Multiple Roles for Saccharomyces cerevisiae Histone H2A in Telomere Position Effect, Spt Phenotypes and Double-Strand-Break Repair

Corresponding author: Arthur J. Lustig, Tulane University Health Sciences Center, 1430 Tulane Ave., New Orleans, LA 70112., alustig{at}tulane.edu (E-mail)

Communicating editor: F. WINSTON

	ABSTRACT

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Telomere position effects on transcription (TPE, or telomeric silencing) are nucleated by association of nonhistone silencing factors with the telomere and propagated in subtelomeric regions through association of silencing factors with the specifically modified histones H3 and H4. However, the function of histone H2A in TPE is unknown. We found that deletion of either the amino or the carboxyltails of H2A substantially reduces TPE. We identified four H2A modification sites necessary for wild-type efficiency of TPE. These "hta1tpe" alleles also act as suppressors of a insertion allele of LYS2, suggesting shared elements of chromatin structure at both loci. Interestingly, we observed combinatorial effects of allele pairs, suggesting both interdependent acetylation and deacetylation events in the amino-terminal tail and a regulatory circuit between multiple phosphorylated residues in the carboxyl-terminal tail. Decreases in silencing and viability are observed in most hta1tpe alleles after treatment with low and high concentrations, respectively, of bleomycin, which forms double-strand breaks (DSBs). In the absence of the DSB and telomere-binding protein yKu70, the bleomycin sensitivity of hta1tpe alleles is further enhanced. We also provide data suggesting the presence of a yKu-dependent histone H2A function in TPE. These data indicate that the amino- and carboxyl-terminal tails of H2A are essential for wild-type levels of yKu-mediated TPE and DSB repair.

GENES positioned adjacent to a telomere are epigenetically repressed, or silenced, a phenomenon known as telomere position effect (TPE; GOTTSCHLING et al. 1990 ). TPE has been observed in organisms ranging from yeast to humans (BAUR et al. 2001 ). In yeast, TPE is a metastable process in which repressed and derepressed states are each maintained for multiple generations (GOTTSCHLING et al. 1990 ; MONSON et al. 1997 ; ENOMOTO and BERMAN 1998 ; PARK and LUSTIG 2000 ). Under these conditions, dense, specifically positioned nucleosomes are coupled with multiple associations among nonhistone regulatory proteins in subtelomeric chromatin to form a closed structure, similar to the transcriptionally quiescent heterochromatin of higher eukaryotes (GOTTSCHLING et al. 1990 ; GOTTSCHLING 1992 ; WRIGHT et al. 1992 ; VENDITTI et al. 1999 ).

The genetic regulation of telomeric silencing bears striking similarities to silencing of the cryptic mating-type loci located close to the left (HML) and right (HMRa) telomeres of chromosome III (LUSTIG 1998 ). The telomere contains a subset of factors and cis-acting sequences that act at the HM loci. Silencing is initiated within the telomere by the recruitment of the nonhistone chromatin proteins Sir2, Sir3, and Sir4 by the yKu70/yKu80 heterodimer and Rap1, the major telomeric binding protein that is highly conserved through evolution (COCKELL et al. 1998A ; GASSER and COCKELL 2001 ; LUO et al. 2002 ). This recruitment is coupled with associations among Sir factors, including Sir3/Sir4, Sir3/Sir3, and Sir2/Sir4, that extend into subtelomeric regions (LAROCHE et al. 1998 ; GASSER and COCKELL 2001 ; MOAZED 2001 ; HOPPE et al. 2002 ; LUO et al. 2002 ). In addition, the Sir proteins associate with additional nonhistone proteins, including Dot4 (KAHANA and GOTTSCHLING 1999 ), Sif1 (COCKELL et al. 1998B ), and Ubp3 (MOAZED and JOHNSON 1996 ).

Subsequent unidirectional spreading of the subtelomeric heterochromatic state (RENAULD et al. 1993 ) requires a multiplicity of proteins, including the NAD-dependent histone deacetylase Sir2. This deacetylase targets specific lysine residues within both histone H4 (K16) and histone H3 (K14) (IMAI et al. 2000 ; LANDRY et al. 2000 ; MOAZED 2001 ; SUKA et al. 2001 ; CARMEN et al. 2002 ). In addition, interactions of both Sir3 and Sir4 with histones H3 and H4 are required for telomeric silencing (HECHT et al. 1995 ; LUSTIG 1998 ; CARMEN et al. 2002 ). As in other organisms, recent evidence has also implicated Set1-mediated histone H3 methylation activity in yeast heterochromatin formation (BRIGGS et al. 2001 ; LACHNER et al. 2001 ; KROGAN et al. 2002 ).

The formation of highly condensed "silenced" chromatin states also involves the interaction of telomeric and subtelomeric bound Rap1 and Sir factors. This interaction may be mediated through a "hairpin" looped structure or less-defined interactions within telomeric sequences (GRUNSTEIN 1998 ; DE BRUIN et al. 2000 , DE BRUIN et al. 2001 ; PARK and LUSTIG 2000 ).

Previous studies have indicated that the N-terminal tail of histone H2A is required for repression of basal transcription (LENFANT et al. 1996 ; RECHT et al. 1996 ). In addition, truncation of the amino terminus results in an inability to grow on different carbon sources such as raffinose and galactose (HIRSCHHORN et al. 1995 ). In contrast, phosphorylation within the carboxyl-terminal region of H2A has been linked, among other processes, to replication, DNA damage, and histone deposition (GREEN and POCCIA 1989 ; DOWNS et al. 2000 ; GREEN 2001 ; WARD and CHEN 2001 ; REDON et al. 2002 ). While an extensive body of data supports the involvement of the amino termini of histones H3 and H4 in telomeric silencing, the role of H2A in telomere position effects has remained uncertain (GRUNSTEIN 1998 ).

Like other histones, H2A has a long, positively charged amino terminus (operationally defined as between amino acids 1–20) that extends outside the nucleosome space (LUGER and RICHMOND 1998 ). However, histone H2A is unique in that its carboxyl-terminal region (defined here as residues 120–131) also extends into the extranucleosomal space (LUGER et al. 1997 ; LUGER and RICHMOND 1998 ). This latter region exits the nucleosome between DNA gyres as they wrap around the histone octamer, near the dyad axis, leaving the carboxyl-terminal tails poised to influence access of nucleosomal DNA to regulatory factors. Both extended termini contain multiple potentially modified residues that could regulate the heterochromatin-like subtelomeric structure, possibly by mediating interactions with specific silencing proteins.

Mutations or reductions in the abundance of H2A and H2B confer suppression of auxotrophy caused by insertion of , the Ty1 long terminal repeat, into either upstream activation or 5' coding regions of genes such as HIS4 and LYS2, termed an Spt^- (suppressor of Ty1) phenotype. (CLARK-ADAMS and WINSTON 1987 ; CLARK-ADAMS et al. 1988 ; SWANSON et al. 1991 ). Wild-type and lys2-128 cells initiate transcription at the LYS2 promoter; however, the lys2-128 allele creates an early stop codon that results in a short inactive protein. In contrast, in spt mutant cells, the promoter is functional, leading to a slightly shorter, but active, gene product (SWANSON et al. 1991 ). These Spt^- phenotypes may be the consequence of alterations in the transcriptional initiation (or elongation; see below) mediated through alterations in chromatin structure (HARTZOG et al. 1998 ; YAMAGUCHI et al. 2001 ). Consistent with this, spt11 and spt12 are alleles of HTA1 and HTB1, respectively (CLARK-ADAMS et al. 1988 ).

Mutations in SPT4, SPT5, and SPT6 interact genetically with the amino-terminal tails of histone H2A and H2B fusion proteins to confer transcriptional repression when either histone is tethered adjacent to a reporter gene (RECHT et al. 1996 ). Furthermore, overproduction of H2A and H2B can suppress the phenotypes of mutations in several SPT genes (SHERWOOD and OSLEY 1991 ). The stage at which these gene products act in transcription is not fully resolved, although more recent evidence suggests a role for these proteins in the elongation of transcripts through specific chromatin states (HARTZOG et al. 1998 ; YAMAGUCHI et al. 2001 ; POKHOLOK et al. 2002 ). This view is consistent with studies from Gross' lab suggesting that silencing is likely to be the consequence of defects in Pol II elongation rather than in accessibility of the template (SEKINGER and GROSS 2001 ).

In this study, we investigated the role of histone H2A in telomeric silencing. We find that specific alleles of histone H2A, but not H2B, substantially reduce TPE (collectively termed the hta1tpe alleles). Deletion mutations of either the amino or the carboxyl tails produced significant reductions in TPE efficiency. A site-directed mutational analysis of modifiable sites in H2A led to the identification of residues in both amino- and carboxyl-terminal tails that influence TPE efficiency, implicating specific sites of acetylation and phosphorylation as functionally important. Furthermore, all hta1tpe point mutations also confer a strong Spt^- phenotype.

Interestingly, we find that the silencing defects of hta1tpe alleles are exacerbated by bleomycin-induced double-strand-break (DSB) formation. Furthermore, we find that the TPE defects in hta1tpe alleles are more severe in the absence of yKu70 function. H2A is intrinsically involved in yKu-mediated silencing. Many of the hta1tpe alleles increased the sensitivity to bleomycin and displayed a heightened requirement for yKu70 in DSB repair. These data suggest a mechanistic link between repair of bleomycin-induced DSBs and H2A involvement in the regulation of telomeric, subtelomeric, and Spt⁺ chromatin structure.

	MATERIALS AND METHODS

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Plasmids:
pJH55 is a pRS313 derivative that carries the genomic BamHI-SacII fragment of the HTA1-HTB1 locus as the sole source of histones H2A and H2B. Plasmid pJH55 and its derivatives carrying hta1 mutations 4-20 (N), S19P, S19F, K21E, 120-131 (C), and S121P have been previously described (HIRSCHHORN et al. 1995 ). Plasmids containing the hta1 alleles K4R/K7R, K4M/K7M, K7R, K13R, and K13M were the kind gift of Barbara Dunn and Mary Ann Osley. Derivatives of pRS314 carrying the htb1 deletion mutations (3-32, 3-22, 14-31) were previously described (RECHT and OSLEY 1999 ). EcoRI-BamHI fragments of these plasmids carrying the mutant hta1/HTB1 locus were cloned into pRS313. Additional point mutations in the H2A coding region (S1A, N/S1A, K4R, K21M, K21R, S121A, T125A, T125E, S128A, S128E, and the double mutants T125A/S128A and T125E/S128E) were constructed via site-directed mutagenesis of pJH55 (Stratagene, La Jolla, CA). The S128A mutation was created two times with no difference in phenotype. Mutant plasmids were sequenced to verify the presence of each mutation using flanking primers. Bacterial culture and transformations were performed according to standard procedures.

Yeast strains and media:
Rich (YPD or YPAD), synthetic complete (SC), and fluoroorotic acid (FOA)-containing SC media (SC + FOA) were prepared according to standard procedures. All strains used in this study were derived from FY406 [MATa (hta1-htb1)::LEU2 (hta2-htb2)::TRP1 his3200 trp163 lys2-128 ura3-52 leu21 pSAB6], an S288C derivative, which carries the sole source of H2A and H2B on pSAB6 (HIRSCHHORN et al. 1995 ). A plasmid shuffle was used to replace the wild-type pSBA6 with wild-type or mutant pJH55 derivatives. pSBA6 loss was selected on the basis of resistance to FOA and verified by Southern analysis. Growth rates for each strain were determined in liquid YPAD media at 30°.

pRS306 vectors containing URA3 sequences were linearized using either NcoI or StuI and subsequently transformed into FY406 or hta1 mutants. Since these restriction sites lie upstream of the Ty insertion site of ura3-52, within the URA3 coding region, the resulting integrant contains the wild-type URA3 gene adjacent to the genomic ura3-52 allele.

The yku70 null alleles were generated by targeted PCR. yku70::kan^r were amplified by PCR from BY4741/yku70::kan^r (MATa yku70::kan^r his31 leu20 met150 ura30; WINZELER et al. 1999 ; ATCC) using two primers, 5'-T AGACGGACTCATAATTGAATGGTT-3' and 5'-CACTTGGCGTGGTTTATTAGACTAT-3', corresponding to upstream and downstream sequences of YKU70, respectively. The PCR product was transformed into strains containing the wild-type, N, and T125A alleles. YKU70 disruptions were verified by both PCR and Southern analysis. Additionally, each of the yku70 strains was temperature sensitive at 37° and produced short telomeres, as expected.

Silencing and fluctuation analyses:
V_R or VII_L telomeres were labeled with URA3 as previously described (GOTTSCHLING et al. 1990 ). Telomere labeling was verified by Southern analysis. Telomere length was not affected by the introduction of any hta1 mutation tested in this study (data not shown). Telomeric silencing was measured using fluctuation analysis (LIU et al. 1994 ). Briefly, cells containing either URA3-marked-VII_L or -V_R telomeres were streaked onto rich medium and grown for 3–4 days at 30°. Seven colonies of 1.5 mm in diameter were selected, appropriately diluted, and plated onto SC and SC + FOA plates. Colonies were counted after 3 days of growth at 30° and 7 additional days of growth at room temperature. The data were analyzed for statistical significance relative to wild type both by determination of 95% confidence limits around the mean and by the rank-sum test from at least three independent fluctuation analyses. Significant changes determined by the 95% confidence limits were also significant by the rank-sum test in all silencing assays.

To determine the effect of -irradiation exposure on TPE in the hta1tpe alleles, fluctuation analyses were conducted as described above, except that cells plated on SC and SC + FOA plates were immediately exposed to 20 Gy irradiation (70% viability).

During the course of these studies we found that sensitivity of Ura3^- cells on SC + FOA + bleomycin (SC+ FOA+ bleo) plates is increased relative to growth on SC+ bleo (data not shown). We therefore first determined the percentage viability of Ura3^- cells in each hta1 mutant relative to wild type at the same dosage of bleomycin, represented by [(FOA^r_xbleo/hta1/FOA^r _xbleo/HTA1) x 100], where x is the concentration of bleomycin. The percentage values represent the sensitivity of cells to bleomycin on SC + FOA plates. The results did not qualitatively differ from sensitivity to bleomycin on SC plates lacking FOA (data not shown).

To discern the quantitative effect of bleomycin on TPE, 1.5-mm colonies were dispersed in 300 µl ddH₂O and plated in 5- to 10-fold dilutions onto SC + FOA plates, as indicated. The first lane corresponded to a 5-µl aliquot of a 1:10 dilution of the initial cell suspension. Initial (silencing + sensitivity) values in cells containing a URA3-marked VIIL telomere were obtained relative to wild type at the same bleomycin dose, as described above, and are equal to [(FOA^r_{xbleo/hta1/URA3-VIIL}/FOA^r_{xbleo/HTA1/URA3-VIIL}) x 100]. To take into account bleomycin killing at the same dose and mutant, we normalized the values as [(FOA^r_xbleo/hta1/FOA_{xbleo/hta1/URA3/VIIL})] for each dose. To determine the loss of silencing due to increasing doses of bleomycin in a single mutant, the values are normalized to the 0 bleomycin control. The loss of silencing is then described by [(FOA^r_xbleo/hta1/FOA_{xbleo/hta1/URA3-VIIL})/(FOA^r_{0 bleo/hta1}/FOA^r_{0bleo/hta1/URA3-VIIL})].

The effect of yku70 in wild-type and hta1tpe cells was calculated by comparing the ratios [FOA^r (HTA1 YKU70)/FOA^r (HTA1 yku70)] to [FOA^r (hta1tpeYKU70)/FOA^r (hta1tpe yku70)]. Chi-square analysis, where each value is the sum of five multiple trials, was used to determine the significance of differences between these ratios. Quantitative mating-type tests for HML silencing were conducted as previously described (KYRION et al. 1993 ).

Spt^- quantification:
Suppression of lys2-128 was assayed by the ability of cells to grow on lysine omission media. Following growth on YPD or YPAD media, 1.5- to 2.0-mm colonies were dispersed into 300 µl ddH₂O and 10-fold serial dilutions were prepared. A total of 25 µl of a 10^-3 dilution and a 10^-2–10⁰ dilution were plated onto SC and SC-lysine omission media, respectively. Colonies were counted after growth at 25° for 6 days. The mean for each mutant was determined after multiple trials (more than three trials) and is presented as [(SC-lysine colonies)/(SC colonies) x 100] ± 95% confidence limits.

-Irradiation resistance:
Colonies of 1.5 mm were selected after growth on YPAD and dispersed into 10-fold serial dilutions of a 300-µl ddH₂O suspension. For 0–20 Gy exposure, we plated 25 µl of a 10^-3 dilution. For 200 Gy exposure, 25–50 µl of a 10^-2 dilution were plated. Plating was conducted in duplicate to synthetic complete media and immediately exposed to varying levels of -irradiation in a Cs-137 irradiation chamber. Cells were grown at 30° for 3 days postexposure and 7 additional days at room temperature before colony counts and survival ratios were determined. Resistance data were evaluated as the mean survival ratio as defined by [(number of colonies on -exposed plates/number of colonies on unexposed plates) ± the 95% confidence limits] for at least three independent experiments.

Plasmid end-joining assays:
Cells were transformed with 1 µg of SacI-linearized pRS316, releasing a 4-bp overhang, and plated in duplicate for selection on SC-Ura (uracil omission) plates. Dilutions of each transformation were also plated in duplicate on SC plates to control for plating efficiency. Final colony counts were made after 6 days of growth at 30°. The number of successful transformation events was determined as (number of colonies on SC-Ura/number of colonies on SC plates). For all alleles, each transformation was normalized to its wild-type control and the mean values were determined together with the 95% confidence intervals. Circular pRS316 transformation efficiencies were originally used as transformation controls. However, circular and linear efficiencies appear not to be correlated with respect to DNA concentration, apparently due to different optimum transformation efficiencies for linear vs. circular DNA.

Histone quantification:
Chromatin isolation: Yeast cells were grown in 60 ml YPAD with continuous shaking at 25°. Cultures were inoculated with cells from a stationary overnight culture and were grown to mid-log phase (OD at 600 nm = 0.3). Cells were collected by centrifugation (2000 x g for 5 min) and the two cell pellets obtained from each culture were combined by suspension in 30 ml YPAD. The washed cells were collected by centrifugation as described above. The cell pellet was chilled on ice to 4° and mixed with 1.5 ml chromatin isolation buffer [CIB; 0.15 M NaCl, 10 mM Tris-Cl, pH 8.0, 0.5% Triton X-100, 1 mM ß-mercaptoethanol, 1 mM sodium butyrate, 10 mM NaFl, 1 mM NaVO₃, 0.1 mM phenylmethylsulfonyl fluoride (PMSF), pH 8.0]. All subsequent steps were conducted at 4°. Glass beads were added to the top of the liquid surface and the capped tube was vortexed for 1 min. A total of 30 ml CIB was added to the tube and the contents were thoroughly mixed. The glass beads were allowed to settle and the supernatant was decanted into a fresh centrifuge tube. The glass beads were washed with 5 ml CIB, which were combined with the supernatant. Insoluble material, including the yeast chromatin, was recovered by centrifugation (20,000 x g for 5 min). The pellet was then suspended in 30 ml chromatin wash buffer (CWB; 0.15 M NaCl, 10 mM Tris-Cl, pH 8.0, 1 mM ß-mercaptoethanol, 1 mM sodium butyrate, 10 mM NaFl, 1 mM NaVO₃, 0.1 mM PMSF, pH 8.0) and insoluble material was collected by centrifugation. When analyzed for steady-state histone phosphorylation, stationary overnight cultures were inoculated and incubated in the presence of 1.0 mCi ³²PO₄ (New England Nuclear) and allowed to develop to logarithmic growth before collection.

Histone extraction: To extract histones, the washed chromatin pellet was suspended in 1 ml CWB supplemented with 10 µl protamine sulfate (10 mg/ml; Sigma Type X). The suspended chromatin was mixed with 1 ml 0.4 M H₂SO₄ and incubated for 24 hr with occasional mixing. Acid insoluble material was pelleted by centrifugation at 14,000 x g for 10 min and the supernatant was mixed with 1/4 volume 100% trichloroacetic acid and incubated on ice for 1 hr. Precipitated histones were collected by centrifugation at 14,000 x g for 10 min and the pellet was suspended in cold acetone containing 0.1% H₂SO₄. The washed histones were collected by centrifugation (14,000 x g for 5 min), washed with cold acetone, and dried at room temperature. Purified histones were dissolved in 100 µl acid urea loading buffer (5% acetic acid, 8 M urea, 5% ß-mercaptoethanol, 0.01% crystal violet) and analyzed by two-dimensional polyacrylamide gel electrophoresis, as previously described (GREEN et al. 1990 ). The ³²P-histone steady-state labeling patterns of histones were determined by phosphoimager scanning with quantitative software (Image Gauge V3.3). The stoichiometry of histones after Coomassie blue staining of two-dimensional gels was determined after scanning using Image Gauge software.

	RESULTS

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Deletion of the amino or carboxyl tails of histone H2A reduces the efficiency of TPE:
The essential roles of the amino-terminal tails of histone H3 and H4 in yeast silencing have been extensively documented (GRUNSTEIN 1998 ). However, far less is known about the potential TPE functions of the highly conserved amino-terminal and highly variant carboxyl-terminal tails of histone H2A (THOMPSON et al. 1994 ; GREEN 2001 ). To study the effects of mutations in the amino- and carboxyl-terminal tail of H2A, we used a strain in which one of the two copies of HTA-HTB (HTA2-HTB2) was deleted and the second HTA1-HTB1 locus was present on a centromeric plasmid (HIRSCHHORN et al. 1995 ). The strains also contained a URA3-marked telomere at the left arm of chromosome VII (VII_L) and, in some mutants, at the right arm of chromosome V (V_R; GOTTSCHLING et al. 1990 ). The silencing of the URA3 gene was assayed by the ability to grow in the presence of 5-FOA (FOA^R), which allows the growth of Ura3^-, but not of Ura3⁺, cells. Significant differences were determined by nonoverlapping 95% confidence limits and rank-sum tests.

We first examined yeast strains containing amino-terminal deletions of H2A1 or H2B1 and a URA3-marked VII_L telomere. In strains carrying a deletion of amino acids 4–20 (N) of H2A, FOA^R colonies decreased from the wild-type mean of 0.20 to the N value of 0.01 (Fig 1 and Fig 2A). Furthermore, the N deletion in cells carrying the URA3-marked V_R telomere decreased silencing >40-fold to the limit of detection (m = < 0.000053, Fig 2B). The N TPE phenotype was identical to TPE in S1A/N double mutants, which eliminates all amino-terminal sites of modification (Fig 2A). In contrast, although the amino terminus of H2B plays a role in chromatin-mediated transcriptional repression (RECHT et al. 1996 ), a series of analogous H2B amino-terminal deletion mutants conferred wild-type levels of silencing, consistent with the results of a previous study (THOMPSON et al. 1994 ; Fig 1). Hence, the amino-terminal TPE phenotypes are a unique property of the hta1 alleles.

View larger version (62K): In this window In a new window Download PPT slide   Figure 1. The HTA1 N-terminal tail functions in TPE. (Top) Representation of a URA3-marked VIIL telomere. Arrow, direction of URA3 transcription; ovals, subtelomeric nucleosomes; stop mark, silencing of the URA3 gene. (Bottom) Seven FY406 colonies containing plasmid-encoded wild-type H2A and H2B [t (number of trials) = 3, n (pooled sample size) = 21], hta1(4-20)[N] (t = 3, n = 21), htb1(14-31) (t = 3, n = 21), htb1(3-31) (t = 3, n = 21), or htb1(3-22) (t = 3, n = 21) were dispersed, diluted, and spread onto individual SC or SC + FOA plates. Silencing was measured as the mean fraction of FOA-resistant cells from pooled data. Gray stippled bars represent mean values that are not statistically different from those of wild type; the white stippled bar represents mean values that are statistically higher than those of wild type; and the black stippled bar represents mean values that are significantly lower than those of wild type. Error brackets represent the 95% confidence intervals.

View larger version (116K): In this window In a new window Download PPT slide   Figure 2. (A) The identification of hta1tpe alleles. (Top) Representation of the amino- and carboxyl-tail domains of histone H2A1. All mutant residues tested are shaded. , small -helix; , histone fold domains. (Bottom) The silencing phenotype of FY406 cells at the URA3-marked VIIL telomere-containing plasmid-encoded wild-type HTA1 (t = 3, n = 21), hta1(4-20)[N] (t = 3, n = 21), hta1S1A/N (t = 3, n = 21), hta1S1A (t = 3, n = 21), hta1K4R (t = 2, n = 14), hta1K7R (t = 3, n = 21), hta1K4R/K7R (t = 3, n = 21), hta1K13R (t = 4, n = 28), hta1K13M (t = 3, n = 21), hta1K21E (t = 3, n = 21), hta1K21R (t = 3, n = 21), hta1K21M (t = 3, n = 21), hta1(121-131)[C] (t = 12, n = 84), hta1S121A (t = 2, n = 14), hta1S121P (t = 3, n = 21), hta1T125A (t = 3, n = 21), hta1T125E (t = 2, n = 14), hta1S128A (t = 3, n = 21), hta1S128E (t = 2, n = 14), hta1T125A/S128A (t = 3, n = 21), or hta1T125E/S128E (t = 2, n = 14) were determined as described in the legend for Fig 1. The additional alleles that were tested but had no effect on telomeric silencing were K4MK7M, S19F, and S19P. (B) The silencing phenotype of FY406 cells at the URA3-marked VR telomere-containing plasmid-encoded wild-type HTA1 (t = 5, n = 35), hta1N (t = 2, n = 14), hta1K4R (t = 2, n = 14), hta1K7R (t = 6, n = 42), hta1K4R/K7R (t = 3, n = 21), hta1K21E (t = 3, n = 21), hta1C (t = 2, n = 14), hta1T125A (t = 2, n = 14), hta1S128A (t = 2, n = 14), and hta1T125A/S128A (t = 5, n = 35) were conducted as described in the legend for Fig 1. Bar graphs, symbols, and error brackets (95% confidence limits) are defined as in the legend for Fig 1.

Similarly, deletion of amino acids 120–131 (C) of H2A1 resulted in a significant (3-fold) mean decrease in silencing at the VII_L telomere (27.8 ± 2.8%) with an unusually broad 36-fold range of values surrounding the median. A more severe (10-fold) effect was observed at the URA3-marked V_R telomere (Fig 2B). These results indicate that both the amino and carboxyl termini of histone H2A are necessary for wild-type levels of telomeric silencing.

A role for regulated aminoterminal acetylation and deacetylation in TPE:
Previous biochemical studies have led to the identification of modifiable residues within the H2A amino terminus, some of which may influence TPE. In the amino terminus, two of these sites, K4 and K7, can be acetylated (OHBA et al. 1999 ; WHITE et al. 1999 ; VOGELAUER et al. 2000 ; SUKA et al. 2001 ; GOLL and BESTOR 2002 ). Simultaneous mutation of both of these lysine residues to arginine (K4R/K7R) significantly reduces silencing at both the URA3-VII_L (3-fold) and V_R (20-fold) marked telomeres (Fig 2A and Fig B). In contrast, mutations that mimic the charge of an acetylated state (K4M/K7M) had a slightly enhanced TPE phenotype, consistent with a role for histone H2A acetylation in TPE (27 ± 2.8%). However, no amino-terminal single mutation of a lysine to arginine (K4R, K7R, K13R, K21R; Fig 2A) confers a reduction in TPE. Consistent with the known deacetylation of K7 in heterochromatin (SUKA et al. 2001 ), K7R hyper-represses TPE at both V_R and VII_L telomeres (Fig 2A and Fig B), indicating that the K4R/K7R double mutant eliminates the hyper-repression observed in K7R cells. Hence, the K4R/K7R double mutant decreases silencing beyond the value for each single mutant, suggesting an antagonistic interaction between K4 and K7 residues in this micro-environment.

We obtained an additional effect on telomeric silencing after mutation of K21 located within the amino-terminal -helical region (REDON et al. 2002 ). Altering the charge status of residues in this region could produce structural changes resulting in TPE sensitivities. Indeed, mutation of K21 to a neutral or basic charge conferred wild-type levels of silencing. In contrast, mutation of K21 to a negatively charged amino acid led to a modest, but significant, twofold decrease in silencing at the VII_L telomere (Fig 2A). These data raise the possibility that the -helix may be a structural facilitator of telomeric silencing.

Multiple phosphorylation sites within the H2A carboxyl terminus:
As noted above, deletion of the carboxyl-terminal 11 amino acids of histone H2A results in a reduction in telomeric silencing. Interestingly, this region contains three potential sites for phosphorylation: S121, T125, and S128. To test the possible modification of these residues, we analyzed the steady-state labeling of ³²P-labeled histones from different mutants by two-dimensional polyacrylamide gel electrophoresis. The phosphorylation of histone H2A was measured relative to a common nonhistone species. While deletion of the C terminus eliminated 95% of all phosphorylation, single mutations in two of the three residues (T125A and S128A) did not decrease phosphorylation levels more than twofold (Table 1; Fig 3). Even the T125A/S128A allele did not result in loss of phosphorylation comparable to C values, suggesting a role for S121 in C-terminal phosphorylation under specific conditions. This is consistent with our finding that the triple mutant S121A/T125A/S128A has growth rates (2.95 ± 0.25 hr) far slower than those for simple deletion of the C terminus (2.1 ± 0.14 hr) or growth of T125A/S128A mutants (1.71 ± 0.3 hr). Curiously, a single S121A mutation confers a major increase in H2A phosphorylation, suggesting upregulation of phosphorylation at T125 and/or S128 residues. These data suggest that phosphorylation is distributed among multiple residues under non-DNA damage conditions and raises the possibility of regulated switching between phosphorylated residues.

View larger version (63K): In this window In a new window Download PPT slide   Figure 3. A complex pattern of C-terminal hydroxyl-amino acid phosphorylation. Autoradiograms of two-dimensional polyacrylamide gels of 32P-labeled histones isolated from FY406 cells containing plasmid-borne wild-type HTA1, hta1C, hta1S121P, hta1T125A, hta1S128A, or hta1T125A/S128A are shown. The bottom arrow of each panel indicates the position of the H2A species as identified by previous staining of the gel with Coomassie blue. The upper arrow refers to the species used as a control. H2A protein spots migrating to the upper left of the major H2A proteins are acetylated and/or phosphorylated isoforms. Note that overexposure of the hta1C autoradiogram is required to visualize the 32P-labeled histone H2A. H2A histone proteins generated by the different deletion alleles display an alteration of electrophoretic mobility due to size and charge effects.

The effect of phosphorylation site loss on TPE:
To discern the possible function of these residues in TPE, we analyzed the effects of hta1 mutants in each of the potential phosphorylation sites at S121 (S121A and S121P), T125 (T125A, T125E), and S128 (S128A, S128E). Mutations in only one of these three residues, T125A, led to a decrease in TPE at both marked telomeres. A 3-fold and >20-fold decrease was observed in the mean TPE frequencies at the URA3-marked VII_L and V_R telomeres, respectively (Fig 2).

If the negative charge produced by phosphorylation is the regulatory information supplied by this modification (GREEN 2001 ), then a mutation that mimics the negative charge of a phosphorylated residue should overcome the silencing defect produced by the T125A mutation. Accordingly, we found that, unlike T125A, T125E has no silencing defect. Similarly, silencing levels of neither the S128A nor the S128E allele differed from those of wild-type or T125E cells (Fig 2A), consistent with a requirement for a negative charge at T125 for wild-type levels of TPE.

Nonetheless, while S128A does not confer direct defects in telomeric silencing, S128 does appear to play a subtler role in TPE. Specifically, we have uncovered genetic interactions between residues S128 and T125. The presence of an S128A residue rescues the T125A TPE defect in T125A/S128A double mutants (Fig 2). These data suggest a regulatory circuit between T125 and S128 in the carboxyl-terminal microenvironment.

Nonspecific effects of the hta1tpe alleles are unlikely to explain these phenotypes. First, in most mutants, doubling times do not differ from those in wild type, with the exception of N and C strains that grow slightly more slowly than wild type (Table 1). In the case of C, this slower growth rate appears to be associated with the production of inviable or growth-arrested cells during culture (data not shown). However, growth rates do not correlate with the extent of silencing or any other phenotype assayed in this study (cf. Table 1 and Fig 2A). Second, two-dimensional gel electrophoresis of wild-type and mutant histones revealed that histone H2A did not exhibit significant differences in abundance, ruling out effects of underproduction or overproduction (Table 1; Fig 4). Third, a second sensitive telomere phenotype, telomere tract size, is not altered in any of the histone H2A alleles (data not shown).

View larger version (385K): In this window In a new window Download PPT slide   Figure 4. Wild-type abundance of H2A in hta1tpe alleles. Coomassie-blue-stained two-dimensional polyacrylamide gels from independently isolated chromatin from FY406 cells containing a plasmid-borne wild-type HTA1 (WT), hta1C, hta1N, and hta1T125A. Other alleles tested (but not shown here) are hta1S121A, hta1T125A, hta1S128A, and hta1T125A/S128A. In all cases the percentage [H2A/(H2A + H2B + H3 + H4)] was quantified. Values of this ratio did not vary among wild-type and hta1tpe alleles by >20%. Therefore, alterations in H2A abundance cannot account for the phenotypes observed in this study. The position of each histone is shown in A, and the arrows in B–D indicate the positions of H2A. Histone H2A proteins generated by N and C alleles have slightly altered relative positions on the gel.

The hta1tpe1 alleles do not influence steady-state HML silencing:
Yeast telomeric silencing and the cryptic mating-type silencing at the HM loci share common mechanistic elements. These include the histone H3 and H4 deacetylase Sir2 and the silencing proteins Sir3 and Sir4 that dock with the amino termini of both histone H3 and histone H4 in heterochromatic DNA (HECHT et al. 1995 ; MOAZED 2001 ; CARMEN et al. 2002 ). The expression levels of HML in wild-type and hta1tpe alleles were assayed by a quantitative mating-type assay that can detect small shifts in silencing. We observed no changes in steady-state mating efficiencies of either N (45 ± 3.4%) or C mutants (58 ± 10%) or of any other hta1tpe allele tested relative to wild type (50 ± 6.3%; Table 2).

All hta1tpe alleles confer Spt^- phenotypes:
Previous studies have shown that alterations in HTA1 gene dosage can suppress the auxotrophy conferred by the insertion of into the LYS2 5' coding region (CLARK-ADAMS et al. 1988 ; SHERWOOD and OSLEY 1991 ). The suppression or activation of -element transcriptional initiation or elongation has served as a classic model for the identification of factors involved in altering higher-order chromatin, including histones H2A and H2B (SHERWOOD and OSLEY 1991 ).

Interestingly, each hta1tpe allele confers Spt^- suppression of lys2-128, as indicated by the percentage of cells capable of forming a Lys⁺ colony (see Fig 5). The Spt^- phenotypes differed in their severity. Increases of 10- to 20-fold in Lys⁺ cells were exhibited by N, C, and T125A/S128A hta1tpe alleles (Fig 5B). In contrast, more dramatic 1000- to 2000-fold increases were observed for hta1K4R/K7R, hta1K21E, and hta1T125A missense mutations (Fig 5A). Hence, deletions of the amino or carboxyl termini confer phenotypes that differ significantly from mutations within the two tail regions. In contrast, neither hta1S121A nor hta1S128A alleles exhibited an Spt^- phenotype.

View larger version (74K): In this window In a new window Download PPT slide   Figure 5. All hta1tpe alleles confer Spt- phenotypes. For each experiment, suppression was measured by counting the number of cells capable of growth on lysine omission media relative to total cells. The cells assayed were FY406 cells containing a plasmid-borne copy of wild-type HTA1 (n = 4), hta1N (n = 4), hta1K4R (n = 3), hta1K7R (n = 3), hta1K4R/K7R (n = 4), hta1K21E (n = 4), hta1C (n = 8), hta1S121P (n = 4), hta1T125A (n = 5), hta1S128A (n = 4), and hta1T125A/S128A (n = 3). Mean values not statistically different from those of wild type are shown by gray stippled bars, while values higher than those of wild type are indicated by black stippled bars. Error brackets denote 95% confidence limits. All alleles are shown in A, while only lower-suppressing alleles are shown in B. Note that although the mean value of hta1S121P is relatively high, the large error precludes a conclusion regarding this allele.

The two combinatorial effects observed for TPE values are maintained in the Spt^- cells. In the case of K4R and K7R, neither single mutant confers an Spt^- phenotype. In contrast, the Spt^- phenotype of the K4R/K7R double mutant is elevated 1000-fold over wild type. Similarly, T125A confers a 1000-fold increase in Spt^- cells, while S128A does not confer an Spt^- phenotype. The T125A/S128A allele retains only a 10-fold increase in Lys+ cells, far closer to the S128A than to the T125A phenotype (Fig 5). The common presence of these genetic interactions suggests a mechanistic relationship between loss of silencing with an increased severity of the Spt^- phenotype.

Formation of bleomycin-induced DSBs further decreases silencing in hta1tpe alleles:
Recent data have suggested the presence of an equilibrium of silencing and repair factors between the telomere and the sites of DSBs, leading to lower levels of silencing (MARTIN et al. 1999 ; MCAINSH et al. 1999 ; MILLS et al. 1999 ). If the defects of the hta1tpe alleles studied here are acting through the stability of telomeric chromatin structure, then these mutants may influence the loss of telomeric factors after DSB formation. One test of this hypothesis is the level of telomeric silencing during or after DNA damage. To test the effects of DNA damage formation on TPE, we grew cells in the presence of bleomycin. Bleomycin creates DSBs that use homologous recombination (at <20 mU/ml) and yKu-dependent nonhomologous endjoining (NHEJ) at higher doses (MAGES et al. 1996 ; MILNE et al. 1996 ; MARTIN et al. 1999 ; MILLS et al. 1999 ; MOORE et al. 2000 ).

The hta1tpe cells were grown in low doses (5 mU/ml) of bleomycin. Under these conditions, both mutant and wild-type cells display only a 2-fold decrease in viability (see MATERIALS AND METHODS). A striking decrease in TPE was observed in most hta1tpe mutant cells (Fig 6). On the basis of FOA-resistant growth in the presence or absence of the marked telomere, the N allele conferred a >800-fold decrease over the wild-type value containing no bleomycin. This decrease is >40-fold above that of mutant cells lacking bleomycin, leading to the formation of microcolonies on SC + FOA + bleo plates (Fig 6A and Fig B). K21E colonies generate two distinct phenotypes on SC + FOA plates that differ in their response to bleomycin treatment (population 1, n = 2; population 2, n = 3). The first phenotype did not respond to bleomycin at 5 mU/ml, while the second population displayed a 50-fold decrease in silencing compared to mutant cells lacking bleomycin (Fig 6). This variability may be the consequence of epigenetic switches between H2A states or differences in the plasmid copy number. In contrast to other alleles, the K4R/K7R mutation does not respond to DNA damage.

View larger version (64K): In this window In a new window Download PPT slide   Figure 6. hta1tpe silencing is further diminished after DNA DSB formation. (Left) To discern the effect of bleomycin on TPE, 1.5-mm colonies were dispersed in 300 µl ddH2O and fivefold serial dilutions of FY406 strains containing wild-type HTA1, hta1K7R, hta1K4R/K7R, hta1K21E-2, hta1C, hta1T125A, hta1S128A, and hta1T125A/S128A were plated onto SC or SC + FOA media containing either 0 or 5 mU/ml bleomycin. For SC plates, the first lane corresponds to 5-µl of a 1:50 dilution of the initial cell suspension. For SC + FOA plates, the first lane corresponds to a 5 µl aliquot of a 1:10 dilution of the initial cell suspension. Subsequent lanes represent fivefold serial dilutions in the presence or absence of 5 mU/ml bleomycin. Microcolonies are difficult to visualize in these photographs. (Right) To compensate for the increase of bleomycin sensitivity on SC + FOA, values for wild-type HTA1 [n (number of trials) = 6], hta1N (n = 4), hta1K7R (n = 4), hta1K4R/K7R [n (0 mU) = 4, n (5 mU) = 5], hta1K21E (n = 5), hta1C (n = 6), hta1T125A (n = 6), hta1S128A (n = 6), and hta1T125A/S128A (n = 4) were normalized to wild type at 5 mU/ml of bleomycin in the presence or absence of a URA3-marked telomere. For each allele, the relative decrease in silencing was determined as described in MATERIALS AND METHODS. The resulting values are plotted in the graph. Gray stippled bars refer to values that do not statistically differ from wild type; black stippled bars represent >20-fold decreases in silencing relative to wild type. K21E-2 refers to one of two phenotypes that this allele displays on SC + FOA plates following induction of DNA damage.

The C allele did not give rise to a diminished TPE phenotype in response to DNA damage. In contrast, the T125A missense mutation within the carboxyl-terminal tail conferred a fivefold decrease in TPE in response to 5 mU/ml bleomycin. This defect is coupled with a concomitant decrease in growth rate compared to T125A cells grown in the absence of bleomycin. As in the absence of bleomycin, T125A/S128A did not confer a defect in silencing or in growth rate in the presence of bleomycin. Hence, the production of DSBs by bleomycin in most hta1tpe missense alleles exacerbated telomeric silencing.

Numerous studies have suggested that -irradiation is repaired in a yKu-independent mechanism, as is the nucleotide excision repairs following UV-induced damage (MOORE 1989 ; MILNE et al. 1996 ; MILLS et al. 1999 ). Interestingly, cells treated transiently with -irradiation do not display a major TPE response to DNA damage. Only one of the hta1tpe alleles, K21E, conferred a fourfold decrease in TPE after irradiation (m = 0.0252) compared to nonirradiated cells. These data indicate that the hta1tpe response to DNA damage may be dependent on differing mechanisms of DNA repair, such as yKu-dependent and -independent NHEJ and homologous recombination.

A subset of hta1tpe mutants is hypersensitive to specific classes of DSBs:
Growth on bleomycin destabilizes telomeric chromatin through the release of shared silencing/repair factors (e.g., yKu heterodimer and Sir3) to other nuclear sites, including induced DSBs (MARTIN et al. 1999 ; MCAINSH et al. 1999 ; MILLS et al. 1999 ). This depletion leads to decreases in telomeric silencing in response to DNA damage, but does not explain the response of cell viability to bleomycin. In wild-type cells, we observe a twofold decrease in viability at 5 mU/ml and a sixfold decrease at 15 mU/ml, similar to previous reports (MARTIN et al. 1999 ; MCAINSH et al. 1999 ). The "telomere-release" model predicts that hta1tpe alleles should release additional competent DNA repair factors, resulting in a consequential increase in double-strand-break repair, realized as increased bleomycin resistance.

In striking contrast, our results suggest that the converse appears to be true. Differential sensitivity to bleomycin was observed in the range between 15 and 20 mU/ml bleomycin and, with the exception of K4R/K7R, followed the pattern of hta1tpe phenotypes (Fig 7). In the case of the N, C, and T125A, the viability of cells at 15 mU/ml was 450-, 200-, and 17-fold lower, respectively, than that of wild type. In addition, colony growth rates were decreased compared to wild-type cells. S128A and T125A/S128A behave identically to wild type, suggesting once again the dominance of the S128A phenotype over T125A. A qualitatively similar result was obtained after growth of cells on SC + FOA + bleo (Fig 7) or on SC + bleo (Fig 9A; data not shown). These observations suggest two possibilities: that the hta1tpe alleles are defective in both telomeric silencing and DSB repair or that the two phenotypes are both linked to a telomeric defect in response to bleomycin-specific DSBs (see DISCUSSION).

View larger version (75K): In this window In a new window Download PPT slide   Figure 7. hta1tpe alleles are hypersensitive to DSB production. Ura3- cells carrying hta1tpe alleles were grown on SC + FOA in the presence of 15 mU/ml bleomycin. The percentage of viable cells in an hta1tpe allele was normalized to wild-type cells treated with the same level of bleomycin on SC + FOA plates. This value, described in MATERIALS AND METHODS, represents the decrease in bleomycin resistance at 15 mU/ml in a specific allele relative to wild-type cells and allows a direct comparison between viability and silencing on SC + FOA plates. Ura- strains produce a lower number of colonies on FOA media relative to growth on SC, thereby allowing for a direct comparison between silencing and bleomycin sensitivity. Growth on SC produced a qualitatively similar result. FY406 cells carried a plasmid-borne copy of wild-type HTA1, hta1N, hta1K7R, hta1K4R/K7R, hta1K21E-2, hta1C, hta1T125A, hta1S128A, and hta1T125A/S128A. Gray stippled bars refer to values not statistically different from those of wild type; black stippled bars refer to values significantly lower than those of wild-type resistance. N, K21-2, and T125A also had slow growth rates associated with the loss of silencing. Values for N and C (not visible in chart) are 0.112 and 0.031%, respectively.

View larger version (106K): In this window In a new window Download PPT slide   Figure 8. Plasmid repair defects of hta1tpe alleles. SacI-digested pRS316 DNA was transformed into FY406 cells containing plasmid-encoded wild-type HTA1 (n = 23), hta1N (n = 5), hta1K7R (n = 6), hta1K4R/K7R (n = 5), hta1K21E (n = 5), hta1C (n = 14), hta1T125A (n = 7), hta1S128A (n = 6), or hta1T125A/S128A (n = 5). n refers to the number of trials. The number of transformants for each allele was normalized to the internal wild-type control in each trial. The data are presented as the mean of all trials for each allele ± 95% confidence limits. Gray stippled bars indicate values not statistically different from wild type at P = 0.05; black stippled bars refer to alleles that displayed a significantly lower level of plasmid end joining.

View larger version (41K): In this window In a new window Download PPT slide   Figure 9. The influence of yKu70 on hta1tpe silencing and bleomycin sensitivity. (A) Sensitivity of hta1N and hta1T125A to bleomycin in the presence or absence of yKu70. Cells from each allele were plated at varying dilutions on SC and SC + bleomycin plates at the indicated concentrations. The results are presented on a semilog scale. In each case, the 95% confidence limits are indicated. Strain designations are shown on the right. Note that these values differ from those in Fig 7 since cells were grown on SC plates lacking FOA, which displays a lower sensitivity to bleomycin. (B) Sensitivity of TPE in wild-type and hta1tpe alleles that lack yKu. Each strain indicated was plated onto SC + FOA at 10-fold dilutions and the cells counted at the appropriate dilution. The ratio (FOAr/SC) was determined and used to correct for the lowered viability on FOA as described in A and in MATERIALS AND METHODS. Each YKU70 strain was set to a value of 1 and the fold decrease in FOA resistance in yku70 strains was plotted on the semilog plot. HTA1, hta1N, and hta1T125A alleles had a significantly greater loss in silencing than the corresponding allele had in the absence of Yku70.

A second approach used to measure DNA repair in the presence of the hta1tpe mutants is an in vivo plasmid repair assay reflecting predominantly yKu-dependent repair (DOWNS et al. 2000 ; MATERIALS AND METHODS). Following plasmid cleavage with SacI and transformation into wild-type or hta1tpe mutant cells, mean transformation efficiencies relative to wild type were determined (Fig 8). The products of plasmid rejoining were both simple ligations of the SacI site and ligations in which the SacI site was eliminated (data not shown). The eliminated sites are likely to reflect end degradation (or recombinational rearrangement) and religation (BOULTON and JACKSON 1998 ). With the exception of C, all of the hta1tpe alleles that displayed bleomycin sensitivity also displayed defects (ranging from 25 to 50% of wild-type values) in the plasmid repair assay. Once again, S128A and the T125A/S128A double mutant displayed wild-type phenotypes. In this assay, C conferred wild-type transformation levels. This allele may disjoin multiple pathways that process unique substrates through NHEJ or homologous recombination pathways. Consistent with this possibility we have found that the enhanced sensitivity to bleomycin in C cells is not accompanied by an increase in yKu dependence (data not shown).

Genetic interactions between hta1N and hta1T125A and the yku70 pathways:
The effects of bleomycin on TPE suggested that the hta1tpe alleles may more easily titrate the yKu heterodimer from telomeres to DSBs due to either an increased number of bleomycin-specific DSB or an alteration in telomeric chromatin structure. We would therefore predict a greater dependence of DSB on the presence of yKu in hta1tpe alleles. To test this hypothesis, we assayed sensitivity to varying doses of bleomycin in wild-type, N, and T125A alleles in the presence or absence of yKu70. The sensitivity of wild-type cells at 15 mU/ml bleomycin was 40–50% of the 0 mU/ml bleomycin value in both YKU70 and yku70 cells. These data are consistent with the predominant use of homologous recombination in the repair of DSBs at doses of bleomycin below 20 mU/ml (data not shown; MARTIN et al. 1999 ). In hta1N cells, however, viability decreased 2-fold in the yku70 null allele compared to YKU70 cells at 15 mU/ml bleomycin, indicating the higher use of the yKu system of DSB repair in these cells (P < 0.005; Fig 9A). A more dramatic effect was observed in the hta1T125A allele. These cells displayed a 3.5-fold and 120-fold increase in the dependency of DSB repair on yKu at 5 mU/ml and 15 mU/ml of bleomycin, respectively. Hence, the data are consistent with the proposed relationship between yKu dependence in wild-type and hta1tpe alleles.

The second prediction of this hypothesis is an altered equilibrium between telomeric yKu and DSB-bound yKu in the hta1tpe alleles. We therefore tested the dependence of silencing on yKu in wild-type and hta1tpe cells in the presence or absence of bleomycin. This assay is limited by the substantial decrease in silencing in yku70 cells (220-fold in our strains; PORTER et al. 1996 ; EVANS et al. 1998 ). However, both N and T125A alleles displayed a further >7-fold and >2.5-fold decrease in silencing, respectively, in the absence of yKu (P < 0.01; Fig 9B). In summary, these data suggest that histone H2A contributes to the maintenance of yKu-dependent silencing even in the absence of DNA damage.

	DISCUSSION

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

In this study, we demonstrate a role for the amino and carboxyl tails of histone H2A in telomeric silencing, a property not shared by histone H2B (THOMPSON et al. 1994 ). This represents the first evidence for the participation of histone H2A in TPE. Most of these hta1tpe alleles share a common set of phenotypes in several cellular processes (Table 2). First, all hta1tpe alleles suppress the lys2-128 mutation. This Spt^- phenotype is likely to be mediated through alterations in chromatin structure. The T125A allele confers a 2000-fold increase in the Spt^- phenotype relative to wild-type cells. In contrast, the T125AS128A allele conferred only a 7-fold increase relative to wild-type cells, although the S128A allele alone generated a wild-type phenotype. These data suggest that the mutant S128A has a strong influence on the T125A phenotype and therefore clearly participates in this process.

Second, the hta1tpe alleles N, S21E-2, and T125A confer a decrease in TPE in the presence of bleomycin relative to the 0-mU/ml bleomycin values. In addition, hypersensitivity to bleomycin is observed in most hta1tpe alleles (N, S21E-2, T125A, C). Indeed, the two hta1tpe alleles tested (N, T125A) are more sensitive to bleomycin in the absence of yKu70. Third, all of the hta1tpe alleles that conferred bleomycin-induced loss of silencing also displayed a decrease in the efficiency of NHEJ in a cohesive-end plasmid-joining assay. Taken together with the minimal effects of -irradiation, these data implicate a defect in yKu-mediated repair events—a major pathway for repair of bleomycin-induced DSBs and plasmid rejoining (MAGES et al. 1996 ; MILNE et al. 1996 ; MARTIN et al. 1999 ; MILLS et al. 1999 ; MOORE et al. 2000 ). Fourth, the link between silencing, yKu, and histone H2A is further strengthened by the observation that TPE N and T125A is additionally compromised in the absence of yKu. These studies indicate that the amino- and carboxyl-terminal tails of H2A are essential for wild-type levels of yKu-dependent TPE and DSB repair.

These phenotypes are not likely to be the consequence of nonspecific or global effects of hta1tpe alleles. First, as noted in the RESULTS, growth rate, H2A abundance, and telomere size are not significantly altered in the hta1tpe alleles (Table 1). Second, changes in URA3 transcription are unlikely to be the consequence of a loss of H2A-mediated repression of the basal transcription of some genes (LENFANT et al. 1996 ). This latter process is dependent on the Hir proteins and on components of the SWI/SNF complex (DIMOVA et al. 1999 ; RECHT and OSLEY 1999 ). In contrast, the pathway of URA3 transcriptional regulation involves a specific Ppt1 activator system and there is no evidence for the involvement of SWI/SNF in URA3 transcription (ROY et al. 1990 ). Third, some hta1 alleles within the amino terminus are defective in the SWI/SNF-mediated chromatin remodeling (HIRSCHHORN et al. 1995 ). Of the four amino-terminal alleles investigated in that study (N, S19F, S19P, S21E), the order of alleles that conferred swi/snf phenotypes is S19F >> S19P = N > S21E. In contrast, neither S19F nor S19P displayed a loss of telomeric silencing (data not shown). Thus, there is no correlation between the hta1 alleles involved in SWI/SNF gene activation and the hta1tpe alleles. Fourth, the enhanced sensitivity to bleomycin in hta1tpe alleles is not due to random fragmentation since -irradiation has no effect on viability relative to wild type (data not shown).

Both histones H2A1 and H2A3 function in telomeric silencing:
Mutations in HTA3, which encodes the yeast H2A.Z variant, cannot complement hta1 mutations, suggesting that H2A1 and H2A3, at least in part, perform unique functions (JACKSON and GOROVSKY 2000 ; SANTISTEBAN et al. 2000 ; REDON et al. 2002 ). Deletion of HTA3 results in a partial loss of HMR silencing and a significant (25-fold) decrease in FOA^r colonies at the URA3-marked VII_L telomere (DHILLON and KAMAKAKA 2000 ). Interestingly, this magnitude of silencing loss is similar to the defects observed in the N allele of H2A1. It is likely therefore that both H2A1 and H2A3 function in the formation or stability of telomeric heterochromatin, although the functional relationship between these variants remains unknown.

Internal regulatory circuits governing H2A function:
A comparison of phenotypes defines several distinct classes of hta1tpe mutants. The first class consists of mutations that are defective in silencing, the Spt⁺ phenotype, and the DNA damage response. These include N, K21E, and T125A. One allele, C, defines a second class, which is not defective in plasmid rejoining, but is responsive to bleomycin. This allele follows a pathway distinct from the T125A/S128A interaction within the C-terminal tail microenvironment. Finally, the K7R and K4R/K7R alleles define a third class that display defects only in telomeric silencing and in the Spt⁺ phenotype, but not in DNA damage response. These mutations act within the microenvironment of the amino-terminal tail and may define a unique pathway of H2A in TPE and Spt⁺ phenotypes unrelated to DNA damage. Therefore, it is likely that loss of the amino terminus confers defects through either structural modification(s) of histone H2A or loss of as-yet-unidentified binding sites for silencing-related proteins.

One interesting feature of our resul