Spt16/Cdc68, Pob3, and Nhp6 collaborate in vitro and in vivo as the yeast factor SPN, which is homologous to human FACT. SPN/FACT complexes mediate passage of polymerases through nucleosomes and are important for both transcription and replication. An spt16 mutation was found to be intolerable when combined with a mutation in any member of the set of functionally related genes HIR1, HIR2/SPT1, HIR3/HPC1, or HPC2. Mutations in POB3, but not in NHP6A/B, also display strong synthetic defects with hir/hpc mutations. A screen for other mutations that cause dependence on HIR/HPC genes revealed genes encoding members of the Paf1 complex, which also promotes transcriptional elongation. The Hir/Hpc proteins affect the expression of histone genes and also promote normal deposition of nucleosomes; either role could explain an interaction with elongation factors. We show that both spt16 and pob3 mutants respond to changes in histone gene numbers, but in opposite ways, suggesting that Spt16 and Pob3 each interact with histones but perhaps with different subsets of these proteins. Supporting this, spt16 and pob3 mutants also display different sensitivities to mutations in the N-terminal tails of histones H3 and H4 and to mutations in enzymes that modulate acetylation of these tails. Our results support a model in which SPN/FACT has two functions: it disrupts nucleosomes to allow polymerases to access DNA, and it reassembles the nucleosomes afterward. Mutations that impair the reassembly activity cause chromatin to accumulate in an abnormally disrupted state, imposing a requirement for a nucleosome reassembly function that we propose is provided by Hir/Hpc proteins.
THE Saccharomyces cerevisiae proteins Spt16/Cdc68, Pob3, and Nhp6 function together as a complex that has been called CP or SPN (Brewsteret al. 2001; Formosaet al. 2001) and that we now call yFACT. Nhp6 is an HMG1-motif DNA-binding protein that binds either to DNA or to nucleosomes (Yenet al. 1998; Formosaet al. 2001). Spt16 and Pob3 form a stable heterodimer that does not bind to DNA, to nucleosomes, or to Nhp6:DNA complexes, but does bind to Nhp6: nucleosome complexes (Formosaet al. 2001). These “SPN:nucleosome” complexes display increased accessibility of the DNA within the nucleosome to DNase I (Formosaet al. 2001). SPN/yFACT therefore binds to and alters the structure of the fundamental subunit of chromatin.
Spt16 and Pob3 are both essential for viability, but Nhp6 is nonessential (Costiganet al. 1994). Genetic evidence indicates that Nhp6 supports the function of Spt16-Pob3 in vivo and it supports the binding of Spt16-Pob3 to nucleosomes in vitro (Brewsteret al. 2001; Formosaet al. 2001). However, Spt16-Pob3 must be able to execute its essential function without Nhp6. Human and frog cells contain a factor similar to SPN called FACT or DUF, respectively, although the primary sequence elements found in the yeast Pob3 and Nhp6 proteins are fused to form the single polypeptides SSRP1 and DUF87 in these metazoan complexes (Okuharaet al. 1999; Orphanideset al. 1999). An HMG1 motif is therefore covalently associated with FACT and DUF, but yeast Nhp6 interacts with Spt16-Pob3 only weakly outside of the context of SPN:nucleosome complexes (Brewsteret al. 2001; Formosaet al. 2001). SPN, FACT, and DUF therefore represent members of a broadly conserved family of factors, but SPN has a somewhat different subunit architecture.
Nucleosomes normally block the progression of RNA polymerase II, but FACT allows Pol II to elongate through these structures in vitro (Orphanideset al. 1998). Frog oocyte extracts depleted of DUF fail to support DNA replication (Okuharaet al. 1999). These results suggest that FACT/DUF promotes progression of both RNA and DNA polymerases along their natural chromatin templates. Genetic analysis in yeast supports this conclusion in vivo, because mutations that alter yFACT/SPN components cause defects in both the regulation of transcription and the performance of DNA replication (Maloneet al. 1991; Rowleyet al. 1991; Johnet al. 2000; Schlesinger and Formosa 2000; Yuet al. 2000; Brewsteret al. 2001; Formosaet al. 2001). Although SPN and FACT/DUF are organized somewhat differently, the overall similarities in sequence and activity suggest that these factors all use a common mechanism to facilitate both transcription and replication. We therefore refer here to SPN as yFACT with the adjusted acronym of “facilitates chromatin transactions.” This emphasizes the similarities among these factors and incorporates the role they play in both transcription and replication.
The current data suggest that the FACT family functions by making chromatin less inhibitory to the progression of polymerases, perhaps by destabilizing or even disassembling nucleosomes (Orphanideset al. 1999). Nucleosomes are composed of ∼146 bp of DNA wrapped around a histone octamer composed of a core (H3-H4)2 tetramer and two H2A-H2B dimers (van Holde 1988). Human FACT binds to free H2A-H2B dimers, suggesting that it might stabilize disrupted nucleosomes (Orphanideset al. 1999). One or both H2A-H2B dimers can be removed from a nucleosome (van Holde 1988), resulting in a structure that is less inhibitory to the passage of RNA polymerase II (Kireevaet al. 2002). One specific proposal for the mechanism of FACT therefore is that it promotes the disassembly of nucleosomes into incomplete, less restrictive forms (Orphanideset al. 1999; Kireevaet al. 2002).
Nucleosomes are deposited onto DNA chiefly during DNA replication, and the synthesis of histone proteins is tightly regulated to coincide with this period of increased demand. Histone gene transcription is therefore repressed when cells are not in S phase and is also blocked by elevated levels of histone proteins (Osley 1991). Both the cell cycle regulation and feedback controls fail if any of the histone regulatory or histone promoter control (HIR1, HIR2/SPT1, HIR3/HPC1, or HPC2) genes are defective (Osley 1991; Xuet al. 1992). Mutations in these “HIR/HPC” genes therefore lead to increased levels of histone gene transcripts during periods when few nucleosomes are being assembled.
Hir1 and Hir2 proteins are similar to one another and to a family of HIRA proteins that have been identified in several eukaryotes (see references cited in Lorainet al. 1998). These proteins bind to histones (Lorainet al. 1998; Suttonet al. 2001), and the frog HIRA has been shown to act in a replication-independent nucleosome deposition pathway (Ray-Galletet al. 2002). Consistent with a role in chromatin assembly in yeast, the Hir/Hpc proteins display functional overlap and some sequence similarity with the chromatin assembly factor known as CAF-I (Kaufmanet al. 1998). CAF-I mediates the replication-dependent deposition of nucleosomes in vitro and is important for maintaining normal chromatin structure in vivo (Smith and Stillman 1989; Kaufmanet al. 1997). Mild phenotypes result from deletion of individual genes that encode either Hir/Hpc proteins or CAF-I subunits, but combining deletions of genes from both sets causes a strong growth defect, a severe decrease in heterochromatin-mediated gene silencing, and chromosomal instability due to defective kinetochore formation (Kaufmanet al. 1998; Sharpet al. 2002). These data suggest that Hir/Hpc and CAF-I proteins each promote chromatin assembly through independent but partially redundant pathways.
To gain further insight into the function of yFACT, we conducted an unbiased screen for mutations that cause increased reliance on the function of Spt16. Here we report that mutations in any of the four HIR/HPC genes cause poor growth or lethality when combined with mutations in SPT16 or POB3. Synthetic phenotypes such as these have proven valuable in discovering common functions for genes (Guarente 1993) and can reflect any of several relationships between the gene products involved. The evidence presented here supports an interpretation that extends our understanding of both FACT and the Hir/Hpc pathway. We propose that FACT not only disrupts nucleosomes but also is responsible for their reassembly after a polymerase has passed. We further propose that the Hir/Hpc proteins constitute a general nucleosome reassembly-repair pathway that becomes essential when yFACT fails to perform its own reassembly function efficiently. Consistent with this model, a screen for mutations that are intolerable in the absence of HIR1 revealed mutations in the Paf1 complex, another factor required for Pol II elongation. It therefore appears that the processes that allow polymerases to access DNA and are therefore necessary for elongation also disrupt chromatin structure and create a need for a reassembly pathway provided by the Hir/Hpc proteins.
MATERIALS AND METHODS
Screen for mutants: Mutations that are lethal when combined with spt16-11 (T828I P859S) were identified essentially as described in Bender and Pringle (1991) and in Formosa and Nittis (1999). Strain 7915-2-4 (spt16-11 ade2 ade3 ura3) carrying the YEp352 derivative pTF149 (SPT16 ADE3 URA3) was mutagenized with ultraviolet radiation to ∼50% survival on rich medium. About 105,000 colonies were screened for the inability to survive loss of the plasmid and therefore contained no white sectors and were sensitive to 5-fluoroorotic acid (5-FOA; Boekeet al. 1987) at 22°. SPT16 is essential for viability (Maloneet al. 1991), so candidates were tested for null mutations in SPT16 by forming diploids with an spt16-Δ strain. Isolates unable to form white sectors or to grow on medium containing 5-FOA at a permissive temperature for spt16-11 mutants (26°) were considered to represent null mutants in SPT16, dominant synthetic mutants, or plasmid integrants and were discarded. Isolates able to form white sectors and grow on medium containing 5-FOA at 26° but not at a restrictive temperature for spt16-11 mutants (37°) were retained. Most combinations of spt16-11 with defective versions of POB3 were previously shown to cause lethality (Schlesinger and Formosa 2000). Candidates were therefore also tested for complementation of the nonsectoring phenotype and 5-FOA sensitivity by plasmids containing POB3. A total of 29 strains were identified that appeared to harbor mutations that were synthetically lethal in combination with spt16-11 after these tests. A genomic library in the low-copy vector p366 (a generous gift from P. Hieter and F. Spencer) was introduced into some of these strains to identify plasmids that complement the synthetic lethality and therefore produce white colonies resistant to 5-FOA. Plasmids without the SPT16 gene were sequenced to identify candidate genes. Individual genes were then amplified by PCR (primer sequences available upon request) and tested for complementation in low-copy vectors. In the case of HPC2 and HIR1, the synthetic lethality was further confirmed by deleting these genes and then reconstructing double mutants with spt16-11 in standard genetic crosses in the presence of a plasmid containing the HPC2 or HIR1 genes. These were tested for the ability to survive loss of the HIR1 or HPC2 plasmid by testing for growth on medium containing 5-FOA.
Mutations that are lethal when combined with a hir1-Δ gene deletion were identified essentially as described above using strain PKY817 (hir1-Δ ade2 ade3 ura3) carrying the YEp24-derivative pPK171 (HIR1 ADE3 URA3). Approximately 200,000 colonies were screened for the inability to survive loss of pPK171. Two isolates were identified that regained the ability to form white sectors and to grow on medium containing 5-FOA in the presence of the pRS425-derivative pPK174 (HIR1 LEU2). Previous results showed that hir1 mutations are lethal when combined with defects in spt4 (A. Bortvin and F. Winston, personal communication), and tests revealed that one of the mutants from this screen was complemented by a plasmid containing SPT4. The other mutant was complemented by plasmids containing only PAF1. Genetic crosses with marked gene deletions were then performed to confirm and extend these results (Table 2).
Maximal permissive temperature determination: Aliquots of saturated cultures containing similar numbers of cells were distributed onto solid media and incubated at increments of 1°. The highest temperature supporting at least 10% viability was designated the maximal permissive temperature (MPT).
A screen for synthetic defects reveals a genetic interaction between SPT16 and HIR/HPC genes: The spt16-11 (T828I P859S) allele supports robust growth at moderate temperatures, but it causes sensitivity to elevated temperatures, to the dNTP synthesis inhibitor hydroxy-urea (HU), and to the transcription elongation inhibitor 6-azauracil (6-AU; Formosaet al. 2001). HU blocks progression of replication by limiting the availability of dNTPs and 6-AU inhibits elongation of transcription by causing imbalances in the pools of rNTPs (Exinger and Lacroute 1992; Shaw and Reines 2000). Other alleles of SPT16 cause more severe defects than spt16-11 for each of these phenotypes, but no other allele tested affected all of the phenotypes, suggesting a defect in a core function of Spt16 required both for DNA replication and for transcription elongation. Mutations that enhance the spt16-11 mutation could therefore reveal the function of yFACT in either pathway.
We screened for mutations that cause lethality when combined with spt16-11 (see materials and methods). A total of 29 strains carrying mutations that make spt16-11 intolerable were isolated by screening an spt16-11 ade2 ade3 strain for colonies unable to survive loss of a plasmid carrying SPT16, URA3, and ADE3 and therefore uniformly red in color and unable to grow on media containing 5-FOA. A low-copy genomic library was introduced into a set of these mutants, and plasmids that complemented the synthetic defect were identified by their ability to permit growth on media containing 5-FOA and to allow the formation of white sectors. Plasmids with inserts including HIR1, HIR3, and HPC2 were identified in this way. Complementation by these genes alone was confirmed using plasmids containing single open reading frames (ORFs). Since mutations in HIR2/SPT1 cause many of the same phenotypes as mutations in the genes initially identified, a plasmid containing only this gene was constructed and the remaining mutants were tested with all four HIR/HPC genes. Twenty-six of the mutations were complemented by one of these genes (Table 2). The unbiased screen therefore indicates very clearly that spt16-11 mutations cause strong dependence on the function of the HIR/HPC genes.
Reconstructions confirm the interaction with SPT16: Each mutant was complemented by only one of the HIR/HPC genes, suggesting that each mutant had a defect in one of these genes (instead of an alternative in which all synthetic defects with spt16-11 are suppressed by increasing the amount of any HIR/HPC gene). To confirm that individual hir/hpc mutations enhance the defects caused by spt16-11 we constructed deletions of HIR1 and HPC2 and attempted to obtain double mutants with spt16-11 using genetic crosses. Diploids heterozygous for spt16-11 and either hpc2-Δ or hir1-Δ and containing URA3-marked plasmids with a normal copy of either HPC2 or HIR1 were constructed and sporulated. Double mutants were identified among the haploid segregants, and these were tested for the ability to survive loss of the URA3-marked plasmid by selection on media containing 5-FOA. As shown in Figure 1A, the combination of spt16-11 and hpc2-Δ is lethal because cells with this genotype do not produce colonies at any temperature on media containing 5-FOA and therefore do not survive the loss of the plasmid with the HPC2 gene. In contrast, an spt16-11 hir1-Δ combination was found to be viable, but impaired for growth relative to single mutants even at 26° (Figure 1C). The growth defect was more pronounced at elevated temperatures, leading to inviability at 33°, a temperature that is permissive for each single mutant. Combining the spt16-11 and hir1-Δ mutations therefore causes a strong synthetic defect even under permissive conditions, a defect severe enough to allow detection in the nonsectoring assay employed in our screen. We conclude from these reconstructions and the results of the genetic screen that mutation of any of the four HIR/HPC genes causes a strong defect or lethality when combined with the spt16-11 mutation.
The interaction between SPT16 and HIR/HPC genes is allele specific: Different alleles of SPT16 cause significantly different phenotypes (Formosaet al. 2001 and Table 3), suggesting that different mutations might affect different functional domains of the Spt16 protein that are involved in distinct pathways. To determine whether the interaction with defects in HPC/HIR genes is also allele specific, we tested a set of spt16 mutants for the ability to tolerate loss of HPC2. For this purpose a strain with deletions of both HPC2 and SPT16 was constructed and the essential SPT16 gene was provided on a URA3-marked plasmid. Low-copy LEU2-marked plasmids bearing mutant spt16 alleles (Formosaet al. 2001) were then introduced into this strain, and transformants were tested for the ability to survive with only the mutant spt16 allele after loss of the plasmid with the wild-type gene. The spt16 alleles varied in their ability to support growth in a strain lacking HPC2, indicating that the interaction is allele specific (Table 3). Notably, the commonly used spt16-G132D mutation (Evanset al. 1998) supported growth when combined with hpc2-Δ. Surprisingly, while some alleles displayed tight lethality in combination with hpc2-Δ in this test, spt16-11 itself supported very weak growth. This suggests that the plasmid-based SPT16 gene may provide a less stringent test of the synthetic defect with hpc2-Δ than does the endogenous genomic context used in the screen and in Figure 1. This is consistent with previous reports that the temperature sensitivity caused by spt16 mutations can be suppressed by increasing the copy number of the mutated gene (Xuet al. 1993; Evanset al. 1998). We therefore also tested the effect of combining several genomic spt16 mutations with hpc2 using standard genetic crosses (Table 3). The spt16-G132D hpc2-Δ double mutant was recovered from a cross at the expected frequency and displayed only slightly reduced growth at 26° (Figure 1A). However, loss of HPC2 did enhance the temperature sensitivity of the spt16-G132D strain somewhat, since the double mutant displayed a maximal permissive temperature 1°-2° lower than that of the single mutant (Table 3). Combining the spt16-8 or spt16-22 alleles with an hpc2 deletion also produced viable cells with enhanced temperature sensitivity, but the synthetic defect was more severe in these cases than with spt16-G132D, with the maximal permissive temperature decreasing at least 5° compared to the single mutants (Table 3). Combining hpc2-Δ with spt16 mutations was therefore deleterious in all cases tested, but the severity of the defect was strongly allele specific, suggesting that certain defects in Spt16 cause greater reliance on Hpc/Hir proteins.
POB3, but not NHP6, also interacts with HIR/HPC genes: Pob3 and Nhp6 function with Spt16 both in vitro and in vivo (Brewster et al. 1998, 2001; Wittmeyeret al. 1999; Schlesinger and Formosa 2000; Formosaet al. 2001). Spt16 and Pob3 are found in cell extracts only in a complex with one another, while Nhp6 associates only weakly with Spt16-Pob3 (Brewsteret al. 2001; Formosaet al. 2001). We therefore tested the effect of combining mutations in the POB3 or NHP6A/B genes with an hpc2 deletion to see whether defects in all yFACT components cause dependence on HIR/HPC genes. The pob3-L78R hpc2-Δ mutants are viable but grow very slowly at 26° and display increased temperature sensitivity compared to single mutants (Figure 1B). Similar results were obtained with a pob3-7 hir1-Δ combination (Figure 1C). However, an nhp6a-Δ nhp6b-Δ hpc2-Δ strain was no more compromised for growth than was an nhp6a-Δ nhp6b-Δ strain, even at temperatures near the restrictive temperature for nhp6a-Δ nhp6b-Δ strains (data not shown). Therefore, loss of HIR/HPC genes was also severely detrimental to cells lacking normal POB3, but no strong genetic interaction was detected with Nhp6. The components of yFACT therefore can function both together and independently, with Spt16-Pob3 and Nhp6 defects causing very different responses to the loss of Hir/Hpc proteins.
Mutations in HIR/HPC genes are also lethal when combined with other factors that promote elongation: hir1 mutations were previously found to be lethal when combined with some alleles of SPT4, SPT5, or SPT6 (A. Bortvin and F. Winston, personal communication; also see materials and methods). These three genes are functionally related to one another and promote transcriptional elongation (Bortvin and Winston 1996; Hartzoget al. 1998; Wadaet al. 1998). To extend these observations, a screen for additional mutations that cannot be tolerated in the absence of HIR1 was performed, and it revealed that a paf1 mutation has this property (Table 2). Paf1 is a member of a third complex that promotes transcription elongation and has been linked physically and genetically both to Spt4/5/6 and to yFACT (Costa and Arndt 2000; Squazzoet al. 2002). Reconstructions of double mutants with another member of the Paf1 complex, Cdc73, confirm that loss of function of the Paf1 complex causes reliance on the Hir/Hpc proteins (Table 2).
The histone chaperone Asf1 is involved in some Hir/Hpc functions, such as the ability to promote silent heterochromatin formation at telomeres (Sharpet al. 2001; Suttonet al. 2001), but is not required for other functions, such as the role in maintaining kinetochores (Sharpet al. 2002). Asf1 is required in cells lacking members of the Paf1 complex since the asf1-Δ paf1-Δ combination is lethal (Table 2). Mutations in another gene linked to transcription elongation, HPR1 (Chavez and Aguilera 1997), did not cause a defect when combined with hpc2-Δ, demonstrating that not all elongation factors display this effect. Reconstructions also show that combining asf1-Δ with either spt16-11 or pob3-7 results in cells that grow normally under conditions used in our genetic screens (data not shown), but cause a severe decrease in the maximal permissive temperature (Table 2). Asf1 is therefore required for full activity of the Hir/Hpc proteins in maintaining the viability of spt16 and pob3 mutants. Together, these results indicate that defects in three systems that promote transcriptional elongation (Spt4/5/6, the Paf1 complex, and yFACT) all make the Hir/Hpc-Asf1 pathway more important or essential.
SPT16 and POB3 are activators of histone gene transcription: The spt16 alleles that cause the strongest synthetic defects with an hpc2-Δ deletion overlap the set that caused the greatest sensitivity to HU, although the correlation is imperfect (Formosaet al. 2001 and Table 3). This suggests that the same activity of Spt16 that is responsible for allowing growth in the presence of HU is also needed when Hir/Hpc proteins are defective. HU blocks DNA replication by inhibiting the synthesis of dNTP precursors. HU arrest normally represses histone expression, but this response fails in cells lacking normal HIR/HPC genes (Osley and Lycan 1987; Xuet al. 1992). We therefore considered the possibility that Spt16 and Hir/Hpc proteins each contribute to the repression of histone gene transcription, and the synthetic defects observed reflect the inability of a cell to tolerate very high levels of histones. In this model, some spt16 mutations cause sensitivity to HU because they fail to repress histone gene expression during S-phase arrest.
It has previously been reported that the spt16-G132D mutation causes decreased levels of histone gene transcription and that the repression of this transcription is normal during an HU arrest (Xuet al. 1993), but spt16-G132D mutants are relatively resistant both to HU and to loss of Hpc2 and thus might not represent other spt16 alleles. However, as shown in Figure 2, S1 analysis of HTB1 transcripts in several spt16 and pob3 mutants shows that histone gene expression is not increased in any of the mutants tested, but rather is always either unaltered or decreased. The effect is small in some cases, as for spt16-G132D, but over twofold in others, as for spt16-16 and pob3-1. The decrease in HTB1 transcription did not correlate well with sensitivity either to HU or to the loss of Hir/Hpc proteins. For example, spt16-11 causes relatively strong sensitivity to HU and is lethal when combined with hir/hpc mutations but has little or no effect on HTB1 transcript levels. Thus, none of the spt16 or pob3 mutations tested support a role for Spt16-Pob3 as a repressor of histone gene transcription. Further, hir1-Δ and hpc2-Δ mutations in either W303 or A364a genetic backgrounds grew as well as wild-type strains on 100 mm HU (data not shown), indicating that the elevated level of histone gene expression observed under these conditions is not toxic by itself. Finally, the synthetic defect caused by combining spt16-11 with hir/hpc mutations does not seem to be caused by the Hir-phenotype itself (loss of histone gene repression during exposure to HU), since two other mutations that cause this phenotype, spt10-Δ and spt21-Δ (Sherwood and Osley 1991), did not cause severe decreases in viability or rate of growth when combined with spt16-11 (data not shown). We conclude that the interaction between SPT16-POB3 and HIR/HPC genes is not due to overlapping roles as independent repressors of histone gene expression.
SPT16 and POB3 mutants are affected by alterations in the number of copies of histone genes: The S. cerevisiae genome includes two copies of each of the genes that encode the four histone proteins. The four genes that encode H2A and H2B are found as two paired sets (HTA1-HTB1 and HTA2-HTB2), as are the genes that encode H3 and H4 (HHT1-HHF1 and HHT2-HHF2). Deleting some histone gene pairs causes the global misregulation of transcription initiation site selection called the Sptphenotype, as well as other defects in transcription (Clark-Adamset al. 1988; Osley 1991). Unbalanced increases in either H2A-H2B or H3-H4 gene sets also cause the Sptphenotype, as well as chromosome instability (Meeks-Wagner and Hartwell 1986; Clark-Adamset al. 1988). The appropriate availability of a balanced set of histone proteins is therefore important for maintaining appropriate regulation of transcription and genomic stability. The hir/hpc mutants deregulate expression of both pairs of genes that encode H3-H4, but only one of the H2A-H2B sets (HTA1-HTB1; Osley and Lycan 1987; Xuet al. 1992), leading to unbalanced overexpression of the histone gene sets. We therefore considered the possibility that hir/hpc mutations affect the quality of chromatin by altering the ratio of histones available and that this alteration in chromatin quality impairs yFACT activity such that mutations in SPT16 and POB3 cannot be tolerated. To test this, we determined whether direct perturbations of histone gene copy numbers affect the growth of spt16 or pob3 mutants.
As shown in Figure 3, decreasing the histone gene copy number affected the growth of both spt16 and pob3 mutants. Surprisingly, these mutants responded in opposite ways to these changes. The growth of an spt16-11 strain was strongly impaired and the maximal permissive temperature was significantly reduced by deletion of either HHT1-HHF1 or HHT2-HHF2, indicating that a strain with this defect in Spt16 is sensitive to decreased expression of H3-H4 proteins (Figure 3 and Table 4). A similar effect was observed with an spt16-G132D strain (Figure 3), so this genetic interaction with diminished H3-H4 copy number is not as strongly allele specific as the interaction with hpc2. However, when the same H3-H4 gene deletions were combined with a pob3-7 mutation, instead of causing an enhanced defect the double mutant grew slightly more rapidly than a pob3-7 strain and displayed the same maximal permissive temperature (Figure 3 and Table 4). In contrast, deletion of HTA2-HTB2 was strongly detrimental to the pob3-7 strain, but caused a slight increase in the growth of an spt16-11 strain (Figure 3 and Table 4). We have been unable to perform a similar test with a deletion of HTA1-HTB1 because, unlike results obtained in other labs (e.g., Rykowskiet al. 1981; Norris and Osley 1987; Clark-Adamset al. 1988), this mutation alone caused inviability in three distinct strain backgrounds in our hands. Defective POB3 therefore causes sensitivity to decreased expression of H2A-H2B, but not to decreased levels of H3-H4, and defective SPT16 causes the opposite response (Table 4).
Increasing the levels of histone genes also affected the growth of spt16 and pob3 mutants, and the effects were consistent with the results obtained with histone gene deletions. As shown in Figure 4, increasing the expression of H3-H4 by introducing the HHT2-HHF2 gene pair on a low-copy vector had little effect on a pob3-7 strain, but partially suppressed the temperature sensitivity of an spt16-11 strain. [High-copy plasmids carrying these gene pairs impaired the growth of both wild-type and mutant strains significantly (data not shown).] Conversely, increasing the amount of H2A-H2B partially suppressed the growth defect of a pob3-7 strain, but was deleterious to the growth of an spt16-11 strain (Figure 4).
Other studies have suggested that the ratio of H2A-H2B to H3-H4 expression is important for normal chromatin function (Meeks-Wagner and Hartwell 1986; Kaufmanet al. 1998). As summarized in Table 4, if we interpret the changes in histone gene dosage described above in terms of H2A-H2B:H3-H4 ratios, then high ratios are deleterious to spt16 mutants, and low ratios enhance growth of these strains, while pob3 mutants prefer high ratios and are impaired by low ratios. However, both spt16 and pob3 mutations displayed synthetic defects with hir/hpc and asf1 mutations. Therefore, although yFACT function is affected by the levels of histone gene expression, our data suggest that the functional interaction between Spt16-Pob3 and the Hir/Hpc proteins cannot be accounted for by changes in histone gene expression alone.
Spt16-Pob3 does not interact genetically with CAF-I: Hir1 and Hir2 were initially identified as repressors of histone gene expression (Osley and Lycan 1987). However, similar proteins from mammalian and amphibian sources participate in nucleosome assembly in a pathway independent of DNA replication and repair (Ray-Galletet al. 2002). The Hir/Hpc proteins in yeast also overlap functionally with the DNA replication-dependent nucleosome assembly factor CAF-I (Kaufmanet al. 1998; Sharpet al. 2002). We therefore determined whether loss of CAF-I is detrimental to spt16 or pob3 mutants. Neither the growth rate nor the maximal permissive temperature of pob3-L78R, pob3-7, spt16-G132D, or spt16-11 strains was affected by deletion of CAC1, the gene encoding the largest subunit of CAF-I (data not shown). Similar results were obtained with a deletion of CAC2 combined with an spt16-11 mutation (data not shown). Therefore, although CAF-I and the Hir/Hpc pathways are redundant for some functions, CAF-I does not appear to have the activity of the Hir/Hpc proteins that becomes essential when yFACT is defective.
spt16 mutations display interactions with histone acetyltransferase genes: Spt16 interacts with the histone acetyltransferase (HAT) complex NuA3 (Johnet al. 2000). Evidence supporting this interaction includes partial copurification, a two-hybrid interaction between Spt16 and the catalytic subunit of NuA3 (Sas3), and a synthetic sensitivity to 6-AU when a deletion of SAS3 is combined with high-level expression of a truncated version of SPT16 (Johnet al. 2000). Consistent with these data, we found that combining spt16-G132D with a sas3 deletion caused a synthetic growth defect (Figure 5). The double mutant was unable to grow at a temperature permissive for the single mutants and was more sensitive to 6-AU than were the single mutants. The synthetic defect was relatively mild and was not observed with spt16-11 or pob3-7 in strains from the A364a background (data not shown). In the W303 background sas3-Δ somewhat enhanced the Ts-phenotype caused by spt16-11 but did not affect sensitivity to 6-AU (Figure 6 and data not shown). Because of the evidence for a physical interaction between Spt16 and Sas3, this genetic interaction could indicate that the loss of NuA3 enhances some spt16 mutations by destabilizing functional complexes containing Spt16 protein. Alternatively, Spt16-Pob3 could be more generally sensitive to the modification state of histones and nucleosomes. To further characterize the dependence of yFACT on histone acetylation, we determined whether spt16-11 interacts genetically with other HATs.
Strains with combinations of spt16-11 and mutations in various genes encoding HATs were constructed and tested for growth at elevated temperatures and on media containing 6-AU. Deletion of HAT1 had no effect on the growth of an spt16-11 strain (data not shown), but gcn5-Δ, elp3-Δ, and esa1-L254P (Clarkeet al. 1999) were all detrimental when combined with spt16-11 (Figure 6). Spt16 function is therefore sensitive to mutations in many HATs. The defect was particularly severe with gcn5-Δ, with the double mutant displaying at least a 5° decrease in the maximal permissive temperature relative to single mutants and slow growth even under permissive conditions. The gcn5-Δ mutation is synthetically lethal with a sas3-Δ mutation (Howeet al. 2001), suggesting that these two enzymes overlap for some critical function. Our results are consistent with participation of Spt16 and Sas3 in a common function, which, when inactivated, causes greater reliance on Gcn5. Overall, the genetic interactions between SPT16 and multiple HAT genes suggest that yFACT function is affected by the acetylation pattern of histones.
Consistent with a role for histone acetylation in yFACT function, we previously showed that deletion of the histone deacetylase encoded by RPD3 partially suppresses the temperature sensitivity caused by a pob3 mutation (Formosaet al. 2001). Although an rpd3 deletion did not have a strong effect when combined with an spt16-G132D mutation, the rpd3 spt16-11 double mutant has a significant growth defect at 30° compared to that of the single mutants, representing a decrease of ∼5° in the maximal permissive temperature (Figure 5). Therefore, spt16-11 mutants are also affected by the loss of a deacetylase, but as observed with changes in the copy number of histone genes, spt16 and pob3 mutations respond in opposite ways to the loss of RPD3.
Defects in SPT16 cause reliance on normal histone tails: To test whether the effects noted when spt16 and pob3 mutations were combined with mutated acetyltransferases and deacetylases result directly from alterations in histones, we combined yFACT mutations with mutations in the genes that encode the histone proteins H3 and H4. For this purpose, strains were constructed that had deletions of both sets of the genes that encode histones H3 and H4, a low-copy plasmid carrying URA3 and the normal HHT2-HHF2 locus, and spt16, pob3, or nhp6a/b-Δ mutations. Plasmids with mutated alleles of HHT2-HHF2 were then introduced into these strains and tested for the ability to replace the wild-type plasmid by selecting on medium containing 5-FOA.
An otherwise normal strain can tolerate the loss of the N-terminal tail of either H3 or H4, but an spt16-11 strain displayed increased temperature sensitivity when either tail was deleted (Table 5). Point mutations that affect residues known to be acetylated under different circumstances were also tested. Mutations in H3 did not produce strong effects, but changes in the H4 tail caused defects. Changing the lysine residues at positions 5 and 12 in H4 to arginines caused a growth defect in the spt16-11 strain, whereas changing the same residues to glutamines was tolerated. Arginine residues are chemically similar to lysines, but cannot be acetylated, while glutamine is chemically similar to the acetylated form of lysine. These results therefore suggest that the ability to acetylate these residues of H4 contributes to the execution of the Spt16 function. Newly synthesized H4 is usually acetylated on lysines 5 and 12 (Sobelet al. 1995), suggesting that FACT function may be related to the nucleosome assembly process. In contrast, acetylation of lysines 8 and 16 of H4 correlates with transcriptionally active chromatin, and these sites also appear to be important for Spt16 activity. Mutation of residues 8 and 16 to arginines (but not to glutamines) caused a growth defect in an spt16-11 strain, although one less severe than that of the K5,12R mutations (Table 5). Consistent with the defects observed with multiple HAT mutations, Spt16 activity appears to be affected by the ability of histone tails to be modified at multiple sites. It is notable that the N terminus of Spt 16 shares significant homology with a family of aminopeptidases, but lacks the active site residues found in members of the family with hydrolase activity (Aravind and Koonin 1998; C. Kaplan, personal communication). This suggests that Spt16 could bind directly to the N-terminal tails of histones, and this binding could be affected by modifications of these tails.
As noted for other perturbations of histone expression, pob3 mutants were also affected by histone tail mutations, but in ways distinct from and sometimes opposite to those noted for the spt16-11 strain. As with spt16-11 cells, loss of either the H3 or the H4 tail was detrimental to a pob3-L78R strain (Table 5). In contrast, whereas the H4 K5,12R mutations were more detrimental than the K8,16R mutations in an spt16-11 strain, in a pob3-L78R strain the K5,12R mutations had no effect but the K8,16R mutations were intolerable. These experiments again show that the partners Spt16 and Pob3 are affected very differently by changes in the pattern of histone tail modifications.
The function of Nhp6 is also affected by mutations in acetylases and deacetylases (Yuet al. 2000), and, as with Spt16-Pob3, the effects appear to directly involve histone tails. As shown in Table 5, combining deletions of both genes that encode Nhp6 with a deletion of the N-terminal tail of either H3 or H4 is lethal at 30°, representing a drop in the maximal permissive temperature of at least 7°. This defect is considerably more severe than that of phenotypes caused by combinations with spt16-11. The pattern of effects caused by nhp6a/b-Δ is also different from those observed in either the spt16-11 cells or the pob3-L78R cells. For example, the nhp6a/ b-Δ strain did not respond to the H4 K5,12R mutations that produced the strongest effect with spt16-11. Therefore, although Nhp6 and Spt16-Pob3 collaborate in vitro and in vivo, deficiencies in each protein cause different defects with respect to functional modifications of the N-terminal tail of histone H4. Overall, Nhp6 and Spt16-Pob3 appear to have both common and distinct functional roles.
SPN/FACT is a broadly conserved eukaryotic protein complex that binds to and alters the properties of nucleosomes in vitro (Orphanideset al. 1999; Formosaet al. 2001). This factor appears to moderate the inhibitory effects of chromatin, facilitating the progression of polymerases along nucleosomal templates (Orphanideset al. 1998). Previous genetic analysis indicated that this activity is important in vivo for both transcription and replication (Maloneet al. 1991; Rowleyet al. 1991; Johnet al. 2000; Schlesinger and Formosa 2000; Yuet al. 2000; Brewsteret al. 2001; Formosaet al. 2001). To further define the role of the yFACT component Spt16, we used an unbiased genetic strategy to identify factors that support its activity or that rely upon it to complete its function efficiently. Of the 29 mutants identified, 26 affected HIR1, HIR2, HIR3, or HPC2. Both spt16 and pob3 mutations, but not loss of Nhp6, cause this reliance on the HIR/HPC pathway, and the effect with spt16 mutations is strongly allele specific. The Hir/Hpc pathway is therefore revealed as essential for survival when yFACT activity is compromised in specific ways.
Synthetic defects can signal any of several relationships between gene products (Guarente 1993). Spt16-Pob3 and the Hir/Hpc proteins could display mutual dependence because they act within a common complex, but no data have been reported that support physical interactions among these proteins. Each pathway could act independently to repress histone gene expression, but data presented here and elsewhere (Xuet al. 1993) indicate that instead Spt16-Pob3 is an activator of transcription at these loci. hir/hpc mutations cause misregulation of histone gene expression, which could lead to formation of aberrant nucleosomes. FACT acts on nucleosomes, so hir/hpc mutations could impose a requirement for optimal Spt16-Pob3 activity by decreasing the quality of the FACT substrate. However, spt10 and spt21 mutations also cause misregulation of histone gene expression (Sherwood and Osley 1991), but these defects are tolerated in combination with the spt16-11 mutation (although spt10 hir1 combinations are lethal; D. Hess and F. Winston, personal communication). Further, we find that directly perturbing histone gene expression by altering the number and ratio of histone genes does not recapitulate the lethal effects observed with hir/hpc mutations. Instead, to our surprise, spt16 and pob3 mutants responded in opposing ways to altered histone gene ratios and to changes in the structure and modifications of histones. In general, high ratios of H2A-H2B to H3-H4 were detrimental to spt16 mutants, but suppressed the growth defects of pob3 mutants, and low ratios had the opposite effects (Figures 3 and 4, Table 4). Altering either the potential for H3-H4 tails to be acetylated or the enzymes that modify them also caused opposing effects (Figures 5, 6, 7 and Table 5). Appropriate histone availability and modification are therefore important determinants of FACT activity, but this does not appear to be sufficient to explain the genetic relationships observed between the Spt16-Pob3 and Hir/Hpc pathways.
The Hir/Hpc proteins were initially identified through their role in regulating histone gene expression, but recent results show that they also function in nucleosome assembly (Kaufmanet al. 1998; Lorainet al. 1998; Sharpet al. 2001; Suttonet al. 2001; Ray-Galletet al. 2002). This activity suggests a model for interpreting the results presented here (Figure 7). In this model, we propose that SPN/FACT has two functions. First, as suggested previously (Orphanideset al. 1999; Kireevaet al. 2002), it mediates the destabilization or disassembly of nucleosomes to facilitate passage of polymerases. Second, it promotes reassembly of chromatin after the polymerase has passed. This could involve maintaining contact with the nucleosomal components so that the same histone proteins that were separated could be reunited, or it could mean recruiting replacement components from a free pool. In this model, some mutations in SPT16 and POB3 cause increased levels of aberrant chromatin structures to accumulate, perhaps because the disassembly function is unregulated or because the reassembly activity is deficient. We further propose that the Hir/Hpc proteins act as a general “nucleosome repair” pathway that restores nucleosomes with an aberrant composition to a normal one. Mutations in FACT that enhance the formation of aberrant nucleosomes therefore cannot be tolerated when there are defects in the Hir/Hpc pathway because this abrogates the ability to repair the damage.
This model is consistent with the known properties of FACT and also can explain several of the observations presented here. Removing one or both H2A-H2B dimers makes nucleosomes less inhibitory to RNA polymerase II (Kireevaet al. 2002). FACT binds both to nucleosomes and to H2A-H2B dimers and promotes RNA Pol II elongation on nucleosomal templates (Orphanides et al. 1998, 1999; Formosaet al. 2001). Binding to H2A-H2B dimers was initially viewed as a mechanism for disrupting the nucleosome to create a less inhibitory form. In our proposal, the binding of FACT to nucleosomes and to fragments of nucleosomes reflects distinct activities required at different points in a cycle that is completed when the nucleosome is restored to a normal form. Some contacts promote dissociation of H2A-H2B dimers, while others tether the fragments to enhance efficient reassembly after the polymerase has passed. Spt16 and Pob3 are therefore each needed to complete the cycle effectively, and defects in either could lead to the formation of aberrant nucleosomes. Therefore, either spt16 or pob3 mutations could cause dependence on the Hir/Hpc pathway, and this would depend on which part of the cycle was defective in each mutant. While both proteins work together to promote this cycle, they contact different parts of the nucleosome and might have very different responsibilities regarding the disassembly and reassembly phases of the cycle. Changes in the availability, structure, and modification status of the histones could therefore affect spt16 and pob3 mutants very differently, or even in opposite ways, as we have observed.
Nhp6 promotes binding of Spt16-Pob3 to nucleosomes in vitro (Formosaet al. 2001) and supports the function of Spt16-Pob3 in vivo (Brewsteret al. 2001; Formosaet al. 2001). However, nhp6 mutants do not cause reliance on the Hir/Hpc system. We propose that Nhp6 acts primarily to enhance loading of Spt16-Pob3 to nucleosomes at promoters or replication origins, but is not involved in the subsequent elongation steps and therefore does not influence the nucleosome reassembly reaction.
Defects in other elongation factors also cause reliance on the Hir/Hpc pathway. This includes the functionally related Spt4, Spt5, and Spt6 proteins (DeSilvaet al. 1998; F. Winston, D. Hess and A. Bortvin, personal communication), as well as members of the Paf1 complex (Table 2). This could indicate that these factors also promote reassembly of nucleosomes directly (as previously suggested for Spt4-Spt5; Hartzoget al. 1998), that they influence the ability of FACT to promote this activity, or that delayed elongation promotes dissociation of nucleosomal components by prolonging the time spent in a tenuous disassembled state. Supporting the notion that Spt4/5/6 and the Paf1 complex affect yFACT function, proteins in these three complexes have been linked physically and genetically to one another (Costa and Arndt 2000; Squazzoet al. 2002). Further, the Hir/Hpc proteins have been linked physically with TAFII subunits (Sanderset al. 2002), and Asf1 has been linked to the TFIID-associated protein Bdf1 (Chimuraet al. 2002). The functional association revealed by genetic screens between the Hir/Hpc-Asf1 proteins and factors that promote polymerase passage may therefore be supported by the physical association of Hir/Hpc-Asf1 proteins with general transcription factors.
While many features of this model remain to be demonstrated, it is reasonable that the factor responsible for the disassembly of nucleosomes would have an important role in their reassembly and that some additional pathway would exist to ensure completion of this important function. These results therefore provide insight into the functions of both FACT and the Hir/Hpc system.
We thank Jennifer Ginn, Soo Y. Lee, the UC Berkeley MCB140L class, and Jonathon Tuttle’s AP Biology class at Hunter High School for assistance with the genetic screens; Fred Winston for sharing unpublished results; Sharon Dent, Hannah Klein, Loraine Pillus, and Mary Ann Osley for providing plasmids and strains; Rick Singer for suggestions regarding FACT nomenclature; and Craig Kaplan for valuable discussions regarding the aminopeptidase homology of Spt16. This work was supported by grants from the National Science Foundation to T.F. (MCB 9986142) and to P.D.K. (MCB 9982909), grants from the National Institutes of Health to T.F. and D.J.S. (GM64649) and to D.J.S. (GM39067), Department of Energy funds administered through the Lawrence Berkeley National Laboratory to P.D.K., and a postdoctoral fellowship from the American Cancer Society to M.D.A.
Communicating editor: F. Winston
- Received July 31, 2002.
- Accepted September 12, 2002.
- Copyright © 2002 by the Genetics Society of America