Nutritional cell cycle reprogramming reveals that inhibition of Cdk1 is required for proper MBF-dependent transcription

The nutritional environment determines the growth rate, cell size and cell cycle distribution. Because most laboratories use nutrient-rich media to grow cells, most published studies were performed under conditions of maximal growth. Fission yeast cells constantly monitor the nutritional environment, in particular nitrogen content, in order to coordinate cell growth with cell division and cell differentiation. In the presence of nitrogen, fission yeast cells grow and divide; in the complete absence of nitrogen, they stop growing and undergo cell cycle arrest and differentiation (Yanagida, 2009). Most laboratories grow fission yeast cells in nitrogen-rich media, where cells grow fast and divide with a large cell size after a long G2 phase. However, fission yeast cells can also grow and divide in nitrogen-poor media. When fission yeast cells are shifted from nitrogen-rich to nitrogen-poor medium their cell cycle is immediately reprogrammed and they divide with a small cell size. In this nitrogen-poor environment, G2 is short and G1 is long (Fantes and Nurse, 1977; Carlson et al., 1999; Petersen and Nurse, 2007; Chica et al., 2016).

Recently, our group has shown that the Greatwall (Ppk18 and Cek1 in fission yeast)–Endosulfine (Igo1 in fission yeast)–PP2A/B55 (Ppa1/Pab1 and Ppa2/Pab1 in fission yeast) pathway couples cell growth to the mitotic cell cycle machinery by regulating the activity of PP2A/B55, the main protein phosphatase that dephosphorylates the cyclin-dependent kinase-1 (Cdk1)/cyclinB substrates in fission yeast (Chica et al., 2016; Pérez-Hidalgo and Moreno, 2017). In nitrogen-rich media, TORC1 and PP2A/B55 activities are high and G2 is long; as a consequence, the cells are large. In nitrogen-poor media, TORC1 and PP2A/B55 activities are low, the cells enter mitosis with a reduced cell size, G2 is short and G1 is extended to maintain cell size homeostasis.

Extension of G1 in fission yeast depends on the presence of Rum1 and Ste9, two negative cell cycle regulators that maintain a low CDK activity during G1 before entry into S phase. Rum1 is a CDK inhibitor (CKI) of Cdk1/cyclin B complexes (Moreno and Nurse, 1994; Correa-Bordes et al., 1997; Benito et al., 1998), whereas Ste9 (initially named Srw1) is an activator of the anaphase-promoting complex/cyclosome (APC/C) that promotes the degradation of the B-type cyclins in G1 (Yamaguchi et al., 1997, 2000; Kitamura et al., 1998; Kominami et al., 1998; Blanco et al., 2000; Yamano et al., 2000). Cells lacking Rum1 and Ste9 divide with the same size and with the same generation time as wild-type cells (Moreno and Nurse, 1994; Yamaguchi et al., 1997; Kitamura et al., 1998; Kominami et al., 1998), indicating that Rum1 and Ste9 are dispensable in nitrogen-rich media. However, Rum1 and Ste9 are required for G1 arrest and for mating in minimal medium without nitrogen (Moreno and Nurse, 1994; Correa-Bordes et al., 1997; Benito et al., 1998; Kitamura et al., 1998; Blanco et al., 2000; Stern and Nurse, 1998; Martín-Castellanos et al., 2000).

Oscillatory changes in Cdk1 activity are sufficient to entrain periodic transcription through the cell cycle (Banyai et al., 2016). One of these transcription factors, MluI cell-cycle box (MCB) binding factor (MBF), the functional orthologue of mammalian E2F, regulates the expression of genes involved in the G1/S transition, such as cig2⁺ encoding the main G1/S cyclin, and in DNA replication, such as cdc18⁺, cdt1⁺ or cdc22⁺ (Baum et al., 1997; Kelly et al., 1993; Fernández-Sarabia et al., 1993; Hofmann and Beach, 1994; Lowndes et al., 1992; Ayté et al., 2001; Maqbool et al., 2003).

Here, we show that Rum1 and Ste9 proteins are essential in cells growing in nitrogen-poor media. In this nutritional environment, cells lacking Rum1 and Ste9 show premature entry into S phase, with reduced MBF activity. Rum1 and Ste9 are required to extend the G1 phase of the cell cycle by inhibiting Cdk1 activity and to set a proper wave of MBF activity at G1/S, which is crucial in providing high levels of expression of S phase genes in order to avoid DNA damage during S phase and G2. Therefore, under nitrogen restriction, inhibition of Cdk1 in G1 is essential in order to entrain MBF-dependent transcription at the onset of S phase.

Fission yeast has two cell size controls, one operating at the end of G1 before the onset of S phase and the other at the end of G2, before mitosis. In nitrogen-rich medium, EMM2 (Edinburgh minimal medium with 5 g l⁻¹ NH₄Cl) or MMGlu (Edinburgh minimal medium with 20 mM glutamate), the cell size required to initiate mitosis is about 14 µm. In this nutritional environment, the G2/M size control predominates and the G1/S cell size control is cryptic because the newly born daughter cells are larger than the minimal cell size required to initiate S phase. In nitrogen-poor medium, MMPhe (Edinburgh minimal medium with 20 mM phenylalanine as nitrogen source), wild-type fission yeast cells initiate mitosis with approximately 10 µm cell size, generating two daughter cells with a smaller cell size than the minimum required to initiate S phase. In this poor nutritional environment, fission yeast cells extend the G1 phase and grow until they reach the minimum cell size required to initiate S phase (Fig. S1, wt) (Nurse, 1975; Nurse et al., 1976; Nurse and Thuriaux, 1977; Nasmyth et al., 1979; Sveiczer et al., 1996).

In order to understand the molecular basis and the relevance of this G1 extension, we shifted wild-type fission yeast cells from a nitrogen-rich medium (MMGlu) to a nitrogen-poor medium (MMPhe) and determined the protein levels of Rum1 and Ste9, two functionally conserved G1 cell cycle inhibitors. Four hours after the shift, Rum1 protein levels increased in MMPhe, albeit not as much as in minimal medium without nitrogen (MM-N) (Fig. 1A, left panel) (Martín-Castellanos et al., 2000; Daga et al., 2003). In the same experiment, the Ste9 protein showed a shift in mobility (Fig. 1A, right panel), suggesting that it became dephosphorylated in MMPhe, as previously described in MM-N (Fig. 1A, right panel) (Blanco et al., 2000). Consistent with the accumulation of Rum1 and the dephosphorylation of Ste9, a 1C G1 population was observed 4 h after the shift, both in MMPhe and in MM-N (Fig. 1B and Fig. S1).

Unlike wild-type cells, mutant cells lacking rum1⁺, ste9⁺ or both rum1⁺ and ste9⁺ did not show the 1C G1 population in MMPhe. Instead, these mutants showed a population of cells with a DNA content between 1C and 2C, most likely corresponding to cells in S phase (Fig. 1C and Fig. S1, arrowheads). The double mutant rum1Δ ste9Δ showed more cells in S phase than the single mutants rum1Δ or ste9Δ.

To confirm that there were more cells in S phase in cells lacking Rum1 and Ste9 than in the wild type, we shifted wild-type and rum1Δ ste9Δ mutant cells from MMGlu to MMPhe and determined the percentage of cells in S phase by measuring the incorporation of the thymidine analogue 5-bromo-2′-deoxyuridine (BrdU) in fission yeast cells expressing the thymidine kinase gene. This nutritional downshift partially synchronises fission yeast cells because 30–40% of the cells are advanced into mitosis as a consequence of the inhibition of PP2A/B55, after the activation of Greatwall–Endosulfine pathway (Chica et al., 2016). The rum1Δ ste9Δ mutant showed more BrdU-positive cells than the wild type at different times after the nutritional shift, in particular during the first cell cycle in MMPhe (2 and 4 h) (Fig. 1D). This result revealed that the rum1Δ ste9Δ mutant is unable to delay the G1 phase of the cell cycle in MMPhe and shows premature entry into S phase. Taken together, the data indicate that Rum1 and Ste9 are required to extend the G1 phase of the cell cycle and to prevent premature S phase entry in nitrogen-poor medium.

Cells deleted for rum1⁺ and/or ste9⁺ behave similarly to wild-type cells and do not show any apparent phenotypes in nitrogen-rich media (Moreno and Nurse, 1994; Yamaguchi et al., 1997; Kitamura et al., 1998; Kominami et al., 1998; Blanco et al., 2000). When rum1Δ ste9Δ cells were shifted from MMGlu to MMPhe, they initially underwent cell size reduction similar to that seen for wild-type cells (Fig. 2A,B; 6 h, corresponding to the first cell cycle in MMPhe) but eventually they became elongated (Fig. 2A,B; 13 and 20 h, corresponding to the second and third cell cycles in MMPhe). The number of elongated cells in the rum1Δ ste9Δ culture increased steadily over time in MMPhe.

In fission yeast, cells accumulating unrepaired DNA lesions activate the DNA damage checkpoint to delay the cell cycle in G2 in order to repair the damaged DNA before mitosis (Caspari and Carr, 1999). Flow cytometry analysis confirmed that the elongated cells in rum1Δ ste9Δ were delayed in G2 (Fig. S2A, lower panel; Fig. S2B). Moreover, the rum1Δ ste9Δ mutant showed a high number of small cells with a DNA content between 1C and 2C (Figs S1 and S2C). This new population of cells seems to be the consequence of a slow S phase. To investigate the generation of DNA damage, we monitored the formation of Rad52:YFP nuclear foci that accumulate at the sites of double-strand breaks or collapsed DNA replication forks (Kim et al., 2000; Meister et al., 2005). The rum1Δ ste9Δ mutant cells began to accumulate Rad52:YFP nuclear foci after the first cell cycle in MMPhe (Fig. 2C,D; 6 h); the number and intensity of the foci and the percentage of elongated cells increased over time. After 24–48 h in MMPhe, about 50% of the cells were elongated and contained very intense Rad52:YFP foci (Fig. 2C,D). Cell elongation and Rad52:YFP nuclear foci were also observed in rum1Δ and ste9Δ single mutants after 20–48 h in MMPhe (Fig. S3A,B). Cells lacking Rum1 showed a percentage of Rad52:YFP foci very close to that of the double mutant rum1Δ ste9Δ, whereas in cells lacking Ste9 the percentage of Rad52:YFP foci was intermediate, between the wild type and rum1Δ (Fig. S3C). This result indicates that Rum1 plays a major role and Ste9 a minor role in preventing DNA damage in nitrogen-poor medium.

The fission yeast serine-threonine protein kinase ATR homologue, Rad3, is activated in response to DNA damage (Brondello et al., 1999). We examined the phenotype of the rad3Δ rum1Δ ste9Δ triple mutant in MMPhe. These cells showed a significant reduction in the number of elongated cells after 20 h in MMPhe compared with the double mutant rum1Δ ste9Δ (Fig. 3A,B), suggesting that the cell cycle delay in G2 was dependent on Rad3. As shown in Fig. 3C, rad3Δ rum1Δ ste9Δ mutant cells presented a significant percentage of cut and mis-segregation phenotypes, reminiscent of cells undergoing mitotic catastrophe. Moreover, flow cytometry analysis revealed a sub-G1 population of dead cells in the triple mutant rad3Δ rum1Δ ste9Δ that was not observed in the double mutant rum1Δ ste9Δ (Fig. 3D, arrow). Deletion of rad3⁺ also decreased the fitness of rum1Δ ste9Δ cells in MMPhe (Fig. S4). Together, these results suggest that the Rad3 protein kinase is required for the cell cycle delay in G2 necessary to allow the survival of cells lacking Rum1 and Ste9 in MMPhe.

Rad3 is the apical protein kinase that activates the checkpoint kinases Cds1 and Chk1 in response to DNA replication stress (Cds1) or to DNA breaks (Chk1) (Rhind and Russell, 1998). We compared the phenotypes of the rum1Δ ste9Δ cells lacking rad3⁺, chk1⁺ or cds1⁺ after 20 h in MMPhe (Fig. 3E). Cells from the triple mutant chk1Δ rum1Δ ste9Δ behaved similarly to rad3Δ rum1Δ ste9Δ cells in that they did not elongate and showed a similar percentage of cut and mis-segregation phenotypes (Fig. 3C,F). In fission yeast, activation of Rad3 in response to DNA damage leads to the phosphorylation and activation of Chk1 (Walworth and Bernards, 1996; Wan and Walworth, 2001; Capasso et al., 2002). We observed a mobility shift of phosphorylated Chk1 in two independent clones of rum1Δ ste9Δ cells growing in MMPhe for 17 h (Fig. 3G). Mild phosphorylation of Chk1 was detected as early as 4–6 h after the shift from MMGlu to MMPhe (Fig. 3H), indicating that rum1Δ ste9Δ cells activate the DNA damage response during the first cell cycle after the shift and that phosphorylation of Chk1 accumulates gradually in nitrogen-poor medium, reaching a maximum after 12 h.

Cds1 is the fission yeast kinase that becomes activated after DNA replication stress and is required to slow down S phase. Cds1 is phosphorylated and activated during S phase by Rad3 in response to DNA replication fork arrest (Lindsay et al., 1998). We failed to detect a band shift of phosphorylated Cds1 protein in rum1Δ ste9Δ cells, probably because activation of the DNA replication checkpoint was either weak or transient. Instead, we determined the number of elongated cells and the percentage of cells with Rad52:YFP nuclear foci in the cds1Δ rum1Δ ste9Δ triple mutant compared with the rum1Δ ste9Δ double mutant after 20 h in MMPhe. As shown in Fig. 3E and Fig. S5A,B, the triple mutant cds1Δ rum1Δ ste9Δ contained more elongated cells and a higher percentage of cells with Rad52:YFP nuclear foci than the double mutant rum1Δ ste9Δ. Consistent with this data, the rum1Δ ste9Δ mutant showed mild hydroxyurea sensitivity even in nitrogen-rich medium (Fig. S6). Together, these results suggest that in cells lacking Rum1 and Ste9 the Cds1 kinase plays a role in preventing DNA damage during S phase. Therefore, in the absence of Rum1 and Ste9, cells undergo DNA replication stress and activate the two checkpoint kinases, Cds1 and Chk1, leading to a slow S phase caused by mild activation of Cds1 and to a cell cycle delay in G2 as a result of activation of Chk1.

Rum1 and Ste9 are negative regulators of Cdk1 activity in G1. Therefore, rum1Δ ste9Δ cells should have higher Cdk1 activity than wild-type cells. To test whether Cdk1 upregulation in rum1Δ ste9Δ cells is the main cause of genomic instability in MMPhe, we analysed the phenotype of cells with reduced Cdk1 activity. For this purpose we used the analogue-sensitive (as) cdc2-asM17 allele that encodes a protein that can be inhibited by bulky ATP analogues (Aoi et al., 2014). We observed that cdc2-asM17 is a hypomorphic allele of cdc2 because cells expressing this mutant version of cdc2⁺ were slightly larger than wild-type cells in MMGlu and in MMPhe, even in the absence of ATP analogue inhibitors (Fig. 4A,B). The reduced Cdk1 activity of this mutant should affect primarily the G2/M transition, which has a threshold of Cdk1 activity higher than the G1/S transition (Coudreuse and Nurse, 2010). Thus, we reasoned that the cdc2-asM17 cells are larger because they have an extended G2 phase.

As shown in Fig. 4A–C, the cdc2-asM17 rum1Δ ste9Δ mutant showed fewer elongated cells and less DNA damage (measured as percentage of cells with Rad52:YFP nuclear foci) after 20 h in MMPhe than rum1Δ ste9Δ mutant cells. Consistent with this observation, flow cytometry analysis showed that the cdc2-asM17 rum1Δ ste9Δ cells did not contain the population of cells in S phase characteristic of the rum1Δ ste9Δ mutant (Fig. 4D). These results indicate that extension of G2 by reducing Cdk1 activity in cells lacking Rum1 and Ste9 may favour DNA repair before mitosis. A long G2 could be one of the reasons why rum1Δ ste9Δ cells do not show any apparent phenotypes in nitrogen-rich medium.

At the onset of mitosis, Cdk1/cyclinB kinase activity is opposed by PP2A/B55 phosphatase activity that inhibits Cdk1/cyclinB activation and antagonizes the phosphorylation of downstream targets of Cdk1 (Chica et al., 2016; Pérez-Hidalgo and Moreno, 2017). Deletion of the ppa2⁺ gene, encoding one of the two catalytic subunits of PP2A, generates cells smaller than wild-type cells (Fig. 5A,B). This semi-wee phenotype is caused by an accelerated entry into mitosis (Chica et al., 2016; Kinoshita et al., 1993; Navarro and Nurse, 2012). Thus, ppa2Δ cells showed greater 1C G1 population in MMPhe as an indirect effect of shortening G2 (Fig. 5C). We therefore tested the effect of deleting ppa2⁺ in rum1Δ ste9Δ. In MMPhe, we observed an increase in the number of elongated cells in ppa2Δ rum1Δ ste9Δ mutant cells compared with rum1Δ ste9Δ. The difference in cell size between ppa2Δ rum1Δ ste9Δ and ppa2Δ was larger than between rum1Δ ste9Δ and the wild type (2.31 versus 1.85 μm), although the ppa2Δ rum1Δ ste9Δ mutant cells did not reach sizes as large as those of the rum1Δ ste9Δ mutant (Fig. 5A,B). In addition, there was an increase in the number of cells undergoing DNA replication in the triple mutant ppa2Δ rum1Δ ste9Δ compared with the double mutant rum1Δ ste9Δ (Fig. 5C, arrows). Importantly, 7.29% of the ppa2Δ rum1Δ ste9Δ mutant cells showed cut and mis-segregation phenotypes in MMPhe (Fig. 5D), a slightly smaller percentage than rad3Δ rum1Δ ste9Δ (9.57%) (Fig. 3C) or chk1Δ rum1Δ ste9Δ cells (9.21%) (Fig. 3F). These results indicate that a reduction in PP2A activity in the rum1Δ ste9Δ mutant increases the number of cells delayed by the checkpoint and that some of these cells are unable to maintain the cell cycle arrest. These cells enter into mitosis and undergo mitotic catastrophe, indicating that PP2A activity is also important in preventing entry into mitosis in response to the activation of the DNA damage checkpoint, suggesting that both inhibition of Cdk1/cyclin B and activation of PP2A are required to delay the cell cycle in G2 upon DNA damage. In agreement with this idea, upregulation of PP2A phosphatase activity in the presence of DNA damage has been described in Xenopus egg extracts (Wang et al., 2015).

The correct supply of deoxyribonucleoside triphosphates (dNTPs) during S phase is crucial in maintaining genome integrity (Chabes and Stillman, 2007; Fleck et al., 2013). Recently, it has been shown that reduced MBF-dependent ribonucleotide reductase (RNR) expression generates DNA replication stress and checkpoint activation, phenotypes that can be reversed by increasing the dNTP pools (Pai et al., 2017). Low levels of RNR provide sufficient dNTPs for mitochondrial DNA synthesis and for DNA repair in quiescent cells and during the G1 phase of the cell cycle, but RNR levels and activity have to increase to high levels as cells enter S phase (Stillman, 2013).

In fission yeast, Spd1 inhibits assembly of RNR in G1. Spd1 is later degraded when the cells enter S phase or in response to DNA replication stress, which activates RNR to increase the dNTP pools (Liu et al., 2003, 2005; Håkansson et al., 2006). Therefore, we tested whether deletion of spd1⁺ could rescue the DNA damage phenotype of the rum1Δ ste9Δ cells in MMPhe by increasing the dNTP pools. We observed that rum1Δ ste9Δ spd1Δ cells were less elongated (Fig. 6A), showed less DNA damage (Fig. 6B) and grew slightly better in MMPhe (Fig. S4) than rum1Δ ste9Δ. This result suggests that rum1Δ ste9Δ cells may have a problem at the onset of S phase that leads to a reduction in the dNTP pools.

It has been shown that increasing the dNTP pools rescues the DNA damage phenotype of the rum1Δ ste9Δ double mutant; expression of cdc22⁺, encoding the large subunit of RNR, at G1/S is MBF-dependent; Spd1 is targeted for degradation by the CRL4-Cdt2 ubiquitin ligase at the onset of S phase; and Cdt2 is also a target of the MBF transcription factor (Hofmann and Beach, 1994; Liu et al., 2003, 2005; Håkansson et al., 2006; Holmberg et al., 2005). We therefore reasoned that defective MBF activation could be responsible for the reduced expression of RNR and for high levels of Spd1. Taking all this into account, we decided to explore the activity of the MBF transcription factor in wild-type and rum1Δ ste9Δ mutant cells growing in nitrogen-poor medium. MBF, the fission yeast functional orthologue of mammalian E2F, promotes transcription of cell cycle genes at the end of G1. Most MBF target genes encode proteins involved in the initiation of DNA replication, nucleotide biogenesis and the regulation of S phase (Bähler, 2005). Deregulation of MBF promotes genotoxic stress in S phase (Pai et al., 2017; Caetano et al., 2014; Pfister et al., 2015).

To test whether MBF-dependent transcription was affected in the rum1Δ ste9Δ mutant, we monitored the levels of two MBF targets (cdc18⁺ and cdc22⁺ mRNAs) using real-time quantitative polymerase chain reaction (RT-qPCR) in wild-type and rum1Δ ste9Δ cells at different times after the shift from MMGlu to MMPhe. As shown in Fig. 7A (left panel), cdc18⁺ and cdc22⁺ mRNA levels rose to a peak 4 h after the shift from nitrogen-rich (MMGlu) to nitrogen-poor (MMPhe) medium, when cells are undergoing S phase (Fig. 1B,C). Both cdc18⁺ and cdc22⁺ mRNA levels were significantly reduced in the rum1Δ ste9Δ mutant cells compared with the wild type (Fig. 7A). In nitrogen-rich medium (MMGlu, time 0), the levels of cdc18⁺ and cdc22⁺ mRNAs were slightly but reproducibly reduced in rum1Δ ste9Δ mutant cells compared with the wild type. After 4 h in nitrogen-poor medium, the mRNA levels of cdc18⁺ and cdc22⁺ dropped considerably in rum1Δ ste9Δ mutant cells (Fig. 7A). As a control, we determined the mRNA levels of cdc18⁺ and cdc22⁺ in cells lacking Rep2 (rep2Δ), a positive regulator of MBF that shows reduced MBF-dependent transcription (Fig. 7A, right panel) (Baum et al., 1997; Gaspa et al., 2016; Tournier and Millar, 2000).

We also monitored the levels of the Tos4 protein, another MBF target (Kiang et al., 2009), by western blot in wild-type and rum1Δ ste9Δ mutant cells in a similar time course experiment to that described above. In nitrogen-poor medium (MMPhe), Tos4:GFP protein levels were significantly reduced in the rum1Δ ste9Δ mutant compared with levels in wild-type cells (Fig. 7B). Again, in nitrogen-rich medium (MMGlu, time 0), rum1Δ ste9Δ mutant cells contained reduced levels of Tos4:GFP protein compared with wild-type cells (Fig. 7B). Moreover, we counted the number of Tos4:GFP-positive cells using fluorescent microscopy in MMGlu and at 3 and 4 h after the shift to MMPhe. As shown in Fig. 7C, in MMGlu and 3 or 4 h after the shift to MMPhe, rum1Δ ste9Δ cells showed fewer Tos4:GFP-positive cells than the wild type. In addition, the Tos4:GFP signal was less intense in the rum1Δ ste9Δ mutant than in the wild-type strain (Fig. 7D). Together, these results are consistent with a reduction in the amplitude of the MBF-dependent expression wave in cells lacking Rum1 and Ste9 in comparison with wild-type cells.

The core MBF transcription factor is composed of Cdc10 and the DNA-binding proteins Res1 and Res2. Additionally, Rep2 is a co-activator of MBF that it is required for high levels of MBF-dependent transcription, but not for transcription periodicity (Baum et al., 1997; Whitehall et al., 1999; Ayté et al., 1995; Zhu et al., 1997; Tahara et al., 1998). Nrm1 and Yox1 are co-repressors of MBF, forming part of a negative feedback loop to repress MBF activity in S phase, G2 and mitosis. Nrm1 and Yox1 accumulate during S phase, bind to MBF at promoters and repress MBF-dependent transcription outside of G1 (de Bruin et al., 2008; Caetano et al., 2011; Ivanova et al., 2011; Aligianni et al., 2009; Purtill et al., 2011). Deletion of nrm1⁺ increases MBF-dependent transcription, resulting in replication errors and genomic instability (Caetano et al., 2014).

We compared the phenotypes of wild-type, rum1Δ ste9Δ, rep2Δ (with downregulated MBF activity) and nrm1Δ (with upregulated MBF activity) cells in MMGlu and after 20 h in MMPhe. In nitrogen-rich medium (MMGlu), nrm1Δ mutant cells were larger than wild-type cells (Fig. 8A,B), showed a high percentage of nuclei with Rad52:YFP nuclear foci in MMGlu (Fig. 8C) and exhibited activation of Chk1, as previously described by Caetano et al. (2014). Interestingly, in nitrogen-poor medium (MMPhe), the phenotype of the nrm1Δ mutant was similar to the wild type (Fig. 8A,B), whereas cells lacking Rep2 presented a marked increase in cell size and high percentage of Rad52:YFP nuclear foci (Fig. 8A–C), similar to the rum1Δ ste9Δ mutant. Flow cytometry analysis revealed that the nrm1Δ mutant did not display a G1 population in MMPhe (Fig. 8D) because they divided with a greater cell size than wild-type cells (Fig. 8A,B). By contrast, the rep2Δ mutant presented more cells with 1C DNA content peak than the wild type as a result of delay in S phase entry (Fig. 8D).

Therefore, cells lacking Rep2 showed similar levels of DNA damage in MMPhe to the rum1Δ ste9Δ mutant (Fig. 8C). These results indicate that proper MBF regulation is important for maintaining genome integrity, and that downregulation of MBF in nitrogen-poor media has worse consequences than upregulation of MBF.

It has been reported that nrm1Δ cells exhibit upregulated MBF activity (Gaspa et al., 2016; Caetano et al., 2014). We constructed the triple mutant nrm1Δ rum1Δ ste9Δ to test whether upregulation of MBF by deleting nrm1⁺ could rescue the DNA damage phenotype of rum1Δ ste9Δ mutant cells. As shown in Fig. S7A, the nrm1Δ rum1Δ ste9Δ cells were shorter than rum1Δ ste9Δ cells in MMPhe (Fig. S7A), suggesting suppression of the rum1Δ ste9Δ phenotype by deleting nrm1⁺. The triple mutant nrm1Δ rum1Δ ste9Δ presented a phenotype similar to that of nrm1Δ cells, with less Rad52:YFP nuclear foci than rum1Δ ste9Δ cells in MMPhe (Fig. S7B). Deletion of nrm1⁺ also rescued the growth defect phenotype of the rum1Δ ste9Δ in MMPhe (Fig. S4). Flow cytometry analysis revealed that the S phase population present in rum1Δ ste9Δ cells in MMPhe was not detected in nrm1Δ rum1Δ ste9Δ, probably due to the long G2 phase of the nrm1Δ mutant (Fig. S7C, arrowhead). These results are consistent with the idea that deletion of rum1⁺ and ste9⁺ causes DNA replication stress in nitrogen-poor media because of downregulation of MBF activity. It is possible that deletion of nrm1⁺ rescues the phenotype of the rum1Δ ste9Δ mutant cells by increasing MBF activity or by extending the G2 phase, which allows more time for DNA repair, similar to what happens to cdc2-asM17 rum1Δ ste9Δ cells or when the rum1Δ ste9Δ mutant cells are grown in nitrogen-rich medium.

Here, we show that inhibition of Cdk1 activity in G1 by Rum1 and Ste9 is required in order to promote a proper wave of MBF-dependent transcription at the G1/S boundary, which is necessary for an efficient S phase. This function of Ste9 and Rum1 is particularly important in cells growing in nitrogen-poor media, where the G1 phase of the cell cycle has to be extended before cells can undergo DNA replication. Consistent with this idea, cells lacking Rum1 and Ste9 enter prematurely into S phase with significantly reduced expression of cdc18⁺, cdc22⁺ and tos4⁺ MBF targets.

Cdk1 protein kinase activity is high in mitosis, drops during anaphase and G1, rises again at the end of G1 to promote S phase and keeps increasing during G2 until the onset of mitosis (Coudreuse and Nurse, 2010). This oscillatory behaviour of Cdk1 activity is important for the periodic expression of cell cycle genes in fission yeast (Banyai et al., 2016). APC/C–Ste9-dependent degradation of Cdc13 and Rum1 inhibition of the Cdk1 kinase generate a window of low Cdk1 activity in G1 that it is crucial in fission yeast cells growing in nitrogen-poor medium, where the G1 phase of the cell cycle has to be extended. We also show that Rum1 and Ste9 are upregulated in nitrogen-poor media and that they are required to prevent genomic instability in subsequent cell cycles. In nitrogen-rich media, there is no apparent defect of the rum1⁺ and ste9⁺ deletions, although the double mutant shows mild hydroxyurea sensitivity. However, when rum1Δ ste9Δ mutant cells are grown in nitrogen-poor medium, they show a slow S phase, high levels of DNA damage and cell cycle delay in G2 promoted by activation of the DNA damage checkpoint. This phenotype is rescued by extending the G2 phase in cells with reduced CDK activity (cdc2-asM17, a hypomorph allele of cdc2⁺) or aggravated by reducing the G2 phase in cells with reduced PP2A activity (ppa2Δ, lacking one of the two catalytic subunits of PP2A).

At the onset of S phase, Cdk1 activates the MBF transcription factor (Banyai et al., 2016; Reymond et al., 1993), the fission yeast functional orthologue of mammalian E2F, which promotes the expression of genes involved in nucleotide biogenesis, S phase progression and DNA repair (Bähler, 2005). The MBF complex is bound to its target promoters throughout the cell cycle and it is tightly regulated by positive and negative factors that restrict MBF activation to the G1/S transition (Wuarin et al., 2002). Rep2 is an MBF transcriptional activator, whereas Nrm1 and Yox1 are transcriptional repressors of MBF during S phase and G2 (Baum et al., 1997; Aligianni et al., 2009; Nakashima et al., 1995; de Bruin et al., 2006). Although it has been described that mutant cells with increased MBF activity (yox1Δ or nrm1Δ) present genomic instability when they are cultured under standard laboratory conditions (nitrogen-rich media) (Caetano et al., 2014; Gómez-Escoda et al., 2011), here, we show that upregulation of MBF is less detrimental in nitrogen-poor medium (MMPhe). In fact, reduction of MBF activity in rep2Δ and in rum1Δ ste9Δ mutants cells generated high levels of DNA damage in nitrogen-poor medium (MMPhe), supporting the idea that, in nitrogen-poor conditions, higher levels of MBF-dependent transcription activity are required at the end of G1 for an efficient S phase. Moreover, deletion of nrm1⁺, which upregulates MBF activity, partially rescued the genomic instability of rum1Δ ste9Δ cells in MMPhe.

Our findings are in agreement with recent results reporting that loss of the Set2 histone H3 lysine 36 (H3K36) methyl transferase generates reduced MBF-dependent ribonucleotide reductase (cdc22⁺) expression, causing a drop in the dNTP pools, abnormal replication origin firing, S phase delay and genotoxic stress (Pai et al., 2017). As in the rum1Δ ste9Δ mutant, the delay in S phase and the replication stress can be supressed in set2Δ cells by increasing the dNTP levels by deleting the MBF repressor nrm1⁺ (our work) or yox1⁺ (Pai et al., 2017) or the RNR inhibitor spd1⁺ (Liu et al., 2003; Håkansson et al., 2006).

Sic1, the budding yeast orthologue of Rum1, is stabilised in G1 in nitrogen-poor media (Moreno-Torres et al., 2015). Sic1 promotes origin licensing during G1 by inhibiting CDK activity (Lengronne and Schwob, 2002; Tanaka and Diffley, 2002). Furthermore, Cdh1 (the budding yeast orthologue of Ste9) together with Sic1 is required to prevent chromosome instability by promoting efficient origin firing (Ayuda-Durán et al., 2014). In animal cells, acute depletion or permanent ablation of Cdh1 causes genomic instability in primary mouse embryonic fibroblasts by slowing down DNA replication fork movement and increasing origin activity (García-Higuera et al., 2008; Garzón et al., 2017). Moreover, aberrant activation of the Rb–E2F pathway, which is very common in many cancers, leads to overexpression of cyclin E, depletion of the dNTP pool and inefficient DNA replication (Bester et al., 2011). In these cells, DNA replication dynamics are altered and S phase takes longer as a result of reduced licensing and increased origin firing (Jones et al., 2013). However, there is no published evidence, either for budding yeast or animal cells, of the connection between CDK inhibition in G1 and G1/S transcription. In the future, it is important to establish whether this connection is also conserved in all eukaryotic cells and whether reduced MBF/E2F activity also generates DNA replication stress, considering that Cdk1 regulation, MBF/E2F-dependent transcription and S phase activation are highly conserved processes.

In summary, we have found that the cell cycle inhibitor Rum1 and the APC/C co-activator Ste9 are upregulated in nitrogen-poor medium. These two proteins are required to inhibit Cdk1 activity in G1, which delays the G1/S transition and facilitates efficient DNA replication by regulating the amplitude of MBF activity. In nitrogen-poor medium, proper levels of MBF-dependent transcription are necessary to generate an optimal S phase.

The fission yeast strains used in this study are listed in Table S1. Fission yeast cells were grown and manipulated genetically according to standard protocols (Moreno et al., 1991). Genetic crosses were performed on malt extract agar plates. All the strains containing rum1⁺ and ste9⁺ deletions were obtained by direct transformations, because of the sterility of both deletions, using a PCR-based strategy (Bähler et al., 1998). Oligonucleotides with 80 bases of homology to regions before the initiation codon and after the stop codon were used to amplify the kanMX6, natMX6 or hphMX6 marker sequences from plasmids pFA6a-kanMX6/natMX6/pFA6a-hphMX6. The PCR product was used to transform a given strain.

Cells were grown overnight at 32°C in yeast extract supplemented with adenine, leucine, histidine, lysine and uracil (YES) and then transferred to Edinburgh minimal medium containing 20 mM glutamate (MMGlu) instead of 5 g l⁻¹ NH₄Cl, as a nitrogen source. MMGlu is considered a nitrogen-rich medium. The generation time of wild-type cells in MMGlu at 32°C is 2.5 h. Cells grown in MMGlu for 1 day were transferred to minimal medium containing 20 mM phenylalanine (MMPhe) or to minimal medium without nitrogen (MM-N). MMPhe is considered a nitrogen-poor medium. The generation time of wild-type cells in MMPhe at 32°C is 6 h. In all cases, cells grown to mid-exponential phase (5×10⁶–8×10⁶ cells ml⁻¹) were centrifuged and washed in the destination medium. For hydroxyurea (HU) sensitivity experiments, HU (Sigma-Aldrich) was added to YES plates from a 1 M stock solution at a final concentration of 5 or 6 mM. All experiments were performed with prototrophic strains to avoid the addition of supplements (amino acids or uracil, which could be used as a nitrogen source) to the minimal medium.

For cell size measurements and Tos4:GFP detection, 2×10⁶ exponentially growing cells were washed in phosphate-buffered saline (PBS) and resuspended in 3.5 µl of 50 µg ml⁻¹ Blankophor (Bayer) and examined under a Nikon Eclipse 90i fluorescence microscope equipped with a Plan/Apo 603 oil objective lens and a Hamamatsu ORCA-ER camera. Live images were acquired with MetaMorph software (Molecular Devices). Cell length was measured from pictures of 600 cells. Tos4:GFP-positive cells were quantified from 500 cells in three independent experiments. The Tos4:GFP intensity signal was obtained from 50 cells using ImageJ (NIH).

For Rad52:YFP nuclear foci detection, 500 µl of cells growing in exponential phase in MMGlu or MMPhe were collected and Rad52:YFP foci were visualized using a DeltaVision microscope and SoftWoRx Resolve 3D software (Applied Precision). The percentages were measured from maximum-intensity projections of 12 0.3 μm z-sections (Image J, NIH) of 500 nuclei. Rad52:YFP nuclear foci were represented as the mean of nuclei with Rad52:YFP foci, with the standard deviation from three independent experiments.

For the detection of aberrant mitosis, samples of 2×10⁶ cells fixed in 70% (v/v) ethanol were washed in PBS and resuspended in 2.6 µl of 1 µg ml⁻¹ 4′,6-diamidino-2-phenylindole (DAPI) and 1.1 µl of 50 µg ml⁻¹ Blankophor (Bayer) and examined under a Leica DM XRA or Nikon Eclipse 90i fluorescence microscope equipped with a Hamamatsu ORCA-ER camera. The percentage of aberrant mitosis was determined from pictures of 950–1000 cells using ImageJ (NIH).

BrdU incorporation was performed using the h⁺ ura4-D18::P3nmt1-Tk-ura4⁺ strain and following the immunofluorescence protocol (Hodson et al., 2003) with modifications. Cells were labelled for 20 min in MMGlu or MMPhe without thiamine containing 2.5 μM BrdU (Becton-Dickinson) at 32°C. DNA denaturation was performed in 1 ml of 2 M HCl, 0.1% Triton X-100 and incubated for 30 min. Cells were neutralized with 0.1 M sodium tetraborate, pH 9. BrdU-positive cells were counted in pictures of 900 nuclei (Image J, NIH).

Protein extracts were obtained using trichloroacetic acid extraction (Foiani et al., 1994). For western blot analysis, anti-Rum1 (1:250; Benito et al., 1998), anti-Ste9 (1:200; Blanco et al., 2000), anti-HA (1:500; 12CA5, Roche), anti-GFP (1:1500; A11122, Molecular Probes) and anti-tubulin (1:10,000; a gift from Dr Keith Gull, University of Oxford, UK) antibodies were used. The secondary antibodies were HRP-conjugated anti-mouse IgG (1:2500; NA-931, Amersham) and anti-rabbit IgG (1:3000; NA-934, Amersham). Detection was performed using the enhanced chemiluminescence procedure (ECL kit).

Total RNA was isolated from 1×10⁸ cells from exponential phase using an RNeasy Mini Kit (Qiagen) as described by the manufacturer. cDNA synthesis was performed with the SuperScript II First-Strand Synthesis System (Invitrogen) in a final volume of 50 μl using 3 μg of RNA previously treated with DNAse I (Invitrogen). cDNA (1 µl) was then used for the quantitative reactions in an Applied Biosystems 7300 Real-Time PCR System. For each reaction, the SYBR Premix Ex Taq (TaKaRa) reagent was used with oligonucleotide concentrations of 0.2 μM. Specific oligonucleotides were as follows: forward 5′-ACACACTCAAGATGTGTGCTATGAT-3′ and reverse 5′-CATTAGTTCCAGCAATATACGAACC for cdc22⁺; forward 5′-TAAATTACCCACAACACCTCAAACT-3′ and reverse 5′-ATGGAACAGGATTACATGTACGATT-3′ for cdc18⁺; and forward 5′-CCTTGCTTGTTGACTGAGGCTC-3′ and reverse 5′-GCAACATAAAAGGCAGGTGCAT-3′ for the actin gene. The thermal cycling conditions were as follows: 95°C for 45 s, 40 cycles of 95°C for 5 s and 60°C for 31 s, followed by a dissociation step at 95°C for 15 s, 60°C for 1 min and 95°C for 15 s. Reactions were run in triplicate and negative controls without reverse transcriptase or cDNA were included in each run. Determination of the PCR efficiency was performed from serial dilutions of wild-type S. pombe genomic DNA (1:10, 1:100, 1:1000, 1:10,000, 1:100,000) to generate a standard curve for each reaction. Expression ratios were calculated according to the mathematic model described by Pfaffl (2001). The samples were first normalized against the constitutively expressed actin gene, following normalization against the wild-type strain grown in MMGlu. The experiments were performed at least twice with cDNA from different biological repeats.

Column graphs show mean±s.d. Two-tailed Student's t-tests were used to calculate significant differences between two groups and one-way ANOVA between more than two groups; the Prism 6 software (GraphPad) was used. A P value of 0.05 or lower was regarded as statistically significant.

We thank all members of the S.M. laboratory, University of Salamanca for helpful discussions. We are grateful to J. Ayté, T. Carr, R. Daga, H. Masukata and M. Sato for strains.