|Home | About | Journals | Submit | Contact Us | Français|
Tra1 is a 3744-residue component of the Saccharomyces cerevisiae SAGA, NuA4, and ASTRA complexes. Tra1 contains essential C-terminal PI3K and FATC domains, but unlike other PIKK (phosphoinositide three-kinase–related kinase) family members, lacks kinase activity. To analyze functions of the FATC domain, we selected for suppressors of tra1-F3744A, an allele that results in slow growth under numerous conditions of stress. Two alleles of TTI2, tti2-F328S and tti2-I336F, acted in a partially dominant fashion to suppress the growth-related phenotypes associated with tra1-F3744A as well as its resulting defects in transcription. tti2-F328S suppressed an additional FATC domain mutation (tra1-L3733A), but not a mutation in the PI3K domain or deletions of SAGA or NuA4 components. We find eGFP-tagged Tti2 distributed throughout the cell. Tti2 is a component of the ASTRA complex, and in mammalian cells associates with molecular chaperones in complex with Tti1 and Tel2. Consistent with this finding, Tra1 levels are reduced in a strain with a temperature-sensitive allele of tel2. Further agreeing with a possible role for Tti2 in the folding or stabilization of Tra1, tra1-F3744A was mislocalized to the cytoplasm, particularly under conditions of stress. Since an intragenic mutation of tra1-R3590I also suppressed F3744A, we propose that Tti2 is required for the folding/stability of the C-terminal FATC and PI3K domains of Tra1 into their functionally active form.
Tra1 and its human homolog TRRAP are members of the PIKK (phosphoinositide three-kinase–related kinase) family of proteins. The family also includes the key cellular regulators ataxia telangiectasia mutated (ATM), ATM and Rad3-related (ATR), DNA-dependent protein kinase catalytic subunit (DNA-PKcs), mammalian target of rapamycin (mTOR), and suppressor with morphological effect on genitalia family member (SMG-1), many with yeast equivalents (Abraham 2004; Lovejoy and Cortez 2009). Tra1 is essential for viability in Saccharomyces cerevisiae and a major constituent of the multisubunit SAGA and NuA4 transcriptional regulatory complexes (Grant et al. 1998; Saleh et al. 1998; Allard et al. 1999). SAGA and NuA4 have been well studied with regard to their functions, playing roles in multiple aspects of gene regulation and DNA repair (Doyon and Cote 2004; Rodriguez-Navarro 2009; Koutelou et al. 2010). Both possess histone acetyltransferase activity, the catalytic subunits being Gcn5 and Esa1 for SAGA and NuA4, respectively (reviewed in Roth et al. 2001). SAGA contains additional modules, critical for regulation, that function in the deubiquitylation of histone H2B (Henry et al. 2003) and interaction with the TATA binding protein (Dudley et al. 1999; Mohibullah and Hahn 2008). Structural data indicate that Tra1 comprises a separate module in SAGA and NuA4 (Wu et al. 2004; Chittuluro et al. 2011). The position of Tra1 in these complexes is not indicative of a scaffold function, a result consistent with Tra1 not being required for stability of Schizosacccharomyces pombe SAGA (Helmlinger et al. 2011). This function is primarily ascribed to Spt7 for SAGA and Eaf1 for NuA4 (Sterner et al. 1999; Auger et al. 2008). TRRAP was first identified through its interactions with the transcription factors c-myc and E2F (McMahon et al. 1998). Similarly Tra1 interacts with yeast transcription factors, targeting SAGA and NuA4 to specific promoters (Brown et al. 2001; Bhaumik et al. 2004; Fishburn et al. 2005; Reeves and Hahn 2005). Independent functions for Tra1 likely also exist since Helmlinger et al. (2011) found that in S. pombe, deletion of the SAGA-specific molecule spTra1 results in changes in gene expression distinct from loss of other SAGA subunits.
On the basis of the association of the members in a series of immunoprecipitation experiments, Shevchenko et al. (2008) identified Tra1 in a novel complex they termed ASTRA (ASsembly of Tel, Rvb, and Atm-like kinase). The components of ASTRA are encoded by essential genes in S. cerevisiae and include: Tra1, Rvb1, Rvb2, Tel2, Tti1, Tti2, and Asa1. In S. pombe, Tra1 but not Tra2, is found in ASTRA (Helmlinger et al. 2011). Rvb1 and Rvb2 are components of multiple regulatory complexes (in S. cerevisiae the Ino80, Swr1, and R2TP complexes), likely in multimeric form (Jha and Dutta 2009; Huen et al. 2010). Both belong to the AAA+ (ATPase associated with diverse cellular activities) family of ATP binding proteins. Lustig and Petes (1986) identified TEL2 in S. cerevisiae through mutations that result in generation-dependent telomere shortening. The association of mammalian and fission yeast Tel2 with several PIKK family members suggested a broader role (Hayashi et al. 2007; Takai et al. 2007). Tti2 and Tti1 were implicated in Tel2-dependent processes on the basis of their mutual association in yeast and mammalian cells (Hayashi et al. 2007; Takai et al. 2007, 2010; Hurov et al, 2010; Kaizuka et al. 2010). At least one role for Tel2, Tti1, and Tti2 (TTT complex, Hurov et al. 2010) is likely in the folding/maturation of PIKK proteins since they affect the steady-state levels and bind newly synthesized PIKK molecules (Takai et al. 2007, 2010; Horejsí et al. 2010; Hurov et al. 2010; Kaizuka et al. 2010). Furthermore, the TTT complex is functionally associated with Hsp90, Hsp70, Hsp40, and the R2TP/prefoldin-like complex (Horejsí et al. 2010; Takai et al. 2010). In a recent study of essential genes, tti1 and asa1 were implicated in chromosome instability (Stirling et al. 2011). Stirling et al. (2011) also demonstrate that knockdown mutants of these genes, as well as tti2 and tel2, result in telomere shortening and a reduction in Tor1 protein levels.
The PIKK family members are large proteins (3744 residues for Tra1), characterized by a C-terminally positioned domain that resembles the phosphatidylinositol-3-kinases (PI3K; Lempiäinen and Halazonetis 2009). Unlike the other PIKK family members, Tra1/TRRAP lack kinase activity (McMahon et al. 1998; Saleh et al. 1998). Despite this, altering residues that parallel key regions of the kinase members affect Tra1 function (Mutiu et al. 2007). Interestingly one of these mutations, tra1-SRR3413 results in age-dependent telomere shortening. On the N-terminal side of the PI3K domain is a FAT (FRAP-ATM-TRRAP) domain that consists largely of HEAT (Huntington, elongation factor 3, PR65/A, TOR) and TPR (tetratricopeptide) repeats; indeed much of the protein is likely helical (Bosotti et al. 2000; Perry and Kleckner 2003; Sibandi et al. 2010; Knutson and Hahn 2011). To the C-terminal side of the PI3K domain is the less highly conserved PRD (PIKK regulatory domain), the site of acetylation of ATM by TIP60 (Sun et al. 2005; for a linear schematic of the domains see Lempiäinen and Halazonetis 2009).
At the extreme C terminus of the PIKK molecules is the 30–35 residue FATC domain (FAT C-terminal; Bosotti et al. 2000); the conservation of the domain is evident from the finding that some FATC domains can be exchanged without loss of function (Jiang et al. 2006). Addition of as little as a single glycine to the C terminus of Tra1 results in loss of viability (Hoke et al. 2010). Other FATC mutations in Tra1 cause growth defects, such as temperature sensitivity and slow growth on media containing ethanol, Calcofluor white, or rapamycin. The FATC domain is similarly important for the other PIKK family members (Priestley et al. 1998; Beamish et al. 2000; Takahashi et al. 2000; Sun et al. 2005). Of note, the parallel mutation to L3733A of Tra1 results in a dramatic loss in the kinase activity of SMG-1 (Morita et al. 2007). The structure of the isolated FATC domain of S. cerevisiae Tor1 has been determined (Dames et al. 2005). It is largely helical in structure with a C-terminal loop held in place by a disulphide linkage. While the helical structure is likely conserved, the absence of the cysteines in the other FATC domains suggests the loop is not. FATC domains are proposed to serve as the target site for protein interactions. In a two-hybrid analysis the FATC domain of Mec1, the S. cerevisiae homolog of ATR, was required for association with the RPA components Rfa1 and Rfa2 (Nakada et al. 2005). Similarly the FATC domain of ATM was shown to interact with Tip60 (Sun et al. 2005); however, a recent report suggests that this is indirect (Sun et al. 2010). Lempiäinen and Halazonetis (2009) speculate that the FATC and PRD domains regulate the kinase domain through interactions with the activation loop similar to the helical domains of the PI3 kinases.
We selected for suppressors of the temperature and ethanol sensitivity of tra1-F3744A with the goal of identifying roles for the FATC domain. Two mutations in tti2 were found in independent selections as partially dominant suppressors of the growth-related phenotypes and transcriptional changes caused by tra1-F3744A. tti2-F328S suppressed a second FATC mutation but not a mutation within the PI3K domain or deletion of SAGA or NuA4 components. Consistent with the documented role of the TTT complex (Takai et al. 2007, 2010; Horejsí et al. 2010; Hurov et al. 2010; Kaizuka et al. 2010), Tra1 levels were reduced in a strain temperature sensitive for Tel2, and an increased level of cytoplasmic Tra1-F3744A was reversed by tti2-F328S. We predict that the basis for the suppression results from Tti2 having a role in the formation, stability, and/or localization of active Tra1. Our finding that an intragenic mutation of arginine 3590 to isoleucine within the putative activation loop of Tra1 also suppresses the F3744A mutation supports the models of Lempiäinen and Halazonetis (2009) and Sturgill and Hall (2009) that folding of the PIKK molecules involves the interaction of FATC and PI3K domains.
Yeast strains, listed in Table 1, are derivatives of BY4741 and BY4742 (Winzeler and Davis 1997). HIS3-linked tra1 strains, wild-type TRA1 (CY4353), tra1-L3733A (CY4057 and CY4103) and tra1-F3744A (CY4350 and CY4351), are described in Hoke et al. (2010). The HIS3 allele marking these alleles is placed at the BstBI site of YHR100C/GEP4, 11 codons from the C terminus. A copy of YHR100C on YCPlac111 or YCplac33 was present in each of these strains to ensure its functionality. Diploid strains containing one Flag5-tagged TRA1 allele marked with URA3 (CY4398) containing TRA1-HIS3 (CY4419) or tra1-F3744A-HIS3 (CY4421) were similarly constructed as were N-terminally eGFP-tagged tra1 strains. The haploid tra1-F3744A strain (CY5524) containing 5′-URA3-Flag5 and linked to 3′ HIS3 was obtained after sporulation. CY5639 and CY5640 containing Flag5-tra1-R3590I and Flag5-tra1-3590I, F3744A were also generated after integration into CY4398 and sporulation. The tra1-F3744A tti2-F328S (CY5567 and CY5669) and tra1-F3744A tti2-I336F (CY5843) double mutant strains were obtained in the selection process described below. The tti2-F328S strain CY5665 was obtained after crossing CY5667 with BY4742. Double mutants with ada2Δ0 (CY5876), eaf3Δ0 (CY5916), and eaf7Δ0 (CY5917) were generated after mating CY5667 with the consortium knockout strains (BY4842, BY7143, and BY4940 obtained from Open Biosystems) and screening Kanr-resistant spore colonies for the tti2 allele by sequencing of PCR products. The tti2-F328S double mutant with tra1-L3733A was obtained after a cross with CY4103. The spt7::LEU2 deletion strain used for double mutant analysis (CY5873) was created after five backcrosses of FY1093 (kindly provided by Fred Winston) with BY4741. RFP-tagged strains in the EY0987 background (MATα his3Δ1 lys2Δ0 ura3Δ0; Huh et al. 2003) were kindly provided by Peter Arvidson. Heterozygous diploid strains containing an RFP-tagged allele and eGFP-tra1-F3744A were made by mating the EY0987 derivatives with CY6018. CY6146 and CY6148 containing tra1-F3744A and either tti2-1 (Stirling et al. 2011) or tel2-15 (Grandin et al. 2012), both temperature-sensitive alleles, were produced after mating of CY4350 with PSY561 or PSY42 (kindly provided by Peter Stirling), respectively, and sporulation. CY6141 with Flag5-tagged Tra1 and tel2-15 were produced from a cross of PSY43 with CY5919. CY6137 was derived after a cross of CY2222 and CY5665.
Growth comparisons were performed on YP media containing 2% glucose (YPD) selective plates after 3–5 days at 30° unless stated otherwise. Standard concentrations used for the selections were: 0.03% methyl methanesulfonate (Sigma-Aldrich), 7.5 μg/ml Calcofluor white (Sigma-Aldrich), 1.0 μg/ml tunicamycin (Sigma-Aldrich), 6% ethanol, 60 μg/ml brefeldin (LC Laboratories, Woburn MA), 1.2 M NaCl, 1.0 M sorbitol, 20 μg/ml geneticin (Sigma-Aldrich), 1.0 μg/ml staurosporine (LC Laboratories), 1.0 μg/ml phleomycin (Sigma-Aldrich), and 1 ng/ml rapamycin (LC Laboratories).
Promoter-lacZ fusions cloned as his3-lacZ fusions into the LEU2 centromeric plasmid YCp87 and have been described (Brandl et al. 1993; Mutiu et al. 2007; Hoke et al. 2010). Molecules for integrating tra1 alleles linked to HIS3 are described in Hoke et al. (2010). Integrative vectors to generate Flag5 and eGFP fusions of Tra1 were constructed in cassettes using as the base a molecule synthesized by Integrative DNA Technologies. This molecule pCB2143 (see Supporting Information, Figure S1), cloned into pTZ19 (lacking the polylinker HindIII site) as a SphI–BamHI fragment contains an SphI site at −549 relative to the translational start site of TRA1, a HindIII site inserted at −351, the sequence ACAAAATGTCAGGATCC surrounding the translational start and a BamHI–SalI fragment encoding a Flag5 tag followed by a NotI site in the alanine frame, to which the coding region of TRA1 to the Kpn1 site is added (Saleh et al. 1998). Inserted into this is the 1.1-kbp HindIII genomic fragment encoding URA3. The integrative plasmid was switched to eGFP by the replacement of the Flag cassette with a BamHI–Not1 cassette encoding eGFP (Hoke et al. 2008a). Myc9-tagged TTI2 and tti2-F328S were expressed from the DED1 promoter in YCplac111 by inserting a Not–SstI fragment amplified from genomic DNA using oligoncleotides 5693-1 and 5693-2 (Table 2) downstream of the DED1 promoter-myc9 cassette (Hoke et al. 2010). N-terminally eGFP-tagged TTI2 was expressed from the DED1 promoter on YCplac33 (Hoke et al. 2008a). pYHR100C/GEP4 genomic DNA was inserted into YCplac111 or YCplac33 as a BamHI–EcoRI fragment after PCR with oligonucleotides 4966-1 and 4966-2.
CY4350 containing YCplac111-YHR100C was grown to stationary phase in YPD. A total of 10 μl of culture, ~2 million cells, was plated onto YPD and UV irradiated at a wavelength of 302 nm for 10 sec. Survival was ~10%. Colonies growing at 37° were colony purified under nonselective conditions and reanalyzed for growth at 37° and on YPD plates containing 6% ethanol. The suppressor strains were crossed with the URA3-tagged tra1-F3744A strain, CY5928, to determine linkage of the suppressor with tra1-F3744A. One strain had an unlinked suppressor mutation that segregated in a 2:2 fashion. A MATα spore colony was backcrossed with CY4350 nine times at each stage selecting for temperature and ethanol-resistant spore colonies. The final isolate was named CY5667. This was sent for genomic sequencing as described below. The tti2 mutations were verified after isolation of genomic DNA, PCR with oligonucleotides 6061-1 and 5693-2, and sequencing of the PCR product using 6061-1 as primer. The selection was repeated independently on three plates using an initial selection at 35° with 4% ethanol. Again one unlinked suppressor mutation was obtained. This was backcrossed seven times with CY4351 with the resulting strain called CY5843. The tti2 allele was sequenced after PCR with oligonucleotides 5693-1 and 5693-2 using PWO polymerase (Roche Applied Science) and cloning of the resulting PCR product in pUC18.
Three genomic DNAs were prepared, two strains containing the suppressor from independent tetrads and a control strain lacking the suppressor from the same tetrad as CY5667. Genomic DNA was prepared from 10 ml of lyticase-treated cells (Ausubel et al. 1988). Removal of RNA was verified by gel electrophoresis. Approximately 5 µg of DNA from each sample was sent to the Centre for Applied Genomics (Toronto, Ontario) for DNA library construction and next-generation sequencing using paired-end reads with the Applied Biosystems SOLiD 4.0 platform. The S. cerevisiae genome sequence was downloaded from the Saccharomyces Genome Database (SGD; http://www.yeastgenome.org) on March 24, 2011. Custom bash and Perl scripts were written for the sequencing analysis. The program Bowtie (Langmead et al. 2009), allowing up to three mismatches per read, was used to map the colorspace reads to each chromosome of the yeast genome and obtain mapped reads in SAM format (Sequence Alignment/Map; Li et al. 2009). The variant call format (VCF) from SAMtools (Li et al. 2009) was used to obtain a raw list of polymorphisms from the mapped reads. Those reads with a Phred quality score <20 were eliminated to obtain a filtered list of polymorphisms. A custom Perl script was written to account for the polymorphisms found in wild-type samples. We do note that ~5% of the TTI2 reads for the suppressor strains were wild type, not containing the F328S codon. We believe these represent errors in the reading or synthesis of the barcodes on the 14 other yeast sequences in the lane.
Yeast strains containing lacZ-promoter fusions were grown to stationary phase in media lacking leucine. Assays with PHO5-lacZ in media depleted of phosphate, and HIS4-lacZ in media lacking histidine were performed as described in Mutiu et al. (2007), GAL10-lacZ in media containing 2% galactose as the carbon source in Brandl et al. (1993), and for RPL35a-lacZ in YPD as described by Hoke et al. (2010). Ο-nitrophenol-β-D-galactosidase was used as substrate and values normalized to cell density.
ChIP assays using anti-H3 (Abcam; ab1791), anti-AcH4 K8 (ab1760) and anti-AcH3 K18 (ab1191) were performed as described by Mutiu et al. (2007). Input chromatin for the immunoprecipitations was prepared from yeast strains CY4353, CY4350, and CY5667, and normalized by PCR analysis of serial dilutions using oligonucleotides 5583-1 and 5583-2 to the PHO5 promoter. Agarose gels were stained with ethidium bromide and bands quantified using AlphaImager 3400 software (Alpha Innotech, San Leandro, CA). Background from a mock immunoprecipitation was subtracted. Values presented are the mean percentages relative to wild type (CY4353; TRA1TTI2) of the ratio of acetylated histone (AcH3 or AcH4) to total histone H3 for two independent experiments. PGK1 promoter primers were 2927-1 and 2927-2.
Yeast cells expressing eGFP and/or RFP fusions were grown in synthetic complete (SC) media, then diluted 1:4 into synthetic complete media with or without 8% ethanol. Growth in ethanol was for 18 hr. Prior to visualization, cells were concentrated 10-fold and 4′6-diamidino-2-phenylindole (DAPI) added to 0.02 mg/ml. Fluorescent images were obtained using a Zeiss Axioskop 2 microscope driven by ImageJ 1.41 software (National Institutes of Health) and a Scion CFW Monochrome CCD Firewire camera (Scion, Frederick MD) using DAPI, RFP, and GFP filter sets. Quantification of GFP signal intensity was performed using ImageJ software (version 1.45). The freehand selections tool was used to trace each whole cell and nucleus separately. The measure function output the signal intensity per unit area for each selection. To correct for background, the average intensity of three background selections adjacent to each cell was subtracted. The corrected nuclear intensity was then divided by the corrected whole cell intensity to give a nuclear-to-cell intensity ratio. The mean from 20 cells was calculated plus/minus a standard deviation.
We previously demonstrated that the extreme C terminus of Tra1, the FATC domain, plays a key role in the protein’s function (Hoke et al. 2010). Altering the terminal Phe of Tra1 results in temperature sensitivity and slow growth in media containing 6% ethanol, Calcofluor white, or rapamycin. As shown in Figure 1A, growth of a strain containing the tra1-F3744A allele is impaired in a variety of other stress conditions including media depleted of phosphate and containing galactose as the carbon source. The allele does not result in sensitivity to all stress, as the strain was relatively resistant to high concentrations of sodium chloride or sorbitol. To address the role of this region of Tra1, we initiated a genetic screen to identify suppressors of the tra1-F3744A allele. We plated ~2 million cells and subjected them to UV radiation at a dose that allowed 10% viability. Three colonies were isolated on the basis of their fast growth in YPD at 37°. Each of these isolated strains also grew in media containing 6% ethanol. To determine whether the suppressor mutations were linked to tra1-F3744A, the suppressor strains were crossed with a strain containing tra1-F3744A positioned downstream of a URA3 marker (CY5928). Analysis of the resulting tetrads revealed that each of the suppressors segregated as a single mutation, and one of the three (SUP3) was not closely linked to tra1-F3744A. A second independent screen was performed starting with ~10 million cells and selecting for growth at 35° on media containing 4% ethanol. Again one unlinked suppressor (SUPB) was obtained. Growth of a wild-type strain (CY4353), the tra1-F3744A strain (CY4350), and the two suppressor strains on YPD at 30° and 37°, and on YPD containing 6% ethanol is shown in Figure 1B.
We analyzed the expression of Tra1-F3744A in wild-type and SUP3-containing strain backgrounds to address whether the suppressor altered the expression of Tra1. Diploid strains with one copy of Flag5-Tra1 or Flag5-Tra1-F3744A were grown in YPD media. As shown in Figure 1C, the amount of Flag5-tagged Tra1-F3744A (lanes 3 and 6) is similar to wild-type Tra1 (lanes 2 and 5). This level is not altered by the SUP3 mutation (lanes 1 and 4), indicating that suppression is not mediated through a change in Tra1 expression. The protein shown in Figure 1C was prepared by lysis with glass beads to minimize the processing time. We have seen changes in the profiles when extracts are prepared by grinding in liquid nitrogen and subjected to a further hour-long centrifugation. Figure 1D shows a representative profile for Flag5-tagged Tra1 or Tra1-F3744A in haploid strains grown in YPD media. We observe a pronounced proteolytic product of ~300 kDa for Flag5-Tra1-F3744A (arrow, lanes 3–5), which is minimal for wild-type Flag5-Tra1 (lanes 1 and 2), suggesting that the F3744A change may alter the conformation of Tra1. This proteolytic fragment was reduced in a strain containing a plasmid copy of SUP3 (lanes 6–8).
To identify the SUP3 mutation, the suppressor-containing strain was backcrossed with the parent CY4350 (or CY4351) nine times, each time selecting for spore colonies that grew in ethanol and at 37°. After the ninth backcross, genomic DNA was isolated from three strains: an ethanol/37°-resistant spore colony and a sensitive spore colony from the same tetrad and an ethanol/37°-resistant spore colony from a distinct tetrad. Libraries were prepared and genomic sequencing was performed using the ABI SOLiD 4.0 platform at the Centre for Applied Genomics at The Hospital for Sick Children (Toronto, Ontario, Canada). The sequencing was performed in a single lane, multiplexing with 12 additional unrelated samples. Approximately 50 million reads were obtained for each sample, 60% of which aligned to the reference genome from the Saccharomyces Genome Database. Four single nucleotide differences were shared between the 37°/ethanol resistant strains and not found in the sensitive strain. Only one of these differences, a T- to-C change within codon 328 of TTI2, was found upon manual inspection of individual sequence reads. This T- to-C transition converts amino acid residue Phe328 to Ser (Figure 2A), a position highly conserved in the fungal Tti2 proteins (Figure 2B; a closely related region is not apparent in human Tti2, C8ORF41). As shown by the Western blot in Figure 2C, the F328S mutation did not alter expression of Tti2.
Two approaches were used to demonstrate that suppression of tra1-F3744A was the result of tti2-F328S. As shown in Figure 3A, SUP3 is partially dominant, with suppression being slightly less in a heterozygous diploid as compared to a haploid strain. We compared this effect with the addition of wild-type TTI2 to CY4350 (tra1-F3744A) on a centromeric plasmid (Figure 3B). As was found for the heterozygous diploid, addition of the TTI2-containing plasmid partially, but not completely, reversed the effect of tti2-F328S. Second, we analyzed eight independent spore colonies, six resistant and two sensitive, from a cross of CY5667 (tti2-F328 tra1-F3744A) and CY4350 (TTI2tra1-F3744A). The TTI2 allele from each spore colony was isolated by PCR and sequenced. For each of the eight alleles, tti2-F328S was found in all of the resistant strains and none of the sensitive strains. As the characteristics of the independently derived SUPB and SUP3 strains were highly similar, having identified tti2-F328S as SUP3, we sequenced the TTI2 allele in a derivative of the original SUPB strain that had been backcrossed vs. its parent seven times. One mutation was observed, an A-to-T at the first position of codon 336, converting isoleucine 336 to phenylalanine. The TTI2 allele from four spore colonies was sequenced; the tti2-I336F mutation segregated 2:2 with the suppressor phenotype.
Tti2 is a component of the Tra1-containing ASTRA complex and associates with Tel1 and Tor1 in part of what may be a chaperone complex with Tel2 and Tti1 (Shevchenko et al. 2008; Hurov et al. 2010; Takai et al. 2010; Helmlinger et al. 2011). To better understand the interactions of the tti2 mutations with TRA1, we examined the growth of single and double mutant strains under a variety of conditions (Figure 4; also see Figure S2). Both the tti2-F328S and I336F mutations suppressed the effects of tra1-F3744A under all of the conditions examined. Interestingly, this included reversal of the enhanced growth resulting from tra1-F3744A seen in media containing geneticin (G418), and the slow growth due to the DNA damaging agents MMS and phleomycin. Also of note the tti2-F328S allele in isolation resulted in minimal phenotypes; the tti2-F328S strain had a very slight reduction in growth in media depleted of phosphate.
To determine whether the tti2-F328S allele restores the transcriptional competency of strains containing tra1-F3744A, we performed β-galactosidase assays with a PHO5-lacZ promoter fusion (Figure 5A). The tra1-F3744A allele decreases expression of PHO5-lacZ approximately fivefold as compared to wild-type TRA1. tti2-F328S suppressed this effect, resulting in expression that was 80% of wild type. tti2-F328S in the context of otherwise wild-type TRA1 had only a marginal effect on PHO5-lacZ expression. GAL10, HIS4, and RPL35a lacZ-reporter fusions were also analyzed (Figure 5B). Similar to PHO5, tra1-F3744A decreased activated expression of GAL10-lacZ and expression of HIS4-lacZ approximately threefold. In both cases tti2-F328S restored expression to approximately wild-type levels. In contrast, the RPL35a promoter was relatively unaffected by tra1-F3744A or tti2-F328S.
We examined the effects of Tra1-F3744A on histone H4 acetylation (K8) as a ratio of the total histone H3 at the PHO5 and PGK1 promoters. Figure 6A shows the AcH4/H3 ratio for yeast strains CY4350 (tra1-F3744A TTI2) and CY5667 (tra1-F3744A tti2-F328S) as a percentage of that found for the wild-type strain CY4353 (TRA1TTI2). Since histone H4 acetylation of PHO5 is required prior to induction (Nourani et al. 2004), the chromatin immunoprecipitation was performed for cells grown in YPD media. The ratio of acetylated histone H4 to total histone H3 at PHO5 was reduced to ~40% in the tra1-F3744A strain. In comparison, the ratio was ~90% of wild type at PGK1. The ~40% decrease for PHO5 is somewhat less than the fourfold reduction seen upon deletion of the NuA4 component Eaf1 (Auger et al. 2008). Histone H4 acetylation returned to near wild-type levels in the presence of tti2-F328S. Figure 6B shows the analysis for acetylated histone H3 (K18) at PHO5 for cells grown in low phosphate media (Nourani et al. 2004). In contrast to histone H4 acetylation, Tra1-F3744A had virtually no effect on histone H3 acetylation (AcH3/total H3). Under the same conditions, deletion of Ada2, a direct regulator of Gcn5, reduced histone H3 acetylation by ~20-fold.
We examined the ability of tti2-F328S to suppress deletions within the genes of other components of the SAGA and NuA4 complexes. Similar to the effect of tra1-F3744A, deletions of ada2 or spt7 result in slow growth on media containing ethanol. However, unlike tra1-F3744A, slow growth caused by ada2Δ and spt7Δ was not suppressed by tti2-F328S (Figure 7A). Disruption of the gene encoding the NuA4 component Eaf3 results in slow growth in media containing Calcofluor white plus staurosporine (Figure 7B). Slow growth of an eaf7 disruption is observed for cells grown in 6% ethanol at 35°. Neither of these phenotypes was suppressed by tti2-F328S (Figure 7B). As a strain with a disruption of YNG2 was not available in an isogenic background, suppression by tti2-F328S was analyzed by taking advantage of its dominant nature. YCplac111-tti2-F328S was transformed into QY202 (yng2Δ; kindly supplied by Jacques Côtè) and QY202 (YNG2). As shown in Figure 7C, the plasmid copy of tti2-F328S did not suppress slow growth of the yng2Δ strain at 30°. We also analyzed the ability of tti2-F328S to suppress a second tra1 FATC domain mutation, tra1-L3733A, as well as a triple alanine scanning mutation of residues 3413–3315 (tra1-SRR3413; Mutiu et al. 2007) within the PI3K domain. As shown in Figure 7D, tti2-F328S partially suppressed the temperature sensitivity and slow growth in ethanol seen for the tra1-L3733A strain, though not as efficiently as for tra1-F3744A (compare to Figure 7A). In contrast, tti2-F328S did not suppress the slow growth of the tra1-SRR3413 strain at 37°, but rather augmented the slow growth phenotype (Figure 7E).
We expressed N-terminally eGFP-tagged Tti2 and Tti2-F328S in BY4741 to determine whether the F328S mutation altered its localization (Figure 8). The wild-type molecule was found throughout the cell with both cytoplasmic and nuclear localization. When stressed with 6% ethanol, the distribution was relatively unchanged though some foci were observed. The proximity to the vacuole suggests that these may be late endosomes. The localization of eGFP-Tti2-F328S was almost identical to the wild-type protein. In media containing 6% ethanol there was a slight reduction in vacuolar proximal foci, but this was variable.
We engineered yeast strains containing N-terminally eGFP-tagged Tra1 and Tra1-F3744A to examine their localization (Figure 9A). To avoid complications of slow growth due to the tra1-F3744A allele, the analysis was performed in heterozygous diploid strains with an untagged wild-type copy of Tra1. When grown in synthetic complete media wild-type eGFP-Tra1 was almost exclusively in the nucleus. Tra1-F3744A was found in the nucleus, but punctate fluorescence was also apparent in the cytoplasm. The amount of the cytoplasmic eGFP-Tra1-F3744A was reduced in the heterozygous tti2-F328S/TTI2 strain. In media containing 6% ethanol (Figure 9B), Tra1 was more disperse but the majority of the protein remained in foci, which we suggest are nuclear (DAPI staining was ineffective in this media). In ethanol-containing media, eGFP-Tra1-F3744A was diffusively distributed throughout the cytoplasm. Again, this appeared partially reversed in the tti2-F328S/TTI2 strain. We used imaging software to quantify the fluorescent intensity of GFP-Tra1 and GFP-Tra1-F3744A per unit area in the nucleus as compared to the whole cell (Table 3). For cells grown in ethanol, we assumed that the most pronounced focus was the nucleus. This quantification agrees with the visual conclusions drawn from the images of Figure 9 that Tra1-F3744A is more pronounced in the cytoplasm and partially relocalized to the nucleus by tti2-F328S.
To address the nature of the foci in which Tra1-F3744A is found, we visualized eGFG-Tra1-F3744A in strains containing RFP-tagged membrane constituents (Huh et al. 2003; strains kindly provided by Peter Arvidson). Figure 10A shows the analysis with RFP-tagged Anp1 (Golgi apparatus), Sec13 (ER-to-Golgi vesicles), and Nic96 (nuclear periphery) in diploid strains (eGFG-tra1-F3744A/TRA1), when the cells were grown in media containing 6% ethanol. Of these, the closest overlap was seen with RFP-Anp1; however, precise colocalization with any one type of membrane was not evident, including additional analyses with Cop1, Pex3, and Snf7 (Figure S3). For comparison in Figure 10B, we show the localization of eGFP-Tra1-F3744A with the RFP-tagged proteins in synthetic complete media (no ethanol). As shown above, eGFP-Tra1-F3744A was more evident in the nucleus than in the presence of ethanol, but cytoplasmic eGFP-Tra1-F3744A was still apparent.
If Tti2 has a role in the folding and/or stabilization of Tra1, strains with defects in the TTT complex may have reduced levels of Tra1. We introduced a tel2-15 temperature-sensitive allele (Stirling et al. 2011; Grandin et al. 2012) into a strain expressing Flag5-Tra1 and examined Tra1 levels after growth at 30°, 35°, and 37° (Figure 11A). The level of Flag5-Tra1 was unchanged by tel2-15 at 30°, but substantially reduced at the two elevated temperatures. We also examined the phenotypes of strains containing tel2-15, a temperature-sensitive tti2 allele (tti2-1, Stirling et al. 2011), or double mutations of these alleles in combination with tra1-F3744A (Figure 11B). The recessive tel2-15 and tti2-1, alleles in an otherwise wild-type background, resulted in slow growth on media containing 6% ethanol at 30°. Under all of the conditions, tel2-15 resulted in synthetic slow growth in combination with tra1-F3744A. A slight synthetic slow-growth phenotype was evident for the tti2-1 tra1-F3744A strain at 30°.
Two recent analyses of PIKK structure suggest that the FATC domain may interact with and regulate the kinase domain (Lempiäinen and Halazonetis 2009; Sturgill and Hall 2009). To perhaps provide support for such a model, we sequenced the tra1 allele 3′ of base 9730 in one of the intragenic suppressors of tra1-F3744A. A transversion of G-to-T at base 10,769, which converts arginine 3590 to isoleucine, was found. Alignments position R3590 in the putative activation loop, between β-sheet 10 and α-helix 7 corresponding to PI3K-γ (Figure 12A; Walker et al. 1999). To verify that R3590I is responsible for the suppression, the mutation was integrated into CY4398 and a haploid spore colony isolated after sporulation. The growth of this strain (CY5640; tra1-R3590I-F3744A) at 37°, and on media containing ethanol or rapamycin, confirmed that the R3590I mutation conferred suppression (Figure 12B). Similar to tti2-F328S, the F3590I mutation in isolation had no apparent phenotype.
The FATC domain is essential for the function of Tra1 and other PIKK family members (Priestley et al. 1998; Beamish et al. 2000; Takahashi et al. 2000; Sun et al. 2005; Hoke et al. 2010). For Tra1, this is apparent from the fact that a protein containing an additional C-terminal glycine residue will not support viability (Hoke et al. 2010). Altering the terminal phenylalanine of Tra1 to alanine is less severe, but results in slow growth in rich media at 30° and under conditions of stress. We have shown that alleles of TTI2 suppress tra1-F3744A. tti2-F328S restored all of the measured properties of the strains to ~80% of the wild-type level. Suppression by tti2-F328S was specific for mutations in the FATC domain of Tra1; tra1-L3733A was suppressed, whereas alleles altering the PI3K domain, or other SAGA or NuA4 components were not. The suppression by alleles of TTI2, whose product with Tel2 and Tti1 is proposed to act as a chaperone (Horejsí et al. 2010; Hurov et al. 2010; Kaizuka et al. 2010; Takai et al. 2010), the reduced levels of Tra1 in the tel2-15 strain, and the increased number of proteolytic products seen after Western blotting Tra1-F3744A lead us to suggest that the FATC domain is important for Tra1 to acquire or stabilize a fully functional conformation. The finding that the F3744A mutation increased levels of cytoplasmic Tra1 is consistent with this model, or alternatively for roles of the FATC domain and Tti2 in protein trafficking. Loss of any of these possible roles would deplete functional Tra1 and would be expected to act broadly, given the importance of independent Tra1 (Helmlinger et al. 2011), as well as the SAGA and NuA4 complexes; indeed we find numerous phenotypic consequences of tra1-F3744A.
The eGFP-Tra1-F3744A present in the cytoplasm was not uniformly distributed, but appeared in foci. Though not specific for any one membrane type, we propose that these foci represent Tra1-F3744A associated with membranes and that the altered FATC domain potentially traps these molecules on the membranes. In turn, this finding predicts that the folding of Tra1 and perhaps the formation of some of its multisubunit complexes may occur on membranes. This is appealing because the membrane would provide a platform for the process to occur, and perhaps protect the C-terminal domains from proteolysis. A requirement for membrane interactions provides a rationale for the large number of synthetic interactions observed between membrane trafficking components and either tra1-SRR3413 or deletions of NuA4 component genes (Hoke et al. 2008a; Mitchell et al. 2008). Membrane interactions are also consistent with the lipid binding properties of some of the SAGA components (Hoke et al. 2008b). In the event of reduced complex formation the molecules could be targeted to the vacuole, perhaps providing an explanation for the Pep4-dependent cleavage of Spt7 (Spedale et al. 2010). Interestingly, Han and Emr (2011) have recently shown that Cti6 and Tup1 assemble with Cyc8 on late endosomal membranes, mediated through their binding of phosphatidylinositol-3,5-diphosphate. Lipid binding is required for nuclear import of Cti6-Tup1-Cyc8, interaction with SAGA, and activation of galactose-regulated genes. Membranes are inherently sensitive to many environmental cues. As Han and Emr (2011) suggest, the membrane assembly of transcriptional complexes provides a tight link with the environmental state.
Our results in combination with the association of Tra1 and Tti2 determined by Shevchenko et al. (2008) clearly indicate a functional relationship between these molecules. The connection between Tel2, Tti2, and Tti1 demonstrated in mammalian cells suggests that the TTT complex is also functionally associated with Tra1 (Hayashi et al. 2007; Takai et al. 2007, 2010; Hurov et al. 2010; Kaizuka et al. 2010). How this relates to other components of the ASTRA complex is less clear. In that it contains Rvb1 and Rvb2, ASTRA resembles an assembly of the R2TP (Huen et al. 2009 and TTT complexes with Tra1, similar to that seen for mTOR (Horejsí et al. 2010). The potential transient nature of ASTRA and its possible role in the folding/stability of Tra1 agrees with it not yet being isolated as an intact biochemical entity. Alternatively the suppression by Tti2 may take place in the context of an independent TTT complex, with ASTRA required for additional functions.
tti2-F328S acts in a partially dominant fashion to suppress tra1-F3744A. The suppression was most notable with tti2-F328S as the sole copy of the gene, but was still apparent in the context of the wild-type allele. With the specific mechanism of the TTT complex unknown, we can only speculate on how Tti2-F328S and Tti2-I336S act. A strict gain of function is possible, but perhaps less so given the two alleles and the partial dominance. Alternatively, the two mutations may disrupt an interaction or property of Tti2 that otherwise results in its inhibition. The dominant nature of the allele also suggested that the FATC domain of Tra1 might interact closely with the region of Tti2 surrounding F328. Tra1-F3744A may be unable to interact, and suppression result from the restored interaction with Tti2-F328S. The possibility of a direct interaction was attractive, given that a hydrophobic contact for the wild-type proteins could be replaced by a hydrogen bond between the serine of Tti2-F328S and the C terminus of Tra1. However, additional experiments were inconsistent with a direct contact. First this model would not easily explain suppression of tra1-L3733A or the ability of tti2-I336F to suppress. Second, if the domain of Tti2 were to directly contact the FATC domain, one might expect negative effects on the other FATC domain containing proteins. The only discernible phenotype of tti2-F328S in isolation was a slight slow growth in media depleted of phosphate. Tti2-F328S did not lead to sensitivity to the DNA damaging agents MMS or phleomycin, suggesting a minimal effect on Mec1 and Tel1. Finally, we were unable to detect an interaction between a fragment of Tra1 containing the PI3K and FATC domains with the C-terminal half of Tti2 using recombinant proteins. We conclude that Tti2-F328S enhances the activity of Tra1-F3744A, likely by affecting folding, through a mechanism that does not restore interaction between the molecules or involve increased levels of Tti2. We note also that tti2-F328S does not suppress a mutation converting the terminal tryptophan of Mec1 to alanine (Figure S4). This suggests either that folding of the FATC domain of Mec1 does not require Tti2 function or that the change in function of Tti2-F328S is specific for Tra1.
Expression of the NuA4 (Nourani et al. 2004) and SAGA-regulated (Gregory et al. 1998) PHO5 promoter was reduced approximately fivefold by tra1-F3744A. Consistent with an effect of this mutation on NuA4 function, Tra1-F3744A reduced histone H4 acetylation of the PHO5 promoter. In contrast, and despite Gcn5 being required for activated expression, the F3744A mutation had little effect on histone H3 acetylation at the PHO5 promoter. Since the breadth of the phenotypes attributable to tra1-F3744A suggests that SAGA function is altered, we propose that the lack of change in histone H3 acetylation is due to the ability of the Ada complex (Eberharter et al. 1999), including Gcn5, Ada2, and Ngg1, to act independently of SAGA. These results with Tra1-F3744A are similar to what we observe at PHO5 upon deletion of Spt7: partially reduced acetylation, significantly decreased expression (D. Dobransky and C. J. Brandl, unpublished results). The lack of correlation between the importance of these molecules to PHO5 expression and their effect on acetylation suggests that specific targeting of PHO5 acetylation by SAGA is required for expression.
Our study is a direct demonstration of a functional link between Tra1 and Tti2. This link is likely the result of a role for Tti2, as part of the TTT complex, in the folding/maturation of Tra1 as has been found for other PIKK proteins (Takai et al. 2007, 2010; Horejsí et al. 2010; Hurov et al. 2010; Kaizuka et al. 2010; Stirling et al. 2011). In addition, our study points to a putative role for the FATC domain in the regulated folding/maturation of Tra1 in the cytoplasm. We cannot exclude that there are additional roles for the FATC domain. Recent models for the structure of the C-terminal domains of the PIKK family members predict that the helical FATC domain interacts and regulates the kinase domain (Lempiäinen and Halazonetis 2009; Sturgill and Hall 2009). The suppression of tra1-F3744A by mutation of arginine 3590 to isoleucine in the putative activation loop supports such an interaction. Interestingly, the models by Lempiäinen and Halazonetis (2009) and Sturgill and Hall (2009) place the C terminus in proximity to the active site where the FATC domain could have a role in catalysis.
We thank Megan Davey and Stephen Hoke for critically reading the manuscript; Aaron Simkovich, Lance DaSilva, Esther Rosenthal, and Brittany Rasmussen for technical assistance; Peter Arvidson for providing RFP tagged strains; Bernard Duncker, Philip Hieter, Peter Stirling, Fred Winston, and Jacques Côtè for yeast strains; Patricia Kane for helpful suggestions; Joe Mymryk for yeast plasmids; and Ivan Sadowski for sharing sequencing lanes. This work was supported by Canadian Institutes of Health Research grant MOP10845 to C.J.B. D.D. was supported by an Ontario graduate studentship. S.K. was supported by a National Science and Engineering Research Council of Canada studentship.
Communicating editor: K. Arndt