Synergistic action of the Arabidopsis spliceosome components PRP39a and SmD1b in promoting posttranscriptional transgene silencing

Abstract

Besides regulating splicing, the conserved spliceosome component SmD1 (Small nuclear ribonucleoprotein D1)b promotes posttranscriptional silencing of sense transgenes (S-PTGS [post-transcriptional genesilencing]). Here, we show that the conserved spliceosome component PRP39 (Pre-mRNA-processing factor 39)a also plays a role in S-PTGS in Arabidopsis thaliana. However, PRP39a and SmD1b actions appear distinct in both splicing and S-PTGS. Indeed, RNAseq-based analysis of expression level and alternative splicing in prp39a and smd1b mutants identified different sets of deregulated transcripts and noncoding RNAs. Moreover, double mutant analyses involving prp39a or smd1b and RNA quality control (RQC) mutants revealed distinct genetic interactions for SmD1b and PRP39a with nuclear RQC machineries, suggesting nonredundant roles in the RQC/PTGS interplay. Supporting this hypothesis, a prp39a smd1b double mutant exhibited enhanced suppression of S-PTGS compared to the single mutants. Because the prp39a and smd1b mutants (i) showed no major changes in the expression of PTGS or RQC components or in small RNA production and (ii) do not alter PTGS triggered by inverted-repeat transgenes directly producing dsRNA (IR-PTGS), PRP39a, and SmD1b appear to synergistically promote a step specific to S-PTGS. We propose that, independently from their specific roles in splicing, PRP39a and SmD1b limit 3'-to-5' and/or 5'-to-3' degradation of transgene-derived aberrant RNAs in the nucleus, thus favoring the export of aberrant RNAs to the cytoplasm where their conversion into double-stranded RNA (dsRNA) initiates S-PTGS.

Figure 1.

Mutations in PRP39a cause late flowering and suppress S-PTGS. A) Photographs of 40-day-old plants of the wild-type Col-0, prp39a-7 (sgs15-1), and prp39a-8 (sgs15-2) mutants. B) Schematic diagram of the top of chromosome 1 in wild-type Col-0, prp39a-7 (sgs15-1), and prp39a-8 (sgs15-2). A deletion removed the 5′ half of PRP39a in prp39a-7 (sgs15-1), while an inversion disrupted the 5′ end of PRP39a in prp39a-8 (sgs15-2). C) Photographs of 20-day-old 2a3 prp39a-8 (sgs15-2) plants, which show no visible NIA S-PTGS, and 2a3 prp39a-8 (sgs15-2) proUBQ10:PRP39a transformants, which undergo NIA S-PTGS.

Figure 2.

PRP39a subcellular localization. A) Confocal images of N. benthamiana leaves infiltrated with proUBQ10:PRP39a-GFP reveal the localization of PRP39a in the nucleoplasm. proUBQ10:XRN2-RFP, proUBQ10:XRN3-RFP, and proUBQ10:XRN4-RFP constructs were used as controls for nucleolar, nucleoplasmic, and cytoplasmic compartments. B) Confocal images of Arabidopsis root cells from seedlings harboring the proUBQ10:PRP39a-GFP or proUBQ10:SmD1b-GFP transgene reveal distinct nuclear localization.

Figure 3.

Different subsets of mRNAs and lncRNAs are deregulated in prp39a and smd1b compared to Col-0. A) Number of upregulated and downregulated genes for each class of RNAs for prp39a and smd1b compared to WT. B) Overlap of upregulated and downregulated lincRNAs, NATs, and mRNAs in prp39a and smd1b compared to Col-0. C) Hierarchical clustering of normalized RNA abundance in Col-0, prp39a, smd1b. Only differentially expressed NATs, lincRNAs, and mRNAs are shown. Co-expression clusters of mRNAs showing distinct expression profiled are highlighted with different colors.

Figure 4.

Changes in RNA abundance in smd1b and prp39a compared to Col-0. A) Number of each class of alternative splicing (AS) events showing a significant change in prp39a or smd1b compared to Col-0. B) Comparison of DEGs and DAS genes in prp39a or smd1b compared to Col-0. C) Comparison of Δpsi in prp39a relative to Col-0 with Δpsi in smd1b relative to Col-0 for all intron retention (IR) events. Significant changes in prp39a vs Col-0, smd1b versus Col-0 or both are highlighted with distinct colors, respectively.

Figure 5.

Comparative effect of prp39a and smd1b mutations on different PTGS systems. A) RNA gel blot analyses of NIA mRNA and siRNAs in the indicated genotypes. Ethidium bromide staining of 25S rRNA served as loading controls for HMW RNA blots. U6 snRNA hybridization served as loading controls for LMW RNA blots. Part of this blot was previously shown in Elvira-Matelot et al. (2016). Original scans are shown in Supplemental Figure S11. B) Kinetics of NIA S-PTGS in the indicated genotypes. Seeds were sown in vitro and transferred to soil after 10 d. The percentage of silenced plants was determined by scoring the number of chlorotic plants (n = 96 plants for each genotype). C) Kinetics of GUS S-PTGS at the intron-less L1 locus in the indicated genotypes. Seeds were sown in vitro and harvested for GUS activity analysis at 5, 11, and 17 d after germination. GUS activity was expressed as arbitrary units of fluorescence per min per µg of protein. (n = 24 seedlings for each genotype at each time point). D) Kinetics of GUS S-PTGS at the intron-containing 159 locus in the indicated genotypes. Seeds were sown in vitro and harvested for GUS activity analysis at 8 and 13 d after germination. GUS activity was expressed as arbitrary units of fluorescence per min per µg of protein (n = 8 seedlings for each genotype at each timepoint). E) Analysis of NIA2 RNA splicing as determined by RT-PCR (Polymerase Chain reaction) using primers spanning an intron. A schematic diagram of the transgene is shown above the gel picture; red arrows mark the relative position of RT-PCR primers. F) Efficiency of GUS S-PTGS at the intron-less Hc1 locus in the indicated genotypes. Seeds were sown in vitro and transferred to soil after 10 d. The percentage of silenced plants was determined by GUS activity at the flowering stage (n = 96 plants for each genotype). G) Photographs of plants carrying the GxA transgenic loci in the indicated genotypes. H) Photographs of plants carrying the JAP3 transgene (PDS hairpin driven by the Arabidopsis SUC2 promoter) in the indicated genotypes. I) RNA gel blot analyses of PDS siRNAs in the indicated genotypes. miR390 hybridization served as loading control.

Figure 6.

Interplay between prp39a and smd1b mutations and RQC mutations. Time course analysis of GUS S-PTGS at the 159 locus in the indicated genotypes. Seeds were sown in vitro and harvested for GUS activity analysis at 5, 11 and 17 d after germination. GUS activity was expressed as arbitrary units (A.U) of fluorescence per min per µg of protein, and is represented as boxplots (n = 24 seedlings for each genotype at each time-point). Results for each time point were compared by a 1-way ANOVA followed by Tukey's post hoc test. Different letters indicate significant differences between genotypes (P < 0.05) at the corresponding developmental time point.

Figure 7.

Transgene RNA immunoprecipitation in 159 prp39a proUBQ10:PRP39a-GFP and 159 smd1b proUBQ10:SmD1b-GFP plants. Lines exhibiting a full restoration of GUS S-PTGS were used for immunoprecipitation using anti-GFP antibodies. A) RIP-RT-qPCR analysis using GUS and NPTII primers. As positive control for PRP39a-GFP RIP, we included 3 primer sets targeting pre-mRNA transcripts identified as differentially spliced in the prp39a RNA-seq analysis. Data represent the mean percentage of input from 3 biological replicates. Values are means ± SE of the mean. Significant enrichment over the IgG RIP was determined using a Student-t test (*P < 0.05). B) Immunoblot analysis of PRP39a-GFP immunoprecipitation. Membranes were probed with anti-GFP antibodies (α-GFP). Input, unbound, and IP fraction from immunoprecipitation with GFP antibodies or normal Rabbit IgG was separated on a gel. GFP IP was performed on 2 independent proUBQ10:PRP39a-GFP transgenic lines. Mock IP using rabbit IgG was performed on one single transgenic line (Line 1).

Figure 8.

PRP39a generally limits the degradation of HEN2 targets by the nuclear exosome. A) Overlap between upregulated and downregulated genes in prp39a and hen2 mutants, respectively, compared to WT. Only genes detected in both RNA-seq were kept for the analysis. B) Relative expression of selected HEN2 targets as determined by RT-qPCR in wild-type plants and prp39a, hen2 and prp39a hen2 mutants. The red lines show the relative position of qPCR amplicons. On the gene model, dark and light blue boxes mark coding region and exons, respectively. Light blue thick lines show untranslated regions; light blue thin lines show introns; black thin lines show the surrounding genomic DNA. Results are shown as boxplots of log₂ fold-change of 3 to 4 biological replicates, individual data point are also plotted. Significant difference between genotypes was determined using one-way ANOVA with post hoc pairwise Tukey's HSD test (P < 0.05). For each gene, 2 pairs of primers were used, one measuring the steady-state abundance of the transcript and one measuring the abRNA (3′ extension, unspliced transcript, see Lange et al. 2014 for details).

Figure 9.

Tentative model for PRP39a and SmD1b action in S-PTGS. Transgene loci that undergo S-PTGS likely produce abRNA. Degradation of these abRNAs by nuclear RQC 5′-to-3, and 3′-to-5′ RNA degradation pathways limits the amount of abRNAs that enter cytoplasmic siRNA-bodies where RDR6 transform them into dsRNA that are processed by DCL2 and DCL4 into siRNAs that are loaded on AGO1 to promote the degradation of regular mRNA. By binding to transgene abRNAs, SmD1b likely impairs XRN3-mediated 5′-to-3′ degradation and partly limits exosome-mediated 3′-to-5′ degradation. By contrast, PRP39a does not bind transgene abRNAs, but somehow limits the action of HEN2 or other components of the nuclear exosome, without altering the activity of XRN3 on these abRNAs. In the absence of PRP39a and SmD1b, transgene abRNAs are more sensitive to 3′-to-5′ and 5′-to-3′ degradation, respectively, thus reducing the probability that a sufficient amount of abRNAs reach siRNA-bodies to activate S-PTGS.