The nuclear RNA and DNA helicase Sen1 is essential in the yeast Saccharomyces cerevisiae and is required for efficient termination of RNA polymerase II transcription of many short noncoding RNA genes. However, the mechanism of Sen1 function is not understood. We created a plasmid-based genetic system to study yeast Sen1 in vivo. Using this system, we show that (1) the minimal essential region of Sen1 corresponds to the helicase domain and one of two flanking nuclear localization sequences; (2) a previously isolated terminator readthrough mutation in the Sen1 helicase domain, E1597K, is rescued by a second mutation designed to restore a salt bridge within the first RecA domain; and (3) the human ortholog of yeast Sen1, Senataxin, cannot functionally replace Sen1 in yeast. Guided by sequence homology between the conserved helicase domains of Sen1 and Senataxin, we tested the effects of 13 missense mutations that cosegregate with the inherited disorder ataxia with oculomotor apraxia type 2 on Sen1 function. Ten of the disease mutations resulted in transcription readthrough of at least one of three Sen1-dependent termination elements tested. Our genetic system will facilitate the further investigation of structure–function relationships in yeast Sen1 and its orthologs.
TRANSCRIPTION termination by eukaryotic RNA polymerase II (Pol II) uses at least two pathways, one that is coupled to cleavage and polyadenylation of the nascent transcript [the poly(A)-dependent pathway] and one that involves the activity of the RNA/DNA helicase Sen1 (the Sen1-dependent pathway) (Kuehner et al. 2011). The Sen1-dependent pathway was first identified in the budding yeast Saccharomyces cerevisiae and is responsible for transcription termination of many short, noncoding RNA genes, including small nuclear (sn) and small nucleolar (sno) RNAs (Winey and Culbertson 1988; Steinmetz and Brow 1996; Rasmussen and Culbertson 1998; Steinmetz et al. 2001). It also restricts the elongation and accumulation of pervasive cryptic unstable transcripts (Arigo et al. 2006b; Thiebaut et al. 2006) and regulates transcription of some protein-coding genes by premature termination, i.e., attenuation (Steinmetz et al. 2001; Arigo et al. 2006a; Steinmetz et al. 2006b; Jenks et al. 2008; Kuehner and Brow 2008). A set of core factors distinguish the Sen1-dependent pathway from the poly(A)-dependent pathway, including Sen1 and the RNA-binding proteins Nrd1 and Nab3 (Steinmetz and Brow 1996, 1998; Conrad et al. 2000; Steinmetz et al. 2001; Carroll et al. 2007). However, some short messenger RNA (mRNA) genes, such as CYC1, may have hybrid terminators that require factors from both pathways (Steinmetz et al. 2006b).
S. cerevisiae Sen1 is a 252-kDa superfamily 1 helicase encoded by the essential SEN1 gene. Its helicase domain is located in the C-terminal half of the protein (Figure 1A), and the N-terminal 975 amino acids are dispensable for viability (DeMarini et al. 1992). The ortholog of Sen1 in the fission yeast Schizosaccharomyces pombe has ATP-dependent, 5′-to-3′ DNA and RNA unwinding activities in vitro (Kim et al. 1999). Two hypomorphic alleles of S. cerevisiae SEN1, sen1-1 and nrd2-1, have G1747D and E1597K substitutions in the helicase domain, respectively, and exhibit terminator readthrough on a subset of Pol II-transcribed genes (DeMarini et al. 1992; Steinmetz and Brow 1996; Steinmetz et al. 2001, 2006a; Kuehner and Brow 2008; Mischo et al. 2011; Hazelbaker et al. 2013). These findings indicate that the Sen1 helicase domain plays an important role in Pol II termination.
Two main models have been proposed for the mechanism of Sen1-dependent termination. In one model, Sen1 is proposed to be analogous to bacterial Rho helicase, translocating 5′ to 3′ along the nascent transcript until colliding with paused Pol II and terminating its elongation (Steinmetz and Brow 1996; Porrua and Libri 2013). In the other model, Sen1 unwinds RNA/DNA hybrids (R loops) formed between the nascent transcript and the template strand, thereby allowing recognition of termination elements in the nascent transcript by RNA-binding proteins (Mischo et al. 2011). Both models can account for an essential role of the Sen1 helicase domain in termination.
The human ortholog of yeast Sen1, called Senataxin, is encoded by the gene SETX. Mutations in SETX cosegregate with two progressive neurodegenerative disorders with juvenile onset: autosomal recessive ataxia oculomotor apraxia type 2 (AOA2) and autosomal dominant amyotrophic lateral sclerosis type 4 (ALS4) (Chen et al. 2004; Moreira et al. 2004; Lemmens et al. 2010). The mechanism by which the SETX mutations cause AOA2 and ALS4 has not been determined, but many of the missense disease mutations are located in the conserved helicase domain (Figure 1A and Supporting Information, Supporting References, and Table S1), suggesting that the diseases are associated with dysfunction of the helicase activity. Senataxin has been implicated in Pol II transcription termination on protein-coding genes (Skourti-Stathaki et al. 2011). Gene expression profiling showed that AOA2 mutations in SETX cause changes in gene expression profiles in patient fibroblasts, including genes involved in neurogenesis and neuronal functions (Fogel et al. 2014). Thus, it is possible that AOA2 and ALS4 result from defects in expression of genes essential for long-term neuron survival.
To better understand the role of Sen1 in the transcription termination process, we created a plasmid-based assay to characterize Sen1 function in vivo. We found that the essential region of Sen1 corresponds to the helicase domain and a flanking nuclear localization signal, but that efficient transcription termination by Sen1 requires its N-terminal domain as well. Using the crystal structure of the related helicase Upf1 as a guide, we determined that the growth and termination defects caused by nrd2-1 (sen1-E1597K) result from disruption of an intradomain salt bridge in the first RecA repeat. Examining the utility of Sen1 as a surrogate for Senataxin, we analyzed the effects of 13 human AOA2 mutations and showed that six are recessive lethal and that an additional five result in recessive termination defects. Our studies establish a facile genetic system for studying structure–function relationships in Sen1 and underscore the importance of the helicase domain in Sen1-dependent termination of Pol II transcription.
Materials and Methods
The SEN1 gene, including 434 bp upstream of the start codon and 212 bp downstream of the stop codon, was amplified from genomic DNA of the strain 46α by two PCR reactions. One created an upstream SalI site and an internal PstI site via a silent mutation in codon 965. The other created the same internal PstI site and a downstream SpeI site. (The sequences of all primers used in this study are available on request.) The PCR products were digested with the appropriate restriction enzymes and ligated into SalI/SpeI-cut pRS425 to create pRS425-SEN1(PstI). The SalI/SpeI insert from pRS425-SEN1(PstI) was subcloned into pRS313, pRS315, and pRS316. Point mutations in SEN1 were created by the QuikChange procedure (Stratagene) and subcloned into pRS313.
Deletions of SEN1 were created using pRS313-SEN1(PstI). The C-terminal sequences were deleted by the QuikChange procedure, leaving the stop codon intact. The initial N-terminal deletion was made by deleting codons 4–964 of the SEN1 ORF and adding a silent mutation to codon 3 to recreate the internal PstI site. In this construct, codons 3, 965, and 966 have the sequence TCT-GCA-GAA, with the PstI site underlined. In subsequent deletions, codons 1–3 and 965 were retained, and codon 966 was replaced with the indicated codon (e.g., 1004, 1089, etc.), which always begins with a G to retain the PstI site. The presence of this PstI site as well as one in the pRS313 vector sequence allowed construction of each plasmid by ligation of two PCR fragments. All deletion plasmids were sequenced across the deletion junction to confirm their identity. C-terminal deletions were added to N-terminal deletion constructs using the NcoI site at codons 1890–1892 and the downstream SpeI site.
Sen1-GFP fusion constructs were made from pRS315-SEN1(PstI). Using the QuikChange procedure, an internal AvrII site was created by silent mutation of codon 2230 to generate pRS315-SEN1(AvrII). Codons 2–238 of the GFP(S65T) variant were amplified from genomic DNA of a yeast strain (KES000) containing an IMD2-GFP gene fusion (derived from Life Technologies #95700). AvrII sites were created at both ends of the GFP(S65T) coding sequence to allow insertion into pRS315-SEN1(AvrII), generating pRS315-SEN1-GFP. N-terminal deletions of SEN1 in pRS313 were subcloned into pRS315-SEN1-GFP. To make GFP fusions of C-terminally deleted SEN1, sequences upstream of the AvrII site were deleted from pRS315-SEN1(AvrII) by the QuikChange procedure and the GFP(S65T) fragment was inserted into the AvrII site afterward. The only exception is that the C-terminal deletion after codon 1929 was made by inverse PCR from pRS315-SEN1-GFP and blunt-end ligation, removing the upstream AvrII site in the process.
The Senataxin/Sen1 chimeric constructs were made from pRS313-SEN1(PstI). For the entire SEN1-SETX ORF swap construct, SEN1 codons 4–2232(stop) were deleted and replaced with a GGA (Gly) codon to create a BspEI site by inverse PCR, digestion, and self-ligation. SETX codons 3–2678(stop) were PCR-amplified from pSETX (Suraweera et al. 2007) with a BspEI site at both ends. The PCR product was digested with BspEI and ligated into the BspEI-digested SEN1 ORF deletion construct to generate pRS313-SEN1/SETXORF. For the SEN1-SETX helicase domain swap construct, unique PmlI and NcoI sites were used to excise SEN1 codons 1175–1890 and replace them with SETX codons 1769–2484 to generate pRS313-SEN1/SETXhelicase.
pGAC24-CYC1 (Steinmetz and Brow 2003) and the pRS316-NRD1-CUP1 reporter plasmids (Kuehner and Brow 2008) have been described elsewhere. The pGAC24-SNR47 reporter plasmid was generated by amplifying the SNR47 terminator (Carroll et al. 2004) (positions +100 to +215, relative to +1 transcription start site) from DAB206 genomic DNA with an XhoI site introduced at each end. The PCR product was digested with XhoI and ligated to XhoI-cut pGAC24 (Lesser and Guthrie 1993).
Plasmids were transformed into S. cerevisiae by the lithium acetate procedure (Schiestl and Gietz 1989). DAB206 [MATα cup1Δ ura3 his3 trp1 lys2 ade2 leu2 sen1Δ3::TRP1 (pRS316-SEN1)] is derived from 46α (MATα cup1Δ ura3 his3 trp1 lys2 ade2 leu2) (Lesser and Guthrie 1993). The sen1Δ3::TRP1 allele was PCR-amplified from the genomic DNA of DDY44-5B (MATα leu2-Δ1 ura3-52 trp1-Δ1 lys2-801 his3-Δ200 ade2-101 sen1-Δ3::TRP1 YCp50-SEN1C), which is a haploid spore of DD44-N3 (DeMarini et al. 1992). 46α was cotransformed with pRS316-SEN1 and the sen1-Δ3::TRP1 fragment and plated to synthetic complete medium lacking uracil and tryptophan, resulting in replacement of most of the SEN1-coding sequence with the TRP1 gene. Disruption of the chromosomal SEN1 allele in the resulting strain, DAB206, was confirmed by PCR. pRS316-SEN1 was replaced with pRS313-borne SEN1 alleles by standard plasmid shuffle techniques (Boeke et al. 1987). Loss of pRS316-SEN1 was selected for on synthetic complete medium containing 0.75 mg/ml of 5-fluoroorotic acid (5-FOA).
XCY361 (MATα his3-Δ1 leu2-Δ0 lys2-Δ0 ura3-Δ0 NHP6A-TagRFP-T:nat) was derived from BY4742 (Research Genetics, Inc.) by integrating the NHP6A gene fused with TagRFP-T followed by the nourseothricin resistance (nat) gene into the endogenous NHP6A locus. The NHP6A-TagRFP-T/nat DNA fragment for chromosomal integration was made as follows. The NHP6A-mRFP1.3/nat DNA was amplified from strain EY2426 (Kim and O’Shea 2008) with HindIII and XbaI sites added to the ends. The DNA fragment was digested with HindIII and XbaI and inserted into the cloning vector pRS315. To improve photo-stability for time-lapse imaging, the mRFP1.3 ORF was replaced with the TagRFP-T ORF (Shaner et al. 2008). Splicing by overlap extension PCR was used to join NHP6A with TagRFP-T (Warrens et al. 1997). This NHP6A-TagRFP-T fragment was then subcloned into the existing pRS315-NHP6A-mRFP1.3/nat plasmid with HindIII and AscI sites, leaving the nat gene in place but replacing mRFP1.3 with TagRFP-T. The linear NHP6A-TagRFP-T/nat DNA was produced by HindIII and XbaI digestion of the resulting plasmid and transformed into BY4742, followed by selection for resistance to 100 µg/ml nourseothricin (Enzo Life Sciences). Integration at the NHP6A locus was confirmed by PCR of genomic DNA with flanking oligos and sequencing.
Rabbit antiserum against amino acids 1095–1876 of Sen1 purified from Escherichia coli was raised by Harlan Laboratories (Madison, WI). IgG was purified from antiserum using the Melon Gel IgG Spin Purification Kit (Thermo Scientific). For immunoblots of Sen1, 10 OD units of yeast cells growing logarithmically (OD600 of 0.5–1.0) at 30° in yeast extract/peptone/dextrose (YEPD) were pelleted by centrifugation and lysed by vortexing with glass beads (425–600 µm diameter; Sigma-Aldrich) in 100 µl phosphate-buffered saline (PBS) (137 mM NaCl, 2.7 mM KCl, 10 mM Na2HPO4, 1.8 mM KH2PO4, pH 7.4) with Protease Inhibitor Cocktail P2714 (Sigma-Aldrich), and 100 µl of 2× SDS/PAGE loading buffer [100 mM Tris–Cl, pH 6.8, 4% (w/v) SDS, 0.2% (w/v) bromophenol blue, 20% (v/v) glycerol, 200 mM β-mercaptoethanal] was added, and the mixture was heated at 100° for 5 min. Samples were electrophoresed using 4–15% Mini-PROTEAN TGX Precast Gels (Bio-Rad) at 160 V for 50 min and electroblotted onto Hybond-C extra nitrocellulose membranes (Amersham Biosciences) at 6 V/cm, 4° for 12 hr in 20% (v/v) methanol in Towbin buffer (25 mM Tris base, 192 mM glycine, pH 8.3). Blots were blocked with 5% dried milk in PBS buffer with 0.1% Tween-20 at 23° for 1 hr, incubated with purified Sen1 IgG (1:2000 dilution) for 1 hr and then with HRP-conjugated goat anti-rabbit IgG (Thermo Scientific) (1:50,000 dilution) for 1 hr. Blots were developed with Immobilon Western Chemiluminescent HRP Substrate (Millipore) and exposed to HyBlot CL Autoradiography Film (Denville Scientific). Immunoblots of Nrd1 were performed by the same protocol except the Nrd1 polyclonal antiserum (Steinmetz and Brow 1998) was diluted 1:1000.
XCY361 cells transformed with GFP-tagged SEN1 constructs in pRS315 were grown in synthetic complete medium lacking leucine at 30° to an OD600 of 0.5 and then diluted to an OD600 of 0.2. Two hundred microliters of culture was placed on the center of a glass coverslip precoated with 1 mg/mL Concanavalin A, Type IV (Sigma-Aldrich) solution, and viewed with a Nikon Eclipse Ti fluorescent microscope with a Nikon DS-Qi1Mc digital camera and NIS elements AR 4.12.01 software. Filter sets with excitation/emission wavelengths of 465–495/515–555 nm and 530–560/573–648 nm were used for the GFP images and red fluorescent images (RFP) images, respectively. Image files were processed and analyzed by ImageJ (National Institutes of Health) following the ImageJ User Guide (http://rsbweb.nih.gov/ij/docs/guide/146.html). For background subtraction, the command “Process>Subtract background>Create Background” was used to create the background, and the command “Process>Image Calculator” was used to subtract background from the original image. For contrast enhancement, the command “Image>Adjust>Brightness/Contrast” was used. In most cases, appropriate contrast was achieved by a single click on “Auto.”
Yeast cells were grown in YEPD at 30° to an OD600 of 0.8–1.0, quickly shifted to 37° by mixing with the same volume of 44° YEPD, continued growing for 1 hr, and were pelleted by centrifugation. Total cellular RNA was prepared by the glass bead-guanidinium isothiocyanate–hot phenol extraction method (Wise 1991). For Northern blots, 15–20 μg RNA was loaded in each lane, run on a 1% agarose/0.8 M formaldehyde gel, and blotted onto Zeta-probe membrane (Bio-Rad) by capillary transfer. The random-primed radioactive probes for NRD1 and SNR47 were generated using gel-purified PCR products as templates. The probe labeling protocol was adapted from the Prime-a-Gene Labeling System (Promega) using [α-32P]dTTP (PerkinElmer). After labeling, unincorporated nucleotides were removed by a G-50 Sephadex resin column (GE Healthcare). Probe hybridization was performed in Ultrahyb solution (Ambion) with probe concentration ∼106 cpm/ml. The blot was hybridized at 42° for >12 hr, washed in saline-sodium citrate (SSC) wash buffers (2× SSC, 0.1% SDS, 5 min per wash, twice, at 23°; 0.1× SSC, 0.1% SDS, 15 min per wash, twice, at 42°), and visualized with a Phosphorimager (Molecular Dynamics). Primers for PCR amplification of random-priming templates were the following: SNR47 fwd—5′-GGTCTCGAGATATATTTTCGCGTCATTCTTG-3′; SNR47 rev—5′-GGTCTCGAGCAGAAATAAAGAAAATGAAAGC-3′; NRD1 fwd—5′-CGAGTTACAGGAAAGGAACC-3′; and NRD1 rev—5′-TCTATTCCTTCTTGAGTTATATC-3′.
A plasmid-based system for characterizing yeast SEN1 function
Previous genetic studies of S. cerevisiae SEN1 used primarily chromosomal mutations and tested a small number of alleles (Winey and Culbertson 1988; Steinmetz and Brow 1996; Finkel et al. 2010). The essential chromosomal SEN1 gene was deleted previously (DeMarini et al. 1992), but the complementing plasmid used, although functional, was subsequently found to lack the SEN1 promoter, 5′-UTR, and part of the protein-coding region. To allow more facile analysis of mutant SEN1 alleles, we created a yeast strain, DAB206, that contains the sen1-Δ3 chromosomal gene disruption constructed by DeMarini et al. (1992), but the plasmid-borne wild-type SEN1 allele contains 434 bp upstream of the start codon and 212 bp downstream of the stop codon. Furthermore, the wild-type SEN1 allele is on a URA3-marked low-copy plasmid, pRS316, that can be selected against with 5-FOA to allow its replacement with a mutant allele (i.e., “plasmid shuffle”). The expression level of plasmid-borne SEN1 in strain DAB206 appears similar to the level of expression of chromosomal SEN1 in its parental strain, 46α (Figure 1B). We chose 46α as the parent of DAB206 because its chromosomal CUP1 genes are deleted (Lesser and Guthrie 1993), which allows the use of CUP1 reporter constructs to measure the extent of Sen1-dependent terminator readthrough (Figures 1, C–E). Although the CUP1 reporter assay itself cannot distinguish increased terminator readthrough from stabilization of a constitutive readthrough transcript, we previously used nuclear run-on and/or Pol II chromatin immunoprecipitation to show that Sen1-E1597K induces readthrough of all three terminators used in the CUP1 reporter assay (Steinmetz et al. 2006b).
Essential region of Sen1 corresponds closely to the helicase domain
We used the DAB206 strain to determine the minimal essential region of the Sen1 protein. Previously, a plasmid lacking the promoter, 5-UTR, and the first 975 codons of the SEN1 gene was found to complement a heat-sensitive mutation in SEN1 (DeMarini et al. 1992). We further delineated the essential region of Sen1 by systematically deleting its N- and C-terminal codons while leaving the first three codons, stop codon, and all flanking noncoding sequences intact. For most constructs, an additional alanine codon is present between codon 3 and the first downstream codon (see Materials and Methods). Mutant SEN1 alleles were introduced into DAB206 in a HIS3-marked, low-copy-number pRS313 vector (Figure 1C). Recessive viability of each mutant was determined by plating transformants onto medium containing 5-FOA (Figure 2A). For viable SEN1 mutants, their recessive growth phenotypes were scored at four temperatures: 16°, 23°, 30°, and 37° (Figure 2B).
Deletions of codons 4–964 and 4–1003 resulted in strong heat sensitivity and mild cold sensitivity. N-terminal deletion through codon 1088 strongly increased the cold sensitivity, whereas deletion through codon 1134 is recessive lethal at 30°. Sen1 1089-2231 appears to be overexpressed in vivo (Figure 2C), suggesting either that Sen1 levels are autoregulated or that the compromised function of the truncated protein at 30° selects for cells that have amplified the plasmid containing this allele. Deletion of the C-terminal 324 residues (codons 1908–2231) caused only weak heat sensitivity at 37° (Figure 2B) and no obvious change in expression level (Figure 2C). However, deletion of an additional 49 C-terminal codons (to codon 1858), including the extreme C terminus of the helicase domain, resulted in severe growth defects at all temperatures tested.
When N- and C-terminal codon deletions were combined, we identified the minimal length of the SEN1 ORF for viability, which includes codons 1004–1907 or 1089–1929 (Figure 2A). Both of these alleles confer strong cold and heat sensitivity (Figure 2B). Thus, the essential region of Sen1 is ∼900 amino acids long and corresponds closely to the helicase domain. However, a construct with only the 819 residues present in both viable constructs [codons 1089–1907; hereafter called “Sen1(HD)”] is not viable, so an essential function is supplied by either the N- or the C-terminal residues immediately flanking the helicase domain.
Analysis of copper sensitivity in the presence of CUP1-based termination reporter constructs showed that viable deletions of both the N- and the C-terminal regions of Sen1 result in terminator readthrough, although the N-terminal deletion exhibits stronger readthrough on all three terminators tested, consistent with its more severe high- and low-temperature growth defects (Figure 2D).
Nuclear localization of Sen1(HD) is required for cell viability
We sought to elucidate the essential function that is supplied by the residues flanking the helicase domain. Residues 1862–2093 of Sen1 were previously shown to contain a functional nuclear localization sequence (NLS) (Ursic et al. 1995), the N-terminal border of which was subsequently localized to residues 1890–1967 (Nedea et al. 2008). Moreover, a bipartite NLS-like sequence spans Sen1 residues 1908–1929 (DeMarini et al. 1992), which corresponds to the C-terminal essential flanking region (Figure 2A). We speculated that the flanking regions function redundantly in targeting the helicase domain to the nucleus. To test this hypothesis, we investigated localization of Sen1-GFP fusion proteins by fluorescent microscopy in the strain XCY361, which contains an intact chromosomal SEN1 gene and RFP-labeled nonhistone protein 6A (Nhp6A-RFP) as a nuclear marker (Figure 3).
Consistent with a previous study (Ursic et al. 1995), we found that wild-type Sen1-GFP predominantly localizes to the nucleus (Figure 3A). We further showed that the putative bipartite NLS is sufficient to direct nuclear localization, since GFP fused to Sen1 1907–1929 concentrates in the nucleus (Figure 3B). The subcellular distribution of Sen1(1089–1929)-GFP, which contains the bona fide NLS, resembles that of wild-type Sen1 (Figure 3C). Sen1(HD)-GFP, which contains only the helicase domain with no flanking sequences, displayed uniform distribution throughout the cell, with no nuclear enrichment (Figure 3D). However, Sen1(1004–1907)-GFP, which contains the helicase domain with N-terminal flanking sequence, exhibits weak nuclear localization (Figure 3E). This result indicates that the N-terminal flanking sequence has weaker NLS activity than the C-terminal NLS. Nevertheless, either the N- or the C-terminal flanking sequence is able to direct sufficient Sen1 helicase core to the nucleus to support cell viability at 30°.
To test if nuclear localization is the only essential function of the flanking sequences, we fused the SV40 NLS (PKKKRKVGIPQ) to the N-terminal end of Sen1(HD)-GFP. This fusion construct efficiently localized to the nucleus (Figure 3F). However, a strain containing SV40 NLS-Sen1(HD) grew extremely slowly even at 30° (Figure 2B). Therefore, while the SV40 NLS can rescue Sen1(HD) function, it does so much less efficiently than either the N-terminal or the C-terminal flanking sequences.
We noted that the Sen1(1004–1907)-GFP signal and the Nhp6A-RFP signal tend to segregate from one another within the nucleus, with the former exhibiting predominantly a crescent-shaped distribution (Figure 3G). Such a distribution pattern was not observed with any of the other Sen1-GFP constructs, suggesting that a subnuclear, perhaps nucleolar, localization signal exists in the N-terminal flanking region (see Discussion).
sen1-E1597K substitution disrupts an intradomain salt bridge
Using a positive selection for Sen1-dependent terminator readthrough, we previously isolated the nrd2-1 allele of SEN1, which harbors the missense mutation Glu1597 to Lys in the helicase domain (Steinmetz and Brow 1996). This residue is adjacent to motif II (Figure S1), which makes up part of the ATP-binding site, but the molecular defect of the E1597K mutation is unknown. Since no structural information is available for the Sen1 helicase domain, we used the yeast Upf1 helicase domain as a structure model (Chakrabarti et al. 2011). Based on sequence alignment (Figure S1), E1597 in Sen1 is equivalent to Upf1 E579, which forms a salt bridge with R623 in Upf1 that likely stabilizes the 1A domain (Figure 4A). Upf1 R623 aligns with R1641 in Sen1, so this salt bridge may be conserved in Sen1.
To test whether the molecular defect of sen1-E1597K is disruption of the putative salt bridge, we introduced a R1641E substitution. If the proposed salt bridge exists, sen1-R1641E and -E1597K should exhibit the same defects by breaking this salt bridge, but in combination they should suppress such defects by mutually restoring the salt bridge. As shown in Figure 4, B and C, R1641E alone conferred heat sensitivity and terminator readthrough phenotypes similar to those caused by E1597K. Moreover, the E1597K/R1641E double mutant was less heat-sensitive than either single mutant and exhibited no terminator readthrough. These results support our hypothesis that the nrd2-1 mutation disrupts a glutamate–arginine salt bridge between residues E1597 and R1641 that can be functionally replaced by a lysine–glutamate salt bridge. The success of this structure-guided mutational analysis demonstrates the utility of the Upf1 helicase domain structure as a predictive model for the Sen1 helicase domain structure, at least in the 1A and 2A domains, where the proteins are the most similar.
AOA2 disease mutations in yeast SEN1 cause growth defects and terminator readthrough
We next used our yeast genetic system to examine the effects of AOA2 disease mutations on Senataxin function. Initially, we attempted to complement the SEN1 deletion in DAB206 with pRS313-SEN1/SETXORF or pRS313-SEN1/SETXhelicase, in which the entire protein-coding region or just the helicase domain of SEN1, respectively, are replaced with the corresponding regions of the SETX cDNA (see Materials and Methods). However, the resulting strains did not grow on media containing 5-FOA, indicating that neither the entire SETX ORF nor its helicase domain can functionally replace their counterparts in SEN1. Therefore, we sought to characterize the disease mutations in the context of Sen1. This approach is restricted to the helicase domain as it is the only region where the sequence similarity is high enough to allow unambiguous identification of homologous residues. Among the 50 currently identified missense mutations that cosegregate with AOA2 or ALS4, 28 are located in the Senataxin helicase domain (Table S1). Most AOA2 mutations in the helicase domain are in residues conserved in Sen1. In contrast, the two ALS4 mutations in the Senataxin helicase domain (R2136C and R2136H) are in a residue that is not conserved in Sen1, so they cannot be tested in Sen1 (Figure S2).
We introduced 13 AOA2 mutations at analogous residues in Sen1 and characterized their effects in vivo (Table 1 and Figure S2). Wild-type Sen1 and the two previously isolated mutations, E1597K and G1747D, were included for comparison. Plasmid-borne mutant SEN1 alleles were transformed into the haploid yeast strain DAB206 to make merodiploid strains and observe their dominant effects. All 13 merodiploid strains were viable and grew normally at 30°, indicating that none of the disease alleles exhibit a dominant growth defect at this temperature. The merodiploid strains were then grown in medium containing uracil and plated on medium containing 5-FOA to select for clones that had lost the URA3-marked plasmid harboring the wild-type SEN1 allele. Six of 13 AOA2 mutants (Table 1 and Figure 5A) did not grow on medium containing 5-FOA, indicating that these AOA2 mutations render Sen1 inactive for an essential function. When these six alleles were tested for a dominant effect on readthrough of the SNR47 terminator, five exhibited snoRNA terminator readthrough, with sen1-I1370R being the only exception (Figure 5B). Thus, these five nonfunctional AOA2 mutant Sen1 proteins inhibit the function of the wild-type protein on the SNR47 terminator and therefore must accumulate to an appreciable level. In contrast, given that the I1370R substitution replaces a buried hydrophobic residue with a charged residue, it could potentially result in protein misfolding and degradation, which would preclude a dominant effect on wild-type Sen1.
All the viable haploid mutants grew similarly to the wild-type Sen1 strain at 16°, 23°, and 30°, but five of them displayed some degree of heat sensitivity at 37° (Figure 6, A and B). The W1166S substitution displays the strongest defect, followed by T1779P and then I1371F, N1413S, and K1788E with minor growth defects. F1767L and R1850H exhibited no temperature sensitivity. In summary, the severity of the AOA2 mutations in Sen1 is variable. Of the 13 mutants we tested, 6 of them are recessive lethal and another 5 cause detectable heat sensitivity.
We next tested if the viable AOA2 mutations cause defects in Sen1-dependent transcription termination. The haploid mutant strains were characterized in the copper resistance assay using the SNR47, CYC1, and NRD1 terminators. The viable mutants exhibited different levels of copper resistance (Table 1). Four of five heat-sensitive AOA2 mutants, sen1-W1166S, N1413S, T1779P, and K1788E, caused readthrough of one or more terminators used in our assay, and the rest of the mutants showed no readthrough effects. The terminators responded differently to the readthrough mutants. The NRD1 attenuator was responsive to all four AOA2 readthrough mutants, with sen1-N1413S showing less readthrough than the other three. Only sen1-T1779P caused SNR47 terminator readthrough, and no readthrough of the CYC1 terminator was observed in any of the AOA2 mutants. None of the AOA2 mutants caused readthrough of SNR47 or CYC1 terminators to the levels of the previously isolated readthrough mutant sen1-E1597K. In contrast, the founder sen1-1 mutation, G1747D, exhibits readthrough similar to the nrd2-1 mutant, consistent with its defect in pre-transfer RNA splicing being an indirect effect of transcriptional readthrough (Winey and Culbertson 1988; Steinmetz et al. 2001).
AOA2 mutations in Sen1 induce terminator readthrough at the NRD1 and SNR47 loci
Since the SNR47 terminator and NRD1 attenuator are in an artificial context in their respective CUP1 reporter genes, we sought to determine if termination defects also occur at the corresponding genomic loci in the presence of the AOA2 mutations. The strains carrying plasmid-borne wild-type SEN1, sen1-E1597K, -W1166S, -T1779P, or -K1788E were chosen for analysis by Northern blotting. All the cultures were shifted to 37° for 1 hr before RNA extraction to enhance the defects from the mutants.
The Northern blot results are shown in Figure 7A and quantified in Figure 7B. Previously, the chromosomal nrd2-1 mutant (sen1-E1597K) was found to accumulate excess NRD1 mRNA (due to readthrough of the attenuator) and extended SNR47 transcripts (Steinmetz et al. 2001). We observed similar results with the plasmid-borne sen1-E1597K allele. Terminator readthrough at these loci was also observed in some of the disease mutants. For the NRD1 gene (Figure 7, A and B, left), a small increase in transcripts in the sen1-W1166S and -T1779P strains implies decreased transcription attenuation. We did not observe excess NRD1 mRNA in the sen1-K1788E strain despite its strong copper resistance in the presence of the NRD1/CUP1 reporter gene (Table 1). This inconsistency could be explained by the fact that the endogenous NRD1 attenuator extends well into the Nrd1-coding region (Arigo et al. 2006a), which is not included in the NRD1/CUP1 fusion construct. K1788E may promote readthrough of the truncated NRD1 attenuator present in the reporter construct, but not the complete attenuator present at the endogenous NRD1 locus.
For the SNR47 locus (Figure 7, A and B, right), accumulation of SNR47-YDR042C readthrough transcripts was observed in the sen1-W1166S and T1779P mutants but not in sen1-K1788E. T1779P caused strong readthrough of SNR47 to a level similar to E1597K. Modestly elevated readthrough of SNR47 in the sen1-W1166S strain was detected in the Northern blot assay but not in the copper resistance assay, perhaps because this mutant protein is sensitive to increased temperature and RNA was prepared after a shift to 37°, while the copper resistance assay was done at 30°. In summary, our Northern blotting assays revealed termination defects caused by the AOA2 mutants at the endogenous loci, which in general agree with our CUP reporter results.
Several different functions in DNA and RNA transactions have been attributed the essential yeast Sen1 protein and its human homolog, Senataxin (Brow 2011). However, little is known about the natural substrates of its helicase activity and which are related to its essential function. We have developed a genetic system for in vivo analysis of structure–function relationships in Sen1 and have demonstrated the utility of this system by addressing several open questions. In particular, we determined that (1) the minimum essential region of Sen1 is the helicase domain plus a flanking nuclear localization signal; (2) the nrd2-1 mutation (E1597K) disrupts an intradomain ionic bond with R1641; and (3) AOA2 disease mutations in the Senataxin helicase domain have a broad range of effects when introduced into the homologous position of Sen1.
Multiple domains of Sen1 are involved in transcription termination
Only two conditional-lethal missense mutations in the SEN1 gene have previously been identified, and both are in conserved residues of the helicase domain (DeMarini et al. 1992; Steinmetz and Brow 1996). Both mutations are known to cause readthrough of certain transcription terminators by Pol II, indicating the important role of the helicase domain in transcription termination. Our deletion analysis shows that nuclear localization of the Sen1 helicase domain is sufficient for viability, but that efficient transcription termination and normal growth at a range of temperatures also requires the N- and C-terminal domains of Sen1 (Figure 2D). The strong readthrough effects of Sen1 965–2231 agree with the previous observation that the N-terminal part of Sen1 interacts with the Ser5-phosphorylated Pol II C-terminal domain (CTD) and that impairment of this interaction by the mutation R302W decreases Sen1 occupancy on noncoding RNA genes (Chinchilla et al. 2012). Therefore, deleting the N-terminal region could dissociate Sen1 from the Pol II CTD, rendering it unable to efficiently recognize terminators. The C-terminal deletion of Sen1 (Sen1 1–1907) resulted in readthrough of the NRD1 attenuator. This could be caused by disrupting its interaction with Nab3 or other termination factors, such as the phosphatase Glc7 (Nedea et al. 2008). However, we observed only moderate readthrough in this deletion mutant strain, suggesting that the interactions occurring in the C-terminal end are not critical for snoRNA termination or are functionally redundant.
By tracking Sen1-GFP fusion proteins in merodiploid strains (Figure 3), we showed that either the N- or the C-terminal flanking region can target the Sen1 helicase domain to the nucleus. The C-terminal flanking region contains a classical bipartite NLS that directs efficient import and is likely recognized by importin-α (Srp1) (Marfori et al. 2011). The N-terminal flanking region does not contain a classical NLS and provides weak yet site-specific subnuclear localization activity. The resulting distribution pattern of Sen1(1004–1907)-GFP resembles the shape of the nucleolus, which is not observed in the full-length protein, suggesting a different mechanism for the localization activity of the N-terminal flanking region. The importance of this region in Sen1 has been revealed by our deletion analysis, as removing it strongly increased cold sensitivity (Figure 2B). We hypothesize that the N-terminal flanking region interacts with snoRNP proteins, which serves to bring it into the nucleus and localize it to the nucleolus. Such an interaction is presumably more transient in full-length Sen1, where it may serve to recruit Sen1 to cotranscriptionally assembling snoRNPs, increasing the efficiency of transcription termination on snoRNA genes (Ballarino et al. 2005; Yang et al. 2005).
The heterologous SV40 monopartite NLS directs very efficient nuclear localization of the Sen1(HD), but results in very low viability. It is possible that a strong NLS at the N terminus of the helicase domain interferes with helicase function in some way.
Terminator readthrough is a common effect of the AOA2 mutations in Sen1
We tested 13 AOA2 mutations from the Senataxin helicase domain in Sen1, and 77% (10 of 13) caused dominant or recessive terminator readthrough (Figure 5B and Table 1), indicating that defects in transcription termination is a common feature of AOA2 mutations in yeast. The Upf1 helicase domain structure provides insights into the molecular processes disrupted by the AOA2 mutations (Figure 5A and Figure 6A). Some of the mutations that cause readthrough localize to the catalytic core and are expected to disrupt binding of nucleic acid (N1413S and T1779P) or ATP (D1616V and R1820Q) or the coupling of these binding events (P1622L) (Fairman-Williams et al. 2010). Other mutations are in the hydrophobic cores of the RecA domains (I1370R, I1371F, C1409Y, L1569W, and K1788E) and may alter the overall fold of the helicase domain. No growth or termination defects were observed with the F1767L and R1850H AOA2 mutations. These residues are on the exterior surface of the helicase domain and may be involved in Senataxin-specific interactions.
In this study we examined the effects of SEN1 mutations on Pol II transcription termination. Functions of Sen1 in other nuclear processes have been proposed, including Pol I transcription termination, DNA repair, RNA processing, and maintenance of genome stability (Ursic et al. 2004; Kawauchi et al. 2008; Finkel et al. 2010; Mischo et al. 2011; Alzu et al. 2012). Defects in these other processes may contribute to the growth defects that we observed in the presence of the AOA2 substitutions.
Potential of yeast for identifying Senataxin disease variants
Given that 2 of the 13 AOA2 mutations that we tested in Sen1 cause no growth or termination defect, our yeast genetic system has a significant false-negative rate in detecting disease-causing mutations in Senataxin. To determine the false-positive rate, we would need to test single nucleotide polymorphisms (SNPs) in Senataxin that are known not to cause disease in humans and can be aligned accurately with the Sen1 sequence. However, no such SNPs have been identified.
We further wondered if there is a correlation between the yeast growth phenotypes and the severity of AOA2 mutations in humans. As shown in Table 1, the tested AOA2 mutations were categorized into “lethal,” “heat sensitive,” and “normal” groups based on their growth phenotypes in yeast. Using published data, we determined the average age of onset and the average serum α-fetoprotein (AFP) level for patients with the three observed combinations of alleles from these three phenotypic groups, namely two normal alleles, two lethal alleles, and one each of the heat-sensitive and lethal alleles (Table 2). For comparison, we also calculated average values for patients who have two presumptive null alleles (nonsense mutation, frameshift mutation, or a large deletion in the helicase domain). If the yeast phenotype were predictive of the human disease severity, we would expect the individuals with two “normal” alleles to have a significantly later age of onset and/or a lower serum AFP level. However, this analysis yielded no significant difference (using a t-test with P < 0.05) in either the age of onset or AFP level between any two of the yeast phenotype groupings. Therefore, although our yeast genetic system provides a means for identifying potential disease-causing sequence variants in Senataxin, the severity of AOA2 mutations in yeast shows no apparent correlation with severity in the AOA2 population for which we could access clinical data.
Notably, we found that all but one of the AOA2 mutations that are lethal in Sen1 exhibit dominant readthrough of the SNR47 terminator (Figure 5B). Whether the corresponding SETX alleles cause any dominant effects in AOA2 carriers is not known. Although AOA2 is classified as a recessive disorder, our results in yeast suggest that a carrier trait may be detectable at the molecular level.
While the utility of our yeast genetic system for the characterization of the phenotypic consequences of human genetic variants is as yet uncertain, we have clearly established its utility for investigating structure–function relationships in yeast Sen1. More detailed analysis of the effects of the AOA2 mutations in Sen1, for example, by transcriptome studies, may yield insight into the nature of the molecular defects that lead to the symptoms of AOA2.
We thank Michael Culbertson, Christine Guthrie, Harold Kim, and Erin O’Shea for yeast strains; Jon Audhya and Martin Lavin for plasmids; Tom Gisel for plasmid construction; Elaine Brow and Christine Treba for creating mutant SEN1 alleles; Steve Martin-Tumasz for generating Sen1 antiserum; Lynn Weaver for technical assistance; Aaron Hoskins and David Wassarman for use of their microscopes; and Brow lab members for helpful discussions. This work was funded by grant R01 GM082956 from the National Institutes of Health (NIH) (to D.A.B.). K.E.S. was supported in part by NIH training grant T32 GM08349.
Supporting information is available online at http://www.genetics.org/lookup/suppl/doi:10.1534/genetics.114.167585/-/DC1.
Communicating editor: A. Hinnebusch
- Received June 20, 2014.
- Accepted August 11, 2014.
- Copyright © 2014 by the Genetics Society of America