Mutations in SPT10 and SPT21 of Saccharomyces cerevisiae have been previously shown to cause two prominent mutant phenotypes: (1) defects in transcription of particular histone genes and (2) suppression of Ty and δ-insertion mutations (Spt− phenotype). The requirement for Spt10 and Spt21 for transcription of particular histone genes suggested that they may interact with two factors previously shown to be present at histone loci, SBF (Swi4 and Swi6) and MBF (Mbp1 and Swi6). Therefore, we have studied swi4Δ, mbp1Δ, and swi6Δ mutants with respect to histone gene transcription and for interactions with spt10Δ and spt21Δ. Our results suggest that MBF and SBF play only modest roles in activation of histone gene transcription. In addition, we were surprised to find that swi4Δ, mbp1Δ, and swi6Δ mutations suppress the spt21Δ Spt− phenotype, but not the spt21Δ defect in histone gene transcription. In contrast, both swi4Δ and mbp1Δ cause lethality when combined with spt10Δ. To learn more about mutations that can suppress the spt21Δ Spt− phenotype, we performed a genetic screen and identified spt21Δ suppressors in seven additional genes. Three of these spt21Δ suppressors also cause lethality when combined with spt10Δ. Analysis of one spt21Δ suppressor, reg1, led to the finding that hyperactivation of Snf1 kinase, as caused by reg1Δ, suppresses the Spt− phenotype of spt21Δ. Taken together, these genetic interactions suggest distinct roles for Spt21 and Spt10 in vivo that are sensitive to multiple perturbations in transcription networks.
THE insertion of Ty retrotransposons or Ty long terminal repeat sequences (LTRs or δ-elements) into or near the promoters of genes can cause transcriptional defects (Boeke and Sandmeyer 1991). The isolation of suppressors of these transcriptional defects has identified a large set of genes referred to as SPT genes (Winston et al. 1984; Winston and Sudarsanam 1998; Yamaguchi et al. 2001). Two functionally related genes, SPT10 and SPT21, were isolated in two different selections for spt mutants (Fassler and Winston 1988; Natsoulis et al. 1991). Studies of spt10 and spt21 mutants have shown that they share several related mutant phenotypes in addition to their Spt− phenotype: impaired transcription of particular histone genes, suppression of the transcriptional defects caused by loss of upstream activating sequences or activators at particular genes, and defects in chromatin structure (Denis 1984; Natsoulis et al. 1991; McKenzie et al. 1993; Prelich and Winston 1993; Yamashita 1993; Dollard et al. 1994; Kaplan et al. 2003). Furthermore, Spt10 and Spt21 have been shown to physically interact (Hess et al. 2004). While mutant analysis has suggested that Spt10 and Spt21 overlap in some of their functions, other results indicate that each protein also has distinct roles, as spt10 mutants grow significantly more poorly and display broader transcriptional defects than spt21 mutants (Denis and Malvar 1990; Natsoulis et al. 1994).
The regulation of histone synthesis is a critical aspect of eukaryotic cell division because histones play essential roles in all aspects of chromosome function. In Saccharomyces cerevisiae, altered histone levels have been shown to impair chromosome segregation, transcription, and other processes (Osley 1991). The regulation of histone gene transcription in S. cerevisiae is complex. Two sets of divergently transcribed gene pairs encode histones H2A and H2B and two other sets of loci encode histones H3 and H4 (Hereford et al. 1979; Smith and Murray 1983). While transcription of all four histone loci is cell-cycle regulated, with peak expression during S phase (Osley 1991), the promoters at the four loci are highly divergent and are dependent on overlapping but distinct sets of transcription factors (Osley and Lycan 1987; Xu et al. 1992; Dollard et al. 1994; Iyer et al. 2001; Simon et al. 2001; Hess et al. 2004).
Spt10 and Spt21 play important roles in the transcription of histone genes. Previous work has shown that Spt10 and Spt21 are necessary for the transcription of two of the four histone loci and that they bind to the promoters of all four histone loci (Dollard et al. 1994; Hess et al. 2004). However, neither Spt10 nor Spt21 contains recognizable DNA-binding motifs, leading to the hypothesis that a DNA-binding factor may be required for Spt10 and Spt21 recruitment to histone gene promoters. Two candidates for such a factor are MBF (MCB-binding factor) and SBF (SCB-binding factor), which are known to regulate transcription during G1/S (Breeden and Nasmyth 1987; Breeden and Mikesell 1991; Dirick et al. 1992). Whole-genome binding studies of these factors (Iyer et al. 2001; Simon et al. 2001) have suggested that they are associated with the four histone loci during G1/S. In one of these studies, SBF was suggested to be associated with the HTA1-HTB1 and HTA2-HTB2 promoters (Simon et al. 2001). In the second study, MBF and SBF were suggested to be associated with the HHT2-HHF2 promoter and MBF was associated only with HTA2-HTB2 (Iyer et al. 2001). While these studies differ in their results, taken together they suggest that MBF and SBF are involved in histone gene regulation. Thus, MBF and/or SBF are candidates to help recruit Spt10 and Spt21 to histone gene promoters.
MBF and SBF are related and partially redundant transcription activation complexes. Each is a heterodimeric complex that contains Swi6. The complexes require different DNA-binding proteins, as MBF contains Mbp1 and SBF contains Swi4 (Andrews and Herskowitz 1989; Koch et al. 1993; Moll et al. 1993). Although MBF and SBF bind to distinct sequences, each can activate transcription with reduced efficiency from the other's binding site (Dirick et al. 1992). Genetic evidence for redundancy derives from the observation that swi4 and mbp1 single mutants have only mild phenotypes, while a swi4 mbp1 double mutant is inviable (Koch et al. 1993). Surprisingly, swi6 single mutants are viable, but this has been attributed to the observation that Swi4 has a low level of activity as an activator in the absence of Swi6 (Nasmyth and Dirick 1991; Koch et al. 1993).
To address the roles of MBF and SBF in histone gene transcription and in possible interactions with Spt10 and Spt21, we have examined the levels of histone gene transcription in swi4Δ, mbp1Δ, and swi6Δ single mutants. Our results suggest that MBF and SBF serve redundant and modest roles in histone gene transcriptional activation. Surprisingly, we found that swi4Δ, mbp1Δ, and swi6Δ mutations suppress the Spt− phenotype caused by an spt21Δ mutation, although they do not suppress the defect in histone gene transcription. These results prompted us to screen for additional suppressors of the Spt− phenotype of spt21Δ. We identified suppressors in seven genes, as well as one enhancer of the Spt− phenotype. Unexpectedly, many of the spt21Δ suppressors cause lethality when combined with spt10Δ. We investigated one spt21Δ suppressor in detail, a mutation in REG1, and discovered that suppression of spt21Δ by reg1Δ occurs via hyperactivation of the Snf1 kinase. Taken together, these genetic results suggest distinct roles for Spt10 and Spt21 that functionally interact with other transcriptional networks.
MATERIALS AND METHODS
Yeast strains and genetic methods:
All S. cerevisiae strains used in this study (Table 1) are descended from a GAL2+ derivative of S288C (Winston et al. 1995). Standard methods for mating, sporulation, transformation, and tetrad analysis were used, and all media were prepared as previously described (Rose et al. 1990). The null mutations for swi6, swi4, and mbp1 (swi6Δ201::kanMX, swi4Δ201::kanMX, and mbp1Δ201::kanMX) were constructed using standard methods (Baudin et al. 1993; Lorenz et al. 1995), and spt10Δ201::HIS3 and spt21Δ201::HIS3 have been previously described (Hess et al. 2004). All null mutations are complete deletions of the open reading frame and are referred to by a Δ-designation (Giaever et al. 2002). The genomic deletions were verified by PCR analysis. The Spt− phenotype was tested using the insertion mutation lys2-128δ. In a wild-type background, lys2-128δ causes a Lys− (Spt+) phenotype and in spt mutants, lys2-128δ causes a Lys+ (Spt−) phenotype (Simchen et al. 1984). Previous results have shown that these phenotypes correlate with transcription at lys2-128δ (Clark-Adams and Winston 1987).
RNA isolation and RT-PCR:
RNA isolation and reverse transcription were performed as previously described (Hess et al. 2004). Two dilutions of cDNA were subjected to quantitative radioactive PCR and resolved on acrylamide gels as previously described for chromatin immunoprecipitation analysis (Hess et al. 2004). The (experimental primer/ACT1) ratio was set at a value of 100 for wild type. All other strains are internally normalized to ACT1 and then expressed as a percentage of the wild-type ratio.
Identification of spt21Δ suppressors:
Strain FY2199 was mutagenized by transformation with an insertional mutagen library based on the transposon Tn3::lacZ::LEU2 (Burns et al. 1994). Transformed cells were plated on SC-Leu medium at a density empirically determined to yield ∼250 colonies per plate, and plates were incubated for 2 days at 30°. The plates were replica plated to SC-Lys-Leu medium to screen for colonies that were Lys−, indicating suppression of the Spt− (Lys+) phenotype. Candidates that retested as Lys− after colony purification were then tested for linkage of the Leu+ and Lys− phenotypes by crossing each candidate to strain FY2268. For those candidates that displayed linkage, the exact position of the insertion element was identified via the Vectorette PCR-based technique and DNA sequence analysis (Riley et al. 1990). For each gene identified in this way, the complete deletion from the systematic deletion collection was then crossed to FY2268 to test for its ability to suppress (Giaever et al. 2002). Further analysis of the suppressors was carried out using these complete deletions.
Introduction of spt10Δ into the suppressor strains:
Following the crosses between the deletion set and FY2268 described above, spores that contained the following genotype were isolated: MATα SPT21+ lys2-128δ supΔ::kanMX. These strains were then crossed to the spt10Δ mutant, FY2191. Tetrads from these crosses were grown on medium lacking uracil to maintain the wild-type copy of SPT10 on a URA3-marked plasmid (pFW217) (Natsoulis et al. 1994). Before testing for mutant phenotypes, these strains were grown on medium containing 5-flouroorotic acid (5-FOA) to select for cells that no longer maintained pFW217 (Boeke et al. 1984).
MBF and SBF play modest roles in activation of histone gene transcription:
Whole-genome binding studies have suggested that the histone gene promoters are bound by MBF and/or SBF (Iyer et al. 2001; Simon et al. 2001). To examine the transcriptional dependence of the histone loci on MBF and SBF, we measured the mRNA levels for all eight histone-encoding genes in wild-type, mbp1Δ, swi4Δ, and swi6Δ strains. An spt21Δ mutant was included for comparison. Mbp1 is a component of MBF and Swi4 is a component of SBF, while Swi6 is a component of both complexes. The mRNA levels were measured by RT-PCR analysis (materials and methods).
Our results show an overlapping requirement for MBF and SBF (Table 2). First, mbp1Δ causes no significant reduction in histone gene mRNA levels, suggesting little dependence upon MBF. Second, swi4Δ causes a mild reduction in histone gene mRNA levels at all four loci, suggesting a modest requirement for SBF. To test for possible redundancy between MBF and SBF, we also tested swi6Δ. Loss of both complexes via swi6Δ reduces all eight histone gene mRNA levels, with the greatest effects at HHT1-HHF1. These data suggest that both MBF and SBF contribute to activation of histone gene transcription, with SBF playing a more prominent role. Even with the loss of both complexes in swi6Δ, the effects at HTA2-HTB2 and HHT2-HHF2 are not as severe as the defect observed in an spt21Δ mutant. However, in contrast to spt21Δ, loss of MBF and SBF causes transcriptional defects at HTA1-HTB1 and HHT1-HHF1. These results suggest that MBF and SBF act independently of Spt10 and Spt21 in transcription of histone genes.
mbpΔ, swi4Δ, and swi6Δ suppress the Spt− phenotype of the spt21Δ mutant:
To test for possible redundancy between Spt21, SBF, and MBF, we constructed three double mutants, spt21Δ mbp1Δ, spt21Δ swi4Δ, and spt21Δ swi6Δ, and tested them for an Spt− phenotype and for histone gene mRNA levels. Surprisingly, we found that swi4Δ, mbp1Δ, and swi6Δ each suppress the Spt− phenotype caused by spt21Δ, with mbp1Δ and swi4Δ being moderate suppressors and swi6Δ being a strong suppressor (Figure 1).
To determine if mbp1Δ, swi4Δ, and swi6Δ also suppress the spt21Δ histone gene transcription defect, we measured mRNA levels for HTB2 in the double mutants (Table 3). This analysis revealed that the spt21Δ defect in histone gene transcription is not suppressed by any of these mutations. These results strongly suggest that the loss of MBF or SBF suppresses a defect in spt21Δ mutants other than the defect in transcription of HTA2-HTB2.
The loss of MBF or SBF in spt10Δ mutants causes inviability:
Since spt21Δ and spt10Δ mutants share many mutant phenotypes, we set out to test the phenotypes of spt10Δ when combined with mbp1Δ, swi4Δ, and swi6Δ. To do this, we constructed double mutants that also contained wild-type SPT10 on a URA3 CEN plasmid and then tested the ability of strains to grow on medium containing 5-FOA. Our results revealed that, in contrast to spt21Δ, spt10Δ causes lethality when combined with mbp1Δ or swi4Δ (Figure 2). Diploids heterozygous for both spt10Δ and swi6Δ failed to sporulate. This striking difference between spt10Δ and spt21Δ when combined with mbp1Δ or swi4Δ is explored in the discussion.
Isolation of new suppressors of the Spt− phenotype of the spt21Δ mutant:
We have discovered two classes of Spt− suppressors of spt21Δ mutants. The first class is defined by two alleles of SPT10 that suppress spt21Δ with respect to both the Spt− phenotype and the defect in HTA2 and HTB2 transcription (Hess et al. 2004). The second class, defined by the mbp1Δ, swi4Δ, and swi6Δ mutations, suppresses the Spt− phenotype but does not suppress the defect in HTA2 and HTB2 transcription. Many members of this second class, including swi6Δ and reg1Δ (described later), are stronger suppressors of the Spt− phenotype than the SPT10 suppressor alleles described previously (Hess et al. 2004). Therefore, there is not a correlation between the strength of suppression of the spt21Δ Spt− phenotype and suppression of the spt21Δ histone gene transcription defect.
As the spt21Δ suppressors studied so far resulted from analysis of only four specific genes (SPT10, MBP1, SWI4, and SWI6), it seemed likely that suppressors of spt21Δ could be obtained in other genes, possibly revealing additional insights into Spt21 function. Therefore, we performed a screen to identify additional suppressors of spt21Δ. We used a transposon library (Burns et al. 1994) as described in materials and methods and screened ∼20,000 candidates for suppression or enhancement of the spt21Δ Spt− phenotype using the lys2-128δ insertion allele. We identified seven suppressors and one enhancer of the Spt− phenotype that are stable and segregate 2:2 in crosses, indicating that suppression in each case was caused by a single mutation.
For the seven suppressor mutations and one enhancer mutation, the mutant genes were identified (Table 4; materials and methods). For each gene identified, we tested whether the suppression phenotype was due to loss of function. To do this, we crossed strains that contain complete deletions of each gene to an spt21Δ mutant. In all but two cases, GTR1 and UBR2, the deletion allele caused an equivalent or stronger suppression phenotype than did the original insertion mutation (Table 4). We observed a range of suppression phenotypes that were scored as weak, moderate, or strong (Table 4, Figure 3). To determine if any of these mutations also suppress the spt21Δ HTB2 transcription defect, we performed RT-PCR analysis. Our results show that none of the suppressors significantly alter this phenotype (Table 5). Thus, we have identified a set of mutations that suppress the Spt− phenotype of spt21Δ but do not suppress the defect in HTA2 and HTB2 transcription, similar to mbp1Δ, swi4Δ, and swi6Δ.
The spt21Δ suppressors cause distinct interactions with spt10Δ:
Strains with either spt21Δ or spt10Δ share many phenotypes, including Spt− and a defect in histone gene transcription. To determine if some of the spt21Δ suppressors genetically interact with spt10Δ, we constructed double mutants. Our results show that the suppressors fall into distinct groups with respect to interactions with spt10Δ, and they generally differed from the interactions with spt21Δ (Table 4). Of the five spt21Δ suppressors tested, three (reg1Δ, gtr1Δ, and gup1Δ) cause synthetic lethality when combined with spt10Δ, thus behaving similarly to mbp1Δ and swi4Δ. In addition, ubr2Δ suppresses the spt10Δ growth defect but not the Spt− phenotype, and spf1Δ suppresses the Spt− phenotype of spt10Δ. These data further support the hypothesis that although Spt10 and Spt21 are both required to activate histone gene transcription, they also have distinct roles.
A bcy1 mutation enhances the Spt− phenotype of spt21Δ:
The one identified enhancer of spt21Δ was determined to be an insertion in the 3′ end of the BCY1 coding region. This mutant displayed phenotypes previously identified for bcy1 mutants, such as the inability to grow on media containing galactose or glycerol as the sole carbon source (Toda et al. 1987; data not shown). Unfortunately, our mutant displays the inability to enter stationary phase and loses viability when stored at 4° or −70°. This phenotype, observed previously for bcy1 mutants (Paz et al. 1999), prevented further analysis of this mutant.
Constitutive activation of the Snf1 protein kinase suppresses the spt21Δ Spt− phenotype:
The identification of reg1 as a strong suppressor of spt21Δ suggested an interaction of Spt21 with the Snf1 kinase. Reg1 controls Snf1 activity by controlling the Snf1 phosphorylation state, as Reg1 is the regulatory subunit of Glc7, the phosphatase that dephosphorylates Snf1. Snf1 is an important signaling molecule whose direct targets include cytoplasmic proteins, transcription factors, and histone H3 (Hardy et al. 1994; Jiang and Carlson 1997; Sanz et al. 2000; Lo et al. 2001). Snf1 is activated by phosphorylation, as this modification of Snf1 controls its association with Snf4, a Snf1 stimulatory factor (Carlson 1999; Hong et al. 2003). Therefore, in a reg1 mutant, Snf1 is hyperphosphorylated, constitutively bound to Snf4, and highly active (Jiang and Carlson 1997; Sanz et al. 2000).
The identification of reg1 as an spt21Δ suppressor suggested that Snf1 hyperactivation causes suppression of the Spt− phenotype of spt21Δ. To test this idea we first tested if Snf1 is required for suppression of spt21Δ by reg1Δ. Our results (Figure 4) demonstrate that indeed, a snf1Δ mutation abolishes reg1Δ suppression of spt21Δ. Therefore, Snf1 is required for suppression by reg1Δ. We then tested whether hyperactivation of Snf1 by a different mutation, SNF4-204 (Shirra and Arndt 1999), will suppress spt21Δ. As expected, SNF4-204 suppresses spt21Δ. These two results strongly suggest that Snf1 hyperactivation suppresses the Spt− phenotype of spt21Δ.
Previous studies of Spt21 and Spt10 suggested that they have closely related functions, one of which is the transcriptional activation of particular histone genes (Natsoulis et al. 1991, 1994; Dollard et al. 1994; Hess et al. 2004). Results in this article have provided new information about the roles of Spt21 and Spt10 in histone gene transcription and have provided strong support that Spt21 and Spt10 have distinct roles in vivo in addition to their related roles in controlling histone gene transcription. First, the finding that mbp1Δ, swi4Δ, and swi6Δ mutations have only mild effects on histone gene transcription shows that Spt21 and Spt10 are not strongly dependent upon MBF and SBF for this process. Second, the demonstration that several suppressors of the spt21Δ Spt− phenotype, including swi4Δ, mbp1Δ, and swi6Δ, cause synthetic lethality with spt10Δ suggests that some aspects of Spt21 and Spt10 functions are distinct from each other. Third, the identification of Snf1 kinase activity as important in some aspect of Spt21 function, as well as the identification of several other suppressors of spt21Δ, indicates that the defects in spt21Δ mutants are sensitive to a broad set of perturbations.
Previous analysis suggested that Spt21 is required for a subset of Spt10 functions, as spt10 mutants had been shown to have more extensive phenotypes than spt21 mutants, including a slower growth rate and other transcriptional defects (Denis 1984; Natsoulis et al. 1991; Yamashita 1993). In addition, the identification of spt10 mutations that partially suppress spt21Δ suggested that Spt21 may play a less direct role than Spt10 in their common functions, such as transcription of histone genes (Hess et al. 2004). The results in this article alter this view of Spt21 and Spt10 to one in which their functions overlap in some cases, such as histone gene transcription, but are distinct in other cases. If Spt21 were simply serving as an auxiliary factor for a subset of Spt10 functions, then the suppressors of spt21Δ would be expected to also suppress spt10Δ or be genetically neutral. Instead, in many cases the spt21Δ suppressors are synthetically lethal with spt10Δ.
The spt21Δ suppressors identified in this work suppress the spt21Δ Spt− phenotype, but do not affect the spt21Δ histone gene transcription defect, raising the question of whether suppression is related to histone levels and chromatin structure. We propose a model for suppression that addresses this issue and accounts for the wide range of suppressors identified. As previously described, spt21Δ mutants have reduced histone levels and are likely to have altered chromatin structure, resulting in suppression of insertion mutations such as lys2-128δ. However, they have only a slight growth defect on rich media. If an additional mutation, such as mbp1Δ, swi4Δ, or swi6Δ, lengthened S phase, this change might result in sufficient time for the assembly of a normal chromatin structure even with reduced histone levels, resulting in suppression of the spt21Δ Spt− phenotype. This model can explain why such a wide range of suppressors was identified, as many pathways feed into S-phase processes. Furthermore, this model can also explain why mutations that suppress spt21Δ are lethal in combination with spt10Δ. Unlike spt21Δ, spt10Δ has a severe growth defect that may activate S-phase checkpoints. Thus, mutations such as mbp1Δ, swi4Δ, or swi6Δ, which may impair checkpoint activation (Sidorova and Breeden 1997, 2002, 2003) or lengthen S phase, may be lethal when combined with the severe spt10Δ defects.
Several additional issues remain to be answered regarding Spt21 and Spt10. Prominent among them is their mechanism of recruitment to histone gene promoters, as well as the target(s) of the putative Spt10 acetyltransferase. In addition, more extensive analysis of the functional interaction between Spt21 and Snf1, a key signaling molecule, is likely to uncover interesting aspects regarding the function of both of these proteins and may identify Snf1 targets relevant to the function of Spt21.
We are grateful to Karen Arndt and Peggy Shirra for providing the SNF4-204 allele and reg1 alleles. We thank Krista Dobi, Jenny Wu, and Rolf Sternglanz for helpful comments on the manuscript. This work was supported by National Institutes of Health grant GM32967 to F.W.
↵ 1 Present address: Lewis-Sigler Institute for Integrative Genomics, Princeton University, Princeton, NJ 08544.
Communicating editor: M. D. Rose
- Received December 2, 2004.
- Accepted January 31, 2005.
- Genetics Society of America