Functional consequences of splicing of the antisense transcript COOLAIR on FLC transcription

Abstract

Antisense transcription is widespread in many genomes; however, how much is functional is hotly debated. We are investigating functionality of a set of long noncoding antisense transcripts, collectively called COOLAIR, produced at Arabidopsis FLOWERING LOCUS C (FLC). COOLAIR initiates just downstream of the major sense transcript poly(A) site and terminates either early or extends into the FLC promoter region. We now show that splicing of COOLAIR is functionally important. This was revealed through analysis of a hypomorphic mutation in the core spliceosome component PRP8. The prp8 mutation perturbs a cotranscriptional feedback mechanism linking COOLAIR processing to FLC gene body histone demethylation and reduced FLC transcription. The importance of COOLAIR splicing in this repression mechanism was confirmed by disrupting COOLAIR production and mutating the COOLAIR proximal splice acceptor site. Our findings suggest that altered splicing of a long noncoding transcript can quantitatively modulate gene expression through cotranscriptional coupling mechanisms.

Graphical abstract

Figure 1

sof81 Is a Mutation in the Splicing Factor PRP8 that Disrupts FLC Expression (A) The different pathways that antagonistically regulate FLC expression are shown. (B) Mutant screening. Three representative Arabidopsis seedlings of sof81, the C2 progenitor, and F1 individuals from sof81 or C2 crossed to Columbia are shown. False-colored FLC-LUCIFERASE bioluminescence activity is superimposed onto the seedlings. (C) Sequence alignment of the conserved PRP8 RNase H domain region containing the G1891E mutation in prp8-6. At, Arabidopsis thaliana; Hs, Homo sapiens; Sc, Saccharomyces cerevisiae. (D) A schematic representation of the PRP8 protein. Domains and positions of different mutations are shown. Yellow box indicates the epitope of hPRP8 antibody BMR 00434 (Figure S1C). See also Figures S1 and S2.

Figure 2

PRP8 Functions with FCA to Repress FLC and Regulate Flowering (A) Northern analysis comparing FLC (FLC-Ler) levels in C2 prp8-6 to progenitor C2. The asterisk indicates the FLC-LUC transcript (FLC-Col); APT is the loading control. (B) Flowering time determined by rosette leaf number at bolting, values are means ±SEM (n = 12). (C) Flowering time phenotype of different genotypes; data determined by rosette leaf number at bolting. Values are means ±SEM (n ≥ 12). Student’s t test was performed; p values <0.05 are denoted by (^∗) in (B) and (C). (D) Northern analysis of FLC levels in prp8-6 mutants and indicated controls. Two biological repeats are shown; ethidium bromide (EtBr) stained gel is loading control.

Figure 3

prp8-6 Influences Splicing Efficiency of COOLAIR Introns (A) Schematic representation of transcripts from the FLC locus. FLC and COOLAIR isoforms are shown including positions of primers (gray arrows and lines) used to determine splicing efficiencies. Black rectangles denote exons. The asterisk indicates the position of the flc-5 splice acceptor cis element mutation of intron 2 (AG to AA) (Greb et al., 2007). Class Ii and class IIi are abundant COOLAIR isoforms in seedlings grown at warm temperatures, and class Ii and IIii are most abundant in cold-grown seedlings (Swiezewski et al., 2009; Liu et al., 2010; Hornyik et al., 2010). (B) Splicing of alternative FLC sense introns is not affected by prp8-6. FRI in black, and FRI prp8-6 in gray. Levels of spliced and unspliced alternative introns in FLC sense variants FLC.1-4 were determined by qRT-PCR. Splicing efficiencies (spliced/unspliced) are given normalized to FRI background control. Details of the assays are given in extended experimental procedures in the Supplemental Information. (C) Splicing efficiency of introns in UBC and EF1a transcripts assayed using RT-qPCR is not affected by prp8-6 (gray), normalized to the FRI background control (black). (D) Splicing efficiency of COOLAIR transcripts assayed using qRT-PCR of RNA from C2 (black) and C2 prp8-6 mutants (gray). (B–D) Values are means ±SEM (n = 3); Student’s t test was performed. p values <0.05 are denoted by (^∗). See also Figures S3 and S4.

Figure 4

COOLAIR Plays a Role in FLC-Mediated Repression by PRP8 (A) Relative usage of a proximal and distal poly(A) sites of COOLAIR in prp8-6. Proximal and distal poly(A) site usage was assessed by qRT-PCR, as described in Supplemental Experimental Procedures, using primers listed in Table S1 and expressed as relative to total COOLAIR. Genotypes are indicated as FRI wild-type (black) and FRI prp8-6 mutants (gray). Values are means from three biological repeats ±SEM. (B) Three representative FLC^tex transgenic lines and two FLC genomic DNA transgenic controls were crossed to prp8-6 and genotypes homozygous for prp8-6 and all T-DNAs identified. RNA was pooled from each genotype to obtain an average expression value. Averages qRT-PCR values from three independent pooling experiments ±SEM are shown. (A and B) Student’s t test was performed. p values <0.05 are denoted by (^∗). See also Figure S5.

Figure 5

Mutation of COOLAIR Intron 1 Splice Acceptor Site Disrupts PRP8-Dependent Regulation of FLC Expression (A) Schematic representation of intron 1 junction sequence in the wild-type transcript (COOLAIR^WT) or at the mutated 3′ splice site (COOLAIR^AA). (B) qRT-PCR analysis of FLC unspliced and spliced RNA and relative levels of the different COOLAIR forms in the different transgenic genotypes. Ten to fifteen independent transgenic lines for each genotype were harvested and analyzed in pools; values are means ±SEM. qRT-PCR was used to analyze transcript levels relative to UBC. For unspliced, spliced, proximal polyadenylation, and spliced class1i, the COOLAIR^WT value was significantly different (p < 0.05) from the other three genotypes. For distal polyadenylation, the COOLAIR^AA mean is significantly higher than that of COOLAIR^WT (p < 0.05). See also Figure S6.

Figure 6

Coupling between PRP8 Function, Autonomous Pathway Activity, and Histone Demethylation at FLC (A) Chromatin immunoprecipitation (ChIP) assays to analyze H3K4me2 enrichment at FLC. C2 prp8-6 (gray) values are normalized to C2 control (black). Schematic representation of FLC with black rectangles denoting exons. Horizontal bars in A, C, G, and H denote regions analyzed by qPCR in ChIP experiments. H3K4me2 levels in regions A, G, and H are significantly higher in prp8-6 than in C2 (p < 0.05). (B) Pol II occupancy in different regions of FLC in C2 prp8-6 (gray) normalized to the C2 control (black). PolII levels in all regions are significantly higher in prp8-6 than in C2 (p < 0.05). (A and B) Values are means ±SEM (n ≥ 3). (C and D) Splicing efficiency of COOLAIR introns is affected by fca (gray) and fld (white). Averages are from three biological repeats ±SEM in fca-1 and fld-6 normalized to Ler wild-type. (E and F) Relative usage of a proximal poly(A) site of COOLAIR is reduced, and a distal poly(A) site increased in fca and fld mutants. Proximal and distal poly(A) site usage is expressed as relative to total COOLAIR. Averages are from three biological repeats ±SEM in fca-9 and fld-4 normalized to Col wild-type. (C–F) Student’s t test was performed. p values <0.05 are denoted by (^∗). See also Figures S6 and S7.

Figure 7

Cotranscriptional Coupling of Chromatin Modification with COOLAIR Processing (A–C) Inhibition of histone deacetylase activity by trichostatin A (TSA) increases FLC expression and reduces use of the proximal COOLAIR polyadenylation site. Arabidopsis seedlings (C2 genotype) were treated (white) or not (black) with TSA. qPCR data were normalized to UBC. Averages are from three biological repeats ±SEM. All comparisons are statistically significant at p < 0.05. (D) The autonomous pathway promotes use of the proximal antisense splice acceptor site and increases relative use of the antisense proximal poly(A) site. This results in FLD-dependent H3K4me2 demethylation in the gene body of FLC. We envisage that the chromatin modifications influence transcription elongation rate of both strands, leading to low FLC expression and reinforcing the choice of the proximal splice acceptor site of the class Ii intron and proximal polyadenylation of COOLAIR.