Replication stress induced by the ribonucleotide reductase inhibitor guanazole, triapine and gemcitabine in fission yeast

Abstract

Schizosaccharomyces pombe is an established yeast model for studying the cellular mechanisms conserved in humans, such as the DNA replication checkpoint. The replication checkpoint deals with replication stress caused by numerous endogenous and exogenous factors that perturb fork movement. If undealt with, perturbed forks collapse, causing chromosomal DNA damage or cell death. Hydroxyurea (HU) is an inhibitor of ribonucleotide reductase (RNR) commonly used in checkpoint studies. It produces replication stress by depleting dNTPs, which slows the movement of ongoing forks and thus activates the replication checkpoint. However, HU also causes side effects such as oxidative stress, particularly under chronic exposure conditions, which complicates the studies. To find a drug that generates replication stress more specifically, we tested three other RNR inhibitors gemcitabine, guanazole and triapine in S. pombe under various experimental conditions. Our results show that guanazole and triapine can produce replication stress more specifically than HU under chronic, not acute drug treatment conditions. Therefore, using the two drugs in spot assay, the method commonly used for testing drug sensitivity in yeasts, should benefit the checkpoint studies in S. pombe and likely the research in other model systems.

Effects of ribonucleotide reductase inhibitors on Schizosaccharomyces pombe.

Introduction

During the S phase of the cell cycle, numerous endogenous and exogenous factors can perturb accurate DNA replication by the replication machinery, causing replication stress. These include the difficult-to-replicate regions, collision with the transcription machinery, DNA damage on the templates, and insufficient supply or imbalanced dNTPs, the precursors for DNA replication. If not dealt with properly, perturbed replication forks collapse, causing chromosomal DNA damage, genome instability or even cell death. When the replication stress increases, the replication checkpoint (also called the S phase checkpoint) is activated to overcome the stress (see reviews in Yazinski and Zou 2016, and Iyer and Rhind 2017b). The activated replication checkpoint mobilizes several pathways such as increased production of dNTPs, cell cycle delay, stabilization of replication forks and suppression of late firing origins, which all work in concert to minimize the mutation rate and ensure accurate duplication of the genome. Consistent with its importance in genome stability, the replication checkpoint is highly conserved from yeasts to humans. Defects in this pathway cause genetic diseases or cancer predisposition syndromes. Although the replication checkpoint is crucial for genome stability in all eukaryotic organisms, the exact mechanism of how the replication checkpoint initiates at the stressed forks remains to be fully understood (Bandhu et al. 2014, Yue et al. 2014, Ma et al. 2020).

Studies of the replication checkpoint in a genetically tractable model organism require a reliable agent that can generate replication stress without a significant side effect. As mentioned earlier, various types of DNA damaging agents such as methyl methanesulfonate or ultraviolet light can also perturb the replication, causing replication stress. However, DNA damage only pauses a subset of ongoing forks on the leading strand template and, if it occurs outside the S phase, DNA damage also provokes the DNA damage checkpoint pathway. This may complicate the study, leading to an ambiguous description of the checkpoint mechanisms (Iyer and Rhind 2017a). The most reliable replication stress inducer currently used in many laboratories is hydroxyurea (HU). HU has been of clinical and scientific interest for >100 years (see reviews in Spivak and Hasselbalch 2011, and Singh and Xu 2016). It perturbs DNA replication by inhibiting ribonucleotide reductase (RNR), a highly conserved enzyme that produces dNTPs for DNA replication and repair (Nordlund and Reichard 2006). RNR contains a catalytically essential diferric tyrosyl radical inside the smaller subunits. HU suppresses RNR by quenching the tyrosyl radical and thus slows down the polymerase movement at all ongoing forks, generating replication stress (Krakoff et al. 1968, Larsen et al. 1982). Consistent with this mechanism, HU resistance has been observed in cells overexpressing RNR or expressing a mutant RNR (Akerblom et al. 1981, Choy et al. 1988, Sneeden and Loeb 2004). Activation of the replication checkpoint is the primary response in all eukaryotic model organisms. In addition to generating replication stress, HU is also commonly used in the laboratories for S phase cell cycle synchronization due to its reversible action on RNR.

While screening new replication checkpoint mutants in the fission yeast Schizosaccharomyces pombe with enhanced sensitivity to HU, we found that ∼30% of the screened mutants whose lethality is unrelated to the HU-induced replication stress, but caused by other mechanisms such as oxidative stress or cytokinesis arrest (Xu et al. 2016, Singh and Xu 2017). This result suggests that HU causes side effects under certain experimental conditions, which may compromise the description of the cell-killing mechanism and hamper the checkpoint studies. To search for an agent that can produce the replication stress more specifically, we have tested three other RNR inhibitors guanazole (GZ), triapine (3-AP), and gemcitabine (GEM) in S. pombe and compared their effects with HU under various assay conditions. Our results show that while GEM does not show any significant cytotoxic effect in wild-type or mutant S. pombe, 3-AP and GZ generate the replication stress more specifically than HU under chronic drug exposure conditions such as the spot assay. Considering the lower cost of GZ, using GZ in the spot assay to produce more specific replication stress should be helpful to the checkpoint research in S. pombe. Under acute drug treatment conditions, 3-AP is not cytotoxic, whereas GZ shows a significant side effect, arresting the cells at G2/M that indirectly suppresses its effect on DNA replication at S phase. Therefore, using GZ or 3-AP in chronic exposure conditions and HU in acute treatment should complement each other and thus benefit the checkpoint studies by specifically inducing replication stress under various experimental conditions in S. pombe and likely in other model organisms for the related research.

Materials and methods

Yeast strains and chemical agents

Standard methods were used for yeast cell cultures (Moreno et al. 1991). Yeast strains used in this study are listed in Table1. The media used for cell culturing was YE6S (0.5% yeast extract, 3% glucose and the six essential supplements, adenine, uracil, histidine, lysine, leucine and arginine). Stock solutions of GZ (Alfa Aesar, Haverhill, MA, USA or Sigma-Aldrich, St. Louis, MO, USA) and GEM (AmBeed, Inc., Arlington Hits, IL, USA) were prepared in the YE6S medium. 3-AP (Apexbio Tech., LLC, Houston, TX, USA) was dissolved in DMSO. HU (SIgma-ALdrich, St. Louis, MO, USA) was dissolved in deionized water.

Drug sensitivity

For chronic drug treatment, we employed the spot assay, in which 2 × 10⁷ cells/ml of logarithmically growing S. pombe were diluted in 5-fold steps and spotted onto YE6S plates containing HU, GZ, GEM or 3-AP at the indicated concentrations. The control plates for 3-AP received an equal amount of DMSO as DMSO suppresses cell growth at higher concentrations (data not shown). The plates were incubated at 30°C for 3 days and then photographed. For acute drug treatment, we employed the spot assay and the colony recovery assay. For the acute spot assay, the drugs were added to 2 × 10⁶ cells/ml logarithmically growing cultures at the indicated concentrations. Every 2 hours during the drug treatment, a small aliquot of the cultures was removed, washed once, diluted 10 times in deionized water, and then dropped on YE6S plates. The plates were incubated at 30°C for 3 days and then photographed. For the colony recovery assay (Enoch et al. 1992), after adding the drugs to their final concentrations, a small aliquot of the cultures was removed every hour during incubation, diluted 1000-fold in deionized water, spread onto YE6S plates and incubated at 30°C for the cells to recover for 3 days. The colonies were counted and presented as percentages of the untreated cells.

Western blotting

Phospho-specific antibody against phosphorylated Mrc1-Thr⁶⁴⁵ was generated using the chemically synthesized phosphopeptides described in our previous studies (Xu et al. 2006, Xu and Kelly 2009). The S. pombe cells were fixed in 15% TCA on ice overnight and lysed by a mini-bead beater. The lysate was separated by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS–PAGE) on an 8% gel and transferred to a nitrocellulose membrane for western analysis. The membrane was first blotted with the phosphor-specific antibody for 3 h to reveal phosphorylated Mrc1. The blot was stripped, extensively washed, then reprobed with the antibody against Mrc1 to reveal the protein. The blotting signals were detected by electrochemiluminescence using the ChemiDoc XRS imaging system (Bio-Rad). Band intensities were quantified using ImageLab (Bio-Rad).

Flow cytometry

0.5 OD cells were collected by centrifugation and fixed with 70% ice-cold ethanol overnight at 4°C. The fixed cells were treated with 0.1 mg/ml RNase A in 50 mM sodium citrate at 37°C for ≥7 h, stained with 4 μg/ml propidium iodide, extensively vortexed, and analyzed by Accuri C6 flow cytometer using FL2-A channel. The data were analyzed using FCS Express 4Flow software.

Microscopy

Schizosaccharomyces pombe cells were collected by pulse centrifugation, dropped onto a glass slide, and fixed by briefly heating at 70°C. The cells were stained with Blankophor and Hoechst before examination under the fluorescent microscope (Olympus EX41). Images were captured with an IQCAM camera (Fast1394) using Qcapture Pro 6.0 software and extracted into Photoshop (Adobe) to generate the Fig.

Results

Effect of the RNR inhibitors on cell growth determined by spot assay

Spot assay is a method commonly used in many laboratories for assessing the inhibitory effects of genotoxins or chemical agents on cell growth. It is fast, easy, cost effective and widely used in genetic screens. Furthermore, it can be easily scaled up for genome-wide studies. The spot assay is usually performed by harvesting the logarithmically growing cells, diluted in a series of steps, and then spotted on plates containing a genotoxin or a chemical agent. The plates are usually incubated at an appropriate temperature for a period of time so that the cell growth inhibitory effects of the genotoxin or chemical agent can be examined semi-quantitatively.

We examined the inhibitory cell growth effect of GZ, GEM and 3-AP (Fig. 1) in fission yeast by spot assay. HU was included in the experiment for comparison. These RNR inhibitors are all commercially available and have been used in many laboratories for related research. To examine the replication stress induced by the inhibitors, we used following S. pombe strains: wild-type (TK48) cells, the checkpoint mutants rad3∆ (NR1826), mrc1∆ (YJ15), cds1∆ (GBY191), chk1∆ (TK197), and the two metabolic mutants erg11-1 (YJ1296) and hem13-1 (APS19) with a defective biosynthesis pathway of ergosterol and hemin, respectively (Xu et al. 2016, Singh and Xu 2017) (see Table 1). Schizosaccharomyces pombe Cds1 is the homolog of Rad53 in budding yeast and Chk2 in human cells. It is the effector kinase of the replication checkpoint pathway. Chk1, on the other hand, is the effector kinase responsible for activation of the DNA damage checkpoint, which mainly functions during the G2 phase, the longest cell cycle stage of S. pombe. Rad3 is the ortholog of human ataxia telangiectasia and Rad3 related and budding yeast Mec1. It is the master checkpoint kinase that activates Cds1 via the mediator of replication checkpoint Mrc1 in the presence of replication stress or Chk1 when DNA damage occurs outside of the S phase or stressed replication forks collapse such as the situation inside the HU-treated mrc1∆ or cds1∆ cells. The second checkpoint sensor kinase Tel1, the ataxia telangiectasia mutated ortholog in humans, plays a minimal role in the checkpoints in S. pombe. Deletion of rad3 sensitizes S. pombe to replication stress and DNA damage due to a lack of the two checkpoint pathways. Because the activation of the Rad3-Mrc1-Cds1 replication checkpoint pathway is essential to overcome replication stress, the sensitivity of rad3∆,mrc1∆ and cds1∆ cells is an excellent readout of the replication stress. The two metabolic erg11-1 and hem13-1 mutants are highly sensitive to HU in the spot assay. However, they are resistant to acute HU treatment in liquid cultures as they are not killed by the replication stress, but by oxidative stress or cytokinesis arrest induced by HU (Xu et al. 2016, Singh and Xu 2017) (see later). Therefore, the metabolic mutants are included as the indicators of other potential stresses that are induced by the three RNR inhibitors.

Figure 1.

Chemical structures of hydroxyurea, guanazole, triapine and gemcitabine, the four RNR inhibitors used in this study. The structures are drawn by using ChemDraw 20.1 (PerkinElmer Informatics, Inc.).

Table 1.

List of S. pombe strains used in this study.

Strain	Genotype	Sources
TK48	h– leu1-32 ade6-M216	Kelly lab
NR1826	h– rad3::ura4 leu1-32 ura4-D18 ade6	Russell lab
TK197	h+ chk1::ura4 leu1-32 ura4-D18 ade6-M210	Kelly lab
GBY191	h+ cds1::ura4 leu1-32 ura4-D18 ade6	Lab stock
YJ15	h+ mrc1:ura4 leu1-32 ura4-D18 ade6-M210	Lab stock
YJ1296	h+ erg11-G189D:ura4 leu1-32 ura4-D18 ade6-M210	Lab stock
APS19	h– hem13-1(T263I) cds1-6his2HA(Int) leu1-32 ura4-D18 ade6-M210	Lab stock

As shown in the top left panels of Fig. 2, while rad3∆,mrc1∆ and cds1∆ cells were sensitive to HU, chk1∆ cells were less sensitive, suggesting that the replication stress induced by HU is mainly dealt with by the replication checkpoint, not the DNA damage checkpoint pathway. However, chk1∆ cells were also sensitive when HU was increased to 5 mM, indicating that although the replication checkpoint is functional, the DNA damage checkpoint is required to deal with a minimal amount of DNA damage caused by collapsed forks. Therefore, the rad3∆ mutant is more sensitive to HU than mrc1∆ and cds1∆ cells. Under replication stress, Mrc1 contributes to ∼90% of Cds1 phosphorylation by Rad3 (Yue et al. 2011). The slightly higher resistance to HU and GZ observed in mrc1∆ than in cds1∆ cells is likely due to the basal ∼10% phosphorylation of Cds1 by Rad3. Although the spot assay has several abovementioned advantages, it requires chronic exposure of the cells to the tested drugs (∼3 days for S. pombe). Therefore, the cytotoxic effect of HU observed in a previously unknown mutant may not be caused by the replication stress. For example, the erg11-1 and hem13-1 mutants are sensitive to HU in the spot assay (lower part of the top left panels in Fig. 2) and, the sensitivity is comparable or even higher than rad3∆ cells. However, they are not sensitive to HU in acute treatment in liquid cultures (Xu et al. 2016, Singh and Xu 2017) (see Fig. 3C later), indicating that in addition to replication stress, HU also causes other stresses, leading to the chronic cell death of the metabolic mutants. To find an RNR inhibitor that can produce replication stress more specifically than HU in the spot assay, we examined GZ, GEM, and 3-AP under similar conditions. We found that while GEM did not show any inhibitory effect on cell growth to all tested strains at concentrations up to 77.5 mM (lower left panels, results with lower GEM concentrations are not shown), GZ and 3-AP showed a much higher cell-killing effect to rad3∆ and cds1∆ than chk1∆ and wild-type cells, suggesting that GZ and 3-AP can also induce the replication stress. Importantly, unlike HU that kills rad3∆ and the two metabolic mutants indistinguishably, GZ and 3-AP showed a much less cell-killing effect on the two metabolic mutants under similar conditions (compare 15 mM GZ and 0.25 mM 3-AP with 1.25 mM HU). These results show that GZ and 3-AP can generate the replication stress more specifically than HU under chronic exposure conditions. Because of the lower cost, using GZ in the spot assay to eliminate the complications induced by HU under chronic exposure conditions should benefit the checkpoint studies in fission yeast. To show that it is GZ, not the impurity of the sample, that produces the replication stress, we compared the cytotoxic effects of the GZ from two different manufacturers. The results are very similar, if not identical (not shown). Interestingly, the two metabolic mutants also showed a sensitivity to 30 mM GZ and 0.75 mM 3-AP, suggesting that at higher doses, the two inhibitors also generate other stresses (see later).

Figure 2.

The drug sensitivities of wild-type, rad3∆,mrc1∆, cds1∆,chk1∆,erg11-1 and hem13-1 were determined by spot assay. A series of 5-fold dilutions of the logarithmically growing wild-type S. pombe or the cells with the indicated mutations were spotted on YE6S plates or plates containing the drugs at the indicated concentrations. The plates were incubated at 30°C for 3 days and then photographed. While HU, GZ and GEM were directly dissolved in the medium, 3-AP was dissolved in DMSO, which has a negative growth effect in S. pombe at high concentrations (not shown). A representative of repeated experiments is shown.

Figure 3.

Acute drug sensitivity of wild-type and rad3∆ cells determined by spot assay and colony recovery assay. (A) Wild-type S. pombe and the rad3∆ mutant cells were incubated with HU or 3-AP in YE6S medium at the indicated concentrations. Every 2 hours during the drug treatment, a small aliquot of the cultures was removed. The cells were washed once and spotted on an YE6S plate. The plates were incubated at 30°C and then photographed. The experiment was repeated two times and a representative plate is shown. (B) Wild-type and rad3∆ cells were treated with 15 mM HU and the increasing concentrations of GZ. The cell recovery was examined as in panel (A). Dash line indicates discontinuity. (C) The sensitivities of wild-type, rad3∆,cds1∆ and erg11-1 cells to HU and GZ were determined by colony recovery assay. The logarithmically growing cells were treated with 15 mM HU (left panel) or 200 mM GZ (right panel) in the YE6S medium. Every hour during the drug treatment, an aliquot of the cultures was removed, diluted 1000 times in distilled water and the cells were spread onto three YE6S plates. The plates were incubated at 30°C for 3 days for cells to recover. The colonies were counted and shown in percentage against time. Error bars represent the means and standard deviations of triplicates.

Acute inhibitory effects of GZ and 3-AP on cell growth

We then investigated whether GZ and 3-AP can produce acute replication stress in liquid cultures. Because GEM did not show any inhibitory effect on cell growth to the checkpoint mutants in the spot assay, it was not further examined in this study. For assessing the acute drug sensitivity, we employed two commonly used methods: acute spot assay and colony recovery assay. In the acute spot assay, GZ or 3-AP was added to the logarithmically growing wild-type and rad3∆ cells at the indicated concentrations. For comparison, the cultures were also treated with 15 mM HU or 6–10 times higher than the concentrations used in the spot assay. The cultures were incubated at 30°C. Every hour during the drug treatment, a small aliquot of the culture was removed, washed once, diluted 10-fold and the cells were spotted on YE6S plates. The plates were incubated at 30°C for 3 days for cell recovery. As shown in Fig. 3A, while the HU showed a significant cell-killing effect to rad3∆, not the wild-type cells after 2 h treatment, 3-AP did not show any cytotoxic effect during the 8 h long treatment at 5 mM and 7 mM, or 9.3–28 times higher than used in the spot assay. This result suggests that 3-AP may penetrate S. pombe cells much slower than HU (see Discussion). On the contrary, GZ showed a robust cell-killing effect to rad3∆, not the wild-type cells from 50 to 250 mM or 1.7–13.3 times higher than used in the spot assay (Fig. 3B). This result shows that unlike 3-AP, GZ can induce acute replication stress like HU in liquid cultures. Interestingly, although the concentrations of GZ were increased from 100 mM to 250 mM, the cell-killing effect did not increase but slightly decreased and never reached the cytotoxic level of HU. This result show that consistent with the results by spot assay (Fig. 2), GZ may increasingly cause another stress that suppresses the acute replication stress, leading to the slightly reduced cytotoxicity (see later).

To further investigate the acute replication stress induced by GZ, we examined its cytotoxicity by colony recovery assay. In this assay, 15 mM HU or 200 mM GZ was added to the cultures containing logarithmically growing wild-type, rad3∆, cds1∆ or erg11-1 cells. The cultures were incubated at 30°C. Every hour during the treatment, a small aliquot of the cultures was removed, diluted 1000-fold and the cells were spread onto three YE6S plates for recovery. The recovered colonies were counted to calculate the cell survival rates in percentages. In the presence of HU (Fig. 3C, left panel), cds1∆ cells were sensitive but less than rad3∆, which is consistent with the apex function of Rad3 in the checkpoint pathways. The erg11-1 cells, although very sensitive to HU in the spot assay (Fig. 2), were resistant to HU almost like the wild-type cells, confirming our previous report (Xu et al. 2016). In the presence of GZ, rad3∆ and cds1∆ cells were much more sensitive than wild-type and the erg11-1 cells (Fig. 3C, right panel), which confirms the induction of the acute replication stress. However, unlike the HU treatment, the sensitivity of rad3∆ to GZ is quite similar to that of cds1∆ cells. This result, together with that shown Fig. 2 and the acute spot assay in Fig. 3B, suggests that in addition to the replication stress, GZ causes another type of stress that arrests the cell at G2/M, which reduces the sensitivity of rad3∆ to a similar level as in cds1∆ cells (see later).

Effect of GZ on the replication checkpoint

Next, we examined Rad3-dependent phosphorylation of Mrc1 (Claspin in humans) using the phospho-specific antibodies described in our previous studies (Xu et al. 2006, Xu and Kelly 2009, Yue et al. 2011). Since 3-AP did not show any cytotoxic effect under acute drug treatment conditions, it was not included in this experiment. In response to replication stress, Rad3 phosphorylates two functionally redundant residues Thr⁶⁴⁵ and Thr⁶⁵³ in the mid of Mrc1. Phosphorylated Mrc1 facilitates the phosphorylation of Thr¹¹ in Cds1 by Rad3, which promotes the autophosphorylation of Cds1-Thr³²⁸ in the activation loop and the autoactivation of the effector kinase in the replication checkpoint pathway (Xu et al. 2006, Xu and Kelly 2009). Therefore, phosphorylation of Mrc1 by Rad3 is an excellent molecular marker for the stressed forks.

We first examined the phosphorylation of Mrc1-Thr⁶⁴⁵, a representative of the two functionally redundant phosphorylation sites in Mrc1, in wild-type cells treated with increasing concentrations of GZ (Fig. 4A). After the cells were treated with GZ for 3 h or about one cell cycle time, Mrc1 phosphorylation was significantly increased in a concentration-dependent manner (Fig. 4B, blue line). At the 200 to 300 mM range, the levels of Mrc1 phosphorylation were comparable to the cells treated with 15 mM HU. This result is consistent with the observed sensitivity of rad3∆ and cds1∆ cells to GZ under both chronic and acute conditions (Figs 2 and 3). Because Mrc1 is expressed specifically in the S phase (Xu et al. 2006) and the activated replication checkpoint stimulates the expression of Mrc1 (Ivanova et al. 2013), the protein level of Mrc1 is much higher in the drug-treated cells than in untreated cells (Fig. 4B, orange line). Interestingly, when the concentration of GZ was increased to > 150 mM, the level of Mrc1 protein began to decrease (Fig. 4, orange line). This result is consistent with the data in Fig. 3B and C, showing that in addition to the replication stress, GZ may generate another stress that arrests the cell cycle at G2/M (see Fig. 5 later), particularly when higher concentrations are used. The HU-induced G2/M cell cycle arrest has been shown previously in erg11-1 and hem13-1 mutants, which indirectly affects the phosphorylation and protein level of the S-phase specific protein Mrc1 (Xu et al. 2016, Singh and Xu 2017).

Figure 4.

Rad3-dependent Mrc1 phosphorylation in the presence of GZ and HU. (A) Mrc1 phosphorylation in the presence of GZ. Wild-type S. pombe was treated with 15 mM HU or increasing concentrations of GZ in YE6S for 3 h. An equal number of cells were collected and lysed by standard trichloroacetic acid (TCA) and mini-bead beater method. The whole-cell lysate was separated by an 8% SDS–PAGE gel for western blotting analysis using a phosphor-specific antibody (top panel). The same blot was stripped and reprobed with antibodies against Mrc1 (mid panel). A section of the Ponceau S-stained membrane is shown for loading. (B) The band intensities in panel (A) were quantified and shown in percentages in comparison with that in HU-treated cells. Orange line: the protein levels of Mrc1. Blue line: the levels of Mrc1 phosphorylation in the presence of GZ. (C) Wild-type and the S. pombe cells with the indicated mutations were treated with 15 mM HU or 200 mM GZ for 3 h and examined as in panel (A).

Figure 5.

Flow cytometry analysis of the cell-cycle progression of wild-type and rad3∆ cells in the presence of HU or GZ. HU and GZ were added to the cultures at the indicated concentrations. Every hour during the treatment, an equal number of cells were removed and fixed for flow cytometry analysis. The green and red lines indicate 1C and 2C DNA content, respectively.

We next examined whether the GZ-induced Mrc1 phosphorylation is dependent on Rad3 and whether the phosphorylation occurs in erg11-1 cells. The results showed that in the presence of 200 mM GZ, Mrc1 phosphorylation was increased in wild-type cells but slightly lower than in the HU-treated cells (Fig. 4C). Like HU, the GZ-induced Mrc1 phosphorylation depended on Rad3, which confirms that GZ produces the acute replication stress inside the cells. Similar to HU, the GZ-induced Mrc1 phosphorylation was lower in the erg11-1 cells, which is consistent with the G2/M arrest. Together, these results strongly suggest that in addition to the replication stress, GZ causes another type of stress that arrests the cell cycle progression at G2/M.

GZ induces G2/M cell cycle arrest in S. pombe

To investigate the side effect of GZ in S. pombe, we analyzed the cell cycle progression of wild-type and rad3∆ cells in the presence of 200 mM GZ. For comparison, the cells were also treated with 15 mM HU. Every hour during the drug treatment, an equal number of cells were removed from the cultures and fixed for flow cytometry analysis. Because the S phase is very brief (20–30 min vs 3 h cell cycle time) and DNA replication occurs even before the completion of mitosis in S. pombe, majority of the cells in a logarithmically growing culture show the 2C DNA content (indicated by red lines at 0 h time point in Fig. 5). The small peak of 4C DNA contents are the daughter cells that are still attached to each other. During the first 3 h of incubation with HU, both wild-type and rad3∆ cells were increasingly arrested in the S phase (Fig. 5, green lines). In the presence of GZ, however, the cells remained at the G2/M phase of the cell cycle during the 6 h long incubation. This result clearly shows that although the acute replication stress is induced by GZ (Fig. 4), most of the cells are arrested at G2/M, which is likely caused by another stress. The G2/M arrest, which may function similarly as the DNA damage checkpoint, is consistent with the lower cytotoxicity in rad3∆ and the decreased Mrc1 protein level observed in Fig. 3C and Fig. 4B, respectively.

Microscopic examination of the GZ-treated S. pombe

Acute treatment with HU generates replication stress and arrests the cells in the S phase, which is mainly dealt with by the replication checkpoint. One of the key functions of the activated replication checkpoint is to suppress the mitosis so that the cells have enough time to finish the DNA synthesis before cell division. The rad3∆ cells, however, proceed into mitosis in HU, causing catastrophic premature mitosis, generating the so-called cut (cell untimely tor) cell phenotype (Hirano et al. 1986) in which the two daughters have unequal amounts of genomic DNA or even without detectable DNA. In contrast, acute HU treatment arrests the erg11-1 cells at cytokinesis (Xu et al. 2016). To further investigate the side effect of GZ, we examined the cells under the microscope. In this experiment, wild-type, rad3∆,mrc1∆ and erg11-1 cells were treated with 200 mM GZ or 15 mM HU for 3 h. The drug-treated cells were stained with Hoechst for genomic DNAs and Blankophor for septum and then examined under the microscope. We found that most of the HU-treated wild-type cells were mononuclear and significantly elongated, indicating the activated replication checkpoint (Fig. 6, top panel). The HU-treated rad3∆ cells were short and, a significant number of the cells showed the cut phenotype (Fig. 6, red arrows). The mrc1∆ cells also elongated in HU and most of the cells did not show the cut phenotype as the DNA damage checkpoint remains functional. Consistent with the cytokinesis arrest, HU treatment significantly increased the number of erg11-1 cells arrested in cytokinesis (Fig. 6, green arrows). The GZ treatment did not elongate the wild-type and the three mutant cells. Importantly, the rad3∆ cells did not show the cut phenotype although Mrc1 phosphorylation was significantly increased under this condition in wild-type cells. This result indicates that the GZ-induced G2/M arrest prevents the catastrophic mitosis in rad3∆ cells. Together, these results provide a strong support to the notion that besides the replication stress, GZ also generates another stress, arresting cells at the G2/M phase. Although the exact step of this arrest is still unclear, it is likely at late G2 or mitosis because it suppresses the cut phenotype in rad3∆ and the cytokinesis arrest in erg11-1 cells.

Figure 6.

Microscopic examination of the HU- and GZ-treated S. pombe. Wild-type and the mutant cells were treated with 15 mM HU or 200 mM GZ for 3 h. The drug-treated cells were fixed by brief heating, stained with Hoechst and Blankophor, and examined under the microscope. Red and green arrows indicate the HU-induced cut cells in rad3∆ and cytokinesis arrested cells in erg11-1, respectively.

Discussion

In this work, we tested three commercially available RNR inhibitors GEM, GZ and 3-AP in S. pombe, aiming to find an inhibitor that can produce the replication stress at chronic drug exposure conditions more specifically than HU, the drug that is currently used in many laboratories for generating replication stress and the S phase cell cycle synchronization. Among the three inhibitors, GEM has been used in clinics for the treatment of various carcinomas (see review in Mini et al. 2006). We found that GEM does not have any cytotoxic effect in S. pombe at concentrations as high as ≥70 mM. GEM is a pyrimidine nucleoside analog that requires nucleoside transporters for its influx into mammalian cells. Once inside the cells, GEM undergoes a complex intracellular conversion to the nucleotides diphosphate and triphosphate forms that are responsible for its cytotoxic effects. While the triphosphate form competes with dCTP or is incorporated into DNA to block the DNA synthesis, the diphosphate form is a potent inhibitor of RNR (Mini et al. 2006, Parvathaneni and Sharma 2019). The insensitivity of S. pombe to GEM suggests that fission yeast likely lacks the transporter for the influx of GEM or the enzymes that convert GEM to the nucleotides. Overexpression of an exogenous nucleotide transporter and an exogenous thymidine kinase in S. pombe has been shown to promote the incorporation of thymidine analogs in genomic DNA (Rhind 2009). Whether this expression system can be used to enhance the cell-killing effect of GEM in S. pombe needs further investigation.

Compared with HU, 3-AP is also a bulky drug like GEM. However, it shows a highly cytotoxic effect in S. pombe under chronic exposure conditions. The cell-killing effect of 3-AP is likely caused by replication stress because the replication checkpoint mutant rad3∆ and cds1∆ cells are much more sensitive than the wild-type and the metabolic mutant cells. Unfortunately, 3-AP does not show, if any, acute cytotoxic effect in S. pombe. We suspect that the bulky and hydrophilic nature of 3-AP makes it more difficult to pass through the cell wall and penetrate the S. pombe cells. Consistent with this possibility, a minimal influx of 3-AP, although slowly, can inhibit the cell growth under chronic exposure conditions (Fig. 2). We also found that among the three RNR inhibitors, GZ is the only drug that shows the cytotoxic effect at both chronic and acute treatment conditions. Under chronic conditions, GZ produces the replication stress more specifically than HU because it has a much less cytotoxic effect on the metabolic mutants. Under acute conditions, however, it arrests the cell cycle mainly at the G2/M, not the S phase, showing that in addition to the replication stress, GZ has a significant side effect in S. pombe. One possible explanation for the side effect is that a much higher concentration of GZ is required to generate the replication stress and activate the replication checkpoint in liquid cultures, which likely causes the spread-over effects and thus the G2/M arrest. In fact, increasing HU to ≥50 mM in liquid cultures also arrests S. pombe at the G2/M phase (data not shown). The G2/M delay induced by GZ likely explains the sensitivity of the two metabolic mutants in the spot assay (Fig. 2), the reduced cytotoxicity and the lack of cut cells in rad3∆ mutant (Figs 3C and 6). Nonetheless, because of its low cost, using GZ in the spot assay to eliminate the complications associated with HU under the chronic exposure conditions should benefit the checkpoint research in S. pombe.

It is generally believed that HU and GZ inhibit RNR by the same mechanism, i.e. directly quenching the diferric tyrosyl radical center in the smaller subunits of the enzyme by one-electron transfer from the drugs (Wright and Lewis 1974, Moore and Hurlbert 1985). However, the radical center is buried deep inside the smaller subunit of RNR. Therefore, the exact mechanism of how HU, particularly the bulkier GZ, accesses and subsequently quenches the radical center remains to be fully understood. The tyrosyl radical in the smaller subunits of RNR has to be transferred along a long path to a cysteine residue in the larger subunits to generate a thiyl radical for catalysis. Therefore, HU or GZ may also affect the radical transfer and thus suppress the RNR. Nevertheless, our results show that GZ and 3-AP can replace HU in chronic exposure such as the spot assay to produce more specific replication stress in S. pombe. Considering the lower cost, GZ is a better choice for this assay. However, none of the three tested RNR inhibitors can replace HU under the acute drug treatment conditions to generate the replication stress for checkpoint studies in S. pombe. Therefore, using GZ in spot assay and HU in the acute treatment should complement with each other and thus simplify the research that requires replication stress to be produced under various experimental conditions. Since the RNR enzyme is conserved from bacteria to yeasts and humans, these results may also benefit the research in other model organisms.

Acknowledgements

We thank Tom Kelly and Paul Russell for sharing the yeast strains and other members of the Xu lab for their help and support.

Funding

This work was supported by the NIH grants R01GM110132 and R35GM144307 to YJX.

Conflict of interest statement

None declared