|Home | About | Journals | Submit | Contact Us | Français|
Tgs1 is the enzyme that converts m7G RNA caps to the 2,2,7-trimethylguanosine (TMG) caps characteristic of spliceosomal snRNAs. Fungi grow vegetatively without TMG caps, thereby raising the question of what cellular transactions, if any, are TMG cap-dependent. Here, we report that Saccharomyces cerevisiae Tgs1 methyltransferase activity is essential for meiosis. tgs1Δ cells are specifically defective in splicing PCH2 and SAE3 meiotic pre-mRNAs. The TMG requirement for SAE3 splicing is alleviated by two intron mutations: a UAUUAAC to UACUAAC change that restores a consensus branchpoint and disruption of a stem–loop encompassing the branchpoint. The TMG requirement for PCH2 splicing is alleviated by a CACUAAC to UACUAAC change restoring a consensus branchpoint and by shortening the PCH2 5′ exon. Placing the SAE3 and PCH2 introns within a HIS3 reporter confers Tgs1-dependent histidine prototrophy, signifying that the respective introns are portable determinants of TMG-dependent gene expression. Analysis of in vitro splicing in extracts of TGS1 versus tgs1Δ cells showed that SAE3 intron removal was enfeebled without TMG caps, whereas splicing of ACT1 was unaffected. Our findings illuminate a new mode of tunable splicing, a reliance on TMG caps for an essential developmental RNA transaction, and three genetically distinct meiotic splicing regulons in budding yeast.
Hypermethylated 2,2,7-trimethylguanosine (TMG) cap structures are characteristic of the small nuclear RNAs that program mRNA splicing (U1, U2, U4 and U5) (1). TMG is formed from m7G caps by the enzyme Tgs1 (2), which catalyzes two successive methyl transfer reactions from AdoMet to the N2 atom of 7-methylguanosine (3–8). Whereas guanylate caps are essential in all eukarya that have been examined genetically, the TMG cap is conspicuously not. A tgs1Δmutant of fission yeast Schizosaccharomyces pombe grows normally (5). The tgs1Δ mutation of budding yeast S. cerevisiae causes a growth defect at cold temperatures, although tgs1Δ cells grow as well as TGS1 cells at 34°C (2,6). The tgs1Δmutants of budding and fission yeast lack any detectable TMG caps on their U1, U2, U4 and U5 small nuclear RNAs (snRNAs) and small nucleolar RNAs (snoRNAs) (2,5), signifying that there is no Tgs1-independent route to generate TMG caps in vivo. Tgs1 depletion by RNAi in HeLa cells, which reduced Tgs1 protein levels to below the limit of detection, had no effect on cell growth (9). On the other hand, TMG synthesis plays an essential role during animal development, whereby mutations in the Drosophila Tgs1 methyltransferase active site caused lethality at the early pupal stage, which correlated with depletion of TMG-containing RNAs (10).
The initially surprising conclusion that fungi and human somatic cells grow in the absence of Tgs1 suggested that there might be backup mechanisms to ensure the function of the many essential TMG-capped RNAs when the TMG modification is missing. This idea was confirmed via synthetic enhancement genetics in budding yeast (6,11), which highlighted a redundant role of the TMG cap in spliceosome assembly. The output of the genome-wide screen reported in Hausmann et al. (6) for mutational enhancement of tgs1Δ was highly enriched in proteins implicated in U1 and U2 snRNP function during pre-mRNA splicing. The two strongest interactors with Tgs1, resulting in synthetic lethality, were Mud2 and Nam8. Nam8 is an RNA-binding component of the U1 snRNP and is present in the commitment complex of U1 snRNP at the 5′ splice site (12,13). Nam8 is a putative homolog of the mammalian RNA-binding protein and splicing factor TIA-1 (14–16). Mud2, the yeast homolog of metazoan splicing factor U2AF65, interacts with the pre-mRNA/U1snRNP commitment complex in a manner that depends on the branchpoint sequence of the intron; Mud2 is proposed to facilitate subsequent recruitment of the U2 snRNP (17,18). Synthetic interactions of Tgs1 with other splicing factors (Brr1, Lea1, Ist3, Mud1, Isy1, Cwc21 and Bud13) fortify the case for a genetically redundant role of the TMG cap in mitotically growing yeast (6). This raises the question of what cellular transactions, if any, are TMG cap-dependent per se.
Among the vegetatively optional yeast splicing factors in the Tgs1 genetic ‘neighborhood’, Nam8 stands out because it is strictly essential for meiosis. During sporulation, Nam8 promotes the splicing of specific mRNAs that encode proteins required for meiotic recombination and cell division (19–22). Nam8-dependent splicing of four meiotic mRNAs—AMA1, MER2, MER3 and SPO22—is activated by the meiotic splicing regulator Mer1 (16,23–25), which is produced only in meiotic cells under the control of the meiotic transcription factor Ime1 (26). Mer1 activates splicing by binding to an intronic splicing enhancer sequence (5′-AYACCCUY-3′) present in the AMA1, MER2, MER3 and SPO22 pre-mRNAs (16,20). Mer1 bound to the intronic enhancer is thought to promote assembly of the U1 and U2 snRNPs on the pre-mRNA. We recently showed that Nam8 plays a distinct Mer1-independent role in splicing of the PCH2 meiotic pre-mRNA, thereby establishing the existence of two genetically distinct meiotic splicing regulons (16). Transcripts subject to Mer1 and Nam8 splicing regulation have either suboptimal 5′ splice sites, suboptimal branchpoints or a large 5′ exon that apparently dictate their reliance on otherwise inessential splicing factors (16,21,27,28).
Here, we queried whether Tgs1 and TMG caps might also be specifically required for RNA transactions during meiosis. Previously, a screen of the yeast single-gene deletion library for defects in sporulation and meiosis had identified YPL157W (TGS1) as a sporulation gene (29), but the connection between Tgs1 and cap trimethylation had not yet been made when this screen was conducted and no insights were made to how YPL157W might act during meiotic development. We now report that Tgs1 and, most importantly, its methyltransferase activity are required for sporulation, because TMG caps promote splicing of a novel meiotic pre-mRNA regulon. By evaluating the splicing of the known meiosis-specific pre-mRNAs (16,30–34) in wild-type and tgs1Δyeast diploids during attempted sporulation, we discovered that Tgs1 is required for efficient splicing of two meiotic mRNAs that lack a Mer1-binding site: SAE3 and PCH2. In each case, we identified the distinctive features of the pre-mRNA that dictate Tgs1-dependence in vivo. Analysis of in vitro splicing in extracts of TGS1 versus tgs1Δ cells showed that SAE3 intron removal was enfeebled selectively absent TMG caps, while splicing of the ACT1 and U3 introns was unaffected. Our findings highlight new complexity in regulated splicing in yeast and the context-dependent reliance on TMG caps for an essential RNA transaction.
The meiosis/sporulation experiments were carried out with isogenic diploids in the SK1 background. To generate tgs1Δ diploids, haploid derivatives of the SK1 strain, SKY163 (MATa ho::LYS2 lys2 ura3 leu2::hisG) and SKY164 (MATα ho::LYS2 lys2 ura3 leu2::hisG) were used. In brief, DNA segments encompassing the tgs1::kanMX and tgs1::natMX disruption cassettes were generated by polymerase chain reaction (PCR) amplification of genomic DNA from the respective tgs1Δ yeast strains (6) using primers flanking the cassette. One of the PCR products (tgs1::kanMX) was introduced into SKY163 and geneticin-resistant integrants were selected. SKY164 was transformed with a tgs1::natMX deletion cassette and nourseothricin-resistant natMX integrants were selected. The targeted insertions were confirmed by diagnostic Southern blotting. The SKY haploids were then mated and homozygous tgs1Δ diploids were selected on yeast peptone dextrose (YPD) agar containing 100µg/ml nourseothricin and 150µg/ml geneticin.
Integration of TGS1 and tgs1-D126A at the leu2 locus of SKY tgs1Δ strains was performed as follows. To generate the targeting cassettes, DNA fragments (1.5kb) spanning the TGS1 and tgs1-D126A genes under the control of the native TGS1 promoter were excised from pUN100-based plasmids (6) and inserted into the polylinker of the integrative pRS305 (LEU2) vector. The resulting plasmids were restricted with EcoRV within the LEU2 gene and the linearized plasmids were transformed into the SKY haploids tgs1Δ::kanMX and tgs1Δ::natMX. Leu+ transformants were selected and diagnostic Southern blotting was used to confirm the correct integrations of the LEU2 plasmid at leu2. The corresponding haploids were then mated and tgs1Δ leu2::[LEU2 TGS1] and tgs1Δ leu2::[LEU2 tgs1-D126A] diploids were selected.
Single colonies of diploid yeast strains were patched on agar plates with glycerol as the carbon source for at least 6h to select for cells with healthy mitochondria. Cells were streaked on YPD agar plates and incubated for 3 days at 30°C. Single colonies were then inoculated into YPD liquid medium and grown at 30°C to stationary phase (A600 of 6–8). Aliquots were inoculated into 12.5ml of presporulation medium [0.5% yeast extract, 1% peptone, 0.67% yeast nitrogen base (without amino acids), 1% potassium acetate, 0.05M potassium biphthalate (pH 5.5), 0.002% antifoam 204] to attain an A600 of 0.8. The cultures were incubated for 7h at 30°C and added to 100ml of fresh presporulation medium to attain an A600 of 0.025 (wild-type cells) or 0.1 (tgs1Δ cells). These cultures were incubated for 16h until A600 reached ~3.0. The cells were harvested by centrifugation, washed twice with sporulation medium (2% potassium acetate, 0.001% polypropylene glycol) and then resuspended in sporulation medium at A600 of 6. Aliquots were withdrawn from synchronous meiotic cultures at 6, 8, 10, 12, 14 and 24h post transfer to sporulation medium. The cells were fixed in an equal volume of 100% ethanol and then examined by light microscopy (100× magnification) to assess the abundance of 4-spore asci. Two-hundred cells from each sample were scored. The extents of sporulation (% asci) are plotted as a function of time in Figure 1. Each datum is the average of three separate experiments±SEM.
For analysis of meiotic RNA, 2-ml aliquots of the cultures were withdrawn immediately prior to transfer to sporulation medium and at 4h and 8h post-transfer to sporulation medium. Cells were harvested by centrifugation and RNA was extracted from the cells by using the MasterPure Yeast RNA Purification Kit (Epicentre BioTechnologies). The DNase digestion step in the RNA purification protocol was modified such that the samples were incubated for 1h with 20U of RNase-free DNase I (New England Biolabs) to eliminate genomic DNA. First-strand cDNA was synthesized in reaction mixtures containing 10ng/μl total RNA, 25ng/μl oligo(dT)12–18 primer, 10U/μl SuperScript II (Invitrogen), 1X First Strand Buffer, 10mM dithiothreitol (DTT) and 0.5mM deoxynucleotide triphosphates (dNTPs). RNA, dNTPs and primers were preincubated for 5min at 65°C before quick chilling on ice. DTT and First Strand Buffer were added, and the mixture was incubated at 42°C for 2min. Finally, reverse transcriptase was added and the reaction mixture was incubated at 42°C for 50min and then at 70°C for 15min. The cDNAs for meiotic transcripts were then PCR-amplified in 25μl reaction mixtures containing 1X native Pfu buffer, 0.2mM dNTPs, 0.8μM gene-specific sense strand primer and 0.8μM 5′-32P-labeled gene-specific antisense strand primer (16), 0.05U/μl Pfu DNA Polymerase (Agilent Technologies) and 2μl of each cDNA sample. The PCR cycles (n=31) entailed incubations at 94°C for 30s, 55°C for 90s and 72°C for 2min. The reverse transcriptase (RT)–PCR products were analyzed by electrophoresis through 2% native agarose gels. After electrophoresis, the gels were stained with ethidium bromide and then dried under vacuum on DEAE paper. The 32P-labeled PCR products were visualized by autoradiography and quantified by scanning with a Fuji BAS-2500 imager.
A haploid ‘wild-type’ TGS1 yeast strain (MATa TGS1 leu2Δ ura3Δ), derived from S288c, and a tgs1::natMX variant thereof (6) were transformed with plasmids expressing MER1, SPO22, SAE3 or PCH2. The expression plasmids were as follows. pYX212-MER1 (2μ URA3 TPI1-MER1) has the MER1 open reading frame under the transcriptional control of the yeast TPI1 promoter. pRS425TPI-SPO22 (2μ LEU2 TPI1-SPO22) carries the intron-containing SPO22 gene driven by the TPI1 promoter. pRS425-PCH2 (2μ LEU2 PCH2) bears the intron-containing PCH2 gene (from 540bp upstream of the translation start codon to 265bp downstream of the stop codon) under the control of its native promoter. pRS425-SAE3 (2μ LEU2 SAE3) includes the intron-containing SAE3 gene (from 370-bp upstream of the start codon to 255-bp downstream of the stop codon) under the control of its native promoter. SAE3 and PCH2 intron mutants (see Figures 4 and and8)8) were generated by two-stage overlap extension PCR with mutagenic primers; the mutated SAE3 and PCH2 DNAs were inserted into pRS425. The inserts of all plasmid clones were sequenced to exclude the acquisition of unwanted mutations during amplification and cloning.
The haploid plasmid-bearing yeast strains were grown in SD-(Ura−Leu−) liquid medium at 30°C until A600 reached 2–4. Cells were harvested by centrifugation from 2-ml aliquots of the cultures. RNA extractions and RT–PCR were performed as described above, with exceptions as follows: (i) cDNA synthesis was primed with 0.1µM gene-specific antisense primers (16) instead of oligo(dT)12–18; (ii) the numbers of PCR cycles for amplification of MER2, PCH2, SAE3 and SPO22 cDNAs were 28, 27, 29 and 25, respectively; (iii) the PCR reactions were quenched by adding ethylenediaminetetraacetic acid (EDTA) and sodium dodecyl sulfate (SDS) to a final concentrations of 12.5mM and 3.3%, respectively; and (iv) the 32P-labeled PCR products were analyzed by electrophoresis through native 5% polyacrylamide gels containing 90mM Tris-borate, 1.2mM EDTA.
Strains used for extract preparation were BJ2168 (MATa leu2 trp1 ura3-52 prp1-1122 pep4-3 prc1-407 gal2) and BJ-Δtgs1, a derivative of BJ2168 in which the TGS1 locus was replaced by the tgs1::natMX deletion cassette (6). Cultures (6l) were grown in YPD medium at 30°C until A600 reached 2.0–2.5. Whole-cell extracts were prepared as described (42). 32P-GMP-labeled intron-containing precursor RNAs were transcribed from linearized plasmid templates by T7 RNA polymerase and purified by gel-filtration through Sepharose CL-6B (42). Splicing reaction mixtures containing 40% (v/v) extract, 60mM potassium phosphate (pH 7.0), 2mM adenosine triphosphate (ATP), 2mM MgCl2, 3% (w/v) PEG8000 and 1nM precursor RNA were incubated at 28°C. The reaction products were analyzed by electrophoresis through a 6% polyacrylamide gel containing 7M urea in TBE. The labeled RNAs were visualized by autoradiography of the dried gel and quantified by scanning with a Typhoon phosphorimager (Molecular Dynamics). The splicing efficiencies [spliced/(spliced+ unspliced)]×100 were calculated after normalizing the values for the 32P-GMP content in the two RNA species.
To generate integration cassettes for the HIS3 reporter genes, a 1-bp genomic DNA fragment spanning the HIS3 ORF plus 297 and 20bp of upstream and downstream sequences was amplified by PCR using primers His3-F and His3-R (Supplementary Table S1) and inserted in between the XmaI and PstI sites in pUC19. To insert the SAE3 intron (or the mutated versions) at nucleotide position +48 in the HIS3 ORF, we individually amplified the HIS3-5′ gene fragment (345bp) with primers HIS3-F and SAE3-exon1R (Supplementary Table S1) and the HIS3-3′ gene fragment (635bp) with primers HIS3-R and SAE3-exon2F (Supplementary Table S1). The SAE3, SAE3-BP and SAE3Δhp introns were amplified by PCR with primers SAE3-intronF and SAE3-intronR. The exon and intron DNA fragments with overlapping terminal sequences were then assembled into intron-punctuated HIS3 cassettes by overlap-extension PCR and the cassettes were inserted into pUC19 plasmids. Using the same strategy and primers listed in Supplementary Table S1, the PCH2 intron was inserted to generate HIS3-[PCH2]-5′. The pUC19-based plasmids harboring HIS3-[PCH2] and HIS3-[PCH2-BP] (in which the PCH2 intron is inserted at position 430 within the HIS3 ORF) were described previously (16). The cassettes were excised with SmaI and PstI and transformed into tgs1Δ::natR p360-TGS1(URA3 TGS1) cells and isogenic wild-type cells (harboring a URA3 plasmid) that were histidine-auxotrophs. His+Ura+ transformants were selected and analyzed for integration of the respective cassettes at the HIS3 locus by diagnostic Southern blotting and by sequencing of DNA fragments PCR-amplified from genomic DNA using primers flanking the HIS3 gene. The cells were then streaked to medium containing 5-FOA (and histidine) to select for cells that had lost the URA3 plasmids. Individual colonies were patched to YPD agar. Cells were grown in liquid YPD medium until the cultures attained A600 of 0.7–0.9. The cultures were diluted in water to A600 of 0.01. Serial 10-fold dilutions were prepared and aliquots (3µl) of each were spotted in parallel on SD agar medium with or without histidine. We also introduced HIS3, HIS3-[SAE3] and HIS3-[PCH2] reporter cassettes into isogenic mud2Δ p360-MUD2(URA3 CEN MUD2) and mud1Δ p360-MUD1(URA3 CEN MUD1) strains and then analyzed them as outlined above. swt21Δ strains harboring the reporter genes were generated through genetic manipulations involving crossing swt21Δ::natR cells to isogenic wild-type strains containing the HIS3, HIS3-[SAE3], or the HIS3-[PCH2] reporter alleles.
We monitored the appearance of 4-spore asci as a function of time after transfer of a culture of yeast SK1 diploid cells to sporulation medium. Wild-type SK1 efficiently and synchronously formed asci between 8 and 12h and attained 93% sporulation efficiency at 24h (Figure 1A). By contrast, an isogenic tgs1Δ diploid was grossly defective in completion of the meiotic program, with only 4% ascus formation after 24h (Figure 1A). Normal kinetics of sporulation were restored to the tgs1Δstrain by integration of a wild-type TGS1 gene at the chromosomal LEU2 locus (Figure 1B). The key finding was that no rescue of the sporulation defect of the tgs1Δ diploid was achieved by integration of the tgs1-D126A allele (Figure 1B). Asp126 coordinates the ribose hydroxyls of the AdoMet methyl donor in the Tgs1 active site (35). The Asp126 equivalent is essential for guanine–N2 methyltransferase activity in vivo and in vitro in all Tgs enzymes that have been studied (2,4–6). We conclude that the catalytic activity of Tgs1, and thus the TMG cap structure, is critical for the yeast meiotic developmental program.
RNA was isolated from wild-type and tgs1Δ diploids immediately prior to (time 0) and 4h and 8h after transfer from pre-sporulation medium to sporulation medium. cDNA was prepared from each RNA sample by reverse transcription and then used for gene-specific PCR amplification of meiotic spliced transcripts (Figure 2 and Table 1). The sense and antisense primers corresponded to sequences flanking the introns so that the longer products of amplification of cDNA derived from unspliced pre-mRNAs could be easily resolved by native gel electrophoresis from the shorter products of amplification of cDNAs copied from spliced mRNA (Figure 2). One of the primers was 5′ 32P-labeled in each PCR reaction so that we could quantify the distributions of unspliced and spliced cDNAs for each gene of interest (Figure 2, lanes 3–5). An aliquot of a PCR amplification reaction using genomic DNA as template provided a marker for the unspliced species (lane 2). No labeled products were generated from PCR reactions programmed by RNA that had not been subjected to prior treatment with reverse transcriptase (lane 1), indicating that the RNA samples were effectively free of contaminating genomic DNA.
Figure 2 shows exemplary data demonstrating induction of meiotic transcription and regulated meiotic splicing in wild-type cells (16), while focusing on some of the meiotic transcripts for which splicing efficiency was most acutely affected by the absence of Tgs1. Transcriptional induction of the SPO22 gene in wild-type cells was evinced by the increase in total RT–PCR products at 4 and 8h post-sporulation (lanes 4 and 5) compared to the level at time 0 (lane 3), which was accompanied by a sharp increase in the percentage of the RT–PCR product derived from spliced versus unspliced SPO22 RNA. In agreement with previous studies (36,37), we also observed transcriptional induction of PCH2 (Figure 2) and of several other intron-containing meiotic genes analyzed (data not shown). The induction of SPO22, PCH2 and other meiotic genes was also evident in tgs1Δcells after 4 and 8h in sporulation medium (Figure 2 and data not shown). This result signifies that the gross defect in yeast sporulation in the absence of Tgs1 and TMG caps is not caused by an early failure in the meiotic transcriptional program. The salient finding was that the tgs1Δmutation sharply inhibited the splicing of SAE3 and PCH2 pre-mRNAs, while having only a modest effect on splicing of the Nam8/Mer1 target SPO22 (Figure 2). By contrast, the control GLC7 RNA was constitutively expressed and very efficiently spliced in wild-type and tgs1Δcells (Figure 2).
The results of our survey of splicing of 14 meiotic transcripts at 4h post-induction of sporulation are compiled in Table 1, wherein each datum for splicing efficiency—[spliced/(spliced+unspliced)]×100—is the average of three independent sporulation experiments and RT–PCR analyses. We operationally defined a severe mutational effect on meiotic splicing as one that elicits a ≥4-fold reduction in splicing efficiency compared to wild-type controls and a modest mutational effect as one that reduces efficiency by 2-fold (16). SAE3 and PCH2 were the only two transcripts that met our criterion for a severe meiotic splicing defect in the tgs1Δ strain. Ablation of Tgs1 reduced SAE3 and PCH2 splicing efficiencies by 9-fold and 5-fold, respectively (Table 1).
Regulated splicing of Mer1-dependent meiotic mRNAs can be recapitulated in vegetative yeast cells by forced expression of the Mer1 splicing enhancer protein (16,20,22,23). For example, MER2 pre-mRNA is constitutively transcribed in vegetative cells, but is spliced inefficiently (20%) because Mer1 is absent (Figure 3A) (16). Expression of Mer1 during vegetative growth (by transformation with a 2µ MER1 plasmid in which MER1 is linked to a constitutive promoter) increased MER2 splicing efficiency to 87% in wild-type cells (Figure 3A). Mer1 expression increased MER2 splicing efficiency in tgs1Δ cells from 9% to 72% (Figure 3A). To gauge the effects of Tgs1 ablation of splicing of another Mer1 target, SPO22, we cotransformed vegetative cells with a 2µ plasmid bearing SPO22 plus either a 2µMER1 plasmid (Mer1+) or an empty vector control (Mer1–). Splicing of SPO22 mRNA in wild-type yeast was acutely dependent on Mer1 (86% in Mer1-expressing cells versus 14% in control cells) (16), but not Tgs1 (56% splicing in Mer1-expressing tgs1Δcells versus 3% in tgs1Δ control cells) (Figure 3B). We conclude that whereas loss of Tgs1 diminished basal splicing of MER3 and SPO22 RNAs, it did not compromise the activation of MER3 and SPO22 splicing by Mer1. Previous studies showed that Mer1-activated splicing of MER2 and SPO22 was strictly codependent on Nam8 (16).
By contrast, the splicing of PCH2 and SAE3 transcripts, which we identified here as Tgs1-dependent, was unaffected by Mer1 expression (Figure 3C). This makes sense, given that neither of the Tgs1-dependent meiotic pre-mRNAs has a Mer1 enhancer element in the intron. The efficiencies of PCH2 and SAE3 splicing in wild-type vegetative cells (67% and 65%, respectively) were comparable to those seen in meiotic cells (Table 1) (16). Whereas PCH2 splicing was reduced to similar degrees in either nam8Δ cells (24%) or tgs1Δ cells (18%), SAE3 splicing was inhibited selectively in tgs1Δ cells (26% efficiency) (Figure 3C). These results consolidate the following points: (i) both examples of Tgs1-dependent splicing are independent of Mer1 and can be modeled in vegetative cells and (ii) Tgs1-dependent SAE3 and PCH2 pre-mRNAs differ in their codependence on Nam8.
The Mer1/Nam8-dependent meiotic pre-mRNAs share the property that their 5′ splice sites deviate from the consensus 5′-GUAUGU motif that promotes base-pairing with the U1 snRNA during the first step of spliceosome assembly. By contrast, the Tgs1-dependent SAE3 and PCH2 pre-mRNAs adhere perfectly to the consensus 5′ splice site sequence. However, the SAE3 and PCH2 transcripts have introns that deviate from the consensus yeast branchpoint motif 5′-UACUAAC. The branchpoint is 5′-UAUUAAC in SAE3 and 5′-CACUAAC in PCH2; these variants are extremely rare among yeast intron-containing genes (38). SAE3 is further distinguished by a rare 3′ splice site element, AAG-3′, that deviates from the canonical 3′ splice site YAG-3′.
To query whether these deviant features of the SAE3 intron confer Tgs1 dependence, we introduced mutations that restored the consensus branchpoint and 3′ splice site elements and then assayed splicing efficiency in vegetative cells by RT–PCR. Installing a perfect UAG 3′ splice site (in mutant SAE3-3′SS) had no salutary effect on splicing efficiency in wild-type (66%) or tgs1Δ(21%) cells (Figure 4). By contrast, the consensus branchpoint (entailing a U-to-C mutation in SAE3-BP) enhanced splicing in wild-type cells to 88%. Whereas the corrected branchpoint also raised splicing efficiency slightly in tgs1Δcells (to 47%), it did not efface the dependence of SAE3 splicing on Tgs1. Simultaneous correction of the branchpoint and 3′ splice sites in mutant SAE3-(BP+3′SS) provided no advantage over the single mutation of the branchpoint (Figure 4). We conclude that the non-consensus branchpoint affects SAE3 splicing efficiency but is not the sole decisive factor in SAE3’s dependence on TMG caps.
There is an emerging wealth of evidence attesting to the role of RNA secondary structures in regulating the splicing of specific pre-mRNAs (39). Interrogation of the potential secondary structure of the 86-nt SAE3 intron in Mfold (40) revealed a single predicted fold (ΔG -11.7) containing a stem–loop structure of immediate interest with respect to regulatory potential (Figure 4). This stem–loop, from nucleotides 57 to 78, consists of a 9-bp stem with one bulged base plus an apical triloop. This predicted structure embraces the 5′-UAUUAAC branchpoint element within the apical stem and loop, in a manner that would mask the intermolecular base-pairing interactions of the branchpoint sequence to U2 snRNA (Figure 4) that are critical for the next step of spliceosome assembly. Thus, this intronic secondary structure might pose a barrier to efficient SAE3 pre-mRNA splicing that is overcome by Tgs1 and TMG caps.
To evaluate this scenario, we made a short internal deletion in the SAE3 intron—of nucleotides 57–65 (Δhp)—that is predicted to abolish the RNA hairpin structure (Figure 4). An instructive finding was that disruption of the stem–loop covering the branchpoint by the Δhp mutation increased splicing efficiency in both wild-type and tgs1Δ cells, to 84% and 53%, respectively, but still did not efface the contribution of Tgs1 to SAE3 splicing. The decisive maneuver was to introduce the Δhp deletion into the SAE3-BP intron, which resulted in equivalently high (90%) splicing efficiencies in wild-type and tgs1Δ cells (Figure 4). We conclude that the reliance of SAE3 splicing on Tgs1 and TMG caps reflects the doubly unfavorable primary and secondary structure of the branchpoint in the SAE3 intron.
We inserted the 86-nt SAE3 intron near the 5′ end of the otherwise intron-less chromosomal yeast HIS3 gene in both TGS1 and tgs1Δ haploid yeast cells (Figure 5A). HIS3 provides a convenient reporter for gene expression, manifest as histidine prototrophy (HIS3 ‘on’) or auxotrophy (HIS3 ‘off’). The native HIS3 gene is functional in TGS1 and tgs1Δstrains, both of which grow on agar medium lacking histidine (Figure 5B). By contrast, the HIS3-[SAE3] reporter containing the inserted SAE3 intron was functional in TGS1 cells, but not in the tgs1Δbackground (Figure 5B). The salient finding was that the Δhp mutation disrupting the stem–loop covering the branchpoint restored HIS3 function in tgs1Δcells (Figure 5B). By contrast, the U-to-C mutation in SAE3-BP that restored a consensus branchpoint sequence did not suffice per se to confer histidine prototrophy in tgs1Δcells. However, combining the consensus branchpoint U-to-C mutation with an upstream 4-nt substitution that prevents base pairing of this segment to the branchpoint restored HIS3 function in tgs1Δ cells (Figure 5B). These results show that the SAE3 intron, by virtue of its branchpoint context, is an autonomous determinant of Tgs1-dependence in the absence of other SAE3 gene elements (i.e. promoter, 5′ and 3′ exons, 5′ and 3′ untranslated regions). Thus, TMG caps are implicated in recognition and/or utilization of the deviant SAE3 branchpoint.
The specificity of the splicing factor requirements for HIS3-[SAE3] reporter function was tested by introducing the expression cassette into the chromosome of yeast strains deleted for four other vegetatively inessential splicing factors—Nam8, Mud1, Mud2, and Swt21—that display synthetic lethal/sick interactions with one another and with Tgs1 (6,8,16). We found that HIS3-[SAE3] was active in nam8Δ, mud1Δ and swt21Δ cells, but not in mud2Δ cells (Figure 5C). The finding that Mud2 facilitates expression of a pre-mRNA with a suboptimal branchpoint region is consistent with Mud2 being a component of a heterodimeric complex with the yeast branchpoint binding protein Msl5 (41).
We extended our analysis of intronic determinants of SAE3 splicing by testing the wild-type and mutated SAE3 RNAs as substrates for splicing in vitro in a whole-cell extract prepared from a TGS1 yeast strain. The SAE3 DNA constructs were cloned downstream of a bacteriophage RNA polymerase promoter and SAE3 pre-mRNAs with 5′ m7G caps and internal 32P-GMP-labels were prepared by cap dinucleotide-primed transcription in vitro by phage RNA polymerase (Supplementary Figure S1). The isolated pre-mRNAs were incubated with yeast whole-cell extract under optimized splicing conditions (42) and the radiolabeled products were resolved by polyacrylamide gel electrophoresis (PAGE) and visualized by autoradiography. Splicing was manifest as the conversion of the unspliced precursor into a shorter mature spliced product (Figure 6). Splicing efficiency was quantified by the distribution of radiolabel [spliced/(spliced+unspliced)] after correction for the 32P-GMP content of the two species. The salient finding was that in vitro splicing of the wild-type SAE3 intron was conspicuously inefficient (7%), but was responsive to manipulations of the intron in a manner concordant with the in vivo data presented above. To wit, splicing was enhanced by installation of a consensus branchpoint (36%) and disruption of the predicted branchpoint secondary structure (38%), whereas there was no salutary effect of restoring a consensus 3′ splice site (9%) (Figure 6). Combining the consensus branchpoint with the relief of secondary structure over the branchpoint exerted additive effects that resulted in 89% SAE3 splicing efficiency in vitro (Figure 6), a value comparable to the 83% efficiency of splicing of actin pre-mRNA in the same experiment (data not shown).
Having established that several of the SAE3 pre-mRNAs can be spliced in vitro, we compared their efficacies in whole-cell extracts derived from otherwise isogenic TGS1 and tgs1Δ yeast strains. The kinetic profiles of actin pre-mRNA splicing were similar in the two extracts (Figure 7A), as were the rates and extents of splicing of the intron-containing U3 pre-snoRNA (data not shown). These results signify that TMG cap structures on the endogenous spliceosomal snRNAs are not limiting for removing the introns of pre-RNAs that are normally spliced with high efficiency in vitro. By contrast, the rates and extents of splicing of the SAE3-BP and SAE3-Δhp pre-mRNAs in the TGS1 extract were lower than that of actin pre-mRNA; moreover, the SAE3-BP and SAE3-Δhp substrates experienced a kinetic delay in the initial appearance of their respective mature spliced products (Figure 7B and C). A key finding was that the rate and extent of SAE3-BP splicing was ~3-fold lower in tgs1Δextracts than in TGS1 extracts (Figure 7B). The SAE3-Δhp substrate was spliced nearly 2-fold slower in tgs1Δextracts than in TGS1 extracts (Figure 7C). The disparity in SAE3 versus actin pre-mRNA splicing in the TGS1 extract was erased by the SAE3-(Δhp+BP) intron (Figure 7D), which also corrected the lag in splicing onset seen with the SAE3-BP and SAE3-Δhp substrates. The SAE3-(Δhp+BP) intron also expunged the disparity in SAE3 splicing rates in TGS1 versus tgs1Δextracts (Figure 7D). Thus, the requirements for bypassing TMG cap-dependence of SAE3 splicing in vivo were recapitulated in vitro.
The PCH2 transcript is distinguished by its exceptionally long 5′ exonic open reading frame (1551-nt) and a non-consensus intron branchpoint sequence, 5′-CACUAAC (Figure 8). We probed the PCH2 splicing determinants by: (i) installing a consensus branchpoint sequence in an otherwise native PCH2 gene; (ii) deleting most of the long upstream exon (while installing a new in-frame AUG codon) to create a PCH2-Δ5′ variant with a 51-nt 5′ exon; and (iii) combining the BP and Δ5′exon changes. We tested the PCH2 mutants on 2µplasmids for their splicing efficiency in TGS1 and tgs1Δ yeast cells. As noted previously (16), the results implicate the non-consensus branchpoint and long 5′ exon as separate negative influences on PCH2 splicing in wild-type cells. For example, shortening the 5′ exon per se increased splicing efficiency to 89% for PCH2-Δexon versus 70% for the PCH2 transcript (Figure 8). The consensus branchpoint change per se increased splicing efficiency to 91% for PCH2-BP (Figure 8). There was no apparent effect of combining the BP and Δexon changes in wild-type cells. The PCH2 transcript was spliced with 21% efficiency in tgs1Δ cells, a 3.5-fold decrement compared to TGS1 cells. The PCH2-Δexon transcript was spliced at 66% efficiency in tgs1Δ cells, signifying that shortening the 5′ exon overrode much of the Tgs1 requirement (Figure 8). The PCH2-BP transcript was spliced with 69% efficiency in tgs1Δ cells, implying that the non-consensus branchpoint is an independent determinant of Tgs1-dependence of PCH2 splicing. Combining the BP and exon changes elicited a further gain of splicing in tgs1Δ cells to 91% (Figure 8), the same as the maximum level attained in wild-type cells. Thus, the TMG requirement for PCH2 splicing is governed by two features of the pre-mRNA: the intron and its deviant branchpoint sequence and the long 5′ exon.
We inserted the 113-nt PCH2 intron within the chromosomal yeast HIS3 gene of TGS1 and tgs1Δ haploid yeast cells (Figure 9A). The PCH2 intron was placed near the 3′ end of the HIS3 ORF to reflect its distal position in the native PCH2 pre-mRNA. The HIS3-[PCH2] gene containing the inserted PCH2 intron was functional in TGS1 cells (as gauged by growth on agar medium lacking histidine) but not in the tgs1Δbackground (Figure 9B). The instructive finding here was that a single C-to-U mutation in the intron of the HIS3-[PCH2] reporter that restored a consensus branchpoint (Figure 9A) sufficed to restore HIS3 function in tgs1Δcells (Figure 7B). These results show that the PCH2 intron, by virtue of its deviant branchpoint, is an autonomous determinant of Tgs1-dependence in the absence of other PCH2 gene elements. To evaluate the contributions of 5′ exon length to the Tgs1-dependence of PCH2 intron removal, we repositioned the intron proximally within the HIS3 reporter (inserting it at the same site used to generate the HIS3-[SAE3] reporter). This maneuver sufficed to allow tgs1Δ cells to grow in the absence of added histidine (Figure 9B). These results implicate TMG caps in recognition and/or utilization of the non-consensus PCH2 branchpoint in the context of a pre-mRNA substrate with a relatively long 5′ exon.
The specificity of the splicing factor requirements for HIS3-[PCH2] reporter function was tested by introducing the expression cassette into the chromosome of yeast strains deleted for other vegetatively inessential splicing factors that display synthetic lethal/sick interactions with one another and with Tgs1. As expected, expression of the HIS3-[PCH2] reporter was impaired in nam8Δcells (Figure 9C), consistent with the Nam8 requirement for PCH2 pre-mRNA splicing during meiosis (16). The instructive findings were that HIS3-[PCH2] was active in mud1Δ and mud2Δcells (Figure 9C) and in swt21Δ cells (data not shown).
The inessentiality of Tgs1 and TMG caps for vegetative growth of budding and fission yeast (2,5) is paradoxical in light of the presence of TMG caps on the essential spliceosomal U1, U2, U4 and U5 snRNAs of nearly all eukaryal taxa (43). Synthetic genetic array analysis in budding yeast revealed that the effects of ablating the TMG cap are buffered by spliceosome assembly factors, that are themselves inessential for vegetative growth (6,11). Thus, Nature has overlaid genetic redundancy on an ancient eukaryal-specific RNA modification (the TMG cap) that participates in a defining step of eukaryal RNA biogenesis (spliceosome-catalyzed intron removal). This raises the important question of what RNA transactions, if any, are TMG cap-dependent per se.
The present finding that the absence of TMG caps on all yeast snRNAs does not impact in vitro splicing of intron-containing actin and U3 pre-RNAs is a biochemical validation of the genetic data that TMG caps are dispensable for yeast cell growth and, perforce, for adequate levels of in vivo splicing of the many intron-containing RNAs needed for vegetative growth under laboratory conditions. Earlier depletion–reconstitution studies with mammalian cell extracts had shown that that individual human U1, U2, U4 and U5 snRNPs could be reconstituted singly in a splicing-competent state from the respective synthetic snRNAs lacking a TMG cap structure (44–47). However, in U2 snRNA-depleted frog oocytes, a TMG cap was required for reconstitution of splicing by injected synthetic U2 snRNA (48). These seemingly contradictory findings regarding U2 function hinted that reliance on TMG caps for splicing might vary with cell type or developmental stage. The idea of context-dependent TMG requirements gained traction with the report that Tgs1 methyltransferase activity is critical during a discrete temporal phase of Drosophila early pupal development (10).
Here, we show that Tgs1 and TMG caps are essential for execution of the meiotic developmental program in budding yeast. We correlate the failure of tgs1Δdiploids to sporulate with a specific deficit in splicing of two meiotic pre-mRNAs—SAE3 and PCH2. SAE3 had not been recognized previously as a genetically vulnerable target of splicing control. PCH2 was identified recently as a novel target for Mer1-independent Nam8-dependent meiotic splicing control (16). Sae3 is essential for meiosis and sporulation (49), whereas Pch2, a meiotic checkpoint protein, is not normally required for sporulation in an otherwise wild-type background (50). Although it is conceivable that the severe defect in SAE3 splicing might explain the sporulation deficiency of the tgs1Δ strain, the modest decrements in splicing of AMA1 and MER3 pre-mRNAs in tgs1Δ cells (2.5-fold at 4h; Table 1) could also contribute to the phenotype. Note that AMA1 and MER3 each have one of the characteristics of Tgs1-dependent meiotic pre-mRNA targets, to wit: (i) AMA1, like PCH2, has an exceptionally long 5′ exon (1183-nt); and (ii) MER3 has a deviant branchpoint sequence (5′-GACUAAC).
Tgs1-dependent meiotic splicing is novel vis-à-vis the Mer1-driven mechanism characterized previously (23). Mer1 upregulates splicing of four meiotic transcripts: AMA1, MER2, MER3 and SPO22. All Mer1-regulated meiotic transcripts share three properties: (i) they have non-consensus 5′ splice sites; (ii) they have an intronic enhancer sequence to which Mer1 binds and (ii) they are co-dependent for splicing on Nam8 (Figure 10). By contrast, the Tgs1-dependent SAE3 and PCH2 transcripts have consensus 5′ splice sites and lack Mer1 intronic enhancer elements. The two Tgs1-dependent transcripts diverge with respect to whether they do (PCH2) or do not (SAE3) co-depend on Nam8 for their splicing during meiosis (Figure 10). The Tgs1-dependent transcripts also differ as to whether excision of their introns does (SAE3) or does not (PCH2) co-depend on Mud2 (Figures 5 and and99).
The positive impact of TMG caps on SAE3 and PCH2 pre-mRNA splicing is evinced in vegetative yeast cells, by comparing splicing efficiency in TGS1 versus tgs1Δ haploids. We infer that there are probably no strictly meiosis-specific positive coregulators of TMG dependency. Indeed, we see that vegetative splicing of SAE3 and PCH2 is unaffected by forced expression of Mer1, the only known meiosis-specific splicing regulator (26). The residual levels of SAE3 and PCH2 splicing in tgs1Δhaploids (26% and 18%, respectively) are somewhat higher than the corresponding splicing levels in tgs1Δ diploids undergoing meiosis (14% and 8%, respectively). Nonetheless, the reliable differences in splicing efficiencies of Tgs1-dependent transcripts in vegetative cells allowed for analysis of the RNA determinants of meiotic splicing in vegetative cells that do or do not have TMG caps.
Here, we focused predominantly on SAE3, the meiotic transcript that depends uniquely on Tgs1 (but not Nam8) for efficient splicing in vivo. The SAE3 intron has non-consensus branchpoint and 3′ splice site signals, either of which could conceivably enfeeble SAE3 splicing and make it more sensitive to factors that are inessential for splicing of consensus introns. It proved the case that the primary and secondary structures of the SAE3 branchpoint are key determinants of splicing efficiency and Tgs1 dependence, whereas the 3′ splice site had no obvious impact on either parameter. We demonstrated additive negative effects of the non-consensus branchpoint sequence and the predicted hairpin fold in which the branchpoint is embedded. Simultaneously restoring a consensus UACUAAC sequence and eliminating the potential for hairpin formation enhanced the efficiency of SAE3 splicing in TGS1 cells and abolished the differential in SAE3 splicing efficiency in tgs1Δ cells. The SAE3 intron per se is a portable determinant of Tgs1-dependent gene expression when transferred to the otherwise intron-less HIS3 gene. This result is important, because it militates against a scenario in which the requirement for TMG caps is enforced in cis, i.e. if the SAE3 pre-mRNA is itself targeted for TMG capping in TGS1 cells via some distinctive property of the SAE3 transcription unit (a la TMG capping of a fraction of RRE-containing HIV-1 pre-mRNAs in human cells; (51). In the HIS3 reporter context, the secondary structure of the SAE3 branchpoint is the dominant factor in Tgs1 dependence, i.e. deleting the proximal stem that embeds the SAE3 branchpoint restored histidine prototrophy to tgs1Δ cells. Changing the sequence of the proximal stem so as to prevent base pairing to the branchpoint (without deleting any nucleotides) also restored histidine prototrophy to tgs1Δ cells, when combined with U-to-C change in the branchpoint.
We recapitulated features of SAE3 splicing control in an in vitro splicing system. The yeast whole-cell extract efficiently excised introns from synthetic pre-RNAs derived from yeast transcripts that are known to be spliced well in vivo (e.g. actin and U3). By contrast, a synthetic SAE3 pre-mRNA was a conspicuously feeble substrate for splicing in vitro, as a consequence of the same negative influences at the branchpoint that affected SAE3 splicing in vivo. The extent of splicing of wild-type SAE3 intron in vitro (7%) was lower than the steady-state splicing level in vivo in vegetative cells (65%). This disparity could reflect the beneficial effects of cotranscriptional pre-mRNA processing in the yeast nucleus in vivo that are not attainable during splicing of an exogenous pre-mRNA in a yeast extract in vitro. Yet, our finding that simultaneous installation of a consensus branchpoint and elimination of the branchpoint hairpin allowed highly efficient splicing of the SAE3-(BP+Δhp) pre-mRNA—in TGS1 and tgs1Δextracts—validated the in vitro system as a gauge of cis-acting determinants of SAE3 splicing.
The partial gain of in vitro splicing elicited by the single SAE3-BP and SAE3-Δhp intron mutations allowed us to probe whether the absence of TMG caps on spliceosomal U snRNAs affected SAE3 splicing. The findings that splicing of these SAE3 introns was enhanced in TGS1 versus tgs1Δ extracts provided direct confirmation of the in vivo splicing phenotypes for these pre-mRNAs seen in TGS1 and tgs1Δhaploid cells and verified the independent influences of branchpoint sequence and secondary structure on the TMG cap-dependency of splicing. To our knowledge, this is the first instance in which an aspect of yeast meiotic splicing control has been modeled in vitro.
Our presumption, based on the gain-of-function branchpoint context changes, is that the TMG requirement for SAE3 splicing in meiosis pertains to a particular TMG-capped U snRNA (e.g. U2) or subset of snRNAs (e.g. U2 and U1) that orchestrate early spliceosome assembly and branchpoint recognition. We observed no accumulation of SAE3 splicing intermediates (e.g. lariat intron and 5′ exon) in TGS1 or tgs1Δ extracts, hinting that TMG exerts its effect during spliceosome assembly. A key question is whether TMG caps act directly to facilitate early steps in splicing (e.g. via TMG-to-RNA or TMG-to-protein interactions) or indirectly, whereby the absence of TMG caps affects snRNP structure or stability. It was reported that tgs1Δ yeast cells have grossly normal steady-state levels of spliceosomal snRNAs (2,5) and display no aberration in the sedimentation profiles of their spliceosomal snRNPs (2). We recently documented that affinity-purified snRNPs from TGS1 versus tgs1Δ S. cerevisiae cells are indistinguishable with respect to overall yield, snRNA content, and the sizes of the individual snRNAs (Schwer,B., Erjument-Bromage,H. and Shuman,S., manuscript in preparation). Our analysis of the polypeptide compositions of SmB affinity-purified snRNPs showed that the snRNPs from tgs1Δ cells were not lacking in any of the protein components found in the snRNPs from TGS1 cells (Schwer,B., Erjument-Bromage,H. and Shuman,S., manuscript in preparation). Such results suggest that the splicing phenotypes associated with Tgs1 deficiency are not caused by gross changes in snRNP biogenesis or structure. Further dissection of the role of the TMG cap will ultimately hinge on reconstituting SAE3 splicing in vitro with combinations of individual TMG-capped and m7G-capped snRNPs purified from TGS1 cells and tgs1Δcells, respectively.
We find that the PCH2 pre-mRNA has different determinants of Tgs1-dependency than SAE3. PCH2 has a non-consensus branchpoint sequence, 5′-CACUAAC, albeit without any predicted secondary structure when the intron is analyzed by Mfold. The PCH2 transcript is distinguished by its exceptionally long 5′ exon. Our studies of PCH2 splicing implicate the non-consensus branchpoint and long 5′ exon as separable negative influences on PCH2 splicing in TGS1 cells and as concerted determinants of the Tgs1-dependence of PCH2 splicing (Figure 10). The separable negative effects of 5′ exon length and deviant BP sequence were also evident when the PCH2 intron was imported to the HIS3 gene. (Here again, we take this to mean that the Tgs1 requirement is not imposed by hypermethylation of the PCH2 pre-mRNA.)
The elucidation of three genetically distinct meiotic splicing regulons, each with its own governing RNA determinants (Figure 10), highlights new modes of tunable RNA splicing in yeast (i.e. by TMG caps) and an unexpected richness of RNA controls during the budding yeast meiotic program.
Supplementary Data are available at NAR online.
U.S. National Institutes of Health grants GM52470 (to S.S.) and GM50288 (to B.S.). Funding for open access charges: NIH grants (GM52470 and GM50288).
Conflict of interest statement. None declared.
We thank Olivia Orta for expert technical assistance. S.S. is an American Cancer Society Research Professor.