RNA Silencing Induced by an Artiicial Sequence ... - Plant Physiology

Breakthrough Technologies

RNA Silencing Induced by an Artificial Sequence That Prevents Proper Transcription Termination in Rice1[W] Taiji Kawakatsu 2, Yuhya Wakasa 2, Hiroshi Yasuda 2,3, and Fumio Takaiwa* Genetically Modified Organism Research Center, National Institute of Agrobiological Sciences, Tsukuba, Ibaraki 305–8602, Japan

Posttranscriptional gene silencing (PTGS) is a sequence-specific mRNA degradation caused by small RNA, such as microRNA (miRNA) and small interfering RNA (siRNA). miRNAs are generated from MIRNA loci, whereas siRNAs originate from various sources of double-stranded RNA. In this study, an artificial RNA silencing inducible sequence (RSIS) was identified in rice (Oryza sativa). This sequence causes PTGS of 59 or 39 flanking-sequence-containing genes. Interestingly, two target genes can be simultaneously suppressed by linking a unique target sequence to either the 59 or 39 end of RSIS. Multiple gene suppression can be also achieved though a single transformation event by incorporating the multisite gateway system. Moreover, RSISmediated PTGS occurs in nuclei. Deep sequencing of small RNAs reveals that siRNAs are produced from RSIS-expressing cassettes and transitive siRNAs are produced from endogenous target genes. Furthermore, siRNAs are typically generated from untranscribed transgene terminator regions. The read-through transcripts from the RSIS-expression cassette were consistently observed, and most of these sequences were not polyadenylated. Collectively, this data indicates that RSIS inhibits proper transcription termination. The resulting transcripts are not polyadenylated. These transcripts containing RSIS become templates for double-stranded RNA synthesis in nuclei. This is followed by siRNA production and target degradation of target genes.

RNA silencing represents a powerful technology that has been successfully used for reverse genetics and crop improvement. RNA interference (RNAi) is generated via posttranscriptional gene silencing (PTGS). PTGS occurs when the artificial double-stranded RNA (dsRNA) of a target gene is processed into small interfering RNA (siRNA) molecules, triggering target mRNA degradation. This same process is observed during antisense RNA silencing, cosuppression (sense strand RNA silencing), inverted repeat-mediated RNA silencing, and virus-induced gene silencing (Ossowski et al., 2008). MiR-based artificial microRNA (miRNA) that mimics the endogenous miRNA pathway is another choice to produce RNA silencing (Schwab et al., 2006; Warthmann et al., 2008). The expression of antisense RNA results in dsRNA formation with sense mRNA (Waterhouse et al., 1998). In inverted-repeat-RNAi, dsRNA is easily formed within sense and antisense arrayed self-complementary

1 This work was supported by research grants from the Ministry of Agriculture, Forest, and Fisheries of Japan (Genomics and Agricultural Innovation no. GMC0003 to F.T.) and by a Grant-in-Aid for Young Scientists (grant no. 22688001 to T.K.) from the Ministry of Education, Culture, Sports, Science, and Technology of Japan. 2 These authors contributed equally to the article. 3 Present address: National Agricultural Research Center for Hokkaido Region, Sapporo, Hokkaido 062–8555, Japan. * Corresponding author; e-mail [email protected]. The author responsible for distribution of materials integral to the findings presented in this article in accordance with the policy described in the Instructions for Authors (www.plantphysiol.org) is: Fumio Takaiwa ([email protected]). [W] The online version of this article contains Web-only data. www.plantphysiol.org/cgi/doi/10.1104/pp.112.202689

regions (Smith et al., 2000). Virus-induced gene silencing generates dsRNA during viral RNA replication or by RNA-dependent RNA polymerase (RdRP) activity (Ratcliff et al., 2001; Vaistij and Jones, 2009). High expression of sense strand RNA is subject to RdRP, and forms dsRNA that leads to cosuppression (Elmayan and Vaucheret, 1996; Dalmay et al., 2000). mRNA that does not contain the proper 59 cap or poly (A) tail is recognized as aberrant mRNA. This mRNA is subjected to dsRNA formation as a template of RdRP and then degraded (Gazzani et al., 2004). Glucagon like peptide-1 (GLP-1), a 30-amino acid peptide hormone (7–36) secreted from intestinal L cells, promotes insulin secretion and lowers blood Glc, depending on blood Glc concentrations. In an attempt to accumulate GLP-1 peptides in rice (Oryza sativa) endosperm, a codon-optimized modified GLP-1 (mGLP-1) gene was expressed under the control of the endospermspecific glutelin GluB-1 promoter. This construct contained the GluB-1 59 untranslated region (UTR) and signal peptide, as well as the GluB-1 terminator containing the 39 UTR (Yasuda et al., 2005). This expression system using the GluB-1 promoter and terminator is routinely used for the expression of functional proteins/ peptides in rice endosperm (Kawakatsu and Takaiwa, 2010). However, cosuppression of endogenous GluB-1 as well as mGLP-1 was observed in the endosperm of transgenic rice plants (Yasuda et al., 2005; Kawakatsu et al., 2010). When the terminator was substituted with the globulin Glb-1 terminator, GluB-1 and Glb-1 were suppressed (Kawakatsu et al., 2010). Substitution of the promoter and terminator with those of 13-kD prolamin RM1 and RM2 resulted in silencing of the 13-kD prolamin genes (Kawakatsu et al., 2010). Additionally, other seed storage protein (SSP)-less lines were produced.

Plant PhysiologyÒ, October 2012, Vol. 160, pp. 601–612, www.plantphysiol.org Ó 2012 American Society of Plant Biologists. All Rights Reserved.

601

Kawakatsu et al.

Furthermore, even multiple target genes with no homology to each other can be simultaneously suppressed by expressing RNA silencing inducible sequence (RSIS) linked to unique sequences derived from the target genes. Using these lines, the mGLP-1 sequence was shown to act as an RSIS. The status of the transgene and endogenous target genes was characterized in the GluB-less variants, GluB$Glb-less and 13-kD Pro-less. siRNAs were generated from both transgenes and endogenous target genes, indicating that siRNAs derived from transgenes can trigger the degradation of target genes. This study reports that RSIS-mediated RNA silencing is PTGS and occurs in nuclei. Transgenes containing RSIS are consistently expressed as read-through transcripts, and a large quantity of the transcripts are not polyadenylated. Such aberrant transcripts can serve as templates of RdRPs, and synthesized dsRNAs are likely processed by DCL proteins to produce primary siRNAs. The results of this study indicate that RSIS inhibits proper transcription termination of RSIS-containing transcripts and enhances cosuppression.

RESULTS RSIS Induces RNA Silencing in Various Tissues

Previously, we generated transgenic rice in which the levels of SSP were reduced using RSIS. Eighteenkilodalton oleosin (Ole18) is specifically accumulated in the aleurone layer of endosperm and embryo in wildtype grains (Fig. 1B). Accumulation of Ole18 in the embryo and endosperm (aleurone layer) was severely depressed in transgenic rice with Ole18-less construct (Fig. 1A), in which Ole18 promoter and terminator were linked to RSIS (Fig. 1B). When the GluB-1 endospermspecific promoter and the Ole18 terminator were linked to RSIS (Fig. 1A, GluB$Ole18-less), accumulations of Ole18 and GluB subfamily glutelins were significantly suppressed in the endosperm of transformants (Fig. 1B). In contrast, accumulation of Ole18 in the embryo was not affected (Fig. 1B). RNA silencing derived from RSIS did not spread from the endosperm to the embryo. Such nonsystemic RNAi between endosperm and embryo has been reported in maize (Zea mays; Houmard et al., 2007).

Figure 1. RSIS-mediated silencing in various tissues. A, Diagrams of the constructs. SP, Signal peptide derived from each rice SSP. In all cases, the promoter contains the 59 UTR and the terminator contains the 39 UTR sequence. B, Suppression of Ole18 in embryo and endosperm. C, Suppression of Glx I in leaf, stem, and seed. D, Silencing of CYP90D2 in seedling. E, Semidwarf phenotype of D2-less plants at 1 week after germination. F, Leaf length of wild-type and D2-less plants. 1L, First leaf; 2LS, second leaf sheath; 2LB, second leaf blade; 3LS, third leaf sheath; 3LB, third leaf blade. * and *** indicate significant differences P , 0.05 and P , 0.001, respectively, between the wild type and D2-less (Student’s t test). 602

Plant Physiol. Vol. 160, 2012

RNA Silencing Inducible Sequence in Rice

Thirty-three-kilodalton Glyoxalase I (Glx I) is constitutively expressed in rice plants and is encoded by a single gene. Transformants containing Glx I-less construct had little Glx-I in leaves, stems, or seeds (Fig. 1, A and C). CYP90D2, that catalyzes the steps from 6-deoxoteasterone to 3-dehydro-6-deoxoteasterone and from teasterone to 3-dehydroteasterone in the late brassinosteroid biosynthesis pathway, is also expressed in rice seedling, and ebisu dwarf/d2/cyp90d2 shows semidwarf phenotype (Hong et al., 2003). CYP90D2 expression was suppressed in transformants containing D2-less construct, and they showed semidwarf phenotype (Fig. 1, A and D–F). These results indicate that RSIS-mediated RNA silencing is effective not only in seed organs, but also in vegetative tissues. Multiple Silencing by Combination of RSIS and MultiSite Gateway System

We have showed that two genes could be suppressed simultaneously by a single RSIS expression cassette in which sequences derived from different genes were ligated to the 59 and 39 ends of RSIS (Fig. 1; Kawakatsu et al., 2010). We have developed an expression system that allows the insertion of three expression cassettes into a single binary vector using the MultiSite Gateway (MSG) system (Wakasa et al., 2006). We could then assess whether multiple silencing is induced through a single transformation event by the combination of RSIS and MSG. We constructed RSIS expression cassettes with different SSP gene promoters containing 59 UTRs and terminators, then inserted them into a single binary vector. The 3-gluless is constructed

with different glutelin promoters (GluB-1, GluB-4, and GluA-2) and terminators (GluA-3, GluA-1, and GluB-2; Fig. 2). The sspless construct is composed of glutelin (GluA-2 and GluB-1) and 13-kD prolamin (RM1) promoters, and 13-kD prolamin (RM1 and RM2) and Glb-1 terminators (Fig. 2). Eleven of 60 independent transgenic lines (18%) containing the 3-gluless construct, were silenced at different levels for the three individual target genes. In four lines (7%), almost all glutelins were suppressed (Fig. 2). In the case of sspless, several SSPs were depressed in 31 of 60 independent transgenic lines (52%). Glutelins, prolamins, and globulin were reduced in 13 (21%), 21 (33%), and 27 (43%) lines, respectively. Seven lines (12%) contained reduced glutelins, prolamins, and globulin (Fig. 2). We compared the level of GluB-1 suppression by the single target construct GluB-less with that by multigene target constructs GluB$Glb-less, GluB$Ole18-less, 3-gluless, and sspless. In all lines, GluB-1 was suppressed to less than 5% of the wild type (Supplemental Fig. S1). Suppression level was most mild in 3-gluless, and strongest in GluB-less (Supplemental Fig. S1). Simultaneous suppression was less effective than single target suppression, but these levels are enough for many cases. We also compared suppression level among all transgenic plants containing GluB-less construct. In all suppressed lines, GluB-1 mRNA were suppressed by less than 1% of the wild type, and GluB-1 protein were reduced to undetectable level (Supplemental Fig. S2). In summary, multiple genes can be successfully suppressed by a single transformation event by combining RSIS with MSG. The RSIS-MSG combination can be a good method for generating multiple gene

Figure 2. Multiple silencing by a combination of RSIS and the MSG system. The construction strategy for binary vectors using the MSG LR clonase reaction, and resultant 3-gluless and sspless constructs are schematically represented. Total seed proteins were analyzed by SDS-PAGE. wt, Wild type; 3g, 3-gluless; Ssp, sspless. Plant Physiol. Vol. 160, 2012

603

Kawakatsu et al.

knockdown lines, substituting for conventional crossing between single knockdown lines. GluB-1 Silencing Induction Depends on the Intact RSIS Sequence in Transgenic Rice Seed

In an effort to determine factors that are critical for inducing RNA silencing, GluB-less construct derivatives with various modifications were expressed in rice. GluB-less (2RSIS), in which the RSIS sequence had been removed from GluB-less, did not induce RNA silencing (Fig. 3). GluB-less (GFP-C), in which RSIS was substituted with a comparable length of GFP (91 bp between 630 and 720 nucleotides [nt] of the GFP coding region), also failed to induce RNA silencing (Fig. 3). These results suggest that the RSIS sequence is necessary for silencing, and the observed silencing is sequence specific. Neither inversion of the RSIS sequence (anti), nor frame shift by insertion of a guanine residue at the beginning of the coding region (+G), significantly affected the silencing efficiency (53.3% and 35.7%, respectively, versus 54.2% of the original GluB-less), indicating that translation of RSIS is not required for silencing (Fig. 3). The 39 or 59 halves of RSIS were deleted (N-RSIS or RSIS-C), and RSIS was arranged as a tandem repeat of dimer (32) or trimer (33; Fig. 3). Induction of silencing was completely or almost completely abolished by deleting either half of the RSIS sequence (Fig. 3). By contrast, multiple repeats of RSIS decreased the silencing efficiency with the dimer (32) and eliminated the silencing with the trimer (33; Fig. 3). Thus, the size of the silencing construct seems important, with the optimal size likely ranging from 90 to 180 nt. Previous experiments showed that GFP-fused, and pentameric and hexameric forms of RSIS were expressed, and their

products accumulated to high levels in seed (Yasuda et al., 2006). Based on these results, the specific monomeric sequence of RSIS dictates silencing efficiency, and the intact RSIS sequence is critical for silencing. RSIS-Mediated RNA Silencing Is Nuclear PTGS

siRNAs were previously reported as being produced from the RSIS-expression cassette in GluB-less seed (Yasuda et al., 2005). To confirm whether specific siRNAs derived from the suppressed transcripts accumulate in transgenic plants, northern analysis was conducted to analyze small RNAs prepared from developing seeds of wild-type, GluB-less, GluB$Glb-less, and 13-kD Pro-less transgenic plants. siRNAs (21–24 nt) derived from target genes were accumulated: GluB-1 (in GluB-less and GluB$Glb-less), Glb-1 (in GluB$Glb-less), and RM1 (in 13-kD Pro-less; Supplemental Fig. S3). In addition to target gene-derived siRNAs, 21- to 24-nt siRNAs derived from RSIS also accumulated in all of the analyzed lines (Supplemental Fig. S3). These results suggest that RNAs containing RSIS sequence form dsRNAs that trigger siRNA biogenesis, followed by RNA silencing. PTGS is mediated by 21- and 22-nt siRNAs, whereas transcriptional gene silencing (TGS) is mediated by 24- to 26-nt siRNAs (Hamilton et al., 2002). TGS is associated with RNA-dependent DNA methylation (RdDM) within promoter regions. Since 24-nt siRNA is a hallmark of RdDM (Matzke et al., 2009), the cytosine methylation of the GluB-1 coding and promoter regions were analyzed in wild-type, GluB-less, and GluB$Glb-less plants via bisulfite sequencing using genomic DNA extracted from developing seeds at 10 to 15 d after flowering (DAF). However, only small differences in the degree of methylation were observed within the analyzed regions (Fig. 4, A and B). These regions contained 16 cytosines in the coding region and 30 cytosines in the promoter region. Next, the presence of unspliced GluB-1 pre-mRNA was investigated as evidence of GluB-1 transcription. PremRNA is typically very short lived, such that GluB-1 premRNA is undetected in total RNA extracted from seed (Fig. 5). On the other hand, when nuclei (nuclear RNA) were prepared, GluB-1 pre-mRNA was similarly detectable for both the wild-type and GluB-less transgenic rice (Fig. 5). These data suggest that the GluB-1 transcription rate was not affected by RSIS-mediated RNA silencing. By contrast, GluB-1 mRNA was significantly decreased in nuclear RNA extracted from a GluB-less plant compared with a wild-type plant (Fig. 5). These results suggest that RSIS induces nuclear PTGS via 21and 22-nt siRNAs, but not TGS associated with RdDM. Profiling of siRNAs from RSIS-Containing Transcripts and Target Transcripts

Figure 3. Silencing efficiency of GluB-less variant cassettes. Silencing efficiencies (the number of silenced plants/the number of independent transformants) are indicated to the right of the constructs. Red color indicates no silencing. 604

To understand the mechanisms underlying RSISmediated RNA silencing, high-throughput sequencing analysis of small RNA was performed. Total small RNAs were extracted from developing seeds of the Plant Physiol. Vol. 160, 2012


Figure 4. DNA methylation levels within the GluB-1 genomic region. A, Schematic diagram of GluB-1 locus. Trigger regions used for GluB-less construct are indicated by black bars. Bisulfite sequenced regions (BS) and methylation-sensitive PCR-amplified regions (MSP) are indicated by blue and green bars, respectively. AluI sites are indicated in red. B, Bisulfite sequence of GluB-1 locus. Black, gray, and white bars indicate methylation levels at CpG, CpHpG, and CpHpH. C, Methylation-sensitive PCR of GluB-1 locus.

wild type, GluB-less, GluB$Glb-less, and 13-kD Proless transgenic plants at 10 to 15 DAF. These small RNAs were subsequently sequenced by the Illumina genome analyzer IIx. siRNAs of 24 nt were most abundant in each line (Fig. 6). In the wild type, reads were rarely mapped on RSIS-containing constructs. In transgenic lines, reads that perfectly matched the RSIScontaining constructs both in sense and antisense strands accounted for 1.5% to 4.1% of the total reads (Fig. 6). Over 90% of the reads were 21-, 22-, or 24-nt in length (Fig. 6). The mapped reads in both strands were rarely distributed on the untranscribed promoter region (Fig. 6). In GluB-less transgenic rice, the distribution patterns were similar among 21-, 22-, and 24-nt siRNAs (Fig. 6; Supplemental Fig. S4). Mapped reads exhibited bias for the sense strand in each line (Fig. 6). The majority of the siRNAs mapped in the sense strand located in the 39 flanking region of RSIS, whereas certain amounts of siRNAs in the antisense strand were mapped on the first half of RSIS in GluB-less plants (Fig. 6). siRNAs that mapped on RSIS in the sense strand were uniformly distributed (Fig. 6). In GluB$Glb-less and 13-kD Pro-less transgenic rice, the distribution pattern of siRNAs showed tendencies similar to GluB-less, with the exception that the siRNAs mapped in the antisense strand peaked in the 59 flanking region of RSIS in the 13-kD Pro-less transgenic rice (Supplemental Fig. S4). Next, the 18- to 25-nt reads were mapped on GluB-1, Glb-1, RM1, and RM2 (Fig. 7). Reads were mapped on suppressed genes (GluB-1 in GluB-less and GluB$Glbless, Glb-1 in GluB$Glb-less, RM1 and RM2 in 13-kD Pro-less) in both sense and antisense strands, suggesting that these genes were suppressed by transitive RNAi machinery to some extent. Unexpectedly, there were small RNA reads mapped on these SSP genes even in wild-type and nonsilenced lines (Fig. 7). However, these reads were mapped in only sense strand. We found small RNA reads library from another study (Zhu et al., 2008) were also mapped on these genes. Small RNA Plant Physiol. Vol. 160, 2012

reads from this study and another study were also mapped in only sense strand of genes highly expressed in endosperm. These small RNAs may be produced during preparing library for high-throughput sequencing because abundant RNAs are easily subject to contaminated nucleases. Although the 21- to 22-nt mapped reads from the GluB-less plant were distributed throughout the GluB-1 transcript, the 24-nt mapped reads only mapped to the trigger region used in the GluB-less construct (Supplemental Fig. S5). This suggests that the 24-nt reads were primarily derived from the transgene, rather than from endogenous GluB-1. Similarly, the 24-nt reads from GluB$Glb-less mapped almost exclusively to the 59 UTR and signal peptide region of GluB-1 and to the 39 UTR of Glb-1 (Supplemental Fig. S5). These sequences were all derived from the GluB$Glb-less construct (Supplemental Fig. S5). The 21- and 22-nt reads from the GluB$Glb-less construct that mapped on GluB-1 peaked at the 59 UTR, and gradually decreased toward the 39 distal region (Supplemental Fig. S5). By contrast,

Figure 5. Detection of unspliced GluB-1 pre-mRNA in nuclear RNA. A, Schematic representation of the GluB-1 locus and primer sets used in B. Boxes and the lines between them indicate exons and introns, respectively. White labels indicate UTRs. B, RT-PCR of GluB-1 mRNA and premRNA. With or without RT is indicated by +RT or 2RT, respectively. The three lanes indicate three biological replicates. 605

Kawakatsu et al.

Figure 6. Profiling of total small RNAs mapped on RSIS-expression constructs. Values are normalized to coverage per million reads. Sense and antisense polarity reads were plotted on the y axis in the positive and negative directions, respectively. Pie graphs indicate the relative abundance of different sizes of small RNAs in total reads, mapped reads in sense (+) and antisense (2) strands. Gray, light-purple, yellow, and light-green shades indicate UTR, signal peptide (SP) of endogenous target genes, RSIS, and right border of T-DNA (RB), respectively.

those mapped on Glb-1 peaked at the 39 UTR and gradually decreased toward the 59 distal region (Supplemental Fig. S5). Similarly, the 21- and 22-nt reads from 13-kD Pro-less that mapped on RM1 and RM2 peaked at the 59 UTR and 39 UTR, respectively, and gradually decreased toward the opposite distal region (Supplemental Fig. S5). These results suggest that 21- and 22-nt siRNA production spread from the trigger regions used in the RSIS-containing construct. RdDM on GluB-1 locus by 24-nt siRNAs derived from trigger regions was confirmed using methylationsensitive restriction enzyme (Fig. 4, A and C). Methylation at AluI site inhibits cutting of the genome DNA, and allow PCR amplification of the region. On the other hand, unmethylated AluI sites are cleaved, and PCR amplification is limited. MSP1 and MSP4 region within trigger regions were more efficiently amplified in GluBless and GluB$Glb-less, and in GluB-less, compared with the wild type, showing that MSP1 region is methylated both in GluB-less and GluB$Glb-less, but MSP2 region is methylated only in GluB-less (Fig. 4, A and C). MSP2 and MSP3 regions were efficiently or inefficiently amplified in the wild type, GluB-less, and GluB$Glb-less, suggesting that MSP2 region is similarly methylated and MSP3 region is not methylated in these lines (Fig. 4, 606

A and C). These results indicate methylation status was only affected within trigger regions. Consistent Read-Through Transcript from Constructs Containing RSIS

The RSIS-expression cassettes contain a terminator that includes a few dominant polyadenylation (A) signal (PAS) sites since most SSP transcripts are polyadenylated. The siRNAs mapped throughout the terminator region of the RSIS-expression cassettes beyond the PAS (Fig. 6). The siRNAs were mapped on the endogenous and transgene terminator and their distal region. Notably, siRNA did not map on the rice genome beyond the GluB-1 terminator (Fig. 7), whereas siRNAs did map on the right border of the transgene (Fig. 6). These results suggest that a large portion of the siRNA that mapped on the GluB-1 terminator was derived from the transgene. Since siRNAs are derived from dsRNA, this region is suspected to be stably transcribed. Although northernblot analysis did not detect any transgene or endogenous target transcripts, they were detected by PCR. Primer sets were designed to amplify corresponding regions between transgene and endogenous target gene locus (Fig. 8A). Forward primers were fixed, and PAS Plant Physiol. Vol. 160, 2012


Figure 7. Profiling of total small RNAs mapped on endogenous target genes. Values are normalized to coverage per million reads. Sense and antisense polarity reads were plotted on the y axis in the positive and negative directions, respectively. Horizontal black lines indicate the trigger regions common with the RSIS-expression cassettes. Gray and light-purple shades indicate UTR and coding regions, respectively. The scales are capped at 200.

located between reverse primers of region 1 and region 2. In GluB-less transgenic rice, transgene transcripts of an extended length greater than 400 bp from the dominant PAS (region 4) were detected (Fig. 8B). In contrast, corresponding region 4 of endogenous GluB-1 locus was scarcely expressed. Transgene transcripts of an extended length (region 2) were more efficiently amplified than the corresponding endogenous regions in GluB$Glb-less and 13-kD Pro-less transgenic rice (Fig. 8, C and D). These results suggest that readthrough transcriptions are enhanced for all transgene cassettes containing RSIS. Next, the status of the read-through transcripts was examined. To determine whether these aberrant transcripts contain a poly (A) tail, complementary DNAs (cDNAs) were reverse transcribed using oligo(dT) or random primers. In GluB-less transgenic rice, the relative amount of read-through transcripts amplified (region 2) in the randomly generated cDNA pool was about 26-times higher than that generated from the poly (A) tail (Fig. 8D). Furthermore, in GluB$Glb-less and 13-kD Pro-less transgenic rice, relative amounts of read-through transcripts in the randomly generated cDNA pools were approximately 10- and 12-times Plant Physiol. Vol. 160, 2012

higher than those generated from the poly (A) tail, respectively (Fig. 8D). These results suggest that the extended lengths of transgene transcripts largely have no poly (A) tail, or have a poly (A) tail at a far distal site. In Arabidopsis (Arabidopsis thaliana), unpolyadenylated transcripts serve as templates for RDR6. As a result, dsRNAs are produced (Luo and Chen, 2007), and 21- to 23-nt siRNAs are generated. Since RSIScontaining transcripts were unpolyadenylated, the relationship between RSIS-mediated RNA silencing and RDR6 activity was examined. Additionally, since both AGO7 and RDR6 act in the TAS pathway, the potential role of AGO7 in RNA silencing was analyzed (Adenot et al., 2006). The rice mutants shl2-6/rdr6 and shl4-1/ ago7 showed embryo-lethal or severe seedling defect phenotypes, but endosperm development is not affected (Nagasaki et al., 2007). Thus, heterozygous mutants of shl2-6/rdr6 and shl4-1/ago7 and 13-kD Proless transgenic rice were crossed. The recovery of silencing was analyzed in the developing endosperm of F2 seeds. Each single seed was individually genotyped, and the RM1 expression level was determined. However, even in shl2-6/rdr6 or shl4-1/ago7 backgrounds, the RM1 expression level was severely suppressed (Supplemental 607

Kawakatsu et al.

Figure 8. Transcription termination of RSIS-expression cassettes and endogenous target genes. A, The diagram of amplified region. B to D, Amplification of corresponding regions of endogenous target genes (e) and transgenes (t) terminator. B, GluBless. C, GluB$Glb-less. D, 13-kD Pro-less. E, Relative expression levels of region 2 of transgenes by quantitative RT-PCR using cDNA pools prepared using oligo(dT) or random primers. Expression levels were normalized relative to UBQ expression.

Fig. S6), indicating that RSIS-mediated RNA silencing does not require RDR6 or AGO7 activity. DISCUSSION

Plant endogenous siRNAs that are generated from dsRNAs are derived from internal hairpin structures of a transcript, hybridization with a natural antisense transcript, and synthesis by RdRPs (Ghildiyal and Zamore, 2009). No hairpin structure can be deduced from the RSIS sequence, and the linked sequences are not complementary. Thus, RNAs containing the RSIS sequence do not likely form internal dsRNA structures. In addition, since there are no RSIS-like sequences in the rice genome, hybridization with a natural antisense transcript is unlikely. As an alternative, RSIS may act like the TAS locus in that RSIS is targeted by miRNA: Dicer splices RNA transcripts containing RSIS, and aberrant RNAs without a 59 cap or 39 poly (A) tail are converted into dsRNA by RDR6 (Allen et al., 2005; Yoshikawa et al., 2005). However, no miRNA targeting RSIS sequence exists within the miRNA database. In addition to cleaved TAS transcripts, unpolyadenylated transcripts can also serve as templates for RDR6 in Arabidopsis (Luo and Chen, 2007). However, RSIS-mediated silencing did not require SHL2/RDR6 or SHL4/AGO7 activity (Fig. 8). RDR6 and SGS3 coordinately synthesize dsRNA in the cytoplasmic SGS3/RDR6 body (Glick et al., 2008; Elmayan et al., 2009; Kumakura et al., 2009). Recently, AGO7 was reported to function in the cytoplasm but not in nuclei (Jouannet et al., 2012). These facts support the data of this study, showing that RSIS-mediated RNA silencing is nuclear PTGS (Fig. 4). Although nuclear localization of AtRDR6 was reported (Luo and Chen, 2007; Hoffer et al., 2011), the function of RDR6 in nuclei may be different from that in the cytoplasm. RNA-containing RSIS may be loaded into an exogenous siRNA pathway, where it is recognized as aberrant RNA without a poly (A) tail. This aberrant RNA may be converted into dsRNA by nuclear-localized RdRPs that are also involved in siRNA biogenesis or antiviral silencing (Willmann et al., 2011). RDR1 and 608

RDR2 may be involved in such a function. AtRDR2 is localized within the nucleus (Pontes et al., 2006). However, the cellular localization of RDR1 has not been reported. The possibility that RDR6 is involved in RSISmediated silencing and that other RDRs can compensate for the absence of RDR6 cannot be excluded. miRNA- or siRNA-mediated silencing is generally considered to occur within the cytoplasm. The processing body is a key cellular structure, and ribonucleases degrade cleaved target gene transcripts (Xu and Chua, 2011). However, the data of this study indicate that RSIS-mediated PTGS occurs in nuclei (Fig. 5). Since transitive secondary siRNAs were bidirectionally generated from endogenous target genes (Fig. 7), a reasonable conclusion is that DCL proteins must be responsible for target degradation. The nuclear localization of DCL proteins implies that RSIS-mediated RNA silencing is nuclear PTGS (Xie et al., 2004; Hiraguri et al., 2005; Pontes et al., 2006; Song et al., 2007; Hoffer et al., 2011). In plants, transitive secondary siRNAs can be generated from target transgenes, such as GUS and GFP, but rarely from target endogenous genes (Miki et al., 2005). Nuclear PTGS in soybean (Glycine max) is accompanied by transitive secondary siRNAs, suggesting that the production of transitive secondary siRNAs and nuclear PTGS are linked (Hoffer et al., 2011). Five functional DCL protein genes exist in the rice genome (OsDCL1a, -2a, -3a, -3b, and SHO1/OsDCL4; Kapoor et al., 2008). OsDCL1 belongs to the same subgroup as Arabidopsis DCL1 and AtDCL2. OsDCL2 and -3 are similar to AtDCL3. SHO1/OsDCL4 is an ortholog of AtDCL4. Dicer-like proteins have distinct enzymatic activities, although there are some partially redundant biological functions (Xie et al., 2004). AtDCL1 and AtDCL4 produce approximately 21-nt siRNAs. AtDCL2 and AtDCL3 produce 22-nt and 24- to 26-nt siRNA, respectively. DCL1 is specifically involved in miRNA biogenesis both in Arabidopsis and rice (Kurihara and Watanabe, 2004; Liu et al., 2005). Since 21-, 22-, and 24-nt siRNAs were generated from the whole transcripts containing RSIS (Fig. 6), the resulting dsRNAs are likely processed by DCL2, DCL3, and DCL4. On the other hand, although 21- and 22-nt siRNAs were Plant Physiol. Vol. 160, 2012


distributed throughout the whole GluB-1 mRNA, 24-nt siRNAs only mapped to the trigger region. This suggests that endogenous GluB-1 mRNA was degraded by DCL2 and DCL4 (Fig. 7). Furthermore, production of 21- and 22-nt, or 24-nt siRNA occurs in distinct compartments within the nuclei. DCL proteins are localized in the nucleus. Notably, DCL3 is concentrated in the nucleolus, whereas DCL4 is excluded from the nucleoplasm (Pontes et al., 2006; Hoffer et al., 2011). Heterochromatin-forming 24-nt siRNA accumulate in the nucleolus (Pontes et al., 2006). It is possible that the 21- and 22-nt siRNA are generated in the nucleoplasm, while the 24-nt siRNA is generated in the nucleolus. The GluB-1 terminator robustly terminates transcription at dominant sites, especially in the endosperm (Yang et al., 2009). By contrast, read-through transcription was consistently observed in the transcripts containing RSIS in spite of using the sufficient length (0.65 kb) of the GluB-1 terminator. This indicates that RSIS inhibits proper transcription termination (Fig. 8). Recently, MORPHEUS’ MOLECULE1, involved in methylation-independent TGS, was reported to prevent aberrant read-through transcription (Zhou et al., 2010). Although evidence is lacking, the production mechanism of aberrant read-through transcripts caused by RSIS or MORPHEUS’ MOLECULE1 may be related to each other. The mechanism of mRNA 39 end formation is highly conserved among eukaryotes (Proudfoot, 2011). After RNA polymerase II (pol II) transcribes the PAS, containing the AAUAAA consensus and dinucleotides of guanine and uracyl (GU)-rich region, the polyadenylation complex (PAC) recognizes and cleaves the signal. This is followed by polyadenylation via poly(A) polymerase. Faithful 59-capped and 39-polyadenylated transcripts (mRNAs) are exported to the cytoplasm, where they are translated into proteins. Exonuclease degrades the residual 39 region of nascent RNA. When it reaches pol II, the pol II transcription terminates. In actuality, the transcription termination has plasticity. More than 50% of eukaryotic genes contain more than one poly (A) site (Meyers et al., 2004; Shen et al., 2008). Alternative polyadenylation is important for the regulation of gene expression (Wu et al., 2011). In Arabidopsis, the alternative polyadenylation of FCA, involved in flowering control, has been well studied. This involves at least two protein complexes. FY/cleavage and polyadenylation specificity factor are

responsible for recognizing the polyadenylation signal, and the cleavage stimulatory factor recognizes functional GU-rich sequence. The core components of PAC play important roles in the cleavage and polyadenylation (39-end processing) of precursor mRNA (Simpson et al., 2003; Liu et al., 2010). In addition, novel RNAbinding proteins have been identified as gene-specific transacting factors of poly (A) site selection (i.e. transcription termination site selection; Hornyik et al., 2010). Considering transcripts that contain RSIS, unidentified RNA-binding proteins may bind to RSIS and inhibit the recognition or cleavage of PAS by PAC. Considering that an equal length of GFP coding sequence did not cause silencing, and that the monomer RSIS was the most effective, the RSIS secondary or tertiary structure other than a hairpin structure may determine the recognition (Fig. 3). Certain structures flanking the RSIS may interfere with the accessibility by unknown RNA-binding proteins. The tandem repeated structure of triple RSIS had no silencing effect. Unconventional length, nonpolyadenylated transcripts containing RSIS are retained in nuclei. Without cleavage, pol II continues transcription and RSIS-containing transcripts are subjected to dsRNA synthesis by RdRP as aberrant RNA. This results in the generation of primary siRNA and target degradation. Future Perspectives

RSIS enhanced cosuppression by preventing proper transcription termination followed by dsRNA formation. A design rule for RSIS-based RNA silencing, especially how to choose trigger region, would be similar to the conventional RNAi system. Usually we use terminal regions of transcripts containing UTR (Table I). In many cases, we use regions harboring both UTR and 59 terminal coding region. Their ratio varied among constructs, but they work well. Since UTR region is relatively gene-specific region compared with coding region, UTR trigger would yield target-specific silencing. On the other hand, if family genes should be suppressed, conserved coding regions are good choice. It may be no matter even though either UTR or coding region is used as triggers. We succeeded in suppressing allergen gene family in rice endosperm using RSIS with about 100 bp of highly conserved coding region (Wakasa et al., 2012). According to trigger length, around 100 bp

Table I. Trigger region used for RSIS-expression cassettes and silencing efficiency Construct

59 Trigger

39 Trigger

GluB-less GluB$Glb-less 13-kD Pro-less 10-kD Pro-less 16-kD Pro-less Glx I-less Ole18-less GluB$Ole18-less D2-less

GluB-1 59 UTR (42 bp) + signal peptide coding region (57 bp) GluB-1 59 UTR (42 bp) + signal peptide coding region (57 bp) RM1 59 UTR (27 bp) + signal peptide coding region (63 bp) RP10 59 UTR (67 bp) + signal peptide coding region (72 bp) RP16 59 UTR (65 bp) + signal peptide coding region (57 bp) Glx I 59 UTR (124 bp) Ole18 59 UTR (63 bp) GluB-1 59 UTR (42 bp) + signal peptide coding region (57 bp) None

GluB-1 39 UTR (108 bp) Glb-1 39 UTR (153 bp) RM2 39 UTR (78 bp) RP10 39 UTR (102 bp) RP16 39 UTR (118 bp) Glx I 39 UTR (253 bp) Ole18 39 UTR (362 bp) Ole18 39 UTR (362 bp) CYP90D2 39 UTR (191 bp)


Silencing Efficiency

52.1% 40.0% 58.6% 40.0% 42.9% 50.0% 58.3% 36.4% 53.3%

(12/23) (6/15) (17/29) (6/15) (12/28) (8/16) (7/12) (4/11) (8/15) 609

Kawakatsu et al.

may be suitable and sufficient. Longer trigger region would interfere with formation of unknown structure of RSIS-containing transcripts, although about 300 bp 59 trigger could induce silencing (Yang et al., 2012). Surprisingly, we have not succeeded in inducing RSISmediated silencing of Arabidopsis or tomato (Solanum lycopersicum) genes (F. Takaiwa, unpublished data), and RSIS-mediated silencing works only in rice at present. We speculate that unknown structure of RSIScontaining transcripts is subject to dsRNA formation in rice, but not in other plants such as Arabidopsis or tomato. Rice gains the mechanism of RSIS-mediated silencing during evolution to remove aberrant transcripts, such as viruses. Relation between Arabidopsis and RSIS may be similar to that between nonhost plant and virus. We are looking for such silencing-inducible sequence in Arabidopsis. Further study is needed to determine the nature of RSIS structure and the unknown RSIS-binding protein. Such knowledge should help to elucidate the mechanisms of transcription termination. This should also help to tune up RSIS-mediated silencing system, and to avoid unintended cosuppression when transgenes are highly expressed to produce highvalue proteins/peptides in plants.

MATERIALS AND METHODS Construction of Chimeric Genes and Plant Transformation The details of the GluB-less, GluB$Glb-less, 13-kD Pro-less, 10-kD Pro-less, 16-kD Pro-less constructs have been described previously (Kawakatsu et al., 2010). For Glx I-less and Ole18-less constructs, promoter-containing 59 UTR and terminator-containing 39 UTR of Glx I and Ole18, respectively, were fused to RSIS. For GluB$Ole18-less construct, GluB-1 terminator was substituted by Ole18 terminator. For D2-less construct, cauliflower mosaic virus 35S promoter and CYP90D2 terminator were fused to RSIS. Trigger lengths were summarized in Table I. RSIS sequence of GluB-less was removed (2RSIS) or replaced by guanine-inserted RSIS at 59 end (+G), antisense RSIS (anti), halves of RSIS (N-RSIS: 1–45 bp of RSIS; RSIS-C: 46–90 bp of RSIS), 90-bp 39 end of GFP cDNA (GFP-C), or two (32) or three (33) tandem fused RSIS. Constructs were introduced into calli derived from a mature rice (Oryza sativa ‘Kita-ake’) seed via the Agrobacterium tumefaciens-mediated transformation as described previously (Goto et al., 1999).

RNA Extraction and Reverse Transcription PCR Immature seeds were frozen in liquid nitrogen and ground into powder. Total RNA was extracted in RNA extraction buffer (50-mM Tris-HCl [pH 9.0], 1% SDS, 0.1 M NaCl, 5 mM EDTA-2Na), followed by a minimum of threephenol-chloroform (1:1, v/v) extractions. Nuclear fractions of wild-type or GluB-less developing seed cells (15 DAF) were prepared with the plant nuclei isolation/extraction kit (Sigma-Aldrich). Nuclear RNA was extracted using the RNeasy plant mini kit (Qiagen). One microgram of total or nuclear RNA was treated with DNase I, and single-strand cDNA was synthesized using Superscript III reverse transcriptase (Invitrogen). Quantitative reverse transcription (RT)-PCR was performed using the SYBR Premix Ex TaqII (Perfect Real Time) kit (Takara) on an ABI Prism 7000 HT sequence detector (Applied Biosystems). The expression levels were normalized to the UBQ expression level. Semiquantitative RT-PCR was performed using Ex Taq polymerase (Takara). Amplification was conducted using 35 cycles for GluB-1 pre-mRNA, 25 cycles for GluB-1 mRNA, and 30 cycles to detect transgene transcriptional read through. Specific amplification of GluB-1 pre-mRNA and mRNA was confirmed by sequencing the PCR products. The primers used in this study are listed in Supplemental Table S1. 610

Database Analysis The secondary structure of the 90-nt RSIS sequence was deduced using RNA structure version 5.02 (Mathews et al., 1999, 2004). RSIS-like sequences were used as query sequences and searched by BLAST against the rice genome or individual genes. miRNA-targeting RSIS was searched in miRBase (http:// www.mirbase.org/).

Small RNA Gel Blot For the detection of siRNA, a small RNA fraction was concentrated from total RNA by polyethylene glycol precipitation. Three-microgram aliquots of small RNA were separated on 15% acrylamide gels and electroblotted onto Hybond N+ membranes (GE Healthcare, http://www.gelifesciences.co.jp). Digoxigenin-labeled antisense RNA probes were prepared from the full-length cDNAs of GluB1, Glb1, RM1, and RSIS. Blots were hybridized with digoxigenin-labeled RNA probes in PerfectHyb plus buffer (Sigma-Aldrich, http://www.sigmaaldrich.com) overnight at 37°C.

Methylation Analysis Genomic DNA was extracted from premature seed (14 DAF) using a cetyl trimethyl ammonium bromide method. Three micrograms of genomic DNA were denatured by 0.3 M NaOH at 37°C for 10 min. Denatured DNA was treated with sodium bisulfite in 300 mL of reaction mixture (32.5% [w/v] urea, 35% [w/v] sodium bisulfite, 0.5 mM hydroquinon). Amplification was conducted using 30 cycles of 95°C for 30 s, 55°C for 15 min, followed by 55°C for 15 h. The reaction mixture was purified with a Wizard DNA clean-up system (Promega, http://www.promega.co.jp), and purified DNA was eluted with 50 mL of water. Next, 5.5 mL of 3 M NaOH was added and the sample was incubated at 37°C for 15 min. The reacted DNA sample was repurified by ethanol precipitation, eluted with 20 mL of water, and used as a PCR template. PCR products were sequenced, and the methylation rate was estimated by calculating the percentage rate of Cys exchange. For methylation-sensitive PCR, genomic DNA was treated with AluI overnight, then 100-fold diluted. The primers used in this study are listed in Supplemental Table S1.

Small RNA Profiling and Data Analysis Total RNA (5 mg) was separated by denatured urea-PAGE, and 18- to 30-nt small RNA was extracted using the SmRNA sample prep kit (Illumina). After an adenylated adapter was ligated to the 39 end of the small RNA using T4 RNA ligase 2, truncated (Illumina), a tagged adaptor was ligated to the 59 end of the small RNA using T4 RNA ligase 1 (Illumina). Small RNA-containing adaptors was reverse transcribed and then amplified for 12 cycles. The PCR products were separated by PAGE, and approximately 100-bp DNA was extracted as a tagged DNA library. This was followed by sequencing using a Genome Analyzer IIx (Illumina). After adaptor sequences were removed, small RNA data were analyzed using CLC Genomics Workbench software (CLC bio). The 18- to 25-nt reads were extracted as small RNA, and the 21-, 22-, and 24-nt reads were separated. Reads were mapped to reference using the RNA-seq program sense strand specifically or nonspecifically, and counted. Complementary mapped reads were represented as their differences. Data were presented as coverage (number of reads that cover the base) per million reads. Sequence data from this article can be found in the GenBank/EMBL data libraries under accession number DL489594.1.

Supplemental Data The following materials are available in the online version of this article. Supplemental Figure S1. Simultaneous suppression level of GluB-1. Supplemental Figure S2. Suppression level among all transgenic plants containing GluB-less construct. Supplemental Figure S3. Detection of siRNA in the developing seed of the wild type, GluB-less, GluB$Glb-less, and 13-kD Pro-less. Supplemental Figure S4. Comparison of siRNAs mapped on RSIS-expressing constructs in GluB-less, GluB$Glb-less, and 13-kD Pro-less seeds. Supplemental Figure S5. Comparison of siRNAs mapped on the 39 distal region of the endogenous GluB-1 locus or RSIS-expression cassette. Plant Physiol. Vol. 160, 2012


Supplemental Figure S6. RSIS-mediated RNA silencing in shl2/rdr6 and shl4/ago7. Supplemental Table S1. Primers used in this study.

ACKNOWLEDGMENTS The authors would like to thank Drs. Jun-Ichi Itoh and Yasuo Nagato for their kind gift of shl2 and shl4 mutant seeds, Dr. Manabu Yoshikawa for discussion, and Ms. Yukie Ikemoto for technical assistance. Received June 26, 2012; accepted July 25, 2012; published July 27, 2012.

LITERATURE CITED Adenot X, Elmayan T, Lauressergues D, Boutet S, Bouché N, Gasciolli V, Vaucheret H (2006) DRB4-dependent TAS3 trans-acting siRNAs control leaf morphology through AGO7. Curr Biol 16: 927–932 Allen E, Xie Z, Gustafson AM, Carrington JC (2005) microRNA-directed phasing during trans-acting siRNA biogenesis in plants. Cell 121: 207–221 Dalmay T, Hamilton A, Rudd S, Angell S, Baulcombe DC (2000) An RNAdependent RNA polymerase gene in Arabidopsis is required for posttranscriptional gene silencing mediated by a transgene but not by a virus. Cell 101: 543–553 Elmayan T, Adenot X, Gissot L, Lauressergues D, Gy I, Vaucheret H (2009) A neomorphic sgs3 allele stabilizing miRNA cleavage products reveals that SGS3 acts as a homodimer. FEBS J 276: 835–844 Elmayan T, Vaucheret H (1996) Expression of single copies of a strongly expressed 35S transgene can be silenced post-transcriptionally. Plant J 9: 787–797 Gazzani S, Lawrenson T, Woodward C, Headon D, Sablowski R (2004) A link between mRNA turnover and RNA interference in Arabidopsis. Science 306: 1046–1048 Ghildiyal M, Zamore PD (2009) Small silencing RNAs: an expanding universe. Nat Rev Genet 10: 94–108 Glick E, Zrachya A, Levy Y, Mett A, Gidoni D, Belausov E, Citovsky V, Gafni Y (2008) Interaction with host SGS3 is required for suppression of RNA silencing by tomato yellow leaf curl virus V2 protein. Proc Natl Acad Sci USA 105: 157–161 Goto F, Yoshihara T, Shigemoto N, Toki S, Takaiwa F (1999) Iron fortification of rice seed by the soybean ferritin gene. Nat Biotechnol 17: 282–286 Hamilton A, Voinnet O, Chappell L, Baulcombe D (2002) Two classes of short interfering RNA in RNA silencing. EMBO J 21: 4671–4679 Hiraguri A, Itoh R, Kondo N, Nomura Y, Aizawa D, Murai Y, Koiwa H, Seki M, Shinozaki K, Fukuhara T (2005) Specific interactions between Dicer-like proteins and HYL1/DRB-family dsRNA-binding proteins in Arabidopsis thaliana. Plant Mol Biol 57: 173–188 Hoffer P, Ivashuta S, Pontes O, Vitins A, Pikaard C, Mroczka A, Wagner N, Voelker T (2011) Posttranscriptional gene silencing in nuclei. Proc Natl Acad Sci USA 108: 409–414 Hong Z, Ueguchi-Tanaka M, Umemura K, Uozu S, Fujioka S, Takatsuto S, Yoshida S, Ashikari M, Kitano H, Matsuoka M (2003) A rice brassinosteroid-deficient mutant, ebisu dwarf (d2), is caused by a loss of function of a new member of cytochrome P450. Plant Cell 15: 2900–2910 Hornyik C, Terzi LC, Simpson GG (2010) The spen family protein FPA controls alternative cleavage and polyadenylation of RNA. Dev Cell 18: 203–213 Houmard NM, Mainville JL, Bonin CP, Huang S, Luethy MH, Malvar TM (2007) High-lysine corn generated by endosperm-specific suppression of lysine catabolism using RNAi. Plant Biotechnol J 5: 605–614 Jouannet V, Moreno AB, Elmayan T, Vaucheret H, Crespi MD, Maizel A (2012) Cytoplasmic Arabidopsis AGO7 accumulates in membraneassociated siRNA bodies and is required for ta-siRNA biogenesis. EMBO J 31: 1704–1713 Kapoor M, Arora R, Lama T, Nijhawan A, Khurana JP, Tyagi AK, Kapoor S (2008) Genome-wide identification, organization and phylogenetic analysis of Dicer-like, Argonaute and RNA-dependent RNA polymerase gene families and their expression analysis during reproductive development and stress in rice. BMC Genomics 9: 451 Kawakatsu T, Hirose S, Yasuda H, Takaiwa F (2010) Reducing rice seed storage protein accumulation leads to changes in nutrient quality and storage organelle formation. Plant Physiol 154: 1842–1854 Plant Physiol. Vol. 160, 2012

Kawakatsu T, Takaiwa F (2010) Cereal seed storage protein synthesis: fundamental processes for recombinant protein production in cereal grains. Plant Biotechnol J 8: 939–953 Kumakura N, Takeda A, Fujioka Y, Motose H, Takano R, Watanabe Y (2009) SGS3 and RDR6 interact and colocalize in cytoplasmic SGS3/ RDR6-bodies. FEBS Lett 583: 1261–1266 Kurihara Y, Watanabe Y (2004) Arabidopsis micro-RNA biogenesis through Dicer-like 1 protein functions. Proc Natl Acad Sci USA 101: 12753–12758 Liu B, Li P, Li X, Liu C, Cao S, Chu C, Cao X (2005) Loss of function of OsDCL1 affects microRNA accumulation and causes developmental defects in rice. Plant Physiol 139: 296–305 Liu F, Marquardt S, Lister C, Swiezewski S, Dean C (2010) Targeted 39 processing of antisense transcripts triggers Arabidopsis FLC chromatin silencing. Science 327: 94–97 Luo Z, Chen Z (2007) Improperly terminated, unpolyadenylated mRNA of sense transgenes is targeted by RDR6-mediated RNA silencing in Arabidopsis. Plant Cell 19: 943–958 Mathews DH, Burkard ME, Freier SM, Wyatt JR, Turner DH (1999) Predicting oligonucleotide affinity to nucleic acid targets. RNA 5: 1458–1469 Mathews DH, Disney MD, Childs JL, Schroeder SJ, Zuker M, Turner DH (2004) Incorporating chemical modification constraints into a dynamic programming algorithm for prediction of RNA secondary structure. Proc Natl Acad Sci USA 101: 7287–7292 Matzke M, Kanno T, Daxinger L, Huettel B, Matzke AJ (2009) RNAmediated chromatin-based silencing in plants. Curr Opin Cell Biol 21: 367–376 Meyers BC, Tej SS, Vu TH, Haudenschild CD, Agrawal V, Edberg SB, Ghazal H, Decola S (2004) The use of MPSS for whole-genome transcriptional analysis in Arabidopsis. Genome Res 14: 1641–1653 Miki D, Itoh R, Shimamoto K (2005) RNA silencing of single and multiple members in a gene family of rice. Plant Physiol 138: 1903–1913 Nagasaki H, Itoh J, Hayashi K, Hibara K, Satoh-Nagasawa N, Nosaka M, Mukouhata M, Ashikari M, Kitano H, Matsuoka M, et al (2007) The small interfering RNA production pathway is required for shoot meristem initiation in rice. Proc Natl Acad Sci USA 104: 14867–14871 Ossowski S, Schwab R, Weigel D (2008) Gene silencing in plants using artificial microRNAs and other small RNAs. Plant J 53: 674–690 Pontes O, Li CF, Costa Nunes P, Haag J, Ream T, Vitins A, Jacobsen SE, Pikaard CS (2006) The Arabidopsis chromatin-modifying nuclear siRNA pathway involves a nucleolar RNA processing center. Cell 126: 79–92 Proudfoot NJ (2011) Ending the message: poly(A) signals then and now. Genes Dev 25: 1770–1782 Ratcliff F, Martin-Hernandez AM, Baulcombe DC (2001) Technical advance: tobacco rattle virus as a vector for analysis of gene function by silencing. Plant J 25: 237–245 Schwab R, Ossowski S, Riester M, Warthmann N, Weigel D (2006) Highly specific gene silencing by artificial microRNAs in Arabidopsis. Plant Cell 18: 1121–1133 Shen Y, Ji G, Haas BJ, Wu X, Zheng J, Reese GJ, Li QQ (2008) Genome level analysis of rice mRNA 39-end processing signals and alternative polyadenylation. Nucleic Acids Res 36: 3150–3161 Simpson GG, Dijkwel PP, Quesada V, Henderson I, Dean C (2003) FY is an RNA 39 end-processing factor that interacts with FCA to control the Arabidopsis floral transition. Cell 113: 777–787 Smith NA, Singh SP, Wang MB, Stoutjesdijk PA, Green AG, Waterhouse PM (2000) Total silencing by intron-spliced hairpin RNAs. Nature 407: 319–320 Song L, Han MH, Lesicka J, Fedoroff N (2007) Arabidopsis primary microRNA processing proteins HYL1 and DCL1 define a nuclear body distinct from the Cajal body. Proc Natl Acad Sci USA 104: 5437–5442 Vaistij FE, Jones L (2009) Compromised virus-induced gene silencing in RDR6-deficient plants. Plant Physiol 149: 1399–1407 Wakasa Y, Ozawa K, Takaiwa F (2012) Agrobacterium-mediated cotransformation of rice using two selectable marker genes derived from rice genome components. Plant Cell Rep (in press) Wakasa Y, Yasuda H, Takaiwa F (2006) High accumulation of bioactive peptide in transgenic rice seeds by expression of introduced multiple genes. Plant Biotechnol J 4: 499–510 Warthmann N, Chen H, Ossowski S, Weigel D, Herve P (2008) Highly specific gene silencing by artificial miRNAs in rice. PLoS ONE 3: e1829 611

Kawakatsu et al.

Waterhouse PM, Graham MW, Wang MB (1998) Virus resistance and gene silencing in plants can be induced by simultaneous expression of sense and antisense RNA. Proc Natl Acad Sci USA 95: 13959–13964 Willmann MR, Endres MW, Cook RT, Gregory BD (2011) The functions of RNA-dependent RNA polymerases in Arabidopsis. The Arabidopsis Book 9: e0146, doi/10.1199/tab.0146 Wu X, Liu M, Downie B, Liang C, Ji G, Li QQ, Hunt AG (2011) Genome-wide landscape of polyadenylation in Arabidopsis provides evidence for extensive alternative polyadenylation. Proc Natl Acad Sci USA 108: 12533–12538 Xie Z, Johansen LK, Gustafson AM, Kasschau KD, Lellis AD, Zilberman D, Jacobsen SE, Carrington JC (2004) Genetic and functional diversification of small RNA pathways in plants. PLoS Biol 2: E104 Xu J, Chua NH (2011) Processing bodies and plant development. Curr Opin Plant Biol 14: 88–93 Yang L, Hirose S, Takahashi H, Kawakatsu T, Takaiwa F (2012) Recombinant protein yield in rice seed is enhanced by specific suppression of endogenous seed proteins at the same deposit site. Plant Biotechnol J (in press)

612

Yang L, Wakasa Y, Kawakatsu T, Takaiwa F (2009) The 39-untranslated region of rice glutelin GluB-1 affects accumulation of heterologous protein in transgenic rice. Biotechnol Lett 31: 1625–1631 Yasuda H, Hayashi Y, Jomori T, Takaiwa F (2006) The correlation between expression and localization of a foreign gene product in rice endosperm. Plant Cell Physiol 47: 756–763 Yasuda H, Tada Y, Hayashi Y, Jomori T, Takaiwa F (2005) Expression of the small peptide GLP-1 in transgenic plants. Transgenic Res 14: 677–684 Yoshikawa M, Peragine A, Park MY, Poethig RS (2005) A pathway for the biogenesis of trans-acting siRNAs in Arabidopsis. Genes Dev 19: 2164–2175 Zhou Y, Zhang J, Lin H, Guo G, Guo Y (2010) MORPHEUS’ MOLECULE1 is required to prevent aberrant RNA transcriptional read-through in Arabidopsis. Plant Physiol 154: 1272–1280 Zhu QH, Spriggs A, Matthew L, Fan L, Kennedy G, Gubler F, Helliwell C (2008) A diverse set of microRNAs and microRNA-like small RNAs in developing rice grains. Genome Res 18: 1456–1465
