DRB2, DRB3 and DRB5 function in a non-canonical microRNA pathway in Arabidopsis thaliana

Abstract

DOUBLE-STRANDED RNA BINDING (DRB) proteins have been functionally characterized in viruses, prokaryotes and eukaryotes and are involved in all aspects of RNA biology. Arabidopsis thaliana (Arabidopsis) encodes five closely related DRB proteins, DRB1 to DRB5. DRB1 and DRB4 are required by DICER-LIKE (DCL) proteins DCL1 and DCL4 to accurately and efficiently process structurally distinct double-stranded RNA (dsRNA) precursor substrates in the microRNA (miRNA) and trans-acting small-interfering RNA (tasiRNA) biogenesis pathways respectively. We recently reported that DRB2 is also involved in the biogenesis of specific miRNA subsets.1 Furthermore, the severity of the developmental phenotype displayed by the drb235 triple mutant plant, compared with those expressed by either drb2, drb3 and drb5 single mutants, or double mutant combinations thereof, indicates that DRB3 and DRB5 function in the same non-canonical miRNA pathway as DRB2. Through the use of our artificial miRNA (amiRNA) plant expression vector, pBlueGreen2,3 we demonstrate here that unlike DRB2, DRB3 and DRB5 are not involved in the dsRNA processing stages of the miRNA biogenesis pathway, but are required to mediate RNA silencing of target genes of DRB2-associated miRNAs.

The Arabidopsis thaliana DRB protein family

DOUBLE-STRANDED RNA BINDING (DRB) proteins have been isolated in viruses, prokaryotes and eukaryotes and regulate all aspects of RNA biology, including RNA synthesis, transport, processing, translation and degradation.^4-6 DRB1 and DRB4, two of the five closely related Arabidopsis thaliana (Arabidopsis) DRB family members,^7,8 assist DICER-LIKE (DCL) proteins, DCL1 and DCL4 to process structurally distinct double-stranded RNA (dsRNA) precursor substrates into microRNAs (miRNAs) and trans-acting small-interfering RNAs (tasiRNAs) respectively.^9,10 In the initial nuclear-localized processing stages of the miRNA biogenesis pathway, DRB1 assists DCL1 to select and accurately process a specific short segment of dsRNA, the miRNA/miRNA* duplex, from a much larger stem-loop dsRNA precursor.^11,12 Similarly, in tasiRNA biogenesis, DRB4 is required by DCL4 to accurately and efficiently process short, phased tasiRNA/tasiRNA* duplexes from long perfectly dsRNA substrates.^10,13 More recently, DRB2 was shown to be antagonistic to DRB4 in the production of RNA Polymerase IV (PolIV)-dependent siRNAs and synergistic to DRB4 in the biogenesis of DRB4-dependent miRNAs.¹⁴ We reported similar findings to demonstrate that DRB2 is also antagonistic and synergistic to DRB1 in the biogenesis of specific miRNA subsets.¹ This opposing, yet dual regulatory role of DRB2 on either DRB1- or DRB4-mediated small RNA (sRNA) biogenesis suggests that the collective action of all three nuclear-localized DRB family members is essential to ensure that the accumulation of these specific classes of endogenous sRNA is tightly controlled throughout Arabidopsis development.

In contrast to DRB1, DRB2 and DRB4, DRB3 and DRB5 are expressed in the cytoplasm, and their involvement in the parallel RNA silencing pathways of Arabidopsis remains to be determined.^1,8,14 Arabidopsis lines harboring loss-of-function mutations in either DRB3 or DRB5 (drb3 and drb5 plants respectively) are wild-type in appearance, as is the drb35 double mutant (Fig. 1A). However, when the drb2 mutation is added to drb35 by standard genetic crossing, the resulting drb235 triple mutant plant expresses a severe developmental phenotype, characterized by pale green lanceolate shaped rosette leaves with serrated margins (Fig. 1A) and fused inflorescence stems (Fig. 1B). The similarity of the drb235 phenotype to those expressed by dcl1, drb1 and ago1 plants: loss-of-function mutant plant lines of the miRNA machinery proteins DCL1, DRB1 and ARGONAUTE1 (AGO1) respectively, prompted us to investigate miRNA accumulation in the triple mutant. In four week old Arabidopsis plants, DRB2, DRB3 and DRB5 expression is concentrated and overlapping in the shoot apical meristem (SAM) region and rosette leaf petioles.^1,8 Bioinformatic analysis of the sRNA populations extracted from these tissues of Col-0 and drb235 plants revealed that the accumulation of miRNA subsets was altered in the triple mutant.¹ Subsequent molecular analyses performed on drb2, drb3 and drb5 single and double mutant combinations clearly demonstrated that alterations to miRNA accumulation, either enhancement or reduction thereof, were a result of changes to the processing efficiency of the dsRNA precursor transcript in the absence of DRB2 activity.¹ In accordance with their expression of wild-type-like phenotypes, no change in precursor transcript processing efficiency, miRNA accumulation or target gene expression was observed in drb3, drb5 or drb35 plants.¹ However, several lines of evidence indicate that in Arabidopsis, DRB3 and DRB5 also function in the same non-canonical miRNA pathway as DRB2. These lines of evidence include:

Figure 1. Phenotypes displayed by drb knockout mutants. The drb235 developmental phenotype, characterized by pale green lanceolate shaped rosette leaves with serrated margins and fused inflorescence stems, is specific to the drb235 triple mutant plant. (A) Phenotypes expressed by four week old wild-type (Col-0) and drb knockout mutant plant lines. The drb35 double mutant is essentially wild-type in appearance. The drb13, drb15 and drb135 mutants all express the drb1 single mutant phenotype. The drb23 and drb25 double mutants express a slightly enhanced drb2 mutant phenotype and the drb235 triple mutant displays a severe developmental phenotype. Scale bars = 5 mm. (B) A section of the inflorescence stem sampled from six week old Col-0 plants and triple mutants, drb135 and drb235. Scale bars = 5 mm.

i) DRB2, DRB3 and DRB5 have high amino acid sequence identity and their expression overlaps in vegetative tissues of wild-type Arabidopsis plants;

Our initial bioinformatic assessment of the five DRB proteins identified in the Arabidopsis proteome revealed that DRB2, DRB3 and DRB5 are highly similar at the amino acid level, forming a single phylogenetic clade that is distinct to DRB1 and DRB4.⁸ Expressing the β-glucuronidase (GUS) reporter gene under the control of the putative promoter sequences of DRB2, DRB3 and DRB5 revealed that the expression of these three closely related DRB family members overlaps in the vegetative tissue of wild-type Arabidopsis plants. In four week old DRB2pro:GUS, DRB3pro:GUS and DRB5pro:GUS plants (transformed Arabidopsis lines expressing the DRB2, DRB3 and DRB5 promoter driven GUS reporter gene plant expression vectors), GUS expression is concentrated in the shoot apical meristem (SAM) region and rosette leaf petioles.^1,8 Taken together, these analyses strongly suggest that DRB2, DRB3 and DRB5 have a similar biochemical function, or are involved in the same biological pathway in Arabidopsis.

ii) The drb235 triple mutant expresses a severe developmental phenotype, however drb2, drb3 and drb5 single or double mutant plants express wild-type-like or mild phenotypes;

Our previous molecular analyses, either at the RNA or sRNA level, have shown that the involvement of DRB2 in the miRNA pathway is restricted to specific tissues and miRNA subsets.¹ Consistent with these findings, drb2 plants express a mild phenotype, primarily characterized by changes to rosette leaf structure and morphology (Fig. 1A).^1,8 Phenotypic assessment of drb2, drb3 and drb5 single and double mutants, showed that drb3, drb5 and drb35 plants are essentially wild-type in appearance (Fig. 1A; compare drb35 to Col-0). Furthermore, Figure 1A shows that the drb2 phenotype is only weakly enhanced in either the drb23 or drb25 double mutant. However, drb235 triple mutant plants express a severe developmental phenotype. The severity and specificity of the developmental phenotype expressed by drb235 plants (the pale green rosette leaf coloration and fusion of inflorescence stems is specific to the drb235 triple mutant), again suggests that all three of these DRB proteins function in the same sRNA-mediated gene regulatory pathway in Arabidopsis.

Unlike DRB2, the involvement of DRB1 in the Arabidopsis miRNA pathway is not limited to specific tissues or miRNAs. DRB1 is required by DCL1 for the accurate and efficient processing of the dsRNA precursor transcripts of almost all known Arabidopsis miRNAs, in all tissues and throughout development.^15,16 Curiously however, the addition of either the drb3 and drb5 single mutations, or the drb35 double mutant to drb1, does not appear to further alter the development of drb1 plants. Figure 1 clearly illustrates that drb13, drb15 and drb135 mutant plants all essentially display the drb1 phenotype. Taken together, the rudimentary interpretation of these results is that DRB3 and DRB5 have a strong genetic interaction with DRB2, but a comparatively weak interaction with DRB1.

iii) Artificial miRNA-directed RNA silencing is defective in specific tissues in drb2, drb3 and drb5 single mutant plants, but is completely lost in the drb235 triple mutant;

Bioinformatic analysis of the sRNA populations sampled from the SAM region and rosette leaf petioles of Col-0 and drb235 plants revealed that miRNA accumulation was reduced in the triple mutant for approximately 40% of the 57 MIR gene families detected by sRNA sequencing (Table S3 in Eamens et al.¹). The Arabidopsis MIR159 family, which consists of three members, MIR159A, MIR159B and MIR159C, is one such MIR gene family. Sequencing showed that the accumulation of the sRNA product (miR159a, miR159b and miR159c) derived from all three MIR159 loci was reduced (Table S1 in Eamens et al.¹) by approximately 2.2-fold in drb235 plants (Col-0; 5194 reads, drb235; 2314 reads).¹ northern blotting was therefore used to assess miR159 accumulation in the SAM region and rosette leaf petioles of wild-type Col-0 plants and drb1, drb2, drb3, drb5 and drb235 mutant lines. Figure 2A shows that miR159 accumulation is reduced in drb1, drb2 and drb235 plants, but remains at approximate wild-type levels in drb3 and drb5. As previously demonstrated for other MIR gene families with reduced accumulation in drb235 plants,¹ RT-PCR analysis of PRI-MIR159B expression, the precursor transcript of miR159b, revealed that reduced miR159 accumulation in drb235 plants (and drb2 plants) is a result of inefficient precursor transcript processing in the absence of DRB2 activity (Fig. 2A).

Figure 2. Artificial miRNA-directed RNA silencing in drb mutants. (A) northern blot and RT-PCR analyses revealed that miR159 accumulation is reduced in the SAM region and rosette leaf petioles of four week old drb1, drb2 and drb235 plants due to inefficient processing of the precursor transcript PRI-MIR159B. Panels 1 and 2, sRNA northern blotting to assess the accumulation of the miR159 sRNA and U6 loading control. Panels 3 and 4, RT-PCR assessment of the expression of the PRI-MIR159B precursor transcript and the ACTIN reference gene. (B) Artificial miRNA-directed silencing of PDS in Col-0 plants and drb mutants. RNA silencing of PDS is only defective in drb2, drb3 and drb5 single mutants only in the specific tissues where DRB2, DRB3 and DRB5 are expressed in wild-type plants. However, amiRNA-directed silencing of PDS is completely lost in drb1 and drb235 plants. Scale bars = 5 mm. (C) northern blot analysis of amiR-PDS accumulation and RT-PCR assessment of PRI-MIR159B processing efficiency and PDS expression in whole plant samples of Col-0 plants and drb mutants. Panels 1 and 2, sRNA northern blotting to assess the accumulation of the amiR-PDS sRNA and the U6 loading control. Panels 3 and 4, RT-PCR assessment of the expression of the PRI-MIR159B-PDS precursor transcript and the amiR-PDS target gene PDS. Panel 5, RT-PCR analysis of the reference gene ACTIN.

Our artificial miRNA (amiRNA) plant expression vector, pBlueGreen^2,3 uses the Arabidopsis PRI-MIR159B sequence to deliver the amiRNA sRNA and was therefore used to assess amiRNA-directed RNA silencing of the endogenous target gene PHYTOENE DESATURASE (PDS) in wild-type plants and drb mutant lines. Figure 2B shows that Col-0 plants transformed with the PDS-targeting amiRNA vector (amiR-PDS) are completely photo-bleached. However, drb1/amiR-PDS plants, which are defective in DRB1 activity, the canonical partnering protein of DCL1 in the miRNA pathway, only develop green tissues. When compared with wild-type plants, the molecular data presented in Figure 2C (sRNA northern blotting and RT-PCR), clearly shows that amiRNA-directed RNA silencing is defective in drb1/amiR-PDS transformant lines due to inefficient PRI-MIR159B-PDS processing in the absence of DRB1 activity, resulting in significantly reduced amiR-PDS accumulation and PDS silencing. Interestingly, amiRNA-directed RNA silencing of PDS is also completely abolished in the drb235 triple mutant, with drb235/amiR-PDS plants indistinguishable to the parental line (compare the drb235 plants presented in Figure 2B with Figure 1A). AmiRNA-directed RNA silencing is also defective in drb2, drb3 and drb5 single mutants, but only in the tissues where these three DRB proteins are expressed in wild-type Arabidopsis plants (Fig. 2B).^1,8 Unlike in drb2/amiR-PDS plants, the tissue-specific loss of PDS silencing in drb3/amiR-PDS and drb5/amiR-PDS transformant lines does not correlate with any change to PRI-MIR159B-PDS processing efficiency, amiR-PDS accumulation or PDS expression (Fig. 2C). However, although no change in the level of amiRNA-directed RNA silencing is observed in drb235/amiR-PDS plants, when compared at the molecular level with drb2/amiR-PDS plants, the tissue-specific defects in amiRNA-directed RNA silencing in drb2, drb3 and drb5 single mutants, compared with the complete loss of PDS silencing in the triple mutant, indicates that DRB3 and DRB5 are required to regulate the expression of genes targeted by DRB2-associated miRNAs.

Concluding remarks

We recently demonstrated that in the SAM region and rosette leaf petioles of Arabidopsis plants, DRB2 is involved in the biogenesis of specific miRNA subsets.¹ Here, through the use of our amiRNA plant expression vector, pBlueGreen,^2,3 which is based on the PRI-MIR159B precursor transcript of the DRB2-associated miRNA, miR159b, we provide strong evidence that DRB3 and DRB5 function in the same non-canonical miRNA pathway as DRB2. Phenotypic analyses suggest that the involvement of these two family members in the miRNA pathway is restricted to DRB2-associated miRNAs, with drb13, drb15 and drb135 mutants all displaying the classical drb1 phenotype. This is in direct contrast to the severe developmental phenotype expressed by the drb235 triple mutant, when compared with the wild-type or mild phenotypes displayed by drb2, drb3 and drb5 single and double mutant plants.

Curiously, at the molecular level, the results presented here also indicate that DRB3 and DRB5 are required to regulate the expression of target genes of DRB2-associated miRNAs via a cleavage-independent mechanism of RNA silencing (Fig. 2). The expression of DRB3 and DRB5 in the plant cell cytoplasm,¹ strongly identifies these two DRB family members as functioning downstream of the nuclear-localized dsRNA processing and miRNA maturation stages of the Arabidopsis miRNA pathway. Figure 3 compares and contrasts our proposed DRB2-mediated non-canonical miRNA pathway to the canonical DRB1-mediated miRNA pathway. In the DRB1-mediated pathway, following transcription of a non-protein-coding transcript from a MIR gene (black-colored lines) by RNA Polymerase II (PolII) and folding of this transcript to form the pri-miRNA, DCL1 with the assistance of DRB1 performs sequential cleavage events to liberate the miRNA/miRNA* duplex from the pri-miRNA and pre-miRNA precursor molecules. The miRNA guide strand is then separated from the miRNA* passenger strand and loaded onto AGO1-catalyzed RISC in the cytoplasm. AGO1 uses the loaded miRNA as a sequence-specificity guide to identify complementary mRNAs and cleaves these transcripts to regulate target gene expression. We propose that in the DRB2-mediated non-canonical miRNA pathway, DRB2 recognizes and preferentially binds structurally distinct miRNA precursor transcripts (red-colored lines). The sequential cleavage of these DRB2-associated precursor transcripts by DCL1 could identify the resulting miRNA for entry into an alternate miRNA pathway. In the cytoplasm, the DRB2-associated miRNA could then be shuttled to an alternate RISC complex that may contain a different AGO protein family member to AGO1 at its catalytic core. The cytoplasmic DRB proteins, DRB3 and DRB5 could complex with this alternate RISC that uses the loaded miRNA to guide a cleavage-independent mechanism of RNA silencing to regulate DRB2-associated miRNA target gene expression.

Figure 3. Proposed model of the DRB2-mediated non-canonical miRNA biogenesis in Arabidopsis. In the canonical miRNA pathway (represented by black colored lines), the pri-miRNA precursor transcript is transcribed from a MIR gene by RNA PolII. In the nucleus, the pri-miRNA is recognized by the DCL1/DRB1 partnership and undergoes two rounds of dsRNA cleavage, releasing the miRNA/miRNA* duplex. Following duplex strand separation and export to the cytoplasm, the miRNA is loaded onto RISC and used as a guide for AGO1-catalyzed cleavage of complementary mRNAs to regulate target gene expression. In the proposed DRB2-mediated non-canonical miRNA pathway (represented by red colored lines), following RNA PolII transcription, the pri-miRNA transcript is recognized and sequentially processed by the alternate DCL1/DRB2 partnership to liberate the miRNA/miRNA* duplex. The duplex strands are separated and the miRNA guide strand transported to the cytoplasm. The miRNA guide strand could then be loaded to; i) AGO1-catalyzed RISC to regulate target gene expression via the canonical mRNA cleavage-based mechanism of miRNA-directed RNA silencing in plants, or; ii) an alternate RISC complex containing DRB3 and DRB5 and that directs a translational repression mechanism of RNA silencing to regulate the expression of target genes of DRB2-associated miRNAs.

Unlike in plants, translational repression, and not target mRNA cleavage, is the predominant mechanism of miRNA-directed RNA silencing in animals.^17,18 However, recent studies have shown that in plants several proteins, or specific miRNAs, also direct a translational repression mechanism of RNA silencing for target gene expression regulation.^19-21 Taken together, the phenotypic and molecular data presented here identifies DRB3 and DRB5 as strong candidates for mediating a translational repression mechanism of miRNA-directed RNA silencing in Arabidopsis. Therefore, we have now developed a green fluorescent protein (GFP) reporter gene based system to functionally characterize, at the protein level, the involvement of DRB3 and DRB5 in regulating the expression of DRB2-associated miRNA target genes.

Materials and Methods

Vector construction

The amiRNA plant expression vector, pBlueGreen has been described previously.^2,3 The endogenous miR159b/miR159b* sequences of the PRI-MIR159B precursor transcript were replaced by PCR using primers, pAMIR-PDS-F (5′ - TATATGCTCTTCGAGAG GGTGCAACTTAAGGCTAGGATTGGAGGGTTTAGCAGGGTGAAGTAAAG - 3′) and pAMIR-PDS-R (5′ - TATATGCTCTTCGAGATGGTGCAACTTAATTCTAGGATAGAAGAG TGAAGCCATTAAAGGG - 3′). A plasmid preparation of the amiR-PDS plant expression vector was extracted from Escherichia coli (strain DH5α) and used to transform Agrobacterium tumefaciens (strain GV3101) via electroporation.

Plant material

The drb1, drb2, drb3, drb5, drb35 and drb235 T-DNA insertion loss-of-function mutant plant lines have been described previously by Curtin et al.⁸ The drb13, drb15 and drb135 mutants presented in Figure 1 of this study were generated by standard genetic crossing. Wild-type Col-0 plants and drb mutant lines were transformed with an Agrobacterium culture of the amiR-PDS plant expression vector by floral dipping as described by Clough and Bent.²² Transformant lines expressing the amiR-PDS plant expression vector were selected by germinating the dipped seed on plant growth media supplemented with 10 mg/mL of phosphinothricin. All Arabidopsis lines used in this study were grown under standard growth conditions of 16 h of light/8 h of dark at 24°C.

RNA analysis

For RNA analyses, sRNA northern blotting and RT-PCR, total RNA was extracted from the SAM region and rosette leaf petioles (Fig. 2A) or whole plant (Fig. 2C) samples using TRIzol Reagent according to the manufacturer’s instructions (Invitrogen). Northern blot analysis to assess miRNA and amiR-PDS accumulation was performed according to Eamens et al.¹ The DNA oligonucleotide probes used in this study to assess U6 (U6 probe), miR159 (miR159 probe) and amiR-PDS (amiR-PDS probe) accumulation, are; U6 probe (5′ - AGGGGCCATGCTAATCTTCTC - 3′); miR159 probe (5′ - TAGAGCTCCCTTCAATCCAAA - 3′), and; amiR-PDS probe (5′ - GTGCTTGAATTAAGACCTTAT - 3′). RT-PCR assessment of miRNA precursor transcript processing efficiency and target gene expression was performed according to Eamens et al.¹ All RT-PCRs were repeated three times and ACTIN (ACTIN2; AT3G18780) was used as the reference gene. RT-PCR products for ACTIN, PRI-MIR159B, PRI-MIR159B-PDS and PDS were amplified with primer pairs, pACT-RTF (5′ - TCTTCCGCTCTTTCTTTCCA - 3′)/pACT-RTR (5′ - GAGAGAACAGCTTGGATGGC - 3′), pPRI-159B-F (5′ - ATGGCTTCACTCTTCTTTGGA - 3′)/pPRI-159B-R (5′ - CCTACTCAAGA TCCATCATCC - 3′), pPRI-159B-F/pPRI-159B-PDS-R (5′ - GTGCTTGAATTAAGACCTTAT - 3′), and pPDS-RTF (5′ - CCAAACTGTGAACCATGTCG - 3′)/pPDS-RTR (5′ - AGCAGAATTTGCCAGAGAGG - 3′), respectively.

Glossary

Abbreviations:

amiRNA

artificial microRNA

dsRNA

double-stranded RNA

DRB protein

double-stranded RNA binding protein

miRNA

microRNA

PDS

PHYTOENE DESATURASE

Disclosure of Potential Conflicts of Interest

No potential conflicts of interest were disclosed.