samples were polyadenylated at 37°C for 60 min in a 50-μl reaction volume containing 0.5 μg RNA and 1.5 U poly (A) polymerase (Ambion). The reaction product was diluted to 300 μl. An equal volume of acid-phenol:chloroform was added, mixed and centrifuged. The aqueous phase was carefully removed and transferred to a new tube. Two volume of ethanol and one-tenth volume of sodium acetate (3 M, pH 5.2) were added and mixed thoroughly. The mixture was allowed to precipitate at -20°C for 2 h and then centrifuged at 12000 rpm for 20 min. The supernatant was discarded and the pellet was dissolved with 20 μl RNA-free water. To generate a small RNA cDNA (srcDNA) library, 20 μl of the tailed RNA and 1 μl of RTS primer were mixed in a 26-μl reaction volume, incubated at 65°C for 10 min, and annealed at 4°C for 20 min. Reverse transcription was carried out with reverse-transcriptase (Invitrogen) at 50°C for 60 min. Finally, the reverse transcriptase was inactivated by incubation at 70°C for 15 min. A small RNA-specific primer and a universal reverse primer were used for amplification of individual small RNAs (Additional file 3). The annealing temperature was adjusted according to the T_m of individual small RNAs. After PCR, an aliquot of the PCR products was analyzed on a 2.5% agarose gel [43].

Differential expression analysis of microRNAs

Total RNA was extracted from the samples using TRIzol solution (Trans, Beijing, China) and treated with RNase-free DNase I (Promaga, Beijing, China). First strand cDNA was synthesized using total RNA and reverse transcriptase (Promaga, Beijing, China). Expression levels of mature miRNAs were analyzed by Semi-quantitative RT-PCR stem-loop method [97–100]. A stem-loop containing RT primer with its 5'-end complementary to target miRNA's last 6-nt at 3'-end was designed. Reverse transcription was performed at 16°C for 30 minutes, followed by 60 cycles of pulsed RT at 30°C for 30 seconds, 42°C for 30 seconds and 50°C for 1 second. Semi-quantitative RT-PCR was performed using a forward primer containing the 5' part sequence of miRNA and a universal primer complementary to the stem-loop part of RT primer at 94°C for 2 min, followed by 21 cycles of 94°C for 15 s and 60°C for 1 min. The reaction products (about 60 bp) were analyzed by electrophoresis on a 2.5% agarose gel in 1× TAE. The primers used in this study were listed in the Additional file 4.