Widespread Long Noncoding RNAs as Endogenous Target Mimics for MicroRNAs in Plants^1,[W]

Identification of eTMs for Conserved miRNAs in Arabidopsis and Rice

To systematically screen for eTMs, sequences of all intergenic regions, annotated noncoding genes, and short genes with predicted open reading frames (ORFs) of less than 100 amino acids from the genomes of Arabidopsis and rice were collected and used as the eTM prediction libraries. Twenty miRNAs conserved between Arabidopsis and rice were selected from miRBase (Table I). We developed a method to predict eTMs of these 20 miRNAs in the collected libraries in both species according to the pairing pattern between IPS1 and miR399. A total of 36 potential eTMs were identified for 11 Arabidopsis miRNAs, and 189 eTMs were identified for 19 rice miRNAs (Table I; Supplemental Tables S1 and S2; Supplemental Data S1 and S2).

To investigate the evolutionary conservation status of the eTMs, sequences of the predicted eTM binding sites and surrounding regions for the same miRNA in both Arabidopsis and rice were aligned. The predicted miRNA binding sites were well conserved among eTMs of the same miRNA either within or across species, but sequences outside of the predicted miRNA binding sites varied a lot, showing no constraints on sequence composition (Fig. 1A). Previous studies have shown that the double-stranded RNA duplexes formed by eTMs and their corresponding miRNAs usually have a three-nucleotide bulge at the middle of miRNA binding site within eTM sequences, which is supposed to abolish the target cleavage functions of miRNAs (Ivashuta et al., 2011; Rubio-Somoza et al., 2011). In our prediction, we also required a three-nucleotide bulge on eTMs that located between the 5′ end ninth to 12th positions of miRNAs. The nucleotides forming the bulge region were not conserved among eTMs of the same miRNA, either within each species or between Arabidopsis and rice (Fig. 1). This phenomenon indicated that the conservation of miRNA binding sites among eTMs was constrained by evolution in accordance with the sequences of the corresponding miRNAs.

Expression Analysis of eTMs

Among the predicted eTMs, one was an annotated noncoding gene, six were annotated short ORFs, and the rest were from nonannotated intergenic regions (Table I). To examine whether these predicted eTMs were indeed transcribed, two RNA-Seq data sets including samples from different tissues and developmental stages of Arabidopsis and rice were downloaded and analyzed (Fig. 2, A and B). A total of 18 Arabidopsis eTMs and 84 rice eTMs were detected in at least one sample, of which more than 60% (11 Arabidopsis eTMs and 58 rice eTMs) were expressed in at least two samples (Supplemental Tables S1 and S2). Among the eTMs with RNA-Seq expression data, seven eTMs for three Arabidopsis miRNAs and 10 eTMs for seven rice miRNAs also had corresponding transcripts among the complementary DNA (cDNA), EST, or Plant Genome Database (PlantGDB)-assembled unique transcript databases (Table I; Supplemental Tables S1 and S2).

We next used reverse transcription (RT)-PCR to validate the expression of some eTMs in selected samples (Supplemental Tables S1 and S2). A total of 12 selected ath-eTMs including seven eTMs with RNA-Seq reads and five eTMs without RNA-Seq reads were examined in mixed samples (Supplemental Fig. S1). All seven eTMs with RNA-Seq reads together with three eTMs without RNA-Seq reads were validated to be expressed, and the other two eTMs were not detected either among RNA-Seq reads or by the RT-PCR experiment among the selected samples (Supplemental Fig. S1). We then used semiquantitative RT-PCR to analyze the expression of some eTMs in separate samples from different tissues and developmental stages. Except for ath-eTM171-2, which was mainly detected in seeding roots, other examined eTMs were all expressed in multiple tissues (Fig. 2, C and D). The expression of some eTMs showed great variation among tissues. In addition, the expression patterns of eTMs for some homologous miRNAs were also conserved across species. For example, in both Arabidopsis and rice, the eTMs for miR156 had lower expression in leaves than in other tissues. Among the examined multiple eTMs for the same miRNA in Arabidopsis, the two eTMs for miR160 had well-correlated expression patterns, whereas the expression patterns of eTMs for miR171 were diverged.

eTMs Repress miRNA Function and May Induce miRNA Degradation

To investigate whether the predicted eTMs can indeed function to abolish the binding between corresponding miRNAs and their targets, we overexpressed one eTM for miR160 (ath-eTM160-1) in Arabidopsis. The pairing between ath-eTM160-1 and miR160 contains a three-nucleotide bulge between the 5′ end ninth and 10th positions of miR160 (Fig. 3A). The transgenic plants overexpressing ath-eTM160-1 (OE-ath-eTM160-1) had smaller and serrated leaves compared with the wide-type plants (Fig. 3, B and C). The known target genes of Arabidopsis miR160 are auxin response factor10 (ARF10; AT2G28350), ARF16 (AT4G30080), and ARF17 (AT1G77850; Mallory et al., 2005). Consistent with the expected effects of eTMs, the expression of all three ARF genes were significantly increased in OE-ath-eTM160-1 plants (Fig. 3D). Moreover, the expression level of miR160 was decreased markedly (Fig. 3E), in agreement with the previous report that some eTMs can induce miRNA degradation (Todesco et al., 2010; Ivashuta et al., 2011; Banks et al., 2012; Yan et al., 2012).

The pairing sequence and bulge position of ath-eTM160 were also conserved in rice (Fig. 4A). Similar to ath-eTM160-1, the rice eTMs for miR160 (osa-eTM160-3) also form a central bulge with miR160 in the position of the traditional cleavage site (Fig. 4A). As miR160 is highly conserved during evolution with identical sequence in Arabidopsis and rice, theoretically, eTMs for miR160 in one species should also be functional once it is overexpressed in the other species. Thus, to accelerate the experimental process, we overexpressed rice osa-eTM160-3 in Arabidopsis to study its functions. Transgenic plants overexpressing osa-eTM160-3 (OE-osa-eTM160-3) had severe developmental defects, with dwarf size and serrated leaves (Fig. 4, B and C), as well as accelerated flowering time (data not shown). These phenotypes resembled those of plants overexpressing artificial miR160 target mimics (Todesco et al., 2010; Yan et al., 2012). Correspondingly, the expression of known miR160 target genes was also significantly increased in the osa-eTM160-3 overexpression plants (Fig. 4D). We also found that the mature miR160 level was decreased in the transgenic plants (Fig. 4E).

To examine whether the identified eTMs indeed function via pairing with miRNAs, we introduced four point mutations to osa-eTM160-3 within sequences pairing with miR160 (osa-eTM160-3M; Fig. 4A). Overexpressing osa-eTM160-3M under the control of a 35S promoter in Arabidopsis did not cause observable phenotype or expression changes of either miR160 or its targets (Fig. 4, C and D), suggesting an abolishment of osa-eTM160-3 function. Furthermore, when overexpressing osa-eTM160-3 in plants already with the overexpression of ath-miR160c, no phenotypes or expression increment of miR160 targets was observed (Fig. 5).

In addition to miR160, we also investigated the functions of eTMs for miR166. The pairing between miR166 and its eTM ath-eTM166-1 also had a bulge at the middle of the miRNA sequence (Fig. 6A), and such a pairing pattern was conserved in rice (Supplemental Fig. S2). The transgenic plants overexpressing ath-eTM166-1 (OE-ath-eTM166-1) had spoon-shaped cotyledons and abnormal rosette leaf shapes (Fig. 6B). The known target genes of Arabidopsis miR166 are three homeodomain-Leu zipper III genes, ATHB-9 (AT1G30490), ATHB-14 (AT2G34710), and ATHB-15 (AT1G52150; Kim et al., 2005; Zhou et al., 2007; Zhu et al., 2011). The expression of all three of these genes was significantly increased in OE-ath-eTM166-1 plants (Fig. 6C). Similar phenotypes were also observed in plants overexpressing the rice miR166 eTM osa-eTM166-2 (Supplemental Fig. S2).

We next used a transient agroinfiltration assay to test whether other eTMs were also functional. We constructed expression vectors containing eTMs of four miRNAs (miR156, miR159, miR169, and miR172) separately and overexpressed them in the leaves of Nicotiana benthamiana. The above functionally confirmed ath-eTM160-1 and ath-eTM166-1 were also included as positive controls. The sequences of these tested miRNAs and their target sites were all well conserved between Arabidopsis and N. benthamiana. The previously published small RNA sequencing data showed that all these miRNAs were expressed in the leaves of N. benthamiana with relatively high abundance (Tang et al., 2012). Therefore, overexpression of the exogenous eTMs should repress the functions of the corresponding endogenous miRNAs. As examined at 2 d after agrobacteria infiltration, all the examined eTMs except for ath-eTM169-1 dramatically increased the mRNA abundance of the corresponding miRNA targets in their transiently expressed leaves, suggesting that they indeed inhibited the functions of the corresponding miRNAs (Supplemental Fig. S3).

Prediction of eTMs

A total of 20 conserved miRNAs between Arabidopsis (Arabidopsis thaliana) and rice (Oryza sativa) were collected from miRBase (version 18). Noncoding RNAs, short ORF-encoding genes (with coding potential of less than 100 amino acids), and intergenic sequences in The Arabidopsis Information Resource (TAIR) 10 for Arabidopsis and the Michigan State University Rice Genome Annotation Project (release 7) annotations were collected and used as the eTM prediction libraries. eTMs for the 20 selected miRNAs were predicted using local scripts with the following rules: (1) bulges were only permitted at the 5′ end ninth to 12th positions of miRNA sequence; (2) the bulge in eTMs should be composed of only three nucleotides; (3) perfect nucleotide pairing was required at the 5′ end second to eighth positions of miRNA sequence; and (4) except for the central bulge, the total mismatches and G/U pairs within eTM and miRNA pairing regions should be no more than three. The distance between qualified intergenic eTMs and their upstream/downstream genes should be longer than 200 nucleotides.

Conservation Analysis of eTMs

The miRNA pairing sites along with 50-nucleotide sequences upstream or downstream of the miRNA pairing sites of the predicted eTMs for each miRNA were extracted and aligned. The conservation status of the eTMs was analyzed as follows: among all homologous eTMs for each orthologous miRNA pair in Arabidopsis and rice, the occurrence frequency of the most common nucleotide at each position was calculated as the conservation score. Conservation scores of all eTMs for the same position were plotted by the box-plot package in the R environment. ClustalW and WebLogo were employed to generate the multiple alignment and sequence logos of 13 predicted eTMs of Arabidopsis and rice for miR160.

Expression Analysis of eTMs

Two RNA-Seq data sets from the National Center for Biotechnology Information Gene Expression Omnibus database, SRP007511 for Arabidopsis and SRP008821 for rice, were downloaded and assembled to evaluate the expression of predicted eTMs. The raw RNA-Seq data were mapped to Arabidopsis TAIR 10 and the Michigan State University Rice Genome Annotation Project (release 7), respectively, using TopHat version 2 with default parameters. The reads per kilobase per million mapped reads values of transcripts derived from the eTM corresponding genomic loci were taken as the expression of eTMs. Further expression evidence for eTMs was collected by checking their presence among the sequences of cDNAs, ESTs, and PlantGDB-assembled unique transcripts collected in the PlantGDB.

Plant Materials and Transgenic Plant Construction

Arabidopsis (ecotype Columbia) plants were grown under standard conditions (16 h of light/8 h of dark) at 22°C. The eTMs were amplified by PCR from cDNA using the specific primers shown in Supplemental Table S3. The PCR products of eTMs were placed behind the cauliflower mosaic virus 35S promoter in the pCAMBIA1300 vector. The floral dip method (Clough and Bent, 1998) was used to generate transgenic Arabidopsis plants.

The OE-ath-miR160c (MIR160c) transgenic plants (Wang et al., 2005) were provided by Xiao-Ya Chen.

RT-PCR Analysis of Gene Expression

Total RNA was isolated using Trizol (Invitrogen) from each sample. In the quantitative RT-PCR experiments, 1 μg of total RNAs treated with DNase I (New England Biolabs) was synthesized to cDNA by Moloney murine leukemia virus (New England Biolabs) using poly(dT) oligonucleotides. Each RT-PCR was combined with the RT-minus (without reverse transcriptase) negative control run to confirm that the RNA samples had no DNA contamination. SYBR Green PCR Master Mix (Applied Biosystems) was used in all quantitative RT-PCR experiments. The relative fold expression changes of genes were calculated using the comparative threshold cycle method (Livak and Schmittgen, 2001). Semiquantitative PCR products were amplified for no more than 30 cycles to avoid the reaction reaching plateau stage. Primers used in all RT-PCR experiments are listed in Supplemental Table S4.

Small RNA Northern-Blot Hybridization

Small RNA northern-blot hybridization was carried out following previously reported procedures (Zhao et al., 2012) using 20 μg of total RNA in each experiment. The complementary sequences of miRNAs were labeled by [γ-³²P]ATP and used as probes.

Transient Agroinfiltration Assay in Nicotiana benthamiana

To construct agroinfiltration transient expression vectors, the HindIII-GUS-EcoRΙ fragment of the pBI121 vector was inserted into HindIII/EcoRΙ-digested pCAMBIA3301 vector, producing the pCAMBIA3301-121 vector. For constructing pCAMBIA1300-121-GFP (3-1-G-GUS) vector, the GFP fragment was amplified and inserted into NcoΙ/BstEII-digested pCAMBIA3301-121 vector. All eTM fragments were inserted into XbaΙ/SacΙ-digested 3-1-G-GUS, generating the pCAMBIA1300-121-GFP-eTM (3-1-G-eTM) vector. The amplification primers for eTMs are shown in Supplemental Table S3.

The eTM overexpression vector (3-1-G-eTM) was transformed into Agrobacterium tumefaciens strain EHA105. Agrobacterial cells were infiltrated into leaves of N. benthamiana using 3-1-G-GUS as the control vector. The transient agroinfiltration assay was performed as described previously (He et al., 2008). The leaves were harvested 2 d after infiltration. The expression profiles of target genes for corresponding miRNAs were detected. All primers are shown in Supplemental Table S4. Target genes of miRNAs in N. benthamiana were predicted by psRobot (Wu et al., 2012).

Supplemental Data

The following materials are available in the online version of this article.

• Supplemental Figure S1. RT-PCR assay of ath-eTMs.

• Supplemental Figure S2. Functional analysis of osa-eTM166-2.

• Supplemental Figure S3. Functional analysis of eTMs by transient agroinfiltration assay.

• Supplemental Figure S4. Expression analysis of miR160 and target genes.

• Supplemental Table S1. Summary of predicted eTMs in Arabidopsis.

• Supplemental Table S2. Summary of predicted eTMs in rice.

• Supplemental Table S3. Primers used for vector construction.

• Supplemental Table S4. Primers used for semiquantitative or quantitative RT-PCR.

• Supplemental Data S1. Predicted eTMs of 20 conserved miRNAs in Arabidopsis.

• Supplemental Data S2. Predicted eTMs of 20 conserved miRNAs in rice.