mRNA Decay of Most Arabidopsis miRNA Targets Requires Slicer Activity of AGO1

Abstract

MicroRNAs (miRNAs) are key posttranscriptional regulators of gene expression in animals and plants. They guide RNA-induced silencing complexes to complementary target mRNA, thereby mediating mRNA degradation or translational repression. ARGONAUTE (AGO) proteins bind directly to miRNAs and may catalyze cleavage (slicing) of target mRNAs. In animals, miRNA target degradation via slicing occurs only exceptionally, and target mRNA decay is induced via AGO-dependent recruitment of deadenylase complexes. Conversely, plant miRNAs generally direct slicing of their targets, but it is unclear whether slicer-independent mechanisms of target mRNA decay also exist, and, if so, how much they contribute to miRNA-induced mRNA decay. Here, we compare phenotypes and transcript profiles of ago1 null and slicer-deficient mutants in Arabidopsis (Arabidopsis thaliana). We also construct conditional loss-of-function mutants of AGO1 to allow transcript profiling in true leaves. Although phenotypic differences between ago1 null and slicer-deficient mutants can be discerned, the results of both transcript profiling approaches indicate that slicer activity is required for mRNA repression of the vast majority of miRNA targets. A set of genes exhibiting up-regulation specifically in ago1 null, but not in ago1 slicer-deficient mutants was also identified, leaving open the possibility that AGO1 may have functions in gene regulation independent of small RNAs.

Figure 1.

Developmental phenotype of ago1 slicer-deficient plants. A, Seedlings of Col-0 wild type, ago1-27 (hypomorphic allele), ago1-3, the complementation line of ago1-3 with wild-type AGO1P:FLAG-AGO1, and slicer-deficient point mutants D762A, E803A, D848A, and H988F. B, Rosettes of ago1-3 null and slicer-deficient mutants. C, Cauline leaves of ago1-3 null and slicer-deficient mutants. D, Flowers and aberrant reproductive organs of ago1-3 null and slicer-deficient mutants. E, Representative T1 plants of slicer-deficient (SD) AGO1P:FLAG-AGO1 in Col-0 wild type or ago1-3/+. A wide variability of defects was found among T1 plants of the four point mutants, but not in plants transformed with the equivalent wild-type transgene.

Figure 2.

miRNA populations in slicer-deficient AGO1 plants. Analysis of miRNAs and miRNA*s in total RNA extracted from 8- or 13-d-old seedlings of Arabidopsis stable transgenic lines expressing the wild type or slicer-deficient FLAG-AGO1 in the ago1-3 null background. A, Northern-blot analyses. The same membranes were rehybridized with different miRNA probes. U6 was used as loading control. B, Normalized reads/million of miRNAs, averaged from two biological replicates. The SD (slicer-deficient) column corresponds to the average between the normalized reads obtained for the three AGO1 slicer-deficient mutants analyzed (D762A, E803A, and H988F). Data from miRNA families showing a similar distribution among genotypes were pooled. Only miRNA families with more than 20 reads in total are shown (the complete data set can be found in Supplemental Table S1). Green-yellow color codes indicate relative expression levels of the same miRNA across the different genotypes. Orange-white color codes in the right-most column indicate abundance of the different miRNAs. The membranes 1 to 4 used to hybridize miRNA probes in A are the same as the ones used for siRNA analysis in Figure 3A of Arribas-Hernández et al. (2016). Accordingly, the U6 loading controls are the same in the corresponding panels.

Figure 3.

AGO1 slicer-deficient alleles do not provide evidence for slicer-independent miRNA target regulation. A, Western-blot analysis of protein extracts from 8- or 13-d old whole seedlings of ago1-3 and stable transgenic lines expressing the wild type or slicer-deficient FLAG-AGO1 (D762A, E803A, and H988F) in the ago1-3 null background. Three identical membranes were cut horizontally according to the expected protein sizes and were used for incubation with antibodies detecting the miRNA targets CSD1 (miR398), CSD2 (miR398), AGO2 (miR403), and TIR1 (miR393). AGO1 and FLAG antibodies were used to monitor expression of FLAG-AGO1 in the transgenic lines. Coomassie staining serves as loading control. B to G, Analysis of RNA-seq data from poly(A⁺) RNA isolated from seedlings described in A. Two biological replicates of each genetic background were analyzed. B, PCA. KO, ago1-3 null (knockout); SD, ago1-3 expressing slicer-deficient AGO1 mutants. C and D, Volcano plot showing differential expression between 13-d-old seedlings of ago1-3 (KO) compared to ago1-3 FLAG-AGO1 (WT) and averaged data from slicer-deficient samples (SD) compared to the wild type (C) or SD compared to KO (D). Only significantly regulated genes are plotted (|log₂(fold change)| > 0.5, adjusted P value < 0.05). Validated or predicted miRNA targets are highlighted in red. E, miRNA targets upregulated in SD compared to KO. F, miRNA targets downregulated in SD compared to KO. G, Genes significantly downregulated in SD compared to KO and significantly upregulated in KO compared to wild-type plants. In E to G, only genes with |log₂(fold change)| > 0.5, adjusted P value < 0.05 are included. Adjusted P values were calculated using Wald test with Benjamini-Hochberg adjustment in all cases.

Figure 4.

A Cre-Lox allele of AGO1 to study the effect of slicing on miRNA regulation. A, Schematic representation of the T-DNA containing AGO1P:3xHA-AGO1-AGO1T in Cre-LoxP DNA-excision system used to construct conditional loss of AGO1 function in the ago1-3 background (adapted from Zuo et al., 2001). Expression of the XVE chimeric transcription factor is controlled by the constitutive G10-90 promoter. XVE binding to β-estradiol leads to transcriptional activation of CRE recombinase from the O^LexA-46 promoter. CRE excises the DNA segment placed between the Lox P sites from the host genome, leading to loss of 3xHA-AGO1 and to fusion of the G10-90 promoter with the downstream GFP. B, Phenotype of 20-d-old ago1^Cre-Lox seedlings germinated on MS-agar media with or without 10 μm β-estradiol. GFP fluorescence in root tips of induced plants is shown below. C, Western-blot analyses of protein extracted from the aerial part of seedlings germinated on ±10 μm estradiol and harvested 15 d after germination. All lines analyzed contain the pX6 3xHA-AGO1 transgene in the different backgrounds indicated. Coomassie staining is used as loading control. Note efficient disappearance of 3xHA-AGO1 protein upon CRE induction only in lines with an additional wild-type copy of AGO1.

Figure 5.

RNA-seq analysis upon conditional loss of AGO1 function. A to C, Analysis of RNA-seq data from poly(A⁺) RNA isolated from the aerial part of seedlings germinated on 10 μm estradiol and harvested 15 d after germination. Two biological replicates of each genetic background were analyzed. All lines contain the same pX6 3xHA-AGO1 transgene in different backgrounds: Col-0 wild type, ago1-3 FLAG-AGO1^WT, ago1-3 (‘c-KO’ or Cond. KO), or three different ago1-3 FLAG-AGO1 slicer-deficient mutants (Cond. SD): ‘c-D762A’, ‘c-E803A’, and ‘c-H988F’. Two biological replicates of each genotype were analyzed. A, PCA. B, Volcano plots showing differential expression of validated and predicted miRNA targets. “Cond. SD” refers to the average between ‘c-D762A’, ‘c-E803A’, and ‘c-H988F’ values, and ‘WT’ refers to average values between Col-0 wild type and ago1-3 FLAG-AGO1^WT. C, miRNA targets with significantly different expression (|log₂(fold change)| > 0.5, adjusted P value < 0.05) between Cond. KO and Cond. SD, as described above. Adjusted P values were calculated using the Wald test with Benjamini-Hochberg adjustment.