Cell-Cycle Perturbations Suppress the Slow-Growth Defect of spt10Δ Mutants in Saccharomyces cerevisiae

Spt10 is a putative acetyltransferase of Saccharomyces cerevisiae that directly activates the transcription of histone genes. Deletion of SPT10 causes a severe slow growth phenotype, showing that Spt10 is critical for normal cell division. To gain insight into the function of Spt10, we identified mutations that impair or improve the growth of spt10 null (spt10Δ) mutants. Mutations that cause lethality in combination with spt10Δ include particular components of the SAGA complex as well as asf1Δ and hir1Δ. Partial suppressors of the spt10Δ growth defect include mutations that perturb cell-cycle progression through the G1/S transition, S phase, and G2/M. Consistent with these results, slowing of cell-cycle progression by treatment with hydroxyurea or growth on medium containing glycerol as the carbon source also partially suppresses the spt10Δ slow-growth defect. In addition, mutations that impair the Lsm1-7−Pat1 complex, which regulates decapping of polyadenylated mRNAs, also partially suppress the spt10Δ growth defect. Interestingly, suppression of the spt10Δ growth defect is not accompanied by a restoration of normal histone mRNA levels. These findings suggest that Spt10 has multiple roles during cell division.


Spt21 histones suppressors
The Saccharomyces cerevisiae Spt10 protein plays important roles in gene expression and growth. Mutations in the SPT10 gene have been identified in many different ways, including as suppressors of the transcriptional defects caused by Ty and Ty LTR insertion mutations (Fassler and Winston 1988;Natsoulis et al. 1991), suppressors of glucose repression of ADH2 (Denis and Malvar 1990), and suppressors of loss of an upstream activation sequence (Prelich and Winston 1993;Yamashita 1993). Several subsequent studies have demonstrated that Spt10 is a site-specific DNA binding protein that binds cooperatively at the regulatory regions of the four S. cerevisiae histone loci where it activates transcription (Dollard et al. 1994;Eriksson et al. 2005Eriksson et al. , 2011Hess et al. 2004;Mendiratta et al. 2006Mendiratta et al. , 2007Xu et al. 2005). DNA binding is dependent upon both a zinc finger domain and an adjacent region required for cooperative binding (Mendiratta et al. 2006(Mendiratta et al. , 2007. Spt10 also plays a negative role in histone gene transcription, as it is required for repression of several histone loci outside of S phase (Sherwood and Osley 1991). An intriguing feature of the Spt10 amino acid sequence is a conserved acetyltransferase domain (Neuwald and Landsman 1997). Although this domain is required for Spt10 function , no acetyltransferase activity or acetyltransferase substrates have yet been identified for Spt10, despite efforts by several laboratories.
The SPT21 gene is functionally related to SPT10. Mutations in SPT21 were isolated in two of the same mutant selections as mutations in SPT10 (Natsoulis et al. 1991;Prelich and Winston 1993), including one large-scale selection that identified only these two genes (Natsoulis et al. 1991). In addition, mutations in SPT21 appear to cause the same pattern of histone locus transcription defects as do mutations in SPT10 (Dollard et al. 1994;Hess et al. 2004;Sherwood and Osley 1991). In vivo, Spt21 is also recruited to all four histone loci, and this recruitment is required for the recruitment of Spt10 during S-phase . Mutations in SPT10 and SPT21 share other phenotypes, including silencing defects (Chang and Winston 2011). Mutations have been identified in SPT10 that suppress the requirement for SPT21, suggesting that Spt21 is an accessory factor, required for optimal Spt10 function .
In addition to the close functional relationships between SPT10 and SPT21, obvious differences between them suggest that they do not always function together. There are three especially striking differences between the two. First, SPT10 is transcribed throughout the cell cycle, whereas SPT21 is transcribed only during S phase, at the same time as histone genes (Cho et al. 1998;Spellman et al. 1998). Second, a complete deletion of SPT10 (spt10D) causes a severe growth defect, whereas a complete deletion of SPT21 (spt21D) causes a only a mild growth defect (Natsoulis et al. 1994). Finally, mutations that suppress an spt21D mutation do not suppress spt10D and, in fact, sometimes cause lethality when combined with spt10D (Hess and Winston 2005). Taken together, the common and distinct phenotypes of spt10D and spt21D mutants suggest that Spt10 and Spt21 function together to regulate histone gene expression and that, in addition, Spt10 plays other roles that are critical for normal growth.
To gain insight into other possible roles for Spt10, we have screened for both enhancers and suppressors of the spt10D growth defect. The identification of mutations that cause lethality when combined with spt10D suggests that Spt10 has overlapping roles with the SAGA coactivator complex. In addition, Spt10 appears to be functionally related to Asf1, the Hir complex, and the Caf-1 complex, whose functions are connected in histone gene regulation, transcriptional silencing, and chromatin assembly (Amin et al. 2012;Eriksson et al. 2012;Kaufman et al. 1998;Sutton et al. 2001). The identification of partial suppressors of the spt10D growth defect suggests that Spt10 plays important roles throughout the cell cycle. In support of the idea that these functions are independent of the role of Spt10 as an activator of histone gene transcription, suppressors of the spt10D growth defect do not reverse the defects in histone gene transcription.

MATERIALS AND METHODS
Yeast strains, media, and crosses All S. cerevisiae strains (Table 1) are GAL2 + derivatives of the S288C background (Winston et al. 1995). Capital letters denote wild-type genes, lowercase letters denote mutant alleles, and D indicates a complete open reading frame deletion. To construct spt10D haploids, the open reading frame of SPT10 was first replaced with the LEU2 gene or a kanamycin resistance marker in a diploid strain. Then, plasmid pFW217 (SPT10-URA3-CEN) was used to transform the diploid to Ura + , followed by sporulation of the diploid to obtain haploids with the spt10D mutation and pFW217. Whenever possible, spt10D strains were grown in the presence of pFW217 to minimize selection for spontaneous growth suppressors. Then, the spt10D phenotypes were tested after growth on medium with 5-fluoroorotic acid (5-FOA) to select for cells that had lost pFW217. For the nap1D::kanMX, hsl1D:: kanMX, mih1D::kanMX, swe1D::kanMX, and pat1D::kanMX alleles, a 2.4-kb cassette was amplified by polymerase chain reaction (PCR) from genomic DNA isolated from the corresponding deletion set strain (Giaever et al. 2002), then used to transform a wild-type strain. The cassette contains a replacement of the entire open reading frame with a kanamycin resistance marker. The cln3D::HIS3, lsm1D::natMX, and bck2D::hphMX alleles were generated by PCR-mediated disruption of the entire open reading frame (Goldstein and McCusker 1999). All deletions were confirmed by PCR. The cdc28-T18A Y19F allele was generated by digesting p433 (a generous gift from A. Amon) with EcoRI and using the fragment containing the cdc28-T18A Y19F allele and the URA3 marker to transform a wild-type strain. The URA3 gene was then replaced with the KanMX drug resistance cassette of pRS400. Media, basic yeast techniques, mating, sporulation, and tetrad dissection were as previously described (Rose et al. 1990). Crosses to test double mutant lethality generally contained one parent with an spt10D mutation and also carrying plasmid pFW217 (SPT10-URA3-CEN). Double-mutant lethality was assayed by replica plating the spore col-onies to 5-FOA plates to determine whether strains that had lost pFW217 were viable.

Transposon mutagenesis screen
The transposon mutagenesis screen was performed as described (Burns et al. 1994). In summary, the LEU2-marked library DNA was digested with NotI, then used to transform strain FY2191. Transformant colonies were selected on SC-Leu-Ura medium then replica plated to 5-FOA medium to select for cells that had lost pFW217 (SPT10-URA3), leaving colonies containing the library insertion in an spt10D genetic background. Colonies that failed to grow were designated synthetic lethal candidates, and colonies growing more quickly than FY2191 were designated growth suppressor candidates. All candidates were purified to single colonies, which were then individually patched on SC-Leu medium followed by replica plating to verify the growth phenotype. All candidates remaining after this rescreening were purified and tested a third time. Each candidate was then crossed to an spt10D leu2 strain to test whether the mutant phenotype cosegregated with the LEU2 marker on the transposon. For the confirmed mutants, genomic DNA was isolated, and vectorette PCR was used to identify the location of each transposon insertion (Arnold and Hodgson 1991). As one growth suppressor candidate was tightly linked to the SPT10 locus, instead of vectorette PCR, we used a candidate gene approach and by a combination of PCR and sequencing, demonstrated the insertion to be within LSM1.
Synthetic genetic array (SGA) screen A collection of yeast strains containing deletions of every nonessential gene was screened for phenotypes in an spt10D background using an SGA screen (Tong et al. 2001). The collection was spotted onto YPD plates with deletion set strains hoD::KanMX, lys2D::KanMX, and lys12D::KanMX spotted separately at the top and bottom of each plate as controls that do not affect spt10D growth. The array was mated by replica plating to a lawn with an spt10D strain (FY2923) containing a can1::STE2pr-HIS3 allele and carrying the pFW217 (SPT10-URA3) plasmid. Diploids were selected on SC-Leu-Ura and sporulated on solid 1% potassium acetate medium supplemented with histidine, uracil, leucine, and lysine. MATa haploids that contain the deletion set mutation, spt10D, and the SPT10 plasmid were selected by replica plating onto SC-Arg-His-Leu-Ura+canavanine+G418 medium. The cells were then replica plated to SC + 5-FOA medium to leave the mutant spt10D as the only SPT10 allele present. Strains with better or worse growth compared with the control strains were identified and retested, and then tetrads were dissected to assay for 2:2 segregation and cosegregation of the suppression phenotype with the kanamycin resistance marker.

Dilution spot tests
For dilution spot tests, unless noted otherwise, strains harboring the pFW217 (SPT10-URA3-CEN) plasmid were single colony purified on 5-FOA medium to select for plasmid loss, and single colonies were then patched to YPD media. After 2 d, the cells were resuspended in water to a density of 4 · 10 6 cells/mL ( Figure 2) or 1 · 10 7 cells/mL (Figures 1, 326). Fivefold serial dilutions were spotted onto the media indicated. Plates were scanned after 223 d at 30°, unless otherwise indicated.

RESULTS
Identification of mutations that enhance or suppress the spt10D slow-growth phenotype To study the basis of the spt10D slow growth phenotype, we screened for mutations that enhance or suppress the growth defect by using both transposon insertion mutagenesis (Burns et al. 1994) and the S. cerevisiae deletion set (Giaever et al. 2002), both as described in Materials and Methods. As spontaneous suppressors of the spt10D slow growth phenotype arise at a high frequency, we maintained a low-copy SPT10 plasmid (pFW217) in the spt10D strains until the final screening step for each method. We began with a transposon insertion mutagenesis screen (Burns et al. 1994;Kumar and Snyder 2002) in which we tested 9000 independent transformants for improved or impaired growth compared with the spt10D parent (Materials and Methods). By this approach, we identified eight mutations in a total of six genes (Table 3). Three mutations that confer suppression of spt10D poor growth were in two genes and five mutations that cause lethality when combined with spt10D were identified in four genes. For all six genes, we tested a complete deletion of the identified gene and found the same suppression phenotype, suggesting that all of the insertion mutations cause null phenotypes. For all subsequent experiments, the deletion mutations were used.
From this initial screen, a concern of bias arose, as we had obtained two different transposon insertions within ASF1 without obtaining any insertions in other genes whose deletions were previously shown to be lethal in combination with spt10D. These genes include HTA1, HTB1, HHF1, HIR1, ASF1, RKR1, and MBP1 (Braun et al. 2007;Fassler and Winston 1988;Hess 2004;Hess and Winston 2005;Sutton et al. 2001). Therefore, rather than saturate the transposon mutagenesis screen, which would require testing 30,000 transformants (Burns et al. 1994), we switched to the more systematic approach of screening the deletion set.
We screened the deletion set for mutations that either suppress or enhance the spt10D slow growth defect (Materials and Methods). Our screen yielded 44 mutations that cause lethality in combination with spt10D (Table 4) and 13 mutations that improve spt10D growth (Table  5). Interestingly, there was no overlap with the mutations identified from the transposon mutagenesis screen, although some functionally related genes were identified (LSM genes). The lack of overlap indicates that the deletion set screen had many false-negative results. There was also a class of 12 mutants that appeared to cause lethality during the original screen but showed little or no growth defect upon
The loss of specific classes of SAGA genes is lethal in combination with spt10D Our screens identified four genes encoding components of the SAGA coactivator complex whose deletion is lethal when combined with spt10D: SPT3, SPT8, SGF11, and SGF29. These four factors are believed to be involved in distinct activities of the multifunctional SAGA complex, as Spt3 and Spt8 modulate the recruitment of the TATAbinding protein (TBP) to promoters Green 2001, 2002;Dudley et al. 1999;Larschan and Winston 2001), Sgf11 is part of the DUB module of SAGA (Kohler et al. 2010;Samara et al. 2010), and Sgf29 has recently been shown to bind to H3K4me2/3, to be required for Gcn5-dependent histone acetylation in vivo, and to help recruit TBP to promoters (Bian et al. 2011;Shukla et al. 2012). To test whether the double-mutant lethality with spt10D is general for all SAGA deletion mutants or specific for certain classes, we tested deletions of SPT20, encoding a core component of SAGA, UBP8, encoding a histone deubiquitylase, and GCN5, encoding the histone acetyltransferase. Our results (Figure 1) show that the spt20D spt10D double mutant is inviable, whereas both the ubp8D spt10D and gcn5D spt10D double mutants are viable but grow poorly, even worse than the spt10D single mutant. Our genetic analysis, then, demonstrates that Spt10 shares essential or important roles with distinct functions of the SAGA coactivator complex. In light of the spt10D-gcn5D genetic interaction, we note that we did not see a genetic interaction between spt10D and rtt109D (RTT109 encodes a histone acetyltransferase that has been implicated in histone gene transcription) (Fillingham et al. 2009).
Double-mutant lethality of spt10D with asf1D and hir/hpc2D mutations suggests functional overlaps Among the genes identified as causing double-mutant lethality with spt10D were asf1D, hir2D, hir3D, and hpc2D. Previous studies also showed that spt10D asf1D double mutants are inviable (Sutton et al. 2001). Asf1 has been shown to be a histone chaperone (Munakata et al. 2000), the Hir complex (comprised of Hir1-3 and Hpc2) has been implicated in chaperone and nucleosome assembly activities (Green et al. 2005;Prochasson et al. 2005), and both Asf1 and the Hir complex have been shown to regulate histone gene transcription (Osley and Lycan 1987;Sutton et al. 2001;Xu et al. 1992). Furthermore, these factors are believed to function both physically and genetically with each other and with the Caf-1 complex (Green et al. 2005;Kaufman et al. 1998;Liu et al. 2012;Sutton et al. 2001). The isolation of asf1D and hir/hpc2D mutations as causing lethality when combined with spt10D suggests that Spt10 participates in this set of functions. To test this further, we crossed spt10D by hir1D and by cac1/rlf2D (CAC1 encodes a component of the Caf-1 complex) to test for double mutant lethality. Our results (Table 6) show that spt10D causes inviability with asf1D and hir/hpc mutations, but not with cac1D. This pattern is reminiscent of earlier studies that showed that both asf1D and hir/hpc mutations cause double-mutant sickness with cac mutations, but not with each other (Kaufman et al. 1998;Sutton et al. 2001). We note that our screens did not identify mutations in RTT106, which encodes a histone chaperone that has been shown to regulate histone gene transcription by interactions with Asf1/Hir/Caf-1 (Fillingham et al. 2009;Huang et al. 2007;Kurat et al. 2011;Silva et al. 2012;Zunder and Rine 2012). Similarly, a screen for mutations that cause double-mutant lethality with rtt106D did not identify spt10D (Imbeault et al. 2008). In contrast to spt10D, an spt21D mutation allowed viability when combined with hir1D or asf1D (Table 6). Taken together, our results suggest that Spt10, but not Spt21, contributes to an essential function in collaboration with Asf1 and the Hir complex, likely either in histone gene activation or an aspect of chromatin assembly.

Genes involved in silencing show mutant phenotypes in combination with spt10D
One notable class of mutants appeared to show lethality in combination with spt10D during our systematic screen. However, upon retesting by tetrad dissection, viable double mutant spores were obtained at the expected frequency, without substantial growth defects. This class of mutants included sir1D, ard1D, and pol32D, all of which have roles in silencing (Pillus and Rine 1989;van Welsem et al. 2008;Whiteway et al. 1987). Others have reported a similar pattern of apparent lethality for sir1D dot1D and pol32D dot1D in another deletion set screen (van Welsem et al. 2008). They discovered that the pattern actually resulted from mating type silencing defects, which prevent growth when the SGA screening method is used. Our studies of Spt10 have demonstrated it to be required for silencing (Chang and Winston 2011).
The slow growth of spt10D mutants can be suppressed through multiple genetic pathways The mutations that we identified that suppress the spt10D growth defect fall into several functional categories. For the remainder of our analysis, we focused on the four mutations that individually caused the strongest suppression of the spt10D growth defect: hsl7D, nap1D, bck2D, and lsm1D ( Figure 2). Hsl7 is an arginine methyltransferase with a role in the bud morphogenesis checkpoint (Lew 2000). Nap1 is a histone chaperone involved in the nuclear import of histones, and it regulates cell-cycle progression in G2/M (Zlatanova et al. 2007). Bck2 regulates the transition from G1 to S phase of the cell cycle (Epstein and Cross 1994;Lee et al. 1993), and Lsm1 is part of a heteroheptameric complex involved in RNA decapping and n processing (Tharun 2009). Lsm1 has recently been shown to control histone mRNA stability (Herrero and Moreno 2011). All of the deletion mutations are partial suppressors individually, but when lsm1D is combined with hsl7D or bck2D, strong additive effects are seen (Figure 2). Little or no additivity is seen with other combinations. This finding suggests that hsl7D and bck2D suppress the spt10D growth defect through a different genetic pathway than does lsm1D.
To study these effects, we conducted a more detailed genetic analysis of each suppressor.
Perturbations of the G2/M transition allow spt10D mutants to grow faster HSL7, along with HSL1, initially was isolated in a histone synthetic lethal screen, which identified genes that become essential when the tail of either histone H3 or histone H4 is deleted (Ma et al. 1996).
Although the basis of this synthetic lethality remains unknown, Hsl1, a protein kinase, and Hsl7 have been shown to regulate the bud morphogenesis checkpoint through the Hsl2Swe12Cdc28 pathway, which monitors whether cytoskeletal events have been properly completed prior to mitosis ( Figure 3A) (Lew 2000). The cyclin-dependent kinase Cdc28 controls cell-cycle progression through the G2/M transition; its activity is inhibited by the kinase Swe1 and activated by the phosphatase Mih1. When an S. cerevisiae cell buds, Hsl1 recruits Hsl7 to the bud neck and phosphorylates both proteins. This recruits Swe1, leading to Swe1 degradation, causing decreased phosphorylation of Cdc28 and thereby promoting progression through G2/M. Thus, an n  Table 5 Genes found by SGA analysis whose deletion suppresses the spt10D poor growth phenotype

BCK2
Protein kinase C signaling pathway and the G1/S transition CLB2 B-type cyclin involved in G2 to M progression HAL5 Putative protein kinase HDA2 Component of a class II histone deacetylase complex IES3 Component of the INO80 complex ITR1 Myo-inositol transporter LAS21 Synthesis of the glycosylphosphatidylinositol (GPI) core structure LSM6 Part of complexes involved in RNA processing, splicing, and decay LSM7 Part of complexes involved in RNA processing, splicing, and decay NAP1 Bud morphogenesis, microtubule dynamics, and transport of histones H2A and H2B SIF2 Component of the Set3C complex SLM4 Component of the EGO complex SYH1 Protein of unknown function, influences nuclear pore distribution Figure 1 Mutations in genes encoding SAGA subunits lead to lethality or poor growth in an spt10D background. Shown are fivefold dilution spot tests. All strains were grown to saturation in SC-Ura medium in the presence of the pFW217 SPT10-URA3-CEN plasmid. They were serially diluted fivefold and spotted onto SC-Ura and 5-FOA plates to select for cells that have maintained or lost the SPT10 plasmid, respectively. The SC-Ura plate is shown after 2 d of incubation at 30°a nd the 5-FOA plate after 5 d. Upper and lower panels are from the same plate. The strains were wild type (FY2200), spt10D (FY2924), spt8D spt10D (FY2925) spt20D spt10D (FY2926), gcn5D spt10D (FY2927), and ubp8D spt10D (FY2928).
hsl7D single mutant has increased Swe1 activity, resulting in decreased Cdc28 activity. We tested the effects of other mutations in the Hsl2Swe12Cdc28 pathway on spt10D growth. Consistent with our findings for hsl7D, both hsl1D and mih1D, which also impair progression through the bud morphogenesis checkpoint, suppress the spt10D growth defect, whereas a mutation (swe1D) that promotes progression does not ( Figure 3B). As additional evidence that impairment of G2/M progression suppresses the spt10D growth defect, we identified clb2D as a suppressor in our screen (Table 5).
To test whether suppression of the spt10D growth defect by hsl7D occurs within the Hsl2Swe12Cdc28 pathway, we tested combinations of mutations in this pathway. First, we found that swe1D is epistatic to hsl7D with respect to suppression of the spt10D growth defect ( Figure 3B), suggesting that suppression by hsl7D is mediated through Swe1 activity. Second, we tested whether the inhibitory phosphorylation of Cdc28 by Swe1 plays a role in hsl7D suppression of the spt10D growth defect. To do this, we used the cdc28-T18A Y19F allele (Amon et al. 1992;Sorger and Murray 1992), which makes cells insensitive to mutations upstream in the Hsl-Swe1-Cdc28 pathway, thus mimicking loss of Swe1. We found that hsl7D no longer suppresses the spt10D growth defect in the presence of the cdc28-T18A Y19F allele ( Figure 3C), further supporting that hsl1D-and hsl7D-mediated suppression occurs through the Hsl2Swe12Cdc28 pathway. Taken together, our genetic analysis suggests that mutations that activate the bud morphogenesis checkpoint can confer improved growth of spt10D cells.
Perturbations at the G1/S transition also suppress the spt10D growth defect Bck2 was originally isolated as a factor important in protein kinase C signaling, and it has been found to be important in controlling the G1/ S transition of the cell cycle (Epstein and Cross 1994;Lee et al. 1993). A related protein involved in regulating the G1/S transition is Cln3, a cyclin that binds to Cdc28 to regulate the transition through START (Richardson et al. 1989). We asked whether a cln3D mutation can also suppress the spt10D growth defect. Spot tests demonstrate that cln3D spt10D mutants grow better than spt10D single mutants (Figure 4), suggesting that different perturbations in the G1/S transition can suppress the spt10D growth defect. Taken together with the hsl7D suppression data, our genetic analysis demonstrates that the spt10D slow growth can be suppressed by mutations that delay cell cycle progression at either the G1/S transition or the bud morphogenesis G2/M checkpoint.
Impairment of the Lsm1-72Pat1 complex suppresses the spt10D slow growth phenotype Next we conducted a more detailed genetic analysis of three closely related suppressors: lsm1D, lsm6D, and lsm7D. The eight S. cerevisiae LSM (like Sm) genes form two distinct, ring-shaped, heteroeptameric complexes (Tharun 2009). The first complex, containing Lsm2-8, localizes to the nucleus and regulates pre-mRNA splicing. The second complex, containing Lsm1-7, is localized to the cytoplasm and regulates the decapping of polyadenylated mRNAs, in conjunction with Pat1 (protein associated with Topoisomerase II). We note that in both larger eukaryotes (Tharun 2009) and in yeast (Herrero and Moreno 2011), the Lsm1-72Pat1 complex has been implicated in promoting the degradation of histone mRNAs.
The result that lsm1D suppresses the spt10D slow growth phenotype suggests that it is the Lsm1-72Pat1 complex, rather than the Lsm22Lsm8 complex that is related to spt10D growth. We therefore also tested whether pat1D suppresses the spt10D growth phenotype. Our results ( Figure 5) show that pat1D does suppress the spt10D n Table 6 spt10D is inviable with hir1D and asf1D

Double Mutant
Phenotype a spt10D hir1D Inviable b spt10D asf1D Inviable c spt10D cac1D Viable d spt21D hir1D Viable e spt21D asf1D Viable f spt21D cac1D Viable g a The phenotype was determined by testing the ability of the double mutant to survive loss of plasmid pFW217 (SPT10-URA3-CEN) by assaying growth on 5FOA plates as described in Materials and Methods. The cross done for each combination is listed below.
Environmental conditions that slow cell division also suppress the spt10D slow growth phenotype Considering that genetic means of slowing cell-cycle progression can suppress the spt10D slow growth phenotype, we asked whether altered growth conditions that slow cell cycle progression will also suppress this phenotype. First, we assayed the growth of spt10D strains on medium containing 25 mM hydroxyurea (HU), a ribonucleotide reductase inhibitor that impedes S-phase progression. We found that addition of 25 mM HU causes modest suppression of the spt10D growth defect relative to wild-type growth ( Figure 6A).
Second, we slowed growth using medium that contains glycerol rather than glucose as a carbon source. Relative to wild-type, spt10D growth modestly improves on this medium ( Figure 6B). These findings are consistent with the possibility that slowing cell cycle progression through multiple means improves spt10D growth.
Suppressors of the spt10D growth phenotype do not restore histone mRNA levels Because Spt10 binds to histone gene promoters and regulates histone gene transcription (Dollard et al. 1994;Eriksson et al. 2011;Hess et al. 2004;Sherwood and Osley 1991;Xu et al. 2005), we wanted to test whether the suppressors improve spt10D growth by increasing histone gene mRNA levels. We therefore measured mRNA levels for all eight histone genes in the suppressor strains, using reverse transcription and real-time PCR. We used primer pairs highly specific for their corresponding transcripts (Table 2; N. McLaughlin and D. Clark, personal communication) to distinguish the two nearly identical copies of each histone gene.

Figure 4
A mutation perturbing the G1/S transition can partially suppress the spt10D growth defect. Fivefold dilution spot assays were performed as in Figure 3. Strains were wild type (FY2200), cln3D (FY2962), spt10D (FY2924), and cln3D spt10D (FY2963). Pictures were taken after 2 d. background. The increased level of histone mRNAs observed for lsm1D agrees with previous results (Herrero and Moreno 2011). Overall, our results suggest that restoration of normal histone mRNA levels is not necessary for suppression of the spt10D slow growth phenotype.
We note that, like spt10Δ mutants, spt21Δ mutants show decreased levels of HTA2, HTB2, and HHF2 mRNA, but unlike spt10Δ mutants or the suppressor strains, the spt21Δ mutants show modest increases in mRNA levels for HTA1, HTB1, HHF1, and to a lesser degree HHT1. These results suggest that Spt10 and Spt21 have some nonoverlapping roles in histone gene regulation.

DISCUSSION
In this work, we have identified a broad spectrum of mutations that either cause lethality when combined with spt10D or that suppress the slow growth phenotype caused by spt10D. The first set of genes suggests that the function of Spt10 partially overlaps with the SAGA coactivator complex as well as with two factors involved in chromatin assembly and histone gene transcription, Asf1 and the Hir complex. Given the pleiotropic nature of mutants lacking these functions, as well as the documented role of Asf1 and the Hir complex in histone gene regulation (Osley and Lycan 1987;Sutton et al. 2001;Xu et al. 1992), these double mutant lethalities are not surprising. Several additional genes were identified in the screen for double-mutant lethality (Tables 3 and 4), and the results suggest that functional overlaps also exist between Spt10 and both the Elongator complex and the Ino80 complex. As there are no known roles for SAGA, Elongator, or Ino80 in histone gene expression, further studies of these interactions will be required to understand whether the essential process in which Spt10 and these other factors participate involves histone gene expression or a previously uncharacterized role for Spt10.
The suppressors of the spt10D growth defect led us to conclude that perturbations at multiple points of the cell cycle can suppress the slow growth of spt10D mutants. Although it seems paradoxical that an impairment of cell-cycle progression would enhance growth, there is precedent for a defect in one process suppressing a defect in a related process. For example, a cold-sensitive spt5 mutation is suppressed with 6-azauracil, which decreases the rate of transcription elongation (Hartzog et al. 1998). Furthermore, perturbations in multiple different cell cycle phases can suppress a silencing defect at the S. cerevisiae silent mating type loci and telomeres (Laman et al. 1995).
One model to explain our findings is that spt10D mutants grow slowly due to the shortage of a factor or factors necessary for normal Figure 6 Nongenetic means of suppressing the spt10D slow growth phenotype. (A) Fivefold dilutions were made as in Figure 3, then spotted onto YPD medium or YPD + 25 mM HU. Pictures were taken after 2 d. Strains were WT (FY2200), spt10D (FY2924), and mec1D sml1D (FY2967). mec1D sml1D mutants are hypersensitive to HU. (B) Wildtype (FY2200) and spt10Δ (FY2924) strains were subjected to fivefold serial dilutions as in Figure 3 and grown on YPD medium for two days or on YP + 3% glycerol medium for 5 d. Figure 7 mRNA abundance for the core histone genes in growth suppressor strains. RNA was isolated and reverse transcribed, and real-time PCR with gene-specific primers (Table 2) was used to quantitate histone mRNA levels for (A) HTA1 and HTA2; (B) HTB1 and HTB2; (C) HHT1 and HHT2; and (D) HHF1 and HHF2. All values were normalized to ACT1 mRNA levels and are shown relative to wild type, which was assigned a value of 1. Shown is the mean 6 SEM for at least three independent experiments. Strains were wild type (FY2200 and FY1856), spt10D (FY2924 and FY2938), spt21D (FY2816 and FY2817), hsl7D spt10D (FY2930 and FY2950), nap1D spt10D (FY2931 and FY2968), bck2D spt10D (FY2932 and FY2969), lsm1D spt10D (FY2933 and FY2970), hsl7D lsm1D spt10D (FY2941 and FY2971), bck2D lsm1D spt10D (FY2944), hsl7D nap1D bck2D lsm1D spt10D (FY2949 and FY2972), hsl7D (FY2934 and FY1924), nap1D (FY2935, FY2973), bck2D (FY2936, FY2974), and lsm1D (FY2937, FY2975). growth, and that cell cycle perturbations compensate for this growthlimitation, either by allowing more time for the factor to be produced, or by adjusting the relative levels of factors with which it interacts. Considering the well-characterized role of Spt10 in activating histone gene transcription, obvious candidates for such factors are histone proteins. We note that histone levels are clearly a factor in spt10D growth, as a plasmid that encodes all four core histones (with the HTA1-HTB1 and HHT1-HHF1 loci) restores spt10D growth to nearly wild-type levels (Eriksson et al. 2005;Silva et al. 2012). However, we found that suppressors of the spt10D growth defect do not suppress the spt10D defect in histone mRNA levels, suggesting that the slow growth can be affected by other routes, possibly independent of histone gene transcription. Alternatively, the suppressors might partially alleviate the requirement for normal histone levels.
Left unresolved by these and other studies of Spt10 is the role of the Spt10 acetyltransferase domain. While it is required for Spt10 function , its target(s) remain unknown. The elucidation of these targets will go a long ways toward helping us understand the roles of Spt10 in growth.