FACULTY OF SCIENCE UNIVERSITY OF COPENHAGEN DENMARK

PhD Thesis

Katrine Vyff Løvschal

Regulation of and Screening for Species-Specific Inhibitors in Fission Yeast

Academic advisor: Associate Professor Christian Holmberg Submitted: 21/03/2019

Regulation of Ribonucleotide Reductase and Screening for Species-Specific Inhibitors in Fission Yeast

Katrine Vyff Løvschal PhD Thesis March 2019

This thesis has been submitted to the PhD School of Science at the University of Copenhagen, Denmark

Preface

This thesis has been submitted to the PhD School of Science at the University of Copenhagen. The research presented herein was carried out between March 2016 and March 2019 in the Cell Cycle and Genome Stability Lab, Department of Biology, University of Copenhagen under supervision of Associate Professor Christian Holmberg.

The thesis covers two main areas presented in two chapters. Chapter 1 provides a general introduction to the cellular functions of eukaryotic class Ia ribonucleotide reductases. It explores the genetic interactions involving ribonucleotide reductase and genome stability factors in fission yeast, most of which are conserved in humans. Chapter 2 presents a strategy to screen for novel and species-specific inhibitors of ribonucleotide reductase using fission yeast.

This thesis presents original, unpublished work, except where references are made to previous work. In addition to my work presented in the two chapters, I have contributed to the following publication:

Paper I Deoxynucleoside Salvage in Fission Yeast Allows Rescue of Ribonucleotide Reductase Deficiency but Not Spd1-Mediated Inhibition of Replication

Oliver Fleck, Ulrik Fahnøe, Katrine Vyff Løvschal, Marie-Fabrice Uwamahoro Gasasira, Irina N. Marinova, Birthe B. Kragelund, Antony M. Carr, Edgar Hartsuiker, Christian Holmberg and Olaf Nielsen

Genes (2017) 8(5), DOI: 10.3390/genes8050128

v

Acknowledgements

First of all, I would like to thank Associate Professor Christian Holmberg and Professor Olaf Nielsen for the provision of the laboratory facilities in the Cell Cycle and Genome Stability Lab and for creating an inspiring research environment. I also wish to thank the Villum Foundation for funding my PhD project. I am sincerely grateful to my supervisor Christian Holmberg for sharing his expertise and providing me with continued guidance throughout my studies, as well as for commenting on this thesis. I also wish to express my gratitude for the opportunity to attend the 9th International Fission Yeast Meeting in Banff, Alberta, Canada and the Yeast Genetics and Genomics course in Cold Spring Harbor, New York, USA. Olaf Nielsen is thanked for assisting as co-supervisor and always being helpful and open for discussions. I would also like to thank past and present members of the Cell Cycle and Genome Stability Lab, as well as former and current office mates for creating a wonderful work environment and for inspiring discussions. I would especially like to thank Michaela Rasmussen for her technical assistance in the lab. Finally, I take this opportunity to express my gratitude to my wonderful family and friends for always being there for me with kind words and inspirational patience. Thank you, Andreas, for your unconditional love and support. I could not have done this without you.

vii

Contents

Preface …………………………………………………………………………………………….…...v

Acknowledgements………………………………………………………………………………..vii

Abstract………………………………………………………………………………………...……..ix

Resumé………………………………………………………………………………………...……..xi

Abbreviations…………………………………………………………………………………….....xv

Chapter 1 Aspects of Ribonucleotide Reductase and Species-specific Regulation……………...1

Introduction……………………………………………………………………………….…...3

Schizosaccharomyces pombe……………………………………………………….3 Life cycle……………………………………………………………..………....4 Mating types………………………………………...…………………….……5 S. pombe as model organism…………………………………………….….7

DNA integrity checkpoints and genome stability………………………….…..8

Ribonucleotide reductase……………………………………………………….….11

Regulation of ribonucleotide reductase…………………………………………16 Transcriptional and post-translational regulation...……………….…..16 Regulation by allostery and oligomerization...………………………...17 Regulation by small protein inhibitors…………………………………..21 Ubiquitin-mediated regulation of RNR inhibitors…………………….23 Cdt2 CRL4 inactivation……………………………………………………….25

Results………………………………………………….………………..……………………27

Generation of fission yeast strains relying on human RNR……….…….…27

Strain characterization.……………..……….………………..……………………29

The effect of spd1 overexpression on cell cycle progressio………………..33

Human RNR in fission yeast confers checkpoint activation and dependency…………………………………….………………..……………………35

Elevated levels of the R2 homologs suppress checkpoint activation and dependency…………………………………….………………..……………………38

In vivo function(s) of Spd1…………………………………….………...………..42

Spd1 is destabilized in cells with human RNR……………………………….47

Discussion………………………………………………………………………………...... 50

Chapter 2 Co-culturing System to Screen for Species-Specific Inhibitors of RNR………….59

Introduction……………………………………………………………………………….....61

Results………………………………………………………………………………...……...63 Experimental setup…………………………………………………………….…...63 Strain creation…………………………………………………………………….....65 Proof of principle…………………………………………………………….……..66

Discussion and future perspectives…………………………………………………….69

Materials and Methods………………………………………………………………………...…73

References………………………………………………………………………………………...... 89

Paper 1………………………………………………………………..……………………………...97

Appendix – Supplementary Figures………………………………………………………….113

Abbreviations

aa − amino acid dGTP − deoxyguanosine triphosphate Amp − ampicillin Dif1 − damage-regulated import facilitator 1 APC − anaphase promoting complex DmdNK − Drosophila melanogaster ATM − ataxia-telangiectasia mutated deoxyribonucleoside kinase ATP − adenosine triphosphate DMSO − dimethyl sulfoxide ATR − ataxia telangiectasia and Rad3- DNA − deoxyribonucleic acid related dNDP − deoxynucleoside diphosphate BF − bright field dNTP − deoxyribonucleoside triphosphate bp − base pairs dTTP − deoxythymidine phosphate BSA − bovine serum albumin DSB − double-strand break C. albicans − Candida albicans Dun1 − DNA-damage uninducible 1 Cdc − cell division cycle E. coli − Escherichia coli cDNA − complementary DNA EtOH − ethanol Cdh1 − cadherin 1 FACS − fluorescence-activated cell sorting Cds1 − checking DNA synthesis 1 G418 − genetecin

Cdt2 − Cdc-10-dependent transcript 2 G1 − gap 1

CFP − cyan fluorescent protein G2 − gap 2 ChimR2 − chimeric ribonucleotide reductase h − hour small subunit h+/- − heterothallic Chk − checkpoint kinase h90 − homothallic COP9 − constitutive photomorphogenesis 9 HA − hemagglutinin CPT − camptothecin hENT1 − human equilibrative nucleoside CRL − cullin RING E3 transporter 1 CSN − COP9 signalosome hRNR − human ribonucleotide reductase C-terminal − carboxy terminal HRP − horse radish peroxidase Cul4 − cullin 4 HU − hydroxyurea cut − cell untimely torn IDP − intrinsically disordered protein dATP − deoxyadenosine triphosphate IgG − immunoglobulin G dCTP − deoxycytidine triphosphate IP − immunoprecipitation Ddb1 − DNA-damage binding 1

xi IRBIT − IP3R binding protein released with Pol − polymerase inositol 1,4,5-triphosphate R1 − ribonucleotide reductase large subunit Kana − kanamycin R2 − ribonucleotide reductase small subunit KanMX − kanamycin A selection marker Rad − radiation sensitive kDa − kilodalton RLU − relative light units LiOAc − lithium acetate RNA − ribonucleic acid MBF − Mlu1 binding factor RNR − ribonucleotide reductase MBP − myelin basic protein ROS − reactive oxygen species mCherry − monomeric derivative of DsRed RPA − replication protein A fluorescent protein RRM1 − ribonucleotide reductase catalytic Mec1 − mitosis entry checkpoint 1 subunit M1 [Homo sapiens] Mik1 − mitosis inhibitory kinase 1 RRM2 − ribonucleotide reductase regulatory MMS − methyl methanesulfonate subunit M2 [Homo sapiens] MSA − minimal sporulating agar RRM2B − ribonucleotide reductase MSL − minimal sporulation liquid regulatory TP53 inducible subunit M2B Mbu1 − multi budding 1 [Homo sapiens] NDP − ribonucleoside diphosphate RT − room temperature NDPK − nucleoside diphosphate kinase S. cerevisiae − Saccharomyces cerevisiae NER − nucleotide excision repair S. pombe − Schizosaccharomyces pombe NLS − nuclear localization sequence SCF − Skp1-Cul1-F-box nmt − no message in thiamine S.d. − standard deviation N-terminal − amino terminal SDS-PAGE − Sodium dodecyl sulphate O/N − overnight polyacrylamide gel electrophoresis ORF − open reading frame siRNA − small interfering RNA P. falciparum − Plasmodium falciparum Sml1 − suppressor of Mec1 lethality 1 PCNA − proliferating cell nuclear antigen Spd1 − S phase delayed 1 PCR − polymerase chain reaction ssDNA − single-stranded DNA − tumor protein 53 (TP53) Ste9 − sterile 9 p53R2 − p53-inducible ribonucleotide Suc22 − suppressor of cdc22ts reductase small subunit homolog TCA − trichloracetic acid PIKK − phosphatidylinositol 3-kinase- TMPK − thymidylate phosphate kinase related kinase ts − temperature sensitive

xii TS − thymidylate synthase Wt − wild type Ub − ubiquitin YES − yeast extract solid Ubr2 − E3 ubiquitin-protein ligase YEL − yeast extract liquid UV − ultraviolet

xiii

Abstract

Cellular proliferation and genome integrity depend on accurate DNA replication and repair of damaged DNA, which in turn are impacted by the level of deoxyribonucleotides. Ribonucleotide reductase is the responsible for the de novo synthesis of deoxyribonucleotides in all organisms and is thus indispensible for life. The key requirement for ribonucleotide reductase has made it an attractive target for both anti-cancer and anti-microbial therapy, and during the last few decades, a considerable amount of effort has been devoted to developing novel and more specific inhibitors of this enzyme. Thorough knowledge of ribonucleotide reductase, and the mechanisms that interlock to restrain it, may lead to new rationales for developing novel inhibitors. Here, we have used fission yeast as model organism to explore how human ribonucleotide reductase can complement for lack of the yeast enzyme, and to explore the genetic interactions involving ribonucleotide reductase and genome stability factors, which are highly conserved from yeast to humans. We have demonstrated good cross-species complementation of the human enzyme regarding the production of the DNA precursors and cellular growth. Other functions were shown to be less complemented leading to checkpoint dependency and meiotic defects, independently of deoxyribonucleotide pools. Our results indicate that fission yeast ribonucleotide reductase has another cellular function, in addition to generating deoxyribonucleotides, and that the small protein inhibitor Spd1 serve as a negative regulator of both functions. Furthermore, our results provide evidence that Spd1 is able to restrain the human R1-R2 holoenzyme to regulate deoxyribonucleotide pools. We also present a strategy to identify novel and species-specific inhibitors of ribonucleotide reductase using fission yeast as selective system. Based on preliminary experiments, we evaluate this strategy to have promising potential to identify novel inhibitors that may be used to treat various human diseases.

xv

Resumé

Korrekt DNA-replikation og reparation af beskadiget DNA er vitalt for cellulær vækst og opretholdelse af genomisk integritet. Dette afhænger af et korrekt og afbalanceret niveau af deoxyribonukleotider. Enzymet ribonukleotidreduktase er ansvarlig for de novo syntetisering af deoxyribonukleotider i alle organismer, og er derfor uundværlig for liv. På grund af dets essentielle rolle, er ribonukleotidreduktase et attraktivt mål for både anti-cancer og anti-mikrobielle medikamenter. Man har i mange år arbejdet, med stor indsats, på at identificere nye og mere specifikke inhibitorer af dette enzym. Dybtgående viden omkring ribonukleotidreduktase, og de mekanismer som arbejder sammen om at regulerer det, kan lede til alternative rationaler for udvikling af nye inhibitorer. Vi har brugt spaltegær som model- organisme til at undersøge hvordan humant ribonukleotidreduktase kan komplementere for tabet af det tilsvarende enzym i gær. Endvidere har vi undersøgt genetiske interaktioner som involverer ribonukleotidreduktase og faktorer vigtige for opretholdelse af genomisk integritet i spaltegær, hvoraf de fleste er konserverede i mennesker. Vi har demonstreret god komplementering hvad angår syntetisering af deoxyribonukleotider. Andre funktioner var i mindre grad komplementerede, hvilket resulterede i afhængighed af et funktionelt checkpoint system samt problemer under meiosen – begge uafhængigt af niveaet af deoxyribonukleotider. Vores resultater indikerer, at ribonukleotidreduktase i spaltegær muligvis har en anden funktion, foruden den involveret i syntetisering af deoxyribonukleotider. Begge disse funktioner synes at være negativt reguleret af det lille ustrukturerede protein, Spd1. Vores resultater indikerer også at Spd1 kan hæmme aktiviteten af det humane R1-R2 holoenzym. I dette studie, præsenterer vi også en strategi for identificering af nye og arts- specifikke inhibitorer af ribonukleotidreduktase, ved brug af spaltegær som model organisme. Baseret på foreløbige eksperimenter, evaluerer vi denne strategi til at have lovende potentiale i at identificere nye inhibitorer, som muligvis kan bruges i behandling af forskellige sygdomme i mennesker.

xvii

Chapter 1

Aspects of Ribonucleotide Reductase and Species-specific Regulation

Chapter objectives To investigate Schizosaccharomyces pombe cells reliant on human ribonucleotide reductase for de novo production of deoxyribonucleotides, as well as to study the conservation of the mechanisms that interlock to restrain the enzyme and maintain genome stability

Introduction In all living organisms, active DNA synthesis is required prior to cell division, and in response to DNA damaging insults threatening the integrity of the genome. A prerequisite for precise DNA synthesis is the generation of a correct and balanced pool of the deoxyribonucleoside triphosphates (dNTPs), which are the building blocks of DNA. The enzyme ribonucleotide reductase (RNR) is responsible for de novo biosynthesis of the dNTPs and is thus indispensable for all living organisms. The essential role of RNR in cell proliferation have made it an attractive target for anti- cancer, anti-bacterial, and anti-viral therapy (Eklund et al., 2001). In this study, the yeast Schizosaccharomyces pombe (S. pombe) has been used as a model organism to investigate the regulation and enzymatic properties of class Ia RNRs. Emphasis has been on fission yeast cells relying on the human counterpart of the enzyme. The following sections will provide an introduction to some of the main concepts regarding RNR. First, S. pombe will be introduced as a model organism, followed by a review on DNA integrity checkpoints and genome stability. A thorough description of class Ia RNRs, including their structure, regulation and mechanisms of action, will be provided with special emphasis on the differences between S. pombe RNR and mammalian RNR. In chapter 2, the role of RNR in human diseases will be touched upon, including strategies for developing novel RNR inhibitors.

Schizosaccharomyces pombe S. pombe, also known as fission yeast, is a unicellular, free-living fungus, belonging to the phylum Ascomycota (Hoffman et al., 2015). Fission yeast has the smallest sequenced eukaryotic genome of just 14 Mb, distributed on three , (compared with 16 chromosomes found in S. cerevisiae) (Forsburg, 1999; Wood et al., 2002). Currently, roughly 5000 protein-coding are annotated, of which 67% are conserved in humans. The best online resource for S. pombe is PomBase (https://www.pombase.org). This database for fission yeast provides structural and functional annotations. It contains detailed information on genomic sequences, genomic features and genome-wide datasets (Wood et al., 2012).

3 Life cycle Fission yeast is rod-shaped with a length of ∼7-14 µm and a width of ∼4 µm (Figure 1.1) (Hayles and Nurse, 2001). It grows by tip elongation until it attains a critical length, whereupon it divides by medial fission within a very short cell cycle of 2-4 hours (Egel, 2005). The mitotic cell cycle of fission yeast resembles that of a typical eukaryote, which is separated into the G1 (Gap 1), S (DNA synthesis), G2 (Gap 2), and M (Mitosis) phases (Figure 1.2, left). Following mitosis, a transverse septum is laid down medially, and the cell is cleaved to produce two essentially identical daughter cells. Concurrently, the newly segregated nuclei enter the next cell cycle and undergo G1 and S phase prior to completion of the cytokinesis, meaning that a single cell almost always has a 2C DNA content; as a binucleate G1/S phase cell or as a G2 cell (Forsburg and Rhind, 2006; Hoffman et al., 2015). After cytokinesis, cells remain in G2 phase, which is particularly long (70% of the cell cycle time) in exponentially growing cells. In contrast, G1, S and M phase each take up only 10% of the division time (Forsburg and Rhind, 2006). After sufficient growth in G2 phase, the next mitotic phase is initiated where sister chromosomes segregate and the nucleus divides. Instead of proceeding with the next mitotic cycle, cells can also enter a stationary phase and become quiescent cells (Costello et al., 1986).

………………………………………………………….… Figure 1.1. S. pombe Scanning electron micrograph of S. pombe cells (from Hayles and Nurse, 2001) ………………………………………………………….…

4 In addition to asexual reproduction, fission yeast cells are also able to reproduce sexually by mating with each other to generate diploid cells (Figure 1.2, right). This occurs only when subjected to nitrogen starvation in the presence of a mating partner with opposite mating type (Egel, 2004; Hoffman et al., 2015). Mating involves adhesion of the cells, followed by digestion of the walls separating them, to form a single fusion cell containing the two parental nuclei. The nuclei subsequently fuse to form a zygote, a single cell containing a single diploid nucleus (Hoffman et al., 2015). Newly formed zygotes immediately proceed to pre-meiotic S phase, followed by two successive meiotic divisions, which result in the formation of four haploid spores in a zigzag-shaped zygotic ascus. The four dormant ascospores can each germinate and re- enter the cell cycle when optimal conditions are restored. Diploid outgrowth from a zygote can occur – albeit rarely. Thus, the default lifestyle of S. pombe is as a haploid (Forsburg and Rhind, 2006; Hoffman et al., 2015).

Mating types The mating type of fission yeast is determined by expression of mating type genes at the mat1 locus on II. These genes encode either P (plus) or M (minus) genetic information, resulting in the two mating types h+ and h-, respectively (Beach et al., 1982; Willer et al., 2015). h+ and h- strains are heterothallic and require a compatible partner for sexual reproduction to take place, which is controlled by the action of mating pheromones; h+ cells produce P-factor and h- cells produce M-factor. In contrast, homothallic strains referred to as h90 (because 90% of the cells mate and produce spores when they are starved for nitrogen), are capable of mating type switching, and are thus self-fertile. Homothallic strains therefore form colonies containing mixed populations of h+ and h- cells. In this way, a mating partner is always present if unfavorable conditions require the formation of ascospores, which are more resistant to the environment (Egel, 2004). Most laboratory h+ strains can revert to h90 – albeit at very low frequency (10-3/generation), whereas h- strains are stable (Klar, 1992; Klar et al., 1991). Mating pheromones and the capability of homothallic fission yeast to mate within a colony, are frequently exploited in yeast genetics.

5

……………………………………………………………………………………………………………. Figure 1.2. The life cycle of S. pombe.

In S. pombe, G1, S, and M phases are short (10% each of the cell cycle), while G2 is relatively long. DNA replication starts before the two daughter cells separate physically, resulting in cells almost always having a 2C DNA content. The default lifestyle of S. pombe is as a haploid, and haploid cells with opposite mating types (depicted as blue and yellow nuclei) can reproduce sexually. First, the cells conjugate to form a transient diploid zygote, which then proceed through meiosis and sporulation to generate four ascospores. Each of the ascospores can return to the mitotic cycle and proliferate as a haploid cell. Diploid outgrowth from the zygote occurs rarely, and diploid S. pombe cells are generally unstable as they are highly prone to sporulate (Sabatinos and Forsburg, 2010). …………………………………………………………………………………………………………….

6

S.pombe as model organism S. pombe has become one of the best-known model organisms in the entire fungal kingdom – only next to S. cerevisiae (budding yeast) (Egel, 2004). As an eukaryote, S. pombe is a valuable model organism to investigate processes that are universal and well conserved from yeast to human, including genome duplication, DNA recombination and repair, as well as cell-cycle and checkpoint responses (Hoffman et al., 2015). Findings from such studies in fission yeast may be directly extrapolated to higher eukaryotes. Like S. cerevisiae, S. pombe offers the strengths of excellent yeast genetics including the ease of genome manipulation. Further benefits of using fission yeast as a model organism include its simplicity and low cost. The short doubling time is a huge advantage as it allows large experimental populations, which is beneficial for the significance of the obtained data. In this chapter, fission yeast was used as a model organism to provide insight into class Ia RNRs, their role in dNTP biosynthesis, as well as their importance in maintaining genome stability.

7 DNA integrity checkpoints and genome stability DNA is constantly subjected to insults threatening the stability of the genome. Both extrinsic and intrinsic factors, including ultraviolet light (UV), ionizing radiation (IR), and oxidative products arising from cellular metabolism, contribute to the threat (Shechter et al., 2004). To constrain the detrimental effects of these factors and to protect the genome, cells have active checkpoint mechanisms that respond to DNA damage and replication stress by a well-coordinated cellular response (Nyberg et al., 2002). Activation of these signaling pathways results in a delay in cell cycle progression to allow time for coping with the DNA insults. In this way, any event of one cell cycle phase is properly completed before entering the next, which is important for maintaining genome stability (Hartwell and Weinert, 1989; Yin et al., 2008). Consistent with their importance in all eukaryotes to ensure error-free cell cycle progression, the checkpoint pathways are highly conserved throughout evolution (Hoffman et al., 2015) (Figure 1.3). In eukaryotic cells, checkpoint responses to damaged or unreplicated DNA are initiated by the key regulators Ataxia-telangiectasia-mutated (ATM) and ATM- and Rad3-related (ATR). ATR (Rad3 in S. pombe, Mec1 in S. cerevisiae) and ATM (Tel1 in S. pombe and S. cerevisiae) belong to the family of phosphatidylinositol-3-kinase- like kinases (PIKKs) and respond to different types of DNA insults (Abraham, 2001). ATM mainly responds to DNA double-strand breaks (DSBs), whereas ATR respond to a broad spectrum of DNA damage including any lesion that perturbs replicating chromosomes (Maréchal and Zou, 2013). Common to these lesions is the generation of single-stranded DNA (ssDNA) covered with replication protein A (RPA). This DNA structure (RPA:ssDNA) arises during DNA repair and perturbed replication forks, and is proposed to trigger the checkpoint responses (Awasthi et al., 2016; Zou et al., 2003). Once activated, ATR and ATM mediate the signal to the immediate downstream effector kinases Chk1 (Chk1 in S. pombe and S. cerevisiae) and Chk2 (Cds1 in S. pombe and Rad53 in S. cerevisiae) (Liu et al., 2000; Matsuoka et al., 1998; Reinhardt et al., 2009). Unlike ATM and ATR, which work at the chromatin close to the site of damage, the effector kinases can transduce the signal to distant substrates eventually leading to a cell cycle delay (Abraham, 2001).

8 In fission yeast, Rad3 is the major signal transducing kinase (Bentley et al., 1996). Rad3 is involved in responses to both DNA damage and replication stress and is required to propagate checkpoint signals to both of the effector protein kinases Chk1 and Cds1 (Lindsay et al., 1998; Martinho et al., 1998). Unlike ATR, Rad3 is a non-essential protein and its loss does not significantly affect the cell cycle of undamaged cells. However, Rad3 becomes required for cells to survive treatments that perturb DNA synthesis. That is, mutants lacking Rad3 function are checkpoint defective and die in the presence of DNA damaging agents, as they fail to prevent inappropriate mitosis (Humphrey, 2000; Lindsay et al., 1998; Walworth and Bernards, 1996). Though a sequence homolog of mammalian Chk2, fission yeast Cds1 is a functional equivalent to mammalian Chk1 (Iyer and Rhind, 2017). Cds1 is phosphorylated and activated in response to perturbations to DNA replication, as well as to DNA damage during S phase (Lindsay et al., 1998). Once activated, it functions to stabilize perturbed replication forks to prevent them from collapsing, suppress firing of late replicating origins, and delay mitosis so that DNA synthesis can resume when proper conditions are restored (Boddy et al., 1998; Lindsay et al., 1998; Rhind and Russell, 2015; Sabatinos and Forsburg, 2010; Sabatinos et al., 2012; Xu, 2016)

In the G2 phase of the cell cycle, DNA damage from exposure to DNA- damaging agents causes activation of Chk1 (Martinho et al., 1998). The main function of Chk1 is to delay the onset of mitosis. Both Chk1 and Cds1 block the G2-M transition by preventing activation of Cdc2, the kinase that drives mitosis (Rhind and Russell, 2015). Cdc2 is inactive when phosphorylated on Tyr15, which is maintained by simultaneous stimulation of the inactivating protein kinases Wee1 and Mik1 and inhibition of the activating phosphatase Cdc25 (Boddy et al., 1998; Borgne and Nurse, 2000; Furnari et al., 1997). In the absence of Cds1, Chk1 can act to execute a checkpoint delay (Lindsay et al., 1998). Consequently, ∆cds1 mutants do not have an apparent checkpoint defect and they elongate and cease division like wild type cells in response to replication stress. In contrast, ∆rad3 and ∆cds1∆chk1 mutants continue to divide and exhibit a “cut” phenotype, indicative of checkpoint failure and consequently untimely segregated chromosomes (Tanaka and Russell, 2001).

9 In addition to controlling proper cell cycle progression, the checkpoint pathways have been shown to be involved in a multifaceted response. In metazoans, the checkpoint pathways involve the tumor suppressor protein p53. Depending on the magnitude of the DNA damage leading to checkpoint activation, p53 can execute, prolonged cell cycle arrest, DNA repair, senescence or apoptosis (Vogelstein et al., 2000; Wahl and Carr, 2001). Even though no homolog of the p53 protein exists in unicellular organisms, yeast cells also have integrated checkpoint and repair pathways (Moss et al., 2010; Wahl and Carr, 2001). A highly conserved strategy of the checkpoints to facilitate DNA repair involves up-regulation of RNR, the enzyme responsible for de novo synthesis of the DNA building blocks (Jordan and Reichard, 1998).

……………………………………………………………………………………………………………. Figure 1.3. Checkpoint pathways in fission yeast and mammalian cells The checkpoint pathways are conserved in the two systems, resulting in cell cycle delay and time to repair any insults to the DNA. The p53 pathway in mammalian cells provides additional cell fates that are not needed in unicellular organisms. For simplicity, only the main components of the checkpoints are depicted. …………………………………………………………………………………………………………….

10 Ribonucleotide reductase Ribonucleotide reductase (RNR) is an essential and highly conserved enzyme, which is responsible for de novo biosynthesis of deoxyribonucleoside triphosphates (dNTPs), the building blocks of DNA (Reichard, 1988). RNR catalyzes the reduction of ribonucleoside diphosphates (NDPs) into their deoxy forms (dNDPs) by controlled free-radical chemistry (Stubbe and Riggs-Gelasco, 1998; Thelander and Reichard, 1979). RNRs can be grouped into three classes based on their metal cofactors, the mechanism by which they generate the free radical, and whether they interaction with oxygen (Nordlund and Reichard, 2006). These classes are further subcategorized. However, as this study is limited to RNRs found in eukaryotes, only the class Ia will be described in detail. The classic class Ia RNR is a multimer of two non-identical subunits denoted α and β (R1 and R2, respectively). The composition of class Ia enzymes are complex and remains poorly understood. However, it is commonly accepted that the active state of the enzyme is a α2β2 hetero-tetramer consisting of 2xR1 and 2xR2 subunits (Nordlund and Reichard, 2006) (Figure 1.4). Various inactive forms of class Ia RNR have been reported for different species, all of which are induced in a dATP- dependent manner (see below). R1, the larger of the RNR subunits, contains the catalytic site as well as two allosteric binding sites required for its regulation; the activity site and the specificity site. The smaller R2 subunit contains a differic iron center, in which a stable tyrosyl radical is generated in a reaction involving molecular oxygen. The generated free radical is continuously transferred over a 30-35 Å distance to a structurally conserved cysteine residue at the catalytic site. This ultimately leads to the generation of a thiyl radical, which provides the reducing power for catalysis (Ekberg et al., 1996; Minnihan et al., 2014; Nordlund and Reichard, 2006). Upon substrate binding, the thiyl radical abstracts a hydrogen atom from the 3ʹ-carbon generating an intermediate substrate radical (Figure 1.4, enlarged catalytic site). Two proximal reducing cysteines of the large R1 subunit reduce the substrate by protonation of the 2ʹ- hydroxyl group, which result in the formation of a disulfide bridge and removal of a water molecule. The reaction is completed by re-oxidation of the initial cysteine to a thiyl radical, and re-reduction of the reducing cysteines by disulfide exchange reactions involving thioredoxin and thioredoxin reductase (Eklund et al., 2001; Nordlund and Reichard, 2006).

11 …………………………………………………………………………………………………………..... Figure 1.4. Catalytic mechanism of class Ia RNRs

Illustration of a class Ia RNR, depicted as a α2β2 hetero-tetramer. The large R1 subunit contains the catalytic site (C) as well as two allosteric sites: the activity site (A) and the specificity site (S). The smaller R2 subunit contains a diferric-oxygen center (Fe-O-Fe), which generates a tyrosyl free radical. The generated radical is transferred to a conserved cysteine residue at the catalytic site to generate a thiyl radical required for catalytic activity. The enlarged catalytic site shows the reaction mechanism of substrate reduction. The generated thiyl radical (S•) abstracts a hydrogen atom from the 3ʹ-carbon of the ribose generating a substrate radical (1). The hydroxyl group of the 2ʹ- carbon is lost in the form of water (2) as two cysteines at the catalytic site reduce the substrate (3). After substrate reduction, the initial thiyl radical is re-generated (4), and the reducing cysteines are re- reduced via a series of disulfide exhange reactions involving thioredoxin (TR), thioredoxin reductase (TRR), and NADPH (5). RNR act on all four NDPs (B=A, U, G or C). …………………………………………………………………………………………………………....

12 In mammalian cells, the RRM1 encodes the large R1 subunit. Two distinct genes, RRM2 and RRM2B, encode isoforms of the small R2 subunit denoted R2 and p53R2, respectively (Eklund et al., 2001; Tanaka et al., 2000). Both R2 and p53R2 can generate the important tyrosyl free radical and are able to form an active enzyme with the R1 protein (Guittet et al., 2001). The RRM2B gene is a downstream target for the tumor suppressor p53 and its expression is induced by DNA damage providing a link between RNR and cancer. Consistent with p53R2 being a DNA damage- responsive gene product, the R1-p53R2 holoenzyme is responsible for providing dNTPs for DNA repair in quiescent cells. Furthermore, it is crucial for mitochondrial homeostasis as it support both replication and repair of mitochondrial DNA. In contrast, the R1-R2 holoenzyme supports DNA replication and repair during S phase (Bourdon et al., 2007; Guittet et al., 2001; Nordlund and Reichard, 2006; Pontarin et al., 2007, 2012). In fission yeast, the homologs of R1 and R2 are denoted Cdc22 and Suc22, respectively (Fernandez Sarabia et al., 1993). Budding yeast has two homologs of R1 (Rnr1 and Rnr3) and two homologs of R2 (Rnr2 and Rnr4), possible resulting from when S. cerevisiae underwent genome duplication during its evolution (Wolfe and Shields, 1997). Rnr2 and Rnr4 are both required forming a functional reductase together with a Rnr1 dimer (α2ββ’). Thus, budding yeast relies on a class Ia reductase with unique features (Elledge et al., 1992; Huang and Elledge, 1997; Nordlund and Reichard, 2006). In general, there is a high sequence identity among the eukaryotic RNR proteins. The amino acid sequence of fission yeast Suc22 is depicted in Figure 1.5. Comparison of this sequence with those of human R2 and p53R2 revealed 70% and 69% sequence identity, respectively. Furthermore, Suc22 shares 72% and 53% sequence identity with the S. cerevisiae proteins Rnr2 and Rnr4, respectively (not shown). Especially the functionally important regions within the R2 homologs, including the radical site, iron ligands, residues involved in radical transfer, and the C-terminal heptapeptide essential for binding to R1, show great conservation (Guittet et al., 2001). Similarly, the R1 homologs are highly conserved. Fission yeast Cdc22 shares 69% sequence identity with human R1 (not shown).

13

……………………………………………………………………………………………………………. Figure 1.5. Alignments of the amino acid sequences of fission yeast Suc22, human R2 and p53R2 Sequences of R2 subunits from different organisms are very similar with high conservation of important regions. The conserved iron ligands are marked with red boxes. The radical site is marked with a green box. Residues involved in the long-range radical transfer pathway are marked with blue boxes. The heptapeptide required for binding to the R1 subunit is marked with a yellow box. The KEN- box required for degradation of human R2 in late mitosis is marked with a black box. …………………………………………………………………………………………………………….

14 Despite high sequence identity between R2 homologs, the amino ends of the proteins differ considerable in length. This is most evident from the two human R2 isomers, which shares 81% amino acid sequence identity. The human p53R2 protein has a 33 amino acid truncation compared with the human R2 protein (Chabes et al., 2003b). This region within R2 contains a KEN-box at position 30-32 (marked with a black box in Figure 1.5). The KEN-box is required for cell cycle stage-specific degradation of the protein by binding of the Cdh1-anaphase-promoting complex (APCCdh1, APCSte9 in S. pombe) (see below) (Chabes et al., 2003b; Kitamura et al., 1998; Trickey et al., 2008). Since the amino ends differ between the R2 homologs, they are not likely needed for enzymatic activity. In support of this, it has been shown that a N-terminal truncation of the R2 protein in mice did not significantly alter the enzymatic activity of RNR (Chabes et al., 2003b; Kauppi et al., 1996). Furthermore, protein structure predictions suggest that both the N- and C-termini of the R2 homologs of RNR show a similar pattern of disorder, which is known to mediate binding diversity. (Brignole et al., 2018; Eklund et al., 2001; Kriwacki et al., 2002; Smith et al., 2009; Sugase et al., 2007). The N-terminal residues of R2 homologs could thus be involved in other functions including cell-cycle regulation, protein-protein interactions, and proteolysis, as is the case for the mammalian R2 protein.

15 Regulation of ribonucleotide reductase The maintenance of a correct and balanced pool of the dNTPs is a prerequisite for DNA replication and repair, which in turn are essential for genomic integrity (Reichard, 1988). Disturbance to the dNTP pool, being an increase or decrease, is mutagenic and may lead to genetic abnormalities and cell death in higher eukaryotes. Also distortions in the dNTP ratios, with one nucleotide being in excess, increase the likelihood of polymerase incorporation errors (Aye et al., 2014; Kunz and Kohalmi, 1991; Reichard, 1988). Due to this, RNR is among the most regulated enzymes. RNR is regulated throughout the cell cycle by diverse intricate mechanisms including transcriptional regulation, proteolysis of RNR subunits, as well as allosteric and oligomeric regulation. Further ways of modulating RNR activity, that until recently were found to be employed only in yeast, are via binding of small protein inhibitors. Thus, multilayered mechanisms interlock to regulate RNR, highlighting the critical importance of dNTP homeostasis (Nordlund and Reichard, 2006).

Transcriptional and post-translational regulation Proliferating cells require great amounts of the DNA precursors when the genome is duplicated during S phase. Thus, RNR activity is regulated with regard to the cell cycle. Transcription of the mammalian RRM1 gene peaks in S phase and is negligible in G0/G1 phase. At the protein level, mammalian R1 is basically constant throughout the cell cycle due to its long half-life (>24 hours) (Engstrom et al., 1985; Håkansson et al., 2006). Transcription of mammalian RRM2 also peaks in S phase and is minimal in G0/G1 phase. At the protein level, the abundance of R2 correlates with its mRNA level. The mammalian R2 protein has a short half-life (3-4 hours) and is in addition regulated via proteolysis (Chabes et al., 2003b; Engstrom et al., 1985). In late mitosis, the APCCdh1 mediates the degradation of R2 via binding to its N- terminal KEN-box. The APC is an E3 ubiquitin ligase, which is activated upon association with Cdh1 in late mitosis and targets R2 for degradation by ubiquitylation

(Chabes et al., 2003b; Jaspersen et al., 1999). In G2 phase, R2 is degraded via the SCF (Skp1-Cul1-F-box)Cyclin F ubiquitin ligase. This occurs after CDK-mediated phosphorylation of R2 on Thr33, which promotes binding of Cyclin F (Angiolella et al., 2012). Thus, fluctuations in R2 protein levels keep the activity of the R1-R2 complex in check and coordinate enhanced dNTP production with DNA replication.

16 In contrast to R2, low levels of p53R2 can be detected throughout the cell cycle, which answers the question of how proliferating G1 cells, as well as quiescent cells, obtain dNTPs for DNA repair when the R2 protein is absent (Chabes et al., 2003b; Nordlund and Reichard, 2006). Only in response to DNA damage are p53R2 levels high, which coincides with repression of the R2 protein levels. Both events are dependent on p53 activation by the checkpoint pathways (Tanaka et al., 2000; Yamaguchi et al., 2001) Activation of the checkpoints also induces RRM1 expression (Zhang et al., 2009). RNR is similarly controlled in both fission and budding yeast with the highest activity in S phase (Elledge et al., 1992; Fernandez Sarabia et al., 1993). In fission yeast, the cdc22 gene is transcriptionally induced at the G1-S phase transition. The suc22 gene produces two alternative transcripts: a short constitutively expressed transcript and a longer S phase-specific transcript. Both the cdc22 transcript and the longer suc22 transcript are induced by DNA damage (Fernandez Sarabia et al., 1993; Harris et al., 1996; Taylor et al., 1996).

Regulation by allostery and oligomerization The eukaryotic R1 subunit contains two binding sites for allosteric effectors termed the activity site and the specificity site, respectively. Overall enzymatic activity is regulated at the activity site, where binding of ATP induces enzyme activity to increase the dNTP pool (1) (Figure 1.6), and binding of dATP inhibits the enzyme (8). The regulation at the specificity site makes it possible for RNR to provide the correct proportions of each of the four dNTPs by binding and responding to dNDP species that are present in excess. Substrate specificity is determined by binding of ATP, dATP, dTTP, or dGTP, each inducing conformational changes of the protein structure, which rapidly adapt the catalytic site to specific substrates (Brown and Reichard, 1969). In particular, one loop of the protein structure (loop 2), forming a bridge between effector and substrate binding sites, has been shown to be important for this conformational change (Brown and Reichard, 1969; Jordan and Reichard, 1998; Xu et al., 2006).

17 Binding of ATP or dATP at the specificity site (2) (Figure 1.6) signals CDP or UDP to bind to the catalytic site (3), where they are reduced to their respective deoxy forms (dCDP and dUDP). dNDPs are in turn phosphorylated by nucleoside diphosphate kinase (NDPK) yielding dNTPs to be used for DNA synthesis. In addition, the biosynthesis of dTTP from dUDP requires thymidylate synthase (TS) and thymidylate phosphate kinase (TMPK) prior to phosphorylation by NDPK (Aye et al., 2014; Kunz and Kohalmi, 1991). Generated dTTP binds to the specificity site (4) (Figure 1.6) and selects for GDP to be converted at the catalytic site (5). After its generation from dGDP, dGTP binds to the specificity site (6) and signals the binding of ADP to the catalytic site (7). The conversion of ADP into dATP constitutes the final step in this sophisticated mechanism, as generated dATP binds to the activity site of RNR to inactivate the enzyme when the dNTP pool has reached a sufficient level (8). A mutation in the activity site within R1, where the conserved aspartic acid at position 57 is substituted by asparagine (R1D57N), eliminates the ability of RNR to discriminate between ATP and dATP. Thus, this mutation abolishes the dATP feedback inhibition of RNR (Caras and Martin, 1988). In mammalian cells, the dATP- feedback inhibition is strict and RNR is 50% inhibited in vitro at 5-10µM dATP (Eriksson et al., 1979; Reichard et al., 2000). In contrast, budding yeast has a relaxed feedback inhibition, where concentration higher than 50µM of dATP is required to inhibit RNR (Domkin et al., 2002). The R1D57N mutation leads to an elevated dNTP pool and a mutator phenotype in budding yeast as well as in a murine T-lymphoma cell line (Caras and Martin, 1988; Chabes et al., 2003a). In fission yeast, the Cdc22D57N mutation leads to a 6-12 fold increase in the concentrations of the four dNTPs, which also correlated with increased mutation rates. This demonstrates that fission yeast similarly to mammalian cells relies on a very stringent dATP feedback inhibition mechanism (Fleck et al., 2013).

18

……………………………………………………………………………………………………………… Figure 1.6. of RNR ATP and dATP can bind to both the activity site (A) and the specificity site (S). Binding of ATP to the activity site stimulates enzyme activity (1), whereas binding of dATP inhibits the enzyme (8). Binding of ATP and dATP at the specificity site (2) stimulate the reduction of pyrimidine ribonucleoside diphosphates (3). Generation of dTTP (4) provides the specificity for GDP reduction (5), and generated GTP (6) provides the specificity for ADP reduction (7). In this way, a single enzyme can supply the cell with roughly equimolar amounts of the dNTPs. Abbreviations used: nucleoside diphosphate kinase (NDPK), thymidylate synthase (TS), thymidylate phosphate kinase (TMPK). Sequential arrows indicate further processing by other enzymes. ……………………………………………………………………………………………………………....

19 It has been shown that the dATP-mediated inhibition of RNR is associated with oligomerization of the RNR subunits. Indeed, formation of complex higher-order multimers seems to be a common feature to allosteric inhibition of class Ia RNRs, although the mechanism for this is poorly understood (Ando et al., 2011; Crona et al., 2016; Fairman et al., 2011; Kashlan et al., 2002; Rofougaran et al., 2006). In E. coli, inhibition by dATP occurs through formation of an α4β4 state of RNR, in which the distance between the radical generating residue within β is too far (∼60Å) from the receiving cysteine in the catalytic site to support radical transfer. In addition, the disruption of a compact α2β2 state exposes the amino acid residues involved in radical transfer, to bulk water (Ando et al., 2011; Chen et al., 2018).

In human cells, a dynamic equilibrium between α, α2, and α6 forms of human

R1 has been shown by structure-based data. α6 forms increased with increasing dATP levels and associated with β to form complex high-order multimers (Fairman et al., 2011; Rofougaran et al., 2006). The R1D57N mutation in the activity site prevents formation of hexamers, which support the finding that oligomerization is a prerequisite for RNR inhibition by dATP (Chanpimol et al., 2017; Fairman et al., 2011).

20 Regulation by small protein inhibitors In addition to transcriptional regulation and regulation by allosteric effectors, RNR can be controlled via binding of small protein inhibitors belonging to the family of intrinsically disordered proteins (IDPs). IDPs are characterized as fully functional proteins lacking a well-structured three-dimensional fold (Wright and Dyson, 2015). Rather, they fluctuate through a range of conformations, adopting diverse structures on different targets. This characteristic enables IDPs to interact promiscuously with multiple binging partners, which occurs with high specificity but very low affinity (Sugase et al., 2007). IDPs are important for the regulation of signaling pathways and mutations within these proteins are often associated with disease (Wright and Dyson, 2015). In fission yeast, Spd1 (for S phase delayed) was identified in a screen for proteins that block the cell cycle, when overexpressed (Woollard et al., 1996). Spd1 was later shown to belong to the family of IDPs and is known to be an inhibitor of RNR operating through at least two mechanisms (Nestoras et al., 2010). In vitro studies have shown that Spd1 can inhibit fission yeast RNR activity by binding to the large Cdc22 subunit (Hakansson et al., 2006). Spd1 can also bind and sequester the small Suc22 subunit in the nucleus of cells (Liu et al., 2003; Nestoras et al., 2010; Nielsen, 2003). The bulk of Cdc22 and Suc22 constitutively localize to the cytoplasm, where the holoenzyme produces dNTPs that diffuse into the nucleus for DNA synthesis. By nuclear sequestration of Suc22, Spd1 ensures less holoenzyme formation and thus less RNR activity. Even though both mechanisms result in restricted RNR activity, they can be genetically separated (Nestoras et al., 2010). Additional roles for Spd1 have been suggested, including a role in modulation of the overall architecture of the RNR holoenzyme (Nestoras et al., 2010). A second Spd1-related protein, Spd2, has been identified in fission yeast that likewise belongs to the family of IDPs. Spd2 appears to play a role in a subset of Spd1-regulated processes, however unlike Spd1, Spd2 is not required for nuclear sequestration of Suc22 and does not seem to effect cellular dNTP levels (Vejrup- hansen et al., 2014).

21 RNR in budding yeast is likewise regulated by the small protein inhibitors Sml1 and Dif1 that according to synteny analysis arose from a whole genome duplication of the S. cerevisiae genome (Lee et al., 2008; Wolfe and Shields, 1997). Sml1 binds directly to the large Rnr1 subunit to restrict enzyme activity (Chabes et al., 1999; Zhao et al., 1998), whereas Dif1 is required for nuclear import of the small subunits (Rnr2 and Rnr4), when cells are outside of S phase (Wu and Huang, 2008). Thus, fission yeast Spd1 is a functional ortholog of both Sml1 and Dif1 (Lee et al., 2008). Limited sequence conservation of the small RNR inhibitors in different species complicated the process of identifying orthologs in mammalian cells. The process was further complicated by their disordered nature (Salguero et al., 2012). However, a conserved metazoan protein termed IRBIT (IP3R binding protein released with inositol 1,4,5-triphosphate) was identified in 2003 (Ando et al., 2003), of which depletion in HeLa cells caused imbalanced dNTP pools and altered cell cycle progression (Arnaoutov and Dasso, 2014). It has been shown that IRBIT stabilizes dATP in the activity site of RNR thereby inhibiting the enzymatic activity. Thus, IRBIT is a novel regulator of RNR in higher eukaryotes. The N-terminal domain of IRBIT is intrinsically disordered and a further domain (amino acid 64 to 87) shows some similarity to the central region of budding yeast Sml1 (amino acid 46 to 72). In line with this, Sml1 has been shown to bind and inhibit the mammalian R1 subunit (Chabes et al., 1999; Zhao et al., 2000).

22 Ubiquitin-mediated regulation of RNR inhibitors In fission yeast, Spd1 inhibits RNR outside of S phase and in the absence of DNA damage. At this stage, the majority of the Suc22 subunits are compartmentalized to the nucleus (Liu et al., 2003). During S phase and in response to DNA damage, RNR must escape the Spd1 restrain to ensure that dNTPs are available for replicative and repair DNA synthesis. This is achieved by targeted degradation of Spd1 mediated by the CRL4Cdt2 complex. The CRL4Cdt2 complex is an E3 ubiquitin ligase consisting of the cullin scaffold protein Cul4 (Pcu4 in fission yeast), the adapter protein Ddb1 and the substrate-recruiting factor Cdt2 (Holmberg et al., 2005; Liu et al., 2003, 2005). Activation of the CRL4Cdt2 complex occurs by transcriptional induction of Cdt2, which is mediated by the MluI binding factor (MBF) prior to S phase in unperturbed cells (Liu et al., 2005). After DNA damage in

G2 cells, the DNA damage checkpoint enforced by Rad3 and Chk1 is required for transcriptional induction of Cdt2 (Liu et al., 2005; Salguero et al., 2012; Watson et al., 2004). Polyubiquitylation of Spd1 by the CRL4Cdt2 complex directs it for degradation, which requires interaction with chromatin-associated proliferating cell nuclear antigen (PCNA) (Salguero et al., 2012), and the Csn1 and Csn2 subunits of the COP9/signalosome (CSN) (Liu et al., 2003) (Figure 1.7). PCNA is a polymerase processivity factor during replication and damage induced DNA synthesis, and thus provides a direct link between RNR activity and DNA synthesis (Salguero et al., 2012). The net reduction of Spd1 levels results in delocalization of Suc22 from the nucleus and RNR complex formation, which in turn results in elevated dNTP levels (Holmberg et al., 2005; Liu et al., 2003). Mammalian cells also contain the CRL4Cdt2 pathway, which drives many cell cycle transition events (Salguero et al., 2012). However, whether the metazoan RNR inhibitor IRBIT is regulated trough this pathway remains to be determined. The CRL4Cdt2 pathway is not conserved in budding yeast, and thus not responsible for degradation of Sml1 and Dif1 in S phase and after DNA damage. Instead, activation of the Mec1/Rad53 DNA damage checkpoint leads to phosphorylation of the protein inhibitors by Dun1. The phosphorylates inhibitors are subsequently ubiquitylated by the Rad6-Ubr2-Mub1 E2/E3 ubiquitin ligase and eventually degraded (Andreson et al., 2010; Zhao et al., 2001).

23 ……………………………………………………………………………………………………………. Figure 1.7. RNR activity controlled by Spd1 Outside of S phase, Spd1 inhibits RNR by direct binding to Cdc22 in the cytoplasm and by nuclear sequestration of Suc22. DNA synthesis associated with DNA replication or DNA damage repair leads to polyubiquitylation of Spd1 by the CRL4Cdt2 complex consisting of Cul4, Ddb1, and Cdt2. This reaction requires interaction with chromatin-bound PCNA. Spd1 is subsequently degraded by the COP9/signalosome (CSN), which results in activation of RNR and subsequently increased levels of the four dNTPs. ……………………………………………………………………………………………………………. ….

24 CRL4Cdt2 inactivation CRL4Cdt2 defective fission yeast cells fail to degrade Spd1 and thus undergo DNA replication in its presence. This leads to decreased dNTP pools and consequently slow progression through S phase, increased mutation rates and DNA damage sensitivity. Furthermore, Spd1 accumulating cells are completely unable to undergo meiotic differentiation as they fail to enter pre-meiotic S phase, and their survival relies notably on the activation of the Rad3 checkpoint (Bondar et al., 2003; Holmberg et al., 2005; Liu et al., 2003). The checkpoint dependency and meiotic defect of Spd1 accumulating cells can be suppressed by Spd1 loss or alternatively by Suc22 overexpression, which previously has been attributed the re-establishment of sufficient levels of dNTPs (Holmberg et al., 2005; Liu et al., 2003). Interestingly, Fleck et al., 2013 showed that the presence of a Cdc22D57N mutation neither suppressed checkpoint activation nor dependency in cells that fail to degrade Spd1, even though the mutation led to a 6-fold increase in the dATP levels. Furthermore, we have recently demonstrated that the Rad3 checkpoint dependency in Spd1 accumulating cells cannot be suppressed by an exogenous supply of deoxynucleosides to cells with an established salvage pathway ( Fleck et al., 2017). These findings have led to the suggestion that Spd1 interferes with other unknown processes in addition to restraining RNR activity to regulate dNTP pools, plausibly assisted by Spd2 (Fleck et al., 2013, 2017; Vejrup-hansen et al., 2014). The indications of Spd1 and Spd2 having multiple roles are in accordance with the promiscuous nature of IDPs (Milles and Lemke, 2014).

25 It is unknown how exactly the different pathways work together to regulate RNR and maintain balanced and correct levels of the four dNTPs. However, the vast number of regulatory mechanisms emphasizes the importance of doing so. As a complementary route to de novo dNTP biosynthesis, most cells have salvage pathways and substrate cycles to fine-tune the dNTP levels (Reichard, 1988). Fission yeast does not have a salvage pathway for uptake and phosphorylation of deoxynucleosides, and thus depends solely on the ribonucleotide reduction pathway. Recently we have shown that a salvage pathway could be established in fission yeast by engineering strains to express the genes for the human equilibrative nucleoside transporter 1 (hENT1) and the Drosophila melanogaster deoxynucleoside kinase (DmdNK) (Fleck et al., 2017). We showed that this salvage pathway is functional as 2 µM of deoxynucleosides in the culture medium could rescue the growth of two different temperature-sensitive alleles controlling RNR. Furthermore, we showed that salvage could suppress the killing of ∆rad3 cells exposed to the known RNR inhibitor hydroxyurea. This work is presented in Paper I.

26 Results

Generation of fission yeast strains depending on human RNR RNR has long been considered an excellent target for both anti-cancer, anti- bacterial and anti-viral drugs because of its essential role in cells, and due to the fact that RNR subunits are overexpressed in various cancer tissues. Much effort is being devoted to identify and develop novel inhibitors of this enzyme. In this context, thorough knowledge of how RNR is regulated with respect to both intracellular and external cues is of utmost importance. For the purpose of studying the properties and regulation of human RNR in a simple system, recombinant fission yeast strains were created, which expressed either the human R1-R2 or R1-p53R2 holoenzyme, respectively. Through a four-step strategy, the genes encoding yeast RNR were replaced with the genes encoding human RNR. First, a construct containing human RRM2R2 cDNA under control of the heterologous, thiamine-repressible nmt1 promoter (Maundrell, 1990) was used for targeted integration at the endogenous leu1 locus (Figure 1.8, step I). In this background, a construct containing human RRM1R1 cDNA and a KanMX cassette conferring resistance to G418 were recombined into the cdc22 locus, thereby replacing the cdc22-coding region (step II). A DNA fragment containing either RRM2R2 or RRM2Bp53R2 cDNA was used for targeted integration at the chromosomal suc22 locus (step III). As the requirement for both human RNR subunits constitutes a built-in selection approach (Supplementary Figure S1), the positive transformants were efficiently achieved upon plating onto medium containing thiamine (repressing the expression of RRM2R2 from the leu1 locus (Supplementary Figure S2)). Successful integrations were confirmed by appropriate selection and PCR. The RRM2R2 gene integrated at the leu1 locus was subsequently crossed out (step IV), resulting in two recombinant yeast strains expressing the genes encoding human RNR from the native fission yeast promoters; RRM1R1RRM2R2 (created by H. B. Nielsen (Nielsen, 2013)) and RRM1R1RRM2Bp53R2 (created in this study). The two strains will collectively be referred to as the human RNR (hRNR) strains, but will when appropriate be clearly specified as either hR1-R2 or hR1-p53R2.

27

……………………………………………………………………………………………………………… Figure 1.8. Generation of recombinant fission yeast strains with human RNR Step I: Integration of a nmt1-RRM2 construct at the chromosomal leu1 locus using a linear pDUAL fragment (Matsuyama et al., 2004). Positive clones were selected based on leucine prototrophy and PCR analysis. Step II: A cloned DNA fragment containing human RRM1 cDNA and a KanMX cassette was integrated at the cdc22 locus. Positive clones were selected based on resistance to G418 and PCR analysis. Step III: A linear DNA fragment containing either human RRM2 or RRM2B cDNA was integrated at the chromosomal suc22 locus. Positive clones were efficiently achieved upon plating onto medium with thiamine and by PCR analysis. Step IV: Successful integration of RRM2/B at the suc22 locus allowed cross-out of the nmt1-RRM2 construct at the leu1 locus. Arrows indicate primers used to verify integrations. ………………………………………………………………………………………………………………

28 Strain characterization The two recombinant fission yeast strains with human RNR were characterized and compared with wild type. By microscopy, it was shown that the hRNR cells appeared normal (Figure 1.9A). Furthermore, the viability of the strains was similar to wild type (Figure 1.9B), and they showed a normal DNA content profile (data not shown). When cultured in rich medium, the hRNR strains grew slightly slower than wild type (Figure 1.9C-D).

……………………………………………………………………………………………………………… Figure 1.9. Characterization of fission yeast depending on human RNR (A) Cell morphology of hRNR cells compared with wild type (wt). Images of exponential growing cells from indicated cultures were obtained using the Xcyto® Quantitative Cell Imager (ChemoMetec). (B) Viability of hRNR cells analyzed by spot assay and compared with wild type. 10-fold serial dilutions of each strain were spotted onto solid rich medium (YES) and incubated at 30°C for 3 days. (C) Growth curves. Cells were grown to exponential phase, diluted to 1.5 x 106 cells/mL and grown for 8 hours. The cell density of each culture was counted every hour using the Xcyto® Quantitative Cell Imager. (D) Average cell lengths and doubling times. The average doubling times were calculated from data presented in (C). The average cell lengths were based on 100 cells from each strain. ………………………………………………………………………………………………………………

29 To monitor the effect of making fission yeast cells depend on human RNR for dNTP biosynthesis, dNTP pools were determined from asynchronous, exponential growing cultures. Small-molecule extracts were prepared from the samples as described in Materials and Methods and used as source of nucleotides in four different primer extension assays, each specific for a unique dNTP. The protocol makes use of 32P-labeled primer/template hybrids, which allow incorporation of a given dNTP at specific positions resulting in products of different lengths. The reaction products were separated in a 10% polyacrylamide-urea gel. To determine the amount of dNTPs incorporated into the primer/templates, a dNTP standard was included (Supplementary Figure S3). The measurements are presented as the mean percentage of the dNTP to ATP normalized to wild type levels (Figure 1.10A). No substantial change in the dNTP levels was observed in the hRNR strains compared with wild type. Similar results with different approaches have previously been obtained. One experiment used an HPLC column, whereas the other used the primer extension assay described here (Figure 1.10B). In both experiments, hR1-R2 cells were shown to have dNTP levels comparable to wild type. Furthermore, both studies showed that deletion of spd1 in hR1-R2 cells resulted in elevated dNTP pools (1.7-fold increase in the dATP level). By comparison, deletion of spd1 in the wild type background resulted in a 1.78-fold increase in the dATP level. Interestingly, the dNTP pools measured in this study were not elevated in the hR1-p53R2 ∆spd1 cells compared with the hR1-p53R2 cells (Figure 1.10A).

Most asynchronous cells are in G2 phase, which is particularly long in the cell cycle of fission yeast. However, as the amount of dNTPs increases during S phase to allow an appropriate amount of DNA precursors for DNA synthesis, I speculated that even a small variation in the fraction of S phase-cells between strains could result in dNTP pool inconsistencies. I therefore decided to measure the dNTP pools in S phase of synchronized cells.

30 Cultures were synchronized in G1 by starving the cells for nitrogen in the presence of M-factor pheromone as described in Materials and Methods. By this method, 80%-90% of the cells arrest in G1 (Nielsen, 2016). Release from the G1 arrest was induced by addition of arginine, and the progression through the cell cycle was followed by taken samples every 30 minutes (Figure 1.10C). Samples for dNTP measurements were taken at the time point with the most S phase cells. Small- molecule extracts were prepared as described above. Only dATP levels were measured in S phase-cells and are presented as the mean percentage of dATP to ATP normalized to wild type in Figure 1.10D. The table in Figure 1.10E summarizes the dATP levels as well as the fold increase between asynchronous cells (G2 phase) and S phase-cells. In some strains, the dNTP pool was measured only in G2 phase or S phase.

Consistent with the dNTP pools measured in G2 phase, the dATP level in S phase of hR1-R2 cells was comparable to that of wild type and increased upon deletion of spd1 – albeit more modest. Furthermore, deletion of spd1 in the hR1- p53R2 background had no effect on the dATP levels. However, whereas the dNTP pool of hR1-p53R2 cells was comparable to that of wild type in G2 phase, the level of dATP was significantly lower in S phase. Almost no increase in the dATP level was observed between G2 and S phase, which could indicate a problem of the hR1-p53R2 holoenzyme in providing the cells with an appropriate amount of the dNTPs for DNA synthesis in S phase. However, it should be noted that G1 arrest was difficult to obtain in all the strains with human RNR, which is evident from the cell cycle profiles depicted in Figure 1.10C. Furthermore, these cells had problems recovering from the arrest, which is consistent with the delay in S phase progression compared to wild type, and the broader peaks in later time points. The cause of this observation, and whether it affects the dNTP pool measurements, is unknown. Taken together, these results show that human RNR can complement for lack of the yeast enzyme to support viability and cellular growth. Minor alterations in the dNTP levels might account for the slightly slower growth of hRNR cells compared with wild type. Similar to observations in wild type cells where Spd1 restrains the Cdc22-Suc22 complex, Spd1 seems to restrain the hR1-R2 complex to regulate dNTP pools. In contrast, it does not seem to target the hR1-p53R2 complex.

31 ………………………………………………………………………………………………………………… Figure 1.10. Deoxynucleotide measurements

(A) G2 phase dNTP pools. Small-molecule extracts from asynchronous cultures were used as source of nucleotides in four different primer extension assays, each specific for a unique dNTP. dNTP pools are presented as the mean percentage of the dNTP to ATP normalized to wt. Error bars, s.d. (n=3). (B) G2 phase dNTP pool measurements performed by Christian Holmberg as described in (A) (n=1). (C) Cell cycle profiles from synchronized cells. Cells were synchronized by nitrogen starvation in the presence of

M-factor pheromone (Nielsen, 2016). Arginine was added at time point 0’ to release the cells from G1 arrest. Samples were taken with a 30 minutes interval and the samples with the most S phase cells (marked in red) were used for dNTP pool measurements. (D) S phase dATP pools. dATP levels are presented as the mean percentage of dATP to ATP normalized to wt. Error bars, s.d. (n=2). (E) Table summarizing the dATP pools measured in G2 and S phase cultures, as well as the fold increase. …………………………………………………………………………………………………………………

32 The effect of spd1 overexpression on cell cycle progression To further investigate the possibility of the hR1-R2 complex being susceptible for Spd1 inhibition, I ectopically overexpressed spd1 in the recombinant fission yeast strains. In general, the phenotypes associated with excess Spd1 are much more severe than those caused by Spd1 loss (Holmberg and Nielsen, 2012). Elevated Spd1 levels cause increased mutation rates, damage sensitivity and checkpoint activation and dependency. Furthermore, cells accumulating Spd1 fail to undergo premeiotic S phase (Holmberg et al., 2005). I transformed different strains with plasmids containing spd1 under control of the strong thiamine-repressible nmt1 promoter. Cells were grown to exponential phase in medium containing thiamine (promoter off). The cultures were then washed twice and grown in fresh medium without thiamine (promoter on) for 16 hours. The effect of Spd1 accumulation was examined by cell cycle progression analysis (Figure 1.11A) and microscopy (Figure 1.11B). Before Spd1 accumulation, all the cells exhibited normal cell morphology and a 2C DNA content peak characteristic for fission yeast. After 16 hours of Spd1 accumulation, wild type cells were dramatically elongated and arrested with both 1C and 2C DNA contents, as previously described (Woollard et al., 1996). In contrast, minor cell elongation was observed in cells with human RNR. In hR1-p53R2 cells, the cell cycle profiles before and after Spd1 accumulation were comparable. Intriguingly, a broader peak representing a more heterogeneous distribution of DNA content was observed in hR1-R2 cells after accumulation of Spd1. This may present slowly replicating hR1-R2 cells as a consequence of elevated Spd1 levels, which further supports the proposition that Spd1 might be able to restrain the hR1-R2 complex to regulate dNTP pools. Whether Spd1 accumulated to similar levels in the different strains have not been demonstrated.

33

……………………………………………………………………………………………………………... Figure 1.11. Effects of ectopic overexpression of spd1 Ectopic expression of spd1 was driven by the strong thiamine-repressible nmt1 promoter. Cells were grown to exponential phase in medium with thiamine to repress expression. The cells were subsequently washed and incubated in medium lacking thiamine to allow overexpression of spd1 for 16 hours. (A) Cell cycle profiles of indicated strains with the promoter off (black) or on for 16 hours (blue). (B) The morphology of cells with the promoter off (upper images) or on for 16 hours (lower images) analyzed using the Xcyto® Quantitative Cell Imager. ……………………………………………………………………………………………………………...

34 Human RNR in fission yeast confers checkpoint activation and dependency A previous study using the recombinant fission yeast strain with the hR1-R2 complex, showed that viability could only be maintained in the presence of a functional Rad3ATR checkpoint pathway (Nielsen, 2013). In fission yeast, Rad3 is the major transducing kinase and it is activated by a wide range of DNA-damaging agents and in response to S phase perturbations (Lindsay et al., 1998; Martinho et al., 1998). To study the need for a functional checkpoint in the hR1-p53R2 strain, I crossed it to a rad3 temperature-sensitive (rad3-ts) mutant. Cells harboring the rad3-ts allele are checkpoint-proficient at 25°C and checkpoint-deficient at 35°C (Martinho et al., 1998). A spot assay was made, in which the viability of the strains were studied when shifted from the permissive to the restrictive temperature (Figure 1.12A). The result demonstrated that rad3 is required for normal cell growth in both strains depending on human RNR, as the viability of the hR1-R2 rad3-ts and hR1-p53R2 rad3-ts double mutants decreased when grown at the restrictive temperature (Figure 1.12A, right panel). In contrast, wild type cells were independent of rad3 under unperturbed conditions. Thus, human RNR in fission yeast confers checkpoint dependency. There are two branches of the Rad3 checkpoint pathway, the replication branch in which Cds1 is the major effector, and the DNA damage branch in which Chk1 is mediating the checkpoint signal (Lindsay et al., 1998). To define whether both of the branches were required for tolerating human RNR, ∆cds1 and ∆chk1 mutants were crossed to all the genetic backgrounds and isolates were used in a spot assay (Figure 1.12B). Under unperturbed conditions, the ∆cds1 and ∆chk1 mutants grew like wild type cells. In contrast, both mutations conferred reduced viability when crossed into cells with human RNR. However, whereas deletion of the cds1 gene resulted in severe synthetic sickness, deletion of the chk1 gene had a more modest effect. This indicates that the replication checkpoint is required for normal viability of fission yeast cells with human RNR. Despite considerable effort, no cells with human RNR and both of the checkpoint gene deletions were isolated, indicating that the presence of human RNR in checkpoint deficient cells results in synthetic lethality. As Cds1 and Chk1 work downstream of Rad3, synthetic lethality would likewise be expected in the hRNR rad3-ts strains grown at the restrictive temperature. However, the checkpoint deficiency of the temperature-sensitive mutant is not as severe as that of the deletion (∆rad3) mutant, which might explain the discrepancy (Anda et al., 2016).

35 To support these findings, the activation of Chk1 and Cds1 was examined. Chk1 is believed to be directly phosphorylated and activated by Rad3 in response to DNA damage, which is readily monitored by a reduction of its electrophoretic mobility in SDS-PAGE (Capasso, 2002). Therefore, I performed a Western blot of hemagglutinin-tagged Chk1 (Chk1-HA). Protein extracts for Western blotting were prepared from asynchronous, exponentially growing cultures by TCA extraction, and the phosphorylation of Chk1 was visualized using anti-HA antibodies. Exposure of wild type cells to 40 µM CPT, a topoisomerase 1 poison causing DNA damage, elicited a characteristic mobility shift (Figure 1.12C, lane 1). In contrast, no mobility shift was detected in untreated cells, indicating that Chk1 is not phosphorylated and activated in wild type (lane 2) and hRNR cells (lane 3-4) under unperturbed conditions. To investigate Cds1 activation, the phosphorylation of myelin basic protein (MBP, an artificial substrate of Cds1) was monitored in an in vitro kinase assay. Briefly, HA-tagged Cds1 (Cds1-HA) was immunopurified using anti-HA antibodies and Protein G Sepharose beads. The purified kinase was incubated together with MBP in the presence of radioactive [γ-32P]ATP and the phosphorylated products were separated in SDS-PAGE. The lower part of the gel containing MBP (~18.4 kDa) was used for phosphorimaging, whereas the upper part containing Cds1-HA (~53 kDa) was used for Western blotting. As evident from Figure 12D, Cds1 activity was low in wild type (lane 2). In contrast, the presence of human RNR caused strong induction of the kinase activity (lane 3-4) with the strongest activation found in hR1-p53R2 cells (lane 4). A wild type strain with untagged Cds1 was included as control to demonstrate specific immunoprecipitation (lane 1). Taken together, the presence of human RNR in fission yeast leads to checkpoint activation and dependency. Whereas the DNA damage checkpoint (Chk1) might play a minor role, the replication checkpoint (Cds1) is absolutely required to prevent synthetic sickness in cells depending on human RNR. Presumably, the absence of Cds1 in hRNR cells may cause a need for the Chk1 sub-pathway. Supporting this notion, hRNR cells are not very sensitive to DNA damaging agents including methyl methanesulfonate (MMS), camptothecin (CPT), and zeocin (Nielsen, 2013) but are hypersensitive to DNA replication stress induced by hydroxyurea (HU) (data not shown).

36 HU is a well-established inhibitor of RNR and it has been used for the treatment of multiple human diseases including cancer (Huang et al., 2016; Krakoff et al., 1968; Singh and Xu, 2016). HU specifically scavenges the catalytically important tyrosyl free radical within the small subunit of RNR, leading to decreased dNTP levels, which causes replication forks to stall and cells to arrest in S phase. In fission yeast, Cds1 is activated in response to HU treatment and serves to prevent detrimental mitosis with unreplicated or damaged chromosomes. This is associated with dramatic cell elongation. Curiously, the constitutive activation of Cds1 in hRNR cells was not associated with cell cycle arrest and cell elongation. Instead, these cells appeared normal and grew similar to wild type. When I examined the microscopic appearance of the hRNR cells exposed to HU, I saw that these cells elongated similar to wild type (Figure 1.12E). This indicates that the function of Cds1 is intact and serves to arrest the cells upon nucleotide depletion. It might also indicate that Cds1 is constitutively activated in hRNR cells independently of dNTP levels, which would be consistent with the results presented above.

Figure 1.12 Cont. 37

…………………………………………………………………………………………………………… Figure 1.12. Human RNR in fission yeast confers checkpoint activation and dependency (A) Viability of rad3-ts mutants. 10-fold serial dilutions of indicated strains were spotted onto YES medium and incubated at the permissive (25°C) and restrictive temperature (35°C) for 3-4 days. (B) Viability of ∆cds1 and ∆chk1 mutants. Performed as described in (A) and incubate at 30°C. (C) Western blot detecting phosphorylation of Chk1. All strains contained an HA-tagged copy of Chk1 replacing the endogenous Chk1. Extracts were obtained from mid-log phase cells treated with either 40 µM CPT in DMSO or DMSO alone as indicated in the figure. CPT treatment caused a characteristic mobility shift representing phosphorylated states of Chk1 (Chk1-HA-P). (D) In vitro kinase assay. All strains contained an HA-tagged copy of Cds1 replacing the endogenous Cds1. Extracts for IP were obtained from mid-log phase cells, and Cds1 kinase activity against MBP was monitored (phosphorylation of MBP is proportional to Cds1 activation). Untagged cells were used as control for HA-specific IPs (Control). Lower panel shows a Western blot of Cds1-HA with immunoglobulin G (IgG) heavy chain serving as a loading control. The kinase assay is a representative of three independent experiments with similar results. MBP band intensities were quantified with respect to the loading control and normalized to wt. Values are presented under the immunoblot as mean±s.d. (n=3) (E) Microscopic appearance of cells treated with HU. Cells were treated with 0.4 mM HU for 6 hours before imaging (lower images) and compared to untreated cells (upper images). ……………………………………………………………………………………………………………

38 Elevated levels of the R2 homologs suppress checkpoint activation and dependency To reconcile the characteristics of wild type cells and cells with human RNR, a chimeric RNR small subunit (chimR2) was constructed. This subunit consists of the first 69 amino acid residues of Suc22 in frame with the residues 69-389 of human R2 (Supplementary Figure S4). The recombinant gene encoding chimR2 was integrated in the suc22 locus of cells already containing the human R1 subunit. The strain will be referred to as the hR1-chimR2 strain (created by A. C. Eidesgaard under my supervision (Eidesgaard, 2017)). The rationale for creating the chimR2 subunit was that the less conserved region between yeast Suc22, human R2, and p53R2 is located to the N-terminus of their sequences. Thus, the N-terminus of Suc22 could possibly be engaged in an unidentified function, which is not complemented by the subunits of human origin and therefore leads to constitutive checkpoint activation. The reasoning of a function of the Suc22 N-terminus was further augmented by its unstructured nature. Protein structure predictions suggest that both the N- and C-termini of RNR β subunits show a similar pattern of disorder (Brignole et al., 2018; Eklund et al., 2001; Smith et al., 2009). IDPs and proteins containing intrinsically disordered regions (IDRs) are known to interact promiscuously with multiple interaction partners (Kriwacki et al., 2002; Sugase et al., 2007), and thus yet unidentified functions might map to the N-terminal sequences of the R2 homologs. The dependency of cds1 observed in the hRNR strains (Figure 1.12B) was indeed suppressed in the hR1-chimR2 strain (Figure 1.13A). Consistently, as evident from the in vitro kinase assay depicted in Figure 1.13B, Cds1 activity in hR1-chimR2 cells was low as in wild type. To make sure that the suppression was not due to altered protein levels, I performed a Western blot using antibodies against hR2/p53R2 (Figure 1.13C). Surprisingly, a higher level of the chimR2 protein was detected compared with the levels of the hR2 and p53R2 proteins.

39 The anti-R2/p53R2 antibodies were raised against amino acid 90-389 mapping at the C-terminus of R2 of human origin, but cross-react with human p53R2 and yeast Suc22. Due to the high sequence conservation between hR2 and hp53R2, I expect the antibodies to bind with equal affinity to the hR2, hp53R2 and chimR2 subunits. In Figure 1.13C, the protein levels of these subunits have been normalized to the level of Suc22 in wild type for better comparison between individual experiments. However, I do no expect the antibodies to bind with similar affinity to Suc22, which is why I do not draw any parallels between the protein levels of yeast and human proteins. The result was unexpected as the genes encoding the respective small subunits are all expressed from the endogenous suc22 promoter. The protein level of hp53R2 was also ∼2-fold higher than the level of hR2, although lower than the level of chimR2. The reason for this fluctuation in protein levels between strains has not been clarified. It was not possible to determine if the suppression of checkpoint dependency in hR1-chimR2 cells was due to the presence of the N terminus of Suc22 or due to stabilization of the chimR2 protein. Furthermore, the hR1-chimR2 strain was found to have very high dNTP levels (the dATP level in S phase was ∼4.4-fold higher than that of wild type, Figure 1.10E), which might also confer suppression. To test the influence of RNR protein levels in relation to the checkpoint dependency observed in hRNR cells, a second RRM2R2 gene was crossed into the hR1-R2 genetic background. The second gene was expressed from the endogenous leu1 locus under control of the nmt1 promoter. The protein level of hR2 in cells with the nmt1-RRM2R2 construct was ∼2.3-fold higher than that of cells with RRM2 expressed only from the endogenous suc22 locus (Figure 1.13C). Elevated hR2 levels did indeed suppress the checkpoint activation in the hR1-R2 background (Figure 1.13B). Similarly, ectopic overexpression of RRM2Bp53R2 rescued the checkpoint dependency of cells with human RNR (data not shown). To assess whether the suppression was due to elevated dNTP pools as a consequence of excess hR2, I included an hR1D57N-R2 mutant in the analysis. The hR1D57N mutation destroys the dATP feedback inhibition of RNR and thus leads to higher dNTP pools. However, no suppression of the checkpoint dependency was observed in hR1D57N-R2 rad3-ts cells grown at the restrictive temperature (Figure 1.13D, right panel), indicating that high dNTP pools alone are not the cause of suppression.

40 Taken together, the results show that elevated levels of the R2 homologs can suppress the checkpoint activation and dependency of cells relying on human RNR. Thus, the N-terminus of Suc22 itself is not important for suppression of the checkpoint dependency. Rather it is indirectly involved in the suppression by stabilizing the chimR2 subunit via an unknown mechanism. Furthermore, the suppression is independent of dNTP pools.

…………………………………………………………………………………………………….……… Figure 1.13. Elevated levels of the R2 homologs suppress checkpoint activation and dependency (A) Viability assay. 10-fold serial dilutions of indicated strains were spotted onto YES medium and incubated at 30°C for 3 days. (B) In vitro kinase assay. Performed as described in Figure 12D (n=1). (C) Western blot of the R2 homologs. Cell extracts were obtained from mid-log phase cells and anti- R2/p53R2 antibodies were used for detection. The presented immunoblot is a representative of two independent experiments. Band intensities were quantified with respect to the loading control (anti- tubulin) and normalized to wt. Values are presented under the immunoblot as mean±s.d. (n=2). (D) Viability assay performed as described in (A). Cells were incubated at the permissive (25°C) and the restrictive (35°C) temperature for 3-4 days. ……………………………………………………………………………………………………….…… … 41 In vivo function(s) of Spd1 Similar results have been obtained in fission yeast cells accumulating Spd1. The checkpoint activation and dependency in these cells can be suppressed by spd1 deletion or alternatively by overexpression of Suc22 (Holmberg et al., 2005; Liu et al., 2003). A previous interpretation of this observation was that excess Suc22 serves to counterbalance the lower cytoplasmic levels of this subunit promoting RNR complex formation and dNTP synthesis (Nestoras et al., 2010). However, recent findings have provided evidence that the suppression is independent of dNTP levels (Fleck et al., 2013, 2017). This has led to the suggestion that Spd1 interferes with other unknown processes in addition to restraining RNR activity to regulate dNTP pools. The other function(s) of Spd1, plausibly assisted by Spd2, results in checkpoint activation and dependency independently of the dNTP levels (Fleck et al., 2013, 2017; Vejrup-hansen et al., 2014). As Suc22 can compete with the interaction between Spd1 and its unknown target(s), an elevation in the level of Suc22 leads to suppression of the checkpoint activation and dependency. If the mechanism by which Spd1 elicit a checkpoint dependency is independent of its ability to restrain RNR, then this mechanism could be the cause of the constitutive checkpoint activation in hRNR cells. The rescue of checkpoint dependency by elevating the level of hR2, hp53R2 or chimR2 would then be dependent on direct interaction with Spd1 to compete with the interaction between Spd1 and its unknown target. To assess whether Spd1 could interact directly with the R2 homologs, I performed indirect immunofluorescence microscopy to determine their sub-cellular localization. In wild type, Spd1 binds directly to Suc22 and sequesters it in the nucleus, outside of S phase. Thus, the majority of Suc22 subunits can be detected in the nucleus of asynchronous cell. I reasoned that hR2, p53R2 and chimR2 would likewise be localized in the nucleus if the ability to interact with Spd1 had been preserved.

42 Asynchronous, exponentially growing cultures were fixed in cold methanol and subjected to staining with anti-R2/p53R2 primary antibodies (cross-reactivity facilitated Suc22 detection) and Alexa Fluor® 594 conjugated secondary antibodies. The fluorescent signal of Suc22 was predominantly detected in the nucleus of wild type cells as expected (nuclear signal was detected in 92.5% of the cells) (Figure 1.14A). In ddb1-depleted cells, Suc22 also accumulated in the nucleus (92.5% nuclear signal). Ddb1 is a component of CRL4Cdt2 ubiquitin ligase required for Spd1 degradation. Concomitant deletion of spd1 resulted in complete loss of Suc22 nuclear localization (100% cytoplasmic signal). The fluorescent signals of hR2, hp53R2, and chimR2 were exclusively detected in the cytoplasm of the cells, indicating that Spd1 can sequester neither of them in the nucleus. The hR2, hp53R2, and chimR2 proteins were also localized in the cytoplasm of cells accumulating Spd1 (hR1-R2 ∆ddb1, hR1- p53R2 ∆ddb1, and hR1-chimR2 ∆ddb1, data not shown). This result indicates that Spd1 cannot interact directly with the hR2, hp53R2 and chimR2 subunits. Alternatively, the interactions are too weak to allow nuclear sequestration of the proteins. Interestingly, our previous findings showed that Spd1 could restrain the activity of the hR1-R2 complex to regulate the dNTP pool (Figure 1.10). If this holds true, the restrain does not correlate with nuclear sequestration of the hR2 subunit, and thus likely occurs through binding to the hR1-R2 holoenzyme. This result is consistent with the literature (Nestoras et al., 2010), showing that the major regulatory role of Spd1 on RNR activity is not directly dependent on subcellular localization of Suc22. To further investigate the ability of Spd1 to restrain the hR1-R2 complex, I performed two assays, each depending on robust phenotypes for Spd1-dependent restraint of RNR activity (Nestoras et al., 2010). The first is the ability of ∆spd1 mutants to suppress the synthetic inviability associated with concomitant loss of the canonical Spd1 degradation pathway and the rad3 checkpoint gene (Liu et al., 2003). The second is the ability of ∆spd1 mutants to suppress the meiotic defect evident in ∆ddb1 mutants (Holmberg et al., 2005; Liu et al., 2003).

43 The first assay showed, consistent with the literature (Holmberg et al., 2005), that deletion of ddb1 in checkpoint deficient cells resulted in synthetic lethality, which was partially suppressed by concomitant deletion of spd1 (Figure 1.14B). Similar to these observations, deletion of ddb1 in checkpoint-deficient hRNR cells resulted in decreased viability, albeit less pronounced. However, whereas deletion of spd1 could partially rescue the phenotype of hR1-R2 rad3-ts ∆ddb1 cells grown at the restrictive temperature, no rescue was observed when deleting spd1 in hR1-p53R2 rad3-ts ∆ddb1 cells, indicating that Spd1 might exert its function only on the former. This is consistent with our previous findings. The effect of ddb1 deletion in the hR1-p53R2 background, thus reflect other functions of Ddb1, independently of its function in degrading Spd1. This is expected, as Ddb1 is also involved in nucleotide excision repair (NER). It should be noted that the hRNR rad3-ts strains grew like wild type at the restrictive temperature in this spot assay, in contrast to the observations in Figure 1.12A. I expect that the incubation temperature might have been lower than intended, which has led to some sustain of Rad3 activity, but not enough to deal with the effects associated with ddb1 loss. In the hR1-chimR2 background, deletion of ddb1 did not result in checkpoint dependency. Consequently, no effect was seen upon concomitant deletion of spd1. This observation is probably due to the higher chimR2 protein level or higher dNTP pool found in the hR1-chimR2 strain. In the second assay, meiosis was induced in self-fertile h90 strains by nitrogen starvation. After formation of zygotic asci, the percentage of asci with either zero, one, two, three or four spores was scored. The data is presented in Figure 1.14C as a semiquantitative measure of the meiotic competence. Consistent with the literature (Holmberg et al., 2005), meiosis in wild type cells resulted in four-spored asci (∼90% of the asci contained four spores). The ∆ddb1 mutant was completely unable to undergo meiosis to form ascospores due to accumulating Spd1 (∼96% empty zygotes). Concomitant deletion of spd1 suppressed this defect to almost wild type levels (∼76% zygotes with four spores).

44 Interestingly, hR1-R2 cells were not able to undergo proper meiosis. Besides empty asci (∼44%), asci with one (∼30%) or two (∼26%) spores were formed. The meiotic defect was neither exacerbated nor alleviated upon deletion of ddb1 or ddb1 and spd1. A meiotic defect, albeit less severe, was also observed in hR1-p53R2 cells. These cells could partially undergo meiosis and generate four-spored asci (∼35%). However, asci containing less than four spores were dominating. Similar to the hR1- R2 cells, no effect on meiosis was observed upon deletion of ddb1 or ddb1 and spd1. hR1-chimR2 cells were able to undergo meiosis and they generated four-spored asci near wild type levels (∼89%). The meiotic competence of these cells was slightly altered by deletion of ddb1, however, this result was not consistent between experiments. Taking the two assays together, I conclude that Spd1 has an in vivo inhibitory function in cells with the hR1-R2 complex. Based on findings presented above, this function is to regulate the dNTP pools. In contrast, Spd1 is not able to restrain the activity of the hR1-p53R2 complex. Both hRNR strains displayed meiotic defects, independently of Spd1 function(s). The 2-fold increase in protein level of hp53R2 compared with hR2 (Figure 1.13C) might account for the less severe meiotic defect observed in hR1-p53R2 cells. Consistent with this proposition, the hR1-chimR2 strain was fully able to undergo proper meiosis. Thus the broader trends in the data make clear that the level of the small RNR subunit influences the ability of cells to undergo proper meiosis. Similar trends were observed in relation to checkpoint activation and dependency. Here it was shown that suppression of checkpoint dependency by elevated hR2 levels was independent of dNTP pools. Analogous to this observation, elevated levels of hR2 rescued the meiotic defect observed in hR1-R2 cells, whereas the presence of the hR1D57N allele conferring high dNTP pools did not (Figure 1.13C, last two columns).

45 ……………………………………………………………………………………………………………… Figure 1.14. Spd1-dependent restraint of RNR activity (A) Subcellular localization of the small RNR subunits. Asynchronous, exponential growing cells of indicated genotypes were fixed in cold methanol and stained with anti-R2/p53R2 primary antibodies, Alexa Fluor® 594 conjugated secondary antibodies, and DAPI. Cellular localization was visualized by fluorescence microscopy. Nuclear and cytoplasmic distributions of the small subunits are depicted for each strain on the right (n = 40). (B) Viability assay. 10-fold serial dilutions of indicated strains were spotted onto YES medium and incubated for 3-4 days at the permissive (25°C) and the restrictive (35°C) temperature. (C) Meiotic competence. Meiosis was induced in h90 strains by nitrogen starvation and the number of spores per asci was counted on day 3. Spores in ≥ 200 asci were counted for each strain. ………………………………………………………………………………………………………………

46 Spd1 is destabilized in cells with human RNR The finding that Spd1 has an inhibitory function in hR1-R2 cells but not in hR1- p53R2 cells was peculiar. The two recombinant fission yeast strains both depend on the large subunit of human RNR (hR1) and therefore only differ in the proteins constituting the small subunit. Moreover, these two proteins share 81% amino acid sequence identity. The simplest explanation to why we only see an effect of Spd1 in hR1-R2 cells would be alterations in the Spd1 levels between the two strains. To examine this possibility, Western blotting was performed on cell extracts from untreated cells and cells treated with 200 mM HU. HU is known to trigger CRL4Cdt2-dependent degradation of Spd1. A band representing Spd1 (∼14 kDa) was observed in wild type cells, which disappeared almost completely in the presence of HU (Figure 1.15A, lane 1-2). As expected, Spd1 accumulated in the ∆ddb1 background (lane 3-4), regardless of treatment. In untreated ∆ddb1 cells, the level of Spd1 was increased ∼4 fold compared to untreated wild type cells (Figure 1.15B). Interestingly, the Spd1 level was lower in the hR1-R2 cells (Figure 1.15A, lane 5) and negligible in the R1-p53R2 cells (lane 9). Even when these cells were defective in the CRL4Cdt2-dependent Spd1 degradation pathway (lane 7-8 and lane 11-12, respectively), the protein levels were low compared to that of ∆ddb1 cells (lane 3-4). In the hR1-chimR2 background (lane 13), the Spd1 level resembled that of wild type, however, in hR1-chimR2 ∆ddb1 cells (lane 15-16), Spd1 accumulated to the same level as in hRNR ∆ddb1 cells, which is less than half of the level observed in ∆ddb1 cells. The lower levels of Spd1 in cells with human RNR were observed in three independent experiments (quantified in Figure 1.15B). Taken together, the result indicates that the CRL4Cdt2 pathway is intact and mediates Spd1 degradation in response to HU. The result also indicates that Spd1 is somehow destabilized in cells with human RNR independently of the CRL4Cdt2 degradation pathway. This is possibly due to diminished interaction with a binding partner.

47 If the interaction between Spd1 and the hR1-R2 complex is weak, more Spd1 proteins will be unbound and prone to degradation via Ddb1-independent pathways. In the hR1-chimR2 background, the elevated level of chimR2 protein (2.8 times that of hR2) will provide more interaction sites for Spd1, and thus more Spd1 proteins are stabilized. In contrast, but consistent with the above findings, a higher level of p53R2 (2 times that of hR2) does not lead to stabilization of Spd1, as the interaction between Spd1 and the hR1-p53R2 complex is negligible (or non-existing).

…………………………………………………………………………………………………….……… Figure 1.15. Spd1 is destabilized in cells with human RNR (A) Spd1 degradation profiles. Cells were grown to exponential phase. Aliquots were treated with 200 mM HU or left untreated. Untreated cells were harvested after 2 hours, whereas treated cells were harvested after 3 hours. Western blotting was performed using anti-Spd1 antibodies. The Western blot is a representative of three independent experiments with similar results. (B) Relative Spd1 levels. Spd1 levels were quantified with respect to the internal control (Ponceau S), and normalized to wt. Error bars, s.d. (n=3). ………………………………………………………………………………………………………….… 48 Discussion

In the present study, recombinant fission yeast strains that rely on human RNR were created and examined. The strains were created by replacing the genes encoding yeast RNR with those encoding human RNR, which resulted in two strains differing in their subunit composition. The results showed a requirement for both RNR subunits of human origin to generate a functional enzyme (Supplementary Figure S1), even though RNR is highly conserved from yeast to human. Evidence was provided that the human enzyme could complement for lack of the yeast enzyme in terms of cellular growth (Figure 1.9), and the production of dNTPs (Figure 1.10), which is the major role of RNR in all organisms. Based on the dNTP pool measurements presented in Figure 1.10, the activity of the hR1-p53R2 complex might be lower than that of the hR1-R2 complex, but sufficient to support cellular growth. In contrast, the fission yeast strains with human RNR failed to undergo proper meiosis (Figure 1.14C) and they were dependent on the replication checkpoint for survival (Figure 1.12). In fact, Cds1, the major mediator of the replication checkpoint was shown to be constitutively activated in these cells (Figure 1.12D). As Cds1 activation is restricted to S phase (Lindsay et al., 1998), this result might indicate that the hRNR cells suffer from S phase problems. In general, a higher induction of Cds1 kinase activity was observed in hR1-p53R2 cells compared with hR1-R2 cells, indicating more severe S phase problems. This is possibly due to the slightly lower activity of the hR1-p53R2 enzyme. Even though constitutively activated, the Cds1 kinase was shown to be intact, as it served to arrest hRNR cells in response to HU, which depletes the dNTP pool (Singh and Xu, 2016) (Figure 1.12E). Furthermore, as untreated hRNR cells are not cell cycle arrested, this result also indicates that constitutive activation of Cds1 is not because of insufficient dNTP pools, which is consistent with the results presented above.

49 Elevated levels of R2 homologs suppress meiotic defects and checkpoint dependency independently of deoxynucleotide pools The hR1-chimR2 strain was created to reconcile the characteristics of wild type cells and cells with human RNR. As the major difference between the respective small subunits is located to the N-termini, we proposed that the N-terminus of Suc22 could be engaged in a yet unidentified function, which is not complemented by the subunits of human RNR. The lack of complementation could perchance account for the constitutive checkpoint activation. Indeed, the presence of amino acid 1-69 of Suc22 fused to hR2 (amino acid 69-389) in the hR1-chimR2 strain resulted in suppression of the checkpoint dependency (Figure 1.13A-B). Furthermore it suppressed the meiotic defects observed in cells with human RNR (Figure 1.14C). However, it was demonstrated that the chimR2 subunit was somehow stabilized resulting in higher protein (Figure 1.13C) and dNTP levels (Figure 1.10E), which in turn could be the cause of the suppression. Consistently, I showed that excess hR2 also suppressed checkpoint activation (Figure 1.13B-C) and the meiotic defects (Figure 1.14C) in the hR1-R2 background. To address whether elevated dNTP pools as a consequence of excess hR2 protein were the actual cause of suppression, I included the hR1D57N-R2 mutant in the analysis. High dNTP pools alone did not reverse the checkpoint dependency (Figure 1.13D) and meiotic defects (Figure 1.14C). This supports the notion that cells with human RNR suffer replication problems that are independent of dNTP pools. In this case, excess hR2 might contribute to a function that is independent of the catalytic role of the RNR complex. It is interesting that replacement of the N terminus of human R2 with that of Suc22 leads to stabilization of the protein. The reason for this was not elucidated but one explanation is that the replacement removes the KEN-box normally found in position 30-33 of human R2. In mammalian cells, the KEN-box is recognised by the APCCdh1 (APCSte9 in fission yeast), which targets R2 for degradation by polyubiquitination (Chabes et al., 2003b; Kitamura et al., 1998; Trickey et al., 2008). This explanation is based on the assumption that the hR2 protein is degraded in a KEN-box dependent manner by APCSte9 in fission yeast. Interestingly, neither the p53R2 protein nor the Suc22 protein contains a KEN-box. However, as the level of p53R2 protein does not equal the level of chimR2, the absence of a KEN-box cannot alone account for the stabilization of the chimR2 protein.

50 Spd1 can bind and restrain the hR1-R2 complex to regulate dNTP levels Based on deoxynucletide measurements performed in this study as well previously, Spd1 was found to restrain the hR1-R2 complex to regulate dNTP levels (Figure 1.10). In contrast, Spd1 was unable to restrain hR1-p53R2 activity. Consistent with these findings, overexpression of spd1 resulted in a small fraction of slowly- replicating hR1-R2 cells, whereas no effect was observed in the hR1-p53R2 background. In further support of Spd1 having an inhibitory function on the hR1-R2 complex, deletion of spd1 partially rescued the synthetic sickness associated with loss of ddb1 in checkpoint-deficient hR1-R2 cells (Figure 1.14B). In contrast, no rescue was observed in checkpoint-defective hR1-R2 ∆ddb1 cells. The apparent inhibitory function of Spd1 in hR1-R2 cells is independent of its ability to sequester the hR2 subunit in the nucleus (Figure 1.14A). This is consistent with the findings that the major regulatory role of Spd1 on RNR activity in wild type cells is not directly dependent on subcellular localization of Suc22 (Nestoras et al., 2010). It also indicates that Spd1 does not interact directly with the hR2 subunit. Alternatively, Spd1 is able to interact with the hR2 subunit but the interaction is too weak to support nuclear sequestration. Consistent with a weak interaction between Spd1 and hR2, the Spd1 level was found to be low in hR1-R2 cells (Figure 1.15). Moreover, in hR1-R2 cells defective in the canonical Spd1 degradation pathway, Spd1 accumulated to less than half of the level observed in the ∆ddb1 mutant. This argues that Spd1 is somehow destabilized independent of the CRL4Cdt2 pathway. One explanation for this destabilization could be diminished interaction with a binding partner. Spd1 belongs to the family of IDPs, which lack a well-structured three-dimensional fold but adopts diverse structures on different targets. If the interaction between Spd1 and human RNR is weak, the unstructured nature of soluble Spd1 might lead to its degradation. In this way, Spd1 is destabilized outside of S phase in hRNR cells, where it is normally stable in wild type cells due to interaction with both Cdc22 (in the cytoplasm) and Suc22 (in the nucleus).

51 This is consistent with literature, stating that IDPs are inherently sensitive to proteolysis due to their lack of fold, and that mechanisms by which they prevent degradation by default, include folding upon interaction with a partner (Suskiewicz et al., 2011). Degradation of Spd1 by broad-specific proteases or via ubiquitin- independent degradation by the CSN, can both explain the modest accumulation of Spd1 in hR1-R2 ∆ddb1 cells. This hypothesis can also be applied to the hR1-p53R2 and hR1-chimR2 strains. The level of Spd1 in hR1-p53R2 cells resembled that of wild type cells treated with HU, which leads to CRL4Cdt2-mediated degradation of Spd1. This indicates that Spd1 does not interact with the hR1-p53R2 complex, which is consistent with the results obtained in this study. In the hR1-chimR2 strain, the Spd1 level was comparable to that of wild type. However, the level of Spd1 in CRL4Cdt2-defective hR1-chimR2 cells did not exceed those observed in CRL4Cdt2 defective hRNR cells. Elevated levels of chimR2 in the hR1-chimR2 strain likely explain this observation. Despite weak interactions, elevated chimR2 levels will counterbalance the level of soluble Spd1 susceptible for proteolysis. It should be noted that this theory is build on the assumption that the Spd1 levels do not fluctuate between strains due to altered transcription.

52 The in vivo inhibitory role(s) of Spd1 In general, there was a good correlation between the characteristics of cells reliant on human RNR, and wild type cells accumulation Spd1. Fission yeast cells defective in CRL4Cdt2 mediated protein ubiquitylation are challenged at S phase because Spd1 is not degraded. The effects of Spd1 accumulation include severe S phase stress, Rad3 checkpoint activation and dependency, and meiotic defects (Holmberg and Nielsen, 2012; Holmberg et al., 2005). The checkpoint dependency and meiotic defect can be suppressed by spd1 loss or alternatively by Suc22 overexpression. Interestingly, accumulating evidence suggest that this suppression is independent of dNTP levels. By using a Cdc22D57N mutant, Fleck et al., 2013 showed that high dNTP pools alone could suppress neither checkpoint activation nor dependency associated with Spd1 accumulation. Recently, we confirmed this by establishing a salvage pathway to allow uptake and phosphorylation of exogenous applied deoxynucleosides. Our findings showed that salvage did not rescue Spd1- mediated checkpoint activation and dependency (Fleck et al., 2017). Finally, the latest discovery is that ectopic overexpression of a catalytic dead version of Suc22 (Suc22Y173F) can rescue the checkpoint dependency of Spd1 accumulating cells (Olaf Nielsen, unpublished data). Collectively, these findings have led to the suggestion that Spd1 interferes with other unknown processes in addition to restraining RNR activity to regulate dNTP pools. The other function(s) of Spd1, plausibly assisted by Spd2, results in checkpoint activation and dependency (Fleck et al., 2013, 2017; Vejrup- hansen et al., 2014). If the mechanism by which Spd1 elicit a checkpoint dependency is independent of its ability to restrain RNR, we would expect to see an effect of ectopic overexpression of spd1 in cells with human RNR. That is, even though the interaction between Spd1 and human RNR is weak (or non-existing), Spd1 would still be able to exert its function on the unknown target. However this was not the case (Figure 1.11). Only a small effect was observed in hR1-R2 cells, which corresponded to the proposed ability of Spd1 to restrain enzyme activity to regulate the dNTP pools. It should be noted that this experiment wad performed once and should be performed again in combination with a Western blot to prove that Spd1 is indeed overexpressed to similar levels in the different strains.

53 A possible explanation for this deviation from the suggested model is that the additional function(s) of Spd1 requires interaction with RNR. Alternatively, the unknown target of Spd1 is a yet undescribed function of RNR.

I propose the following model: The catalytic function of RNR essential for dNTP biosynthesis is evolutionary conserved. Therefore, human RNR can complement for lack of the yeast enzyme regarding dNTP production. In addition to the catalytic function, RNR has another function which is less conserved through evolution. Spd1 can restrain both of these functions in wild type fission yeast, however, whether Spd1 can restrain the functions of RNR from different species is determined by the degree of conservation between the enzymes. In the following, this model has been put into context of some of the findings from this study and previous studies in fission yeast. The model is presented in Figure 1.16.

Wild type: The catalytic function of RNR supplies the cell with the appropriate amount of dNTPs to be used for DNA synthesis. This function is restrained by Spd1 outside of S phase when the requirement for dNTPs is low. Furthermore, binding of Spd1 to Cdc22 (in the cytoplasm) and Suc22 (in the nucleus) protects Spd1 from proteolysis by default. In addition to its catalytic role, RNR executes another function (function X), which can also be restrained by Spd1. Accumulation of Spd1 leads to a decrease in the dNTP pool due to inhibition of the catalytic activity of RNR, which in turn leads to a S phase delay (and cell elongation), checkpoint activation and DNA damage sensitivity. Accumulation of Spd1 also leads to inhibition of function X and this too leads to checkpoint activation through an unknown mechanism. Deletion of spd1 or overexpression of Suc22 suppresses the defects associated with Spd1 accumulation. In contrast, elevating the dNTP pools (by using the Cdc22D57N mutant or by salvage) only counteracts the catalytic restriction of RNR. Thus, function X will still be restrained, which is the major cause of the checkpoint activation and dependency in Spd1 accumulating cells.

54 hR1-R2: The catalytic function of hR1-R2 is evolutionary conserved and therefore hR1- R2 can complement for lack of Cdc22-Suc22 in providing fission yeast cells with an appropriate amount of dNTPs. The interaction between Spd1 and hR1-R2 is semi- conserved and weak. Consequently, a small increase in the dNTP levels is evident upon deletion of spd1. The ability to execute function X is less complemented by hR1-R2, which leads to checkpoint activation independently of Spd1. Because of this, overexpression of hR2, but not deletion of spd1, can suppress the checkpoint activation and dependency. As in wild type cells, elevating the dNTP pool (by using the hR1D57N mutant) only counteracts the catalytic restriction of RNR. The weak interaction between Spd1 and hR1-R2 protects some Spd1 proteins from proteolysis. However, as Spd1 fails to bind and sequester hR2 to the nucleus, more soluble Spd1 exists in hR1-R2 cells, which are susceptible for proteolysis via CRL4Cdt2-independent pathways. Collectively, this leads to the occurrence of a moderate level of Spd1 in hR1-R2 cells. hR1-p53R2: The catalytic function of hR1-p53R2 is evolutionary conserved and therefore hR1-p53R2 can complement for lack of Cdc22-Suc22 in providing fission yeast cells with an appropriate amount of dNTPs. However, the interaction between Spd1 and hR1-p53R2 is less conserved than the interaction between Spd1 and hR1-R2 (or non- existing). Consequently, no increase in the dNTP levels is evident upon deletion of spd1. The ability to execute function X is less complemented by hR1-p53R2, which leads to checkpoint activation independently of Spd1. Because of this, overexpression of hp53R2, but not deletion of spd1, can suppress the checkpoint activation and dependency. The negligible interaction between Spd1 and hR1-p53R2 (if any), together with the inability of Spd1 to sequester hp53R2 in the nucleus, leads to unprotected Spd1 proteins, which are subjected to proteolysis via CRL4Cdt2- independent pathways. Therefore, the level of Spd1 is negligible in cells with hR1- p53R2.

55 hR1-chimR2: The catalytic function of hR1-chimR2 is preserved and sufficient to provide fission yeast cells with dNTPs. Because the chimR2 protein is somehow stabilized, these cells have high dNTP pools. The interaction between Spd1 and hR1-chimR2 is weak. The ability of hR1-chimR2 to execute function X might be poor. Because chimR2 is present in excess, hR1-chimR2 cells phenocopy cells overexpressing Suc22 (or hR2/p53R2), which leads to suppression of the checkpoint activation and dependency. Furthermore, stabilization of chimR2 provides more binding sites for Spd1 (even though the interaction is weak), which in turn leads to protection of Spd1 from proteolysis via CRL4Cdt2-independent pathways. Therefore, the level of Spd1 is higher in hR1-chimR2 cells compared to the levels observed in hR1-R2 and hR1- p53R2 cells.

How the meiotic data can be related to this model, remains to be determined. Especially the ability of hR1-p53R2 cells to partially undergo proper meiosis, while hR1-R2 cells fails to do so, is peculiar. As higher levels of the R2 homologs were shown to rescue meiotic defects, the higher level of hp53R2 compared with hR2 might explain the discrepancy (the level of hp53R2 was ∼2-fold higher than hR2, Figure 1.13C). In contrast, the higher level of p53R2 contradicts with the more pronounced induction of Cds1 kinase activity in hR1-p53R2 cells compared with hR1- R2 cells (the level of 32P-MBP was 1.5-fold higher in hR1-p53R2 cells compared with hR1-R2 cells, Figure 1.12D). The slightly lower dATP pool observed in S phase- synchronized hR1-p53R2 cells might reconcile these findings as it would possibly reinforce the checkpoint activation. That is, the hR1-p53R2 complex is less efficient in the production of dNTPs than the hR1-R2 complex. This leads to a small reduction in the level of dNTPs, which in turn activates the checkpoint through a dNTP- dependent pathway. However, the dNTP pool reduction is not severe enough to execute a cell cycle arrest, and thus the hR1-p53R2 cells grow like wild type. Consistent with this suggestion, the specific activity of the human and mouse R1- p53R2 complex has been shown in vitro to be less than that of the human and mouse R1-R2 complex (Guittet et al., 2001).

56 ………………………………………………………………………………………………………………… Figure 1.16. Model of RNR function(s) (A) RNR has (at least) two functions in the cell (yellow arrows; the thickness of the arrows indicates strength of activity). One functions is de novo biosynthesis of dNTPs, a function that is highly conserved through evolution. RNR also functions through a yet undescribed pathway (function X), which is independent of dNTP biosynthesis and evolutionary less conserved. Failure to perform either function leads to checkpoint activation and dependency. (B) In the wt background, accumulating Spd1 inhibits both functions of RNR, which leads to checkpoint activation. (C) Suppression of checkpoint activation can be accomplished by spd1 deletion or overexpression of Suc22, which restores both functions of RNR. (D) Suppression cannot be accomplished by elevated dNTP pools as this will only counteract the dNTP- dependent checkpoint activation, but not the dNTP-independent checkpoint activation associated with function X.

Figure 1.16. Cont.

57 (E) In hRNR cells, function X is less conserved, which leads to constitutive checkpoint activation. Furthermore, the catalytic activity of the hR1-p53R2 holoenzyme is less efficient compared with hR1- R2, which leads to dNTP-dependent checkpoint activation (functions specific for hR1-p53R2 is indicated in blue). (F) Spd1 interaction with the hR1-R2 complex is weak and accumulation of Spd1 leads to a small reduction in the dNTP pool. Spd1 does not interact with the hR1-p53R2 complex and, thus, no reduction in the dNTP pool is observed upon Spd1 accumulation. (G) Deletion of spd1 does not suppress checkpoint activation in hRNR cells. However, overexpression of hR2/hp53R2 boosts both functions of RNR and suppresses checkpoint activation. (H) In contrast, suppression cannot be accomplished by elevated dNTP pools as this will only counteract the dNTP-dependent checkpoint activation, but not the dNTP-independent checkpoint activation associated with lack of function X. …………………………………………………………………………………………………………… …………………………

In the model presented in Figure 1.16, an additional function of the RNR complex has been suggested. This study has focused on RNR enzymes that differ in the proteins that constitute the small subunit of RNR. It is possible that the results observed in the different strains reflect a nonenzymatic role of the RNR subunit proteins, rather than of the RNR complex. Additional roles of the R2 homologs of RNR have been suggested based on the different lengths of their amino acid sequences. In mice, a N-terminal truncation of the R2 protein did not significantly alter the enzymatic activity of RNR (Chabes et al., 2003b; Kauppi et al., 1996). Furthermore, p53R2 can form an active RNR complex with R1 to support dNTP production, despite lacking 33 amino acid residues in the N terminus compared with hR2. It is thus considered likely that RNR subunit-specific functions exist. The presence of a KEN-box in the N terminus of mammalian R2, important for its cell cycle-regulated degradation, constitutes an example of functional domains within a protein subunit that are unrelated to the catalytic function of RNR. Finally, the predicted unstructured nature of the N terminus of R2 homologs supports the idea of various functions, possibly through multiple binding partners (Brignole et al., 2018; Eklund et al., 2001; Kriwacki et al., 2002; Smith et al., 2009; Sugase et al., 2007). Exploring the possible nonenzymatic functions of RNR proteins, as well as enzymatic functions of RNR that differ between species, may lead to new rationales for developing novel RNR inhibitors, which can be used as both anti-cancer and anti- microbial agents. The role of RNR in human diseases and the identification of novel and species-specific inhibitors will be the topic of chapter 2.

58

Chapter 2

Co-culturing System to Screen for Species- specific Inhibitors of Ribonucleotide Reductase

Chapter objectives

To establish a co-culturing screening system that may be used to screen for species-specific inhibitors of ribonucleotide reductase using fission yeast as model organism

59

60 Introduction

The essential role of RNR in DNA synthesis and cellular growth in all organisms, has resulted in an enormous interest in developing RNR inhibitors to be used as anti-cancer, anti-bacterial and anti-viral drugs (Eklund et al., 2001). RNR has been an anti-cancer target for decades, and the interest in RNR increased further with the identification of p53R2 as a p53-regulated subunit of RNR ((Nakano et al., 2000; Tanaka et al., 2000). p53 is a transcription factor with tumor suppressor activity and most human cancers are associated with mutations in p53 or the pathways that directly regulate it (Eklund et al., 2001; Hollstein et al., 1991; Vogelstein, 1990). The genomic instability often seen in tumors lacking p53 may reflect dysfunction of RNR due to failure of inducing p53R2 (Tanaka et al., 2000). Inhibiting RNR activity reduces the intracellular dNTP pools, thereby inhibiting DNA synthesis. As cancer cells are known to be highly proliferative, they are more sensitive to shortage of the DNA precursors than normal cells (Aye et al., 2014; Shao et al., 2013). RNR inhibitors currently in use, or under development, for cancer therapy can be divided into two categories: suppression of and enzyme inactivation. The former involves siRNA-mediated knockdown of genes encoding the subunits of RNR, whereas the latter includes inhibitors targeting the individual RNR subunits, as well as inhibitors prohibiting the formation of an active holoenzyme. Gemcitabine (a substrate analog), Clofarabine (an allosteric effector analog), Cisplatin (a sulfhydryl group inactivator), Triapine (an iron chelator) and HU (a radical scavenger) all belong to the latter (Shao et al., 2013). HU was the first RNR inhibitor to be extensively studied and has been used to treat various human diseases including cancer (Madaan et al., 2012). However, due to its relatively short half-life and its low affinity for human R2, its clinical effectiveness is limited (Wijerathna et al., 2011). In addition to low affinity for RNR, other shortcomings of RNR inhibitors currently in use include resistance development and non-specific mechanisms of action, which can lead to side effects. Especially inhibitors targeting the iron-tyrosyl radical center, including HU, are challenged by the regeneration of the free radical. However, besides acting as a single agent, RNR inhibitors can be employed in combination with other anticancer drugs to enhance their cytotoxic effects by blocking DNA damage repair (Shao et al., 2013). Yet, more specific and efficient RNR inhibitors for clinical use have to be identified and developed.

61 Searching for subunit-specific inhibitors would be an attractive approach, as human R2 and p53R2 have been shown to play different roles in the cell (Bourdon et al., 2007; Guittet et al., 2001; Pontarin et al., 2007, 2012). Furthermore, cancer cells lacking p53 depend on R2 for the production of dNTPs required for DNA replication and repair (Greenblatt et al., 1994; Tanaka et al., 2000). Thus, R2-specific inhibitors in combination with genotoxic agents may be an attractive strategy to produce preferential cytotoxicity to p53-deficient cancer cells with relatively limited impact on normal cells. In cancer cells retaining the p53 gene, a different strategy is to target the p53R2 subunit. It has been shown that suppression of p53R2 reduces cell survival after genotoxic stress (Yanamoto et al., 2005). Inactivating p53R2 has also been suggested to induce apoptosis in a p53-dependent manner to eliminate damaged cells, due to failure in p53R2-dependent DNA repair (Yamaguchi et al., 2001). The biggest challenge to overcome in these approaches will undoubtedly be indiscriminately inhibition of both R2 and p53R2 due to high conservation in their sequence and three- dimensional structure (Guittet et al., 2001). Therefore, identification of possible non- enzymatic roles of the RNR subunits and their regulatory networks, which has been the focus of chapter 1, may prove valuable in the search for new drugs targeting the RNR-related pathways. In this chapter, I present an in vivo screening system that may be used to screen for novel RNR inhibitors. The screening system relies on co-culturing of fission yeast strains that express RNR from different species. To exemplify the applicability of the system, I have performed preliminary experiments using the species-specific RNR inhibitor HU. The increased susceptibility of human RNR to the radical scavenger was readily observed. The applicability of the screening system is wide and may in principle be used to screen for inhibitors of RNR from any species with the prerequisite that the enzyme can be stably expressed in fission yeast. That is, the screening system can ideally be used to identify inhibitors of RNR to be used in anti- cancer therapy as well as other human diseases.

62 Results

Experimental setup In chapter 1, it was shown that human RNR (hR1-R2 and hR1-p53R2) could be stably expressed in fission yeast. Thus, fission yeast was suitable to be exploited as selective system in searching for species-specific RNR inhibitors. The experimental setup of the screening system is depicted in Figure 2.1A. The system is based on co- culturing of two epifluorescence-tagged strains, relying on RNR from different species. The fluorescent proteins CFP and mCherry are chosen due to non- overlapping emission spectra, which should allow clear distinction of the strains using an epifluorescence microscope. Furthermore, a nuclear localization sequence (NLS) is fused to the N-terminus of the fluorescent proteins to restrict the signal to the nucleus, thereby minimizing counting errors due to cell-size variations. After the strains are co-cultured 1:1, they should be exposed to a chemical or small molecule library. Hits targeting one strain but not the other will elicit a shift in the ratio of mCherry-positive and CFP-positive cells, which can be detected by imaging. Thus, by using this system, we might be able to identify hits that show species-specific cytotoxicity, which should then be validated further. Depending on the species from which the RNR holoenzymes originate, these hits may eventually be used in chemotherapy, anti-viral therapy or as antibiotics.

63

………………………………………………………………………………………………………….….. Figure 2.1. Co-culturing screening system (A) Design of the screening system. Two fission yeast strains relying on RNR from different species and labeled with mCherry and CFP, respectively, are co-cultured and exposed to a chemical or small molecule library. Imaging and analyzing the mCherry:CFP ratio might lead to candidate inhibitors that confer differential sensitivity to be validated further. (B) Strains for co-culturing screening system. Top: Schematic of the nmt1-NLS-mCherry and nmt1-NLS-CFP targeting constructs. The constructs were integrated at the leu1 locus of leucine auxotrophic (leu1-32) strains. The nmt1 promoter allows controlled expression of the fluorescent proteins by using thiamine. Bottom: Epifluorescence micrographs of populations co-cultured in medium lacking leucine for continuous selection. The cells were grown in the absence of thiamine to allow maximal expression of the fluorophores, which can be easily distinguished due to their non-overlapping emission spectra. Strains used: wt (CFP-positive), hR1-p53R2 (mCherry-positive). ………………………………………………………………………………………………………….…..

64 Strain creation In order to create the screening system, plasmids containing mCherry and CFP cDNA with a N-terminal NLS was created using Gibson Assembly® cloning, followed by enzymatic digest and ligation into the pDUAL (Matsuyama et al., 2004a) destination vector pON1204, as described in Materials and Methods. In this way, the fluorescent markers were placed under control of the nmt1 promoter, from which they can be strongly expressed in the absence of thiamine. To allow targeted integration at the chromosomal leu1 locus of leucine auxotrophic (leu1-32) strains, the plasmids were digested with NotI prior to transformation (Figure 2.1B, top). Positive clones were selected based on leucine prototrophy, and proper integration was confirmed by PCR. The following strains were established: wild type (mCherry-positive), wild type (CFP-positive), hR1-R2 (CFP-positive), and hR1-p53R2 (mCherry-positive). Their respective doubling times were comparable (data not shown). As expected, do to the compatibility of the fluorescent proteins, two populations with distinct fluorescent markers were clearly distinguishable when co-cultured and examined using an epifluorescence microscope (Figure 2.1B, bottom). A negligible fraction of the cells lost their fluorescent tag due to continuous selection for leucine prototrophy. Growing the cells in medium lacking thiamine resulted in maximal expression of the fluorescent proteins.

65 Proof of principle Once the strains were established, the potential of the system in identifying selective RNR inhibitors was tested based on sensitivity to HU. HU constitutes an example of a species-specific inhibitor. As evident from Figure 2.2, wild type cells are not sensitive to low concentrations of HU, whereas hRNR cells are hypersensitive to this radical scavenger.

…………………………………………………………………………………………………………..... Figure 2.2. Fission yeast cells relying on human RNR are hypersensitive to HU 10-fold serial dilutions of indicated strains were spotted onto YES medium without or with increasing concentrations of HU and incubated for 3 days. Human RNR confers extreme sensitivity to low concentrations of HU, whereas wild type cells displayed normal sensitivity. ………………………………………………………………………………………………………….....

To test if treatment with HU could elicit a shift in the CFP:mCherry ratio, indicative of differential sensitivity to HU, wild type (CFP-positive) and hR1-p53R2 (mCherry-positive) strains were co-cultured. Aliquots were exposed to 0.4 mM HU or left untreated. Samples were taken every 2 hours, for which integrated image analysis and data presentation were obtained using the Xcyto® Quantitative Cell Imager. As depicted in the intensity dot plots in Figure 2.3A (Supplementary Figure S5), live CFP-positive cells and live mCherry-positive cells were gated and counted. Before addition of HU (t=0), the cells were gated approximately 50:50. After 8 hours of treatment, almost all living cells were from the wild type population (CFP-positive). Quantification of the data showed a fairly constant CFP:mCherry ratio throughout the experiment in the control populations (Figure 2.3B). However, in the HU treated populations, the ratio shifted as more hR1-p53R2 cells died or stopped dividing. A substantial change in the ratio between treated and untreated populations was seen after 6 hours.

66 Similar results were obtained when co-culturing wild type and hR1-R2 strains with reversed tags (wt: mCherry-positive) (hR1-R2: CFP-positive) (Figure 2.3C). In this experiment, a substantial change in the ratio between treated and untreated populations was observed after 8 hours of treatment. In order to evaluate the quality of the system, the statistical parameter Zʹ (Z prime) factor was calculated. This parameter is used to evaluate high throughput screening assays, where a Zʹ value ≥ 0.5 indicates a robust assay (Running et al., 1999). The calculation was based on 20 co-cultures; 10 controls and 10 treated with 0.4 mM HU. The co-cultures contained wild type (CFP-positive) and hR1-p53R2 (mCherry-positive) cells. Samples were taken before and 8 hours after addition of HU. A Zʹ score of 0.73 was obtained, indicative of a very robust assay. This preliminary experiment clearly demonstrates the species-specific sensitivity to HU and I evaluate the system to have great potential in identifying species-specific inhibitors of RNR.

67

……………………………………………………………………………………………………………… Figure 2.3. Proof of principle using the species-specific RNR inhibitor HU (A) Intensity dot plots. Wild type (CFP-positive) and hR1-p53R2 (mCherry-positive) populations were co-cultured and aliquots were either treated with 0.4 mM HU or left untreated. Samples were taken every 2 hours and analyzed using the Xcyto® Quantitative Cell Imager (ChemoMetec). The fluorescence intensity of individual cells in the mCherry channel (x-axis) and the CFP channel (y-axis) is depicted at different time points. Each dot plot is based on ∼1000 live cell counts. (B) Quantification of data obtained in (A). Alterations in the CFP:mCherry ratios between treated and untreated populations are indicative of differential sensitivity to HU. A substantial change in the ratios is observed after 6 hours. (C) Quantification of data obtained as described in (A), this time using wild type (mCherry-positive) and hR1-R2 (CFP-positive) strains. A substantial change in the ratios is observed after 8 hours. ………………………………………………………………………………………………………………

68 Discussion and future perspectives

RNR is a requirement for cellular proliferation and a prerequisite for life, as it is responsible for de novo biosynthesis of dNTPs in all living organisms. Because of its essential role, RNR is a well-established target for anti-cancer and anti-microbial drugs (Eklund et al., 2001). Despite the significant amount of RNR inhibitors already identified, of which some are in clinical use, a lot of effort is being devoted to develop new and more specific inhibitors. In this study, an in vivo screening system was created based on co-culturing of fission yeast strains relying on RNR from different species. Preliminary experiments were performed using fission yeast strains with either native RNR or human RNR, to test if differential sensitivity to HU could elicit a detectable response. Indeed, the cell killing effect of HU on cells relying on human RNR was clearly observed after 6-8 hours of treatment (Figure 2.3). Furthermore, the quality of the system was assessed using the statistical parameter Zʹ-factor, based on the differential sensitivity to HU in wild type and hR1-p53R2 strains. The system is very robust and gave good separation between treated samples and control samples, as judges by a Zʹ score of 0.73. Based on the preliminary results, we evaluate the system presented here to have a promising potential to identify novel inhibitors with known intracellular and species-specific target. By surveying RNRs in an intracellular environment, the screening system is expected to select for candidates with desired biological properties and exclude compounds that do not show in vivo activity. Furthermore, the candidate inhibitors will be cell permeable, and will likely be able to enter other eukaryotic cells as well. Specific advantages to this screening system are that co-culturing of the yeast cells eliminates the problem of variability across experiments, which results in improved validations. Moreover, as the only difference between the co-cultured strains is the origin of RNR, we expect candidate inhibitors to act species-specifically on RNR without secondary targets.

69 Unfortunately, due to time limitations an actual screen was never performed using this experimental setup. It would have been interesting to screen for inhibitors that would target R2 but not p53R2 of human RNR and vice versa, as such inhibitors may have different clinical values. A screen like this would, however, be complicated by the high conservation of the two subunits. Nevertheless, based on the results obtained in chapter 1, showing that Spd1 might restrain the hR1-R2 complex but not the hR1-p53R2 complex, the possibility of identifying inhibitors targeting only one of these complexes is not inconceivable. Another problem that may arise using the strains relying on the human enzyme, is that these cells are dependent on the replication checkpoint for survival, as I showed in chapter 1. Thus, compounds might be identified that are not direct inhibitors of RNR but have a cell killing effect because of the checkpoint dependency. In any case, a thorough “clean up” of the hits will be needed after the screening. Hits identified as potent inhibitors of RNR would be expected to confer disruption of the cell cycle and a decrease in the dNTP levels in the cells. Restoring cell cycle progression and the dNTP pools by an exogenous supply of deoxynucleosides (by salvage) would confirm that RNR is the cellular target of the compounds. Indeed we have shown that salvage clearly suppressed the cell-killing effect of HU in checkpoint-deficient cells (Fleck et al., 2017). Alternatively, a reference assay measuring the activity of RNR may be required to eliminate false positives. Nevertheless, identification of compounds targeting the replication checkpoint would still be relevant in a cancer-related perspective. More general limitations to the screening system could be that some compounds themselves are fluorescent (Charles Hoffman, personal communication) which could potentially disturb the signals from the cells. However, such compounds would most likely differ in signal intensities, which could then be excluded during data analysis. Cell permeability of S. pombe might constitute a problem as well. Unlike mammalian cells, S.pombe cells are enclosed in a cell wall. Consequently, it might be necessary to create mutations in the yeast strains to increase the uptake of compounds or reduce the efflux.

70 The focus of this chapter has been to screen for inhibitors that specifically will target the subunits of human RNR, which may be used in anti-cancer therapy. However, the applicability of the screening system is much wider. In principle, it may be used to screen for inhibitors to be used in the treatment of various diseases. A prerequisite is that RNR from a eukaryotic pathogen differs from that of its vertebrate host in such a way that minimizes potential cross-reactivity. A context where RNR could be an optimal target of new therapeutics is in apicomplexan diseases. The C- terminus of the human R2 subunit differs from both of the R2 subunits found in apicomplexan parasites, including the malaria-causing Plasmodium falciparum (Munro and Silva, 2012). This opens the possibility of screening for inhibitors that specifically target apicomplexan pathogens with minimal effect on host cells. The prospect of screening for RNR inhibitors that can be used as pathogen- specific antibiotics also deserves attention at a time of increasing recovery of multidrug-resistant bacteria threatening the public health. The chances of identifying species-specific inhibitors in this case are promising, given that the sequence identities between human and bacterial RNRs are generally below 50% (Tholander and Sjoberg, 2012). However, whether bacterial RNRs can support dNTP production and cellular growth in fission yeast remains to be examined.

71

72 Materials and Methods

Strains, primers and plasmids The strains used were obtained by standard genetic techniques as described previously (Forsburg and Rhind, 2006) and are listed in Table S1 together with their genotype and source. The primers used are listed in Table S2 together with their sequence (5´to 3´) and purpose of use. The plasmids used are listed in Table S3 together with their content and their purpose of use. Strains expressing human RNR were created as follows: RRM2 cDNA was integrated into the leu1 locus using a pDUAL plasmid (Matsuyama et al., 2004b). Subsequently, a linear construct containing RRM1 cDNA and a KanMX cassette flanked by 5ʹ- and 3ʹ-UTR of cdc22, was integrated into the cdc22 locus. RRM2 or RRM2B cDNA flanked by 5ʹ- and 3ʹ-UTR of suc22 was integrated into the suc22 locus, and the RRM2 gene at the leu1 locus was subsequently crossed out. Correct integration was verified by PCR. The coding sequences of human RRM1, RRM2, and RRM2B were codon optimized for S. pombe. The strain with RRM1 and RRM2 was created previously (H. B. Nielsen, 2013, unpublished). The strain with RRM1 and RRM2B was created in this study. Strains expressing NLS-CFP or NLS-mCherry were created as follows: The mCherry and CFP ORFs were amplified by PCR using plasmid pON1226 and pON1228, respectively, with the primer set KVLFP9 + KVLRP9. The NLS was amplified by PCR using the plasmid pON1234 and the primer set KVLFP8 + KVLRP8. Gibson Assembly® cloning was performed to create linear fragments containing mCherry or CFP, harboring the NLS at the N-terminal. The assembled fragments as well as the plasmid pON1204 were cut with NdeI and BglII and ligation was performed using T4 DNA ligase, resulting in the two expression vectors pON1237 and pON1238. To allow targeted integration at the chromosomal leu1 locus of leucine auxotrophic (leu1-32) yeast strains, the plasmids were cut with NotI prior to transformation. Positive clones were selected based on leucine prototrophy, and proper integration was confirmed by PCR using the primer set ONP601 + ONP602 (Matsuyama et al., 2004b).

73 Table S1. Strain list

Strain Genotype Source

Eg282 h90 Laboratory stock Eg544 h- d-mat2/3::LEU2+ Laboratory stock Eg545 h+ d-mat2/3::LEU2+ Laboratory stock Eg1072 h- ste9::ura4+ ura4-D18 leu1-32 Laboratory stock Eg1288 h+ chk1::ura4+ ura4D18 Laboratory stock Eg1324 h- leu1-32 Laboratory stock Eg1407 h90 ddb1::kmx spd1::ura4+ ura4-D18 Laboratory stock Eg1583 h- spd1::ura4+ ura4-D18 ade6-485 Laboratory stock Eg1787 h- cds1-2HA-His6::ura4+ ura4-D18 leu1-32 Laboratory stock Eg2447 h+ ddb1::nat Laboratory stock Eg2686 h+ rad3-ts Laboratory stock Eg3153 h- leu1+::nmt1-hR2 ura4-D18 cdc22-M45::hR1-kmx Laboratory stock Eg3157 h+ leu1-32 ura4-D18 H. B. Nielsen Eg3173 h- leu1+::nmt1-hR2 ura4-D18 cdc22-M45::hR1-kmx suc22::hR2 H. B. Nielsen Eg3175 h+ leu1-32 ura4-D18 cdc22-M45::hR1-kmx suc22::hR2 H. B. Nielsen Eg3219 h90 Δddb1::kmx ura4-D18 Laboratory stock Eg3220 h90 cdc22::hR1-kmx suc22::hR2 ura4-D18 Laboratory stock Eg3222 h90 cdc22::hR1-kmx suc22::hR2 ddb1::kmx ura4-D18 Laboratory stock Eg3223 h90 cdc22::hR1-kmx suc22::hR2 ddb1::kmx Δspd1::ura4+ ura4-D18 Laboratory stock Eg3224 h+ cdc22::hR1-kmx suc22::hR2 leu1-32 H. B. Nielsen Eg3236 h- cdc22::hR1-kmx suc22::hR2 leu1-32 Laboratory stock Eg3237 h- cdc22::hR1-kmx suc22::hR2 Δddb1::nat Laboratory stock Eg3238 h+ cdc22::hR1-kmx suc22::hR2 Δspd1::hygro Laboratory stock Eg3242 h- rad3-ts Δddb1::nat ura4-D18 Laboratory stock Eg3272 h+ rad3-ts Δddb1::nat Δspd1::hygro ura4-D18 Laboratory stock Eg3324 h- cdc22::hR1-kmx suc22::hR2 Δspd1::hygro Δddb1::nat rad3-ts leu1-32 Laboratory stock Eg3326 h+ cdc22::hR1-kmx suc22::hR2 Δddb1::nat rad3-ts Laboratory stock Eg3422 h+ cdc22::hR1-kmx suc22::hR2 rad3-ts leu1-32 ura4-D18 H. B. Nielsen Eg3424 h- cds1-2HA6His::ura4+ cdc22::hR1-kmx suc22::hR2 leu1-32 ura4-D18 H. B. Nielsen Eg3428 h+ Δcds1::ura4+ cdc22::hR1-kmx suc22::hR2 leu1-32 ura4-D18 H. B. Nielsen Eg3430 h+ Δcds1::ura4+ leu1-32 ura4-D18 H. B. Nielsen Eg3451 h+ Δchk1::ura4+ cdc22::hR1-kmx suc22::hR2 leu1-32 ura4-D18 Laboratory stock Eg3932 h+ Δmat2,3 cdc22-D57N ura4::adh-dmdNK-nat-adh-hENT ura4-aim Laboratory stock Eg3954 h+ ura4::adh-dmdNK-nat-adh-hENT ura4-aim Laboratory stock Eg3965 h- leu1+::nmt1-hR2 ura4-D18 cdc22-M45::hR1-kmx suc22::hp53R2 This study Eg3983 h+ leu1-32 ura4-D18 cdc22-M45::hR1-kmx suc22::hp53R2 This study Eg3992 h90 cdc22-M45::hR1-D57N-kmx suc22::hR2 Laboratory stock Eg3993 h90 cdc22-M45::hR1-D57N-kmx suc22::hR2 leu1-32 Laboratory stock Eg3995 h+ leu1-32 cdc22-M45::hR1-kmx suc22::hp53R2 This study Eg3997 h+ ura4-D18 cdc22-M45::hR1-kmx suc22::hp53R2 This study Eg3998 h90 cdc22-M45::hR1-kmx suc22::hp53R2 This study Eg4007 h- cdc22-M45::hR1-kmx suc22::hp53R2 spd1::ura4+ ura4-D18 This study Eg4008 h- cdc22-M45::hR1-kmx suc22::hp53R2 Δddb1::nat Δspd1::hygro ura4-D18 This study Eg4009 h- cdc22-M45::hR1-kmx suc22::hp53R2 cds1-2HA-His6::ura4+ ura4-D18 leu1-32 This study Eg4042 h+ cdc22-M45::hR1-kmx suc22::hp53R2 rad3-ts This study

74 Eg4043 h90 cdc22::hR1-kmx suc22::hR2 Δspd1::ura4+ ura4-D18 spd2::nat This study Eg4058 h+ cdc22::hR1-kmx suc22::hR2 leu1-32 ura4-D18 ura4+::pDUAL FFH1c- This study AHCYL1 Eg4059 h+ ura4-D18 cdc22-M45::hR1-kmx suc22::hp53R2 ura4+::pDUAL FFH1c- This study AHCYL1 Eg4060 h+ leu1-32 ura4-D18 ura4+::pDUAL FFH1c-AHCYL1 This study Eg4061 h+ leu1-32 Δsuc22::suc22(1-69)RRM2(69-389) ura4-D18 cdc22-M45::hR1-kmx A. C. Eidesgaard Eg4062 h chk1-HA ade6-704 cdc22-M45::hR1-kmx suc22::hp53R2 This study Eg4063 h+ cds1::ura4+ ura4-D18 cdc22-M45::hR1-kmx suc22::hp53R2 This study Eg4065 h+ chk1::hygro cdc22-M45::hR1 suc22::hp53R2 This study Eg4071 h cds1::ura4+ ura4-D18 leu1-32 cdc22-M45::hR1-kmx suc22::hp53R2 This study Eg4091 h90 Δsuc22::suc22(1-69)RRM2(69-389) cdc22-M45::hR1-kmx A. C. Eidesgaard Eg4092 h- chk1-HA cdc22-M45::hR1-kmx suc22::hR2 This study Eg4094 h- chk1-HA ade6-704 This study Eg4120 h+ suc22::suc22(1-69)-RRM2(69-389) cdc22-M45::RRM1-KMX ddb1::nat A. C. Eidesgaard Eg4122 h cdc22-M45::hR1-kmx Δsuc22::suc22(1-69)RRM2(69-389) cds1-2HA- This study His6::ura4+ ura4-D18 leu1-32 Eg4123 h cdc22-M45::hR1-kmx Δsuc22::suc22(1-69)RRM2(69-389) cds1::hygro ura4- This study D18 leu1-32 Eg4124 h90 cdc22-M45::hR1-kmx suc22::hp53R2 Δddb1::nat Δspd1::hygro This study Eg4125 h90 cdc22-M45::hR1-kmx suc22::hp53R2 Δddb1::nat This study Eg4126 h90 cdc22-M45::hR1-kmx suc22::hp53R2 Δspd1::hygro This study Eg4143 h- suc22::suc22(1-69)-RRM2(69-389) cdc22-M45::RRM1-KMX ura4-D18 rad3-ts A. C. Eidesgaard Eg4175 h90 suc22::suc22(1-69)-RRM2(69-389) cdc22-M45::RRM1-KMX ddb1::nat This study Eg4188 h90 suc22::suc22(1-69)-RRM2(69-389) cdc22-M45::RRM1-KMX ddb1::nat This study spd1::hygro Eg4304 h+ leu1+::nmt1-hR2 ura4-D18 cdc22-M45::hR1-kmx suc22::hR2 This study Eg4317 h cdc22::hR1-kmx suc22::hR2 cds1-2HA6His::ura4+ ura4-D18 leu1+::nmt1-hR2 This study leu1-32 Eg4324 h- cdc22-M45::hR1-kmx suc22::hp53R2 rad3-ts ddb1::nat spd1::hygro This study Eg4325 h+ cdc22-M45::hR1-kmx suc22::hp53R2 rad3-ts spd1::hygro This study Eg4326 h+ cdc22-M45::hR1-kmx suc22::hp53R2 rad3-ts ddb1::nat spd1::hygro This study Eg4328 h+ cdc22-M45::hR1-kmx suc22::hp53R2 rad3-ts ddb1::nat This study Eg4331 h Δcds1::ura4+ leu1-32::nmt1-hR2-leu1+ cdc22::hR1-kmx suc22::hR2 leu1-32 This study ura4-D18 Eg4339 h90 leu1-32::nmt-hR2-leu1+ cdc22-M45::hR1-D57N-kmx suc22::hR2 This study Eg4362 h+ leu1+::nmt1-NLS-mCherry leu1-32 cdc22-M45::hR1-kmx suc22::hp53R2 This study Eg4366 h suc22::suc22(1-69)-RRM2(69-389) cdc22-M45::RRM1-KMX ura4-D18 rad3-ts This study ddb1::nat Eg4367 h suc22::suc22(1-69)-RRM2(69-389) cdc22-M45::RRM1-KMX rad3-ts ddb1::nat Laboratory stock spd1::hygro Eg4372 h cdc22-M45::hR1-kmx suc22::hp53R2 ste9::ura4+ ura4-D18 This study Eg4373 h+ leu1+::nmt1-NLS-mCherry leu1-32 ade6-704 This study Eg4374 h+ leu1+::nmt1-NLS-CFP leu1-32 ade6-704 This study Eg4375 h+ cdc22::hR1-kmx suc22::hR2 ura4-D18 leu1+::nmt1-NLS-CFP leu1-32 This study Eg4388 h- cds1::ura4 ura4::adh-dmdNK-nat-adh-hENT Laboratory stock Eg4390 h+ cds1::ura4 ura4::adh-dmdNK-nat-adh-hENT cdc22-D57N Laboratory stock Eg4404 h- ura4::adh-dmdNK-nat-adh-hENT ura4-aim cdc22::hR1-kmx suc22::hR2 Laboratory stock

75 Eg4406 h+ ura4::adh-dmdNK-nat-adh-hENT ura4-aim cdc22::hR1-kmx suc22::hR2 Laboratory stock Δspd1::hygro Eg4408 h- ura4::adh-dmdNK-nat-adh-hENT ura4-aim cdc22::hR1-D57N-kmx suc22::hR2 Laboratory stock Eg4410 h- ura4::adh-dmdNK-nat-adh-hENT ura4-aim cdc22::hR1-kmx suc22::hp53R2 Laboratory stock Eg4414 h90 Δsuc22::suc22(1-69)RRM2(69-389) cdc22-M45::hR1-kmx ura4-D18 This study Eg4415 h Δsuc22::suc22(1-69)RRM2(69-389) cdc22-M45::hR1-kmx ste9::ura4+ This study ura4-D18 Eg4471 h cdc22::hR1-kmx suc22::hR2 ste9::ura4+ ura4-D18 This study Eg4481 h- cdc22-M45::hR1-kmx suc22::hp53R2 spd1::hygro spd2::nat This study Eg4485 h90 cdc22-M45::hR1-kmx suc22::hp53R2 spd1::hygro spd2::nat This study Eg4496 h- cdc22-M45::hR1-kmx suc22::hR2 rad3-ts Laboratory stock Eg4497 h- cdc22-M45::hR1-kmx suc22::hR2 rad3-ts Laboratory stock Eg4498 h- cdc22-M45::hR1-D57N-kmx suc22::hR2 rad3-ts Laboratory stock Eg4499 h- cdc22-M45::hR1-D57N-kmx suc22::hR2 rad3-ts Laboratory stock Eg4608 h+ cdc22::hR1-kmx suc22::hR2 This study Eg4610 h+ cdc22-M45::hR1-D57N-kmx suc22::hR2 This study Eg4611 h+ cdc22-M45::hR1-kmx suc22::hp53R2 This study Eg4613 h+ cdc22-M45::hR1-kmx suc22::hp53R2 spd1::hygro This study Eg4614 h+ suc22::suc22(1-69)-RRM2(69-389) cdc22-M45::RRM1-KMX This study

76 Table S2. Primer list

Primer Sequence (5ʹ- 3ʹ) Purpose

ACE6 ATGGGCCTTGAGCATCTCGAAGA Forward primer to verify chimR2. Anneals to suc22. Works with ACE10.

ACE10 GGAAAAATAACAAAACGACGAGGGT Reverse primer to verify chimR2. Anneals to RRM2.

FP3 ATGTATCCGCTCATGAGACA Forward primer to amplify pDUAL plasmid.

KVL1 GAAATCGACGACCAGCTATAAA Reverse primer to verify presence of ura4

KVL2 TATTCAGCTAGAGCTGAGGGG Forward primer to verify presence of ura4

KVL3 CCTAATCCTCACCGCTTTG Forward primer to verify presence of suc22

KVL4 GTAAGACCAGGCATAAGTCCA Reverse primer to verify presence of suc22

KVLFP8 GCA GGC TCT CAT ATG GAT AAA GCG Forward primer to amplify NLS sequence. Works with GAA TTA ATT CC KVLRP8.

KVLRP8 TGC TCA CCA TTG GAT CTT CAA CTT Reverse primer to amplify NLS sequence. TTC TTT

KVLFP9 AGA TCC AAT GGT GAG CAA GGG Forward primer to amplify CFP/mCherry ORF. Works CGA G with KVLRP9

KVLRP9 CTC TAG AGT CGA GAT CTC ACT TGT Reverse primer to amplify CFP/mCherry ORF. ACA GCT CGT CCA TGC

Rad3 FP TGCACTGCGAAGTCCTAACC Forward primer used for sequencing of rad3-ts point mutation. Works with Rad3 RP.

Rad3 RP AAGCAGCCCAACCAATGTAC Reverse primer used for sequencing of rad3-ts point mutation.

RC 1 GGGTCTCCCATGGGTGACCCTGAACG Forward primer to add NcoI restriction site upstream of RRM2B. Works with RC primer 2.

RC 2 TGCTTAAGATCTTAAAAGTCAGCATCT Reverse primer to add BglII restriction site AATGTAAATACAT downstream of RRM2B.

RC 3 TGCTTACCATGGGAGACCCTGCTTTTT Reverse primer to amplify fragment of pDUAL TGT plasmid for creation of pON1249

RC 4 CCACTGAGATCTCGACTCTA Forward primer to amplify fragment of pDUAL plasmid for creation of pON1249

RP2 TGTCTCATGAGCGGATACAT Reverse primer to amplify pDUAL plasmid.

ONP601 CTCTTATTGACCACACCTCTACC Forward primer to verify successful integration of a pDUAL fragment at the chromosomal leu1 locus. Works with ONP602.

ONP602 GGTCATAAAGTTGAACGGATGTCG Reverse primer to verify successful integration of a pDUAL fragment at the chromosomal leu1 locus.

77 ONP722 TATTTTAGCGGAACTTTGATG Forward primer to verify insertion of RRM1. Anneals to cdc22 5ʹ-end. Works with ONP723.

ONP723 ATCACGATCGTAAATAATAG Reverse primer to verify insertion of RRM1. Anneals to RRM1. Also used for sequencing of the hR1-D57N mutation.

ONP724 TCAAATTGTTAATCCTCATC Forward primer to verify insertion of RRM1. Anneals to RRM1. Works with ONP725.

ONP725 GCATTGCATTAATCAAGAGAC Reverse primer to verify insertion of RRM1. Anneals to cdc22 3ʹ-end.

ONP726 GAGGCCTCTTGCGAGATTGCC Forward primer to verify absence of cdc22. Anneals to cdc22. Works with ONP725.

ONP757 AGTTGGGGACGAGGAGTACC Reverse primer to verify absence of cdc22. Anneals to cdc22. Works with ONP722.

ONP789 AACACACCAGAAGGGAAAGG Forward primer to verify insertion of RRM2. Anneals to suc22 5ʹ-end. Works with ONP790.

ONP790 CAAGAGACAAACCCTTCAAAGGA Reverse primer to verify insertion of RRM2. Anneals to RRM2.

ONP848 TCTTCGAGATGCTCAAGGCC Reverse primer to verify absence of suc22. Anneals to suc22. Works with ONP849.

ONP849 CCCATTTAATTTGAGATTGTCTTCG Forward primer to verify insertion of RRM2B. Anneals to the suc22 5ʹ-end.

ONP868 AGTCTACGACTACTTGCAATCTC Forward primer to verify HA-tag on chk1.

ONP869 CAAACGAACTGGCACGACCTTTC Reverse primer to verify HA-tag on chk1.

ONP870 TGTAAAACACCACGAGACAA Forward primer used for sequencing of hR1-D57N mutation. Anneals to the nmt1 promoter.

ONP876 GCAGTTATGGCTGAAACTACGG Forward primer to verify insertion of RRM2B. Anneals to RRM2B. Works with ONP877.

ONP877 CCCGACATCATGGCGGTTG Reverse primer to verify insertion of RRM2B. Anneals to suc22 3ʹ-end.

ONP878 CCAAGCGCGCTGAGAAGG Forward primer to verify absence of suc22. Anneals to suc22. Works with ONP877.

ONP879 GCCAGCTGCTTCAGGTCG Reverse primer to verify insertion of RRM2B. Anneals to RRM2B. Works with ONP849

78 Table S3. Plasmid list

Plasmid Content Purpose pON838 pREP3X-spd1 Used for overexpression of spd1. pON910 pDUAL-FFH1c, AmpR Used as destination vector for Gateway cloning. pON916 pDUAL-GFH1c, AmpR pON1049 pDUAL(pON910)-RRM2, AmpR Used for amplification of pDUAL fragment. pON1157 RRM2B with suc22-5ʹ and -3ʹ overhangs, Used for integration of RRM2B at the suc22 locus. KanaR pON1204 pDUAL(pON916)-mis6-CFP Used as vector to create pON1237 and pON1238 pON1211 pENTR221-IRBIT, KanaR Used for Gateway cloning to insert IRBIT in pDUAL vector. pON1212 pDUAL(pON910)-IRBIT, AmpR Used for ectopic expression of IRBIT. pON1226 pDUAL-histone H3.1-mCherry Used for amplification of mCherry ORF. pON1228 pDUAL-histone H3.1-CFP Used for amplification of CFP ORF. pON1234 NLS-yEmRFP-URA3, AmpR Used for amplification of NLS sequence. pON1237 pDUAL-NLS-mCherry Expression vector used for integration at the leu1 (not Gateway compatible) locus to express NLS-mCherry. pON1238 pDUAL-NLS-CFP Expression vector used for integration at the leu1 (not Gateway compatible) locus to express NLS-CFP. pON1249 pDUAL(pON910)-RRM2B Used for overexpression of RRM2B. (not Gateway compatible)

79 Media and growth conditions Cells were grown at 30°C (unless otherwise stated) in rich yeast extract-based medium (YES/YEL) or minimal medium (MSL/MSA). Malt extract was used for mating of strains. For heterologous gene expression from the nmt1 promoter, addition of thiamine (5 µg/mL) to the medium repressed the promoter when desired. To induce replication stress, HU (Merck) was added to the medium at indicated concentrations.

Cloning Cloning was performed according to the manufactures instructions. Either chemically competent or electrocompetent E.coli cells were used for transformation. Plasmids were purified using the GeneJET Plasmid Miniprep Kit (Thermo Fisher), and correct assembly was verified by either restriction enzyme digest, PCR, or sequencing. Sequencing was performed by Eurofins (https://eurofinsgenomics.eu). Gibson Assembly® Cloning Kit (NEB): PCR fragments (0.02-0.5 pmols) containing 20-40 overlaps with adjacent PCR fragments were assembled using 10 µL Gibson Assembly Master Mix, and incubated in a thermocycler at 50°C for 30 min. 4 µL of the reaction were used for transformation. GatewayTM LR ClonaseTM II Enzyme Mix (Thermo Fisher): The entry clone containing the gene of interest (50-150 ng) was mixed with the destination vector of choice (150 ng/µL) in a reaction containing LR ClonaseTM II enzyme mix and TE buffer. The reaction was incubated at 25°C for 1 hour and stopped by addition of proteinase K for 10 min at 37°C. 2 µL of the reaction were used for transformation. T4 DNA ligase (NEB): Ligation of restriction fragments was performed in a reaction containing vector (50 ng), insert (37.5 ng), T4 DNA ligase and H2O. The reaction was incubated O/N at 16°C and 4 µL were used for transformation.

80 Protoplast fusion Exponentially growing cells (100 mL) were harvested by centrifugation and washed in 0.65M KCl. Generation of protoplasts was achieved by digestion of the cell wall by Lallzyme MMX enzyme mix (Lallemand) (0.1 g/mL Lallzyme MMX in KCl) at 36°C. When close to 100% of the cells had been converted to protoplasts, digestion was stopped by adding TS buffer (10 mM Tris-HCl, pH 7.6, 1.2 M Sorbitol). The protoplasts were washed in TS buffer and resuspended in TSC buffer (TS buffer + 10 8 mM CaCl2) to a density of 10 cells/mL. Aliquots of the protoplast samples to be fused were mixed and incubated at RT for 15 min. The mixed protoplasts were centrifuged, resuspended in 0.5 mL PEG solution (20% PEG 4000, 10 mM CaCl2, 10 mM Tris-HCl, pH 7.6), and incubated at RT for 15 min. The cells were subsequently harvested by gentle centrifugation, resuspended in 0.5 mL TSC buffer, and aliquots were plated on selective plates containing 1.2 M sorbitol.

Spot assays Cultures were grown to stationary phase and cell titers were adjusted to 107 cells/mL. Tenfold serial dilutions were prepared for each culture and spotted onto indicated plates, starting with an initial plating of 5 x 104 cells. The plates were incubated at indicated temperatures for 3-4 days before analysis.

Meiotic analysis Homothallic h90 strains of the indicated genotypes were subjected to sporulation at 25°C on malt extract. After 3 days, the number of spores (0-4) in ≥ 200 individual asci was counted.

Doubling time analysis Cultures were grown to exponential phase and diluted to a start concentration of 1.5 x 106 cells/mL. The cultures were grown for 8 hours and samples were taken with a one-hour interval. Cell concentrations were determined using the NucleoCounter® NC-3000TM (ChemoMetec) (protocol: Viability assay (Yeast)).

81 Cell cycle synchronization Exponentially growing cultures were collected on a 0.45-µm Millipore (Merck) filter by vacuum. The cells were washed with MSL without nitrogen (MSL-N) and resuspended in MSL-N to a concentration of 5 x 106 cells/mL. M-factor was added to the culture to a final concentration of 1 µg/mL. The cultures were incubated for 4 hours before collecting the cells by filtration. The collected cells were once more washed and resuspended in MSL-N. To release the cultures into S phase, arginine was added to a final concentration of 2 g/L. Samples were taken every 30 min, spun down and resuspended in 70% EtOH. The samples were subsequently prepared by RNase treatment and staining with the nucleic acid stain SytoxTM Green (Thermo Fisher). To monitor synchrony, cell cycle distribution analysis was performed using the NucleoCounter® NC-3000TM (ChemoMetec) (protocol: Cell-cycle Sytox Green assay (Yeast)). Data were processed by the FlowJo software.

Live cell imaging Strains with the NLS-mCherry or NLS-CFP constructs were cultured individually in MSL to continuously enforce fluorophore expression. After reaching exponential phase, the cultures were co-cultured (1:1) with an initial concentration of 106 cells/mL. The mixed cultures were divided in two; one control and one treated with 0.4 mM HU. Samples were taken every 2 hours until 8 hours after addition of HU. Sample imaging and analysis was performed using an Xcyto® Quantitative Cell Imager (ChemoMetec), gating separately live CFP-positive and live mCherry-positive cells. Each score was based on ≥ 1000 cells counts. Data were further analyzed in Microsoft Excel. Calculation of the Z´-factor was based on 20 co-cultures; 10 controls and 10 treated with 0.4 mM HU. Samples were taken before and 8 hours after HU treatment.

Based on the four parameters: the means of the samples (µs) and controls (µc) and the standard deviations of the samples (σs) and controls (σc), the Z´-factor was calculated using the formula:

3(� + � ) � = 1 − ! ! �! − �!

82 Protein extraction and Western blotting TCA extraction: Cells were grown to 5 x 106 cells/mL and collected on a 0.45- µm Millipore filter by vacuum. The cells were washed with equal volume ice cold

H2O and washed off the filter with 1 mL ice cold 20% TCA. The collected cells were spun down and resuspended in 200 µL 20% TCA. 500 µL acid washed glass beads were added and the cells were lyzed for 3 x 15 seconds in a FastPrep® instrument (MP Biomedicals). 400 µL 5% TCA was added and the cell lysate was spun down. The pellet was subsequently washed in 0.5 mL 80% acetone, and resuspended in 100 µL 1.5x SDS sample buffer (2x: 65.8 mM Tris-HCl, 26.37% (w/v) glycerol, 2.1% SDS, 0.01% bromophenol blue). The samples were heated for 5 min at 95°C before separation by SDS-PAGE. Separated proteins were transferred to a nitrocellulose membrane (Advantec) using semi-dry buffer (25 mM TrisBase, 192 mM glycine, 0.037% SDS, 10% methanol) and a Semidry Blotter (C.B.S. Scientific Co.). Western blotting: The membrane was blocked with PBS-T (PBS, 0.05% Tween) + 5% low-fat milk powder for 1 hour at RT and subsequently incubated with primary antibodies for 1 hour (or O/N at 4°C). The membrane was washed for 3 x 20 min with PBS-T, and subsequently incubated with secondary peroxidase-conjugated antibodies for 1 hour at RT. The washing procedure was repeated and the membrane was developed using an enhanced chemiluminescence (ECL) kit (Amersham) and an ImageQuant LAS 4000 imager (GE Healthcare). Quantifications of protein levels were calculated based on band intensities, measured with TotalLab Quant software (TotalLab).

Antibodies: Detection of Cds1-HA and Chk1-HA: Primary antibody: anti-HA mouse monoclonal IgG (Merck) (1:10000) Secondary antibody: rabbit anti-mouse polyclonal IgG/HRP (DAKO) (1:5000) Detection of human R2/p53R2: Primary antibody: anti-R2/p53R2 mouse monoclonal IgG (Santa Cruz) (1:200) Secondary antibody: rabbit anti-mouse polyclonal IgG/HRP (DAKO) (1:5000) Detection of Spd1: Primary antibody: anti-Spd1 rabbit polyclonal IgG, bleed 3 (1:50) Secondary antibody: swine anti-rabbit polyclonal IgG/HRP (DAKO) (1:5000)

83 Checkpoint activation analysis Cells were grown to 5x106 cells/mL and collected on a 0.45-µm Millipore filter by vacuum. The membrane was subsequently washed once with equal volume ice cold Stop buffer (150 mM NaCl, 50 mM NaF, 100 mM EDTA, 1 mM NaN3) and cells were washed off the filter. Cell pellets were resuspended in 300 mL lysis buffer (50 mM Tris (pH 7.5), 250 mM NaCl, 50 mM NaF, 5 mM EDTA, 1 mM DTT, 0.1mM orthovanadate, and 0.1% NP-40) supplemented with protease inhibitor tablets (cOmpleteTM Protease Inhibitor Cocktail and PhosSTOP (Merck); 1 tablet of each for 10 mL lysis buffer) together with acid washed glass beads. A FastPrep® instrument (MP Biomedicals) was used to lyze the cells for 3 x 15 seconds. Cell lysate was spun down and the supernatant was transferred to new tubes. Total protein concentrations in supernatants were determined by Bradford protein assay (Bio-Rad) and the responses was compared to that of a protein standard (BSA). For immunoblotting of Chk1-HA, a total of 50 µg protein was separated by SDS-PAGE, transferred to a nitrocellulose membrane and subjected to Western blotting. To detect Cds1 kinase activity, Cds1-HA was immunoprecipitated from 500 µg extract with anti-HA antibodies and Protein G Sepharose beads (GE Healthcare). Coupling occurred on a rotating wheel for 1 hour at 4°C. The immune complex attached to the beads were precipitated by centrifugation, and washed 3 times in lysis buffer and 3 times in 2 x kinase buffer (10 mM HEPES (pH 7.5), 150 mM KCl, 10 mM MgCl2, 2 mM EDTA). To each pellet slurry, 20 µL kinase reaction mix was added; (10 µL 2 x kinase buffer, 5µCi [γ-32P]ATP, 0.2 mM ATP, 0.25 mg/mL MBP). Samples were incubated for 15 min at 30°C. 20µl 2x SDS sample buffer was added to stop the reactions and the samples were moved to 80°C for 5 min. Samples were separated by SDS-PAGE. The gel was cut in two; the upper part containing Cds1-HA (~53 kDa) was transferred to a nitrocellulose membrane and subjected to Western Blotting. The bottom part of the gel containing MBP (~18.4 kDa) was transferred to destain buffer (10% CH3COOH, 20% EtOH), and dried for 2 hours at 80°C. A Typhoon FLA 7000 scanner (GE Healthcare) was used for autoradiography.

84 Nucleotide pool determination TCA extraction: Exponentially growing cultures were divided into 3 x 50 mL aliquots and collected on 0.45-µm Millipore filters by vacuum. The filters were subsequently washed with equal volume ice cold H2O and the cells were washed off with 500 µL ice cold 10% TCA/15 mM MgCl2 into fresh tubes and frozen at -80°C. The samples were thawed on ice, sonicated for 2 x 15 seconds at 2°C, and spun down. The supernatant of each sample was transferred to a fresh tube and extracted seven times with equal volumes of ether. Each round of extraction included the addition of ether, vortexing for 15 seconds, centrifugation for 1 min, and discarding the water phase. Extracts were kept at -80°C. ATP determination: ATP concentrations in the samples were measured with the luciferase-based ATP determination kit (Biaffin). The luminescence (Relative light units (RLU)) was measured using a luminometer, and the values were compared to a standard curve (0, 20, 40, 80, 160, 320 pmol ATP). Primer labeling and template annealing: 5ʹ-end labeling of primers p22 and p27 were performed at 37°C for 1 hour by setting up reactions containing 2 µL primer (20 pmol), 1.5 µL T4 PNK (NEB), 6 µL PNK buffer, 6 µL [γ-32P]ATP (10 mCi/mL) and

14.5 µL H2O. The labeled primers were purified on MicroSpin G-50 columns (GE Healthcare) and annealed to appropriate templates by setting up reactions containing 12 µL hot primer, 2.6 µL cold primer, 560 µM template, 52 µL 2.5x Annealing buffer, and H2O to 130 µL. The reactions were heated at 90°C for 5 min and allowed to cool slowly to RT. Hybrids were stored at -20°C. The following primer/template pairs were constructed:

p22/tA: 5ʹ-GGT AGG GCT TCG CAG CCG TCC A-3ʹ 3ʹ-CCA TCC CGA AGC GTC GGC AGG TTA ATA ATA ATA A-5ʹ

p22/tT: 5ʹ-GGT AGG GCT TCG CAG CCG TCC A-3ʹ 3ʹ-CCA TCC CGA AGC GTC GGC AGG TAT TAT TAT T-5ʹ

p27/tC: 5ʹ-GGT AGG GCT ATA CAT CGC AGC CGT CCA-3ʹ 3ʹ-CCA TCC CGA TAT GTA GCG TCG GCA GGT GTT GTT GTT-5ʹ

p27/tG: 5ʹ-GGT AGG GCT ATA CAT CGC AGC CGT CCA-3ʹ 3ʹ-CCA TCC CGA TAT GTA GCG TCG GCA GGT CAA CAA CAA-3ʹ

85 Primer extension assays: The primer extension assays were performed essentially as described previously (Roy et al., 1999). Reactions to determine purine triphosphates contained 2 µL 10x Klenow buffer (NEB#2), 1.5 µL primer-template hybrid (p22/tA or p27/tG), 100 pmol dTTP, and 4.5 µL H2O. Reactions to determine pyrimidine triphosphates contained 2 µL 10x Klenow buffer, 1.5 µL primer-template hybrid (p22/tT or p27/tC), 100 pmol dATP, and 4.5 µL H2O. Total reaction volume was 20 µL including 10 µL of cell extracts or 10 µL of a dNTP standard. 1 µL Klenow (NEB) was added to the samples to start the reactions, which were carried out at 27°C for 20 min. The reactions were stopped by incubation at 90°C for 5 min, and the samples were dried in an Eppendorf Concentrator (Eppendorf) for 40 min and resuspended in 10 µL formamide loading buffer (95% deionized formamide, 0.025% (w/v) bromophenol blue, 0.025% (w/v) xylene cyanol FF, 5 mM EDTA pH 8.0). The DNA products were separated in a 10% polyacrylamide-urea gel by vertical Gel Electrophoresis. The gel was subsequently dried for 2 hours at 80°C and a Typhoon FLA 7000 scanner (GE Healthcare) was used for autoradiography. Densitometric analysis: Densitometric analysis was performed using TotalLab Quant software (TotalLab). The primer/template pairs allow incorporation of dNTPs into one, two, or three positions (or four positions for p22/tA), leading to aborted products pi of different lengths. The amount of incorporated dNTPs was calculated using the following formulas: pmol dCTP incorporated pmol dTTP incorporated ! ! ! ! ! ! ! !

�! + �! + �! �! + �! �! + �! + �! �! + �! !!! !!! !!! !!! !!! !!! !!! !!! pmol dGTP incorporated pmol dATP incorporated ! ! ! ! ! ! !! !" !"

�! + �! + �! �! + �! �! + �! + �! + �! �! + �! !!! !!! !!! !!! !!! !!! !!! !!!" !!!

Where x0 represents the amount of radioactivity of the starting primer/template substrate p0/t and xi represents the amount of radioactivity of the consecutive elongated primer/template products pi/t. (Supplementary Figure S3).

86 Indirect immunofluorescence microscopy Cells were grown to mid-log phase and collected on a 0.45-µm Millipore filter by vacuum. Cells were fixed in 100% methanol for 10 min at -20°C and rehydrated in

1 mL PEM (100 mM PIPES, 1 mM EGTA, 1 mM MgSO4, pH 6.9). Cells were washes trice with PEM and resuspended in PEM to a concentration of 5 x 107 cells/mL. Cells from 1 mL of sample was digested with Zymolyase 20T (20 mg/mL) and lytic enzyme mix (25 mg/mL) in 1 mL PEMS (PEM with 1.2 M sorbitol) for 20 min and washed trice with 1 mL PEMS. After permeabilizing the cell membranes for 30 seconds with 1 mL 1% Triton-X-100 in PEMS, the cells were washed once in PEM, resuspended in 1 mL PEMBAL (PEM with 1% BSA, 100 mM Lysine hydrochloride, and 0.1% NaN3) and incubated on a rotary inverter for 1 hour. For staining, the cells were resuspended in 100 µL PEMBAL with primary antibody (Anti-R2/p53R2 mouse monoclonal IgG (Santa Cruz), 1:50) and incubated O/N. The following day, cells were washed trice in 1 mL PEM, resuspended in 100 µL PEMBAL with secondary antibody (Goat anti-mouse IgG coupled to Alexa Fluor® 594 (ThermoFisher), 1:400), and incubated in the dark for 5 hours. 2.5 µL of cells were mounted in Vectashield® Antifade Mounting Medium with DAPI (Vector laboratories) and microscopy was performed on a Zeiss Axio Imager microscope. Images were captured using Velocity software (Perkin Elmer) and further processed with Adobe Photoshop CC 2017.

87

88 References

Abraham, R.T. (2001). Cell cycle checkpoint signaling through the ATM and ATR kinases. Genes Dev. 15, 2177–2196. Anda, S., Rothe, C., Boye, E., and Grallert, B. (2016). Consequences of abnormal CDK activity in S phase. Cell Cycle 15, 963–973. Ando, H., Mizutani, A., Matsu-ura, T., and Mikoshiba, K. (2003). IRBIT, a novel inositol 1,4,5-trisphosphate (IP3) receptor-binding protein, is released from the IP3 receptor upon IP3 binding to the receptor. J. Biol. Chem. 278, 10602–10612. Ando, N., Brignole, E.J., Zimanyi, C.M., Funk, M.A., Yokoyama, K., Asturias, F.J., Stubbe, J., and Drennan, C.L. (2011). Structural interconversions modulate activity of Escherichia coli ribonucleotide reductase. Proc. Natl. Acad. Sci. 108, 21046–21051. Andreson, B.L., Gupta, A., Georgieva, B.P., and Rothstein, R. (2010). The ribonucleotide reductase inhibitor, Sml1, is sequentially phosphorylated, ubiquitylated and degraded in response to DNA damage. Nucleic Acids Res. 38, 6490–6501. Angiolella, V.D., Donato, V., Forrester, F.M., Jeong, Y., Pellacani, C., Kudo, Y., Saraf, A., Florens, L., Washburn, M.P., and Pagano, M. (2012). Cyclin F-Mediated Degradation of Ribonucleotide Reductase M2 Controls Genome Integrity and DNA Repair. Cell 149, 1023–1034. Arnaoutov, A., and Dasso, M. (2014). iRBIT is a novel regulator of ribonucleotide resuctase in higher eukaryotes. Science (80-. ). 245, 1–4. Awasthi, P., Foiani, M., and Kumar, A. (2016). ATM and ATR signaling at a glance. J. Cell Sci. 129, 1285–1285. Aye, Y., Li, M., Long, M.J.C., and Weiss, R.S. (2014). Ribonucleotide reductase and cancer: Biological mechanisms and targeted therapies. Oncogene 34, 2011– 2021. Beach, D., Nurse, P., and Egel, R. (1982). Molecular rearrangement of mating-type genes in fission yeast. Nature 296, 682–683. Bentley, N.J., Holtzman, D.A., Flaggs, G., Keegan, K.S., DeMaggio, A., Ford, J.C., Hoekstra, M., and Carr, A.M. (1996). The Schizosaccharomyces pombe rad3 checkpoint gene. EMBO J. 15, 6641–6651. Boddy, M.N., Furnari, B., Mondesert, O., and Russell, P. (1998). Replication checkpoint enforced by kinases Cds1 and Chk1. Science (80-. ). 280, 909–912. Bondar, T., Mirkin, E. V., Ucker, D.S., Walden, W.E., Mirkin, S.M., and Raychaudhuri, P. (2003). Schizosaccharomyces pombe Ddb1 is functionally linked to the replication checkpoint pathway. J. Biol. Chem. 278, 37006–37014. Borgne, A., and Nurse, P. (2000). The Spd1p S phase inhibitor can activate the DNA replication checkpoint pathway in fission yeast. J. Cell Sci. 113 Pt 23, 4341– 4350. Bourdon, A., Minai, L., Serre, V., Jais, J.-P., Sarzi, E., Aubert, S., Chretien, D., de Lonlay, P., Paquis-Flucklinger, V., Arakawa, H., et al. (2007). Mutation of RRM2B, encoding p53-controlled ribonucleotide reductase (p53R2), causes severe mitochondrial DNA depletion. Nat. Genet. 39, 776–780. Brignole, E.J., Tsai, K.L., Chittuluru, J., Li, H., Aye, Y., Penczek, P.A., Stubbe, J.A., Drennan, C.L., and Asturias, F. (2018). 3.3-Å resolution cryo-EM structure of human ribonucleotide reductase with substrate and allosteric regulators bound. Elife 7, 1–23. Brown, N.C., and Reichard, P. (1969). Role of effector binding in allosteric control of ribonucleoside diphosphate reductase. J. Mol. Biol. 46, 39–55.

89 Capasso, H. (2002). Phosphorylation activates Chk1 and is required for checkpoint-mediated cell cycle arrest. J. Cell Sci. 115, 4555–4564. Caras, I.W., and Martin, D.W. (1988). Molecular Cloning of the cDNA for a Mutant Mouse Ribonucleotide Reductase Ml That Produces a Dominant Mutator Phenotype in Mammalian Cells. 8, 2698–2704. Chabes, A., Domkin, V., and Thelander, L. (1999). Yeast Sml1, a protein inhibitor of ribonucleotide reductase. J. Biol. Chem. 274, 36679–36683. Chabes, A., Georgieva, B., Domkin, V., Zhao, X., Rothstein, R., and Thelander, L. (2003a). Survival of DNA damage in yeast directly depends on increased dNTP levels allowed by relaxed feedback inhibition of ribonucleotide reductase. Cell 112, 391–401. Chabes, A.L., Pfleger, C.M., Kirschner, M.W., and Thelander, L. (2003b). Mouse ribonucleotide reductase R2 protein: A new target for anaphase-promoting complex-Cdh1-mediated proteolysis. Proc. Natl. Acad. Sci. 100, 3925–3929. Chanpimol, S., Seamon, B., Hernandez, H., Harris-love, M., and Blackman, M.R. (2017). Cladribine and Fludarabine Nucleotides Induce Distinct Hexamers Defining a Common Mode of Reversible RNR Inhibition. 11, 2021–2032. Chen, P.Y.-T., Funk, M.A., Brignole, E.J., and Drennan, C.L. (2018). Disruption of an oligomeric interface prevents allosteric inhibition of Escherichia coli class Ia ribonucleotide reductase. J. Biol. Chem. 293, 10404–10412. Costello, G., Rodgers, L., Beach, D., Spring, C., Box, P.O., and Harbor, C.S. (1986). Fission yeast enters the stationary phase Go state from either mitotic G1 or G2. Crona, M., Codó, P., Jonna, V.R., Hofer, A., Fernandes, A.P., and Tholander, F. (2016). A ribonucleotide reductase inhibitor with deoxyribonucleoside-reversible cytotoxicity. Mol. Oncol. 10, 1375–1386. Domkin, V., Thelander, L., and Chabes, A. (2002). Yeast DNA damage-inducible Rnr3 has a very low catalytic activity strongly stimulated after the formation of a cross-talking Rnr1/Rnr3 complex. J. Biol. Chem. 277, 18574–18578. Egel, R. (2004). The Molecular Biology of Schizosaccharomyces pombe. Springer. Egel, R. (2005). Fission yeast mating-type switching: Programmed damage and repair. DNA Repair (Amst). 4, 525–536. Eidesgaard, A.C. (2017). Mechanism of replication stress in fission yeast depending on human ribonucleotide reductase. Ekberg, M., Sahlin, M., Eriksson, M., and Sjöberg, B.M. (1996). Two conserved tyrosine residues in protein R1 participate in an intermolecular electron transfer in ribonucleotide reductase. J. Biol. Chem. 271, 20655–20659. Eklund, H., Uhlin, U., Färnegårdh, M., Logan, D.T., and Nordlund, P. (2001). Structure and function of the radical enzyme ribonucleotide reductase. Prog. Biophys. Mol. Biol. 77, 177–268. Elledge, S.J., Zhou, Z., and Allen, J.B. (1992). Ribonucleotide reductase: regulation, regulation, regulation. Trends Biochem. Sci. 17, 119–123. Engstrom, Y., Eriksson, S., Jildevik, I., Skogs, S., and Thelander, L. (1985). Cell Cycle-dependent Expression of Mammalian Ribonucleotide. 260, 9114–9116. Eriksson, S., Thelander, L., and Akerman, M. (1979). Allosteric regulation of calf thymus ribonucleoside diphosphate reductase. Biochemistry 18, 2948–2952. Fairman, J.W., Wijerathna, S.R., Ahmad, F., Xu, H., Nakano, R., Jha, S., Prendergast, J., Welin, R.M., Flodin, S., Roos, A., et al. (2011). Structural basis for allosteric regulation of human ribonucleotide reductase by nucleotide-induced oligomerization. Nat. Publ. Gr. 18, 316–322.

90 Fernandez Sarabia, M.J., McInerny, C., Harris, P., Gordon, C., and Fantes, P. (1993). The cell cycle genes cdc22+ and suc22+ of the fission yeast Schizosaccharomyces pombe encode the large and small subunits of ribonucleotide reductase. MGG Mol. Gen. Genet. 238, 241–251. Fleck, O., Vejrup-hansen, R., Watson, A., Carr, A.M., Nielsen, O., and Holmberg, C. (2013). Spd1 accumulation causes genome instability independently of ribonucleotide reductase activity but functions to protect the genome when deoxynucleotide pools are elevated. Fleck, O., Fahnøe, U., Løvschal, K.V., Gasasira, M.F.U., Marinova, I.N., Kragelund, B.B., Carr, A.M., Hartsuiker, E., Holmberg, C., and Nielsen, O. (2017). Deoxynucleoside salvage in fission yeast allows rescue of ribonucleotide reductase deficiency but not spd1-mediated inhibition of replication. Genes (Basel). Forsburg, S.L. (1999). The best yeast? Trends Genet. 15, 340–344. Forsburg, S., and Rhind, N. (2006). Basic Methods for Fission Yeast. 23, 173–183. Furnari, B., Rhind, N., and Russell, P. (1997). Cdc25 mitotic inducer targeted by Chk1 DNA damage checkpoint kinase. Science (80-. ). 277, 1495–1497. Greenblatt, M.S., Bennett, W.P., Hollstein, M., and Harris, C.C. (1994). Mutations in the p53 Tumor Suppressor Gene : Clues to Cancer Etiology and Molecular Pathogenesis. Cancer Res. 54, 4855–4878. Guittet, O., Håkansson, P., Voevodskaya, N., Fridd, S., Gräslund, A., Arakawa, H., Nakamura, Y., and Thelander, L. (2001). Mammalian p53R2 Protein Forms an Active Ribonucleotide Reductase in Vitro with the R1 Protein, Which Is Expressed Both in Resting Cells in Response to DNA Damage and in Proliferating Cells. J. Biol. Chem. Hakansson, P., Dahl, L., Chilkova, O., Domkin, V., and Thelander, L. (2006). The Schizosaccharomyces pombe replication inhibitor Spd1 regulates ribonucleotide reductase activity and dNTPs by binding to the large Cdc22 subunit. J. Biol. Chem. 281, 1778–1783. Håkansson, P., Hofer, A., and Thelander, L. (2006). Regulation of mammalian ribonucleotide reduction and dNTP pools after DNA damage and in resting cells. J. Biol. Chem. 281, 7834–7841. Harris, P., Kersey, P.J., McInerny, C.J., and Fantes, P.A. (1996). Cell cycle, DNA damage and heat shock regulate suc22+ expression in fission yeast. Mol. Gen. Genet. 252, 284–291. Hartwell, L.H., and Weinert, T.A. (1989). Checkpoints : control that ensure the order of cell cycles events. Science (80-. ). 246, 629–633. Hayles, J., and Nurse, P. (2001). A journey into space. 2. Hoffman, C.S., Wood, V., and Fantes, P.A. (2015). An ancient yeast for young geneticists: A primer on the Schizosaccharomyces pombe model system. Genetics 201, 403–423. Hollstein, M., Sidransky, D., Vogelstein, B., and Harris, C.C. (1991). p53 mutations in human cancers. Science 253, 49–53. Holmberg, C., and Nielsen, O. (2012). Replication: DNA building block synthesis on demand. Curr. Biol. 22, R271-2. Holmberg, C., Fleck, O., Hansen, H.A., Liu, C., Slaaby, R., Carr, A.M., and Nielsen, O. (2005). Ddb1 controls genome stability and meiosis in fission yeast. 7, 853– 862. Huang, M., and Elledge, S.J. (1997). Identification of RNR4, encoding a second essential small subunit of ribonucleotide reductase in Saccharomyces cerevisiae. Mol. Cell. Biol. 17, 6105–6113.

91 Huang, M.E., Facca, C., Fatmi, Z., Baïlle, D., Bénakli, S., and Vernis, L. (2016). DNA replication inhibitor hydroxyurea alters Fe-S centers by producing reactive oxygen species in vivo. Sci. Rep. 6. Humphrey, T. (2000). DNA damage and cell cycle control in Schizosaccharomyces pombe. Mutat. Res. 451, 211–226. Iyer, D.R., and Rhind, N. (2017). The intra-S checkpoint responses to DNA damage. Genes (Basel). 8. Jaspersen, S.L., Charles, J.F., and Morgan, D.O. (1999). Inhibitory phosphorylation of the APC regulator Hct1 is controlled by the kinase Cdc28 and the phosphatase Cdc14. Curr. Biol. 9, 227–236. Jordan, A., and Reichard, P. (1998). Ribonucleotide reductases. Annu. Rev. Biochem. 67, 71–98. Kashlan, O.B., Scott, C.P., Lear, J.D., and Cooperman, B.S. (2002). A comprehensive model for the allosteric regulation of mammalian ribonucleotide reductase. Functional consequences of ATP- and dATP-induced oligomerization of the large subunit. Biochemistry 41, 462–474. Kauppi, B., Nielsen, B.B., Ramaswamy, S., Larsen, I.K., Thelander, M., Thelander, L., and Eklund, H. (1996). The three-dimensional structure of mammalian ribonucleotide reductase protein R2 reveals a more-accessible iron-radical site than Escherichia coli R2. J. Mol. Biol. 262, 706–720. Kitamura, K., Maekawa, H., and Shimoda, C. (1998). Fission Yeast Ste9, a Homolog of Hct1/Cdh1 and Fizzy-related, Is a Novel Negative Regulator of Cell Cycle Progression during G1-Phase. Mol. Biol. Cell 9, 1065–1080. Klar, A.J.S. (1992). Developmental choices in mating-type interconversion in fission yeast. Trends Genet. 8, 208–213. Klar, A.J.S., Bonaduce, M.J., and Cafferkey, R. (1991). The mechanism of fission yeast mating type interconversion: Seal/replicate/cleave model of replication across the double-stranded break site at mat1. Genetics 127, 489–496. Krakoff, I.H., Brown, N.C., and Reichard, P. (1968). Inhibition Reductase of Ribonucleoside by Hydroxyurea1 Diphosphate. 1559–1565. Kriwacki, R.W., Wright, P.E., Hengst, L., Tennant, L., and Reed, S.I. (2002). Structural studies of p21Waf1/Cip1/Sdi1 in the free and Cdk2-bound state: conformational disorder mediates binding diversity. Proc. Natl. Acad. Sci. 93, 11504–11509. Kunz, B.A., and Kohalmi, S.E. (1991). MODULATION OF MUTAGENESIS Regulation of Synthesis. Synthesis (Stuttg). Lee, Y.D., Wang, J., Stubbe, J., and Elledge, S.J. (2008). Dif1 is a DNA-damage- regulated facilitator of nuclear import for ribonucleotide reductase. Mol. Cell 32, 70–80. Lindsay, H., Carr, A., Christensen, P., Edwards, R., Griffiths, D., Murray, J., Osman, F., and Walworth, N. (1998). S-specific activation of Cds1 kinase defines a subpathway of the checkpoing response in S pombe. Genes Dev. 12, 382–395. Liu, C., Powell, K. a, Mundt, K., Wu, L., Carr, A.M., and Caspari, T. (2003). Cop9/signalosome subunits and Pcu4 regulate ribonucleotide reductase by both checkpoint-dependent and -independent mechanisms. Genes Dev. 17, 1130–1140. Liu, C., Poitelea, M., Watson, A., Yoshida, S.H., Shimoda, C., Holmberg, C., Nielsen, O., and Carr, A.M. (2005). Transactivation of Schizosaccharomyces pombe cdt2+stimulates a Pcu4-Ddb1-CSN ubiquitin ligase. EMBO J. 24, 3940–3951.

92 Liu, Q., Guntuku, S., Cui, X.S., Matsuoka, S., Cortez, D., Tamai, K., Luo, G., Carattini-Rivera, S., DeMayo, F., Bradley, A., et al. (2000). Chk1 is an essential kinase that is regulated by Atr and required for the G2/M DNA damage checkpoint. Genes Dev. 14, 1448–1459. Madaan, K., Kaushik, D., and Verma, T. (2012). Hydroxyurea: a key player in cancer chemotherapy. Expert Rev. Anticancer Ther. 12, 19–29. Maréchal, A., and Zou, L. (2013). DNA damage sensing by the ATM and ATR kinases. Cold Spring Harb. Perspect. Biol. 5, 1–17. Martinho, R.G., Lindsay, H.D., Flaggs, G., DeMaggio, A.J., Hoekstra, M.F., Carr, A.M., and Bentley, N.J. (1998). Analysis of Rad3 and Chk1 protein kinases defines different checkpoint responses. EMBO J. 17, 7239–7249. Matsuoka, S., Huang, M., and Elledge, S.J. (1998). Linkage of ATM to cell cycle regulation by the Chk2 protein kinase. Science (80-. ). 282, 1893–1897. Matsuyama, A., Shirai, A., Yashiroda, Y., Kamata, A., Horinouchi, S., and Yoshida, M. (2004a). pDUAL, a multipurpose, multicopy vector capable of chromosomal integration in fission yeast. Yeast 21, 1289–1305. Matsuyama, A., Matsuyama, A., Shirai, A., Shirai, A., Yashiroda, Y., Yashiroda, Y., Kamata, A., Kamata, A., Horinouchi, S., Horinouchi, S., et al. (2004b). pDUAL, a multipurpose, multicopy vector capable of chromosomal integration in fission yeast. Yeast 21, 1289–1305. Maundrell, K. (1990). nmt1 of Fission Yeast. J. Biol. Chem. 265, 10857–10864. Milles, S., and Lemke, E.A. (2014). Mapping multivalency and differential affinities within large intrinsically disordered protein complexes with segmental motion analysis. Angew. Chem. Int. Ed. Engl. 53, 7364–7367. Minnihan, E.C., Nocera, D.G., and Stubbe, J. (2014). Reversible, long-range radical transfer in E. coli class Ia ribonucleotide reductase. 46, 1–27. Moss, J., Tinline-Purvis, H., Walker, C.A., Folkes, L.K., Stratford, M.R., Hayles, J., Hoe, K.-L., Kim, D.-U., Park, H.-O., Kearsey, S.E., et al. (2010). Break-induced ATR and Ddb1-Cul4(Cdt)(2) ubiquitin ligase-dependent nucleotide synthesis promotes homologous recombination repair in fission yeast. Genes Dev. 24, 2705– 2716. Munro, J.B., and Silva, J.C. (2012). Ribonucleotide reductase as a target to control apicomplexan diseases. Curr. Issues Mol. Biol. 14, 9–26. Nakano, K., Balint, E., Ashcroft, M., and Vousden, K.H. (2000). A ribonucleotide reductase gene is a transcriptional target of p53 and p73. Oncogene 19, 4283– 4289. Nestoras, K., Mohammed, A.H., Schreurs, A., Fleck, O., Watson, A.T., Poitelea, M., Shea, C.O., Chahwan, C., Holmberg, C., Kragelund, B.B., et al. (2010). Regulation of ribonucleotide reductase by Spd1 involves multiple mechanisms. 1, 1145–1159. Nielsen, H.B.N. (2013). Aspects of Ribonucleotide Reductase Regulation and Genome stability. Nielsen, O. (2003). COP9 Signalosome : A Provider of DNA. 13, 565–567. Nielsen, O. (2016). Synchronization of S Phase in Schizosaccharomyces pombe Cells by Transient Exposure to M-Factor Pheromone. In Cold Spring Harb. Protoc., p. Nordlund, P., and Reichard, P. (2006). Ribonucleotide reductases. Annu. Rev. Biochem. 75, 681–706. Nyberg, K.A., Michelson, R.J., Putnam, C.W., and Weinert, T.A. (2002). Toward maintaining the genome: DNA damage and replication checkpoints. Annu. Rev. Genet. 36, 617–656.

93 Pontarin, G., Ferraro, P., Håkansson, P., Thelander, L., Reichard, P., and Bianchi, V. (2007). p53R2-dependent ribonucleotide reduction provides deoxyribonucleotides in quiescent human fibroblasts in the absence of induced DNA damage. J. Biol. Chem. Pontarin, G., Ferraro, P., Bee, L., Reichard, P., and Bianchi, V. (2012). Mammalian ribonucleotide reductase subunit p53R2 is required for mitochondrial DNA replication and DNA repair in quiescent cells. Proc. Natl. Acad. Sci. U. S. A. 109, 13302–13307. Reichard, P. (1988). Interactions Between Deoxyribonucleotide And DNA Synthesis. Annu. Rev. Biochem. 57, 349–374. Reichard, P., Eliasson, R., Ingemarson, R., and Thelander, L. (2000). Cross-talk between the allosteric effector-binding sites in mouse ribonucleotide reductase. J. Biol. Chem. 275, 33021–33026. Reinhardt, H.C., Aslanian, A.S., Lees, J.A., and Yaffe, M.B. (2009). p53 deficient cells rely on ATM and ATR-mediated checkpoint signaling through the p38 MAPK/MK2pathway for survival after DNA damage. Cancer 11, 175–189. Rhind, N., and Russell, P. (2015). Tyrosine Phosphorylation of Cdc2 Is Required for the Replication Checkpoint in Schizosaccharomyces pombe . Mol. Cell. Biol. 18, 3782–3787. Rofougaran, R., Vodnala, M., and Hofer, A. (2006). Enzymatically active mammalian ribonucleotide reductase exists primarily as an alpha6beta2 octamer. J. Biol. Chem. 281, 27705–27711. Roy, B., Beuneu, C., Roux, P., Buc, H., Lemaire, G., and Lepoivre, M. (1999). Simultaneous determination of pyrimidine or purine deoxyribonucleoside triphosphates using a polymerase assay. Anal. Biochem. 269, 403–409. Running, D.M., Ligon, J.B., and Miskioglu, I. (1999). A Simple Statistical Parameter for Use in Evaluation and Validation of High Throughput Screening Assays. J. Biomol. Screen. 4, 928–940. Sabatinos, S.A., and Forsburg, S.L. (2010). Molecular genetics of schizosaccharomyces pombe (Elsevier Inc). Sabatinos, S.A., Green, M.D., and Forsburg, S.L. (2012). Continued DNA Synthesis in Replication Checkpoint Mutants Leads to Fork Collapse. Mol. Cell. Biol. 32, 4986–4997. Salguero, I., Guarino, E., Shepherd, M.E.A., Deegan, T.D., Havens, C.G., Macneill, S.A., Walter, J.C., and Kearsey, S.E. (2012). Ribonucleotide Reductase Activity Is Coupled to DNA Synthesis via Proliferating Cell Nuclear Antigen. Curr. Biol. 22, 720–726. Shao, J., Liu, X., Zhu, L., and Yen, Y. (2013). Targeting ribonucleotide reductase for cancer therapy. Expert Opin. Ther. Targets 17, 1423–1437. Shechter, D., Costanzo, V., and Gautier, J. (2004). Regulation of DNA replication by ATR: Signaling in response to DNA intermediates. DNA Repair (Amst). 3, 901– 908. Singh, A., and Xu, Y.J. (2016). The cell killing mechanisms of hydroxyurea. Genes (Basel). 7. Smith, P., Zhou, B., Ho, N., Yuan, Y.C., Su, L., Tsai, S.C., and Yen, Y. (2009). 2.6 Å X-ray crystal structure of human p53R2, a p53-inducible ribonucleotide reductase. Biochemistry 48, 11134–11141. Stubbe, J., and Riggs-Gelasco, P. (1998). Harnessing free radicals : formation and function of the tyrosyl radical in ribonucleotide reductase. 0004, 438–443. Sugase, K., Dyson, H.J., and Wright, P.E. (2007). Mechanism of coupled folding and binding of an intrinsically disordered protein. Nature 447, 1021–1025.

94 Suskiewicz, M.J., Sussman, J.L., Silman, I., and Shaul, Y. (2011). Context-dependent resistance to proteolysis of intrinsically disordered proteins. 20, 1285–1297. Tanaka, K., and Russell, P. (2001). Mrc1 channels the DNA replication arrest signal to checkpoint kinase Cds1. Nat. Cell Biol. 3, 966–972. Tanaka, H., Arakawa, H., Yamaguchi, T., Shiraishi, K., Fukuda, S., Matsui, K., and Takei, Y. (2000). A ribonucleotide reductase gene involved in a p53-dependent cell-cycle checkpoint for DNA damage. Taylor, E.M., McFarlane, R.J., and Price, C. (1996). 5-Azacytidine treatment of the fission yeast leads to cytotoxicity and cell cycle arrest. Mol. Gen. Genet. 253, 128– 137. Thelander, L., and Reichard, P. (1979). Reduction of ribonucleotides. Annu. Rev. Biochem. 48, 133–158. Tholander, F., and Sjoberg, B.-M. (2012). Discovery of antimicrobial ribonucleotide reductase inhibitors by screening in microwell format. Proc. Natl. Acad. Sci. U. S. A. 109, 9798–9803. Trickey, M., Grimaldi, M., and Yamano, H. (2008). The Anaphase-Promoting Complex/Cyclosome Controls Repair and Recombination by Ubiquitylating Rhp54 in Fission Yeast. Mol. Cell. Biol. 28, 3905–3916. Vejrup-hansen, R., Fleck, O., Landvad, K., Fahnøe, U., Broendum, S.S., Schreurs, A., Kragelund, B.B., Carr, A.M., Holmberg, C., and Nielsen, O. (2014). Spd2 assists Spd1 in the modulation of ribonucleotide reductase architecture but does not regulate deoxynucleotide pools. 2460–2470. Vogelstein, B. (1990). Cancer. A deadly inheritance. Nature 348, 681–682. Vogelstein, B., Lane, D., and Levine, A.J. (2000). Surfing the p53 network. Nature 408, 307–310. Wahl, G.M., and Carr, A.M. (2001). The evolution of diverse biological responses to DNA damage: Insights from yeast and p53. Nat. Cell Biol. 3. Walworth, N.C., and Bernards, R. (1996). Rad-dependent response of the chk1- Encoded protein kinase at the DNA damage checkpoint. Science (80-. ). Watson, A., Mata, J., Bähler, J., Carr, A., and Humphrey, T. (2004). Global gene expression responses of fission yeast to ionizing radiation. Mol. Biol. Cell 15, 851– 860. Wijerathna, S.R., Ahmad, M.F., Xu, H., Fairman, J.W., Zhang, A., Kaushal, P.S., Wan, Q., Kiser, J., and Dealwis, C.G. (2011). Targeting the Large Subunit of Human Ribonucleotide Reductase for Cancer Chemotherapy. Pharmaceuticals 4, 1328–1354. Willer, M., Nielsen, O., Styrkársdóttir, U., Hoffmann, L., Egel, R., and Davey, J. (2015). Two-step activation of meiosis by the mat1 locus in Schizosaccharomyces pombe. Mol. Cell. Biol. 15, 4964–4970. Wolfe, K.H., and Shields, D.C. (1997). Molecular evidence for an ancient duplication of the entire yeast genome. Nature 387, 708–713. Wood, V., Gwilliam, R., Rajandream, M.-A., Lyne, M., Lyne, R., Stewart, A., Sgouros, J., Peat, N., Hayles, J., Baker, S., et al. (2002). The genome sequence of Schizosaccharomyces pombe. Nature 415, 871. Wood, V., Harris, M.A., McDowall, M.D., Rutherford, K., Vaughan, B.W., Staines, D.M., Aslett, M., Lock, A., Bähler, J., Kersey, P.J., et al. (2012). PomBase: A comprehensive online resource for fission yeast. Nucleic Acids Res. 40, 695–699. Woollard, A., Basi, G., and Nurse, P. (1996). A novel S phase inhibitor in fission yeast. EMBO J. 15, 4603–4612. Wright, P.E., and Dyson, H.J. (2015). Intrinsically disordered proteins in cellular signalling and regulation. Nat. Rev. Mol. Cell Biol. 16, 18–29.

95 Wu, X., and Huang, M. (2008). Dif1 Controls Subcellular Localization of Ribonucleotide Reductase by Mediating Nuclear Import of the R2 Subunit. Mol. Cell. Biol. 28, 7156–7167. Xu, Y. jie (2016). Inner nuclear membrane protein Lem2 facilitates Rad3-mediated checkpoint signaling under replication stress induced by nucleotide depletion in fission yeast. Cell. Signal. 28, 235–245. Xu, H., Faber, C., Uchiki, T., Fairman, J.W., Racca, J., and Dealwis, C. (2006). Structures of eukaryotic ribonucleotide reductase I provide insights into dNTP regulation. Proc. Natl. Acad. Sci. U. S. A. 103, 4022–4027. Yamaguchi, T., Matsuda, K., Sagiya, Y., Iwadate, M., Nakamura, Y., Arakawa, H., and Fujino, M.A. (2001). p53R2-dependent pathway for DNA synthesis in a p53- regulated cell cycle checkpoint. Cancer Res. 61, 8256–8262. Yanamoto, S., Iwamoto, T., Kawasaki, G., Yoshitomi, I., Baba, N., and Mizuno, A. (2005). Silencing of the p53R2 gene by RNA interference inhibits growth and enhances 5-fluorouracil sensitivity of oral cancer cells. Cancer Lett. 223, 67–76. Yin, L., Locovei, A.M., and D’Urso, G. (2008). Activation of the DNA Damage Checkpoint in Mutants Defective in DNA Replication Initiation. Mol. Biol. Cell 19, 4374–4382. Zhang, Y.W., Jones, T.L., Martin, S.E., Caplen, N.J., and Pommier, Y. (2009). Implication of checkpoint kinase-dependent up-regulation of ribonucleotide reductase R2 in DNA damage response. J. Biol. Chem. 284, 18085–18095. Zhao, X., Muller, E.G., and Rothstein, R. (1998). A suppressor of two essential checkpoint genes identifies a novel protein that negatively affects dNTP pools. Mol. Cell 2, 329–340. Zhao, X., Georgieva, B., Chabes, A., Domkin, V., Ippel, J.H., Schleucher, J., Wijmenga, S., Thelander, L., and Rothstein, R. (2000). Mutational and structural analyses of the ribonucleotide reductase inhibitor Sml1 define its Rnr1 interaction domain whose inactivation allows suppression of mec1 and rad53 lethality. Mol. Cell. Biol. 20, 9076–9083. Zhao, X., Chabes, A., Domkin, V., Thelander, L., and Rothstein, R. (2001). The ribonucleotide reductase inhibitor Sml1 is a new target of the Mec1/Rad53 kinase cascade during growth and in response to DNA damage. EMBO J. 20, 3544–3553. Zou, L., Liu, D., and Elledge, S.J. (2003). Replication protein A-mediated recruitment and activation of Rad17 complexes. Proc. Natl. Acad. Sci. 100, 13827–13832.

96

Paper I

Deoxynucleoside Salvage in Fission Yeast Allows Rescue of Ribonucleotide Reductase Deficiency but Not Spd1-Mediated Inhibition of Replication

Oliver Fleck, Ulrik Fahnøe, Katrine Vyff Løvschal, Marie-Fabrice Uwamahoro Gasasira, Irina N. Marinova, Birthe B. Kragelund, Antony M. Carr, Edgar Hartsuiker, Christian Holmberg and Olaf Nielsen

Genes (2017) 8(5), DOI: 10.3390/genes8050128

97

G C A T T A C G G C A T genes

Article Deoxynucleoside Salvage in Fission Yeast Allows Rescue of Ribonucleotide Reductase Deficiency but Not Spd1-Mediated Inhibition of Replication

Oliver Fleck 1,2, Ulrik Fahnøe 1,†, Katrine Vyff Løvschal 1, Marie-Fabrice Uwamahoro Gasasira 2, Irina N. Marinova 1, Birthe B. Kragelund 3, Antony M. Carr 4, Edgar Hartsuiker 2, Christian Holmberg 1 and Olaf Nielsen 1,*

1 Cell Cycle and Genome Stability Group, Department of Biology, University of Copenhagen, DK-2200 Copenhagen, Denmark; o.fl[email protected] (O.F.); [email protected] (U.F.); [email protected] (K.V.L.); [email protected] (I.N.M.); [email protected] (C.H.) 2 North West Cancer Research Institute, Bangor University, Bangor, Gwynedd LL57 2UW, UK; [email protected] (M.-F.U.G.); [email protected] (E.H.) 3 Structural Biology and NMR Laboratory, Department of Biology, University of Copenhagen, DK-2200 Copenhagen, Denmark; [email protected] 4 Genome Damage and Stability Centre, University of Sussex, Brighton BN1 9RQ, UK; [email protected] * Correspondence: [email protected]; Tel.: +45-3532-2102 † Present address: Hvidovre Hospital and Department of International Health, Immunology and Microbiology, University of Copenhagen, DK-2650 Copenhagen, Denmark.

Academic Editor: Eishi Noguchi Received: 9 March 2017; Accepted: 20 April 2017; Published: 25 April 2017

Abstract: In fission yeast, the small, intrinsically disordered protein S-phase delaying protein 1 (Spd1) blocks DNA replication and causes checkpoint activation at least in part, by inhibiting the enzyme ribonucleotide reductase, which is responsible for the synthesis of DNA building blocks. The CRL4Cdt2 E3 ubiquitin ligase mediates degradation of Spd1 and the related protein Spd2 at S phase of the cell cycle. We have generated a conditional allele of CRL4Cdt2, by expressing the highly unstable substrate-recruiting protein Cdt2 from a repressible promoter. Unlike Spd1, Spd2 does not regulate deoxynucleotide triphosphate (dNTP) pools; yet we find that Spd1 and Spd2 together inhibit DNA replication upon Cdt2 depletion. To directly test whether this block of replication was solely due to insufficient dNTP levels, we established a deoxy-nucleotide salvage pathway in fission yeast by expressing the human equilibrative nucleoside transporter 1 (hENT1) and the Drosophila deoxynucleoside kinase. We present evidence that this salvage pathway is functional, as 2 µM of deoxynucleosides in the culture medium is able to rescue the growth of two different temperature-sensitive alleles controlling ribonucleotide reductase. However, salvage completely failed to rescue S phase delay, checkpoint activation, and damage sensitivity, which was caused by CRL4Cdt2 inactivation, suggesting that Spd1—in addition to repressing dNTP synthesis—together with Spd2, can inhibit other replication functions. We propose that this inhibition works at the point of the replication clamp proliferating cell nuclear antigen, a co-factor for DNA replication.

Keywords: DNA replication; checkpoints; ribonucleotide reductase; PCNA; CRL4Cdt2; intrinsically disordered proteins; deoxynucleotide salvage; fission yeast

1. Introduction Proliferating cell nuclear antigen (PCNA) is an essential co-factor for DNA polymerases during DNA replication in eukaryotes. It forms a ring-shaped homotrimer that encircles the double helix and tethers polymerases to DNA, thereby increasing their rate of processivity. PCNA also serves as a

Genes 2017, 8, 128; doi:10.3390/genes8050128 www.mdpi.com/journal/genes Genes 2017, 8, 128 2 of 14 platform for recruiting numerous other proteins to DNA, and hence is important for the metabolism of DNA and chromatin, during replication and repair. Most partner proteins bind PCNA via a conserved sequence called the PCNA-interacting protein-box (PIP-box), which associates with a hydrophobic pocket on the front face of the PCNA ring. The consensus PIP-box has the structure Q-x-x-Φ-x-x-Ω-Ω, in which Φ is a hydrophobic amino acid (L, V, I, or M) and Ω is an aromatic residue (Y or F). However, many PCNA interacting proteins have degenerate PIP-box sequences [1]. Since PCNA is a trimer, it can bind more than one PIP-box protein at a time. Consequently, PCNA has been proposed to function as a “tool belt” that can orchestrate the sequential recruitment of enzymes e.g., during maturation of Okazaki fragments [2]. In addition, a specialized E3 ubiquitin ligase called CRL4Cdt2 is dedicated to the degradation of proteins bound to PCNA, see [3]. A large number of substrates have been identified for PCNA-targeted degradation, including the cyclin-dependent kinase (CDK) inhibitor p21, the replication licensing factor Cdt1, and the histone methyltransferase Set8 [3]. Ubiquitylation-mediated degradation of these substrates occurs only when they associate with chromatin bound PCNA during S phase, or after DNA damage occurs. The proteins degraded through this pathway harbor a special version of the PIP-box, called a PIP-degron: Q-x-x-Φ-T-D-Ω-Ω-x-x-x-B, where B is a basic residue (K or R). CRL4Cdt2 mediated protein turnover at PCNA is thought to contribute to the orderly orchestration of replication and repair events. However, binding of p21 to PCNA can also directly inhibit replication [4]. In fission yeast, cells defective in the CRL4Cdt2 E3 ubiquitin ligase rely on the Rad3ATR (ATR: ataxia telangiectasia- and Rad3-related) checkpoint for survival. Curiously, the checkpoint activation is due to the untimely accumulation of a single CRL4Cdt2 substrate, a small intrinsically-disordered protein (IDP) called S-phase delaying protein 1 (Spd1) [5–10]. Spd1 was originally identified in a screen for proteins that inhibited replication when overexpressed [11]. The CRL4Cdt2 E3 ubiquitin ligase is activated by transcriptional induction of the Cdt2 substrate recruiting factor, which becomes expressed prior to S phase by the MluI cell cycle box (MCB) transcription complex [9]. Consequently, Spd1 is degraded during DNA replication in wild-type cells, whereas CRL4Cdt2 defective cells undergo S phase in the presence of Spd1, which gives rise to severe S-phase stress—cells accumulate during S phase and their survival relies notably on the activation of the Rad3 checkpoint. Furthermore, such cells are hypersensitive to DNA-damaging agents, they are defective in double-strand break (DSB) repair by homologous recombination, they display a more than 20-fold increase in spontaneous mutation rates, and they are also unable to undergo pre-meiotic S phase [5,7,12]. The requirement for a functional Rad3 pathway and the defects in recombination and pre-meiotic S-phase are all fully suppressed by deleting the spd1 gene, suggesting that these phenotypes are caused by Spd1 mediated interference with key replication functions. The first clue towards a replication target of Spd1 came from the observation that overexpression of the suc22 gene could relieve the checkpoint activation caused by Spd1 accumulation [5]. The suc22 gene encodes the small subunit of the enzyme ribonucleotide reductase (RNR) responsible for the rate-limiting step in the synthesis of deoxynucleotide triphosphates (dNTPs). Furthermore, it was found that Spd1 sequesters Suc22 in the nucleus, away from the large pan-cellular RNR subunit Cdc22, thereby reducing the cytosolic level of active RNR complexes [5]. However, mutants in spd1 that were defective in nuclear localization of Suc22, but still required rad3 in a CRL4Cdt2 mutated background, were subsequently identified, suggesting that checkpoint activation is not (or only in part) due to nuclear localization of Suc22 [8]. Consistent with Spd1 inhibiting RNR, deletion of the spd1 gene leads to a twofold increase in cellular dNTP pools [13]. Also, Spd1 can inhibit the enzymatic activity of RNR, and binds to both subunits in vitro [8,14]. However, several observations have challenged the view that Spd1 causes checkpoint activation by inhibiting RNR. First, while overexpression of Suc22 can suppress the checkpoint activation caused by Spd1 accumulation, overexpression of the large RNR R1 subunit Cdc22 fails to do so [15]. Since Suc22 directly binds to Spd1 [8] (B.B.K., unpublished observations), suppression may—in addition to boosting RNR activity—function by titrating Spd1 away from another target. Strikingly, cells harboring the Genes 2017, 8, 128 3 of 14 cdc22-D57N mutation, defective in dATP-mediated feedback inhibition of RNR, have a 2–5-fold increase in cellular dNTP pools. Yet, when Spd1 accumulation is induced in this cdc22-D57N background by CRL4Cdt2 inactivation, the checkpoint is still required, even though dNTP pools are higher than the wild-type levels [13]. The fission yeast genome encodes a second Spd1-related CRL4Cdt2-targeted IDP, called Spd2 [16]. Overexpression of Spd2 can also delay S phase, but Spd2 does not appear to regulate dNTP pools. Interestingly, on their own, cdc22-D57N or ∆spd2 only weakly rescue the Rad3 requirement of CRL4Cdt2-deficient cells. However, in the cdc22-D57N ∆spd2 double mutant, the checkpoint requirement is fully suppressed, similar to the case of spd1 deletion [16]. This observation suggests that Spd1 can cause checkpoint activation via both deoxynucleotide-dependent and -independent mechanisms, and that Spd2 only contributes to the latter. In the present study we have developed a conditional cdt2 allele that allows us to study the immediate effects of Spd1 and Spd2 accumulation. We show that Cdt2 depletion causes a strong inhibition of DNA replication that is dependent on both Spd1 and Spd2. We also report on the successful expression of a deoxynucleotide salvage pathway in fission yeast, by which we can overcome two mutants in RNR. However, consistent with Spd1 having other targets than RNR, nucleotide salvage was completely unable to relieve the replication problems and checkpoint activation induced by Spd1 accumulation.

2. Materials and Methods

2.1. Molecular and Genetic Procedures The Schizosaccharomyces pombe strains used in the present study are listed in Supplementary Table S1. Standard genetic procedures were performed according to [17]. The cdt2 open reading frame (ORF), including its stop codon, was recombined into the vector pDUAL-FFH41 [18] using Gateway technology (Invitrogen, Waltham, MA, USA). The resulting plasmid was digested with NotI and integrated at the leu1 locus, rendering Cdt2 expression repressible by thiamine. The Drosophila melanogaster gene encoding deoxyribonucleoside kinase (DmdNK) [19] under control of the fission yeast adh promoter, was integrated into the S. pombe genome, replacing ura4. Subsequently, the human equilibrative nucleoside transporter (hENT1) gene [20] under control of the adh promoter, coupled to a nourseothricin (natMX) resistance marker, was integrated adjacent to DmdNK.

2.2. Physiological Experiments and Cell Biology Cells were grown at 30 ◦C (unless otherwise stated) in minimal sporulation liquid (MSL) media [21] to a concentration of 5 × 106 cells/mL. Thiamine was added to a concentration of 5 µg/mL, when indicated. For spot test survival assays, 7 µL of 10-fold serial dilutions (starting with 107 cells/mL) were spotted on MSA plates (MSL with 2% agar) with indicated additives, and incubated 2–4 days at the indicated temperature. Cell cycle synchronization at G1 by nitrogen starvation in the presence of M-factor, and analysis of cell-cycle distribution by flow cytometry, were performed as described by [22]. FACS data were processed by the program FlowJo (FlowJo, Asland, OR, USA). Bimolecular fluorescence complementation (BiFC) was performed with the same constructs as described in [6]. Cds1 kinase assays and dNTP pool measurements were performed as previously described [13]. 5-ethynyl-2’-deoxyuridine (EdU) was added to 10 µM to cells growing exponentially at 30 ◦C in MSL; after 20 min, the cells were fixed in 70% ethanol. Incorporated EdU was coupled to Alexa Flour 545 azide as described [23]. Fluorescence and Nomarski microscopy was performed on a Zeiss Axio Imager platform (Zeiss, Jena, Germany). Genes 2017, 8, 128 4 of 14

3. Results

3.1. Generation of a Conditional CRL4Cdt2 Mutant Accumulation of Spd1 and Spd2 in cells with defective CRL4Cdt2 causes slow progression of the S phase and activation of the Rad3 checkpoint, which also becomes essential for cell survival. However, since CRL4Cdt2 defective cells have been deficient for many generations, it is difficult to discriminate between immediate and compensatory effects. In order to circumvent this problem, we constructed a conditional CRL4Cdt2 allele that could be inactivated within a single cell cycle. The E3 substrate recruiting protein Cdt2 has been reported to exhibit rapid turnover with a half-life of 10–15 min throughout the cell cycle [9], making repression of its transcription a good choice for fast down-regulation of CRL4Cdt2 function. We therefore expressed Cdt2 from the thiamine repressible nmt41 promoter [24] integrated at the leu1 locus in a ∆cdt2 background. We refer to this allele as cdt2TR (for thiamine repressible). In the absence of thiamine, cells of this strain appeared normal (Figure1A), grew with a doubling time similar to wild type (data not shown), and showed a normal DNA content profile (Figure1B, samples at t = 0), suggesting that the induced level of cdt2 transcription was sufficient to mediate Spd1 and Spd2 degradation at S phase. However, when we added thiamine to the culture, the cells elongated (Figure1A), and died in an spd1-dependent manner when a temperature sensitive rad3 allele was inactivated (Figure1C). Also, we found that when Spd1 and PCNA were tagged with, respectively, the N- and C-terminus of Venus yellow fluorescent protein (YFP), a bimolecular fluorescence complementation (BiFC) signal indicative of interaction between the two proteins was observed following thiamine addition (Figure1D), as previously reported for ∆cdt2 cells [6]. Flow cytometry showed that cells gradually accumulated in G1 and S phase, indicative of replication problems (Figure1B, first column). Consistent with cdt2TR cells forming colonies on plates containing thiamine (Figure1C), we found that the S phase arrest was only transient (data not shown). Deletion of the spd1 gene largely suppressed cell cycle arrest (Figure1B, second column). The transient S phase arrest observed appeared more severe than that which was seen in cdt2 deleted cells (Figure1E), presumably because the latter have adapted to the absence of Cdt2 by activating compensatory pathways. Overexpression of the spd2 gene can also inhibit replication, but Spd2 does not appear to regulate dNTP pool levels [16]. Spd2 is also degraded via CRL4Cdt2-mediated ubiquitylation, but unlike Spd1, Spd2 accumulation in a CRL4Cdt2 deficient background does not cause a requirement for the Rad3 checkpoint [16]. However, we found that cells with an spd2 deletion were, like ∆spd1, defective in cell cycle arrest upon switching off Cdt2 expression (Figure1B, third column). These observations demonstrate that the S phase arrest enforced by the inhibition of cdt2 transcription is dependent on Genes 2017, 8, 128 5 of 15 both Spd1 and Spd2 accumulation.

Figure 1. Cont.

Figure 1. Characterization of the cdt2TR allele. (A) cdt2TR cells were grown in minimal sporulation liquid (MSL) medium at 30 °C. The culture was divided in two, thiamine was added to the indicated culture, and pictures were taken with Nomarski optics after 12 hours. (B) cdt2TRcells of the indicated genotype were grown at 30 °C in MSL medium. At t = 0, thiamine was added to the four cultures. Samples were passed through flow cytometry at hourly intervals, as indicated. The apparent slight drift to the left at early time points in the Δspd2 strain was due to a DNA staining artefact. (C) Serial dilutions of strains with the indicated genotypes were spotted on plates either with or without thiamine, and incubated at the indicated temperature for three days. (D) cdt2TR cells expressing VN173-pcn1 and spd1-VC155 [6] were propagated in minimal sporulation liquid (MSL). The culture was divided in two, thiamine was added to the indicated culture, and pictures of yellow fluorescent protein (YFP) fluorescence were taken after four hours. (E) DNA content profiles of growing wild type and Δcdt2 cells.

3.2. Spd1 Accumulation Causes S Phase Delay Next, we wanted to examine how accumulation of Spd1 and Spd2 directly affected S phase progression. In order to do so, we used the recently developed method of M-factor treatment to synchronize cells in G1 [22]. When cdt2TR cells were released from G1 in the absence of thiamine, cells entered S phase after one hour and completed S phase within three hours (Figure 2A). This is similar to the kinetics observed with wild-type cells [22]. However, if we added thiamine to the cells at the time of release, both entry into, and progression through S phase were substantially delayed (Figure 2A). In fact, the cells had not completed DNA replication at the 300 min time point. These results confirm that we can rapidly downregulate cdt2 function by repressing its transcription. Deletion of the spd1 gene completely suppressed the thiamine-induced replication delay (Figure 2B), demonstrating that the replication block was dependent on Spd1 accumulation. In cells deleted for spd2, S phase progression was still substantially delayed by thiamine addition, presumably due to accumulation of Spd1 (Figure 2C). However, the completion of S phase was advanced by approximately two hours relative to spd2+ cells (compare Figure 2C and 2A). When both spd1 and spd2 were deleted (Figure 2D), the kinetics were fast, similar to those of Δspd1 cells

Genes 2017, 8, 128 5 of 15

Genes 2017, 8, 128 5 of 14

Figure 1. Characterization of the cdt2TR allele. (A) cdt2TR cells were grown in minimal sporulation Figureliquid (MSL)1. Characterization medium at 30 of◦C. the The cdt2 cultureTR allele. was ( dividedA) cdt2TR in cells two, were thiamine grown was in added minimal to the sporulation indicated liquidculture, (MSL) and picturesmedium wereat 30 °C. taken The with culture Nomarski was divi opticsded in after two, 12 thiamine h; (B) cdt2 wasTR addedcells ofto thethe indicated culture,genotype and were pictures grown were at 30 taken◦C in with MSL Nomarski medium. optics At t =after 0, thiamine 12 hours. was (B) addedcdt2TRcells to the of fourthe indicated cultures. genotypeSamples were were passed grown through at 30 °C flow in cytometryMSL medium. at hourly At t intervals,= 0, thiamine as indicated. was added The to apparent the four slight cultures. drift Samplesto the left were at early passed time through points in flow the ∆ cytometryspd2 strain wasat hourly due to intervals, a DNA staining as indicated. artefact; The (C )apparent Serial dilutions slight driftof strains to the withleft at the early indicated time points genotypes in the Δ werespd2 spottedstrain was on due plates to a either DNA withstaining or withoutartefact. ( thiamine,C) Serial dilutionsand incubated of strains at the with indicated the indicated temperature genotypes for three were days; spotted (D) cdt2 onTR platescells expressingeither with VN173-pcn1or without thiamine,and spd1-VC155 and incubated[6] were propagatedat the indicated in minimal temperature sporulation for three liquid days. (MSL). (D The) cdt2 cultureTR cells was expressing divided VN173-pcn1in two, thiamine and spd1-VC155 was added to[6] the were indicated propagated culture, in andminimal pictures sporulation of yellow liquid fluorescent (MSL). protein The culture (YFP) wasfluorescence divided werein two, taken thiamine after four was hours; added (E to) DNA the indicated content profiles culture, of an growingd pictures wild of type yellow and fluorescent∆cdt2 cells. protein (YFP) fluorescence were taken after four hours. (E) DNA content profiles of growing wild type 3.2. Spd1and Δ Accumulationcdt2 cells. Causes S Phase Delay

3.2. Spd1Next, Accumulation we wanted Causes to examine S Phase how Delay accumulation of Spd1 and Spd2 directly affected S phase progression. In order to do so, we used the recently developed method of M-factor treatment to synchronizeNext, we cells wanted in G1 to [22 examine]. When howcdt2 TRaccumulationcells were releasedof Spd1 fromand Spd2 G1 in directly the absence affected of thiamine, S phase progression.cells entered In S phaseorder afterto do one so, hour we used and completedthe recently S phasedeveloped within method three hoursof M-factor (Figure treatment2A). This isto synchronizesimilar to the cells kinetics in G1 observed[22]. When with cdt2TR wild-type cells were cells released [22]. from However, G1 in ifthe we absence added of thiamine thiamine, to cells the enteredcells at theS phase time after of release, one hour both and entry completed into, andS phase progression within three through hours S (Figure phase 2A). were This substantially is similar todelayed the kinetics (Figure observed2A). Infact, with the wild-type cells had cells not [22]. completed However, DNA if we replication added thiamine at the 300 to min the timecells point.at the timeThese of results release, confirm both entry that weinto, can and rapidly progression downregulate through cdt2S phasefunction were by substantially repressing itsdelayed transcription. (Figure 2A).Deletion In fact, of thethespd1 cellsgene had completelynot completed suppressed DNA replication the thiamine-induced at the 300 min replication time point. delay These (Figure results2B), confirmdemonstrating that we that can the rapidly replication downregulate block was cdt2 dependent functionon by Spd1 repressing accumulation. its transcription. Deletion of the spd1In cells gene deleted completely for spd2, Ssuppressed phase progression the thiamine-induced was still substantially replication delayed bydelay thiamine (Figure addition, 2B), demonstratingpresumably due that to the accumulation replication bloc of Spd1k was (Figure dependent2C). However,on Spd1 accumulation. the completion of S phase was advancedIn cells by deleted approximately for spd2 two, S hoursphase relativeprogression to spd2 was+ cells still (comparesubstantially Figure delayed2C and by Figure thiamine2A). Whenaddition, both presumablyspd1 and spd2 duewere to accumulation deleted (Figure of2 Spd1D), the (Figure kinetics 2C). were However, fast, similar the completion to those of ∆ofspd1 S phasecells was(Figure advanced2B). Taken by approximat together, theseely resultstwo hours show relative that Spd2 to spd2 accumulation+ cells (compare helps Figure to enforce 2C and the replication2A). When botharrest spd1 imposed and spd2 by Spd1. were deleted (Figure 2D), the kinetics were fast, similar to those of Δspd1 cells

Genes 2017, 8, 128 6 of 15

Genes 2017(Figure, 8, 128 2B). Taken together, these results show that Spd2 accumulation helps to enforce the6 of 14 replication arrest imposed by Spd1.

Figure 2. Spd1 accumulation inhibits S phase progression. Cells with the cdt2TR allele were Figure 2. Spd1 accumulation inhibits S phase progression. Cells with the cdt2TR allele were synchronized in G1 by nitrogen starvation in the presence of the M-factor pheromone [22]. The synchronizedpheromone in was G1 washed by nitrogen away, and starvation the cultures inwere the released presence into S of phase the in M-factor MSL medium pheromone at 30 °C, [22]. The pheromoneeither with or waswithout washed thiamine away, at t and= 0. Samples the cultures were taken were for released flow cytometry into S phaseat the indicated in MSL time medium ◦ at 30pointsC, either (min). withGenotypes: or without (A) cdt2 thiamineTR, (B) cdt2 atTR Δ tspd1 = 0., ( SamplesC) cdt2TR Δ werespd2 and taken (D) cdt2 forTR flow Δspd1 cytometry Δspd2. at the indicated time points (min). Genotypes: (A) cdt2TR;(B) cdt2TR ∆spd1;(C) cdt2TR ∆spd2 and (D) cdt2TR ∆3.3.spd1 Both∆spd2 Branches. of the Rad3 Checkpoint Are Involved in Tolerating Replication Problems Caused by Spd1 Accumulation 3.3. Both BranchesWe next defined of the Rad3 the extent Checkpoint to which Are the Involved function in of Tolerating the DNA Replicationdamage checkpoint, Problems in Caused addition by to Spd1Rad3, Accumulation was required for tolerance of Spd1 accumulation during replication. We crossed the conditional cdt2TR allele into various checkpoint mutant backgrounds, either in the presence or We next defined the extent to which the function of the DNA damage checkpoint, in addition to absence of spd1, and spotted cells onto plates with or without thiamine (Figure 3). For comparison, Rad3, was required for tolerance of Spd1 accumulation during replication. We crossed the conditional we also spotted the cells on plates containing a low concentration of the RNR inhibitor hydroxyurea TR cdt2 (HU).allele In intogeneral, various there checkpointwas a good correlation mutant backgrounds, between the checkpoint either in thefunctions presence required or absence to tolerate of spd1, and spottedthe two S cells phase onto inhibitors. plates withHowever, or without whereas thiamine deletion of (Figure the spd13). gene For comparison,completely suppressed we also spottedthe the cellssensitivity on plates to checkpoint containing loss a low caused concentration by Cdt2 ofdepletion, the RNR it inhibitorhad little hydroxyurea effect on HU (HU). sensitivity, In general, therepresumably was a good because correlation Spd1 does between not inhibit the checkpoint RNR in the functionsabsence of thiamine. required to tolerate the two S phase inhibitors.The However, core checkpoint whereas proteins deletion Rad3 ofand the Rad26,spd1 asgene well completelyas the 9-1-1 checkpoint suppressed clamp the (Rad1 sensitivity and to checkpointRad9) and loss its caused loader by(Rad17), Cdt2 depletion,were all absolutely it had littlerequired effect for on the HU survival sensitivity, of both presumably HU-treated and because Spd1 does not inhibit RNR in the absence of thiamine. The core checkpoint proteins Rad3 and Rad26, as well as the 9-1-1 checkpoint clamp (Rad1 and Rad9) and its loader (Rad17), were all absolutely required for the survival of both HU-treated and Spd1-accumulating cells. Mutants in the two branches of the Rad3 pathway, the replication branch (Cds1 and Mrc1) and the DNA damage branch (Chk1 and Crb2) were both partially sensitive to HU and Spd1 accumulation respectively, indicating that both these branches of the Rad3 pathway Genes 2017, 8, 128 7 of 15

GenesSpd1-accumulating2017, 8, 128 cells. Mutants in the two branches of the Rad3 pathway, the replication branch7 of 14 (Cds1 and Mrc1) and the DNA damage branch (Chk1 and Crb2) were both partially sensitive to HU and Spd1 accumulation respectively, indicating that both these branches of the Rad3 pathway can canredundantly redundantly contribute contribute to tolerance to tolerance of imposed of imposed S phase S phaseproblems. problems. Consistent Consistent with this, with we found this, we foundthat thatthe Δ thecds1∆ cds1Δchk1∆ chk1doubledouble mutant mutant was wasas sensitive as sensitive as Δ asrad3∆rad3 to bothto both HU HU treatment treatment and and Spd1 Spd1 accumulation.accumulation. In In agreement agreement withwith aa functionfunction of the Cds1-dependent Cds1-dependent replication replication branch branch in intolerating tolerating Spd1Spd1 accumulation, accumulation, we we found found that that thiaminethiamine addition to to cdt2cdt2TRTR cellscells caused caused spd1spd1-dependent-dependent induction induction ofof Cds1 Cds1 kinase kinase activity activity (see (see below). below).

Figure 3. Checkpoint requirement in Spd1 accumulating cells. Cells of the indicated genotype (all Figure 3. Checkpoint requirement in Spd1 accumulating cells. Cells of the indicated genotype salvage background) were spotted onto MSA plates either without or with thiamine, or plates (all salvage background) were spotted onto MSA plates either without or with thiamine, or plates containing 3 mM hydroxyurea (HU). The plates were incubated at 30 °C for three days. containing 3 mM hydroxyurea (HU). The plates were incubated at 30 ◦C for three days. 3.4. Establishment of a Deoxynucleotide Salvage Pathway in Fission Yeast 3.4. Establishment of a Deoxynucleotide Salvage Pathway in Fission Yeast Spd1 inhibits RNR, and Δspd1 cells have a two-fold elevation of their dNTP pools [13]. We wantedSpd1 to inhibits directly RNR, test andwhether∆spd1 thecells inhibition have a two-fold of DNA elevationreplication of imposed their dNTP by poolsSpd1 [accumulation13]. We wanted tocould directly be testsuppressed whether by the restoring inhibition dNTP of DNAlevels. replication S. pombe does imposed not have by Spd1a deoxynucleotide accumulation salvage could be suppressedpathway for by uptake restoring and dNTP phosphorylation levels. S. pombe of deoxynucdoes notleosides have a(dN). deoxynucleotide We therefore engineered salvage pathway fission for uptakeyeast andcells phosphorylationto express the genes of deoxynucleosides for hENT1 and the (dN). D. melanogaster We therefore DmdNK; engineered both fission from the yeast strong, cells to expressconstitutive the genes adh1 for promoter, hENT1 and theintegratedD. melanogaster at the ura4DmdNK; locus. We both chose from DmdNK, the strong, since constitutive it has broadadh1 promoter,specificity, and and integrated can phosphorylate at the ura4 alllocus. four Wedeoxynucleosides chose DmdNK, [25]. since Clear it hasfluorescence broad specificity, labeling of and cells can phosphorylatefrom accumulation all four of deoxynucleosidesEdU in DNA was observed [25]. Clear in this fluorescence strain when labeling the nucleoside of cells from analogue accumulation EdU was of EdUadded in DNA to the was culture observed medium in (Figure this strain 4A). when the nucleoside analogue EdU was added to the culture mediumWe (Figure next evaluated4A). the functionality of this salvage pathway for unmodified DNA building blocks,We next by evaluatedtesting whether the functionality we could of thisbypass salvage the pathway essential for function unmodified of DNARNR buildingby adding blocks, by testing whether we could bypass the essential function of RNR by adding deoxynucleosides to the culture medium. We first tested whether we could rescue the temperature-sensitive cdc22-M45 allele of the large subunit of RNR. When crossed into the salvage background, growth of this mutant Genes 2017Genes, 8, 2017 128, 8, 128 8 of 15 8 of 14

deoxynucleosides to the culture medium. We first tested whether we could rescue the at the restrictivetemperature-sensitive temperature cdc22-M45 was restored allele of by the addition large subunit of deoxyribonucleosides of RNR. When crossed to into the the culture salvage medium; maximumbackground, rescue wasgrowth observed of this usingmutant a at concentration the restrictive of temperature 2 µM (Figure was4B). restored In addition by addition to cdc22-M45 of , one otherdeoxyribonucleosides temperature-sensitive to the culture allele medium; of cdc22 maximumcalled C11 rescue has beenwas observed isolated using [26]. a We concentration sequenced both of 2 µM (Figure 4B). In addition to cdc22-M45, one other temperature-sensitive allele of cdc22 called alleles; the cdc22-M45 mutant encoded an F518S substitution, while the cdc22-C11 mutant encoded C11 has been isolated [26]. We sequenced both alleles; the cdc22-M45 mutant encoded an F518S a G591Esubstitution, substitution while in the Cdc22. cdc22-C11 When mutant crossed encoded into a the G591E salvage substitution background, in Cdc22. growth When crossed of the intocdc22-C11 mutantthe at thesalvage restrictive background, temperature growth wasof the also cdc22-C11 rescued mutant by addition at the ofrestrictive 2 µM deoxyribonucleosides temperature was also to the culturerescued medium by (Figureaddition4 C),of 2 suggestingµM deoxyribonucleosides that the observed to the rescue culture was medium notallele-specific. (Figure 4C), suggesting Furthermore, for boththat mutants, the observed rescue rescue was dependent was not allele-specifi on the presencec. Furthermore, of the salvage for both pathway mutants, (Figure rescue4C). was Curiously,dependent rescue on the presence of RNR functionof the salvage appeared pathway to (Figure depend 4C). on the two deoxyribonucleosides dA and dC only (FigureCuriously,4C), suggestingrescue of RNR that function currently appeared unidentified to depend cellular on the two deaminase deoxyribonucleosides activities can dA convert and dC only (Figure 4C), suggesting that currently unidentified cellular deaminase activities can dA and dC into dG and dT respectively. Consistent with the existence of such a conversion pathway, convert dA and dC into dG and dT respectively. Consistent with the existence of such a conversion our dNTPpathway, concentration our dNTP measurementsconcentration measurements indicated that indicated the deoxythymidine that the deoxythymidine triphosphate triphosphate (dTTP) pool was elevated(dTTP) pool by extracellular was elevated by provision extracellular ofdA provision and dC of dA in theand culture dC in the medium, culture medium, albeit not albeit to not the same level asto when the same all level four as deoxynucleosides when all four deoxynucleosides (dN) were provided (dN) were (Figure provided4D). (Figure 4D).

Figure 4. Establishment of a salvage pathway in fission yeast. (A) Incorporation of Figure 4. Establishment of a salvage pathway in fission yeast. (A) Incorporation of 5-ethynyl-2’-deoxyuridine (EdU) into cells expressing the salvage pathway. The fluorescence image 5-ethynyl-2’-deoxyuridineshows a cell that incorporated (EdU) into EdU cells into expressingits DNA and the another salvage cell pathway. that did Thenot. ( fluorescenceB) Growth of image showscdc22-M45 a cell that cells incorporated expressing EdUthe salvage into its pathway DNA and at 25 another °C or at cell35 °C. that The did plates not; at ( B35) °C Growth contained of cdc22-M45 the ◦ ◦ ◦ cells expressingindicated concentration the salvage of pathway dN (an equimolar at 25 C mixture or at 35 of C.dA, The dC, platesdG andat dT). 35 (CC) Growth contained of wild the type indicated concentrationcells, or cells of dNharboring (an equimolar two different mixture temperature-sensitive of dA, dC, dG alleles and of dT); cdc22 (C (M45,) Growth C11), either of wild with type or cells, or cellswithout harboring the salvage two differentpathway at, temperature-sensitive respectively, 25 °C or at 35 alleles °C. The of platescdc22 at 35(M45, °C contained C11), either 2 µM of with or withoutthe the indicated salvage deoxynucleoside. pathway at, respectively, (D) Deoxynucleotide 25 ◦C or triphosphate at 35 ◦C. The(dNTP) plates pool at measurements 35 ◦C contained of a 2 µM cdc22-M45 strain expressing the salvage pathway, grown either at 25 °C, or switched to 35 °C for four of the indicated deoxynucleoside; (D) Deoxynucleotide triphosphate (dNTP) pool measurements of hours. When indicated, 4 µM of dN (as defined above) or 4 µM of dA + dC were added to the a cdc22-M45 strain expressing the salvage pathway, grown either at 25 ◦C, or switched to 35 ◦C for cultures at the time of temperature shift up. The relative levels of deoxynucleoside triphosphates four hours.were normalized When indicated, to ATP and 4 µ arbitrarilyM of dN se (ast to defined 1 at 25 °C. above) (nd: not or determined). 4 µM of dA + dC were added to the cultures at the time of temperature shift up. The relative levels of deoxynucleoside triphosphates were normalized3.5. dNTP Salvage to ATP Does and arbitrarilyNot Rescue setSpd1 to Accumulation 1 at 25 ◦C. (nd: not determined).

3.5. dNTP Salvage Does Not Rescue Spd1 Accumulation Having established a functional deoxyribonucleoside salvage pathway, we tested whether salvage could overcome the replication problems caused by Spd1 accumulation when Cdt2 is depleted. In these experiments, we used 2 µM of dN, which was the optimal concentration for rescue of the cdc22 temperature-sensitive mutants. At the time of release, we added dN to the culture medium of Genes 2017, 8, 128 9 of 14

G1-synchronized cdt2TR cells expressing the salvage pathway (Figure5A). Surprisingly, this did not improve the kinetics of S phase progression in cells with repressed cdt2 transcription. Furthermore, salvage did not prevent Cds1 kinase activation following Cdt2 depletion (Figure5B). Since salvage could not overcome the problems caused by Spd1 accumulation, but could rescue the temperature-sensitive cdc22 mutations, we were interested to establish whether the salvage could rescue the DNA damage sensitivity of CRL4Cdt2 defective ∆ddb1 cells. It is proposed that this sensitivity is, in part, caused by cellular dNTP levels being insufficient for repair synthesis [7]. However, as can be seen in Figure5C, addition of 2 µM dN to ∆ddb1 cells expressing the salvage pathway did not reduce sensitivity to the alkylating agent methyl methanesulfonate (MMS). Increasing the concentration of dN above 2 µM appeared to inhibit growth of ∆ddb1 cells (data not shown). Taken together, these results suggested that Cdt2 depletion at S phase causes problems in addition to its inhibition of dNTP synthesis. To substantiate this conclusion, we directly compare the ability of salvage to rescue the checkpoint activation caused by Spd1 accumulation with that invoked by HU mediated inhibition of RNR, which presumably only affects dNTP synthesis. We investigated the ability of salvage to rescue the killing of rad3-TS cells at the restrictive temperature induced by HU addition or cdt2 depletion (Figure5D). Consistent with HU inhibiting RNR only, we found that salvage could improve the survival of the HU-treated rad3 cells substantially (Figure5D, right panels, rows 3 and 4; compare 0 µM and 2 µM dN). Curiously, when performing the experiment investigating the effects of Spd1 accumulation, we found that in the salvage background, deletion of spd1 could no longer rescue the checkpoint requirement of cdt2-depleted cells (Figure5D, second panel, compare rows 3 and 4 with rows 5 and 6). However, when we added 2 µM dN to the salvage strain, rescue by ∆spd1 was restored to a level similar to that observed in cells without the salvage pathway (Figure5D, third panel, rows 3–6). One explanation for this unexpected observation is that the salvage pathway causes an spd1-independent reduction of dNTP pools in cells that can be counteracted by extracellular dN. In any event, as opposed to the case under HU treatment, we did not see any evidence for rescue of Cdt2-depleted spd1+ cells by salvage in this assay (Figure5D, second and third panels, row 3). In conclusion, this difference between HU and Spd1 is consistentGenes 2017 with, 8, 128 Spd1 inhibiting other cellular functions in addition to deoxynucleotide synthesis.10 of 15

Figure 5. Cont.

Figure 5. Salvage does not overcome S phase problems caused by Spd1 accumulation. (A) FACS profiles of G1 synchronized cdt2TR cells expressing the salvage pathway. The cells were released in either the absence of presence of thiamine (to induce Spd1 accumulation). The culture in the right panel shows cells that were supplied with 2 µM of dN in addition to thiamine. (B). cdt2TR or cdt2TR Δspd1 cells expressing the salvage pathway were treated with thiamine and/or 2 µM dN, as indicated. Cells were harvested after four hours, and Cds1 kinase activity against myelin basic protein (MBP) was monitored. Lower panel shows a Western blot of hemagglutinin-tagged Cds1 (HA-Cds1), with immunoglobulin G (IgG) heavy chain serving as a loading control. (C) Cells of the indicated genotypes were spotted on plates containing the indicated concentration of methyl methanesulfonate (MMS). “+ salv” means strains expressing the salvage pathway. Plates in the lower panel contain 2 µM of dN. (D) Strains of the indicated genotypes were spotted on plates with the indicated supplements (thiamine, 3 mM HU, or 2 µM dN) and grown at 30 °C for three days. All strains contain the cdt2TRallele. Strains in the two last rows do not express the salvage pathway.

Genes 2017, 8, 128 10 of 15

Genes 2017, 8, 128 10 of 14

FigureFigure 5. Salvage 5. Salvage does does not not overcome overcome S S phase phase problemsproblems caused caused by by Spd1 Spd1 accumulation. accumulation. (A) FACS (A) FACS TR profilesprofiles of G1 of synchronizedG1 synchronizedcdt2 cdt2TR cells cellsexpressing expressing the the salvage salvage pathway. pathway. The The cells cells were were released released in in eithereither the absence the absence of presence of presence of thiamine of thiamine (to induce(to induce Spd1 Spd1 accumulation). accumulation). The The culture culture in in the the right right panel panel shows cells that were supplied with 2 µM of dN in addition to thiamine. (B). cdt2TR or cdt2TR shows cells that were supplied with 2 µM of dN in addition to thiamine; (B) cdt2TR or cdt2TR ∆spd1 cells Δspd1 cells expressing the salvage pathway were treated with thiamine and/or 2 µM dN, as expressing the salvage pathway were treated with thiamine and/or 2 µM dN, as indicated. Cells were indicated. Cells were harvested after four hours, and Cds1 kinase activity against myelin basic harvestedprotein after (MBP) four was hours, monitored. and Cds1 Lower kinase panel activity shows against a Western myelin blot basic of hemagglutinin-tagged protein (MBP) was monitored.Cds1 Lower(HA-Cds1), panel shows with aimmunoglobulin Western blot of G hemagglutinin-tagged (IgG) heavy chain serving Cds1 as a(HA-Cds1), loading control. with (C immunoglobulin) Cells of the G (IgG)indicated heavy genotypes chain serving wereas spotted a loading on plates control; cont (Caining) Cells the of theindicated indicated concentration genotypes of were methyl spotted on platesmethanesulfonate containing the (MMS). indicated “+ salv” concentration means strains of expr methylessing methanesulfonate the salvage pathway. (MMS). Plates in “+ the salv” lower means strainspanel expressing contain 2 the µM salvage of dN. ( pathway.D) Strains Platesof the indicated in the lower genotypes panel containwere spotted 2 µM on of plates dN; (Dwith) Strains the of the indicatedindicated genotypessupplements were (thiamine, spotted 3 onmM plates HU, or with 2 µM the dN) indicated and grown supplements at 30 °C for (thiamine, three days. 3 mMAll HU, or 2 µstrainsM dN) contain and grown the cdt2 atTR 30allele.◦C forStrains three in days.the two All last strains rows do contain not express the cdt2 theTR salvageallele. pathway. Strains in the two last rows do not express the salvage pathway.

4. Discussion Fission yeast cells defective in CRL4Cdt2 mediated protein ubiquitylation are challenged at S phase because the Spd1 and Spd2 proteins are not degraded. The thiamine-repressible cdt2TR allele described in the present report allowed us to study the immediate effects of Spd1 and Spd2 accumulation. When we switched off CRL4Cdt2 activity, we saw accumulation of cells in the S phase after three hours (Figure1B, first column), indicating a rapid effect on cell cycle progression. Interestingly, cells deleted for either spd1 or spd2 were compromised for cell cycle arrest, showing that Spd1 and Spd2 both contribute to the S phase arrest observed upon Cdt2 depletion (Figure1B, second and third column). However, the double mutants showed better S-phase progression after 5–6 h than the two single mutants (Figure1B, fourth column), indicating that Spd1 and Spd2 on their own can inhibit replication, presumably through a common target. This is consistent with the observation that both Spd1 and Spd2 can block S phase independently of each other when strongly overexpressed [16]. When the ability of G1 synchronized cells to progress through the S phase was scrutinized (Figure2), we obtained a different result. Here, Spd1 accumulation appeared to be absolutely required for blocking replication (Figure2B), whereas Spd2 had a relatively small enhancing effect on the arrest. However, the completion of S phase in ∆spd2 cells was advanced by approximately two hours relative to spd2+ cells (compare Figure2C and Figure2A), again suggesting that Spd2 can inhibit progression through the S phase. One interpretation of these results is that accumulation of both Spd1 and Spd2 can Genes 2017, 8, 128 11 of 14 induce a transient S phase arrest, but maintenance of the arrested state is mostly dependent on Spd1. Presumably, this difference is related to the fact that Spd1 but not Spd2 inhibits dNTP formation [16]. Our analysis of various checkpoint mutants (Figure3) shows that HU exposure and Spd1 accumulation give rise to S phase problems that can be tolerated by both the replication branch (Cds1 and Mrc1) and the DNA damage branch (Chk1 and Crb2) of the Rad3 pathway. Consequently, the ∆cds1 ∆chk1 double mutant is as sensitive as ∆rad3. Presumably, in the absence of Cds1, exposure to HU or induction of Spd1 accumulation causes fork collapse and subsequent need for the Chk1 sub-pathway [27]. Furthermore, loading of the 9-1-1 checkpoint clamp appears to be essential after both types of replication stress. These observations are consistent with Spd1 exerting its function via the inhibition of RNR (similar to HU). To directly test whether elevation of dNTP pools could improve S phase delay in Spd1- and Spd2-accumulating cells, we established a salvage pathway in fission yeast. By using the equilibrative hENT1 transporter, we could define the intracellular pools simply by adding a given level of deoxynucleosides to the culture medium. Furthermore, by applying the DmdNK deoxyribonucleoside kinase, we could salvage all four deoxyribonucleotides [28]. This engineered salvage pathway allowed us to rescue two different temperature-sensitive mutants in the cdc22 gene encoding the essential R1 subunit of RNR (Figure4C). To our knowledge, this is the first example of salvage of an RNR deficiency in any system. However, it is not clear whether salvage can rescue a strain deleted for both RNR subunits, or whether survival somehow relies on residual RNR functions at the restrictive temperature. Finally, salvage appeared to function with only dA and dC, suggesting that these molecules can be converted into dGTP and dTTP by a pathway involving purine/pyrimidine deaminase activities. The fission yeast genome encodes at least five potential deaminase enzymes, but we have not determined whether any of these are required for salvage by dA and dC. Our main goal for establishing the salvage pathway was to investigate if restoring dNTP levels through salvage could circumvent the S phase problems caused by Spd1 and Spd2 accumulation. However, salvage neither improved the slow S phase progression (Figure5A), nor prevented the Cds1 kinase activation (Figure5B) observed in Cdt2 depleted cells. Moreover, salvage did not suppress the damage sensitivity of CRL4Cdt2 defective ∆ddb1 cells (Figure5C). Strikingly, whereas salvage clearly suppressed the killing of rad3 cells exposed to the RNR inhibiting drug HU, it did not improve the survival of Cdt2-depleted rad3 cells (Figure5D). We interpret this observation as evidence for dNTP synthesis-independent inhibition of replication by Spd1. Such an effect likely occurs through interactions with other protein targets, a scenario that is linked to the IDP properties of Spd1, allowing for multi-valency during interactions [29]. But what is this other target of Spd1? Human p21 is a CRL4Cdt2-targeted IDP that can inhibit replication by binding to PCNA [4], and heterologous expression of p21 causes checkpoint activation at PCNA in fission yeast [30]. We propose that Spd1, together with Spd2, can similarly inhibit progression of the replication fork by binding to PCNA (Figure6). Spd1 can bind to PCNA [ 6] (Figure1D), and Spd2 is most similar to Spd1 in the HUG domain containing the PIP degron that binds to PCNA [16]. Furthermore, Spd2 also appears to bind PCNA in vitro (B.K.B., data not shown). Hence, when overexpressed from the strong nmt1 promoter, both Spd1 (Figure6B) and Spd2 (Figure6C) can block replication independent of each other [16]. However, when CRL4Cdt2 is downregulated, we speculate that Spd1 and Spd2 accumulate to an intermediate level, such that both proteins are required for inhibition of PCNA (Figure6D). Deletion of spd1 hence relieves inhibition from both RNR and PCNA, and therefore suppresses the checkpoint activation caused by CRL4Cdt2 inactivation (Figure6E). On the other hand, when spd2 is deleted, it is only the PCNA inhibition that is relieved; Spd1 will still inhibit RNR (Figure6F). Consequently, it is also necessary to elevate dNTP pools by means of the cdc22-D57N mutation in order to suppress checkpoint activation in CRL4Cdt2 defective ∆spd2 cells [16] (Figure6G). At this stage it is unclear whether Spd1 and Spd2 cause checkpoint activation merely by binding to PCNA, or whether they perturb the recruitment of other replication factors. Resolving this issue will require detailed studies of the interactions between Spd1, Spd2 and PCNA. Genes 2017, 8, 128 12 of 15

occurs through interactions with other protein targets, a scenario that is linked to the IDP properties of Spd1, allowing for multi-valency during interactions [29]. But what is this other target of Spd1? Human p21 is a CRL4Cdt2-targeted IDP that can inhibit replication by binding to PCNA [4], and heterologous expression of p21 causes checkpoint activation at PCNA in fission yeast [30]. We propose that Spd1, together with Spd2, can similarly inhibit progression of the replication fork by binding to PCNA (Figure 6). Spd1 can bind to PCNA [6] (Figure 1D), and Spd2 is most similar to Spd1 in the HUG domain containing the PIP degron that binds to PCNA [16]. Furthermore, Spd2 also appears to bind PCNA in vitro (B.K.B., data not shown). Hence, when overexpressed from the strong nmt1 promoter, both Spd1 (Figure 6B) and Spd2 (Figure 6C) can block replication independent of each other [16]. However, when CRL4Cdt2 is downregulated, we speculate that Spd1 and Spd2 accumulate to an intermediate level, such that both proteins are required for inhibition of PCNA (Figure 6D). Deletion of spd1 hence relieves inhibition from both RNR and PCNA, and therefore suppresses the checkpoint activation caused by CRL4Cdt2 inactivation (Figure 6E). On the other hand, when spd2 is deleted, it is only the PCNA inhibition that is relieved; Spd1 will still inhibit RNR (Figure 6F). Consequently, it is also necessary to elevate dNTP pools by means of the cdc22-D57N mutation in order to suppress checkpoint activation in CRL4Cdt2 defective Δspd2 cells [16] (Figure 6G). At this stage it is unclear whether Spd1 and Spd2 cause checkpoint activation merely by binding to PCNA, or whether they perturb the recruitment of other replication factors. Resolving this issue will require detailed studies of the Genes 2017, 8, 128 12 of 14 interactions between Spd1, Spd2 and PCNA.

FigureFigure 6. 6.Model Model for for inhibition inhibition of of DNA DNA replication replication by by Spd1 Spd1 and and Spd2. Spd2. Spd1 Spd1 (red (red squares) squares) can can inhibit inhibit ribonucleotideribonucleotide reductasereductase (RNR,(RNR, greengreen circles),circles), whilewhileboth bothSpd1 Spd1 and and Spd2 Spd2 (blue (blue squares) squares) can can inhibit inhibit replicationreplication by by binding binding to to proliferating proliferating cell cell nuclear nuclear antigen antigen (PCNA, (PCNA, yellow yellow rings). rings). S S phase phase is is inhibited inhibited ifif at at least least one one of of these these two two events events occurs. occurs. ((AA)) InIn wildwild typetype cells,cells, bothboth Spd1Spd1 andand Spd2Spd2 areare degraded,degraded, andand hencehence dNTPdNTP productionproduction andand elongationelongation areare notnot inhibited;inhibited. (B) When Spd1 is is overexpressed overexpressed ( (↑↑),), itit inhibitsinhibits bothboth processes;processes. (C(C)) InIn cellscells overexpressingoverexpressing Spd2,Spd2, onlyonly PCNAPCNA isis inhibited;inhibited. (D(D)) InIn Cdt2Cdt2 depleteddepleted cellscells (↓(↓),), Spd1 Spd1 and and Spd2 Spd2 accumulate accumulate to to a a moderate moderate level. level. Spd1 Spd1 inhibits inhibits RNR, RNR, while while both both Spd1Spd1 and and Spd2 Spd2 are are required required to to raise raise the the concentration concentration to to a alevel level wherewhere inhibitioninhibition ofof PCNAPCNA cancan occur;occur. (E(E)) Deletion Deletion of ofspd1 spd1relieves relieves both both types types of of repression; repression. ( F(F)) When Whenspd2 spd2is is deleted deleted in in Cdt2-depleted Cdt2-depleted cells, cells, repressionrepression of of PCNA PCNA is is lifted, lifted, but but Spd1 Spd1 still still inhibits inhibits RNR; RNR. ( G(G)) The Thecdc22-D57N cdc22-D57Nmutation mutation changes changes the the RNR configuration (green square) so that it can no longer be inhibited by dATP through negative feedback. Hence, sufficient amounts of dNTPs are formed even in the presence of Spd1 inhibition.

Supplementary Materials: The following are available online at www.mdpi.com/2073-4425/8/5/128/s1. Table S1: Strain list. Acknowledgments: We thank Ken Sawin, Stephen Kearsey and Nick Rhind for the strains provided. Essa Alanazi is thanked for assistance in pilot experiments addressing the salvage pathway in ∆ddb1 background and Michaela Lederer for technical assistance. This work was supported by grants from the Danish Research Council 4481-00344 (to O.N. and B.B.K.), the Villum Foundation 11407 (to C.H. and O.N.), the AICR (to A.M.C. and B.B.K.), the North West Cancer Research (O.F.) and the National Institute of Social Care and Health Research-Cancer Genetics Biomedical Research Unit (O.F.). Author Contributions: O.F., E.H., B.B.K., A.M.C., C.H. and O.N. conceived and designed the experiments; O.F., U.F., K.V.L., M.-F.U.G., I.N.M., C.H. and O.N. performed the experiments; O.F., C.H. and O.N. analyzed the data; O.N. wrote the paper. Conflicts of Interest: The authors declare no conflict of interest.

References

1. Choe, K.N.; Moldovan, G.L. Forging Ahead through Darkness: PCNA, Still the Principal Conductor at the Replication Fork. Mol. Cell. 2017, 65, 380–392. [CrossRef][PubMed] Genes 2017, 8, 128 13 of 14

2. Stodola, J.L.; Burgers, P.M. Resolving individual steps of Okazaki-fragment maturation at a millisecond timescale. Nat. Struct. Mol. Biol. 2016, 23, 402–408. [CrossRef][PubMed] 3. Havens, C.G.; Walter, J.C. Mechanism of CRL4Cdt2, a PCNA-dependent E3 ubiquitin ligase. Genes Dev. 2011, 25, 1568–1582. [CrossRef][PubMed] 4. Waga, S.; Hannon, G.J.; Beach, D.; Stillman, B. The p21 inhibitor of cyclin-dependent kinases controls DNA replication by interaction with PCNA. Nature 1994, 369, 574–578. [CrossRef][PubMed] 5. Liu, C.; Powell, K.A.; Mundt, K.; Wu, L.; Carr, A.M.; Caspari, T. Cop9/signalosome subunits and Pcu4 regulate ribonucleotide reductase by both checkpoint-dependent and -independent mechanisms. Genes Dev. 2003, 17, 1130–1140. [CrossRef][PubMed] 6. Salguero, I.; Guarino, E.; Shepherd, M.E.; Deegan, T.D.; Havens, C.G.; MacNeill, S.A.; Walter, J.C.; Kearsey, S.E. Ribonucleotide reductase activity is coupled to DNA synthesis via proliferating cell nuclear antigen. Curr. Biol. 2012, 22, 720–726. [CrossRef][PubMed] 7. Holmberg, C.; Fleck, O.; Hansen, H.A.; Liu, C.; Slaaby, R.; Carr, A.M.; Nielsen, O. Ddb1 controls genome stability and meiosis in fission yeast. Genes Dev. 2005, 19, 853–862. [CrossRef][PubMed] 8. Nestoras, K.; Mohammed, A.H.; Schreurs, A.S.; Fleck, O.; Watson, A.T.; Poitelea, M.; O’Shea, C.; Chahwan, C.; Holmberg, C.; Kragelund, B.B.; et al. Regulation of ribonucleotide reductase by Spd1 involves multiple mechanisms. Genes Dev. 2010, 24, 1145–1159. [CrossRef][PubMed] 9. Liu, C.; Poitelea, M.; Watson, A.; Yoshida, S.H.; Shimoda, C.; Holmberg, C.; Nielsen, O.; Carr, A.M. Transactivation of Schizosaccharomyces pombe cdt2+ stimulates a Pcu4-Ddb1-CSN ubiquitin ligase. EMBO J. 2005, 24, 3940–3951. [CrossRef][PubMed] 10. Holmberg, C.; Nielsen, O. Replication: DNA building block synthesis on demand. Curr. Biol. 2012, 22, R271–R272. [CrossRef][PubMed] 11. Woollard, A.; Basi, G.; Nurse, P. A novel S phase inhibitor in fission yeast. EMBO J. 1996, 15, 4603–4612. [PubMed] 12. Moss, J.; Tinline-Purvis, H.; Walker, C.A.; Folkes, L.K.; Stratford, M.R.; Hayles, J.; Hoe, K.L.; Kim, D.U.; Park, H.O.; Kearsey, S.E.; et al. Break-induced ATR and Ddb1-Cul4Cdt2 ubiquitin ligase-dependent nucleotide synthesis promotes homologous recombination repair in fission yeast. Genes Dev. 2010, 24, 2705–2716. [CrossRef][PubMed] 13. Fleck, O.; Vejrup-Hansen, R.; Watson, A.; Carr, A.M.; Nielsen, O.; Holmberg, C. Spd1 accumulation causes genome instability independently of ribonucleotide reductase activity but functions to protect the genome when deoxynucleotide pools are elevated. J. Cell Sci. 2013, 126, 4985–4994. [CrossRef][PubMed] 14. Hakansson, P.; Dahl, L.; Chilkova, O.; Domkin, V.; Thelander, L. The Schizosaccharomyces pombe replication inhibitor Spd1 regulates ribonucleotide reductase activity and dNTPs by binding to the large Cdc22 subunit. J. Biol. Chem. 2006, 281, 1778–1783. [CrossRef][PubMed] 15. Yoshida, S.H.; Al-Amodi, H.; Nakamura, T.; McInerny, C.J.; Shimoda, C. The Schizosaccharomyces pombe cdt2+ gene, a target of G1-S phase-specific transcription factor complex DSC1, is required for mitotic and premeiotic DNA replication. Genetics 2003, 164, 881–893. [PubMed] 16. Vejrup-Hansen, R.; Fleck, O.; Landvad, K.; Fahnoe, U.; Broendum, S.S.; Schreurs, A.S.; Kragelund, B.B.; Carr, A.M.; Holmberg, C.; Nielsen, O. Spd2 assists Spd1 in the modulation of ribonucleotide reductase architecture but does not regulate deoxynucleotide pools. J. Cell Sci. 2014, 127, 2460–2470. [CrossRef] [PubMed] 17. Moreno, S.; Klar, A.; Nurse, P. Molecular genetic analysis of fission yeast Schizosaccharomyces pombe. Methods Enzymol. 1991, 194, 795–823. [PubMed] 18. Matsuyama, A.; Shirai, A.; Yashiroda, Y.; Kamata, A.; Horinouchi, S.; Yoshida, M. pDUAL, a multipurpose, multicopy vector capable of chromosomal integration in fission yeast. Yeast 2004, 21, 1289–1305. [CrossRef] [PubMed] 19. Vernis, L.; Piskur, J.; Diffley, J.F. Reconstitution of an efficient thymidine salvage pathway in Saccharomyces cerevisiae. Nucleic Acids Res. 2003, 31, e120. [CrossRef][PubMed] 20. Sivakumar, S.; Porter-Goff, M.; Patel, P.K.; Benoit, K.; Rhind, N. In vivo labeling of fission yeast DNA with thymidine and thymidine analogs. Methods 2004, 33, 213–219. [CrossRef][PubMed] 21. Egel, R.; Willer, M.; Kjaerulff, S.; Davey, J.; Nielsen, O. Assessment of pheromone production and response in fission yeast by a halo test of induced sporulation. Yeast 1994, 10, 1347–1354. [CrossRef][PubMed] Genes 2017, 8, 128 14 of 14

22. Nielsen, O. Synchronization of S Phase in Schizosaccharomyces pombe Cells by Transient Exposure to M-Factor Pheromone. Cold Spring Harb. Protoc. 2016.[CrossRef][PubMed] 23. Hua, H.; Kearsey, S.E. Monitoring DNA replication in fission yeast by incorporation of 5-ethynyl-2’-deoxyuridine. Nucleic Acids Res. 2011, 39, e60. [CrossRef][PubMed] 24. Basi, G.; Schmid, E.; Maundrell, K. TATA box mutations in the Schizosaccharomyces pombe nmt1 promoter affect transcription efficiency but not the transcription start point or thiamine repressibility. Gene 1993, 123, 131–136. [CrossRef] 25. Munch-Petersen, B.; Knecht, W.; Lenz, C.; Sondergaard, L.; Piskur, J. Functional expression of a multisubstrate deoxyribonucleoside kinase from Drosophila melanogaster and its C-terminal deletion mutants. J. Biol. Chem. 2000, 275, 6673–6679. [CrossRef][PubMed] 26. Gordon, C.B.; Fantes, P.A. The cdc22 gene of Schizosaccharomyces pombe encodes a cell cycle-regulated transcript. EMBO J. 1986, 5, 2981–2985. [PubMed] 27. Lindsay, H.D.; Griffiths, D.J.; Edwards, R.J.; Christensen, P.U.; Murray, J.M.; Osman, F.; Walworth, N.; Carr, A.M. S-phase-specific activation of Cds1 kinase defines a subpathway of the checkpoint response in Schizosaccharomyces pombe. Genes Dev. 1998, 12, 382–395. [CrossRef][PubMed] 28. Munch-Petersen, B.; Piskur, J.; Sondergaard, L. Four deoxynucleoside kinase activities from Drosophila melanogaster are contained within a single monomeric enzyme, a new multifunctional deoxynucleoside kinase. J. Biol. Chem. 1998, 273, 3926–3931. [CrossRef][PubMed] 29. Milles, S.; Lemke, E.A. Mapping multivalency and differential affinities within large intrinsically disordered protein complexes with segmental motion analysis. Angew. Chem. 2014, 53, 7364–7367. [CrossRef][PubMed] 30. Tournier, S.; Leroy, D.; Goubin, F.; Ducommun, B.; Hyams, J.S. Heterologous expression of the human cyclin-dependent kinase inhibitor p21Cip1 in the fission yeast, Schizosaccharomyces pombe reveals a role for PCNA in the chk1+ cell cycle checkpoint pathway. Mol. Biol. Cell 1996, 7, 651–662. [CrossRef][PubMed]

© 2017 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/). Appendix – Supplementary Figures

…………………………………………………………………………………………………………………………………. Figure S1. Both human RNR subunits are required for complementing lack of yeast Cdc22 Viability assay. 10-fold serial dilutions were spotted onto YES medium with and without thiamine and incubated for 4 days at the permissive (25°C) and restrictive (35°C) temperatures. Human R1 can complement for lack of Cdc22 only when human R2 is expressed from the leu1 locus (in the absence of thiamine). This indicates that a functional enzyme cannot be formed with human R1 and yeast Suc22 subunits, and thus the cells require both human RNR subunits for survival. Performed by H. B. Nielsen (Nielsen, 2013). ………………………………………………………………………………………………………………………………….

…………………………………………………………………………………………………………………………………. Figure S2. Expression of human R2 is turned off in the presence of thiamine Western blot. Human R1 and R2 were detected using anti-hR1 and anti-hR2 antibodies (1:1000). Both subunits were detected in hR1-R2 cells (lane 2). In cells with hR1 and a nmt1-hR2 construct expressed from the leu1 locus, both subunits were detected in the absence of thiamine (promoter on). In contrast, the presence of thiamine (promoter off) completely repressed expression of hR2. Tubulin serves as loading control. Performed by H. B. Nielsen (Nielsen, 2013). ………………………………………………………………………………………………………………………………….

113 …………………………………………………………………………………………………………………………………. Figure S3. Autoradiograph for determination of dATP pool DNA products from the primer extension assay were separated on a polyacrylamide-urea gel as described in Materials and Methods. Densitometric analysis was performed using TotalLab Quant software and calculations of dATP pools were based on band intensities as well as the lengths of the separated products (formulas are described in Materials and Methods). A dATP standard (dA0-dA8) was included for calibration. ………………………………………………………………………………………………………………………………….

114 ……………………………………………………………………………………………………………………….. Figure S4. Alignments of the amino acid sequences of fission yeast Suc22, human R2, and chimR2 From (Eidesgaard, 2017) ………………………………………………………………………………………………………………………..

115 ………………………………………………………………………………………………………………………………………. Figure S5. Intensity dot plots Complete set of intensity dot plots. Wild type (CFP-positive) and hR1-p53R2 (mCherry-positive) populations were co-cultured and aliquots were either treated with 0.4 mM HU or left untreated. Samples were taken every 2 hours and analyzed using the Xcyto® Quantitative Cell Imager (ChemoMetec). The fluorescence intensity of individual cells in the mCherry channel (x-axis) and the CFP channel (y-axis) is depicted at different time points. Each dot plot is based on ∼1000 live cell counts. ……………………………………………………………………………………………………………………………………...

116