We discuss the regulation of the histone genes of the budding yeast Saccharomyces cerevisiae. These include genes encoding the major core histones (H3, H4, H2A, and H2B), histone H1 (HHO1), H2AZ (HTZ1), and centromeric H3 (CSE4). Histone production is regulated during the cell cycle because the cell must replicate both its DNA during S phase and its chromatin. Consequently, the histone genes are activated in late G1 to provide sufficient core histones to assemble the replicated genome into chromatin. The major core histone genes are subject to both positive and negative regulation. The primary control system is positive, mediated by the histone gene-specific transcription activator, Spt10, through the histone upstream activating sequences (UAS) elements, with help from the major G1/S-phase activators, SBF (Swi4 cell cycle box binding factor) and perhaps MBF (MluI cell cycle box binding factor). Spt10 binds specifically to the histone UAS elements and contains a putative histone acetyltransferase domain. The negative system involves negative regulatory elements in the histone promoters, the RSC chromatin-remodeling complex, various histone chaperones [the histone regulatory (HIR) complex, Asf1, and Rtt106], and putative sequence-specific factors. The SWI/SNF chromatin-remodeling complex links the positive and negative systems. We propose that the negative system is a damping system that modulates the amount of transcription activated by Spt10 and SBF. We hypothesize that the negative system mediates negative feedback on the histone genes by histone proteins through the level of saturation of histone chaperones with histone. Thus, the negative system could communicate the degree of nucleosome assembly during DNA replication and the need to shut down the activating system under replication-stress conditions. We also discuss post-transcriptional regulation and dosage compensation of the histone genes.
THE histone genes have been studied intensively for several decades in model organisms and in humans. Initially, they were studied because they encode proteins of major importance and their function in packaging DNA into the nucleus as chromatin was clearly understood and because they are excellent models for cell-cycle-dependent regulation of gene expression. However, the histone genes are atypical in that there are multiple copies of most histone genes in all organisms. In this regard, the yeasts are perhaps the most tractable model organisms. In particular, the budding yeast Saccharomyces cerevisiae possesses only two copies each of the major core histone genes. This review focuses on the regulation of the yeast histone genes.
The histones are highly positively charged proteins that package DNA, which is negatively charged, into the nucleus in the form of chromatin, the substance of chromosomes. Packaging was initially thought to be the only function of the histones, but it is now clear that they also participate in gene regulation via both classical and epigenetic mechanisms. Nevertheless, the regulatory mechanisms appear to act primarily by controlling the extent to which DNA is made accessible to regulatory factors, i.e., by controlling packaging.
The basic structural unit of chromatin is the nucleosome core, which is composed of two molecules each of the four core histones—H2A, H2B, H3, and H4—formed into an octamer, around which is wrapped ∼147 bp of DNA in 1.75 superhelical turns (Luger et al. 1997). The central ∼80 bp of the nucleosome is organized by an H3-H4 tetramer containing two molecules each of H3 and H4. H2A-H2B dimers are bound on both sides of the H3-H4 tetramer. More complex eukaryotes possess several variants of the histones H2A and H3. The only variants found in budding yeast are H2A.Z (a close variant of H2A) and CenH3 (an H3 variant found only in centromeric nucleosomes). The nucleosome is the structural repeat unit of chromatin and includes the linker DNA and the linker histone H1 (Thoma et al. 1979). In vivo, nucleosomes are regularly spaced along the DNA, with a characteristic repeat length. In budding yeast, the repeat is ∼165 bp (Thomas and Furber 1976; Lohr et al. 1977). Many regulatory regions coincide with gaps in the nucleosomal array, referred to as nucleosome-depleted regions (Yuan et al. 2005).
Typical H1 histones have a central globular domain flanked by a short N-terminal tail and a long, very positively charged, C-terminal tail. The globular domain binds near the dyad axis of the nucleosome core, sealing the two turns of DNA, and the C-terminal tail binds along the linker DNA, helping to fold the nucleosomal array into a higher-order structure. Most cells have about one molecule of H1 per nucleosome, but those cells that space their nucleosomes close together, i.e., with short linker DNA (short repeat length), tend to have significantly less H1. Examples include neuronal cells (Thomas and Thompson 1977) and budding yeast (Thomas and Furber 1976; Lohr et al. 1977). The budding yeast H1 is atypical in that it possesses two globular domains (Landsman 1996; Patterton et al. 1998: Sanderson et al. 2005). In yeast, there is only about one H1 per 37 nucleosomes (Friedkin and Katcoff 2001).
The structure of chromatin provides some clues as to the principles that might be important in regulating the histone genes and histone production. First, the equal stoichiometry of the histones in the nucleosome core suggests that the major core histones (H2A, H2B, H3, and H4) would be produced in approximately equal proportions and that production of H1 would be a certain fraction of this. Second, as every scientist who has attempted to reconstitute chromatin using DNA and histones knows from bitter experience, even a very small excess of histones over DNA is sufficient to aggregate and precipitate the chromatin. This problem occurs when the DNA charge neutralization exceeds a critical value (Clark and Kimura 1990). Thus, we would expect the cell to regulate the relative and absolute amounts of each histone produced and its mode of delivery to the DNA very tightly. These requirements account for the coregulation of histone genes and their complex regulation at the transcriptional, post-transcriptional, and post-translational levels.
Histone production must be regulated during the cell cycle because the cell must replicate not only its DNA during S phase, but also its chromatin. Consequently, the histone genes are activated in late G1 to provide sufficient core histones to assemble the replicated DNA into chromatin. Clearly, enough histone to assemble an entire genome’s worth of DNA into nucleosomes must be synthesized in a timely fashion. Inhibition of DNA synthesis results in rapid repression of the histone genes, indicating that their expression is tightly coupled to replication (Osley 1991). In more complex eukaryotes, some histone gene expression is independent of replication, providing histones for chromatin assembly occurring outside S phase (Tagami et al. 2004), but it is unclear to what extent this is true for yeast.
The Yeast Histone Genes
S. cerevisiae possesses two genes encoding each of the four major core histones, plus single genes encoding H2A.Z (HTZ1), centromeric H3 (CSE4), and H1 (HHO1) (Figure 1A). The major core histone genes are organized into four loci, each containing two histone genes divergently transcribed from a central promoter. Two loci, HHT1-HHF1 and HHT2-HHF2, encode identical H3 and H4 proteins (Smith and Murray 1983). Neither of these gene pairs is essential, but the cell requires one HHT-HHF locus for survival, indicating that H3 and H4 are essential (Smith and Stirling 1988). The other two loci, HTA1-HTB1 and HTA2-HTB2, encode almost identical H2A and H2B proteins (Hereford et al. 1979). Initially, it was thought that HTA1-HTB1 and HTA2-HTB2 could also substitute for one another, although it was clear that HTA1-HTB1 is not equivalent to HTA2-HTB2 because deletion of the former is associated with a significant growth phenotype, whereas deletion of the latter has no obvious effect (Norris and Osley 1987). However, it is now clear that hta1-htb1Δ cells survive only if the HTA2-HTB2 locus has been duplicated and that this duplication can occur in the S288C strain background but not in the W303 background because the latter lacks one of two Ty1 elements required for the duplication (Libuda and Winston 2006). Thus, in the strictest sense of the term, the HTA1-HTB1 locus is essential, whereas the HTA2-HTB2 locus is not. The CSE4 gene is essential (Wysocki et al. 1999) because of its critical role in kinetochore formation. The htz1 null mutant grows slowly (Jackson and Gorovsky 2000). The hho1 null mutant has no growth phenotype (Patterton et al. 1998). In conclusion, all four major core histones and centromeric H3 are essential for survival, but H2AZ and H1 are not.
Extensive studies of the regulation of the yeast histone genes have revealed that both positive and negative systems are involved. For an excellent and comprehensive review of the early work in this field, the reader should consult Osley (1991). In our laboratory, the focus is on the HTA1-HTB1 locus. We will use this locus as a guiding example in what follows.
Cell-Cycle-Dependent Activation of the Histone Genes
Spt10 binds specifically to the histone upstream activating sequence elements
Histone upstream activating sequence (UAS) elements are found in all four of the major core histone promoters (Osley 1991; Eriksson et al. 2005), but not in the HHO1, CSE4, or HTZ1 promoters (Figure 1, A and B). Each of the four divergent major histone promoters contains four UAS elements located roughly midway between the genes (Figure 1A). When transposed to a heterologous promoter, multiple copies of the histone UAS are sufficient for cell-cycle regulation with the correct timing (Osley et al. 1986). The histone UAS is specifically recognized by the Spt10 transcription factor, which has the unusual property of requiring two histone UAS elements for high-affinity binding (Eriksson et al. 2005).
SPT10 was first identified as one of a set of SPT genes, mutations in which suppress Ty1 insertion mutations (Fassler and Winston 1988). The SPT genes encode many proteins important in transcription, including subunits of the SAGA histone-modifying complex (Grant et al. 1998), TBP itself, and histones (Clark-Adams et al. 1988; Winston and Sudarsanam 1998; Yamaguchi et al. 2001). SPT10 is not an essential gene, but the null allele is associated with very slow growth and defects in gene regulation (Denis and Malvar 1990; Natsoulis et al. 1991; Prelich and Winston 1993; Yamashita 1993; Dollard et al. 1994; Natsoulis et al. 1994). Spt10 contains a histone acetyltransferase (HAT) domain similar to that of Gcn5 (Neuwald and Landsman 1997), but it has not been possible to demonstrate HAT activity, despite many attempts by our laboratory and others.
Our interest in Spt10 began when we were searching for the HAT responsible for acetylating nucleosomes at the CUP1 promoter (Shen et al. 2002). Histone acetylation of CUP1 promoter nucleosomes is low in spt10Δ cells, and CUP1 transcription is significantly reduced (Shen et al. 2002; Kuo et al. 2005). However, we were unable to show that Spt10 is present at the CUP1 promoter by chromatin immunoprecipitation (ChIP). Eventually, we turned our attention to the fact that Spt10 had been identified as an activator of the histone genes (Sherwood and Osley 1991; Dollard et al. 1994; Hess et al. 2004) and its presence at the histone promoters had been confirmed by ChIP (Hess et al. 2004), indicating that its effects on the histone genes are likely to be direct. Using genome-wide expression microarrays, we showed that the transcription of hundreds of genes is affected in spt10Δ cells, most of which showed increased expression in the mutant, implying that Spt10 acts primarily as a repressor—not the usual activating function often expected of HATs. We confirmed that Spt10 is present at the histone gene promoters, but we could not detect it at any other genes, including those strongly affected in spt10Δ cells (Eriksson et al. 2005).
We proposed that Spt10 acts directly and solely at the major core histone gene promoters, which it activates (Eriksson et al. 2005). Thus, in spt10Δ cells, a shortage of histones results in a wide range of indirect effects on a large number of genes due to a general disruption of chromatin structure. The evidence in support of our hypothesis (Eriksson et al. 2005) includes the following:
Spt10 contains a DNA-binding domain that binds specifically to the histone UAS element with the consensus sequence (G/A)TTC-N7-TTC(G/T)C (Figure 1B). Pairs of histone UAS elements are found only in the major histone promoters.
There is a general disruption of nucleosome spacing in spt10Δ cells, indicated by a smeary nucleosomal ladder after micrococcal nuclease digestion.
The endogenous 2-μm plasmid has fewer supercoils in spt10Δ cells than in wild-type cells, indicating that fewer nucleosomes are assembled in spt10Δ cells.
The severe growth defect exhibited by spt10Δ cells can be rescued completely by a high copy plasmid carrying HTA1-HTB1 and HHT1-HHF1.
Analysis of expression microarray data indicates that there is a good correlation between the genes affected in spt10Δ cells and those affected in cells depleted of H4 (Wyrick et al. 1999).
These observations amount to strong evidence in favor of our model.
We can account for the spt phenotype if it is argued that the Ty1 promoter is activated by depletion of nucleosomes. In addition, in common with mutants in a number of other chromatin-associated genes, spt10Δ cells exhibit cryptic initiation by RNA polymerase II within open reading frames, which can be explained by a more general depletion of nucleosomes, which would tend to expose TATA-like elements in coding regions (Kaplan et al. 2003).
The most serious difficulty for the model is that the levels of histone mRNA are only modestly decreased in spt10Δ cells (Hess et al. 2004; Eriksson et al. 2005). This might be because these measurements were performed in asynchronous cells; histone mRNA peaks are greatly reduced in synchronized spt10Δ cells (Xu et al. 2005). In addition, there is weak activation of some histone genes by SBF (Swi4 cell cycle box binding factor), which could be sufficient to prevent lethality (see below). We argue that spt10Δ cells grow very slowly partly because they must accumulate sufficient histone mRNA to provide sufficient histones for assembly of the replicated genome into chromatin before S phase can be completed.
The domain structure of Spt10 is shown in Figure 2A. Unlike the intact protein, the DNA-binding domain (DBD) binds specifically and with high affinity to a single UAS when expressed by itself (Figure 2B). The N-terminal domain (NTD) is required for Spt10 homodimer formation and dictates the requirement for recognition of two histone UAS elements, presumably through a conformational change in the Spt10 dimer (Mendiratta et al. 2007). The requirement for binding two histone UAS elements should result in a dramatic increase in specificity because pairs of histone UAS elements are present only in the major core histone gene promoters and such pairs occur nowhere else in the yeast genome (Eriksson et al. 2005).
The DBD contains an unusual H2C2 zinc finger, which is homologous to the N-terminal zinc-finger domain of foamy retrovirus integrase (Mendiratta et al. 2006), the enzyme responsible for insertion of retroviral DNA into the host genome. We proposed that the N-terminal domain of integrase might therefore also be a sequence-specific DNA-binding domain (Mendiratta et al. 2006). The recent crystal structure of foamy virus integrase complexed with DNA lends support to this proposal (Hare et al. 2010) because the N-terminal domain of integrase interacts with the ends of the viral DNA, which include the sequence TTC that is found in the half-sites recognized by Spt10 (Figure 1B).
Although the HAT domain has not been shown to possess HAT activity, the putative catalytic residues are required for the activation of a reporter gene (Hess et al. 2004), implying that HAT activity is critical for activation. There is indirect evidence that Spt10 might acetylate H3-K56 at the histone gene promoters (Xu et al. 2005), but it has been shown since that Rtt109 is the HAT responsible for most, if not all, H3-K56 acetylation in yeast (Driscoll et al. 2007; Han et al. 2007).
In conclusion, Spt10 is a very unusual activator in that it contains a sequence-specific DNA-binding domain and a HAT domain. Given that there is no obvious classical activation domain in Spt10, the usual arrangement in which a sequence-specific activator binds at the promoter and recruits a HAT co-activator might have been bypassed in Spt10 by fusing a HAT domain directly to a DNA-binding domain (Figure 2). Perhaps Spt10 must in turn recruit a protein with an activation domain because activation domains have other functions, such as direct or indirect recruitment of RNA polymerase II and general transcription factors. The famous herpes simplex virus VP16 protein provides an example of a recruited activation domain as it has no DNA-binding domain (O’Hare 1993).
Spt21 plays an activating role in histone gene regulation
Another SPT gene, SPT21, has been implicated as a histone gene activator (Sherwood and Osley 1991; Dollard et al. 1994; Hess et al. 2004; Hess and Winston 2005). SPT21 expression is cell cycle regulated, peaking in early S phase (Cho et al. 1998; Spellman et al. 1998). ChIP experiments show that Spt21 is present at all four major core histone promoters and that its binding peaks in S phase (Hess et al. 2004). The expression of HTA2-HTB2 is strongly reduced in spt21Δ cells and that of HHT2-HHF2 is weakly reduced, but HHT1-HHF1 and HTA1-HTB1 are only slightly affected (Dollard et al. 1994; Hess and Winston 2005). Thus, Spt21 activates HTA2-HTB2 and HHT2-HHF2. However, to complicate matters, SPT21 was identified as a repressor of an HTA1 reporter in a screen for regulators of the histone genes (Fillingham et al. 2009).
There is evidence that Spt21 stimulates Spt10 binding at all four major histone promoters during S phase, particularly at HTA2-HTB2 (Hess et al. 2004). However, in cells arrested with α-factor, Spt10 binds to the histone promoters independently of Spt21 (Hess et al. 2004), indicating that Spt21 is important only for Spt10 binding in S phase. There might be a physical interaction between Spt10 and Spt21, as they co-immunoprecipitate when co-expressed in Escherichia coli (Hess et al. 2004), although we have found that both proteins can form inclusion bodies when expressed separately in E. coli (N. M. McLaughlin, unpublished data), which might compromise the interpretation of this experiment. However, a physical interaction between Spt10 and Spt21 is detected when Spt21 is overexpressed in yeast (Hess et al. 2004).
Clearly, Spt21 is an important player in histone gene regulation, but not a critical one, since the null mutant has no growth phenotype, unlike spt10Δ. Unfortunately, Spt21 has no identifiable domains to offer clues as to its mode of action. Genetic evidence implicates both Spt21 and Spt10 in transcriptional silencing (Oki et al. 2004; Chang and Winston 2011), but this is probably an indirect effect, since neither protein could be detected at telomeres by ChIP. Intriguingly, Spt21 inhibits the formation of “T-bodies,” cytoplasmic granules containing Ty1 RNA (Malagon and Jensen 2008). This observation implies that Spt21 might have an additional function that is distinct from histone gene regulation.
SBF and the timing of histone gene activation
The major G1/S cell-cycle-dependent transcription factors are MBF (MluI cell cycle box binding factor) and SBF (Koch and Nasmyth 1994). MBF and SBF are heterodimers containing Swi6 (a regulatory factor) and a sequence-specific DNA-binding factor: Mbp1 in MBF and Swi4 in SBF (Primig et al. 1992; Koch et al. 1993). Nucleo-cytoplasmic shuttling of Swi6 is important for its regulatory function (Queralt and Igual 2003). Genome-wide ChIP studies show that SBF and MBF are present at some of the histone gene promoters, but there is some disagreement over which: Iyer et al. (2001) report that MBF binds at HTA2-HTB2 and HHT2-HHF2 and that SBF binds at HHT2-HHF2, and Simon et al. (2001) report that SBF binds at HTA1-HTB1, HTA2-HTB2, and HHO1. There are modest reductions in the expression of all of the major histone genes in swi4Δ cells and very small but significant reductions in the expression of some histone genes in mbp1Δ cells (Hess and Winston 2005).
There are two high-affinity SBF-binding sites in the HTA1-HTB1 promoter, within UAS3 and UAS4, where they overlap the Spt10 sites (Eriksson et al. 2011; Figure 1C). Spt10 and SBF binding at UAS3 and UAS4 are mutually exclusive in vitro, presumably due to steric hindrance. Reporter studies using the HTA1-HTB1 promoter and cells synchronized using α-factor indicate that both Swi4 and its binding sites in UAS3 and UAS4 are required to drive a small, early peak of expression (Figure 3). SBF and Spt10 are bound at the HTA1-HTB1 promoter but inactive in arrested cells, as measured by ChIP (Eriksson et al. 2011). After removal of the α-factor, SBF binding peaks in early S phase, coincident with the peak of SBF-mediated expression, and then declines precipitously to low levels by the end of S phase. Spt10 binding decreases throughout the same period. Both SBF and Spt10 dissociate from the promoter by the end of S phase (Eriksson et al. 2011).
SBF-driven transcription is relatively minor and mutation of the SBF-binding sites within the native HTA1-HTB1 locus to prevent Swi4 binding does not confer a growth phenotype. Spt10 and the UAS elements are required to drive the major peak of expression, which appears significantly later than the Swi4-dependent peak. Importantly, mutations that prevent Spt10 binding at the native HTA1-HTB1 locus confer a strong growth phenotype (Eriksson et al. 2011).
HTZ1, HHO1, and CSE4 all show cell-cycle-dependent expression (Pramila et al. 2006). SBF might play an important role in their regulation because all three genes possess a single predicted high-affinity SBF-binding site (Figure 1, A and C). We have confirmed the high-affinity Swi4 sites in the HHO1 and CSE4 promoters, as well as a site in the HHT2-HHF2 promoter. There are also some low-affinity sites in the histone promoters that might, or might not, be biologically significant (D. J. Clark, unpublished data).
Primary importance of the Spt10, SBF, and UAS systems
The primary importance of the UAS system in histone gene regulation is indicated by the fact that mutation of all four UAS elements to prevent the binding of both Spt10 and SBF almost completely eliminates HTA1-HTB1 transcription (Eriksson et al. 2011). It may be concluded that activation of the HTA1-HTB1 promoter is primarily due to Spt10. SBF plays a supporting role at HTA1-HTB1, where it activates a small early peak of transcription. Thus, SBF and Spt10 together control the timing of HTA1-HTB1 expression (Eriksson et al. 2011). This conclusion is consistent with the fact that swi4Δ is synthetically lethal with spt10Δ (Hess and Winston 2005).
Negative Regulation of the Histone Genes
The NEG region
Deletion of 54 bp located between UAS2 and UAS3 in the HTA1-HTB1 promoter results in derepression (Osley et al. 1986). We refer to this 54-bp region as the NEG region (Figure 1A). A search for sequence motifs within the NEG region revealed a 19-bp imperfect palindrome, named the CCR′ element. When inserted into a heterologous promoter, CCR′ confers cell-cycle-dependent repression on a reporter driven by a UAS from CYC1, but the window of expression is significantly earlier than that of the histone genes (Osley et al. 1986).
A search for motifs common to the histone promoters identified, in addition to the UAS, a motif found in the HTA1-HTB1, HHT1-HHF1, and HHT2-HHF2 promoters but not in that of HTA2-HTB2, referred to as the NEG element (Osley 1991; Marino-Ramirez et al. 2006). The NEG element in HTA1-HTB1 is located within the NEG region (Figure 1, A and D). The evidence for its role in repression is circumstantial, based on the fact that HTA2-HTB2 has no NEG element and is not subject to the same negative control system.
It has been proposed that the NEG region limits expression of the histone genes solely to S phase by repressing expression at other stages of the cell cycle (Osley et al. 1986). However, this model cannot account for the cell-cycle-dependent expression of HTA2-HTB2, which lacks the NEG region. Furthermore, the negative regulatory region in HHT1-HHF1 is not required for cell-cycle-dependent expression, whereas the UAS elements are (Freeman et al. 1992). Although much is known about the properties of the NEG region in HTA1-HTB1, crucially, the transcription factors that presumably recognize the CCR′ and NEG elements have not yet been identified.
The histone regulatory complex: a negative regulator of the histone genes
Two independent genetic screens for factors that negatively regulate the expression of the histone genes identified the histone regulatory genes (HIR1, HIR2, and HIR3) and the histone periodic control gene (HPC2) (Osley and Lycan 1987; Xu et al. 1992). The Hir1, Hir2, Hir3, and Hpc2 proteins together form the HIR complex, which acts as a histone chaperone (Green et al. 2005; Prochasson et al. 2005). Another histone chaperone, Asf1, is associated with the HIR complex (Green et al. 2005). Null mutants in components of the HIR complex do not have a growth phenotype (Sherwood et al. 1993), but asf1Δ cells grow poorly (Le et al. 1997).
Mutations in HIR genes derepress all the major histone genes except HTA2-HTB2, as shown by significant expression of the histone genes outside S phase (Osley and Lycan 1987; Xu et al. 1992). Although the levels of histone mRNA are increased in hir mutants relative to wild type, expression is still cell cycle dependent in that a peak is observed in S phase (Osley and Lycan 1987; Xu et al. 1992). In contrast, depletion of Asf1 from cells has little effect on the levels of histone transcripts during the cell cycle (Zabaronick and Tyler 2005), suggesting that the HIR complex and associated Asf1 do not have redundant functions at the histone promoters. Deletion of the NEG region also derepresses HTA1-HTB1, implying that HIR-mediated repression works through the NEG region (Osley and Lycan 1987). The HIR complex is present at all of the histone promoters except HTA2-HTB2, as determined by ChIP (Fillingham et al. 2009). Although the HIR complex binds DNA with high affinity, it does not bind specifically to the histone promoters (Prochasson et al. 2005). The HIR complex is presumably recruited to the histone promoter through interaction with an unidentified sequence-specific transcription factor, probably involving the N-terminal domain of Hpc2 (Vishnoi et al. 2011).
The HIR complex has a role in the tight coupling of histone gene expression to DNA replication. When DNA synthesis is inhibited by hydroxyurea, histone message levels are drastically reduced. This reduction is transient; the message levels are restored as cells adapt to replication stress (Libuda and Winston 2010). Histone gene expression is rendered immune to hydroxyurea-mediated repression by deletion of the NEG region, by hir mutations, and by asf1 mutations, again with the exception of HTA2-HTB2 (Hereford et al. 1981, Lycan et al. 1987; Osley and Lycan 1987; Xu et al. 1992; Spector and Osley 1993; Spector et al. 1997; Sutton et al. 2001; Ng et al. 2002).
Rather confusingly, the HIR complex is involved in diverse processes in addition to histone gene regulation, apparently acting as a histone chaperone in replication-independent nucleosome assembly (Green et al. 2005; Prochasson et al. 2005), GAL transcript elongation by RNA polymerase II (Nourani et al. 2006), heterochromatin silencing (Sharp et al. 2001), and kinetochore function (Sharp et al. 2002). Thus, the HIR complex is not a specific repressor for the histone genes. The reader is referred to Amin et al. (2012b) for an in-depth review of the HIR complex.
Mechanisms of Histone Gene Regulation
Replication-dependent histone gene expression is regulated by both positive (UAS) and negative (NEG) elements in the major core histone gene promoters. The sequence-specific factors Spt10, SBF, and perhaps MBF bind to UAS elements and act as transcriptional activators (Eriksson et al. 2005, 2011). Unfortunately, sequence-specific DNA-binding transcription factors that recognize elements in the NEG region, such as NEG and CCR′, have not yet been identified. However, a surprisingly large number of NEG-region-dependent transcription cofactors have been identified, including ATP-dependent chromatin remodelers [RSC (remodels structure of chromatin), SWI/SNF, and Yta7] and histone chaperones (HIR complex, Asf1, and Rtt106). None of these factors appears to be specific to the histone genes, but they play important roles in their activation and repression. We now discuss how these cofactors might fit into the mechanism by which the sequence-specific factors choreograph events during the cell cycle.
Transcription cofactors: chromatin remodelers and histone chaperones
RSC and SWI/SNF are similar ATP-dependent chromatin-remodeling complexes capable of sliding nucleosomes along DNA, driving conformational changes in nucleosomes, and removing histones from DNA (reviewed by Cairns 2009). However, RSC binding correlates with repression of the histone genes (Ng et al. 2002), whereas SWI/SNF binding correlates with their activation (Dimova et al. 1999; Ferreira et al. 2011) (Figure 4). Much evidence indicates that these complexes are recruited to promoters by sequence-specific transcription factors, but this paradigm is challenged by the fact that RSC contains two sequence-specific DNA-binding factors, Rsc3 and Rsc30 (Badis et al. 2008). Indeed, there are several predicted sites for Rsc3 and Rsc30 in the histone promoters, although it is not yet clear how reliable these predictions are. Therefore, RSC could be one of the sequence-specific factors binding in the NEG region, although it is also detected at HTA2-HTB2 (Ng et al. 2002), which has no NEG region.
RSC associates in a cell-cycle-dependent fashion with the histone promoters, as shown by ChIP (Ng et al. 2002) (Figure 4). RSC binding at HTA1-HTB1 is dependent on the HIR complex, but binding at HTA2-HTB2 is independent of HIR (Ng et al. 2002), suggesting that the link between the HIR complex and RSC binding is not straightforward. In contrast, the SWI/SNF complex activates histone gene transcription, but, surprisingly, its recruitment to the promoter also depends on the HIR complex (Dimova et al. 1999), indicating a mechanistic link between the activating and repressing systems. Intriguingly, the binding of both SWI/SNF and RSC is dependent on the histone chaperone Rtt106 (Ferreira et al. 2011). Thus, the two ATP-dependent remodeling complexes, RSC and SWI/SNF, have antagonistic roles at the histone promoters, presumably by mediating changes in chromatin structure during switches between the activated and repressed states. These antagonistic roles are reflected by their cell-cycle-dependent binding profiles, which are out of phase with one another, such that the SWI/SNF peak coincides with activation and the RSC peak coincides with repression (Figure 4).
Histone chaperones bind histones and facilitate nucleosome assembly and disassembly (reviewed by Ransom et al. 2010). Rtt106 is a chaperone for histones H3 and H4 (Huang et al. 2005); it binds preferentially to H3 acetylated on K56 (Li et al. 2008; Zunder et al. 2012). Rtt106 was identified as a repressor of HTA1 expression in a genetic screen, which also identified two other histone chaperones as repressors: the HIR complex and Asf1 (Fillingham et al. 2009). Thus, three different histone chaperones are involved, although physical interactions between all of them have been reported in vitro and in vivo (Sutton et al. 2001; Green et al. 2005; Fillingham et al. 2009; Ferreira et al. 2011), suggesting that they might act together, with apparently redundant functions. The HTA1-HTB1 promoter is more depleted of nucleosomes in rtt106Δ and hir1Δ cells than in wild-type cells, suggesting that these chaperones might facilitate assembly of nucleosomes on the histone promoters, which then repress transcription (Fillingham et al. 2009).
In the same screen, Fillingham et al. (2009) identified some activators and co-activators of HTA1, including Swi4, Rtt109-Vps75 (the HAT complex responsible for acetylating H3-K56), and Yta7. However, Gradolatto et al. (2008) identified Yta7 as a repressor of the histone genes, and so there is some disagreement here. Yta7 is a putative ATP-dependent remodeling protein containing a bromodomain that binds preferentially to the tail domain of H3 (Gradolatto et al. 2009). However, Yta7 has little effect on histone mRNA levels, and its activities are not confined to the histone genes: it has direct effects on many inducible genes (Lombardi et al. 2011). It has been proposed that Yta7 acts post-translationally, probably by regulating the amount of H3 in chromatin (Lombardi et al. 2011).
Mechanism of histone gene regulation during the cell cycle
The factors described above act together to regulate histone gene transcription during the cell cycle. Although we do not understand many of the molecular mechanisms underlying the events that occur during the cycle, the process of switching between activated and repressed states can be outlined.
In α-factor-arrested cells, the HTA1-HTB1 promoter is already bound by two sequence-specific transcriptional activators, Spt10 and SBF, but they are inactive. In the case of SBF, this is probably due to Whi5-mediated inactivation (Costanzo et al. 2004; De Bruin et al. 2004), which involves recruitment of the Rpd3L histone deacetylase (Takahata et al. 2009). SBF is subsequently activated through phosphorylation of Whi5 by the Cln3-Cdc28 kinase (Costanzo et al. 2004; De Bruin et al. 2004). A similar mechanism may be employed to maintain Spt10 in an inactive state.
After release from α-factor arrest, SBF activates an initial small peak of HTA1 and HTB1 transcription as the cells enter S phase. Shortly after, Spt10 activates a much larger peak of transcription (Eriksson et al. 2011). The mechanisms by which Spt10 and SBF activate transcription are not understood. SBF activity has been generally correlated with dimethylation of H3-K79 by Dot1 (Mellor 2009; Schulze et al. 2009), but how this mark facilitates activation by SBF is unclear. The HAT domain of Spt10 is important for activation (Hess et al. 2004), but what it acetylates, if anything, is unknown. Spt10 is required for the appearance of the S-phase-specific H3-K56ac mark at the HTA1-HTB1 promoter, and both Spt10 and H3-K56ac are required for the binding of SWI/SNF prior to transcription (Xu et al. 2005). The Rtt109-Vps75 HAT complex is probably responsible for all H3-K56 acetylation (Han et al. 2007), and rtt109Δ cells show reduced HTA1 transcription, suggesting that Rtt109 is indeed an important co-activator (Fillingham et al. 2009). This leads to a simple model in which Spt10 mediates recruitment of Rtt109, which then acetylates H3-K56, facilitating SWI/SNF binding (Figure 5). However, evidence that Spt10 interacts with Rtt109 is lacking. At the end of S phase, Spt10 and SBF binding at the HTA1-HTB1 promoter are reduced to low levels, accounting for the observed decline in HTA1-HTB1 transcription (Eriksson et al. 2011).
So how does the negative regulatory system fit into this picture? The CCR′ element within the NEG region confers cell-cycle dependence on a reporter driven by a constitutive UAS element, but the timing is early relative to histone gene expression (Osley et al. 1986). Thus, the reporter is repressed except during an early window. We speculate that Spt10 and SBF act maximally during this window of opportunity and are then progressively dampened by the NEG system. In the absence of the NEG system, transcription activated by Spt10 and SBF is at a higher level and maintained longer, such that the expression peak merges into that of the next cell cycle. An additional, potentially vital, role for the NEG system is its ability to mediate a rapid shutdown of histone gene transcription if cells are subjected to replication stress.
The involvement of so many chromatin-remodeling factors suggests that these events are probably accompanied by major changes in the chromatin structure of the histone genes during the cell cycle, but this remains to be confirmed. By analogy with other inducible promoters in yeast, it might be expected that the activated histone promoters are heavily depleted of nucleosomes and that the repressed promoters are packaged into nucleosomes (Figure 5). A reasonable model is that the histone chaperones (the HIR complex, Asf1, and Rtt106) assemble nucleosomes on the histone promoters toward the end of S phase (Fillingham et al. 2009) and that these nucleosomes are then positioned appropriately for repression by the RSC complex, perhaps targeted by Rsc3 or Rsc30. It has been hypothesized that, in late G1, the HIR complex is converted from a corepressor into a co-activator (Dimova et al. 1999) and that then it cooperates with Asf1 and Rtt106 to recruit the SWI/SNF complex (Dimova et al. 1999; Ferreira et al. 2011). This might involve the Rtt109-Vps75 HAT complex because Rtt109 interacts with Asf1 (Tsubota et al. 2007). In addition, HIR-mediated repression is alleviated by the activities of the mitotic Clb1 and Clb2 cyclins, probably through direct phosphorylation of the HIR complex (Amin et al. 2012a). We speculate that SWI/SNF is involved in repositioning and disassembly of repressive nucleosomes to facilitate activation by Spt10 and SBF. Clearly, a detailed description of the changes in chromatin structure that occur at the histone loci is needed to test this model.
It has been proposed that the maintenance of histone chaperone-mediated chromatin domains is the primary mechanism of cell-cycle regulation of the histone promoters (Fillingham et al. 2009). Thus, repression is mediated by Rtt106, the HIR complex, and Asf1, which cooperate to create a repressive chromatin structure at the promoter, the boundaries of which might be limited by the putative ATP-dependent chromatin-remodeling complex Yta7. Repression is alleviated through Rtt109, a HAT for H3-K56. This is a reasonable model that describes how the expected changes in chromatin structure might be effected as the histone genes cycle between on and off states. However, the additional hypothesis of Fillingham et al. (2009)—that sequence-specific activators (i.e., Spt10 and SBF) are not required for cell-cycle-dependent regulation of the histone genes—is untenable. The chromatin factors involved are all general cofactors known to be present at many diverse promoters: Where would the specificity for the histone genes come from?
Instead, we believe that the activating system is the primary source of specificity for the following reasons: (1) Only the histone UAS element confers cell-cycle-dependent regulation with the correct timing (Osley et al. 1986); (2) mutations in the negative regulatory sequence of HHT1-HHF1 do not affect cycling, whereas mutations in the UAS elements do (Freeman et al. 1992); (3) HTA2-HTB2 cycles similarly to the other major core histone loci but has no NEG system (Osley and Lycan 1987; Xu et al. 1992); (4) Spt10 is the only known sequence-specific factor dedicated to histone gene expression (Eriksson et al. 2005)—none of the known NEG system components are specific for the histone genes; (5) mutations in the UAS elements in HTA1-HTB1 that specifically prevent Spt10 binding result in poor growth (Eriksson et al. 2011); (6) there is no expression from an HTA1-HTB1 reporter carrying UAS mutations that prevent the binding of both Spt10 and SBF (Eriksson et al. 2011); and (7) the phenotypes of null mutants of NEG system components, with the exception of rsc mutants, are minor relative to that of spt10Δ, which results in very slow growth.
Post-transcriptional regulation of the histone genes is very important in more complex organisms (Marzluff et al. 2008), and this is also true of yeast. The major core histone genes are regulated post-transcriptionally (Hereford et al. 1982). The peak in histone transcript levels occurs later than that of mRNA synthesis, as measured in pulse-chase experiments. This effect is attributed to an increase in histone mRNA stability as the cells enter S phase. In contrast, inhibition of DNA replication results in a major reduction in the half-lives of the HTA1 and HTB1 mRNAs (Hereford et al. 1981), as well as shutdown of transcription through the NEG system (see above).
The mechanism by which mRNA stability is regulated is unclear, but it might involve the 3′-untranslated regions (3′-UTR) of the histone mRNAs. The 3′-UTR of HTB1 can confer cell-cycle regulation on a heterologous mRNA, although the peak of expression is late relative to native HTB1 expression (Xu et al. 1990; Campbell et al. 2002). A distal downstream element (DDE) in the 3′-UTR of HTB1 influences both the cycling of the transcript and the site of cleavage prior to polyadenylation; the DDE may bind an unidentified S-phase-specific protein believed to affect mRNA stability (Campbell et al. 2002; some of the data in this article have been retracted). It seems likely that the 3′-UTR affects mRNA stability, but it is possible that it acts at the transcriptional level through the DNA rather than the mRNA.
The TRAMP complexes play a role in histone mRNA degradation. TRAMP4 and TRAMP5, containing the poly(A) polymerases Trf4 and Trf5, recognize various mRNAs and deliver them to the nuclear exosome (Reis and Campbell 2007), where they stimulate the activity of the Rrp6 exoribonuclease (Callahan and Butler 2010). Elevated levels of core histone transcripts are present in trf4Δ, trf5Δ, and rrp6Δ cells. In addition, trf4 is synthetically lethal with rad53, asf1, and hir1 (Reis and Campbell 2007), all of which are involved in the negative regulation of histone gene expression.
Histone genes and dosage compensation
An unusual feature of HTA1-HTB1 regulation is that the mRNA level is sensitive to gene copy number: introduction of an extra copy of the HTA1-HTB1 locus results in a doubling of both the total amount of transcript and the transcript turnover rate. Consequently, the levels of HTA1 and HTB1 mRNA are unaltered (Osley and Hereford 1981). Thus, a post-transcriptional dosage compensation mechanism is active, acting through changes in mRNA turnover rates. The HTA2-HTB2 locus does not show dosage compensation: deletion of HTA1-HTB1 leads to a severe growth phenotype, but deletion of HTA2-HTB2 is compensated for by increased HTA1-HTB1 transcription (Norris and Osley 1987). HHT1-HHF1 and HHT2-HHF2 do not compensate for one another (Cross and Smith 1988; Smith and Stirling 1988).
Dosage compensation of the HTA1-HTB1 locus also involves the NEG system (Moran et al. 1990). Transcription of an HTA1 reporter gene can be repressed or activated, depending on the number of HTA and HTB genes in the cell, whereas an HTA2 reporter gene is unaffected. Furthermore, dosage compensation of HTA1-HTB1 requires histone production because a frameshift mutation introduced into HTB1 eliminates dosage compensation. This very interesting observation suggests that a feedback system might simultaneously involve the NEG system, the 3′-UTR, and histones (Moran et al. 1990).
Regulation of histone degradation
Cellular histone levels are monitored by Rad53, a major checkpoint kinase involved in the DNA damage response and required for stabilization of stalled replication forks (Gunjan and Verreault 2003; Singh et al. 2009). In rad53Δ cells, histone overexpression is lethal because excess histones are not degraded, presumably resulting in aggregation of chromatin (Gunjan and Verreault 2003). Although Rad53 can phosphorylate all four core histones in vitro, it is unclear whether it does this in vivo (Singh et al. 2010). Histones associated with Rad53 are both phosphorylated and polyubiquitylated prior to degradation by the proteasome (Singh et al. 2010). The reader is referred to the review by Gunjan et al. (2006) for further details.
Hypothesis: Negative Feedback Links the UAS and NEG Systems
The involvement of so many histone chaperones in histone gene regulation is intriguing and might have a special significance for the histone genes because histones are the end product. The NEG system could be part of a negative feedback loop that communicates the degree to which genomic nucleosome assembly is complete, and therefore when the histone genes should be shut off (Figure 6).
When nucleosome assembly is complete at the end of S phase, the chaperones responsible for delivering histones to newly replicated DNA would be expected to become fully charged with histones because there is no DNA left for nucleosome assembly. Thus, the signal to the NEG system for shutdown of the UAS system could be the extent to which chaperones bound at the histone promoters are charged with histones. Prior to S phase, the chaperones will be charged with histones from the previous cycle. When S phase is initiated, the chaperones will deposit their histones on newly replicated DNA as it becomes available. Eventually, the chaperones will be depleted of histones, thereby relieving inhibition of histone gene transcription. The UAS system would then work maximally because the chaperones are mostly histone-free. Transcription of the histone genes would eventually be shut down as the chaperones become recharged with new histones and S phase is completed. This hypothesis also accounts for the histone gene shutdown that occurs when replication forks are stalled because there would again be no DNA available for nucleosome assembly, and histone-bound chaperones would accumulate. We are currently testing this hypothesis.
We thank Rohinton Kamakaka for helpful comments on the manuscript. This work was supported by the Intramural Research Program of the National Institutes of Health (National Institute of Child Health and Human Development).
Communicating editor: J. Rine
- Received November 20, 2011.
- Accepted February 29, 2012.
- Copyright © 2012 by the Genetics Society of America