Regulation of fragile sites expression in budding yeast by MEC1, RRM3 and hydroxyurea

Fragile sites are specific loci within the genome that exhibit increased tendencies for chromosome breakage. They are conserved among mammals and are also found in lower eukaryotes including yeast and fly. Many conditions, including mutations and exogenous factors, contribute to fragile site expression, but the nature of interaction among them remains elusive. Here, we investigated this by examining the combined effects of rrm3Δ, mec1 and hydroxyurea (HU), three conditions that induce fragile sites, on expression of the replication slow zone (RSZ), a type of fragile site in budding yeast. Contrary to the expectation that each factor would contribute to fragile site expression in an independent manner, we show that rrm3Δ and high concentrations of HU suppressed RSZ expression in mec1-4ts cells. Further analyses revealed that rrm3Δ suppression occurs via promotion of Sml1 degradation, whereas HU suppresses RSZ via a premature commitment to inviability. Taken together, these observations demonstrate that: (1) the yeast genome contains different types of fragile site with regard to regulation of their expression, and (2) each fragile-site-inducing condition does not act independently, but can elicit a cellular response(s) that can paradoxically prevent the expression of a specific type(s) of fragile sites.


Introduction
Accidental double strand breaks (DSBs) arise during unperturbed proliferation. Such breakage is observed in all organisms examined to date ranging from bacteria to humans, suggesting that the break formation is an undesired byproduct of an evolutionarily conserved process(es). Notably, these DSBs do not arise randomly throughout the genome but at specific locations, often referred to as fragile sites. Examples of such breakage-prone regions of a genome include the bacterial ter (Hill, 1996), the budding yeast replication slow zones (RSZs) (Cha and Kleckner, 2002), and the mammalian common and rare fragile sites (Durkin and Glover, 2007;Sutherland and Hecht, 1985).
Many conditions promote fragile site expression. For instance, inactivation of budding yeast Rrm3, an evolutionarily conserved DNA helicase, leads to chromosome breakage at ~1400 discrete fork pause sites in the genome, including tRNA genes, the telomeres, and centromeres (Azvolinsky et al., 2009;Torres et al., 2004a). Fragility at these loci stems from the presence of specific non-nucleosomal DNA-protein complexes, which directly impede fork progression (Torres et al., 2004a). Thermal inactivation of Mec1, the budding yeast ATR homolog, leads to chromosome breakages that are enriched at relatively large (~10 kb) loci referred to as replication slow zones (RSZs) (Cha and Kleckner, 2002). The RSZ is a genetic determinant that occurs alternately with active replication origins along the entire length of a chromosome, except in the centromeric region. Although the mechanism underlying RSZ expression is unknown, its suppression by sml1, a mutation that leads to a twofold increase in the dNTP levels (Zhao et al., 1998), suggest that the breakage might be regulated by changes in dNTP levels (Cha and Kleckner, 2002). Chromosome breakage has also been observed in checkpoint alleles of MEC1 or its downstream target RAD53 following exposure to an acute dose (200 mM) of HU (Admire et al., 2006;Raveendranathan et al., 2006). When Rad53 function is compromised, breakage is observed preferentially around a subset of early firing origins, termed compromised early origins (CEOs) (Raveendranathan et al., 2006).
Notably, the preferred sites of chromosome breakage in rrm3, mec1-4ts or mec1/rad53 + 200 mM HU do not overlap much, suggesting that different conditions lead to expression of a specific type(s) of fragile sites. The situation in mammals is similar, and the fragile sites are often classified according to their inducing conditions, for example, as folate-, aphicolin-or BrdU-sensitive fragile sites (Durkin and Glover, 2007;Sutherland and Hecht, 1985). Currently, the basis for this differential regulation of fragile site expression remains poorly understood.
The aim of this study was to examine the nature of interaction among different conditions that induce fragile sites. To this end, we assessed the effect of combining rrm3, mec1 and HU on expression of RSZs. Unexpectedly, we found that both rrm3 and high concentrations of HU suppressed RSZ expression. The implication of these observations regarding the nature of eukaryotic fragile sites and regulation of their expression is discussed.

Results and Discussion rrm3 is a mild suppressor of mec1-4ts
To examine the potential interaction between rrm3 and mec1 in RSZ expression, we assessed the status ChrIII in mec1-4ts rrm3 using the pulsed-field gel electrophoresis (PFGE) followed by Southern hybridization method (Fig. 1A). As shown before (Cha and Kleckner, 2002), chromosome breakage enriched at RSZs is observed in mec1-4ts cells (Fig. 1C,D). By contrast, the breakage was notably reduced in the mec1-4ts rrm3 sample, suggesting that inactivation of RRM3 suppressed the breakage. Ectopic expression of RRM3 restored the chromosome breakage (Fig. 1D), confirming that the suppression was due to the loss of RRM3. We found that rrm3 also improved viability of mec1-4ts at 30°C and 34°C (Fig. 1E). The extent of suppression conferred by rrm3, however, was not as robust as that conferred by sml1 or RNR1 overexpression (Desany et al., 1998;Zhao et al., 1998). Taken together, we conclude that rrm3 is a weak suppressor of mec1-4ts, which prevents DSB formation at RSZs and improves viability at restrictive temperatures.

rrm3 leads to Sml1 degradation and improves S-phase progression in mec1-4ts
Elimination of Rrm3 leads to chromosomal damage that activates the intra-S-phase checkpoint (Ivessa et al., 2003). The checkpoint activation in turn leads to the removal of Sml1, a direct inhibitor of Rnr1 (Chabes et al., 2003;Zhao et al., 2001), and higher levels of dNTPs in rrm3 (Cheng et al., 2009;Taylor et al., 2005). Given that elimination of Sml1 is an efficient means of suppressing mec1 lethality (Zhao et al., 1998), the rrm3 suppression might be mediated by MEC1-independent intra-S-phase checkpoint activation and the subsequent removal of Sml1. We addressed this possibility by examining the effects of rrm3 on the status of Sml1. Deletion of Rrm3 led to a notable reduction in Sml1 levels in both MEC1 and mec1-4ts backgrounds. For example, whereas the reduction in Sml1 levels was transient in the wild type (WT), where it was observed mostly during early S-phase ( Fig. 2A,B), the level remained low in MEC1 rrm3 culture even after the completion of bulk genome duplication. The rrm3 mec1-4ts double mutant also exhibited a modest decrease in Sml1 levels at the G1-S transition compared with the mec1-4ts control, and a notable improvement in bulk genome duplication ( Fig. 2A).
In the absence of Mec1 (e.g. mec1-4ts at restrictive temperature), a related kinase Tel1 carries out many of the functions of Mec1 (Carballo and Cha, 2007). Thus, if rrm3 suppression was via MEC1-independent checkpoint activation, the suppression is likely to require TEL1. Indeed, we found that the rrm3 suppression of RSZ expression was TEL1 dependent (Fig. 2C).
The current findings imply that chromosome breakage at RSZs and that at the rrm3-sensitive sites are differentially regulated. Specifically, although the rrm3-dependent downregulation of

182
Journal of Cell Science 124 (2)  arrest and release. The signal corresponding to the MYC-Sml1 is quantified and normalized to that of the tubulin band. Normalized signals are then expressed relative to the t0 sample, which was set to 1 for each strain. 'UnT', untagged control strain. (B)FACS profiles of the samples in A. (C)Status of ChrIII in the indicated strains. Two independently derived rrm3 mec1-4ts tel1 strains were analyzed. Sml1 (above) and the subsequent increase in dNTP levels (Cheng et al., 2009;Taylor et al., 2005) can suppress the RSZ expression, these effects apparently exert no impact on the estimated ~1400 rrm3-sensitive fragile sites (Torres et al., 2004a). The observation that rrm3 does not exhibit sensitivity to HU (Torres et al., 2004b), is also consistent with the notion that the nature of defect(s) conferred by rrm3 is unrelated to a reduction in dNTP levels. We propose that RSZs and the rrm3-sensitive sites represent two different types of yeast fragile site whose expression is regulated, respectively, by a reduction in dNTP levels (this study) or the presence of non-nucleosomal proteins that physically impede fork progression (Ivessa et al., 2003).

HU induction of RSZ expression
If a RSZ is a dNTP-sensitive fragile site, exposure to HU might promote its expression. The effects of HU might also depend on the dose as well as the basal level dNTPs in the cell. To test this idea, we assessed the effects of HU in three mec1 situations that differed with respect to the basal level dNTP concentrations: (1) mec1-4ts at 30°C, with less than WT levels of dNTPs as a result of the thermal inactivation of Mec1, (2) mec1-4ts at 23°C, with near-WT levels of dNTPs, and (3) mec1 sml1 with greater than WT levels of dNTPs because of sml1. As expected, in the absence of HU, only the mec1-4ts at 30°C acquired chromosome breakage at RSZs (Fig. 3Ai,Bi; Fig. 4Ai). mec1-4ts cells incubated at 23°C or the mec1 sml1 culture, in the absence of HU, efficiently completed bulk genome duplication without any signs of chromosome breakage or loss of viability ( Fig. 3Bi-iii, Fig. 4A). Addition of HU in the latter two led to RSZ expression, demonstrating that HU-mediated reduction in dNTPs can promote fragile size expression irrespective of the status of Mec1's essential function or SML1. Note that the amount of HU required for RSZ expression in mec1 sml1 cells was four times higher (10 mM) than that in mec1-4ts at 23°C (2.5 mM), reflecting the need for additional HU to overcome the effects of sml1. We conclude that RSZ is a dNTP sensitive fragile site whose expression is triggered when the dNTP levels in the cell fall below certain threshold level(s). dNTP levels, in turn, would be determined by Mec1 and Sml1-HU, which regulate RNR activity in a positive and negative manner, respectively (Fig. 4C).

Suppression of RSZ expression by high concentrations of HU
Both mec1-4ts and mec1 sml1 cells exhibit a dose-dependent sensitivity to HU (Fig. 3Aii,Bii, Fig. 4Aiii). Unexpectedly, however, a notable reduction in RSZ breakage is observed in the presence of highest concentrations of HU (Fig. 3Ai,Bi, Fig. 4A). We reasoned that this reduction could be due to the effects of HU on either or both of the following two requirements for RSZ expression: (1) replication fork stalling at the loci (Cha and Kleckner, 2002), and  unlikely to be due to cell cycle arrest before onset of the required mitotic chromosomal event. Indeed, we did not observe any notable effects of HU on cell cycle progression as assessed by DAPI staining and budding morphology (Fig. 3Aiv,Biv). mec1/rad53 mutants exposed to high concentrations of HU become committed to inviability shortly after cell initiate genome duplication (Desany et al., 1998;Cha and Kleckner, 2002). Commitment to inviability, defined as the inability of the cell to resume cell division following the removal of a sensitizing condition(s), such as HU or high temperature, is an indication that an irrecoverable damage to the cell has occurred. Evidence indicates that the commitment in mec1/rad53 cells exposed to high levels of HU is mediated by irreversible fork collapse, majority of which occurs within 3-7 kb away from an origin during early S-phase (Santocanale and Diffley, 1998;Shirahige et al., 1998;Sogo et al., 2002). Commitment to inviability in mec1ts cells at high temperature, however, is mediated by chromosome breakage at RSZs during G2-M transition (Cha and Kleckner, 2002). These considerations raise the possibility that the suppression of RSZ expression by high concentrations of HU might be due to collapse of replication forks before they reach RSZs, prematurely committing cells to inviability.
To test the above notion, we assessed the effects of HU on the kinetics of commitment to inviability in mec1-4ts cells incubated at 30°C. As expected, HU inhibited bulk genome duplication in a dose-dependent manner with a notable delay at 10 mM and an even earlier block at 50 mM (Fig. 3Aiii). The analysis reveals that chromosome breakage at RSZs is observed in the cultures (0, 0.5 mM, and 2.5 mM HU) where commitment to inviability occurred comparatively late. In these cultures, the approximate time at which 50% of the cells become committed to inviability (T 50 ; asterisks in Fig. 3Aiv) occurred at t150 minutes, almost 100 minutes after the genome duplication reached mid-late S-phase (Fig. 3Aiii,vi), by which time, replication forks have reached RSZs (Cha and Kleckner, 2002). In contrast, the T 50 in the mec1-4ts cultures without RSZ expression (10 mM and 50 mM HU) occurred notably earlier, at 40 and 20 minutes following the release, respectively (Fig. 3Aiv). At these times, the cells in the corresponding cultures were either just exiting (10 mM) or still in (50 mM) the small budded stage, with the genome duplication stuck at its early stage (Fig. 3Aiii).
Note that the reduced break signals in the 10 mM or 50 mM HU cultures was also due to the presence of non-linear replication intermediates that prevent the unresolved chromosomes from entering the gel. However, it is unlikely that these chromosomes contain RSZ breaks given that replication forks in rad53/mec1 mutants exposed to undergo fork collapse before reaching RSZs (Santocanale and Diffley, 1998;Shirahige et al., 1998;Sogo et al., 2002) and that when chromosome breakage occurs under these conditions, it occurs at a different type of fragile sites referred to as compromised early origins (CEOs) (Raveendranathan et al., 2006).
Taken together, we propose that the lack of RSZ expression in the 10 mM or 50 mM HU cultures is due to premature fork collapse, precluding the forks from reaching RSZs. Why the forks stalled at RSZs retain the ability to restart replication, allowing the

184
Journal of Cell Science 124 (2) When dNTP concentrations fall below the viable threshold (asterisks), mec1 or rad53 cells lose viability. If the extent of dNTP depletion is modest (the area marked by the grey box), RSZ expression is observed (grey circles). When the dNTP concentration falls below this level, the cells become prematurely committed to death before RSZ expression (black circles).
cell to resume cell division if returned to permissive temperature before G2-M, whereas those stalled elsewhere undergo rapid collapse, remains unknown.

RSZ expression in other checkpoint mutants
Is RSZ expression is specific to Mec1 inactivation? To address this question, we assessed the effects of HU on various checkpoint mutants that were either sensitive (rad53 and mrc1) or insensitive (rad9 or rad24) to HU (Fig. 4B). RSZ expression was observed only in the rad53 mutant demonstrating that checkpoint inactivation or HU sensitivity are not sufficient for RSZ expression; we propose that the failure to downregulate Sml1 in response to HU is a key requirement for RSZ expression. Note that RSZ expression in rad53 cells was observed in cultures exposed to 2.5 mM or 10 mM HU, but not 50 mM (Fig. 4B). The latter is consistent with the notion that the extent of dNTP depletion should be modest ( Fig.  4C; above). We conclude that RSZ expression is not mec1 dependent per se, but Mec1 inactivation happens to provide all the necessary conditions -i.e. modest reduction in dNTP levels, the inability to upregulate dNTP synthesis via Sml1 downregulation, and the loss of both intra-S and G2-M checkpoint regulation. We began the current study to address the nature of interaction among different fragile-site-inducing factors, predicting that rrm3 or HU would increase RSZ expression in mec1-4ts by elevating fork stalling. The unexpected results presented here reveal that different fragile-site-inducing factors do not act independently, but rather interact in a complex manner that can suppress expression of a specific type of fragile sites. The current study also reveals that the yeast genome contains both dNTP-sensitive (e.g. RSZs) and dNTP-insensitive (e.g. centromeres, tRNA genes, and telomeres) fragile sites. Given the similarities between the nature of mammalian and yeast fragile sites, the current findings provide highly relevant insights into the nature of mammalian fragile sites and regulation of their expression.

Yeast manipulation and media
Standard yeast manipulation procedures and growth media were used. All strains were of the SK1 background; relevant genotypes of the strains used in the current study are listed in supplementary material Table S1. To obtain a synchronous mec1-4ts culture for cell cycle analysis, cells were grown to mid-log phase and arrested for 3 hours with 5 g/ml -factor at 23°C and released at either 23°C or 30°C in the media specified in each experiment. In all cases, samples for RSZ analysis were collected 5 hours after the release.

Spot test
Indicated strains were grown to log phase at 23°C in YPD. Each culture was adjusted to an OD600 of 0.4 and tenfold serial dilution was spotted onto YPD agar. The agar plates were then incubated at the indicated temperature for 3 days.

Pulse-field gel electrophoresis (PFGE)
Agarose plugs containing chromosome-sized DNA were prepared as previously described (Cha and Kleckner, 2002). Electrophoresis was performed using 1/3 plug per lane in a Bio-RAD CHEF Mapper using the following parameters: 6 V/cm, 5-30 seconds, 14°C, 1% agarose, 0.5ϫ TBE and a 24 hour running time.