An Rtf2 Domain-Containing Protein Influences Pre-mRNA Splicing and Is Essential for Embryonic Development in Arabidopsis thaliana

Abstract

Alternative splicing is prevalent in plants, but little is known about its regulation in the context of developmental and signaling pathways. We describe here a new factor that influences pre-messengerRNA (mRNA) splicing and is essential for embryonic development in Arabidopsis thaliana. This factor was retrieved in a genetic screen that identified mutants impaired in expression of an alternatively spliced GFP reporter gene. In addition to the known spliceosomal component PRP8, the screen recovered Arabidopsis RTF2 (AtRTF2), a previously uncharacterized, evolutionarily conserved protein containing a replication termination factor 2 (Rtf2) domain. A homozygous null mutation in AtRTF2 is embryo lethal, indicating that AtRTF2 is an essential protein. Quantitative RT-PCR demonstrated that impaired expression of GFP in atrtf2 and prp8 mutants is due to inefficient splicing of the GFP pre-mRNA. A genome-wide analysis using RNA sequencing indicated that 13-16% of total introns are retained to a significant degree in atrtf2 mutants. Considering these results and previous suggestions that Rtf2 represents an ubiquitin-related domain, we discuss the possible role of AtRTF2 in ubiquitin-based regulation of pre-mRNA splicing.

Keywords: C2HC2 zinc finger; Rtf2 domain; alternative splicing; intron retention; ubiquitin ligase.

Figure 1

T+S transgene silencing system to study RdDM. The two-component transgene silencing system consists of a target locus, T, which contains a GFP reporter gene downstream of a minimal promoter (narrow gray bar) and an upstream virus-derived enhancer (containing a short tandem repeat) that drives GFP expression in shoot and root meristem regions. The silencer locus, S, contains an inverted DNA repeat of distal enhancer sequences (black arrowheads corresponding to thick black bar in T) that is transcribed by RNA polymerase II (Pol II) from a constitutive viral promoter (35S). The resulting hairpin RNA is processed by DICER-LIKE3 (DCL3) to produce 24-nt small interfering RNAs that induce Pol V complex-mediated de novo DNA methylation (blue “m”) of the target enhancer region, leading to transcriptional silencing of GFP expression (Kanno et al. 2008, 2010; Sasaki et al. 2014). Figure is not drawn to scale.

Figure 2

Phenotypes of sdr1 and sdr4 mutants. (A) GFP expression is silenced in wild-type (WT) T+S seedlings, whereas silencing is released in the dms4-1 mutant, which is GFP positive. GFP silencing appears to be reestablished in the dms4-1 sdr1-1 and dms4-1 sdr4-1 double mutants, which are GFP negative (all mutations in T+S background). (B) The dms4-1 sdr1-1 double mutant displays delayed growth and other phenotypic features of the single dms4-1 mutant compared to the age-matched WT control. The dms4-1 sdr4-1 double mutant also appears similar to the dms4-1 single mutant (not shown). (C) Bisulfite sequencing of the target enhancer demonstrates heavy cytosine methylation in all sequence contexts (black, CG; blue CHG; red, CHH; H is A, T, or C) in WT T+S plants, which are GFP negative (A). Release of GFP silencing in the dms4-1 mutant is associated with substantial loss of methylation. The double mutants, dms4-1 sdr1-1 and dms4-1 sdr4-1, appear GFP negative (A) but this is not accompanied by restoration of the WT DNA methylation level in the target enhancer.

Figure 3

Impaired GFP expression in sdr1 and sdr4 mutants in the T line. (A) The wild-type T line expresses GFP in the shoot and root meristem regions. GFP expression is impaired in sdr1-1 and sdr4-1 seedlings (T background only, WT DMS4, S locus absent). (B) Bisulfite sequencing of the target enhancer demonstrated that the impairment of GFP expression in the sdr1-1 single mutant is not accompanied by any significant DNA methylation at the target enhancer (right). The unmethylated enhancer in SDR1 wild-type plants containing the T locus is shown as a control (left). The sdr1-1 mutation also has no effect on DNA methylation genome-wide as indicated by a methylome analysis (Figure S2).

Figure 4

SDR1 is an evolutionarily conserved Rtf2 domain-containing protein. (A) SDR1 (At5g58020), which is 354 amino acids in length, contains an Rtf2 domain (amino acids 84–338) and hence renamed here AtRTF2. AtRTF2 is a single copy gene in Arabidopsis. The position of the G85E amino acid substitution resulting from the sdr1-1/atrtf2-1 point mutation identified in our screen and the sites of two T-DNA insertion alleles (atrtf2-2 and atrtf2-3) are indicated. (B) Phylogenetic tree of Rtf2 orthologs in different organisms. With the exception of budding yeast, Rtf2 orthologs are present in other eukaryotes examined. An unrooted phylogenetic tree was generated by the neighbor-joining method using ClustalW and visualized with TreeView. (C) Amino acid sequence alignments of AtRTF2 orthologs in plants show high similarity in the Rtf2 domain but only partial conservation in the plant-specific N-terminal extension, which is ∼80 amino acids in length. The red arrowhead indicates the location of the sdr1-1/atrtf2-1 G85E mutation at the beginning of the Rtf2 motif. The multiple sequence alignment by ClustalW was performed using GenomeNet (http://www.genome.jp/tools/clustalw/) and consensus amino acid residues were highlighted using BoxShade (http://www.ch.embnet.org/software/BOX_form.html).

Figure 5

Full-length AtRTF2 is required for proper GFP expression. (A) Schematic drawing of C-terminal HA-tagged constructs encoding full-length AtRTF2 (ATRTF2-HA) or a truncated version lacking amino acids 7–63 of the N-terminal extension (ATRTF2ΔN^7–63-HA). These constructs were introduced into the homozygous atrtf2-1 mutant. (B) Western blots probed with a GFP antibody revealed little accumulation in the T+S line, in which GFP expression is silenced (Figure 2A) but wild-type accumulation in the T line. GFP accumulation is low in the atrtf2-1 mutant but returns to the wild-type level when the mutant is complemented with the full-length AtRTF2-HA construct (two examples shown). Levels of GFP protein remain low when using the AtRTF2ΔN-HA construct in the complementation test (two examples shown). RT-PCR (bottom) confirmed that the HA-tagged transgenes are transcribed. The blot was probed with an antibody to tubulin as a control for protein stability and loading levels. The stained membrane is also shown as a loading control. gDNA, genomic DNA (shown for plants without and with the AtRTF2-HA transgene).

Figure 6

AtRTF2 encodes an essential nuclear protein. (A) Self-fertilized plants heterozygous for the null atrtf2-2 mutation (Figure 4A) produce ∼25% defective seedlings (arrowheads, top and enlargement, bottom), which are homozygous for the atrtf2-2 mutation. These seedlings die shortly after germination. The developmental defect can be complemented by full-length AtRTF2 transgenes but not by truncated AtRTF2ΔN versions lacking most of the plant-specific N-terminal extension (Table 1). (B) A seed pigment defect is visible in siliques of heterozygous atrtf2-2 mutant plants (Savage et al. 2013). In our experiment, ∼10–14 days after flowering, ∼25% of the seeds (196/831 counted) appeared white (red arrowheads), whereas 100% of the seeds in an age-matched wild-type control (882/882 counted) were green. (C) AtRTF2-GFP (left) and AtRTF2ΔN-GFP (right) fusion proteins (constructs below) are located predominately in nuclei (shown here in root tip cells).

Figure 7

SDR4 is PRP8. Intron–exon structure of the PRP8 gene (At1g80070) (top) and domain structure of PRP8 (bottom), a core spliceosomal protein of 2359 amino acids. The sdr4-1/prp8-7 mutation (G to A at position 30,125,295 on chromosome 1) creates a G1820E amino acid substitution in the RNase H-like domain. A second point mutation in this region, prp8-6 (G1891E), was reported recently (Marquardt et al. 2014). PRP8 domains were identified in Pfam (http://pfam.xfam.org/).

Figure 8

Alternative splicing of GFP pre-mRNA. (A) As shown by cDNA cloning and sequencing, the T locus encodes three polyadenylated GFP transcripts that result from alternative splicing: a long unspliced transcript; a middle transcript that results from splicing a canonical GT-AG intron; and a short transcript that results from splicing a cryptic intron with noncanonical AT-AC splice sites. Although AT-AC termini are a feature of U12 introns removed by the minor spliceosome, this intron does not contain the highly conserved U12 intron 5′ splice site or branch point sequence and therefore is not a U12 intron (Figure S3). The long and middle transcripts are not translatable into GFP protein (GFP⁻) owing to the presence of numerous PTCs after the initiating methionine (Figure S4). The short transcript can be translated into GFP protein (GFP⁺) (Figure S4) and mutational analysis indicates that it is indeed the major transcript encoding GFP protein (Figure S5). A 5′-RACE experiment demonstrated that transcription initiates in the distal enhancer region ∼45 bp downstream of the short tandem repeat in the target enhancer (arrowheads) (Figure S3). “TATA” indicates an apparently unused minimal promoter directly upstream of the GFP coding region (Figure 1). Arrows indicate the positions of the primers used for the amplification of the three GFP transcripts (B), and the individual long and short transcripts (C). We did not analyze in detail the middle transcript because it is the least abundant and accumulates inconsistently. (B) Semiquantitative RT-PCR showing accumulation of long and short GFP transcripts in the indicated genotypes. Actin is shown as a constitutively expressed control. −RT, no reverse transcriptase. (C) Quantitative RT-PCR showing accumulation of long and short GFP transcripts in the indicated genotypes. Stably expressed At5g60390 was used for normalization (Wang et al. 2014).

Figure 9

Intron retention in atrtf2 and prp8 mutants. (A) Venn diagrams indicate significantly increased IR events in homozygous atrtf2-1 (hypomorphic allele), atrtf2-2 (null allele), and prp8-7 (hyomorphic allele) mutants. Total introns (U2 and U12) are shown at left; U12 introns only are shown at right. The full list is provided in Table S2. (B) Validation of IR events by quantitative RT-PCR. Genes were selected from Table S2 based on P-values. Types of IR events observed are exemplified by At3g63140, At2g47940, and At4g33150 (IR in atrtf2-1 only) and At1g09340, At5g16050, and At3g25690 (IR in prp8-7 only). At3g13920 is shown as a control gene that shows no statistically significant IR changes in the wild-type T line and the mutants. The y-axis indicates the relative IR level normalized to stably expressed At5g60390 (Wang et al. 2014). The primers were designed so that one was inside the target intron and the other was in an adjacent exon. The numbers in parentheses after each gene ID indicate the target intron number for validation as counted from the genomic 5′ end. The error bars indicate standard error of the mean (SEM) of three independent biological replicates. Letters above each bar indicate statistical significance tested by Tukey’s honestly significant difference test (P < 0.05). The same letter (a) indicates no statistically significant difference between the two samples. A different letter (b) indiates a statistical difference between the two samples.