The Arabidopsis thaliana double-stranded RNA binding protein DRB1 directs guide strand selection from microRNA duplexes

Abstract

In Arabidopsis thaliana (Arabidopsis), DICER-LIKE1 (DCL1) functions together with the double-stranded RNA binding protein (dsRBP), DRB1, to process microRNAs (miRNAs) from their precursor transcripts prior to their transfer to the RNA-induced silencing complex (RISC). miRNA-loaded RISC directs RNA silencing of cognate mRNAs via ARGONAUTE1 (AGO1)-catalyzed cleavage. Short interefering RNAs (siRNAs) are processed from viral-derived or transgene-encoded molecules of double-stranded RNA (dsRNA) by the DCL/dsRBP partnership, DCL4/DRB4, and are also loaded to AGO1-catalyzed RISC for cleavage of complementary mRNAs. Here, we use an artificial miRNA (amiRNA) technology, transiently expressed in Nicotiana benthamiana, to produce a series of amiRNA duplexes with differing intermolecular thermostabilities at the 5' end of duplex strands. Analyses of amiRNA duplex strand accumulation and target transcript expression revealed that strand selection (amiRNA and amiRNA*) is directed by asymmetric thermostability of the duplex termini. The duplex strand possessing a lower 5' thermostability was preferentially retained by RISC to guide mRNA cleavage of the corresponding target transgene. In addition, analysis of endogenous miRNA duplex strand accumulation in Arabidopsis drb1 and drb2345 mutant plants revealed that DRB1 dictates strand selection, presumably by directional loading of the miRNA duplex onto RISC for passenger strand degradation. Bioinformatic and Northern blot analyses of DCL4/DRB4-dependent small RNAs (miRNAs and siRNAs) revealed that small RNAs produced by this DCL/dsRBP combination do not conform to the same terminal thermostability rules as those governing DCL1/DRB1-processed miRNAs. This suggests that small RNA processing in the DCL1/DRB1-directed miRNA and DCL4/DRB4-directed sRNA biogenesis pathways operates via different mechanisms.

FIGURE 1.

Vectors used in this study. (A) Schematic representation of the amiRNA plant expression vector pBlueGreen. The endogenous MIR159b precursor transcript sequence (pre-miRNA) was replaced with the LacZ gene (blue box) to separate 5′ and 3′ arms of the MIR159b primary transcript (light orange boxes). PCR product of the MIR159b precursor transcript (dark orange box) containing amiRNA guide and passenger strand sequences and flanked by SapI restriction sites (S) was amplified and cloned into the pri-miRNA sequence (replacing the LacZ gene) using the corresponding SapI restriction sequences of the pBlueGreen vector to produce the amiRNA plant expression vectors. (B) Schematic representation of the GUS sense (pART27:GUS-S) and GUS antisense (pART27:GUS-AS) plant expression vectors. The GUS transgene, either in the sense or antisense orientation (light blue boxes), was constitutively expressed by the Cauliflower Mosaic Virus 35S promoter (35S-P; yellow boxes). (Dark blue, dashed-line box) Position of the recognition site of the transgenes targeted by the series of amiRNA strand choice vectors. (C) The RNA sequences of the GUS sense and antisense target transgene flanking the strand choice amiRNA cleavage site (arrows). The perfectively matched sequences of mature amiRNA processed from amiRNA strand choice vectors SC-7 and SC-8 are given below their respective mRNA targets. (D) Hairpin RNA transgene of the GF portion of GFP.

FIGURE 2.

Target transgene expression in GUS strand choice amiRNA transformants. (A–I) Expression of the GUS sense (blue colored columns) and GUS antisense (red colored columns) target transgenes in amiRNA transformant lines. (Bolded, underlined sequence in A–E,G,H) Preferentially selected duplex strand from each of the seven asymmetrical amiRNA strand choice vectors analyzed in this study. Each sample was normalized to the respective selectable marker genes of the amiRNA vector (Basta) and the target transgene vector (kanamycin). The expression of all analyzed transcripts was also normalized to the Arabidopsis gene Cyclophilin (At2g29960). The relative expression level of each target transgene was then compared with the expression levels of controls (GUS sense or antisense target transgene alone) to determine the silencing efficiency of each GUS strand choice amiRNA duplex. Error bars represent the standard error of the mean (SEM) between three biological replicates.

FIGURE 3.

Small RNA accumulation in GUS strand choice amiRNA transformant lines. (A) Small RNA accumulation (amiRNA duplex strand) in amiRNA transformant lines infiltrated with the GUS sense target transgene alone and in combination with the nine amiRNA duplexes, SC-1–SC-9. Total RNA was probed with U6 (loading control), miR164 (internal control), and an amiRNA duplex antisense (−) strand-specific DNA oligonucleotide. (B) Small RNA accumulation (amiRNA* duplex strand) in amiRNA transformant lines infiltrated with the GUS antisense target transgene alone and in combination with the nine amiRNA duplexes, SC-1–SC-9. Total RNA was probed with U6 (loading control), miR164 (internal control), and an amiRNA duplex sense (+) strand-specific DNA oligonucleotide.

FIGURE 4.

Phenotypes, sRNA accumulation, and target transcript expression in wild-type Arabidopsis plants and in drb1 and ago1 T-DNA insertional mutant knockout lines. (A) Phenotypes expressed by 4-wk-old wild-type Arabidopsis plants (Col-0) and drb1 and ago1 T-DNA insertional mutant knockout lines. (B) Small RNA accumulation of miR159, miR167, and miR168 duplex strands (miRNA and miRNA*) in wild-type, drb1, and ago1 plant lines. (C) Transcript expression in wild-type, drb1, and ago1 plant lines for endogenous mRNAs, Myb33, Arf8, and Ago1, the target transcripts for miRNAs miR159, miR167, and miR168, respectively. Error bars represent the standard error of the mean (SEM) among three biological replicates.

FIGURE 5.

Phenotypes, sRNA accumulation, and target transcript expression in wild-type Arabidopsis plants and drb1 and drb2345 mutant lines. (A) Phenotypes expressed by 4-wk-old drb1 and drb2345 mutant lines compared with wild-type Arabidopsis plants (Col-0). (B) MicroRNA duplex strand accumulation (miRNA and miRNA*) in wild-type plants (Col-0) and in drb1 and drb2345 mutant lines for miR163, miR168, miR169, and miR408, Arabidopsis miRNAs processed from Class II miRNA duplexes. (C) MicroRNA duplex strand accumulation (miRNA and miRNA*) in T-DNA insertional mutant knockout lines drb1 and drb2345, compared with wild-type (Col-0), for miR160, miR161, miR164, and miR172, endogenous miRNAs processed from Class I miRNA duplexes. (D) Target transcript expression of miRNA-regulated mRNAs, Arf17 (miR160; processed from a Class II miRNA duplex) and Cuc2 (miR164; processed from a Class I miRNA duplex), in wild-type (Col-0), drb1, and drb2345 plants. Error bars represent the standard error of the mean (SEM) among three biological replicates.

FIGURE 6.

DCL4/DRB4-dependent sRNA accumulation in wild-type Arabidopsis plants and in drb1 and drb4 T-DNA insertional mutant knockout lines. (A) Phenotypes expressed by 4-wk-old wild-type (Col-0), drb1, and drb4 plants. (B) MicroRNA duplex strand accumulation (miRNA and miRNA*) of DCL4/DRB4-dependent miRNAs, miR822 and miR839, in wild-type Arabidopsis plants, and in drb1 and drb4 T-DNA insertional mutant knockout lines. (C) The abundance ratios (as percentages) of siRNA duplex strands derived from the introduced GFP hpRNA, and endogenous Arabidopsis miR/miR* duplex strands. All possible siRNA duplexes that could be generated from the GFP hpRNA were categorized by their 5′ terminal nucleotide (5′ terminal nucleotides of duplex strands designated X and Y). For siRNAs, each graphed point is calculated using the formula X strand/X+Y strand (×100). For miRNAs, the same formula is used where X = miRNA strand and Y = miRNA* strand. (Filled diamonds) siRNAs or miRNAs with the abundance predicted by the miRNA termini thermostability rules, (open diamonds) siRNAs or miRNAs present at abundances contrary to predictions based on the miRNA termini thermostability rules. (M) Mismatched bases, (W) weak G–U base pairing.

FIGURE 7.

DRB1-directed selection of miRNA guide strands. The upper pathway shows a schematic working model for the Arabidopsis DRB1 protein functioning in a similar fashion to the Drosophila siRNA-specific dsRBP, R2D2, directing the loading of miRNA duplexes to AGO1 for passenger strand degradation through its preferential binding to the more thermodynamically stable end of miRNA duplexes. The lower pathway shows the alternate working model for the combined action of DRB1 with DCL1, forming a heterodimer, similar to the RLC of the Drosophila siRNA biogenesis pathway, directionally loading the miRNA duplex to AGO1 for passenger strand degradation.