Balanced levels of histones are crucial for chromosome stability, and one major component of this control regulates histone mRNA amounts. The Saccharomyces cerevisiae poly(A) polymerases Trf4 and Trf5 are involved in a quality control mechanism that mediates polyadenylation and consequent degradation of various RNA species by the nuclear exosome. None of the known RNA targets, however, explains the fact that trf mutants have specific cell cycle defects consistent with a role in maintaining genome stability. Here, we investigate the role of Trf4/5 in regulation of histone mRNA levels. We show that loss of Trf4 and Trf5, or of Rrp6, a component of the nuclear exosome, results in elevated levels of transcripts encoding DNA replication-dependent histones. Suggesting that increased histone levels account for the phenotypes of trf mutants, we find that TRF4 shows synthetic genetic interactions with genes that negatively regulate histone levels, including RAD53. Moreover, synthetic lethality of trf4Δ rad53Δ is rescued by reducing histone levels whereas overproduction of histones is deleterious to trf's and rrp6Δ mutants. These results identify TRF4, TRF5, and RRP6 as new players in the regulation of histone mRNA levels in yeast. To our knowledge, the histone transcripts are the first mRNAs that are upregulated in Trf mutants.
IN eukaryotic cells, the DNA is packaged into nucleosomes, where each nucleosome consists of an octamer composed of two histone H2A–H2B heterodimers and a histone H3–H4 tetramer wrapped with ∼146 bp of DNA. The four histones are present in the nucleosome, and therefore within chromosomes, in equimolar stoichiometry with respect to each other and to the DNA. Cells have evolved multiple mechanisms that maintain histone abundance at very precise levels, and any disruption of these mechanisms that leads to a prolonged imbalance in the ratio of the histones to each other or to the amount of DNA leads to chromosome instability. A key feature of the control is that expression of the histone genes is tightly coupled to rates of DNA synthesis in Saccharomyces cerevisiae (Hereford et al. 1981, 1982; Osley and Hereford 1982). Histones are transcribed from four sets of gene pairs (HTA1–HTB1 and HTA2–HTB2 for H2A and H2B, and HHT1–HHF1 and HHT2–HHF2 for H3 and H4), which are divergently transcribed from the respective promoters. Transcription is activated at the G1/S transition and repressed in G1, G2, and M phases of the cell cycle (Osley and Hereford 1982; Osley and Lycan 1987; Cross and Smith 1988; Xu et al. 1992; Sutton et al. 2001). In addition, histone mRNAs are also modulated post-transcriptionally through 3′ elements of the genes (Lycan et al. 1987; Xu et al. 1990; Campbell et al. 2002). This complexity has thwarted efforts to fully understand the mechanisms underlying histone mRNA homeostasis. In this work, we describe a previously unanticipated pathway that contributes to modulation of histone mRNA levels.
Despite being initially identified biochemically as DNA polymerases (Wang et al. 2000), it is widely accepted today that TRF4 and TRF5 encode for nuclear poly(A) polymerases in budding yeast, as predicted from their primary sequence (Aravind and Koonin 1999; Saitoh et al. 2002; Haracska et al. 2005). In fact, TRF4 has also been designated PAP2 [poly(A) polymerase 2]. The structure and biochemical functions of Trf4, and of Trf5, which is 58% identical to Trf4, are conserved throughout evolution, as orthologs in Schizosaccharomyces pombe (Read et al. 2002; Saitoh et al. 2002; Win et al. 2006a,b), Caenorhabditis elegans (Wang et al. 2002a), Xenopus (Barnard et al. 2004), human, and mouse (Kwak et al. 2004) have been associated with poly(A) polymerase activity. There is also functional conservation at the physiological level, since S. pombe cid14 can complement the lethality of S. cerevisiae trf4-ts top1Δ at the restrictive temperature (Win et al. 2006a).
Although polyadenylation had generally been thought to increase the stability of eukaryotic mRNAs, current studies challenge this view. In fact, Trf4 was recently shown to be involved in polyadenylation of hypomodified forms of tRNAmet, targeting them for degradation by the nuclear exosome, and as such to participate in an unanticipated RNA quality control mechanism in yeast (Kadaba et al. 2004). RNAs encoded by intergenic regions transcribed by RNA polymerase II, as well as a number of Pol I and Pol III rRNA transcripts, pre-snRNAs, and pre-snoRNAs, are also targets of this RNA surveillance system (LaCava et al. 2005; Vanacova et al. 2005; Wyers et al. 2005; Egecioglu et al. 2006; Kadaba et al. 2006). Trf4 protein has little activity on its own but is the catalytic subunit of a poly(A) polymerase complex, termed TRAMP, that contains Air1 and/or Air2, putative RNA-binding subunits, and Mtr4, a putative helicase (LaCava et al. 2005; Vanacova et al. 2005; Wyers et al. 2005). Trf5 also has poly(A) polymerase activity and is a component of the TRAMP5 complex, which contains Trf5, Air1, Mtr4, and targets such as rRNA (Haracska et al. 2005; Egecioglu et al. 2006; Houseley and Tollervey 2006). Degradation of the Trf targets is mediated at least in part by Rrp6, a component of the nuclear exosome, whose nucleolytic activity is promoted by association with the TRAMP4/5 complexes, which, in turn, are stimulated to make their associated RNAs better substrates for the exosome (Houseley et al. 2006). However, none of the RNAs thus far identified as targets of these new poly(A) polymerases accounts for their protein/protein interactions, the genetic interactions trf mutants display, or the defects in genome stability that are the hallmark of trf4 and trf5 mutants. Both Trf4 and Trf5 are nuclear proteins and are chromatin bound (Wang et al. 2002b; Huh et al. 2003). trf4Δ is synthetic lethal with mutants lacking topoisomerase I, which retains superhelical tension within chromosomes in a viable range, with the DNA replication helicase/nuclease, dna2 (M. E. Budd, C. C. Reis and J. L. Campbell, unpublished data), and with many genes involved in chromatin dynamics and histone modification (Sadoff et al. 1995; Castano et al. 1996a; Pan et al. 2006). trf5Δ is synthetic lethal with genes involved in DNA replication, such as cdc45 and cdc8 (Tong et al. 2004). In addition, Trf5 interacts with pol ϵ, a replicative DNA polymerase, by two-hybrid assays, and both Trf4 and Trf5 interact with pol ϵ in in vitro pull-down assays (Edwards et al. 2003). trf4Δ mutants are viable at 30° but are inviable at 16° (Sadoff et al. 1995). Deletion of TRF5 alone produces no phenotype but is lethal in a trf4Δ background, and overexpression of TRF5 suppresses trf4Δ phenotypes, suggesting that the two genes perform redundant functions (Castano et al. 1996b; Walowsky et al. 1999). trf4-ts trf5Δ mutants are temperature sensitive for growth and show an altered cell cycle (Castano et al. 1996b; Wang et al. 2000). trf4Δ mutants show specific defects in DNA metabolism: hyperrecombination in the rDNA and sensitivity to DNA damaging agents, DNA replication inhibitors, and microtubule poisons (Sadoff et al. 1995; Castano et al. 1996a,b; Walowsky et al. 1999; Wang et al. 2002b; Edwards et al. 2003). The Trf mutants have been reported to have chromosome condensation and cohesion defects (Wang et al. 2000; Carson and Christman 2001; Edwards et al. 2003), although subsequent studies did not find a defect in trf4Δ cells in sister chromatid cohesion (Petronczki et al. 2004). All of these observations suggest that the Trf's may have targets that regulate stable chromosome transmission in addition to the targets identified to date, which consist of stable RNA species, such as rRNA, snRNA, snoRNA, and tRNA.
In this study, we show that inactivation of Trf4 and Trf5 poly(A) polymerases or deletion of Rrp6 nuclease of the exosome leads to abnormally high levels of mRNAs encoding the core histones of the nucleosome. Furthermore, we observe strong synthetic interactions between trf4 mutants and genes required for histone homeostasis both at the protein and at the mRNA levels. Since we also find that the trf and rrp6 mutants are hypersensitive to ectopic histone overexpression, we propose that Trf4, Trf5, and Rrp6 contribute to a previously undocumented level of histone mRNA regulation in yeast.
MATERIALS AND METHODS
Yeast strains and plasmids:
Strains and plasmids used in this study are listed in Tables 1 and 2, respectively. All strains constructed in this study are derived from W303 RAD5+ except HA-tagged versions of TRF4 and TRF5, which were derived from OAy470 (W303 MATa, bar1∷hisG rad5-535). W303 strains received from other labs were crossed at least once to W303 RAD5+ and used in this study (the RAD5 allele was determined by DNA sequencing where indicated in Table 1). Construction of Trf4 and Trf5 C-terminal HA-tagged strains and deletion mutants was performed as described (Longtine et al. 1998). Correct chromosomal integrations were confirmed by colony PCR using appropriate flanking primers. Strains carrying combinations of multiple mutant alleles were generated by genetic crosses. Manipulations and growth of S. cerevisiae were performed by standard procedures. Tetrad analysis was performed following standard techniques.
Oligonucleotides used for cloning, mutagenesis, real-time PCR, and Northern blot probes can be found in Table 3.
For the TRF4 and TRF5 expression vector construction, yeast genomic DNA was used to PCR amplify the coding regions of TRF4 and TRF5 with primers Trf4SacIF/Trf4SacIR and Trf5SpeIF/Trf5SacIR, respectively. The TRF4 amplification PCR product was purified, digested with SacI, and ligated into pRS316 vector. The TRF5 amplification PCR product was purified, digested with SpeI and SacI, and inserted downstream of the GAL1,10 promoter that had been cloned into the polylinker of pRS316.
For site-directed mutagenesis of TRF4 and TRF5, plasmids pDLT4 (Table 2) and pRS316GALTRF5 were amplified with Pfu Ultra (Stratagene, La Jolla, CA) using primers Trf4DXDF/Trf4DXDR and Trf5DXDF/Trf5DXDR, respectively. The PCR reactions were digested with DpnI for 3 hr. DpnI was heat inactivated, and 5 μl of the reactions were used to transform Escherichia coli. Correct mutagenesis was confirmed by DNA sequencing.
Cell synchrony studies:
Trf4 and Trf5 HA-tagged strains were arrested in G1, S phase, and G2/M by supplementing the media of exponential cultures with 10 μg/ml α-factor, 200 mm hydroxyurea, or 10 μg/ml nocodazole, respectively, for 2 hr. For the trf4-ts trf5Δ strain, arrest in G1 was performed by adding 10 μg/ml α-factor to exponentially growing cells at the permissive temperature of 30° for 1 hr 30 min, after which cultures were shifted to 37° and α-factor was readded to a final total concentration of 15 μg/ml. Cells were held at the restrictive temperature for 1 hr before the release from α-factor to reduce the effects of heat shock.
Approximately 5 × 106 cells were fixed in 70% ethanol and kept at −20° until further processing. Cells were washed in 1 ml of 50 mm sodium citrate buffer, pH 7.4 and resuspended in 500 μl of the same buffer. RNase A was added to a final concentration of 1 mg/ml and the reactions were incubated at 50° overnight. Subsequently, proteinase K was added to the samples to a final concentration of 1 mg/ml and incubated for 1 hr at 50°. After the incubation, 500 μl of the sodium citrate buffer and propidium iodide (to a final concentration of 16 μg/ml) were added to each sample. After this point cells were protected from light. The samples were sonicated, filtered with mesh (3-64/32 NITEX; Tetko), and analyzed by flow cytometry.
Cells were grown to saturation in medium lacking leucine (−LEU) overnight. Equivalent numbers of cells were serially diluted and plated on −LEU plates, or on freshly prepared −LEU plates containing indicated concentrations of hydroxyurea (HU) or methyl methanesulfonate (MMS), and grown for 4 days at the indicated temperatures.
Rad53 phosphorylation status was analyzed as described (Budd et al. 2006). For quantification of the cellular levels of HA-Trf4 and HA-Trf5, whole-cell extracts were prepared by alkaline lysis and separated in SDS–8% polyacrylamide gels. After transfer to nitrocellulose membrane, protein blots were probed by anti-PSTAIRE (1:10,000) and anti-HA (1:2000) antibodies. Intensity of bands was quantified by densitometry and normalized against Cdc28 (anti-PSTAIRE).
For real-time PCR total RNA was isolated and cDNA was synthesized as described (Lesur and Campbell 2004). Cultures of wild-type and trf4-ts trf5Δ cells (strain AC1968) were grown to early exponential phase at 30° and treated with 10 μg/ml α-factor for 1.5 hr. Cultures were shifted to 37°, α-factor was readded to a final total concentration of 15 μg/ml, and cells were allowed to grow at the restrictive temperature for 1 hr. Cells were then collected by centrifugation, washed, and resuspended in YPD prewarmed at 37°. Cell cycle progression was monitored by flow cytometry, and RNA extraction was carried out in cells collected 45 min and 65 min after G1 release for wild type and AC1968, respectively. Real-time PCR was performed exactly as described (Lesur and Campbell 2004) except that 12.5 pmol primers were used in each reaction and that normalization was to ACT1 mRNA. For Northern blot analysis, total RNA was extracted as described (Collart and Oliviero 1993). Poly(A)+ purification from total RNA was performed by using the MicroPoly(A)Purist kit (Ambion, Austin, TX). For agarose–formaldehyde Northern blots, 8 μg total RNA or 1 μg poly(A) RNA were dissolved in RNA sample loading buffer (Sigma, St. Louis), heated at 65° for 7 min, and separated in a 1.2% denaturing agarose–formaldehyde gel in 1× MOPS buffer pH 7.0. RNA Marker (no. G3191; Promega, Madison, WI) was used as a ruler. RNA was blotted onto a positively charged nylon membrane (no. 1417240; Roche, Indianapolis) by overnight capillary transfer in 20× SSC. The membranes were washed in 2× SSC after transfer. For acrylamide–urea Northern blots, 8 μg of total RNA were dissolved in loading buffer II (Ambion), heated at 80° for 8 min, and separated in a 6.5% polyacrylamide-bis-acrylamide (19:1) gel containing 6.5 m urea and 0.5× TBE. RNA ladder, low range (Fermentas) was run in adjacent lanes. After the run, RNA was transferred to a nylon membrane by electrotransfer in 0.5× TBE using a Trans-Blot Cell apparatus (Bio-Rad, Hercules, CA). Blots were UV crosslinked using a UV Stratalinker 2400 from Stratagene (autocrosslink at 120 mJ/cm2). Transfer efficiency and equivalence of loadings were assessed by staining the filters with 0.03% methylene blue in 0.3 m NaAC, pH 5.2. The oligonucleotide probes (Table 3) were radiolabeled by incubating [γ-32P]ATP (Amersham Pharmacia, Piscataway, NJ) and T4 polynucleotide kinase (New England Biolabs, Beverly, MA) with 25 pmol of oligonucleotide probe at 37° for 1 hr in 50-μl reactions and purified with Micro Bio-Spin 30 columns (Bio-Rad). The blots were probed using radiolabeled oligonucleotides in ULTRAhyb-Oligo hybridization buffer (Ambion) according to the manufacturer's instructions. Following hybridizations, filters were exposed to Phosphor-Imager screens and analyzed using Storm Scanner (Molecular Dynamics, Sunnyvale, CA).
RNase H reactions:
Total RNA (20 μg) was annealed with 300 ng H4hyb2 oligo or simultaneously to 300 ng H4hyb2 and 400 ng oligo(dT) (Table 3) in 25 mm Tris pH 7.5, 1 mm EDTA, 50 mm NaCl in 100 μl volume. Samples were heated at 65° for 10 min and allowed cool to room temperature. Ten microliters of 10× RNase H buffer (200 mm Tris pH 7.5, 100 mm MgCl2, 500 mm NaCl, 10 mm DTT, 300 μg/ml BSA) were added to the hybridization reaction together with 0.75 unit of RibonucleaseH (Promega). The reactions were incubated at 30° for 90 min. Five hundred microliters of 5:5:1 phenol:chloroform:LET buffer (100 mm LiCl, 20 mm EDTA, 25 mm Tris pH 8.0) were added to the reaction. After vortexing for 3 min and centrifugation for 3 min, the supernatant was recovered and precipitated.
Trf mutants display a slow S phase but are not defective in the DNA damage or replication stress checkpoints:
We wished to further understand the mechanism underlying the phenotypes of TRF4/PAP2 mutants with respect to genome stability. One clue was the fact that Trf4/Pap2 and Trf5 interact with Pol2, the catalytic subunit of pol ϵ (Edwards et al. 2003). Specifically, these interactions occur via the C-terminal domain of the Pol2 protein, which is required not only for DNA replication but also for the DNA replication stress checkpoint and for the replication of specific chromatin domains (Navas et al. 1995, 1996; Edwards et al. 2003; Iida and Araki 2004; Tackett et al. 2005). Here, we have carefully analyzed two of these functions, S-phase progression and replication checkpoint status in trf4Δ, trf5Δ, and trf4-ts trf5Δ mutants. Each trf single mutant was arrested in G1 phase with α-factor, the cells were released from the block, and the ensuing synchronous cell cycle was monitored by flow cytometry. Deletion of TRF4 leads to a delay in entry into and progression through S phase, while trf5Δ cells display a flow cytometry profile similar to wild-type cells (Figure 1A). Although the trf4Δ cells show an S-phase delay, they complete S phase and enter new cell cycles. Since TRF4 and TRF5 are thought to have at least partially redundant functions (Castano et al. 1996b), we also studied trf4-ts trf5Δ, which grows at 30° but not at 37° (Wang et al. 2000). The double-mutant cells, synchronized with α-factor, also present a significant delay in entry into S phase when released from G1 arrest at the restrictive temperature and show a clear delay in progression through S phase (Figure 1B). The delayed entry into and completion of S phase observed in trf mutants are consistent with a role for these genes in S phase. To test this idea further, we compared the steady-state levels of both Trf4 and Trf5 proteins in cells arrested in G1, S, or G2/M phases of the cell cycle. Using HA-tagged versions of Trf4 and Trf5, we found that both Trf4 and Trf5 protein expression levels peak in S-phase-arrested cells (Figure 1C). We also compared the steady-state levels of Trf4 and Trf5 protein in asynchronous cells, and, consistent with other reports (Ghaemmaghami et al. 2003), Trf4 is expressed at about four times the level of Trf5 (Figure 1D). The absence of a growth defect or slow S phase in a single trf5 mutant is perhaps explained by the lower levels of Trf5 compared to Trf4. To investigate the state of the S-phase checkpoint in the trf mutants, we monitored the level of phosphorylation of the checkpoint effector kinase Rad53. We found that trf4Δ cells proficiently phosphorylate Rad53 when exposed to MMS (Figure 2A). Thus, the trf4Δ mutant is proficient in activating the checkpoint response to exogenous DNA damage. We also asked whether the slow S-phase progression (Figure 1A) resulted from endogenous DNA damage that activated the S-phase checkpoint by monitoring Rad53 phosphorylation in the absence of exogenous DNA damage. We observed no phosphorylation of Rad53 in the trf4Δ mutant in the absence of MMS, at 30° (Figure 2A) or at the restrictive temperature of 16° (not shown), indicating that there is insufficient spontaneous damage to activate the checkpoint and therefore probably insufficient DNA damage to account for the slowing of S phase in trf4Δ mutants. Rad53 was also phosphorylated in a trf4-ts trf5Δ mutant after addition of MMS but not in the absence of MMS, even after prolonged incubation at the restrictive temperature (Figure 2B). S-phase checkpoint activation is known to slow the progression of S phase, and we also found that the trf4-ts trf5Δ mutant is competent for slowing S phase in response to HU (Figure 1B, column 4). Thus, the DNA damage checkpoint, while intact, also fails to account for the slow S-phase progression of cells with both Trf4 and Trf5 functions compromised.
TRF4 shows synthetic lethal interaction with RAD53 but not with other DNA damage checkpoint genes:
Since mutations in Trf4/5 do not activate the DNA replication stress checkpoints, we expected that Rad53 would be dispensable for the viability of Trf4/5 mutants. Surprisingly, however, when we crossed trf4Δ cells with rad53Δ cells, tetrad analysis revealed that trf4Δ rad53Δ cells are inviable (Figure 3A). (All strains in the cross carried the sml1Δ allele that is required for the viability of rad53Δ mutants.) We do not think that this is due to a spore germination defect because of results shown below (see Figure 3B and Figure 5).
We next asked whether the kinase activity of Rad53 was required for trf4Δ viability. Strain rad53K227A carries a partially but not fully defective kinase activity, since SML1 deletion or RNR1 overexpression is not required for viability. As shown in Figure 3B, strong synergistic growth defects are observed in rad53K227A trf4Δ double mutants derived from a rad53K227A, trf4Δ cross. Fewer than the expected number of double mutants were recovered, and they were extremely slow growing in comparison to either single mutant. We conclude that the Rad53 kinase activity is important for the viability of trf4Δ.
Rad53 activation/phosphorylation in response to DNA damage or to replication stress relies on the action of the two upstream kinases, Mec1 and Tel1. Rad53 phosphorylation in turn leads to the activation of the downstream protein kinase, Dun1. If the synthetic lethality of trf4Δ rad53Δ were due to failure of the DNA damage or replication stress checkpoint, one would expect the trf4Δ mutant also to require MEC1, TEL1, and DUN1 for viability and, hence, that trf4Δ mec1Δ, trf4Δ telΔ, and/or trf4Δ dun1Δ would be synthetically lethal. However, mec1Δ trf4Δ mutants are viable and do not present a negative synergistic genetic interaction at 30° (Figure 3C). Even at 37° or 16°, temperatures at which the slow growth of trf4Δ is even more evident than at 30° (not shown), double mutants grow like the trf4Δ single mutant. (sml1Δ is present in the trf4Δ mec1Δ crosses, as the lethality of mec1Δ is suppressed by inactivation of SML1.) Moreover, there is no synthetic interaction between trf4Δ and tel1Δ (Figure 3D) or between trf4Δ and dun1Δ (Figure 3E). Although some of the double trf4Δ dun1Δ mutant spores appear smaller than either single mutant, when restreaked the colonies from single and double mutants are similar. Therefore, the essential role of RAD53 in the absence of TRF4 is independent of Mec1 and Tel1 and also independent of the downstream checkpoint effector Dun1 and, thus, does not appear to be related to disruption of the DNA damage or replication stress-induced checkpoint.
Synthetic genetic interactions between TRF4 and genes required for histone mRNA regulation:
In addition to and independent of its role in responding to DNA damage, Rad53 was recently reported to monitor histone protein levels and target excess soluble histones for degradation (Gunjan and Verreault 2003). Interestingly, the role of Rad53 in histone regulation requires its kinase activity and is Mec1 and Tel1 independent, but excess histone levels do not lead to phosphorylation of Rad53 (Gunjan and Verreault 2003). The results shown above are consistent with a role for Trf's, like Rad53, in contributing to the regulation of histone levels. If so, we reasoned that trf4Δ might be synthetically lethal with asf1Δ, a histone H3/H4 chaperone important for nucleosome assembly in vivo during DNA replication, repair, and transcription (Tyler et al. 1999; Schwabish and Struhl 2006). In addition, asf1Δ cells are defective in repression of histone mRNA transcription upon S-phase arrest with HU (Sutton et al. 2001) and are sensitive to histone overexpression (Sharp et al. 2005). We find that deletion of TRF4 in combination with ASF1 deletion results in significantly reduced fitness at 30° and inviability at 37° (Figure 3F). This synthetic lethal interaction agrees with the prediction of a role for Trf4 in regulation of histone levels.
The HIR genes are transcriptional corepressors required for regulation of histone expression (Sherwood et al. 1993; Spector et al. 1997). For that reason, we looked at the effect of disruption of HIR1 on trf4Δ cells. By genetic crosses we find that trf4Δ hir1Δ cells are synthetically sick. Double mutants in this cross were recovered; however, the double mutant spores form extremely slow-growing colonies (Figure 3G).
In conclusion, TRF4 exhibits strong synthetic interactions with ASF1 and HIR1, two genes that have been previously implicated in repression of histone transcription in yeast.
Synthetic genetic interactions between TRF4 and genes involved in histone acetylation:
The synthetic lethality between TRF4 and TOP1 that led to the initial discovery of Trf4 is intriguing, but remains unexplained. The synthetic lethal interaction between TOP1 and YNG2, a nonessential subunit of the NuA4 histone acetyltransferase (HAT) complex (Choy and Kron 2002), raised the possibility that the lethality of trf4Δ top1Δ could be related to chromatin dynamics. To test this idea, we investigated genetic interactions between trf4Δ and genes involved in histone acetylation. Table 4 is a summary of the genetic interactions performed in this study. We found that trf4Δ is synthetically lethal with yng2Δ and yaf9Δ, subunits of the NuA4 HAT complex, and with gcn5Δ, the catalytic subunit if the ADA/SAGA HAT complexes. In addition, trf4Δ is synthetically lethal with a deletion of HTZ1, which codes for a replacement variant of histone H2A (see discussion). Interestingly rad53Δ, but not mec1Δ, is also synthetically lethal with htz1Δ, asf1Δ, and yaf9Δ (Pan et al. 2006). Shared synthetic lethal interactions have been interpreted to imply function in a common regulatory module, and therefore our results taken together with those on rad53Δ suggest parallel roles of Trf4 and Rad53 in histone homeostasis.
trf4-ts trf5Δ cells display abnormally high levels of mRNA coding for the core histones:
Given the involvement of the poly(A) polymerase activity of Trf's in RNA quality control mechanisms, together with the requirement of RAD53 and ASF1 for viability of trf4Δ cells, we hypothesized that Trf4/5 might be involved in a post-transcriptional mechanism, acting either directly or indirectly on histone mRNAs, for maintaining the proper balance of histone transcripts. This prompted us to examine the histone mRNA levels in trf4Δ, trf5Δ, and trf4-ts trf5Δ mutants. We first analyzed the double mutant. In light of the tight cell cycle regulation of transcription of the four core histones and the S-phase defects of the trf4-ts trf5Δ strain (Figure 1B), total RNA was collected from cells passing synchronously through S phase. Wild-type and trf4-ts trf5Δ mutant cells were released from α-factor arrest into a synchronous cell cycle at 37°, as described in materials and methods, and samples from each strain were collected at the same stage of S-phase progression, as determined by flow cytometry (Figure 4B, circled sections). Cells from the trf4-ts trf5Δ strain show significantly increased levels of HTA2, HTB2, HHT2, and HHF2 mRNAs, which encode the four core histones that compose the nucleosome (Figure 4A). The transcript encoding Htz1, the replication-independent H2A variant is, however, normal in the mutant in comparison to wild type. Importantly, other cell cycle-regulated transcripts such as POL2 or CLN2 are not significantly different in the mutant in comparison to wild type. Nor are RNR3 levels significantly elevated, adding independent evidence that the trf4-ts trf5Δ mutant is not undergoing spontaneous DNA damage and inducing the DNA damage checkpoint. We conclude that levels of the replication-dependent histone mRNA levels are specifically upregulated in trf4-ts trf5Δ cells.
We next investigated whether the increased histone mRNA steady-state levels in the trf4-ts trf5Δ strain could be confirmed by Northern blot, focusing our analysis on HHF2 mRNA. As shown in Figure 4C, this is the case. This experiment was performed with asynchronous cells, which extends our results to take into account both replication-dependent histone mRNA expression and potential accumulation outside of S phase. The extent of overexpression of HHF2 mRNA is similar to that found by real-time PCR (Figure 4A).
Since the replication-dependent histones are polyadenylated in yeast, we also asked if the excess RNA was represented in the poly(A)+ RNA fraction. As shown in Figure 4D, analysis of the same RNA samples used in Figure 4, A and B, revealed significant accumulation of poly(A)+ RNA in the mutant compared to wild type. Since Trf4 and Trf5 are either defective or missing, this suggests that Pap1, the conventional poly(A) polymerase, may polyadenylate histone mRNAs. Indeed, we find that thermal inactivation of Pap1 in a pap1-1 mutant leads to a reduction in the amount of HHF2 transcript, with a fraction of HHF2 mRNA migrating faster (Figure 4E, arrows). Thus, the histone mRNAs are targets of Pap1. There is clearly residual polyadenylated HHF2 in the pap1-1 strain (Figure 4E), as verified by enrichment on oligo(dT) cellulose, but deleting TRF4 or TRF5 in the pap1-1 mutant did not reduce the amount of residual polyadenylated HHF2 (data not shown).
Finally, we show that upregulation of histone mRNA levels is specifically attributable to the mutations in TRF4/5, since introduction of TRF4 in a single-copy plasmid into trf4-ts trf5Δ cells reduces the HHF2 mRNA levels back to wild-type amounts (Figure 4F).
We also examined the level of HHF2 RNA in the trf4Δ and trf5Δ single mutants, respectively. Quantitative PCR, performed on exponentially growing trf4Δ cells, shows no significant alteration in the levels of any of the transcripts coding for the four core histones when compared to wild type (Figure 4G). Lack of histone mRNA overexpression was confirmed for trf4Δ and extended to trf5Δ by Northern blotting (Figure 4H). Note that in our hands, quantification of acrylamide–urea Northern blots tends to overestimate the difference in the levels of the RNA species analyzed when compared to agarose–formaldehyde gels. For instance, compare Figure 4, C and D, with Figure 4F in which the extent of HHF2 mRNA overexpression in the trf4-ts trf5Δ strain is 2.4-fold in agarose Northern blots—an estimation in the range of the one obtained by real-time PCR (Figure 4A)—whereas the value rises to 6.8-fold in acrylamide Northern analysis.
Our observations suggest that Trf4 and Trf5 have a redundant role in histone regulation and that each can compensate for the loss of the other.
Deletion of HHT2-HHF2 suppresses the synthetic lethality between TRF4 and RAD53 and histone overexpression is toxic to trf mutants:
Given the multiple functions of Rad53 in yeast cells and the lack of previously reported protein-encoding mRNA targets for Trf4, we sought further evidence that the synthetic lethality of trf4Δ rad53Δ was related to the observed histone mRNA expression and presumed histone imbalance. We asked if the synthetic lethality between rad53Δ and trf4Δ could be suppressed by lowering the histone dosage in the cell. We crossed a double mutant rad53Δ hht2-hhf2Δ with trf4Δ. As observed above, rad53Δ trf4Δ spores either were missing or formed extremely sick spores. On the contrary, all the expected rad53Δ trf4Δ hht2-hhf2Δ spores were recovered and formed viable colonies (Figure 5A), supporting the proposal that the inviability of rad53Δ trf4Δ is due to the combined failure of two pathways to control the cellular levels of histone mRNA and protein, one governed by Trf4 and the other by Rad53.
Conversely, we investigated the effect of histone overexpression in trf mutants. Delivery of the four core histones in a multicopy vector under the control of a constitutive promoter results in balanced overexpression of the four core histones and does not affect wild-type growth (see below). However, as shown in Figure 5B, constitutive histone overexpression exacerbates the ts phenotype of the tr4-ts trf5Δ mutant. In addition, trf4Δ cells become hypersensitive to MMS and HU when histones are overexpressed (Figure 5C). Thus, excess of histones may explain, at least in part, the sensitivity of trf4Δ to exogenous DNA damage and replication stress. Ectopic histone overexpression has no effect on the growth of mutants deficient in the classical poly(A) polymerase, pap1-1 (not shown).
Because the trf4Δ mutant is synthetically lethal with yng2Δ, gcn5Δ (this study), and yaf9Δ (Pan et al. 2006), we decided to evaluate the effect of histone overexpression in these mutants. As shown in Figure 5D, histone overexpression has a strong deleterious effect in each strain. Conversely, we do not detect a reduction in the amount of acetylated bulk histone H4 in the trf4Δ or the trf4-ts trf5Δ strains (data not shown). Altogether, these results suggest that deregulation of histones underlies the synthetic lethality between TRF4 and YNG2/YAF9/GCN5.
rrp6Δ cells, but not air1/2Δ, have increased histone mRNA levels and are sensitive to histone overexpression:
We reasoned that if Trf's are required to prevent excess transcripts coding for the core histones, then mutants in other members of the TRAMP complex or in the Rrp6 exoribonuclease subunit of the exosome should also display unusually high levels of histone mRNAs and hypersensitivity to histone overexpression. When we analyzed the effect of deleting two putative RNA-binding subunits of TRAMP and TRAMP5 complexes, AIR2 and AIR1, respectively, we found that air1Δ and air2Δ single mutants, as well as air1Δ air2Δ double mutants, show wild-type levels of transcript encoding HHF2 (Figure 6). Consistent with this, overexpression of histones has no deleterious effect on the growth of air1Δ, air2Δ, or air1Δ air2 strains even when these mutants are challenged with exogenous DNA damage (not shown). Therefore, Trf4 and Trf5 can maintain appropriate histone mRNA levels in the absence of Air1 and Air2, suggesting that there may be other RNA-binding proteins besides Air1 and Air2 and that mediate Trf interaction with specific substrates that lead to histone mRNA upregulation. Evidence for this is presented in supplemental Figure 1 at http://www.genetics.org/supplemental/.
We next investigated the effect of deleting RRP6 in histone homeostasis. We found that the rrp6Δ mutant is hypersensitive to histone overexpression (Figure 7A). Furthermore, deletion of RRP6 leads to markedly increased levels of total histone HHF2, resembling trf4-ts trf5Δ cells (Figure 7B). In our strain background, we also found that trf4Δ is synthetic lethal with rrp6Δ (Table 4), a result that prevented the analysis of rrp6Δ trf4Δ cells with respect to histone mRNA levels. We conclude that Rrp6, and hence the nuclear exosome, is involved in histone mRNA regulation.
Histone mRNA levels are very low in G1, G2, and M phases of the cell cycle, but histone transcription is derepressed and activated in S phase. We next assessed whether the rrp6Δ mutation affects this temporal pattern of expression. As shown in Figure 7, C and D, when rrp6Δ cells were synchronized with α-factor, histone HHF2 mRNA levels fluctuated in the cell cycle with approximately the same timing as in wild type, taking into account the slower cell cycle progression of the rrp6Δ cells. Nevertheless, there is a dramatic increase in the extent of accumulation of HHF2 mRNA in S phase (as illustrated in Figure 7E). Since some HHF2 mRNA is detectable in G1, G2, and M, it is not clear to what extent repression of histone mRNAs at those points in the cell cycle is affected by the rrp6Δ mutation; however, it is clear that negative regulation is not abolished. Therefore, RRP6 deletion does not disrupt the cell cycle-dependent oscillation of histone levels and leads to an S-phase-specific increase, compared to wild type, in the steady-state levels of histone mRNA.
We next investigated the temporal pattern of HHF2 mRNA expression in trf4-ts trf5Δ, as shown in Figure 7, F–H. We observed that the increase in accumulation of HHF2 mRNA, as in the rrp6Δ strain, is pronounced in S phase and that negative regulation is mostly intact in G1, G2, and M phases. Higher levels of HHF2 mRNA are observed outside of S phase in the trf4-ts trf5Δ mutant when compared to wild type or rrp6Δ. However, the slightly elevated levels of expression seen outside of S phase in the trf4-ts trf5Δ may be related to inability to fully synchronize this mutant with α-factor and resulting asynchrony, or there may be some derepression in trf4-ts trf5Δ cells. Thus, it appears that it is the DNA replication-related histone mRNA accumulation that is deregulated in the trf and rrp6Δ mutants. This phenotype is very similar to that observed with asf1Δ or with hir1Δ mutants (Sutton et al. 2001).
Analysis of poly(A) tail length of HHF2 mRNA in trf's and rrp6 mutants:
One possible substrate for polyadenylation in the Trf/Rrp6-mediated regulatory system is the histone mRNAs themselves, although, to date, no mRNA transcripts have been identified as Trf substrates. We therefore inquired about the length of the HHF2 poly(A) tail in different mutant backgrounds to determine if histone mRNAs are directly polyadenylated by Trf4 and Trf5. To analyze poly(A) lengths at high resolution, RNase H-directed cleavage was used. A DNA oligonucleotide complementary to the terminal portion of the HHF2 transcript 214 nucleotides upstream of the first major transcription termination site was annealed to total RNA extracted from wild-type and tr4-ts trf5Δ cells. Following RNase H digestion we estimated the lengths of the poly(A) tail by Northern blot by comparison to samples hybridized to oligo(dT) before RNase H treatment to remove poly(A) tails. As shown in Figure 8A, HHF2 transcripts carry heterogeneous poly(A) tails that extend up to ∼70 bp. We found no significant increase in the length of the polyadenylated HHF2 mRNA population derived from the wild-type or rrp6Δ strains upon TRF4 (Figure 8A) or TRF5 overproduction (not shown) or decrease in HHF2 mRNA length in the tr4-ts trf5Δ strain (Figure 8B). Perhaps the presence of Pap1 in the strains, instability of the RNAs polyadenylated by Trf polymerases, or the multiple transcription termination sites render any changes in poly(A) tail length undetectable (see discussion).
Tight regulation of histone relative stoichiometry and overall levels is fundamental to the preservation of genome integrity in all eukaryotes. Abnormal histone levels induce defects in mitotic chromosome segregation, chromatin structure, and transcription and lead to loss of viability (Meeks-Wagner and Hartwell 1986; Han et al. 1987; Clark-Adams et al. 1988; Kim et al. 1988; Norris et al. 1988). Defects in chromatin structure caused by inactivation of nucleosome assembly factors cause high rates of chromosomal rearrangements and spontaneous DNA damage and elicit checkpoint activation (Myung et al. 2003; Ye et al. 2003; Ramey et al. 2004).
This study provides genetic and biochemical evidence that Trf4 and Trf5 make a redundant contribution to genome stability in yeast through control of histone mRNA levels during S phase. We show that the mRNAs coding for the four core histones, but not other cell cycle-regulated transcripts tested, accumulate to abnormally high levels in S phase in a trf4-ts trf5Δ mutant whereas deletion of each TRF gene per se does not result in a significant increase in histone mRNAs. So far, no other true mRNA products of RNA Pol II transcription have been found to be upregulated in Trf mutants. That this overexpression has a physiological impact is supported by the fact that additional ectopic overproduction of histones decreases the viability and increases the DNA damage sensitivity of trf mutants with no effect in the wild-type isogenic strain. Moreover, trf4Δ cells require RAD53 for viability. This supports the idea that Trf4 and Trf5, like Rad53, are involved in regulation of histone amounts and that histone abundance reaches lethal levels in the absence of both pathways. Although trf4Δ and trf4-ts trf5Δ mutants exhibit a slow S phase, the synthetic lethal interaction between TRF4 and RAD53 is unlikely to result from DNA damage due to slow or aberrant DNA synthesis, since induction of the DNA damage response is not observed in the trf mutants. This is true even in the trf4-ts trf5Δ double mutant at the restrictive temperature, even though the checkpoint is intact and can be activated in the presence of exogenous DNA-damaging agents. Supporting this, no synthetic interactions are observed between TRF4 and the master checkpoint kinases, MEC1 and TEL1, indicating that a function specifically carried out by Rad53 and independent of the general DNA damage checkpoint activation pathway is required for trf4Δ viability. Finally, trf4Δ rad53Δ lethality is suppressed by lowering histone dosage in the cell, demonstrating that excess histones underlie trf4Δ rad53Δ inviability. We therefore propose that the slow S phase and related defects in chromosome maintenance in trf mutants are related, at least partly, to defects in regulation of the S-phase histone mRNAs.
The effects on histone mRNA accumulation in trf4-ts trf5Δ are furthermore likely attributable to the established role of the interplay between Trf4/Trf5 and the nuclear exosome in RNA degradation. An attractive feature of this model is that it provides an explanation to why loss of Rrp6 leads to the same histone mRNA phenotypes as trf4-ts trf5Δ. In as much as Trf4 and Trf5 are thought to be exclusively nuclear and Rrp6 is found only in the nuclear exosome and not in the cytoplasmic, histone mRNA regulation may constitute an essential function of the nuclear exosome. The rrp6 null mutant is temperature sensitive for growth at 37°, has slow growth at 30°, and, like the trf mutants, shows benomyl and MMS sensitivity (Begley et al. 2002; Daniel et al. 2006). A recent study found that a considerable number of mRNAs are upregulated in rrp6Δ cells (Houalla et al. 2006). However, an increase in histone mRNA levels upon inactivation of Rrp6 was not detected. Interestingly, the effect of a deletion of RRP6 on other transcripts analyzed in this study was reduced at 37° (Houalla et al. 2006), in complete agreement with our observations for HHF2 mRNA levels, whose increase in rrp6Δ cells is less elevated at higher temperatures (data not shown). Air1 and Air2, RNA-binding components of TRAMP4 and TRAMP5 complexes, are not required for the maintenance of normal histone mRNA levels (Figure 6 and supplemental Figure 1 at http://www.genetics.org/supplemental/), suggesting that other RNA-binding subunit/s might be involved in the regulation of histone transcripts, even if indirectly, by Trf's and the exosome. Interestingly, the RNA-binding protein Nrd1p was recently reported to associate with and stimulate the nuclear exosome for 3′-end processing of RNA polymerase II transcripts. Nrd1, which also interacts with the TRAMP complex, has been proposed to assist the exosome in identifying certain mRNAs as targets for degradation (Gavin et al. 2006; Vasiljeva and Buratowski 2006), indicating that other RNA-binding proteins, besides Air1 and Air2, might modulate Trf4 and Trf5 action. Our genetic studies on the toxic effect of TRF5 overexpression on air1Δ air2Δ double mutants (supplemental Figure S1 at http://www.genetics.org/supplemental/) also support this hypothesis.
While the simplest model to explain our results is that histone mRNAs are substrates of Trf4- and Trf5-mediated degradation, to establish this has been more difficult than it has been for other Trf4/5 targets studied to date. There may be several reasons for this. First, the generally rapid turnover of mRNAs relative to the stable rRNA, tRNA, snRNA, and snoRNA species may pose an obstacle to the detection of mRNA degradation intermediates. Second, there are at least two major transcription termination sites for HHF2 [Figure 8, A and B, oligo(dT) lanes], as reported by others (Cross and Smith 1988). Third, HHF2 RNA is a substrate of the classical poly(A) polymerase, Pap1, which adds heterogeneous length poly(A) tails (Figure 4E). In addition to the heterogeneity of histone mRNAs making it difficult to determine whether Trf4 and/or Trf5 polyadenylate histone mRNAs, it is also possible that the instability or low abundance of histone species polyadenylated by Trf's precludes their detection. At this point we still cannot discard the possibility that the inability to see effects of inactivation or overproduction of TRF's on histone mRNA length, even in rrp6Δ strains, is due to the heterogeneity, instability, or low abundance of the mRNAs, putatively polyadenylated by Trf4/5. It is possible that the histones and perhaps other mRNA levels are regulated through a competition between stabilization and degradation involving multiple Paps.
Genetic analysis reported here (Table 4) and additional findings reported elsewhere (Pan et al. 2006) support the existence of a module of interconnected pathways that regulate histone homeostasis and thereby ensure effective chromatin assembly (see Figure 9 for schematic). We show that the trf mutants exhibit synthetic fitness defects with asf1Δ, hir1Δ, and rad53Δ, each of which causes elevated histone levels. Given the wide spectrum of functions assigned to Asf1 and the Hir1 complex (Osley and Hereford 1982; Osley and Lycan 1987; Xu et al. 1992; Sharp et al. 2001, 2005; Sutton et al. 2001; Gunjan and Verreault 2003; Adkins and Tyler 2004; Groth et al. 2005; Recht et al. 2006; Schwabish and Struhl 2006), it is hard to stipulate as to the exact reason why trf4Δ asf1Δ or trf4Δ hir1Δ are synthetically lethal, but the synthetic effects may be due to the shared roles of Asf1 and Hir1 in regulating transcription of the histone mRNAs. Regulation of histone mRNAs by Trf's could operate either at the level of histone transcription or at the level of the stability of histone mRNAs. Negative controls operate on the histone genes during inhibition of DNA replication and in G1, G2, and M phases of the cell cycle, and positive controls activate transcription in S phase. Although the precise mechanism of histone transcriptional regulation in yeast is unknown, the histone promoters (with the exception of HTA2–HTB2) contain well-characterized cis-acting positive and negative regulatory promoter elements (Osley et al. 1986; Osley 1991; Freeman et al. 1992). The Hir1 complex and Asf1 contribute to negative regulation through these sequences (Osley and Lycan 1987; Cross and Smith 1988; Moran et al. 1990; Xu et al. 1992; Sherwood et al. 1993; Dollard et al. 1994; Compagnone-Post and Osley 1996; Spector et al. 1997; Sutton et al. 2001). Since we demonstrate that HTA2 and HTB2 transcripts are upregulated in trf4-ts trf5Δ cells, we do not favor the idea that Trf's are involved in the Hir1 or Asf1 transcriptional repression pathway because the promoter of the HTA2–HTB2 locus lacks the negative element required for this regulation (Osley and Lycan 1987; Sutton et al. 2001). More likely, they participate in a parallel pathway, such as post-transcriptional regulation (Lycan et al. 1987; Xu et al. 1990). The importance of post-transcriptional regulation of histone mRNAs is underscored by the fact that even when expressed from a constitutive promoter, full-length HTB1 mRNA is periodically regulated, indicating that regulation at the post-transcriptional level is sufficient to confer cyclic oscillation of histone mRNA (Lycan et al. 1987; Xu et al. 1990). Genes involved in this mechanism in yeast have not been identified, although 3′-end sequences of HTB1 required for this level of regulation have been mapped (Campbell et al. 2002).
Rad53, a key effector kinase of the DNA damage checkpoint, mediates the degradation of excess histones that are not packaged into chromatin in a mechanism that is independent of the role of Rad53 in the DNA damage response pathway, and this role appears to account for the slow growth of rad53 mutants (Gunjan and Verreault 2003). Our results strongly indicate that the trf4Δ rad53Δ synthetic lethality occurs due to excess histones and is independent of Rad53's role in the DNA damage checkpoint. The Rad53 pathway is clearly nonredundant with Trf4, given the fact that histone overexpression is deleterious to trf mutants even in RAD53 cells, suggesting that Rad53 alone cannot guarantee appropriate histone turnover when transcript levels are misregulated.
An additional aspect of our genetic analysis worth comment is the synthetic lethality between the trf genes and components of the histone acetyltransferase complexes (Table 4) and the deleterious effects of overexpressing plasmid-borne histone genes in mutants in the HAT complexes (Figure 5). Histones are acetylated shortly after their synthesis, and acetylated histones are found in association with histone chaperones, such as Asf1, which aids CAF-1 in chromatin assembly during DNA synthesis, and the Hir1 complex, which in yeast appears to participate in assembly of silent chromatin (Polo and Almouzni 2006). Histone acetylation by the so-called B-type HATs plays a crucial role in the assembly of newly synthesized histones onto chromatin. B-type HATs are defined as being cytoplasmic and acetylating free histones prior to their incorporation into chromatin. A-type HATs are nuclear and act upon histones that are already assembled in chromatin (Brownell and Allis 1996). However, this strict definition does not always apply since Hat1, the only B-type HAT identified in yeast to date, can be found in the nucleus (Parthun et al. 1996), and deletion of HAT1 does not provoke any phenotype (Kleff et al. 1995). Thus, functionally redundant B-type HATs may exist. Our results regarding the toxicity of histone overexpression in mutants of components of NuA3 and NuA4 complexes raise the possibility that these two HAT complexes can also acetylate free histones prior to their incorporation into chromatin in addition to their established function as A-type HATs that are thought to be involved in maturing previously assembled nucleosomes. This could account for our demonstration of the toxic effect of histone overexpression in these mutants as well as the synthetic interaction between them and TRF4 and RAD53 (this study and Pan et al. 2006). An alternative explanation might be that ectopic histone overexpression, as well as the loss of control of histone levels that occurs in rad53 and trf mutants, titrates out these HAT complexes, blocking their action on the chromatin histones. However, both possibilities imply that NuA3 and NuA4 complexes are able to acetylate free histones, which has not been reported to date. The synthetic lethality of trf4 with the HAT complexes (Table 4) suggests a parallel function. Synthetic lethality with htz1 may have a similar explanation. We propose that these synthetic interactions are most likely due to additive defects in a pathway ultimately affecting chromatin assembly.
The elaboration of so many parallel levels of histone regulation uncovered in this and previous studies is not surprising given that rapid deposition of histones following passage of the replication fork is essential in yeast (Han et al. 1987). Histones must be synthesized and become rapidly available but they cannot be maintained in excess because of deleterious interactions with DNA and with the multiple histone modifying enzymes that regulate genomic transactions. Any departure from normal levels of appropriately modified histones leads to faulty chromatin assembly and ensuing genomic instability.
Given the high functional conservation of Trf4 and Trf5 across species it is very likely that human orthologs are endowed with the same cellular function as the yeast proteins. In metazoans, where regulation of histones at the post-transcriptional level is understood in comparatively more detail, the mRNAs encoding for the replication-dependent histones lack a poly(A) tail, ending instead in a conserved stem-loop structure located immediately upstream of the 3′ end of the mRNA. In mammalian cells, post-transcriptional regulation of histone mRNAs accounts for the main mechanism that restricts histone synthesis to S phase. Upon inhibition of DNA replication, histone mRNAs are rapidly degraded by a process that appears to start at the 3′ end of the transcripts (Ross et al. 1986) and that involves an exonuclease, the 3′hExo, that binds to the stem loop (Dominski et al. 2003; Marzluff 2005). The structural disparity between the 3′ ends of histones and polyadenylated mRNAs in metazoans suggests that singular 3′-end post-transcriptional mechanisms aimed to regulate histone mRNA processing take place. Recent studies, however, suggest that factors required for 3′-end processing are shared by histone and polyadenylated mRNAs. In fact, CPSF-73, a cleavage factor required for the formation of polyadenylated mRNAs, has been shown to be involved in histone mRNA 3′-end processing as well (Dominski et al. 2005). Also, symplekin, along with various components of the cleavage and polyadenylation machinery of polyadenylated mRNAs, has been shown to participate in 3′-end maturation of histone mRNAs in mammalian cells (Kolev and Steitz 2005). Interestingly, both symplekin and xGLD-2 poly(A) polymerase, the homolog of Trf's in Xenopus, were found to be essential for cytoplasmic polyadenylation during oocyte maturation (Barnard et al. 2004). It will be interesting to investigate the role of Trf homologs/isoforms in higher eukaryotes and a possible function in histone homeostasis.
We thank Rochelle Diamond for assistance with flow cytometry; J. F. Diffley for the Rad53 antibody; M. Foiani for the rad53K227A strain; M. F. Christman for the CY1243 strain; and R. Rothstein for the W2017-7C, U960-5C, and U953-61A strains. We thank R. Parker for helpful discussions and for supplying the RNaseH protocol. C.C.R. is supported by a predoctoral fellowship from Fundacao para a Ciencia e Tecnologia (SFRH/BD/9612/2002), Portugal. This work was supported by National Institutes of Health grant GM25508 and by a grant from the Research Management Group.
Communicating editor: P. Russell
- Received September 21, 2006.
- Accepted December 6, 2006.
- Copyright © 2007 by the Genetics Society of America