Interaction of small RNA–8105 and the intron of IbMYB1 RNA regulates IbMYB1 family genes through secondary siRNAs and DNA methylation after wounding

Authors


For correspondence (e-mail stjeng@ntu.edu.tw).

Summary

Small RNAs (sRNAs) play important roles in plants under stress conditions. However, limited research has been performed on the sRNAs involved in plant wound responses. In the present study, a novel wounding-induced sRNA, sRNA8105, was identified in sweet potato (Ipomoea batatas cv. Tainung 57) using microarray analysis. It was found that expression of sRNA8105 increased after mechanical wounding. Furthermore, Dicer-like 1 (DCL1) is required for the sRNA8105 precursor (pre-sRNA8105) to generate 22 and 24 nt mature sRNA8105. sRNA8105 targeted the first intron of IbMYB1 (MYB domain protein 1) before RNA splicing, and mediated RNA cleavage and DNA methylation of IbMYB1. The interaction between sRNA8105 and IbMYB1 was confirmed by cleavage site mapping, agro-infiltration analyses, and use of a transgenic sweet potato over-expressing pre-sRNA8105 gene. Induction of IbMYB1-siRNA was observed in the wild-type upon wounding and in transgenic sweet potato over-expressing pre-sRNA8105 gene without wounding, resulting in decreased expression of the whole IbMYB1 gene family, i.e. IbMYB1 and the IbMYB2 genes, and thus directing metabolic flux toward biosynthesis of lignin in the phenylpropanoid pathway. In conclusion, sRNA8105 induced by wounding binds to the first intron of IbMYB1 RNA to methylate IbMYB1, cleave IbMYB1 RNA, and trigger production of secondary siRNAs, further repressing the expression of the IbMYB1 family genes and regulating the phenylpropanoid pathway.

Introduction

Plants are sessile organisms, and are thus always exposed to various environmental stresses. Unlike animals, they are unable to escape from biotic and abiotic stresses by moving. Rain, wind, pathogens and herbivores cause wounding to plants and affect their growth and development (Bowles, 1990; Taye et al., 2011). The expression of specific genes related to defense systems and wound healing may protect plants from attack by herbivores and pathogens (Bowles, 1990; Chen et al., 2005). Previous studies have identified several regulatory components, including hydrogen peroxide, nitric oxide, calcium ion, jasmonate and ethylene, as signals that are involved in defense systems against wounding (Jih et al., 2003; Chen et al., 2005, 2008; Schilmiller and Howe, 2005; Lin et al., 2011). Small RNAs (sRNAs) are also considered important regulators during transcription and post-transcription in the defense response (Carrington and Ambros, 2003; Zhang et al., 2006).

Small RNAs are divided into two major classes, microRNAs (miRNAs) and small interfering RNAs (siRNAs), according to their biogenesis (Mallory and Vaucheret, 2006). miRNAs are processed from single-stranded endogenous transcripts that form stem-and-loop structures with low free energy (Zhang et al., 2006). However, siRNAs are generated from perfectly complementary double-stranded RNA (Sunkar et al., 2007; Hsieh et al., 2009).

sRNAs are 21–24 nt non-coding RNAs that play important roles in plant development through the regulation of their targets (Bartel, 2004). Their biogenesis is dependent on the Dicer-like (DCL) ribonuclease family (Tang et al., 2003; Kurihara and Watanabe, 2004). The mature forms of sRNAs are incorporated into Argonaute (AGO)-associated RNA-induced silencing complexes (RISCs) to recognize the nucleic acid with a perfect or near-perfect complementary sequence, and cause gene silencing (Martinez et al., 2002; Bartel, 2004). Most sRNAs mediate RNA cleavage or translational repression (Martinez et al., 2002; Bartel, 2004). However, sRNAs may induce the production of secondary siRNAs (Chen et al., 2010) or trigger DNA methylation (Chellappan et al., 2010; Havecker et al., 2010; Wu et al., 2010; Cuperus et al., 2011). The functions of sRNAs are dependent on their lengths and binding to AGO (Wu et al., 2010). Most 21 and 22 nt sRNAs associate with AGO1; however, the key determinants of secondary siRNA production are 22 nt rather than 21 nt sRNAs (Chen et al., 2010; Cuperus et al., 2010). The 22 nt miRNAs miR173 and miR828 trigger production of secondary siRNAs. Artificially converting their size to 21 nt abrogates their ability to induce generation of secondary siRNAs (Chen et al., 2010). The 21 nt miRNA miR390 triggers generation of the secondary siRNA by interacting specifically with AGO7 rather than AGO1 (Axtell et al., 2006). The 24 nt sRNAs, however, associate with AGO4 to mediate DNA methylation (Wu et al., 2010; Cuperus et al., 2011). Regulation of the 24 nt miR1863 mediates methylation of Os06 g38480 (Wu et al., 2010). The miR820 precursor may be cleaved into a 21 nt miRNA to mediate RNA cleavage, and into a 24 nt miRNA to mediate DNA methylation (Wu et al., 2010).

Phenylpropanoids are involved in diverse biological activities. They act as precursors for cell-wall formation, as antioxidants, and as barriers to prevent microbial infection (Dixon and Paiva, 1995; Jorgensen et al., 2005; Naoumkina et al., 2010). Expression levels of genes involved in the phenylpropanoid pathway, which comprises the synthesis of flavonoids and lignin (Deluc et al., 2006; Bhargava et al., 2010), are up- or down-regulated after wounding (Dixon and Paiva, 1995; Delessert et al., 2004; Soltani et al., 2006). Wounding induces the activity of phenylalanine ammonia lyase, a key enzyme in the phenylpropanoid pathway, to induce phenylpropanoid accumulation (Dixon and Paiva, 1995). Several transcription factors harboring the R2R3-type MYB domain regulate phenylpropanoid metabolism (Deluc et al., 2006; Bhargava et al., 2010). IbMYB1 (MYB domain protein 1) and IbMYB2 proteins (IbMYB2–1, IbMYB2–2, IbMYB2–3 and IbMYB2–4) activate flavonoid biosynthesis in sweet potato (Ipomoea batatas cv. Tainung 57) (Mano et al., 2007). Expression of IbMYB, which mainly regulates the expression of genes involved in the core phenylpropanoid pathway, is repressed by miR828 during wounding response (Lin et al., 2012). Phenylpropanoid metabolism is also influenced by ptr-miR473 during tension and compression stresses in Populus trichocarpa (Lu et al., 2005). miRNAs thus regulate phenylpropanoid metabolism during stress responses.

Previous research has identified more than 100 miRNAs in Arabidopsis (Griffiths-Jones et al., 2008). These miRNAs have multiple roles in growth, development, miRNA metabolism and stress responses. Wounding also affects miRNA expression to induce defensive responses in tobacco (Tang et al., 2012). However, little is known regarding the wounding-responsive sRNAs. Wounding is one of the most serious stresses for plants, affecting plant growth and facilitating pathogen infection (Takabatake et al., 2006). The present study identified a novel wounding-related sRNA from sweet potato, and analyzed interaction of the wounding-responsive sRNA with its target to elucidate the sRNA-mediated mechanisms involved in wounding.

Results

Microarray-based analyses of wounding-related small RNAs

To identify wounding-responsive sRNAs in sweet potato, a microarray containing a cDNA library of sRNAs of sweet potato was prepared (Table S1). After screening using cDNAs from wounded leaves, ten clones with the highest expression were chosen for sequencing. The results showed that expression of clone105, a 24 nt sRNA (5′-AAUAAGUUCAUGCAUGGGUAAUUG-3′), was increased 2.5-fold after wounding. Its sequence showed no similarity to any known plant miRNAs in miRBase (http://mirbase.org/index.shtml) and the National Center for Biotechnology Information (NCBI) database, indicating that clone105 is a novel sRNA. Other wounding-related 20–24 nt sRNAs from the microarray analysis were also analyzed, and the putative target genes of clone105, clone133, clone167 and clone256 were predicted using the NCBI database (Table S2). Among these sRNAs, only the clone105 precursor was isolated by RACE, and clone105 was named as small RNA-8105 (sRNA8105).

Northern blotting analyses confirmed that expression of sRNA8105 was up-regulated after wounding (Figure 1). The sRNA8105 transcript was detected after 0.5 h, and remained present for up to 3 h after leaf wounding. Northern blotting using antisense sRNA8105 as a probe detected two different-sized RNAs: 22 and 24 nt (Figure 1 and Figure S1), indicating that these two sRNA species may derive from the sRNA8105 precursor (pre-sRNA8105).

Figure 1.

Expression of sRNA8105 upon wounding

Total RNAs from the leaves wounded at the times indicated were analyzed using Northern blotting to detect sRNA8105 and 5S rRNA. Unwounded (W−) leaves were included for comparison; 5S rRNA was used as an internal control.

Processing of stem-loop pre-sRNA8105 to generate sRNA8105 via IbDCL1

Isolation of the sRNA precursor allowed us to classify a novel sRNA. A 171 bp DNA fragment encoding pre-sRNA8105 gene was obtained from low-molecular-weight RNAs of sweet potato, and its secondary structure was predicted using mfold (Zuker, 2003). This precursor structure formed a stem-and-loop hairpin with a free energy of −35.88 kcal mol−1 (Figure 2a and Figure S2), and was considered as a potential miRNA precursor structure with low free energy (−32 to −57 kcal mol−1) (Bonnet et al., 2004; Zhang et al., 2006). In addition, the sequence UAACUCUGGAUGGUA UGU, located at the sRNA8015* region, was identified by sRNA deep sequencing, and its expression was much lower than that of sRNA8105 (Figure S3).

Figure 2.

Processing of pre-sRNA8105 via IbDCL1

(a) Part of the secondary structure of pre-sRNA8105. The complete secondary structure is presented in Figure S2

(b) Expression of sRNA8105 and IbMYB1-siRNA in transgenic sweet potatoes. Total RNA was isolated from WT, transgenic plants expressing an empty vector (EV), and transgenic plants over-expressing pre-sRNA8105 gene (lines 8105–1, 8105–2 and 8105–4), pre-sRNA8105 gene with Ibdcl1-RNAi (lines 8105/dcl1-RNAi–1, 8105/dcl1-RNAi–2 and 8105/dcl1-RNAi–4), or pre-sRNA8105 gene with Ibdcl2-RNAi (lines 8105/dcl2RNAi–1, 8105/dcl2RNAi–2 and 8105/dcl2RNAi–3). Then the expression levels of sRNA8105, IbMYB1-siRNA and 5S rRNA were detected using Northern blotting.

Small RNA biogenesis involves four DCL proteins. DCL1 is mainly involved in generation of miRNAs, while DCL2, DCL3 and DCL4 are responsible for production of siRNAs (Laubinger et al., 2010). In addition, RNA-dependent RNA polymerase (RDR) proteins are required for siRNA generation. We isolated two IbDCL and two RDR genes from sweet potato, namely IbDCL1, IbDCL2, IbRDR2 and IbRDR6, based on their sequence similarities to those genes from other plants (Figure S4). Transgenic sweet potato plants over-expressing pre-sRNA8105 gene with or without the Ibdcl-RNAi constructs (8105OE, 8105OE/Ibdcl1-RNAi and 8105OE/Ibdcl2-RNAi) were generated to examine which IbDCL genes are involved in sRNA8105 biogenesis (Figure 2b). Without wounding, mature forms of sRNA8105 occurred at significantly higher levels in the 8105OE transgenic plants compared to wild-type and transgenic plants expressing the empty vector. In addition, sRNA8105 accumulated in 8105OE/Ibdcl2-RNAi but not in 8105OE/Ibdcl1-RNAi transgenic plants. These results indicate that IbDCL1 rather than IbDCL2 is required for formation of sRNA8105. In addition, sRNA8105 production did not require functional IbRDR2 and IbRDR6 according to sRNA8105 expression analysis in transgenic plants over-expressing pre-sRNA8105 gene with or without the Ibrdr-RNAi constructs (8105OE, 8105OE/Ibrdr2-RNAi and 8105OE/Ibrdr6-RNAi) (Figure S5). Taken together, the results show that sRNA8105 biogenesis is DCL1-dependent, and DCL2-, RDR2- and RDR6-independent.

sRNA8105 targets the first intron of IbMYB1

To identify potential sRNA targets, the sweet potato expressed sequence tag (EST) and whole-genome shotgun (WGS) databases from NCBI were used to BLAST (http://blast.ncbi.nlm.nih.gov/Blast.cgi) search the complementary sequences of sRNA8105. This computational analysis selected complementary sequences containing up to four mismatches (counting a G::U pair as a match), and predicted that the sRNA8105 may target the first intron of IbMYB1 with three mismatches. PCR was further used to amplify the cDNAs transcribed from DNase-treated mRNAs of sweet potato to detect the presence of the novel mature mRNA containing the first intron, indicating that only one 211 bp DNA fragment was generated from PCR using primers IbMYB1 F and IbMYB1 R, which bound to the first exon and the second exon of IbMYB1, respectively (Figure 3a). No DNA fragment of 197 bp was produced by PCR using primers IbMYB1 F and IbMYB1 R2, which bind to the first exon and the first intron of IbMYB1, respectively (Figure 3b). These results indicate that mature IbMYB1 mRNA with the first intron is undetectable, and that removal of the first intron of the IbMYB1 mRNA occurred during RNA splicing. The interaction of sRNA8105 with the first intron of IbMYB1 RNA may occur before generation of mature IbMYB1 mRNA. Using RNAfold (Zuker, 2003), the free energy of sRNA8105::IbMYB1 pairs was predicted to be −24.5 kcal mol−1, which is much lower than the threshold energy (−20 kcal mol−1) of validated miRNA::mRNA pairs in plants (Wang et al., 2004; Rusinov et al., 2005), indicating that IbMYB1 may be a target of sRNA8105.

Figure 3.

The absence of the first intron in mature IbMYB1 mRNA

(a) Schematic representation of IbMYB1. Exons are represented by boxes; introns are represented by lines. Arrows indicate the positions of primer binding

(b) The cDNA and genomic DNA (gDNA) of sweet potato were analyzed by PCR using specific primer pairs to detect IbMYB1 and IbActin. Primers IbMYB1 F (F), IbMYB1 R (R), and IbMYB1 R2 (R2) bind to exon 1, exon 2 and intron 1, respectively.

Wounding suppresses the expression of IbMYB1 family genes

Expression of IbMYB1, the putative target of sRNA8105, was evaluated using quantitative RT–PCR with primers IbMYB1 F/IbMYB1 R. These primers recognize IbMYB1 family genes because the cDNA sequences of the IbMYB1 family genes, i.e. the IbMYB1 and IbMYB2 genes, display up to 95% sequence identity. Surprisingly, expression of all the IbMYB1 family genes was suppressed 0.5 and 1 h after wounding (Figure 4a). However, their expression at 3 h after wounding showed no obvious differences from that in a plant without wounding. In addition, genomic DNA (gDNA) analysis by PCR using primers IbMYB1 F/IbMYB1 R4 generated a 368 bp fragment from IbMYB1, and the primers IbMYB1 F/IbMYB2s R generated a 329 bp fragment from IbMYB2 genes (Figure S6), indicating that these specific DNA primers may be able to distinguish between IbMYB1 and IbMYB2 genes. Quantitative RT–PCR using these specific primers was used to detect the levels of IbMYB1 and IbMYB2 gene expression upon wounding. Decreased expression of IbMYB1 was correlated with induction of sRNA8105 (Figures 1 and 4b), indicating that IbMYB1 is the likely target of the sRNA8105. Similarly, the expression of IbMYB2 genes also decreased 0.5 and 1 h after wounding (Figure 4c). Based on the DNA sequences, the interaction site of sRNA8105 is only present in IbMYB1. It is therefore likely that IbMYB1 and sRNA8105 are involved in the regulation of IbMYB2 genes.

Figure 4.

Expression of IbMYB1 family genes upon wounding

The IbMYB1 family genes include IbMYB1 and IbMYB2 genes, and they are regulated by sRNA8105. Total RNA from the leaves wounded at the times indicated was treated with DNase I and then reverse-transcribed to produce cDNAs, which were then assayed for expression of IbMYB1 family genes (a), IbMYB1 (b) and IbMYB2 genes (c) using quantitative PCR. These data were normalized to IbActin gene expression, and the ratios relative to the plants without wounding are shown as relative expression levels. Values are means ± standard deviation (n = 3).

Direct cleavage of IbMYB1 and triggering of secondary siRNAs by 22 nt sRNA8105 suppresses IbMYB1 family genes upon wounding

The present study analyzed the cleavage sites of sRNA8105 in IbMYB1 using RNA ligase-mediated RACE. Cleavage of IbMYB1 RNA occurred in the first intron, with cleavage sites located at the 8th, 9th and 10th (Figure 5a). Most of these cleavage sites were canonical cleavage sites, indicating that the 22 nt sRNA8105 directly cleaves IbMYB1. Of the IbMYB1 family genes, only the IbMYB1 gene contains the binding site for sRNA8105 (Figure S7). sRNA8105 has 22 and 24 nt mature forms, and 22 nt miRNA is a key component for triggering of secondary siRNA production (Chen et al., 2010; Cuperus et al., 2010). Therefore, these data indicate that binding of 22 nt sRNA8105 to the intron of IbMYB1 induces generation of secondary siRNAs, which further regulate expression of the IbMYB1 family genes, including IbMYB1 and IbMYB2 genes.

Figure 5.

Induction of the secondary siRNA by sRNA8105

(a) Cleavage sites induced by sRNA8105 in the first intron of IbMYB1. A RNA ligase-mediated RACE method was used for mapping of the IbMYB1 cleavage sites. Various sizes of isolated fragments were cloned and sequenced. Vertical arrows indicate the positions of cleavage. Watson–Crick pairings are indicated by vertical dashes; G:U wobble pairings are indicated by dots

(b) Expression of IbMYB1-siRNA upon wounding. After wounding of leaves, their total RNAs were analyzed using Northern blotting to detect IbMYB1-siRNA (upper panel), sRNA8105 (middle panel) and 5S rRNA (lower panel). Their expression in the unwounded (W–) leaves was included for comparison; 5S rRNA was used as an internal control

(c,d) Expression of IbMYB1-siRNA and IbMYB1 family genes in transgenic plants over-expressing pre-sRNA8105 gene without wounding. Total RNAs were isolated from wild-type (WT), 8105OE transgenic plants 8105–1 and 8105–2, and transgenic plants expressing the empty vector (EV). The expression levels of IbMYB1-siRNA (c, upper panel), sRNA8105 (c, middle panel) and 5S rRNA (c, lower panel) were detected using Northern blotting. Expression of IbMYB1 family genes (d) was determined using quantitative RT–PCR. The expression of IbActin was analyzed as a control to normalize the RT–PCR results.

To evaluate the effect of the secondary siRNAs from IbMYB1 upon wounding, Northern blot analysis was performed using the predicted anti-siRNA as a probe. IbMYB1-siRNA was detected 0.5 h after leaf wounding (Figure 5b). Putative IbMYB1-related sRNAs were also found by sRNA deep sequencing (Figure S7). The secondary IbMYB1-siRNA also accumulated in 8105OE transgenic plants without wounding (Figure 5c). Detectable amounts of IbMYB1-siRNA were also observed in transgenic 8105OE/Ibdcl2-RNAi and 8105OE/Ibdcl1-RNAi, but the level was greater in 8105OE/Ibdcl2-RNAi transgenic plants than in 8105OE/Ibdcl1-RNAi transgenic plants (Figure 2b). The background levels of IbMYB1-siRNA in 8105OE/Ibdcl1-RNAi plants may result from RNAi knocking down rather than knocking out Ibdcl1 expression (Figure 2b). The presence of IbMYMB1-siRNA in 8105OE/Ibdcl2-RNAi plants may be due to the presence of functional endogenous DCL3 and DCL4 in plants. These results demonstrate that reduced production of sRNA8105 decreases the generation of IbMYB1-siRNA, and that IbMYB1-siRNA is an siRNA related to sRNA8105. In addition, sRNA8105 triggers the production of a secondary siRNA. Furthermore, expression of the sRNA8105-regulated genes, i.e. IbMYB1 family genes, was significantly suppressed in 8105OE plants (Figure 5d). Taken together, the results from sequence comparison of IbMYB1 family genes, mapping of cleavage sites, and Northern blot analysis of siRNA demonstrate that generation of secondary siRNA induced by sRNA8105 regulates expression of the IbMYB1 family genes.

Mediation of IbMYB1 methylation by 24 nt sRNA8105 upon wounding

The most common target sites for sRNAs are the open reading frame (ORF) and untranslated regions (UTRs) (Rhoades et al., 2002; Carrington and Ambros, 2003; Jones-Rhoades and Bartel, 2004). The sRNA8105 target site is located in the first intron of the IbMYB1 gene, but little is known concerning the function of sRNA binding to introns. Pre-sRNA8105 generates two mature forms with different sizes. Thus, the regulation of IbMYB1 family genes by sRNA8105 may involve mechanisms other than mRNA cleavage. miRNAs may also have dual functions; they may regulate gene expression through mRNA cleavage as well as DNA methylation (Wu et al., 2010). In the present study, McrBC, a methylated DNA-specific restriction enzyme, was applied to gDNA isolated from unwounded and wounded leaves prior to quantitative PCR analysis. Amplification of gDNA that was not treated with McrBC was used as a control. As PCR was unable to amplify the methylated DNA digested by McrBC, the ratio of unmethylated DNA to undigested gDNA reveals the DNA methylation status. These results indicate that wounding leads to IbMYB1 DNA methylation (Figure 6a). Subsequently, McrBC-quantitative PCR also detected partial DNA methylation of IbMYB2 genes after wounding (Figure 6b). Over-expression of sRNA8105 increased DNA methylation in IbMYB1 (Figure 6c) and IbMYB2 genes (Figure 6d). DNA methylation is mediated by 24 nt miRNA (Wu et al., 2010). Therefore, in addition to RNA cleavage and secondary siRNA production, 24 nt sRNA8105 may mediate DNA methylation of the IbMYB1 family genes.

Figure 6.

Methylation of IbMYB1 family genes

(a,b) Methylation of IbMYB1 family genes in the wild-type upon wounding. After wounding of leaves, their gDNA was digested by McrBC, which specifically cleaves methylated DNA. The digested DNAs were amplified using PCR to detect IbMYB1 (a) and IbMYB2 genes (b). Undigested gDNA was simultaneously amplified as controls for quantitative comparison. gDNA from leaves without wounding was also analyzed, and the values were set to 1 to normalize the other data. Values are means ± standard deviation (n = 3)

(c,d) Methylation of IbMYB1 family genes in transgenic plants over-expressing pre-sRNA8105 gene without wounding. gDNA was isolated from wild-type (WT), 8105OE transgenic plants 8105–1 and 8105–2, and EV transgenic plants to analyze DNA methylation of IbMYB1 (c) and IbMYB2 genes (d) using quantitative PCR. Undigested gDNA was simultaneously amplified as controls for quantitative comparison. The gDNA from wild-type was also analyzed, and the values were set to 1 to normalize the other data. Values are means ± standard deviation (n = 3).

Agrobacterium-mediated transient expression confirmed the interaction between sRNA8105 and IbMYB1 in vivo

Expression of pre-sRNA8105 gene, the target mimic inhibitor of sRNA8105 (mimic8105) (mimic8105) and IbMYB1 was almost undetected in tobacco (Nicotiana tabacum) (Figure 7a,c,d). The genomic sequence of IbMYB1 containing the sRNA8105 interaction site or the cDNA of IbMYB1 without the sRNA8105 interaction site, combined with the pre-sRNA8105 gene, mimic8105 or the empty vector, were co-infiltrated into tobacco leaves by Agrobacterium-mediated infiltration. Total RNAs and gDNA of the leaves were simultaneously isolated by Trizol reagent to analyze gene expression (total RNAs) and methylation levels (gDNA) (Figure 7a–d). Quantitative RT–PCR was used for gene expression analyses, evaluating the transcript levels of IbMYB1, pre-sRNA8105 gene, mimic8105, Neomycinphosphotransferase II (NPTII) and NtActin (Figure 7a,c,d). The NPTII and NtActin expression levels were used as controls for quantitative comparisons. Detection of pre-sRNA8105 gene and mimic8105 confirmed their expression in infiltrated leaves (Figure 7c,d). In the presence of sRNA8105, the amount of IbMYB1 was significantly lower than in the presence of the empty vector (Figure 7a). In addition, the sRNA8105 did not efficiently suppress IbMYB1 expression in the presence of mimic8105 (Figure 7a). These results demonstrate that sRNA8105 regulates the expression of IbMYB1 in vivo. To determine DNA methylation levels, gDNA from the infiltrated leaves was treated with McrBC prior to quantitative PCR analysis (Figure 7b). Undigested gDNA was used as a control. Small RNA-8105 induced significantly higher methylation of IbMYB1 compared to the empty vector or mimic8105. IbMYB1ΔsRNA8105 without the sRNA8105 interaction site combined with the pre-sRNA8105 gene, mimic8105 or the empty vector was also infiltrated into tobacco leaves by Agrobacterium-mediated infiltration. No difference was found in the amounts of IbMYB1ΔsRNA8105 transcript and methylation levels in the presence of the pre-sRNA8105 gene, mimic8105 or the empty vector (Figure 7a,b). Taken together, these data suggest that sRNA8105 is able to directly cleave IbMYB1 and methylate its gDNA in vivo.

Figure 7.

Agroinfiltration assays of sRNA8105 and its target IbMYB1

Tobacco leaves were infiltrated by Agrobacterium-mediated infiltration using a vector containing the genomic sequence of IbMYB1 with the sRNA8105 interaction site (IbMYB1) or cDNA of IbMYB1 without the sRNA8105 interaction site (IbMYB1Δ8105), combined with empty vector (1301), sRNA8105 precursor (sRNA8105) or the target mimic inhibitor of sRNA8105 (mimic8105). After 4 days, total RNAs and gDNA of these leaves were simultaneously isolated by Trizol reagent. Total RNAs were used to detect gene expression (a,c,d). The gDNAs were used to analyze DNA methylation (b)

(a) Expression of IbMYB1. The expression of IbMYB1 was analyzed using quantitative RT–PCR, and normalized to the levels of NPTII expression. The ratios relative to the plants with empty vector are shown as relative expression levels

(b) DNA methylation of IbMYB1. Genomic DNAs were treated with McrBC, a restriction enzyme specific for methylated DNA, and amplified using PCR to detect undigested IbMYB1. gDNA without McrBC treatment was simultaneously amplified as controls for quantitative comparison. The gDNA from the plants with empty vector was also analyzed, and the values were set to 1 to normalize the other data

(c,d) Expression of pre-sRNA8105 gene (c) and mimic8105 (d). The levels of pre-sRNA8105 gene and mimic8105 expression were analyzed using quantitative RT–PCR, and their values were normalized to the levels of NtActin expression

Values are means ± standard deviation (n = 3).

Wounding affects genes related to the phenylpropanoid pathway

The IbMYB1 family genes encode activators for flavonoid biosynthesis, and up-regulate expression of anthocyanin synthase (IbANS), chalcone isomerase (IbCHI), chalcone synthase (IbCHS), dihydroflavonol reductase (IbDFR) and flavanone 3–hydroxylase (IbF3H) in sweet potato (Mano et al., 2007). Wounding-induced sRNA8105 repressed expression of IbMYB1 family genes (Figure 4a). The expression levels of these flavonoid biosynthesis-related genes were also analyzed after wounding. Wounding down-regulated the expression of IbANS, IbCHI, IbCHS, IbDFR and IbF3H (Figure 8a), and their repression coincided with suppression of the IbMYB1 family genes following wounding (Figure 4a–c). Expression of genes involved in the general phenylpropanoid and specific lignin biosynthesis pathways, i.e. phenylalanine ammonia lyase (IbPAL), hydroxycinnamoyl CoA shikimate/quinate hydroxycinnamoyl transferase (IbHCT) and cinnamyl alcohol dehydrogenase 1 (IbCAD1), was also induced upon wounding (Figure 8b). An increase of lignin content was observed in sweet potato after wounding (Lin et al., 2012). These results demonstrate that wounding causes metabolic flux in the phenylpropanoid pathway toward lignin biosynthesis rather than the flavonoid biosynthesis pathway. 8105OE transgenic plants showed increased expression of IbPAL02 and IbPAL2–8 (Figure 8c), which encode core enzymes in the phenylpropanoid pathway related to the biosynthesis of flavonoid and lignin (Deluc et al., 2006; Bhargava et al., 2010). In contrast, 8105OE transgenic plants showed differential expression of the branch point enzymes IbCHS (Borevitz et al., 2000; Besseau et al., 2007) and IbHCT (Besseau et al., 2007; Li et al., 2010) in the flavonoid and lignin biosynthesis pathways (Figure 8c). IbCHS was down-regulated while IbHCT was up-regulated compared to wild-type and empty vector transgenic plants (Figure 8c). Lignin contents were also significantly increased in 8105OE transgenic plants without wounding (Figure 8d). These findings indicate that sRNA8105 represses the IbMYB1 family genes to regulate the phenylpropanoid pathway following wounding.

Figure 8.

Expression of genes involved in the phenylpropanoid pathway

(a) Expression of genes involved in flavonoid biosynthesis of wild-type sweet potato (WT). Total RNAs from the unwounded and wounded leaves at the times indicated were isolated and analyzed using quantitative RT–PCR. The levels of IbANS, IbCHI, IbCHS, IbDFR and IbF3H expression were detected. The level of IbActin expression was analyzed as a control to normalize RT–PCR results

(b) Expression of genes involved in the core phenylpropanoid pathway and lignin biosynthesis in WT. Total RNAs from the unwounded and wounded leaves were used to detect IbPAL02, IbPAL2–8, IbHCT and IbCAD1 expression using quantitative RT–PCR. The level of IbActin expression was analyzed as a control to normalize RT–PCR results

(c) Expression of genes involved in the core phenylpropanoid pathway, flavonoid biosynthesis and lignin biosynthesis of 8105OE transgenic plants. Total RNAs isolated from WT and transgenic plants (8105–1, 8105–2 and EV) were used to detect IbPAL02, IbPAL2–8, IbCHS and IbHCT expression using quantitative RT–PCR. The level of IbActin expression was analyzed as a control to normalize RT–PCR results

(d) Relative lignin contents in plants. The lignin levels of transgenic plants expressing empty vector (EV) were treated as the reference, and the values were set to 1 to normalize the other data

(e) The phenylpropanoid pathway. The phenylpropanoid pathway is divided into three parts, consisting of the core, flavonoid and lignin pathways. The core phenylpropanoid pathway involves PAL. The flavonoid pathway involves CHS, CHI, F3H, DFR and ANS. The lignin pathway involves HCT and CAD. Upward triangles indicate genes that are up-regulated by wounding. Inverted triangles indicate genes that are down-regulated by wounding

For (a–d), values are means ± standard deviation (n = 3).

Discussion

sRNA8105 forms 22 and 24 nt sRNAs, and its biogenesis is DCL1-dependent and DCL2-, RDR2- and RDR6-independent according to the genetic analyses (Figure 2b and Figure S5). In addition, the secondary structure of pre-sRNA8105 is an imperfectly complementary dsRNA with two asymmetric bulges (Figure 2a and Figure S2), which is the basic structure of miRNA precursors (Meyers et al., 2008; Munafo and Robb, 2010). The siRNAs are derived from perfectly complementary double-stranded RNA (Meyers et al., 2008; Munafo and Robb, 2010). These results suggest that sRNA8105 is an miRNA not an siRNA. However, the important criterion for miRNA classification is demonstration of precise processing (Meyers et al., 2008). Therefore, we performed three rounds of sRNA deep sequencing using three different organizations (Welgene Biotech, http://www.welgene.com.tw; Technology Commons, http://techcomm.lifescience.ntu.edu.tw/; Genetech Biotech, http://www.gtbiotech.com.tw/). In one of these deep sequencing experiments, part of miR8105* and a related IbMYB1-siRNA were obtained (Figures S3 and S7). However, miR8105 itself was not observed during sRNA sequencing. Using real-time PCR to analyze the same sample as used for sRNA sequencing, expression of miR8105* was much lower than that of miR8105 (Figure S3). This situation may result from rapid degradation of the pre-miRNA species after DCL cleavage (Jones-Rhoades et al., 2006; Morin et al., 2008; Schreiber et al., 2011). In addition, recent studies also indicated that data from sRNA deep sequencing show biases due to sRNA/adapter structures, T4 RNA ligase 1, and especially T4 RNA ligase 2 for 3’–adapter ligation (Hafner et al., 2011; Jayaprakash et al., 2011; Sorefan et al., 2012; Zhuang et al., 2012). In addition, the mechanism of miRNA modification, which may affect the cDNA library of sRNAs used for sequencing, is complex and remains to be investigated in plants. Hence, further study is needed to determine whether sRNA8105 is an miRNA.

Most miRNAs mediate target gene silencing by recognizing perfect or near-perfect complementary mRNA at the target sites, i.e. the ORF and UTRs (Martinez et al., 2002; Rhoades et al., 2002; Carrington and Ambros, 2003; Bartel, 2004; Jones-Rhoades and Bartel, 2004). However, the sRNA8105 target site was in the first intron of IbMYB1. In mouse, AGO involved in miRNA activation interacts with target sites in the ORF and UTRs and also in the intron (Chi et al., 2009), indicating that miRNA may bind to introns. Based on a genome-scale survey of intron-derived sRNAs in rice, sirton-derived siRNAs, processed from introns which form hairpin structures, may methylate the DNA of the host gene (Chen et al., 2011). However, there is a lack of direct experimental validation to suggest that miRNAs bind to the intron in rice. In addition, alternative splicing of the first intron of AtMYBL2 mRNA produces a novel alternative transcript using the first intron, which is bound and cleaved by sRNA (Franco-Zorrilla et al., 2009). This first intron becomes part of the ORF in this alternative transcript; hence sRNA binding occurs at the ORF. In the present study, sRNA8105 was confirmed to induce cleavage in the first intron of IbMYB1 RNA by mapping of sRNA8105 cleavage sites (Figure 5a) and agroinfiltration analyses (Figure 7a).

The present study's experimental findings confirmed that sRNA8105 binds to the first intron of IbMYB1 RNA. Mature IbMYB1 mRNA containing the first intron was not detected (Figure 3), indicating that the interaction of sRNA8105 with the first intron of IbMYB1 occurred before splicing of IbMYB1 mRNA. Kim et al. (2010) found that IbMYB1 in sweet potato generates an alternative splice variant using the second rather than the first intron. Splicing of RNA occurs during gene transcription (Proudfoot et al., 2002; Listerman et al., 2006), and requires a 3′ splice site, a 5′ splice site and a branch site (Harris and Senapathy, 1990; Lal et al., 1999; Simpson et al., 2002). In IbMYB1, the interaction site of sRNA8105 is located in front of the consensus branch site CU(A/G)A(A/U) (Figure S4), which is located between 10 and 50 nt upstream of the 3′ splice site (Harris and Senapathy, 1990; Simpson et al., 2002). It is likely that the interaction of sRNA8105 with IbMYB1 occurs immediately after generation of the first intron of IbMYB1 RNA and before termination of transcription.

The 22 nt sRNA8105 was able to directly cleave IbMYB1 RNA and also regulate the IbMYB1 gene family by triggering production of secondary siRNAs upon wounding. Several sRNAs have the ability to act as a trigger for secondary siRNA production in plants (Chen et al., 2010; Cuperus et al., 2011). These secondary siRNAs enhance or fine-tune sRNA effects (Ronemus et al., 2006; Chen et al., 2007; Howell et al., 2007; Zhai et al., 2011). miR399 cleaves PHO2 RNA and also triggers production of PHO2-siRNAs to reinforce or fine-tune regulation of PHO2 in Arabidopsis (Lin et al., 2008). However, the siRNA id65 strengthens target silencing by production of secondary siRNA from upstream of the id65 cleavage site in grape (Vitis vinifera) (Carra et al., 2009). In addition to cis-acting secondary siRNAs, sRNAs mediate production of trans-acting secondary siRNAs. Binding of miR390 to TAS3 generates ta-siARFs that regulate auxin response factors 2, 3 and 4 (Marin et al., 2010). sRNAs may also target one gene to regulate other members of the gene family by producing siRNA. Binding of ta-siR2140 to At1g63130 RNA, a member of the PPR family, generates secondary siRNA that cleaves the RNA of another PPR gene, At1g62930 (Chen et al., 2007; Howell et al., 2007). Over-expression of the sRNA8105 in transgenic plants represses members of its target family, i.e. IbMYB1 and IbMYB2 genes, and induces secondary siRNAs from IbMYB1 (Figure 5c,d). Thus, the sRNA8105 targets one IbMYB1 gene family member in order to repress the whole IbMYB1 gene family through siRNA.

Mapping of the cleavage sites in the IbMYB1 gene indicated the presence of canonical and non-canonical sites for sRNA8105 interaction (Figure 5a). Similar observations have been made previously for other sRNAs (Jones-Rhoades and Bartel, 2004; Allen et al., 2005; Lauter et al., 2005; Chen et al., 2007; Moxon et al., 2008; Carra et al., 2009; Wu et al., 2010). Instead of the canonical position, random breakages or aberrant directed cleavages occur between miR172 and GLOSSY15 mRNA (Lauter et al., 2005). In Euphorbiaceous plants, the cleavage sites for miR156 regulation substantially differ between stressed and unstressed conditions (Zeng et al., 2010). The siRNA id65 also generates 21 nt phased clusters from the cleavage site in the cytokinin synthase gene (Carra et al., 2009). Non-canonical cleavage sites guided by secondary siRNA may thus be present in IbMYB1 upon sRNA8105 interaction.

miRNAs mediate DNA methylation of target genes (Chellappan et al., 2010; Havecker et al., 2010; Wu et al., 2010; Cuperus et al., 2011). These miRNAs are usually 24 nt long and start with an adenine residue (Mi et al., 2008; Wu et al., 2010). The 24 nt sRNA8105 also starts with an adenine residue (Figure 5a). In this study, McrBC-PCR analyses revealed methylation of IbMYB1 after wounding (Figure 6a), and also methylation of the IbMYB1 family genes in wild-type plants upon wounding (Figure 6a,b) and in 8105OE transgenic plants without wounding (Figure 6c,d). These results strongly indicate that the 24 nt sRNA8105 regulates DNA methylation of IbMYB1 family genes.

The lengths of miRNAs may determine their functions. The MIR genes generate different-sized sRNAs at the same miRNA site to mediate mRNA cleavage and DNA methylation (Chellappan et al., 2010; Wu et al., 2010). The functions of the miRNAs are also dependent on the varied AGO family members associated with them (Wu et al., 2010). The 21 nt miRNAs are generated by DCL1 and associate with AGO1 to cleave target mRNA, while 24 nt miRNAs are produced by DCL3 and associate with AGO4 to mediate DNA methylation (Chellappan et al., 2010; Havecker et al., 2010; Wu et al., 2010; Cuperus et al., 2011). DCL1 also generates 22 nt miRNAs, which associate with AGO1 to trigger secondary siRNA production from the miRNA duplex, which contains a bulge (Chen et al., 2010; Manavella et al., 2012). The sRNA duplex region of pre-sRNA8105 contains two bulges (Figure 2a), and DCL1 affects sRNA8105 biogenesis (Figure 2b). These results indicate that pre-sRNA8105 forms 22 nt sRNA8105. In Northern blot analysis, two bands at 22 and 24 nt were obtained using anti-sRNA8105 as a detecting probe (Figure 1 and Figure S1). However, 24 nt miRNAs are mainly produced by DCL3 rather than DCL1 (Chellappan et al., 2010; Wu et al., 2010). Both 22 and 24 nt sRNA8105 were affected in Ibdcl1-RNAi transgenic plants (Figure 2b), indicating that IbDCL1 catalyzes the cleavage step from primary sRNA8105 to precursor sRNA8105. Thus, the sRNA8105 precursor may be cleaved into different sizes of mature sRNAs to mediate RNA cleavage and DNA methylation of IbMYB1. The 22 nt sRNA8105 may also mediate production of secondary siRNAs to regulate expression of IbMYB1 and IbMYB2 genes. Production of secondary siRNA is mostly dependent on 22 nt sRNAs (Chen et al., 2010; Cuperus et al., 2010). However, 21 nt miR390 and miR399 also trigger production of secondary siRNAs (Axtell et al., 2006; Lin et al., 2008). These 21 nt miRNAs recognize more than two binding sites in their target RNAs, and processing of the phased secondary siRNAs occurs among the two binding sites (Axtell et al., 2006; Lin et al., 2008). sRNA8105 has only one target site in the IbMYB1 gene. Processing of the sRNA8105 precursor generated 22 nt mature sRNA, which triggered production of secondary siRNAs, and 24 nt mature sRNA, which triggered DNA methylation of the target. The interaction of sRNA8105 with IbMYB1 RNA may therefore induce three mechanisms: cleavage of target RNA, triggering of secondary siRNA production, and methylation of target DNA.

The R2R3-MYB transcription factors regulate multiple steps in the phenylpropanoid pathway, which comprises syntheses of flavonoid and lignin (Deluc et al., 2006; Bhargava et al., 2010). In sweet potato, expression of the IbMYB1 family genes up-regulates expression of the flavonoid biosynthesis-related genes ANS, CHI, CHS, DFR and F3H (Mano et al., 2007). The transcription factors AtMYB58 and AtMYB63 increase expression of the lignin biosynthesis-related genes HCT and CAD (Zhou et al., 2009). AtMYB75 has dual roles as an activator for flavonoid biosynthesis (Borevitz et al., 2000) and a repressor for lignin biosynthesis in the phenylpropanoid pathway (Bhargava et al., 2010). In the present study, however, expression of IbHCT, encoding the branch point enzyme for lignin biosynthesis pathway (Besseau et al., 2007; Li et al., 2010), was induced by over-expressing sRNA8105, which suppresses expression of IbMYB1 family genes (Figure 8c). Thus the IbMYB1 family genes play dual roles in the phenylpropanoid pathway, as activators to stimulate flavonoid biosynthesis, and repressors to reduce lignin synthesis (Figure 8e). Wounding also induces expression of genes related to lignin biosynthesis in Arabidopsis and sweet potato (Delessert et al., 2004; Soltani et al., 2006). Lignin content is increased to thicken the cell wall at the wounding site (Yen et al., 2001; Lehner and Cardemil, 2003), and lignin formation generates barriers to increase resistance to bacterial pathogens (Lehner and Cardemil, 2003; Raes et al., 2003; Shao et al., 2010). The present study showed up-regulated IbHCT expression in the wild-type upon wounding (Figure 8b) and in 8105OE transgenic plants without wounding (Figure 8c). Lignin contents also increased in 8105OE transgenic plants without wounding (Figure 8d). Wounding was able to induce expression of sRNA8105 to suppress the IbMYB1 family genes, further affecting the expression of genes related to the phenylpropanoid pathway and enhancing lignin biosynthesis (Figure 8e).

In conclusion, the present study identified a novel wounding-related sRNA, sRNA8105, in sweet potato. The 22 and 24 nt sRNA8105s bind to the first intron of IbMYB1 before splicing to mediate RNA cleavage and DNA methylation of IbMYB1, respectively. The interaction of sRNA8105 with IbMYB1 RNA also triggers production of secondary siRNAs to decrease expression of IbMYB1 family genes. The repression of the IbMYB1 family genes may affect metabolic flux in the phenylpropanoid pathway, increasing lignin biosynthesis by regulating the expression of flavonoid and lignin biosynthesis-related genes (Figure S8).

Experimental Procedures

Plant materials and treatments

Sweet potato (Ipomoea batatas cv. Tainung 57) plants were grown in a controlled environment (16 h light at 25°C per 8 h dark at 22°C; humidity 70%; light 40 μmol photons m−2 sec−1). Plants with 6–8 fully developed leaves were used, and the third fully expanded leaves counted from the terminal bud were excised in this study. Plants were wounded with forceps for the duration indicated in each experiment. Tobacco (Nicotiana tabacum L. cv. W38) was grown in the greenhouse. For sweet potato transformation, Agrobacterium rhizogenes-mediated transformation was used (Lee et al., 2004; Schmidt et al., 2007). Agrobacterium rhizogenes strain 15834 was used for hairy root induction. All experiments in this study were repeated at least three times, and similar gene expression patterns were obtained.

Construction of a cDNA library of sRNAs of sweet potato

Construction of a cDNA library of sRNAs of sweet potato was performed as described by Lu et al. (2007) with minor modifications. Total RNAs were isolated using Trizol reagent (Invitrogen, http://www.invitrogen.com). sRNAs (20–30 nt) were isolated by size fractionation. The 5′ ends of isolated sRNAs were ligated to the 5–RNA adapter (Table S3) using T4 RNA ligase 1 (New England Biolabs, NEB, http://www.neb.com) at 37°C for 1 h. Poly(A) tails were added to their 3′ ends by incubation with poly(A) polymerase (NEB) at 37°C for 0.5 h. Then they were reverse-transcribed to produce cDNAs using MMLV reverse transcriptase (Invitrogen) with the 3–RT adapter (Table S3). These cDNAs were further amplified by PCR using primers 5–adapter and 3–adapter (Table S3). PCR fragments of 80–90 nt were isolated and cloned into yT&A cloning vector (Yeastern Biotech, http://www.yeastern.com/). A total of 384 colonies were obtained and printed on a Corning® epoxide-coated slide (Corning, http://www.corning.com/index.aspx), which was used to identify wounding-responsive sRNAs in further experiments.

Microarray analysis

Microarray slide fabrication, probe labeling reactions, microarray hybridization and microarray image analysis were performed as described by Chuang et al. (2002).

Isolation of sRNA precursors

Isolation of sRNA precursors was performed as described by Lin et al. (2012) with minor modifications. The 5′ sequence of the sRNA precursor was obtained by PCR using primers 5–adapter and anti-oligo8105 (Table S3), and its 3′ sequence was also obtained by PCR using primers 3–adapter and oligo8105 (Table S3). Then these PCR fragments were cloned and sequenced. The secondary structure of the sRNA precursor from sweet potato was predicted using RNAfold (Zuker, 2003).

Isolation of DCL and RDR genes and production of Ibdcl-RNAi and Ibrdr-RNAi constructs

The conserved domains of DCL and RDR genes from Arabidopsis thaliana, tobacco and rice (Oryza sativa) were used to BLAST (http://blast.ncbi.nlm.nih.gov/Blast.cgi) search the Ipomoea EST and WGS databases from NCBI to obtain putative IbDCL and IbRDR genes of sweet potato. Then RNAi constructs were created as described in Method S1.

sRNA blot assays

sRNA blot assays were performed as described by Lin et al. (2012). The details of sRNA blot assays for sRNA8105, IbMYB1-siRNA and 5S rRNA are provided in Methods S1.

Gene expression analyses

Total RNAs were isolated using Trizol reagent (Invitrogen). Then total RNAs were treated with DNase I (Ambion, http://www.lifetechnologies.com), and reverse-transcribed to produce cDNAs using MMLV reverse transcriptase (Invitrogen) using primer T25VN (Table S3). The cDNAs were further amplified by quantitative PCR to determine the expression levels of IbMYB1 (AB258985), IbMYB2 genes (IbMYB2–1, AB258986; IbMYB2–2, AB258987; IbMYB2–3, AB258988; IbMYB2–4, AB258989), IbANS (EF187730), IbCAD1 (EF119213), IbCHI (JN083840), IbCHS (FJ478180), IbDFR (AB019243), IbF3H (EF108572), IbHCT (AB576768), IbPAL02 (M29232), IbPAL2–8 (D78640) and IbActin. The levels of IbActin were used as controls for quantitative comparison. The cDNAs were also amplified by PCR using primer pairs IbMYB1 F/IbMYB1 R and IbMYB1 F/IbMYB1 R2 to detect IbMYB1 alternative splicing (Table S3). Primers IbMYB1 F, IbMYB1 R and IbMYB1 R2 bind to the first exon, the second exon and the first intron of IbMYB1, respectively.

sRNA quantitative real-time PCR assay

DNase I-treated total RNA (1 μg) was reverse-transcribed using an miScript II reverse transcription kit (Qiagen, http://www.qiagen.com) to produce cDNAs. Then the cDNAs were further amplified using iQ™ SYBR Green Supermix (Bio–Rad, http://www.bio–rad.com) with specific primers and universal primer (Qiagen).

DNA methylation assays

The DNA methylation status was analyzed by McrBC-PCR as described by He et al. (2009). Genomic DNA (500 ng) isolated using Trizol reagent (Invitrogen) was digested using the McrBC enzyme (NEB), a methylation-dependent endonuclease, at 37°C for 1 h. The digested DNA was further amplified by PCR using primer pairs IbMYB1 F/IbMYB1 R4 and IbMYB1 F/IbMYB2s R to analyze the methylation levels of IbMYB1 and IbMYB2 genes (Table S3), respectively. Undigested gDNA was simultaneously amplified by PCR as controls. The ratios of PCR values from digested DNAs to those from undigested DNAs indicate the methylation status.

Agrobacterium-mediated transient expression in tobacco

Transient expression of sRNA8105, mimic8105 and its target IbMYB1 in tobacco leaves was performed as described by Kim et al. (2009). The gDNA and total RNAs of these infiltrated leaves were isolated and analyzed for DNA methylation and gene expression, respectively. The details of the transient expression assay are provided in Methods S1.

Mapping of sRNA8105-guided cleavage sites

A modified RNA ligase-mediated RACE method (Kasschau et al., 2003; Lin et al., 2012) was used to map sRNA8105-guided IbMYB1 cleavage sites. Details are provided in Methods S1.

sRNA sequencing

The Illumina platform (Illumina, http://www.illumina.com/) was used to sequence sRNA libraries created from total RNA, and was performed by Welgene Biotech (http://www.welgene.com.tw). The Map with Bowtie for Illumina method on the Galaxy website (https://main.g2.bx.psu.edu/) was used to analyze sRNA sequencing data. The raw data from sRNA deep sequencing were uploaded to the Gene Expression Omnibus (http://www.ncbi.nlm.nih.gov/geo) with accession number GSE46457.

Lignin content assays

The thioglycolic acid lignin assay was used to quantify lignin content (Lange et al., 1995).

Acknowledgements

This work was supported by the Taiwan National Science Council (grant numbers 99-2313-B-002-005-MY3 and 98-2311-B-002-011-MY3) and by the National Taiwan University (grant numbers 10R80917-4 and 101R892004) to S.T. Jeng. We wish to thank Dr. Tzyy-Jen Chiou (Academia Sinica, Taipei, Taiwan) for helpful comments of Northern blotting and Technology Commons (http://techcomm.lifescience.ntu.edu.tw/, College of Life Science, National Taiwan University) for technical support.

Ancillary