The activation loop residue serine 173 of S.pombe Chk1 kinase is critical for the response to DNA replication stress

Why the DNA damage checkpoint kinase Chk1 protects the genome of lower and higher eukaryotic cells differentially is still unclear. Mammalian Chk1 regulates replication origins, safeguards DNA replication forks and promotes fork progression. Conversely, yeast Chk1 acts only in G1 and G2. We report here that the mutation of serine 173 (S173A) in the activation loop of fission yeast Chk1 abolishes the G1-M and S-M checkpoints without affecting the G2-M arrest. Although Chk1-S173A is fully phosphorylated at serine 345 by the DNA damage sensor Rad3 (ATR) when DNA replication forks break, cells fail to stop the cell cycle. Mutant cells are uniquely sensitive to the DNA alkylation agent methyl- methanesulfate (MMS). This MMS sensitivity is genetically linked with the lagging strand DNA polymerase delta. Chk1-S173A is also unable to block mitosis when the G1 transcription factor Cdc10 is impaired. Serine 173 is equivalent to lysine 166 in human Chk1, an amino acid important for substrate specificity. We conclude that the removal of serine 173 impairs the phosphorylation of a Chk1 target that is important to protect cells from DNA replication stress. Summary statement Mutation of serine-173 in the activation loop of Chk1 kinase may promote cancer as it abolishes the response to genetic alterations that arise while chromosomes are being copied.


Summary statement 24 25
Mutation of serine-173 in the activation loop of Chk1 kinase may promote cancer as it 26 abolishes the response to genetic alterations that arise while chromosomes are being 27 copied. 28

Introduction 48
homologous recombination at broken forks through Rad51 and BRCA2, regulates fork 74 elongation and arrests cell cycle progression by promoting the degradation of Cdc25A 75 (reviewed in (González Besteiro and Gottifredi, 2015)). While yeast Chk1 can be deleted 76 (Walworth and Bernards, 1996), mammalian cells depend on the kinase for viability. 77 Interestingly, only S345 phosphorylation is required for the essential roles of Chk1 (Wilsker 78 et al., 2008). Inhibition of human Chk1 in unperturbed cells interfers with S phase 79 (Petermann and Caldecott, 2006) and mitosis (Zachos and Gillespie, 2007). Cdc2 (CDK1) 80 phosphorylates human Chk1 at S286 and S301 during normal mitosis as well as in the 81 response to DNA damage (Shiromizu et al., 2006) (Ikegami et al., 2008) with as yet 82

unknown functional implications. 83
Another open question is how the catalytic activity of Chk1 is regulated. The generally 84 accepted model predicts an auto-inhibitory complex between the N-terminal kinase domain 85 and the C-terminal regulatory domain (Kosoy and O'Connell, 2008) (Palermo et al., 2008). 86 This complex is thought to open up when S345 is phosphorylated by ATR (Rad3) at sites 87 of DNA damage. Whether this model is correct is still unclear since only the N-terminal 88 kinase domain of human Chk1 has been crystallised (Chen et al., 2000). The activation 89 loop adopted an open conformation in this structure which implies that Chk1 does not 90 depend on the modification by an upstream activator as many other kinases do. How Chk1 91 is silenced at the end of the DNA damage response is also not fully understood. Human 92 Chk1 is degraded after its modification at S345 in a process that is independent of the 93 other phosphorylation sites (Zhang et al., 2005). A similar degradation does not occur in 94 yeast. Attenuation of Chk1 correlates with its dephosphorylation at S345 by Wip1 (PPM1D) 95 in human cells (Lu et al., 2005) and by Dis2 in S.pombe (den Elzen and O'Connell, 2004). 96 Interestingly, Wip1 is replaced by PPA2 in undamaged cells where it dephosphorylates 97 Chk1 at S317 and S345 (Leung-Pineda et al., 2006). Currently no information is available 98 on the regulation of Chk1 in unperturbed yeast cells. 99 We report here a rare separation-of-function mutation in Chk1 kinase. Mutation of serine 100 173 (S173A) in the activation loop of S.pombe Chk1 abolishes the G1-M arrest, when cells 101 arrest at start in cdc10 mutant cells, and the S-M arrest in the response to broken DNA 102 replication forks. The G2-M checkpoint responses are largely intact and the mutant 103 kinases is fully phosphorylated by Rad3. Chk1-S173 is also specifically sensitive to the 104 alkylation of the DNA template by methyl-methanesulfonate in a manner related to the 105 lagging strand DNA polymerase delta. We conclude that the S173A mutation impairs the 106 activation of a downstream target of Chk1 that is specifically involved in the response to 107 DNA replication stress. This conclusion is in line with the requirement of the equivalent 108 lysine 166 for substrate recognition in human Chk1 (Chen et al., 2000). 109

Results 110
Reduced S345 phosphorylation of Chk1-S173A in unperturbed cells 112 Lysine 166 occupies a central position in the activation loop of human Chk1 opposite the 113 catalytic aspartate 130 (D155 in S.pombe, Fig. 1A) where it may determine substrate 114 specificity (Chen et al., 2000). The corresponding S.pombe residue is serine 173 (Fig. 1A) 115 and aspartate 189 in S.cerevisiae. 116 To find out whether S173 plays a role in Chk1 activity, we mutated this residue to alanine 117 and integrated the mutant gene with a C-terminal HA 3 tag (chk1-S173A-HA 3 ) at its 118 endogenous locus using the Cre-lox recombination system (Watson et al., 2008). The 119 integrated gene was amplified and the mutation was confirmed by DNA sequencing. We 120 also integrated the wild type gene (chk1-HA 3 ) (Walworth and Bernards, 1996) to exclude 121 any effects of the flanking lox DNA sequences on chk1 expression (Fig. S1). 122 We first used the phos-tag electrophoresis assay (Caspari and Hilditch, 2015) to study the 123 phosphorylation pattern of wild type Chk1 to establish a base line for the analysis of Chk1-124 S173A. Phos-tag acrylamide slows down the mobility of proteins relative to the extend of 125 their phosphorylation (Kinoshita et al., 2006). We activated wild type Chk1 with the 126 topoisomerase 1 inhibitor camptothecin (CPT) that breaks DNA replication forks in S 127 phase (Pommier et al., 2010). As previously reported (Wan et al., 1999), CPT induced the 128 mobility shift of Chk1-HA on normal SDS page which is triggered by the phosphorylation 129 of S345 (Capasso et al., 2002) (Fig. 1B). Analysis of the same samples on a phos-tag gel 130 revealed a larger number of phosphorylated Chk1 forms in untreated cells and a group of 131 additional bands when cells were treated with 10μM CPT for 3.5h (Fig. 1B,C). Since these 132 inducible bands were absent in the S345A mutant (chk1-S345A-HA 3 ) (Janes et al., 2012) 133 and in cells without Rad3 kinase (chk1-HA 3 Δ rad3), they are related to the phosphorylation 134 of serine 345 (Fig. 1C). We also noticed that the hypo-phosphorylated material of Chk1 at 135 the bottom of the phos-tag gel consists of at least two bands (A & B in Fig 1C). Mutation of 136 S173 to alanine (S173A) had no obvious impact on the normal band shift when cells were 137 treated with 12mM hydroxyurea (HU), which stalls DNA replication forks, with 10μM 138 camptothecin (CPT) or with the UV mimetic 4-nitroquinoline 1-oxide (4-NQO) at 10μM (Fig.  139   1D). Also the phosphorylation pattern of Chk1-S173A in untreated cells was not 140 significantly different from wild type (Fig. 1E). 141 To find out whether the unperturbed phosphorylation of Chk1 relates to cell physiology, we 142 grew cells from early logarithmic growth into stationary phase and withdrew samples at 143 different times (Fig. 1F). The band associated with S345 phosphorylation peaked during 144 the most active growth phase of wild type cells (time point 2 in Fig. 1G) and was later 145 replaced by a hypo-phosphorylated form once cells had exited the cell cycle (time point 4, 146 band C in Fig. 1G). The peak in S345 phosphorylation reflects most likely the occurrence 147 of endogenous DNA replication damage. It was however interesting to find that the S173A 148 mutation lowered the amount of S345 phosphorylation during the active growth phase 149 (time point 2 in Fig. 1G) suggesting an impaired response to replication stress. 150 151 Chk1-S173A cells are sensitive to DNA alkylation 152 Since Chk1 is crucial for the G2-M checkpoint (Walworth and Bernards, 1996), we 153 synchronised chk1-HA 3 wild type and chk1-S173A-HA 3 cells in G2 by lactose gradient 154 centrifugation (Luche and Forsburg, 2009) and released them into rich medium with or 155 without MMS (0.05%), 4NQO (10μM) or HU (12mM) at 30°C to measure the delay time. 156 The first telling observation came when we compared the untreated strains. While wild 157 type cells (chk1-HA 3 ) entered the second cycle at around 180 min, chk1-S173A-HA 3 cells 158 were delayed by 20 min (Fig. 2A). Such a second cycle delay is typical for agents like CPT 159 or HU which interfere with DNA replication (Mahyous Saeyd et al., 2014). It is therefore 160 possible that the chk1-S173A-HA 3 strain suffers from a DNA replication problem that 161 triggers this short G2 delay. The UV mimetic 4-NQO and the DNA alkylation agent MMS 162 blocked both the passage through the first G2 since DNA is instantly damaged, whereas 163 HU caused the expected second cycle arrest as cells are only hit once they undergo DNA 164 replication (Lindsay et al., 1998). While the S173A mutation had no impact on the HU 165 arrest (Fig. 2B), it allowed cells to exit G2 earlier in the presence of 4-NQO and MMS (Fig.  166 2C, D). This partial G2-M checkpoint defect was more prominent for MMS as chk1-S173A-167 HA 3 cells started to return to the cell cycle already after 80 min compared with wild type 168 cells which arrested throughout the experiment (Fig. 2D). This checkpoint defect correlated 169 with a high MMS sensitivity of the mutant strain ( Fig. 2F, G). Interestingly, a similar loss of 170 viability was not observed when the chk1-S173A-HA 3 strain was treated with HU, CPT or 171 UV light (Fig. 2E). This is an important finding as it reveals S173A as a separation-of-172 function mutation. MMS modifies both guanine (to 7-methylguanine) and adenine (to 3-173 methlyladenine) thereby inducing mismatches in the DNA that are repaired by base 174 excision repair. Inefficient BER results in single-stranded DNA breaks independently of the 175 cell cycle but causes DNA double-strand breaks when these gaps are encountered by a 176 replication fork (Lundin et al., 2005). The MMS sensitivity of the chk1-S173A-HA 3 mutant 177 was not related to a defect in S345 phosphorylation as the mutant kinase displayed the 178 characteristic band shift on phos-tag SDS page (Fig. 2H). Interestingly, the S345 shift was 179 strongest at the lowest MMS concentration of 0.01% and declined at the higher 180 concentrations. 181 182 Chk1-S173A is defective in the G1-M checkpoint 183 In addition to its key role in G2, Chk1 blocks mitosis when S.pombe cells arrest at start in a 184 cdc10 mutant (Fig. 3A) (Carr et al., 1995). Cdc10 is a subunit of the MBF transcription 185 factor complex that activates S phase genes during the G1-S transition (Lowndes et al., 186 1992). We constructed chk1-HA 3 and chk1-S173A-HA 3 double mutants with the 187 temperature-sensitive cdc10.V50 (H362Y) allele (Marks et al., 1992) and released G2-188 synchronised cells into rich medium at 30°C and 37°C (Fig. 3B, C). As reported previously 189 (Carr et al., 1995), chk1-HA 3 cdc10.V50 cells progressed through the first cycle before 190 arresting in G2 at the restrictive temperature of 37°C (Fig. 3B). Entry into the first cycle 191 was delayed by 60 min due to the increase in the temperature (Janes et al., 2012). While Chk1 is independent of its S345 phosphorylation as the temperature up-shift from 30°C to 198 37°C did not trigger the band shift on normal SDS page (Fig. 3E). 199 Since Chk1 acts also upstream of Cdc10 to prevent entry into S phase when the DNA 200 template is alkylated by MMS ( Fig. 3F) (Ivanova et al., 2013), we synchronised chk1-HA 3 201 and chk1-S173A-HA 3 cells in metaphase using the cold sensitive nda3.KM311 allele 202 (Hiraoka et al., 1984) and released cells into rich medium with or without 0.01% MMS by 203 raising the temperature from 20°C to 30°C. This experiment would allow us to measure the 204 delay in G1-S transition induced by MMS. Untreated wild type cells (chk1-HA 3 205 nda3.KM311) initiated DNA replication between 40 min and 60 min post-release which 206 increased the DNA content from 2C to 4C (Fig. 3G, H). The mutant strain (chk1-S173A-207 HA 3 nda3.KM311) showed a similar behaviour but displayed two interesting differences. 208 Not all cells were able to escape the mitotic arrest as they maintained a 2C DNA content, 209 and the proportion of cells that exited reached the 4C DNA content slightly earlier than wild 210 type cells (Fig. 3H). The delayed exit from the metaphase arrest could be linked with the 211 ability of S.pombe Chk1 to sustain the activation of the spindle checkpoint that delays 212 metaphase-to-anaphase transition (Collura et al., 2005). Hence, the S173A mutation may 213 prolong this mitotic arrest. The faster progression of the mutant cells through S phase is 214 consistent with the reduced S345 phosphorylation during the unperturbed cell cycle (Fig.  215 1G) as this indicates a lower checkpoint activation. Addition of MMS delayed the 216 accumulation of the 4C DNA content in both strains, with the S173A mutant showing a 217 more pronounced effect (Fig. 3I). This led us to conclude that the activation loop mutation 218 affects only the down-stream function of Chk1 that restrains mitosis in the cdc10 mutant, 219 but not the up-stream function which delays G1-S transition in the presence of MMS. 220 221 Chk1-S173A fails to respond to broken replication forks 222 The next decisive observation came when we analysed the S-M checkpoint response to 223 broken DNA replication forks. As long as the structural integrity of a stalled fork is 224 protected by Cds1 kinase, Chk1 activity remains low (Xu et al., 2006). Cds1 (Chk2) kinase 225 shields stalled replication structures from nucleases and recombination enzymes (Kai et al., 226 2005) (Boddy et al., 2003). Chk1 is however strongly activated when forks break in the 227 absence of Cds1, and cells without Chk1 and Cds1 are completely checkpoint defective 228 (Lindsay et al., 1998). To test whether the S173A mutation impairs this response, we 229 combined the chk1-S173A-HA 3 allele with the deletion of cds1 (Δcds1). The double mutant 230 was as HU sensitive as the Δ chk1 Δ cds1 strain strongly implying that the activation loop 231 mutation blocks Chk1 activation when replication forks collapse in the absence of Cds1 232 (Fig. 3A). This conclusion was confirmed when we released G2-synchronised chk1-233 S173A-HA 3 Δ cds1 cells into rich medium with 12mM HU. Like the checkpoint defective 234 Δ chk1 Δ cds1 strain, the chk1-S173A-HA 3 Δ cds1 mutant entered a fatal mitosis 140min 235 post-release (Fig. 3B). The majority of cell died while they re-entered the cell cycle 236 indicated by the cut phenotype where one daughter cells is anuclear or where the new wall 237 cuts through the single nucleus (Fig. 3C). Collectively, these results demonstrate an 238 outright dependency of cells on serine 173 when replication forks break in the absence of 239 Cds1. As in the earlier experiments, Chk1-S173A was fully phosphorylated at S345 in 240 Δ cds1 cells (Fig. 3D). These results imply a defect of Chk1-S173A down-stream of 241 collapsed replication forks in the absence of Cds1. 242 243 Chk1-S173A reduces the viability of DNA polymerase epsilon mutant cells 244 Because deletion of S.pombe chk1 compromises the viability of temperature-sensitive 245 mutants of DNA polymerase delta and epsilon (Francesconi et al., 1995), we combined 246 mutant alleles in the three replicative DNA polymerases alpha (swi7-H4), delta (cdc6.23) 247 and epsilon (cdc20.M10) with either chk1-HA 3 or chk1-S173A-HA 3 . While testing cell 248 growth at the semi-restictive temperature of 33°C, we noticed that the S173A mutation 249 specifically reduced the viability of the pol epsilon (cdc20.M10) mutant as the chk1-S173A-250 HA 3 cdc20.M10 double mutant grew only very poorly compared to the chk1-HA 3 251 cdc20.M10 strain (Fig. 5A). DNA polymerase epsilon synthesises the leading strand 252 (Pursell et al., 2007), is involved in long-patch BER (Wang et al., 1993), associates with 253 the DNA replication checkpoint protein Mrc1 (Claspin) (Lou et al., 2008) and establishes 254 heterochromatin (Li et al., 2011). The reduced viability at 33°C could suggest two roles of 255 Chk1. Either the kinase responds to replication problems associated with the leading 256 strand or it promotes DNA pol delta that can remove mismatches left behind by pol epsilon 257 (Flood et al., 2015). Phos-tag analysis showed that some hypo-phosphorylated material 258 was absent from Chk1-S173A , but this was the case for both, pol delta and epsilon (Fig.  259   5B). 260 We next synchronised the strains in early S phase using the HU protocol (Luche and 261 Forsburg, 2009) and released them back into the cell cycle to follow their progression into 262 G2. While the S173A mutation had no impact in the case of DNA polymerase delta (chk1-263 S173A-HA 3 cdc6.23) (Fig. 5C), it did advance cell cycle progression in the DNA 264 polymerase epsilon strain (chk1-S173A-HA 3 cdc20.M10) (Fig. 5D). The mutation in the 265 activation loop allowed cells to acquire a G2 (2 copies, 2C) DNA content 90 min post-266 release, approximately 30 min earlier than the wild type Chk1 kinase (chk1-HA 3 267 cdc20.M10). We did however find no evidence of S345 phosphorylation in any mutant 268 strain during this experiment (Fig. 5E). The faster progression of the chk1-S173A-HA 3 269 cdc20.M10 mutant could explain why the pol epsilon strain loses viability at the semi-270 permissive temperature. The activation loop mutation S173A might block the 271 phosphorylation of a down-stream target that is crucial for a reduction in leading strand 272 synthesis when DNA polymerase epsilon is impaired or when pol delta needs to remove 273 mismatched nucleotides. 274

275
The MMS sensitivity of Chk1-S173A is linked with DNA polymerase delta 276 Given the requirement of pol delta for the removal of alkylated bases by BER (Blank et al., 277 1994), we tested the genetic relationship between chk1-S173A-HA 3 and cdc6.23. 278 Intriguingly, the mutation in the catalytic subunit of pol delta affected survival on MMS 279 plates differentially depending on whether the chk1-HA 3 wild type or chk1-S173A-HA 3 280 mutant allele was present. While cdc6.23 cells containing the wild type kinase were MMS 281 sensitive, cdc6.23 cells with the mutant kinase displayed some degree of resistance (Fig.  282   6A). We followed this observation up by conducting an acute survival test at 0.025% MMS 283 and noticed that the chk1-HA 3 cdc6.23 double mutant was significantly more MMS 284 sensitive than the pol delta (cdc6.23) single mutant that contains the untagged chk1 gene 285 (Fig. 6B). This implies that the tagged chk1-HA 3 allele, which has been used in many 286 studies (Walworth and Bernards, 1996), differs from the untagged gene in a cdc6.23 287 mutant background. Intriguingly, the mutation in the activation loop suppressed this hyper-288 sensitivity to a level observed for the chk1-S173A-HA 3 single allele (Fig. 6B). Collectively, 289 these data show that the MMS sensitivity of the chk1-S173A mutation is epistatic with the 290 cdc6.23 mutation in the catalytic subunit of pol delta at 30°C and that the mutation also 291 suppresses the damaging activity of the tagged wild type Chk1 kinase. The nature of this 292 activity is as yet unknown. We suspect however that the C-terminal tag interferes with the 293 repair function of pol delta in BER (Blank et al., 1994). To test whether the polymerase 294 mutations interfere with S345 phosphorylation of Chk1 and Chk1-S173A, the 295 corresponding strains were treated with 0.01% MMS at 30°C and also exposed to the 296 semi-permissive temperature of 33°C without MMS. While both Chk1 proteins were 297 phosphorylated at S345 in the presence of MMS, the phosphorylation of the wild type 298 kinase was lower in the pol delta mutant coinciding with its high MMS sensitivity (Fig. 6C). 299 Chk1 was only weakly S345 modified at 33°C in both polymerase mutants indicating that 300 no or very little endogenous DNA damage occurs under these conditions. 301

Discussion 303
The only separation-of-function conditions known so far are the phosphorylation of S317 of 304 human Chk1, which is only required for the DNA damage response but not for its essential 305 functions (Wilsker et al., 2008), and the mutations E92D and I484T in S.pombe Chk1 306 which affect the S-M checkpoint but only at 37°C (Francesconi et al., 1997). We report 307 here a new separation-of-function mutation, S173A in the activation loop of S.pombe Chk1, 308 that abolishes the G1-M and S-M checkpoints independently of S345 phosphorylation 309 under normal growth conditions. When chk1-S173HA 3 cells arrest at start during the G1-S 310 transition due to the cdc10.V50 mutation, they cannot prevent mitosis (Fig. 3C,D). A 311 similar problem arises when DNA replication forks break in HU medium in the absence of 312 Cds1 (Fig. 4B, C). Since cdc10.V50 cells arrest with unreplicated chromosomes at start 313 (Luche and Forsburg, 2009), both Chk1 requirements must reflect distinct G1-M and S-M 314 checkpoint activities of Chk1. What is however intriguing is that the chk1-S173A mutant is 315 not CPT sensitive (Fig. 2E), although camptothecin also breaks DNA replication forks 316 (Pommier et al., 2010). This implies that the activation loop mutation is only critical when 317 Cds1 is absent. Since Cds1 protects damaged forks from nucleases and recombinases 318 (Kai et al., 2005) (Boddy et al., 2003), it is possible that the activation loop mutation 319 activates a DNA repair factor that is redundant as long as Cds1 is active. The DNA 320 damage signal must reach Chk1-S173A as the mutant kinase is phosphorylated at S345 in 321 the presence of CPT (Fig. 1D), 4-NQO (Fig. 1D), MMS (Fig. 2H) and HU (Fig. 3D). It is 322 therefore unlikely that the S173A mutation interferes with Rad3 activation at damaged 323 chromosomes involving Crb2 (53BP1), Rad4 (TopBP1) and the 9-1-1 ring (Furuya et al., 324 2004). Since the corresponding lysine-166 in human Chk1 is involved in substrate 325 specificity (Chen et al., 2000), it is more likely that the activation loop mutation blocks the 326 phosphorylation of a down-stream target that is required to restrain mitosis in cdc10 327 mutant cells and when forks break in cds1 deletion cells (Fig. 7A). This target appears to 328 be distinct from Wee1 and Cdc25 because the chk1-S173A strain is able to block mitosis 329 when cds1+ cells are treated with HU or the UV mimetic 4-NQO (Fig. 2B, C). A clear 330 difference exists however when DNA is alkylated by MMS as chk1-S173A cells have a 331 partial G2-M checkpoint defect (Fig. 2D) and are highly sensitive (Fig. 2G; 6B). 332 The genetic link between DNA polymerase delta and Chk1-S173A may hint at this 333 unknown target. The observation that the activation loop mutation reduces the viability of 334 the pol epsilon (cdc20.M10) mutant at 33°C (Fig. 5A) could be explained by the faster 335 progression through S phase (Fig. 5D). This faster progression may however be linked 336 with DNA polymerase delta given that pol epsilon needs pol delta to repair any remaining 337 mismatches in the leading strand which are not removed by its own 3`-exonuclease 338 activity (Flood et al., 2015). Pol delta is also able to replicate the leading and the lagging 339 strand once a fork has collapsed (Miyabe et al., 2015). The requirement of S173 for 340 viability of the pol epsilon (cdc20.M10) mutant could therefore mean that Chk1 is involved 341 in the repair activities of pol delta either when mismatched bases remain in the leading 342 strand after MMS treatment or when the leading strand is elongated by pol delta during the 343 homologous recombination dependent re-start of collapsed replication forks in HU-treated 344 Δ cds1 cells (Fig. 7A). This conclusion is strengthened by the epistatic relationship between 345 chk1-S173A and cdc6.23, the catalytic subunit of pol delta (Fig. 6B). 346 In summary, S173A is a rare separation-of-function mutation of Chk1 that may help to 347 dissect its role in S phase where it might link post-replication repair by DNA polymerase 348 delta with a block over mitosis. To uncover the identity of its proposed target will however 349 require further work. It is intriguing that one of the other known separation-of-function 350 mutations, E92D (Francesconi et al., 1997), sits at the beginning of a loop opposite the 351 activation loop where S173A is (Fig. 7B). The other interesting notion is that this intra-S 352 activity of Chk1, which is not essential in yeast, may have become essential during the 353 evolution of higher eukaryotic cells (Petermann and Caldecott, 2006). 354

Base strain construction and integration of the Chk1 point mutations 365
The base strain was constructed as described in (Watson et al., 2008). The loxP and loxM 366 Cre-recombinase recognition sequences were integrated 84nt upstream of the start codon 367 and 84nt downstream of the stop codon (Fig. S1A) using the primers Base-1 and Base-2 368 (Fig. S1C). The point mutations S173A and S345A were introduced using fusion PCR as 369 reported in (Janes et al., 2012). Genomic DNA from the chk1-HA 3 strain (Walworth and 370 Bernards, 1996) was used as the PCR template to introduce the C-terminal HA affinity tag. 371 The two overlapping chk1 gene segments were amplified using the primers Base-3 and 372 the mutation reverse primer, and the primer Base-4 and the mutation forward primer (Fig.  373   S1C). The full-length fusion fragments were cloned into the lox-Cre integration plasmid 374 using the restriction enzymes SphI and SacI. Integration of the mutated chk1-HA 3 genes 375 resulted in the loss of 4nt upstream of the start codon and of 17nt downstream of the stop 376 codon (Fig. S1B). 377

Cell synchronisation 379
Cells were synchronised as described in (Luche and Forsburg, 2009). HU was used at a 380 final concentration of 15mM for 3.5h at 30 °C in rich medium. Lactose gradients were 381 centrifuged for 8min at 800rpm. The nda3.KM311 mitotic arrest was performed in rich 382 medium as reported in (Nakazawa et al., 2011). One volume of pre-warmed medium (40°C) 383 was added to the 20°C medium to quickly raise the temperature to 30°C at the up-shift to 384 re-start the cell cycle. 385 386