The histone chaperone FACT is involved in the elongation stages of RNAPII-mediated gene transcription, facilitating the dismantling and reassembly of nucleosomes encountered by the transcription complex 
. For the Spt16 component of yeast FACT, several substitution mutations have been identified that impair transcription-linked nucleosome reassembly but preserve the essential functions of Spt16, and are thus selective in their effects (). We report here that one of these substitution mutations, spt16-E857K
, also compromises an important histone-related activity for transcription elongation, as indicated by biochemical and genetic interactions. Co-purification studies showed that the Spt16-E857K mutant protein, while unaffected for interaction with RNAPII, is impaired for interaction with histone H2B, which probably reflects decreased interaction with the H2A–H2B heterodimer. The spt16-E857K
mutation has deleterious genetic interactions with mutations affecting the transcription elongation factors Bur kinase, Spt4–Spt5, and Paf1C. Most of these mutations, like spt16-E857K
itself, also impair transcription-linked nucleosome reassembly as indicated by cryptic-promoter activity 
. In light of the evidence for functionally distinct classes of cryptic promoters 
, the poor cell growth of spt16-E57K
cells harboring other nucleosome-reassembly mutations could in principle reflect an additive effect resulting in growth inhibition due to progressively more impaired nucleosome reassembly. However, our finding that an Rpd3S histone-deacetylase mutation, which also activates cryptic promoters 
, alleviates the growth defect of spt16-E857K
cells that are also mutant for Bur kinase or Paf1C is more consistent with models that emphasizes the importance of histone acetylation for disassembly of a nucleosome encountered by the transcription elongation complex. Bur kinase, Spt4–Spt5, and Paf1C form a functional pathway: Bur phosphorylates Spt4–Spt5, which in turn recruits Paf1C to the transcription complex. Our findings indicate that this Bur/Spt4–Spt5/Paf1C pathway and FACT have overlapping nucleosome-related activities that are important for transcription elongation.
Bur kinase and Spt4–Spt5, which have been independently implicated genetically in facilitating the transcription of nucleosomal templates 
, most likely exert this effect through Paf1C: yeast Paf1C helps to destabilize nucleosomes upon transcription induction, and human Paf1C, in cooperation with the histone acetyltransferase p300, facilitates in vitro
transcription elongation on a chromatin template 
. The disassembly of normally hypoacetylated nucleosomes is presumably hindered by the combined effects of spt16-E857K
plus a mutation affecting Paf1C or an upstream pathway component, resulting in impaired transcription elongation and consequent growth effects. Conversely, Rpd3S deacetylase inactivation, which increases the acetylated state of nucleosomes 
, most likely allows more effective nucleosome disassembly even when Spt16 and Paf1C are functionally compromised. This interpretation, coupled with the findings that spt16-E857K
, in Rpd3S-dependent fashion, and the Bur/Spt4–Spt5/Paf1C pathway have deleterious genetic interactions with the histone chaperone HirC () 
, supports a model in which FACT, Paf1C, and HirC separately influence transcription-linked disassembly of Rpd3S-deacetylated nucleosomes.
Nucleosome acetylation status is important in spt16-E857K
cells, as indicated not only by the effects of Rpd3S inactivation but also by the genetic interactions resulting from inactivation or mutation of the HAT enzymes Gcn5, the acetyltransferase subunit of SAGA, and Esa1, the acetyltransferase subunit of NuA4. Both SAGA and NuA4 supply acetyltransferase activity during transcription elongation 
, and both can facilitate the transcription of a nucleosomal template in vitro 
. Thus acetylation is important for transcription-linked nucleosome disassembly, and becomes even more important in spt16-E857K
cells. The genetic interactions between spt16-E857K
and Bur/Spt4–Spt5/Paf1C pathway mutations suggest that FACT and/or these elongation factors may mediate transcription-linked nucleosome acetylation by the SAGA and/or NuA4 HAT complexes.
A Bur/Spt4–Spt5/Paf1C activity that is unlikely to be involved in the genetic interactions reported here is the stimulation of Set2 methyltransferase to convert H3K36me2 to H3K36me3. Nucleosomal H3K36me2 recruits Rpd3S in vitro
and in vivo
for the restoration of chromatin repression; H3K36me3 also binds Rpd3S in vitro
, but its role in vivo
is unclear 
. H3K36me3 abundance is significantly decreased by several mutations affecting Bur kinase and Paf1C 
. In contrast, the spt4Δ
, and spt5-194
mutations that also have deleterious genetic interactions with spt16-E857K
leave H3K36me3 levels unaffected 
. Therefore, H3K36me3 status is not correlated with deleterious genetic interactions with spt16-E857K
. Similar considerations dispense with other transcription-related activities that depend on Paf1C, including histone methylation at H3K4 and H3K36, and maintenance of CTD Ser-2 phosphorylation levels 
. We find that H3K4 methylation is normal in spt16-E857K
mutant cells, the loss of H3K4 methylation or the Ser-2 kinase CTDK-1 fails to impair the growth of these spt16
mutant cells, and the absence of H3K36 methylation is actually beneficial. Thus the roles of Paf1C in histone methylation and CTD Ser-2 phosphorylation are unlikely to be involved in the genetic interactions described here.
An Spt4–Spt5 activity that does have relevance to the genetic interactions described here is the regulation of Rpd3S localization. Genome-wide assessment shows that Rpd3S is not uniformly distributed along transcribed regions 
. In contrast, in spt4Δ
cells Rpd3S is more uniformly distributed, matching the distribution of RNAPII itself. This aberrant distribution may reflect the association of Rpd3S with the phosphorylated CTD of RNAPII, a recruitment step that precedes Rpd3S binding to methylated H3K36 in transcribed nucleosomes 
. Without Spt4 to mediate transfer from the CTD to the transcribed nucleosome, Rpd3S may be ineffective at nucleosomal deacetylation. A similar situation may be brought about by the spt5-4
point mutations used here. The lack of influence of Rpd3S on the deleterious genetic interactions between spt16-E857K
and Spt4–Spt5 mutations, as found here, may be due to this aberrant Rpd3S management by the mutant elongation complex.
Cells relying on Spt16-E857K protein are impaired not only for transcription-linked nucleosome reassembly 
, but also for the disassembly of nucleosomes encountered by the transcription elongation complex, as evidenced by the genetic interactions described here with factors involved in transcription elongation. In contrast to the dominant negative effect of mutant FACT (containing the Spt16-E857K mutant protein) on nucleosome reassembly, however, the effects of Spt16-E857K on transcription-linked nucleosome disassembly are recessive, as demanded by the procedure used here to identify deleterious genetic interactions between spt16-E857K
and mutations inactivating Bur2 and HirC. Therefore, mutant FACT competes poorly with normal FACT in the genetic assays of transcription-linked nucleosome disassembly, even though it competes effectively with normal FACT in the genetic assays of transcription-linked nucleosome reassembly. One model to account for this functional difference proposes that FACT association with the transcription elongation complex is contingent on nucleosome disassembly, and that the same FACT complex that associates with an elongation complex as a result of nucleosome disassembly is retained for reassembly of the same nucleosome. This model for FACT interactions extends earlier findings that histone retention during transcription elongation depends on Spt16 activity 
The activities of Spt16 for nucleosome reassembly after passage of the transcription elongation complex and nucleosome disassembly as the transcription elongation complex encounters the next nucleosome may be genetically separable, as suggested by the genetic interactions of spt16-E763G, another substitution allele that impairs nucleosome reassembly. This spt16 mutation did not affect histone H2B interactions, had no deleterious genetic interactions with the HirC mutations or histone H2A–H2B overexpression, caused only mild temperature sensitivity in combination with bur2Δ, and showed no interactions with any of the other bur1 mutant alleles ( and , and data not shown). This mutant allele also has a minimal extended phenotype (). These genetic distinctions suggest that Spt16-mediated transcription-linked nucleosome disassembly is not simply transcription-linked nucleosome reassembly run in reverse.
More striking spt16
allele differences involve spt16-E857Q
, a substitution mutant allele that relieves the cold sensitivity of a histone H3 mutation 
. This mutation, causing a glutamine substitution at residue 857, fails to activate cryptic promoters and lacks genetic interactions with bur2Δ
, while spt16-E857K
, encoding a lysine substitution at residue 857, fails to alleviate the cold sensitivity of the H3 mutation (unpublished observations) 
. These findings, plus others 
, show that the segment of the Spt16 protein encompassing residues 857 and 763 has multiple nucleosome-related activities.
Although almost the entire SPT16
ORF was subjected to mutagenesis in the identification of the spt16
mutant alleles studied here and listed in , the mutations found to have dominant effects on transcription-linked nucleosome reassembly, including spt16-E857K
, alter only a limited segment approximating the middle domain of the large Spt16 polypeptide 
. A bioinformatics approach 
that predicts whether a polypeptide sequence has a folded structure resembling any of the known folds in the PDB protein-structure database indicates, with very high confidence, that the segment of Spt16 encompassing the sequence changes in adopts the configuration of a double PH domain. This structural motif comprises two Pleckstrin Homology (PH) domains, each a 7-strand anti-parallel β-barrel structure capped at one end by a helix, oriented similarly and intimately associated with each other. The double PH domain predicted for Spt16 strongly resembles that determined experimentally for Pob3, the binding partner of Spt16 in yeast FACT 
. Thus both components of FACT may contain a double PH domain. Another protein known to have a double PH domain is Rtt106, a histone chaperone (like Spt16 and Pob3) that is involved in transcription 
. Deletion of the gene encoding Rtt106 is deleterious for cells relying on the Spt16-E857K mutant protein (unpublished observation), suggesting that these structurally related proteins have related functions. The segment of Spt16 housing the predicted double PH domain thus has several transcription-related activities.