Determinants beyond both complementarity and cleavage govern microR159 efficacy in Arabidopsis

Abstract

Plant microRNAs (miRNAs) are critical regulators of gene expression, however little attention has been given to the principles governing miRNA silencing efficacy. Here, we utilize the highly conserved Arabidopsis miR159-MYB33/MYB65 regulatory module to explore these principles. Firstly, we show that perfect central complementarity is not required for strong silencing. Artificial miR159 variants with two cleavage site mismatches can potently silence MYB33/MYB65, fully complementing a loss-of-function mir159 mutant. Moreover, these miR159 variants can cleave MYB33/MYB65 mRNA, however cleavage appears attenuated, as the ratio of cleavage products to full length transcripts decreases with increasing central mismatches. Nevertheless, high levels of un-cleaved MYB33/MYB65 transcripts are strongly silenced by a non-cleavage mechanism. Contrary to MIR159a variants that strongly silenced endogenous MYB33/MYB65, artificial MYB33 variants with central mismatches to miR159 are not efficiently silenced. We demonstrate that differences in the miRNA:target mRNA stoichiometry underlie this paradox. Increasing miR159 abundance in the MYB33 variants results in a strong silencing outcome, whereas increasing MYB33 transcript levels in the MIR159a variants results in a poor silencing outcome. Finally, we identify highly conserved nucleotides that flank the miR159 binding site in MYB33, and demonstrate that they are critical for efficient silencing, as mutation of these flanking nucleotides attenuates silencing at a level similar to that of central mismatches. This implies that the context in which the miRNA binding site resides is a key determinant in controlling the degree of silencing and that a miRNA "target site" encompasses sequences that extend beyond the miRNA binding site. In conclusion, our findings dismiss the notion that miRNA:target complementarity, underpinned by central matches, is the sole dictator of the silencing outcome.

Figure 1. Artificial miR159a variants with up to two central mismatches are potent silencers of MYB33/MYB65.

(A) Structure of the MIR159a transgene used to generate miR159a variant constructs. Numbers indicating relative positions to the transcription start site. Black arrow: Transcription start site; Orange bar: Polyadenylation site. (B) The alignment between MYB33 and the miR159a variant sequences used to transform mir159ab. The percentage of free energy pairing with MYB33 compared to a perfect match is listed on the right. Green: original mismatches; Purple: corrected mismatches; Red: introduced mismatches. (C) Aerial view of rosettes from 30-day-old MIR159a0, MIR159a1 and MIR159a2 plants (mir159ab) grown side by side with Col-0 and mir159ab under long day conditions. Scale bar = 10 mm. (D) The mRNA levels of CP1 in the rosettes of mir159ab plants complemented by the MIR159 variants in comparison with wild type and mir159ab. The value for each variant is the average of three independent transgenic lines. All mRNA levels were normalized with CYCLOPHILIN 5. Error bars represent the SEM. (E) Mature miR159a levels in Col-0 and mir159ab, and mature miR159a variant levels in three independent lines of MIR159a0 (mir159ab), MIR159a1 (mir159ab) and MIR159a2 (mir159ab) transformants. All miR159a variant and miR159a levels were normalized with sno101 by comparative quantitation analysis. Error bars represent the SEM.

Figure 2. MiR159a1 and miR159a2 repress the endogenous MYB33/MYB65 mRNA levels.

(A) Un-cleaved MYB33 mRNA levels in three independent lines of MIR159a0 (mir159ab), MIR159a1 (mir159ab) and MIR159a2 (mir159ab) transformants; (B) Un-cleaved MYB65 mRNA levels in three independent lines of MIR159a0 (mir159ab), MIR159a1 (mir159ab) and MIR159a2 (mir159ab) transformants. All mRNA levels were normalized with CYCLOPHILIN 5. Measurements are the average of three technical replicates. Error bars represent the SEM. Similar results were obtained with two independent biological replicates.

Figure 3. MiR159a variants with up to two central mismatches can direct target cleavage at the canonical miR159 cleavage site.

(A) Schematic representation of MYB33 PCR products after a modified 5′-RACE procedure to determine the proportion of degraded MYB33 mRNA that corresponds to miR159-guided cleavage products. Red box: RNA Oligo adaptor; Yellow box: MYB33 sequences; Red arrow: canonical miR159 cleavage site; Green arrow: MYB33 sequencing primer. (B–E) Sequencing chromatographs of the cDNA-adaptor region from 5′-RACE recovered 3′ MYB33 transcripts in: (B) Col-0, (C) mir159ab, (D) MIR159a1 and (E) MIR159a2 plants. Peaks from MYB33 sequence downstream of the miR159 cleavage sites are highlighted in yellow, while those from the adaptor are left white. Three independent transgenic lines were checked for each construct and the results were identical.

Figure 4. Quantitative comparison of un-cleaved MYB33 mRNA and miR159-guided 3′-end MYB33 cleavage products in MIR159a1 and MIR159a2 variants.

Grey bars represent the un-cleaved MYB33 transcript levels in Col-0, mir159ab, and independent transgenic lines of MIR159a1 (mir159ab) and MIR159a2 (mir159ab). Black bars represent the levels of miR159-guided 3′-end MYB33 cleavage products in the same lines. Numbers show the relative ratio of un-cleaved transcripts to cleavage products. All mRNA levels were normalized with CYCLOPHILIN 5. Measurements are the average of three technical replicates. Similar results were obtained with two independent biological replicates. Error bars represent the SEM.

Figure 5. High levels of un-cleaved MYB33 transcripts remain strongly repressed.

(A) Aerial view of 29-day-old rosettes from MYB33 and mMYB33 transgenic plants (myb33 genotype) grown under short day conditions. Scale bar = 10 mm. (B) Alignments between miR159a and the wild type and mutated miR159 target sites in MYB33 and mMYB33 constructs used to transform Arabidopsis. Nucleotide changes made in mMYB33 are highlighted in red. Both MYB33 and mMYB33 encode the same protein. (C) Average rosette sizes of 29-day-old T2 transgenic plants from MYB33 lines grown side by side with Col-0 and myb33. Unit of measurement shown is relative pixel. Multiple plants for each line were grown on soil side by side with Col-0 and myb33. More than fifteen transgenic plants in each line were analysed. Error bars represent SD. (D) The steady state mRNA levels of un-cleaved MYB33 in the rosettes of multiple MYB33 (lines 1–6) and mMYB33 (lines 1–2) T2 transgenic lines. (E) Relative mRNA levels of CP1 in the rosettes of the same MYB33 and mMYB33 lines. All mRNA levels were normalized with CYCLOPHILIN 5. All measurements are the average of three technical replicates with error bars representing the SEM.

Figure 6. Quantitative comparison of un-cleaved MYB33 mRNA and miR159-guided 3′-end MYB33 cleavage products in MYB33 and mMYB33 plants.

Grey bars represent the un-cleaved MYB33 transcript levels in Col-0, MYB33 and mMYB33 transgenic lines. Black bars represent the levels of miR159-guided 3′-end MYB33 cleavage products in the same lines. Numbers show the relative ratio of un-cleaved transcripts to cleavage products. All mRNA levels were normalized with CYCLOPHILIN 5. Measurements are the average of three technical replicates. Similar results were obtained with independent biological replicates. Error bars represent the SEM.

Figure 7. MYB33 variants with central mismatches are not fully silenced.

(A) Aerial view of rosettes from 35-day-old MYB33-1 cm and MYB33-2 cm plants (myb33 genotype) grown side by side with Col-0, mir159ab and mMYB33 under short day conditions. Homo stands for homozygous plants for the transgene. Scale bar = 10 mm. (B) The alignment between the different MYB33 variants and miR159a. The sequences of the amino acids encoded by the miR159 target site in different MYB33 transgenes are shown below. The percentage of free energy when pairing to miR159a compared to the perfect match is shown on the right. Red: mismatches; Purple: nucleotides mutated in the MYB33 miR159 target site. Amino acid changes compared to the original MYB33 protein are highlighted with green. (C) The steady state MYB65 mRNA level in the rosette of selected MYB33-1 cm and MYB33-2 cm transgenic lines with mutant phenotypes. Levels are not statistically different between the variants and parental myb33 plants (D) The steady-state MYB33 mRNA levels in rosettes of the transgenic lines shown in (A). (E) CP1 mRNA levels in the rosette of the transgenic lines shown in (A). All mRNA levels were normalized with CYCLOPHILIN 5. Measurements are the average of three technical replicates. Error bars represent the SEM.

Figure 8. MiR159-guided cleavage is attenuated in MYB33-1 cm and MYB33-2 cm plants.

A fragment of the sequencing chromatography from 5′-RACE recovered 3′ MYB33 transcripts in (A) MYB33-1 cm plants. (B) MYB33-2 cm plants. (C) MYB33 plants. (D) mMYB33 plants. MYB33-1 cm, MYB33-2 cm, MYB33 and mMYB33 sequences are highlighted yellow, while the adaptor is left white. Red line: canonical miR159 cleavage site.

Figure 9. MYB33:miR159a stoichiometry has a critical impact on the silencing outcome in MIR159a and MYB33 variants.

(A) Aerial view of rosettes of 27-day-old of MIR159a1 (mir159ab) and MIR159a2 (mir159ab) plants transformed with the MYB33 construct; and MYB33-1 cm (myb33) and MYB33-2 cm (myb33) plants transformed with MIR159a. All plants were grown side by side with the original MIR159a1, MIR159a2, MYB33-1 cm and MYB33-2 cm parental lines used for transformation. Scale bar = 10 mm. (B) The mRNA levels of CP1 in transgenic lines shown in (A). (C) The relative un-cleaved MYB33 mRNA level (grey bars) and mature miR159a/variant levels (black bars) in Col-0, mir159ab and transgenic lines shown in (A). All original mRNA levels were normalized with CYCLOPHILIN 5. All original miR159a levels were normalized with snoR101. All fold changes in (C) are relative to the levels in Col-0, which is set as 1. Measurements are the average of three technical replicates. Error bars represent the SEM.

Figure 10. Nucleotides flanking the MYB33 miR159 target site are critical for conferring efficient silencing.

(A) ClustalW2 sequence alignment of MYB33 homologues. Nucleotides are grouped into codons of the reading frame. Conserved nucleotides are marked with *. The miR159 binding site is indicated with a pale blue box. Sequences include genes from monocotyledonous species (Tm: Triticum monococcum, Hv: Hordeum vulgare, Tt: Triticum turqidum, Ta: Triticum aestivum, As: Avena sativa, Lt: Lolium tremulentum, Os: Oryza sativa, Sb: Sorghum bicolor, Zm: Zea mays) and dicotyledonous species (At: Arabidopsis thaliana, Vv: Vitis vinifera, Pt: Populus trichocarpa, Rc: Ricinus communis). (B) Mutations made in the regions flanking the miR159 target site in the MYB33-FM construct. Corresponding nucleotide and protein changes are denoted red. (C) Aerial view of 4-week-old MYB33-FM transgenic plants in comparison with MYB33, myb33 and mir159ab plants grown side by side under long day conditions. The number of primary transformants showing the arbitrary phenotypic classification described in the text is listed in the bracket. Scale bar = 10 mm. (D) The mRNA levels of MYB33, MYB65 and CP1 in multiple biological samples of independent MYB33 and MYB33-FM transformants. Mild, Med (medium) and Sev (severe) denote the strength of phenotypic abnormalities observed in MYB33-FM transformants sampled. All mRNA levels were normalized to CYCLOPHILIN 5. Measurements are the average of three technical replicates with error bars representing the SEM.

Figure 11. Quantification of MYB33 expression using the GUS reporter system.

In situ GUS staining of 8-day-old seedlings of the control genotype, (A) myb33, and independent transformants of (B) MYB33:GUS, (C) MYB33-1 cm:GUS, (D) MYB33-2 cm:GUS, (E) MYB33-FM:GUS; all in the myb33 background. (F) The relative fold changes of MYB33 mRNA levels. All values were averaged from two individual biological replicates and the level in MYB33:GUS was normalized to 1. (G) β-Glucuronidase (GUS) activity as measured by MUG assays. Measurements represent three technical replicates and error bars are too small to be visible.

Figure 12. A molecular model for plant miRNA-mediated gene silencing.

For MIR159a1 or MIR159a2 plants, introduction of central mismatches will increases miRISC-mRNA complex half-life due to an attenuated cleavage efficiency, and so leading to an increased target transcript abundance (Figure 1–4). However, as miR159-guided cleavage fails to clear MYB33/MYB65 transcripts (Figure 2), especially when MYB33 is highly transcribed (Figure 5), a non-cleavage (possibly translational repression) mechanism may be the initial or default state of miR159-mediated silencing, with cleavage being a secondary but inherently linked mechanism completing the silencing process, and simultaneously enabling miRISC recycling. When miRISC recycling does become limiting, the miRNA:target mRNA stoichiometry become an increasingly important factor. For instance, attenuated miR159 cleavage of the highly abundant MYB33-1 cm and MYB33-2 cm transcripts (Figure 7), may result in the miR159-RISC complex becoming limiting, leading to the inefficient silencing of these transgenes (Figure 7, 11). Finally, the non-ideal target structure of the MYB33-FM transgene, results in its poor recognition by miR159, enabling this transgene to be expressed (Figure 10, 11).