SERRATE interacts with the nuclear exosome targeting (NEXT) complex to degrade primary miRNA precursors in Arabidopsis

Abstract

SERRATE/ARS2 is a conserved RNA effector protein involved in transcription, processing and export of different types of RNAs. In Arabidopsis, the best-studied function of SERRATE (SE) is to promote miRNA processing. Here, we report that SE interacts with the nuclear exosome targeting (NEXT) complex, comprising the RNA helicase HEN2, the RNA binding protein RBM7 and one of the two zinc-knuckle proteins ZCCHC8A/ZCCHC8B. The identification of common targets of SE and HEN2 by RNA-seq supports the idea that SE cooperates with NEXT for RNA surveillance by the nuclear exosome. Among the RNA targets accumulating in absence of SE or NEXT are miRNA precursors. Loss of NEXT components results in the accumulation of pri-miRNAs without affecting levels of miRNAs, indicating that NEXT is, unlike SE, not required for miRNA processing. As compared to se-2, se-2 hen2-2 double mutants showed increased accumulation of pri-miRNAs, but partially restored levels of mature miRNAs and attenuated developmental defects. We propose that the slow degradation of pri-miRNAs caused by loss of HEN2 compensates for the poor miRNA processing efficiency in se-2 mutants, and that SE regulates miRNA biogenesis through its double contribution in promoting miRNA processing but also pri-miRNA degradation through the recruitment of the NEXT complex.

Figure 1.

Proteins associated with Arabidopsis SERRATE. (A) Phenotypes of 4 week-old WT, se-1 and se-1 plants expressing an FLAG-SE. Scale bars are 1 cm. (B) Volcano plot showing the proteins co-purified with FLAG-SE as compared to control IPs performed in Col-0. The bait SE is indicated by a triangle. Closed circles label the proteins known to compose or interact with the Arabidopsis Microprocessor complex, diamonds label the two subunits of the CBC, squares label the subunits of the THO/TREX complex and open circles label the subunits of the NEXT complex. The dashed line indicates the significance threshold (adjp < 0.05).

Figure 2.

The interaction of SE with the NEXT complex is confirmed by reciprocal IPs. Volcano plots showing the proteins co-purified using GFP-tagged versions of (A) ZCCHC8A, (B) ZCCHC8B and (C) RBM7 as baits as compared to control IPs performed in Col-0. The bait proteins are indicated by triangles. The proteins above the threshold (adjp < 0.05, dashed line) are significantly enriched.

Figure 3.

SERRATE interacts with the RBM7 and ZCCHC8 subunits of the NEXT complex. (A) Yeast two-hybrid assays. SE was fused to the Gal4 DNA BD. HEN2, ZCCHC8A, ZCCHC8B and RBM7 were fused to the Gal4 AD. Yeast co-expressing the indicated proteins and SE-BD or harbouring the empty pGBDKT7 vector were selected on medium lacking the amino acids leucine (L) and tryptophan (T). Interaction was tested on medium lacking also histidine (H) and adenine (A), as well as on medium complemented with aureobasidin (Aur) and the galactosidase substrate X-αGal. (B) FRET-FLIM analyses of Arabidopsis protoplasts expressing GFP fused to SERRATE as fluorescent donor protein. HEN2, ZCCHC8A, ZCCHC8B, RBM7 and PRP39A (as negative control) fused to RFP were used as acceptor proteins. The bar graph presents the fluorescence lifetime of SE-GFP in picoseconds (ps). Error bars indicate the standard error of the mean, n > 10. The asterisks indicate significant differences (P < 0.001, Mann–Whitney–Wilcoxon test) between the control samples co-expressing SE-GFP and PRP39A-RFP, and the samples co-expressing SE-GFP and the indicated RFP fusion proteins.

Figure 4.

ZCCHC8A and ZCCHC8B interact with both RBM7 and HEN2. (A) Yeast two-hybrid analyses of the interactions between the NEXT subunits. Medium lacking leucine (L) and tryptophan (T) was used to select for pGADKT7 and pGBDKT7 vectors. Interaction was tested on medium lacking also histidine (H) and adenine (A), and by adding aureobasidin (Aur) and the galactosidase substrate X-αGal. The interactions within the NEXT subunits were evaluated using HEN2-BD as the fusion of ZCCHC8A, ZCCHC8B and RBM7 to the BD resulted in autoactivation. (B) FRET-FLIM analyses of protoplasts co-expressing fluorescent fusion proteins. The bar graphs present the fluorescence lifetime of the donor molecules HEN2-GFP (left panel), ZCCHC8A-GFP (middle panel) and ZCCHC8B-GFP (right panel) in picoseconds (ps). The RFP-fusion proteins used as fluorescence acceptors are indicated at the bottom. (C) SE, HEN2 and ZCCHC8A form a trimeric complex. Combined BiFC and FRET-FLIM analysis. The fluorescent donor molecules are produced by the bimolecular interaction of HEN2 and ZCCHC8A fused to the N- and C-terminal fragments of Venus and Cyan fluorescent proteins, respectively. SE and PRP39A were fused to RFP and served as acceptors. The bar graphs present the fluorescence lifetime of the donor molecule produced by BiFC between nVEN-HEN2 and ZCCHC8A-cCFP in picoseconds (ps). Error bars indicate the standard error of the mean, n > 10. The asterisks indicate significant differences (P < 0.001, Mann–Whitney–Wilcoxon test) between the control samples co-expressing the donor fused to GFP and the indicated acceptors fused to RFP to the samples expressing the donor and the PRP39A-RFP control.

Figure 5.

Loss of HEN2 results in the accumulation of miRNA precursors. (A) Box plot showing the levels of pri-miRNA transcripts from 44 fully annotated MIR genes in three biological replicates of WT and hen2-2. The star indicates the significant difference between the two samples (P < 0.0001, Wilcoxon Signed Rank test) (B–H) Coverage of RNA-seq reads shown as reads per million reads (RPM) for selected miRNA genes in WT and hen2-2. The organization of the MIR genes is shown below each diagram. Lines represent introns, boxes indicate exons. Small bars inside the boxes indicate miRNA and miRNA* sequences. TSS, transcription start site. Scale bars are 400 bp. Additional examples are shown in Supplementary Figure S3.

Figure 6.

Accumulation of maturation by-products and miRNA precursor transcripts in NEXT mutants. (A) Quantification of pri-miRNA levels by qRT-PCR in mutants of NEXT components. The diagrams shown at the top illustrate the primary transcripts of pri-miRNAs 159a and 161. The dashed line indicates an intron. The boxes indicate the miRNA and miRNA* sequences within the stem-loop. The first dicing step occurs at the edges of the stem-loop and releases both 5‘ and 3 maturation by-products and the pre-miRNA which is diced again to excise the mature miRNA/mRNA* duplex. The arrows indicate primers situated either in the 5‘ regions or designed to specifically detect pri-miRNAs. The bar graphs show the mean of three biological replicates, the error bar represents the SD, asterisks indicate significant accumulation (P < 0.05, Student's t-test) as compared to WT. (B) Quantification of pri-miRNA levels by qRT-PCR in exosome mutants. Bar graphs show the mean of two replicates, the error bars indicate the SD. (C) Boxplot showing the global levels of 42 miRNA precursors with known gene structures in rrp41 RNAi and control samples. The RNA-seq data for this analysis were retrieved from (60). The star indicates a significant difference between the means (Wilcoxon Signed Rank Test, P < 0.05).

Figure 7.

Mutating HEN2 partially restores the phenotypes of se-2 plants. Five-week-old plants of the genotypes indicated in each panel. Scale bars are 1 cm.

Figure 8.

Common targets of both SE and HEN2 (A) Genome-wide transcriptome analysis by RNA-seq. Each genotype was compared to WT by DESeq2. The Venn diagram displays the numbers of upregulated transcripts in hen2-2, se-2 and se-2 hen2-2 samples. (B) Pie charts displaying the proportions of different types of transcripts upregulated in both hen2-2 and se-2 hen2-2 (left), se-2 and se-2 hen2-2 (middle) and in all three genotypes (right). (C) Pie chart showing the transcript types that accumulate to higher levels in se-2 hen2-2 as compared to each of the single mutants. (D–F) RNA-seq read distribution profiles for representative loci encoding snRNAs (D), protein coding genes with increased reads in their 5′ regions, (E) and MIR genes (F). The diagrams at the bottom of each panel show the location and structure of the annotated genes with boxes as exons and lines as introns. White boxes represent pre-miRNA, bars inside boxes represent mature miRNA and miRNA*.

Figure 9.

Accumulation of pri-miRNA precursors partially restores miRNA production in se-2 hen2-2 double mutants. (A) Boxplot showing the levels of 43 primary miRNA precursors with known gene structures that were detected in at least 6 of 12 RNA-seq libraries. Different letters indicate significant differences (P < 0.0001, Wilcoxon Signed Rank test). (B) Bar graph showing the levels of mature miRNAs determined by small RNA-seq in three biological replicates of each genotype in percent of total mapped reads. Different letters indicate significant difference between the samples (P < 0.05, Student's t-test). (C) RNA-seq and small RNA-seq data were compared by DESeq2 and the fold-change of the 43 primary miRNAs considered in (A) was plotted against the fold-change of the corresponding mature miRNAs. A significant positive correlation was determined when se-2 hen2-2 was compared to se-2, but not for any of the other comparisons. S, Spearman correlation coefficient; p, P-value. (D) RNA-seq read distribution profiles showing the accumulation of known miRNA target mRNA in se-2 and the restoration of near to WT levels in se-2 hen2-2. The diagrams below each panel illustrate the annotated transcripts with boxes for exons and lines for introns.

Figure 10.

Simultaneous loss of HYL1 and HEN2 partially rescues decreased miRNA. (A) Representative pictures of 4-week-old hyl1-2 and hyl1-2 hen2-2 plants. Scale bars are 1 cm. (B) Relative expression levels of three miRNA precursors determined by qRT-PCR in three biological replicates of WT, hen2-2, hyl1-2, se-2, hyl1-2 hen2-2 and se-2 hen2-2. Distinct letters indicate difference between the means (P < 0,05, Student‘s t-test). Error bars show the SD. (C) Small RNA blots of two biological replicates (Rep) were hybridized with probes specific to miRNAs and U6 snRNA as indicated. Hybridization to chloroplast 4.5 ribosomal RNA is shown as a loading control.