ribonucleotide reductase and cancer: biological...

11
REVIEW Ribonucleotide reductase and cancer: biological mechanisms and targeted therapies Y Aye 1,2 , M Li 3 , MJC Long 4 and RS Weiss 3 Accurate DNA replication and repair is essential for proper development, growth and tumor-free survival in all multicellular organisms. A key requirement for the maintenance of genomic integrity is the availability of adequate and balanced pools of deoxyribonucleoside triphosphates (dNTPs), the building blocks of DNA. Notably, dNTP pool alterations lead to genomic instability and have been linked to multiple human diseases, including mitochondrial disorders, susceptibility to viral infection and cancer. In this review, we discuss how a key regulator of dNTP biosynthesis in mammals, the enzyme ribonucleotide reductase (RNR), impacts cancer susceptibility and serves as a target for anti-cancer therapies. Because RNR-regulated dNTP production can inuence DNA replication delity while also supporting genome-protecting DNA repair, RNR has complex and stage-specic roles in carcinogenesis. Nevertheless, cancer cells are dependent on RNR for de novo dNTP biosynthesis. Therefore, elevated RNR expression is a characteristic of many cancers, and an array of mechanistically distinct RNR inhibitors serve as effective agents for cancer treatment. The dNTP metabolism machinery, including RNR, has been exploited for therapeutic benet for decades and remains an important target for cancer drug development. Oncogene (2015) 34, 20112021; doi:10.1038/onc.2014.155; published online 9 June 2014 DEOXYRIBONUCLEOSIDE TRIPHOSPHATE (dNTP) POOLS AND GENOMIC INTEGRITY The importance of ribonucleotide reductase (RNR) for genome maintenance relates to its central role in regulating dNTP levels. In mammalian cells, total dNTP pool sizes peak during S-phase to support nuclear DNA (nDNA) replication and are roughly 10-fold lower in G 0 /G 1 , when dNTPs are needed for DNA repair and mitochondrial DNA (mtDNA) synthesis. 1,2 As DNA polymerase substrates, dNTPs inuence several aspects of the replication program, including origin choice, fork speed, inter-origin distance, and dormant origin usage. 35 Failure of cells to maintain appropriate dNTP concentrations can be highly detrimental, leading to DNA breaks, mutagenesis and cell death. 6 During cancer development, uncoordinated cell proliferation can lead to insufcient dNTPs that cause replication stress and further promote genomic instability. 7,8 Conversely, elevated dNTP pools also contribute to increased mutagenesis. 1,9,10 Imbalanced dNTP pools enhance mutagenesis mainly by DNA misinsertion and impaired proofreading. 6,9 DNA misinsertion can result from competition between dNTPs for pairing with the template base, and a dNTP present in excess can be readily misincorporated. DNA polymerase proofreading is reduced in the presence of elevated dNTP concentrations through a phenomenon known as the next-nucleotide effect. 11 Under such conditions, DNA chain extension following a base misinsertion proceeds before the mismatched nucleotide can be removed. 1214 dNTP imbalances can also stimulate frameshift mutations by facilitating correct base-pairing following template slippage or misalignment. 15 RNR AND dNTP BIOSYNTHESIS dNTPs can be generated through de novo and salvage pathways. In mammals, RNR catalyzes the rate-limiting step of the de novo pathway, reducing the 2carbon of a ribonucleoside diphosphate (NDP) to produce the corresponding deoxy (d)NDP. Subsequently, dNDPs are phosphorylated by nucleoside diphosphate kinase (NDPK) yielding dNTPs. 16 Importantly, the de novo biosynthesis of deoxythymidine triphosphate (dTTP) is much more complex, requiring the conversion of deoxyuridine monophosphate to deoxythymidine monophosphate (dTMP) by thymidylate synthase followed by phosphorylation by thymidylate phosphate kinase (TMPK) and then NDPK. Deoxyuridine monophosphate is generated either from deoxyuridine triphosphate (dUTP) or deoxycytidine monophosphate, the biosynthesis of both of which is dependent on RNR-catalyzed reduction of the corresponding nucleoside diphosphates. 6 RNR enzymes are present in all organisms and feature-conserved radical-mediated nucleotide reduction. 17 Mam- malian RNRs, the focus of this review, consist of two subunits α and β, that associate to form the holoenzyme (Figure 1). α contains the catalytic (C) site and two different allosteric sites. β harbors a di-iron cofactor and tyrosyl radical (Y) essential for RNR activity. Evidence suggests that catalytically active RNR is minimally an α 2 β 2 complex. 1820 RNR oligomerization is inuenced by (d)NTP/ATP binding to α allosteric sites. The negative regulator dATP induces inactive α 6 (or α 6 β 2 ) states, highlighting a role for higher-order oligomeric associations as a regulatory mechanism for RNR. 18,2124 However, as discussed further below, the precise nature of RNR oligomeric states in the presence of allosteric activators remains unsettled (Supplementary Table S1). 1 Department of Chemistry and Chemical Biology, Cornell University, Ithaca, NY, USA; 2 Department of Biochemistry, Weill Cornell Medical College, New York, NY, USA; 3 Department of Biomedical Sciences, Cornell University, Ithaca, NY, USA and 4 Graduate Program in Biochemistry, Brandeis University, Waltham, MA, USA. Correspondence: Assistant Professor Y Aye, 224 Baker Laboratory of Chemistry & Chemical Biology, Ithaca, NY 14853, USA or Associate Professor RS Weiss, Department of Biomedical Sciences, Cornell University, Veterinary Research Tower, Ithaca, NY 14853, USA. E-mail: [email protected] or [email protected] Received 6 January 2014; revised 25 April 2014; accepted 26 April 2014; published online 9 June 2014 Oncogene (2015) 34, 2011 2021 © 2015 Macmillan Publishers Limited All rights reserved 0950-9232/15 www.nature.com/onc

Upload: trinhnguyet

Post on 29-Mar-2018

222 views

Category:

Documents


2 download

TRANSCRIPT

REVIEW

Ribonucleotide reductase and cancer: biological mechanismsand targeted therapiesY Aye1,2, M Li3, MJC Long4 and RS Weiss3

Accurate DNA replication and repair is essential for proper development, growth and tumor-free survival in all multicellularorganisms. A key requirement for the maintenance of genomic integrity is the availability of adequate and balanced pools ofdeoxyribonucleoside triphosphates (dNTPs), the building blocks of DNA. Notably, dNTP pool alterations lead to genomic instabilityand have been linked to multiple human diseases, including mitochondrial disorders, susceptibility to viral infection and cancer.In this review, we discuss how a key regulator of dNTP biosynthesis in mammals, the enzyme ribonucleotide reductase (RNR),impacts cancer susceptibility and serves as a target for anti-cancer therapies. Because RNR-regulated dNTP production caninfluence DNA replication fidelity while also supporting genome-protecting DNA repair, RNR has complex and stage-specific roles incarcinogenesis. Nevertheless, cancer cells are dependent on RNR for de novo dNTP biosynthesis. Therefore, elevated RNR expressionis a characteristic of many cancers, and an array of mechanistically distinct RNR inhibitors serve as effective agents for cancertreatment. The dNTP metabolism machinery, including RNR, has been exploited for therapeutic benefit for decades and remains animportant target for cancer drug development.

Oncogene (2015) 34, 2011–2021; doi:10.1038/onc.2014.155; published online 9 June 2014

DEOXYRIBONUCLEOSIDE TRIPHOSPHATE (dNTP) POOLS ANDGENOMIC INTEGRITYThe importance of ribonucleotide reductase (RNR) for genomemaintenance relates to its central role in regulating dNTP levels. Inmammalian cells, total dNTP pool sizes peak during S-phase tosupport nuclear DNA (nDNA) replication and are roughly 10-foldlower in G0/G1, when dNTPs are needed for DNA repair andmitochondrial DNA (mtDNA) synthesis.1,2 As DNA polymerasesubstrates, dNTPs influence several aspects of the replicationprogram, including origin choice, fork speed, inter-origin distance,and dormant origin usage.3–5 Failure of cells to maintainappropriate dNTP concentrations can be highly detrimental,leading to DNA breaks, mutagenesis and cell death.6 Duringcancer development, uncoordinated cell proliferation can lead toinsufficient dNTPs that cause replication stress and furtherpromote genomic instability.7,8 Conversely, elevated dNTP poolsalso contribute to increased mutagenesis.1,9,10

Imbalanced dNTP pools enhance mutagenesis mainly by DNAmisinsertion and impaired proofreading.6,9 DNA misinsertion canresult from competition between dNTPs for pairing with thetemplate base, and a dNTP present in excess can be readilymisincorporated. DNA polymerase proofreading is reduced inthe presence of elevated dNTP concentrations through aphenomenon known as the next-nucleotide effect.11 Under suchconditions, DNA chain extension following a base misinsertionproceeds before the mismatched nucleotide can be removed.12–14

dNTP imbalances can also stimulate frameshift mutations byfacilitating correct base-pairing following template slippage ormisalignment.15

RNR AND dNTP BIOSYNTHESISdNTPs can be generated through de novo and salvage pathways. Inmammals, RNR catalyzes the rate-limiting step of the de novopathway, reducing the 2′ carbon of a ribonucleoside diphosphate(NDP) to produce the corresponding deoxy (d)NDP. Subsequently,dNDPs are phosphorylated by nucleoside diphosphate kinase(NDPK) yielding dNTPs.16 Importantly, the de novo biosynthesis ofdeoxythymidine triphosphate (dTTP) is much more complex,requiring the conversion of deoxyuridine monophosphate todeoxythymidine monophosphate (dTMP) by thymidylate synthasefollowed by phosphorylation by thymidylate phosphate kinase(TMPK) and then NDPK. Deoxyuridine monophosphate is generatedeither from deoxyuridine triphosphate (dUTP) or deoxycytidinemonophosphate, the biosynthesis of both of which is dependenton RNR-catalyzed reduction of the corresponding nucleosidediphosphates.6 RNR enzymes are present in all organisms andfeature-conserved radical-mediated nucleotide reduction.17 Mam-malian RNRs, the focus of this review, consist of two subunits α andβ, that associate to form the holoenzyme (Figure 1). α contains thecatalytic (C) site and two different allosteric sites. β harbors a di-ironcofactor and tyrosyl radical (Y•) essential for RNR activity. Evidencesuggests that catalytically active RNR is minimally an α2β2complex.18–20 RNR oligomerization is influenced by (d)NTP/ATPbinding to α allosteric sites. The negative regulator dATP inducesinactive α6 (or α6β2) states, highlighting a role for higher-orderoligomeric associations as a regulatory mechanism for RNR.18,21–24

However, as discussed further below, the precise nature of RNRoligomeric states in the presence of allosteric activators remainsunsettled (Supplementary Table S1).

1Department of Chemistry and Chemical Biology, Cornell University, Ithaca, NY, USA; 2Department of Biochemistry, Weill Cornell Medical College, New York, NY, USA;3Department of Biomedical Sciences, Cornell University, Ithaca, NY, USA and 4Graduate Program in Biochemistry, Brandeis University, Waltham, MA, USA. Correspondence:Assistant Professor Y Aye, 224 Baker Laboratory of Chemistry & Chemical Biology, Ithaca, NY 14853, USA or Associate Professor RS Weiss, Department of Biomedical Sciences,Cornell University, Veterinary Research Tower, Ithaca, NY 14853, USA.E-mail: [email protected] or [email protected] 6 January 2014; revised 25 April 2014; accepted 26 April 2014; published online 9 June 2014

Oncogene (2015) 34, 2011–2021© 2015 Macmillan Publishers Limited All rights reserved 0950-9232/15

www.nature.com/onc

NDP reduction to dNDP takes place in the α C site and requiresthe unpaired electron initially localized as Y• in β (Figure 1).19

Although the di-iron center is present in each βmonomer, a singleY• is generated per β2.17 Protein radical formation occurs viareaction of the diferrous ions with molecular oxygen in a multistepprocess.17,19 How diferrous ions are loaded into β2, includingpotential involvement of iron chaperones, remains unknown.In vitro, β2 can exist in three distinct states: apo-β2 lacks both thedi-iron center and Y•; met-β2 contains the di-iron center but withY• reduced; and holo-β2 houses a diferric–Y• cofactor (the‘metallocofactor’) and is the only state that, when complexedwith α2, is active (Supplementary Figure S1). The β Y• is transferredthrough a proton-coupled electron transfer chain to generate atransient thiyl radical (S•) within the α C site.19 S• initiatesreduction of the NDP ribose ring, ultimately generating dNDPand resulting in disulfide bond formation on α. Thioredoxin orglutaredoxin re-reduces the oxidized cysteines on α, facilitatingturnover.25

In mammals, α is encoded by the gene Rrm1, whereas twodifferent genes, Rrm2 and p53R2, encode distinct β isoforms.p53R2 was identified as a p53-inducible, DNA damage-responsivegene.26 Mouse p53R2 and RRM2 share 81% identity and domainconservation except for amino-terminal residues required for cellcycle stage-specific RRM2 degradation.27–29 RRM1–RRM2 holoen-zyme provides dNTPs for S-phase nDNA replication and repair inproliferating cells, whereas RRM1–p53R2 contributes dNTPs fornDNA repair in quiescent cells as well as mtDNA replication and

repair.30–33 dNTPs for mtDNA synthesis can also come fromsalvage pathways localized within mitochondria,34,35 and RNRactivity has been identified within mitochondria as well.36

In humans, p53R2 mutations result in mitochondrial disease,including mtDNA depletion syndrome, a lethal condition inwhich patients have only 1–4% residual mtDNA in muscle;30,37

mitochondrial neurogastrointestinal encephalopathy, a neuro-degenerative disorder associated with mtDNA depletion;38

and autosomal-dominant progressive external ophthalmoplegia,characterized by accumulation of multiple mtDNA deletions inpostmitotic tissues.39 Furthermore, p53R2 knockout mice showmtDNA depletion and die due to renal failure.30,40 Together, thesefindings indicate that the de novo dNTP biosynthesis mediatedby RNR is essential to maintain not only nuclear but alsomitochondrial genome integrity.

RNR ACTIVITY REGULATIONThe critical importance of dNTP levels demands that RNR istightly regulated through several distinct mechanisms, includingallosteric and oligomeric regulation, as well as alterations of thelevel and localization of RNR subunits. Allosteric modulationinvolves effector binding to two separate allosteric sites in α, aspecificity (S) site that regulates substrate choice and an activity(A) site that regulates enzyme activity (Figure 1).29 Substratespecificity is determined by the binding of ATP, dATP, dTTP ordGTP (deoxyguanosine triphosphate) to the S site; the mechanismhas been extensively reviewed.29,41 Overall RNR activity iscontrolled by (d)ATP binding to the A site, with elevated dATP:ATP ratios associated with inhibition. α hexamerization is coupledto dATP-induced inhibition.21,22,24,41

A second major mechanism for RNR regulation occurs throughthe cell cycle stage-specific control of RNR protein levels. Inmammals, RNR activity is induced during S phase.42–44 Rrm1 geneexpression is mainly regulated transcriptionally, being negligiblein G0/G1 and peaking in S phase.45 RRM1 protein, however,remains constant throughout the cell cycle owing to its long half-life.42,46,47 RRM2 is regulated both transcriptionally and by proteindegradation. Similar to that of Rrm1, Rrm2 transcription is minimalin G0/G1 and peaks in S phase.48 Unlike RRM1, RRM2 abundancecorrelates with its mRNA level.42,44 Upon S phase exit, RRM2protein is degraded through two ubiquitin ligases, the Skp1/Cullin/F-box (SCF) complex and the anaphase-promoting complex(APC).27,28 During G2 phase, RRM2 is recognized by cyclin F, an SCFubiquitin ligase F-box protein, and targeted for degradation.28

During mitosis/G1 phase, RRM2 is degraded by the Cdh1-APCcomplex that recognizes a KEN box motif at the RRM2N-terminus.27 By contrast, p53R2, which lacks a KEN box, iscontinuously expressed throughout the cell cycle. After DNAdamage, p53R2 is transcriptionally induced in a p53-dependentmanner and with RRM1 forms an active holoenzyme, providingdNTPs for DNA repair.26,29,33

RNR subcellular localization affords an additional layer ofregulatory control, although this aspect of mammalian RNRbiology remains controversial. The bulk of RRM1 and RRM2constitutively localize to the cytoplasm and produce dNTPs thatdiffuse into the nucleus for DNA replication.49,50 However, there isan increasing evidence that RRM1 and RRM2 accumulate at DNAdamage sites in the nucleus.28,51–53 This accumulation isdependent on interaction between RRM1 and the DNA damageresponse protein Tip60.51,52

The work forging a link between RNR translocation and localdNTP production at DNA damage sites is appealing. However, acautionary note is required, as dNTP synthesis requires multipleenzymes. The reports linking DNA damage to translocation ofdNTP production machinery have focused on only one or two keyplayers, typically RNR subunits. It is unknown whether the wholedNTP production pathway responds en masse to DNA damage or

Figure 1. RNR and de novo dNTP biosynthesis. Mammalian RNRenzymes function to reduce the 2' carbon of NDPs to generatedNDPs that are subsequently phosphorylated by nucleoside dipho-sphate kinase, yielding dNTPs for nuclear and mitochondrial DNAreplication and repair. RNR consists of an α subunit encoded by Rrm1and a β subunit encoded by Rrm2 or p53R2. The subunits interact toform hetero-oligomers, with the active enzyme believed to adoptα2β2 quaternary state. α contains the catalytic site (C), an activity site(A) that governs overall RNR activity through interactions with dATP(inhibitory) or ATP (stimulatory) and a specificity site (S) thatdetermines substrate choice. Each β subunit contains a μ-oxo-bridged di-nuclear iron center (Fe-O-Fe) and protein tyrosyl radical(Y-O•) that is transferred to the α C site for catalysis. The simplifiedschematic depictions of RNR holoenzyme structures shown herewere guided by structural studies.18,21

Ribonucleotide reductase and cancerY Aye et al

2012

Oncogene (2015) 2011 – 2021 © 2015 Macmillan Publishers Limited

whether increased dNTP synthesis actually occurs at DNA repairsites. Evidence does exist that the de novo thymidylate synthesispathway forms a scaffolded multienzyme complex at replicationsites in the nucleus.54 With respect to RNR, it is possible thattranslocation of this rate-limiting enzyme to DNA damage sites issufficient to affect local dNTP pools provided all other enzymesrequired for dNTP synthesis are within the nucleus. An alternativehypothesis related to DNA damage-induced RNR translocation isthat RNR has non-nucleotide reduction properties that promoteDNA repair. Parsing out the mechanisms at play during DNAdamage responses will require careful analysis and ultimately newways to accurately measure local dNTP fluxes in cells.

RNR AND DNA DAMAGE RESPONSESRNR genes were among the first DNA damage-inducible genesto be identified.55 In mammalian cells, genotoxins such aschlorambucil and ultraviolet light induce Rrm1 and Rrm2expression by approximately 10-fold.56,57 RRM2 protein stabilityis also DNA damage responsive,44 at least in part, throughATR-dependent downregulation of cyclin F-mediated RRM2degradation.28 The discovery of p53R2 further connected DNAdamage responses and RNR activity and also resolved thequestion of how DNA repair in quiescent cells could be supportedwhen RRM2 is undetectable. p53 binds the first p53R2 intronand stimulates p53R2 expression.26,58 In addition, DNA damage-induced phosphorylation by ATM stabilizes p53R2 and confersresistance to DNA damage.59 Inhibition of p53R2 expression incells that have an intact p53-dependent DNA damage checkpointreduces RNR activity, DNA repair and cell survival after genotoxinexposure.26,60 p53R2-null mouse fibroblasts also show severelyattenuated dNTP pools under oxidative stress.40

One aspect of the DNA damage response influenced byRNR activity is repair pathway choice.61,62 RNR-mediated dNTPproduction after DNA damage fuels DNA synthesis duringhomologous recombinational repair, whereas RNR inhibitionpromotes break repair by non-homologous end joining,which does not require extensive DNA synthesis.62 However,RNR-mediated dNTP pool increases are accompanied by highermutation rates, which may result from reduced fidelityof replicative polymerases and/or activation of error-pronetranslesion DNA synthesis at elevated dNTP levels.10,63–65

Despite the well-documented increase in RNR subunit levels afterDNA damage, there remains uncertainty about the extent to whichthere is a corresponding change in dNTP pools after genotoxicstress. In non-proliferating cells, a slow, fourfold accumulation ofp53R2 after DNA damage results in less than a two-fold increase indNTP pools, a modest change relative to that which occurs uponS-phase entry.47 Moreover, logarithmically growing cells do notshow significant dNTP pool increases upon DNA damage. This mayreflect the action of additional mechanisms influencing dNTP levelsafter DNA damage. It also remains possible that dNTP biosynthesisis compartmentalized at damage sites during the DNA damageresponse,51 such that measurements of total cellular dNTP levels failto reveal local changes.

RNR DEREGULATION AND GENOMIC INSTABILITYBecause proper dNTP levels are essential for genomic integrity,RNR deregulation is mutagenic.1,10,63 Initial insights into theconsequences of mammalian RNR deregulation came fromanalysis of mouse T-lymphosarcoma cells selected for deoxygua-nosine resistance and determined to have a mutation in the A site(RRM1D57N) that disrupts dATP-mediated enzyme inhibition.The mutagenized cells from which RRM1D57N was cloned showincreased dNTP levels and a 40-fold increase in mutation rate.66–68

RRM1D57N overexpression in CHO cells was subsequently shown tocause a 15–25-fold increase in spontaneous mutation frequency,

although no dNTP pool alterations could be identified.67 ElevatedRNR activity has also been identified in hydroxyurea (HU)-resistant,RRM2-overexpressing mouse cell lines. In one study withfibroblasts, a 3–15-fold increase in RNR activity was accompaniedby moderate changes in dNTP pools.69 By contrast, theHU-resistant, RRM2-overexpressing mouse mammary tumor TA3cell line shows a 40-fold increase in enzymatically active RRM2,estimated based on increased Y• content, but no detectable dNTPpool changes relative to parental cells.43,44 Therefore, the elevatedmutagenesis associated with RNR activity alterations has not beendefinitively linked to dNTP pool perturbations. Failure to detectdNTP pool changes in cells with altered RNR activity could bebecause even relatively small changes in dNTP pools can be highlymutagenic.70 Alternatively, a specific interaction between RNR andthe DNA replication or repair machinery may be involved as notedabove, so that biologically important localized pool alterations areobscured during analysis of total intracellular dNTP levels. It is alsonoteworthy that measurement of holocomplex activity in vivo ischallenging, because cell lysis can disrupt complex equilibriabetween subunits and alter cellular (d)NTP allosteric modulatorconcentrations.

RNR IN CANCERUncontrolled proliferation is a defining feature of cancers andmust be supported by a sufficient dNTP supply. Cancer cellsundergo metabolic reprogramming such that glucose is no longermetabolized to maximize ATP production as in normal cells butinstead is used to drive the production of macromolecules for cellreplication, including dNTPs.71 Studies of rat hepatomas in the1970s established that RNR activity is highly correlated with cancergrowth rate; 200-fold differences in enzyme activity wereobserved between fast- and slow-growing tumor cells.72 RRM2overexpression has been observed in gastric, ovarian, bladder andcolorectal cancers.73–77 RRM2 expression is correlated with tumorgrade for both breast and epithelial ovarian cancers, suggesting arole for RNR in supporting rapid cell division of high-gradetumors.78,79 Similarly, RRM2 levels are low in benign skin lesionsbut are significantly higher in malignant melanoma, with highRRM2 expression additionally correlating with poor overallsurvival.80 Increased p53R2 expression has also been reportedin cancers, including melanoma, oral carcinoma, esophagealsquamous cell carcinoma and non-small cell lung cancer(NSCLC).81–85 To broadly assess how RNR is affected in a largecollection of human cancers, we surveyed RNR gene expression inhuman cancers using the ONCOMINE database (Figure 2).Remarkably, RRM2 was among the top 10% most overexpressedgenes in 73 out of the 168 cancer analyses. These include sarcomaand cancers of the bladder, brain and central nervous system,breast, colorectal, liver and lung. Similarly, RRM1 was among thetop 10% most-overexpressed genes in 30 out of the 170 studies,including brain and central nervous system cancer, lung cancerand sarcoma. By contrast, p53R2 was among the top 10% mostoverexpressed genes in only 5 out of the 90 studies.Elevated RNR expression in cancers could be secondary to cell

cycle alterations, as many neoplasms show an increased S-phasefraction, or the direct result of gene amplification or other geneticor epigenetic alterations. Therefore, we analyzed RNR gene copynumber changes using the TCGA database (SupplementaryFigure S2). Among the RNR genes, RRM2B was the most frequentlyaffected by copy number changes and typically showed gains, inaccord with a recent report on human breast cancers.86 It shouldbe noted that the RRM2B locus (8q22.3) is located near the C-MYConcogene (8q24). Most RRM2B copy number increases in cancersare accompanied by C-MYC amplification, raising the possibilitythat they are simply passenger events. However, RRM2B amplifica-tion does occur independently of C-MYC amplification at lowfrequency in breast, colon, ovarian, prostate and uterine cancers,

Ribonucleotide reductase and cancerY Aye et al

2013

© 2015 Macmillan Publishers Limited Oncogene (2015) 2011 – 2021

as well as glioblastoma. This correlates well with increased p53R2expression in breast and prostate cancers (Figure 2) and supportsa potential role for p53R2 as a tumor promoter. RRM2 is amplifiedat low frequency in breast, ovarian, prostate and uterine cancers,malignancies in which RRM2 gene expression is also increased.RRM1 also underwent rare copy number changes in cancers, withthe direction of change depending on tumor type. This mayreflect the complex and stage-specific roles RRM1 can have intumorigenesis, as discussed below.

RRM1: A TUMOR SUPPRESSOR THAT CAN CONFERCHEMORESISTANCERNR-mediated dNTP biosynthesis can have varied and potentiallyopposing effects on tumorigenesis. Altered dNTP pools can impairDNA replication fidelity, leading to tumor-promoting mutations.On the other hand, RNR-supported dNTP production can protectagainst mutations by facilitating DNA repair. Following malignanttransformation, the same repair mechanisms can protect cancercells against potentially lethal stresses, such as those caused bygenotoxic chemotherapies. These complexities are reflected in thedata concerning RRM1 in cancer. Several studies point to RRM1 asa suppressor of tumor initiation. RRM1 overexpression in cultured

cells leads to reduced transformation and suppression oftumorigenicity and lung metastasis in vivo.87 In addition, RRM1overexpression in human lung cancer cells induces phosphataseand tensin homolog (PTEN) expression and suppresses migrationand invasion, as well as overall tumorigenicity and metastasisfollowing xenotransplantation.88 Rrm1-overexpressing transgenicmice show reduced urethane-induced lung tumorigenesis,89

although in an independent study Rrm1 overexpression was notfound to alter the background incidence of spontaneous lungneoplasms in mice.90 In human NSCLC patients who underwentsurgical resection but received no other form of treatment, high-level tumor-associated RRM1 expression correlated with longerlifespan and later disease recurrence.91 Co-expression of RRM1and excision repair protein ERCC1 was significantly associatedwith disease-free and overall survival, especially in patients whounderwent lung cancer surgery at early stages, although aspectsof this study were later questioned.92 The published mechanismsbehind tumor suppression by α, such as PTEN induction, requirefurther investigation and may indicate that functions other thandNTP production are required for α tumor-suppressor activity.Additional animal studies also are needed to address the distinctoutcomes reported concerning how α overexpression impactslung tumorigenesis.Consistent with a role for RNR in promoting DNA repair, high

level RRM1 expression also correlates with poor responses toplatinum drugs, which induce DNA damage against which highRRM1 levels afford protection, and gemcitabine, which directlytargets RRM1.93–102 Although not all studies have found RRM1levels to affect patient survival after gemcitabine treatment, thereis general consensus that low RRM1 levels improve responsivenessto gemcitabine therapy. Additionally, RRM1 overexpressionhas been reported in gemcitabine-resistant cancer cells.103–106

Consequently, RRM1 expression has been proposed as abiomarker in patients with advanced NSCLC to individualizechemotherapy.107

TUMOR PROMOTION BY RRM2 AND P53R2In contrast to the tumor-suppressing roles of RRM1, RRM2has oncogenic activity. For instance, RRM2 cooperates withoncoproteins to increase focus formation and anchorage-independent growth in mouse cells.108,109 Elevated RRM2 expres-sion in human carcinoma cells correlates with higher invasivepotential in vitro 110 as well as decreased thrombspondin-1 andincreased VEGF production, suggesting that RRM2 can promotetumor angiogenesis.111

The role of p53R2 in mutagenesis and tumorigenesis is lessclear-cut. Based on the p53-inducible nature of p53R2 expressionand its role in DNA repair, it was originally suggested that p53R2would have tumor-suppressor activity.26 Under genotoxic stress,p53R2 promotes p21 accumulation and G1 arrest, which mayfacilitate repair and prevent mutation accumulation.112 p53R2expression is negatively correlated with colon adenocarcinomametastasis.113 Similarly, elevated p53R2 expression suppressescancer cell invasiveness and correlates with markedly bettersurvival in colorectal cancer patients.114 Nevertheless p53R2 ishighly expressed in some human cancers as noted above, andexperimental suppression of p53R2 expression impairs cancer cellproliferation in vitro.81

Widespread overexpression of either Rrm2 or p53R2, but notRrm1, in transgenic mice induces NSCLC but not other tumors.90

RNR-induced lung neoplasms arise with relatively long latency,histopathologically resemble human papillary adenomas andadenocarcinomas and are associated with K-ras proto-oncogenemutations. Rrm2 or p53R2 overexpression causes an elevatedmutation frequency in cultured cells, suggesting that RNR-inducedneoplasms could arise through a mutagenic mechanism.Consistent with this possibility, combining RNR deregulation

Figure 2. RNR expression in human cancers. The bar graph illustratesaltered RNR expression in human cancers. Data were retrieved fromthe ONCOMINE cancer gene expression database (version 4.4.4.4,search done on 27 November 2013). The y axis represents thenumber of analyses with differences in gene expression for the geneof interest. Dark red, red and pink: number of analyses with the geneof interest among the top 1, 5 and 10% most overexpressed genes,respectively, in a given study. Dark green, green and light green:number of analyses with the gene of interest among the top1, 5 and 10% most underexpressed genes, respectively, in a givenstudy. RRM2 is among the top 10% of the most overexpressed genesin 73 out of the 168 analyses, RRM1 is among the top 10% in 30 outof the 170 studies and RRM2B (encoding p53R2) is among the top10% in 5 out of the 96 cases. In addition, RRM2 is among the top10% of the most underexpressed genes in 7 out of the 168 studies,and RRM1 is among the top 10% in 6 out of the 170 studies. Theexpression level of the c-MYC proto-oncogene is shown forcomparison. c-MYC is among the top 10% of the most overexpressedgenes in 40 out of the 174 analyses and among the top 10% of themost underexpressed genes in 17 out of the 174 studies.

Ribonucleotide reductase and cancerY Aye et al

2014

Oncogene (2015) 2011 – 2021 © 2015 Macmillan Publishers Limited

with DNA mismatch repair defects synergistically increasesmutagenesis and tumorigenesis.90

The available data raise fundamental questions about thetransforming activity associated with the RNR-β subunit. Giventhat increased RNR activity can lead to altered dNTP pools, onepossibility is that RNR-β overexpression promotes error-proneDNA synthesis, including increased frequency of base mis-insertions, insertion–deletion events and uracil misincorporation,as RNR also reduces UDP to dUDP. Recent evidence indicatesthat RNR-generated dNTPs are also necessary for neoplastic cellsto avoid oncogene-induced senescence.80,115,116 Senescenceinduced by activated Ras expression or Myc depletion was foundto be associated with repression of Rrm2 expression and could bebypassed by treatment with exogenous nucleosides or ectopicRrm2 expression. Thus aberrant RNR expression could enable cellsto overcome this important barrier to transformation. Apoptosisis another cancer-related pathway that could be impacted byRNR-mediated dNTP biosynthesis. In particular, (d)ATP binding toApaf-1 and cytochrome c regulates the formation of theapoptosome, which is crucial for caspase activation and down-stream apoptotic events.117,118

Reactive oxygen species (ROS) also may contribute to RNR-induced mutagenesis and transformation. Because Y• within RRM2is short-lived,119 its lability and reactivity could lead to secondaryreactive species when RRM2 is overproduced.120,121 It remainsunclear whether RRM2 can indeed propagate ROS.119 Involvementof ROS would be compatible with the lung specificity oftumorigenesis in RRM2-overexpressing mice as well as theobserved synergy between RRM2 deregulation and the mismatchrepair system, which responds to both base mismatches andoxidative DNA damage.90 Interestingly, the cooperativity betweenRrm2 and activated oncogenes in inducing transformation isindependent of ribonucleotide reduction.108 Moreover, RRM2 andp53R2-overexpressing cells show an increased frequency of G→ Ttransversions, a signature of oxidative DNA damage.90

As noted for RRM1, there also is evidence that RRM2 levels incancer cells can influence therapeutic responses. Suppression ofRRM2 expression sensitizes cancer cells to both RNR inhibitors andcisplatin.110,122 Elevated p53R2 expression correlates with resis-tance to genotoxic therapies such as radiation and 5-flurouracil,whereas p53R2 knockdown sensitizes cells to DNA-damagingagents.81,123–125 Interestingly, p53R2 levels in cancer cells do notcorrelate with sensitivity to RNR-targeted therapeutics, such astriapine.104,122,126 RRM2 is part of a 12 gene set that is predictive ofbenefit from adjuvant chemotherapy in NSCLC and is associatedwith poor prognosis.127 RRM2 overexpression is also a marker ofpoor prognosis in ovarian cancer patients receiving gemcitabineor other cytotoxic therapies,128 and low RRM2 expression inpancreatic cancers is predictive of greater response togemcitabine.129,130 Similarly, the absence of p53R2 expressionis associated with responsiveness to chemoradiotherapy foresophageal squamous cell carcinoma.131

Another intriguing example of therapeutic response modulationby RNR is found in the case of small-molecule inhibitors of TMPK,which converts dTMP to dTDP in dTTP biosynthesis.52,132 Both RNRand TMPK localize to DNA damage in the nucleus, and theiractivities impact the relative levels of dUTP (produced via RNR)and dTTP (produced via TMPK) at repair sites. When the dUTP:dTTP ratio is low, there is minimal uracil misincorporation duringDNA repair. However, if the dUTP:dTTP ratio is elevated, highlevels of uracil misincorporation can lead to futile repair cycles,DNA breakage and cell death.52 These observations are the basisfor a therapeutic strategy that combines TMPK inhibition with low-dose chemotherapy. It follows that tumors with elevated RNRactivity, dictated in many cases by RRM2 expression, would beespecially susceptible to such a strategy. This prediction holds truein initial cell culture analyses and offers hope for a targetedapproach that could be effective against the large fraction of

cancers with high-level RRM2 expression. Nevertheless, becausedNTP biosynthesis is a complex process requiring multipleenzymes, further work is necessary to fortify the link betweenTMPK inhibition and uracil misincorporation. For instance, thepresence of another enzyme necessary for dNTP production,NDPK, has not been established in the same complexes. Whetherenzymes other than the rate-limiting factor RNR must be presentdirectly at damage sites to allow local dUTP accumulation remainsunknown, but one possibility is that without NDPK recruitment thedNTP precursors dTDP and dUDP would instead accumulate.

RNR AND CANCER THERAPYBecause RNR is the gatekeeper of dNTP homeostasis,29,133 theenzyme is long recognized as a cancer therapeutic target.Although small-molecule inhibitors continue to represent theprimary strategy for RNR inhibition since the last comprehensivereview, studies have uncovered new approaches to small-molecule-based subunit-specific activity modulation and highlightthe potential of gene therapy.134,135 Small-molecule RNR inhibitorsin active clinical use fall into two classes: nucleoside analogs andredox active metal chelators (Figure 3). The former class targetsRRM1 (α), in line with the ability of α to bind nucleotides, whereasthe latter group targets RRM2 (β), consistent with the dependenceof β on a redox active metallocofactor.19 The existence of thesetwo distinct drug classes reflects the inherent biochemicaldiversities of the two subunits, both of which are required for NDPreduction.29,136 Consequently, most drugs targeting α show littlecross reactivity with β and vice versa.

RNR-α INHIBITORSOne of the first nucleoside analogs targeting α to be clinicallyapproved was gemcitabine (Gemzar, F2C; Figure 3a), whichcontinues to be a frontline therapy against pancreatic, bladder andlung cancers.137 RNR inactivation by gemcitabine has beenextensively reviewed.138 The active form, the substrate analogdiphosphate (F2CDP), is an irreversible, suicide inactivator of α,which results in a covalent complex between α and the sugar of F2CDP.139 Despite being stereotyped as an α-targeting drug, F2CDPcannot inactivate α without β; holocomplex assembly is obligatoryfor transient S• formation within the substrate-binding C site on αthat then reacts with substrate analog F2CDP, affording a reactiveintermediate capable of irreversibly alkylating α (Figure 3a).The second approved α-targeting drug to have its RNR-

targeting mechanism elucidated was clofarabine (Clolar, ClF;Figure 3b and Supplementary Figure S3).22,23 ClF belongs to aclinically successful class of nucleoside prodrugs, includingcytarabine (ara-C), nelarabine, azacitidine, decitabine, cladribineand fludarabine, used to treat hematological malignancies.140 Themajority of these antimetabolites are thought to target RNRas part of their cytotoxic spectrum; however, the molecularmechanisms underlying RNR inhibition remain poorly definedexcept for ClF. As the triphosphate (ClFTP) is the most abundantClF metabolite in cells, it was initially postulated that α inhibitionoccurred through α A site binding by ClFTP, similarly to thefeedback inhibitor dATP.140 However, both the diphosphate andtriphosphate (ClFD(T)P) emerged to be α inhibitors in vitro. ClFTP isan A-site-binding reversible inhibitor. RNR-α can be completelyinhibited by brief exposure to saturating amounts of ClFTP, butactivity is regained upon prolonged drug exposure, resulting in asteady-state level of 50% activity. Whether this restoration ofactivity is important for ClF’s drug efficacy remains unaddressed.Unexpectedly, the substrate analog diphosphate, ClFDP, is areversible α inhibitor that binds the C site with slow releaseproperties. Most strikingly, although ClFDP occupies the same siteas F2CDP, α does not chemically engage with ClFDP, and the drug

Ribonucleotide reductase and cancerY Aye et al

2015

© 2015 Macmillan Publishers Limited Oncogene (2015) 2011 – 2021

can be recovered intact from α even following prolongedincubation.22

ClF is a hybrid of its predecessors cladribine and fludarabine,two adenine-containing nucleosides used previously for leukemicreticuloendotheliosis and hairy cell leukemia or leukemia andlymphoma, respectively.140–142 These drugs were developed asanalogs of deoxyadenosine, which selectively kills lymphocytes.143

The triphosphate of cladribine is thought to inhibit RNR, albeitthrough an unknown mechanism.144 Fludarabine primarily inhibitsDNA polymerases, with minimal RNR inhibition. Cladribine andfludarabine feature undesirable chemotypes that make themboth susceptible to glycolytic cleavage, reducing their efficacy.Particularly with fludarabine, hydrolytic and enzymatic cleavageproduces 2-fluoroadenine, which is subsequently converted to thehighly toxic 2-fluoro-adenosine triphosphate.145 ClF represents atriumph of semi-rational design, incorporating the most desirableproperties of cladribine and fludarabine, with minimizedtoxicity.140

Similar to natural ligands, antimetabolites also modulate RNRactivity through changes to α quaternary structure. Although RNRoligomeric regulation remains poorly understood, significantphenotypic differences exist between the natural and non-natural ligand-induced states.41 The α hexameric state inducedby dATP only persists when the A site is saturated with dATP, thusenabling interconversions between the active and inactive statesas a function of cellular dATP concentration. ClFTP also binds theA site and hexamerizes α in vitro.22 However, unlike dATP-inducedα-hexamers, ClFTP-induced α-hexamers persist subsequent toClFTP dissociation.23 Persistent α hexamerization is also initiatedby ClFDP binding to the C site, suggesting that quaternaryregulation is not uniquely associated with α allosteric sites. F2CDP-mediated irreversible RNR inactivation in which the suicidesubstrate F2CDP interacts with the C site is also proposed to leadto α6β6.

139 The kinetic stability of the ClFD(T)P-induced hexamericstates was recently exploited to demonstrate that α exists as adynamic equilibrium of oligomers in cells.23 α from untreated cellsis a mixture of dimer and monomer, whereas ClF-treated cellsyield α that is mainly hexameric.The allosteric activator ATP has also been proposed to affect

α oligomerization independent of β. However, there are differingreports, based on distinct technical approaches, on the functionaloutcome of this process (Supplementary Table S1). Kashlan andCooperman24 described ATP-concentration-dependent formationof αm (m= 2,4,6), whereas Dealwis and colleagues21 showed that3 mM ATP causes α hexamerization. Hofer and colleagues146 alsodetected the presence of α2 and α6 states at 0.1 mM ATP. One wayto reconcile these data is that ATP, in a similar way to dATP,induces weakly associated hexamers that readily collapse tolower-order oligomers at low ATP levels. This model is consistentwith analysis of α in 0.5 mM ATP showing a mixture of lower-order species.22 These observations imply that persistenthexamerization is uniquely induced by antileukemic nucleotidesthat inhibit RNR.Despite their success, nucleoside analogs suffer from

complications.147 Most are administered as inactive prodrugs,necessitating successive phosphorylation by cellular kinases togain activity.140 Depending upon the nucleoside, a differentsteady-state equilibrium between the monophosphate, dipho-sphate and triphosphate is established.148 These variable drugmetabolites can lead to low steady-state concentrations of theactive variant, side effects149 and ultimately toxicity.150 Nucleos(t)ides are also susceptible to numerous catabolic pathways,which can generate dangerous side products.151 In addition,resistance to certain dNTP analogs is linked to differential miRNAexpression.152

Both F2C and ClF are used in combination therapies. Thecombination of F2C and carboplatin is widely used.153 The theorybehind this and related approaches is that functional RNR is

Figure 3. Inhibitors targeting multiple characteristics of RNR.(a, b) Nucleotide analog inhibitors interacting with α. (a) Gemcita-bine (F2C) is approved for the treatment of lung, pancreatic, breastand ovarian cancers. RNR catalytic turnover is triggered by forwardproton-coupled electron transfer, resulting in transient C-S• forma-tion that initiates NDP reduction. The active diphosphate (F2CDP)form, as a substrate analog, can interact with the C-S• to form asubstrate radical that subsequently decomposes, resulting inenzyme inactivation. Inactivation leads to covalent crosslinkingbetween α and the sugar of decomposed F2CDP. Formation ofactivated RNR holocomplex bearing the transient C-S• is aprerequisite for F2CDP inactivation. (b) Clofarabine (ClF) is usedfor the treatment of refractory pediatric leukemias. The activediphosphate and triphosphate [ClFD(T)P] forms are reversibleinhibitors that exclusively target α. Inhibition in both cases iscoupled with the assembly of hexameric, catalytically non-viablequaternary states that are induced by ClFD(T)P either in thepresence or absence of β. See also Supplementary Figure S3.(c, d) Small-molecule inhibitors interacting with β. (c) HU is used inthe treatment of chronic myeloid leukemia, melanoma, head andneck and refractory ovarian cancers. Treatment of β with HU leads tothe formation of catalytically incapable apo-β2 that lacks both Y-O•and the di-iron center. (d) Triapine (3AP) is currently being evaluatedin clinical trials. β− Specific targeting is mediated by the activeform Fe(II)-(3AP) that reduces Y-O•, converting holo-β2 into thecatalytically incapable variant, met-β2 (Y-OH). C-SH, reduced Cys;C-S•, Cys radical; Y-OH, Tyr; Y-O•, Tyr radical.

Ribonucleotide reductase and cancerY Aye et al

2016

Oncogene (2015) 2011 – 2021 © 2015 Macmillan Publishers Limited

needed to provide dNTPs to repair DNA damage caused bydrugs such as carboplatin. In line with this hypothesis, theF2C/carboplatin combination is successful in treating carboplatin-resistant tumors.154 ClF is commonly used with ara-C, a dCTPanalog that, following phosphorylation to ara-CTP, can block DNAsynthesis. This combination highlights that nucleoside analogshave complex metabolic pathways: ara-CTP inhibits deoxycytidinekinase (dCK), an enzyme responsible for phosphorylating ara-C toara-CMP.155 Steady-state cellular ara-CTP levels are relatively low,impairing efficacy. RNR inhibition by ClFD(T)P stimulates dCKactivity,156 leading to higher ara-CTP levels.157 A similar effect isobserved in cells treated with F2C and staurosporine, whichincreases dCK activity.158

RNR-β INHIBITORSThe di-iron center and Y• within β are logical targets for anti-cancer drugs. One structurally simple β inhibitor, HU (Figure 3c), isa known metal chelator and radical quencher. It is used in thetreatment of CML,159 AML 160 and glioblastoma.161 Although β is aHU target 162 and β overexpression confers HU resistance,69,163,164

this drug is promiscuous, and other metalloenzymes, such ascarbonic anhydrase and matrix metalloproteinases, are also HUtargets.165 HU attacks both the Y• and di-iron center ofmammalian RNR-β on a similar timescale.166 This contrasts withthe mechanism of HU-induced β-inactivation in bacteria, whichinvolves exclusively Y• quenching, leading to a met enzymestate.167 These data underscore the dual properties of HU as ametal-chelator and single-electron donor.A second β-targeting drug, triapine (3AP; Figure 3d) is currently

in clinical trials for CML and various solid tumors.168 3APrepresents an important case study, because it highlights thecomplexities in deconvoluting inhibition mechanisms. 3AP is themost successful of a group of thiosemicarbazone β inhibitors andis active against HU-resistant tumors.169 Initially, it was assumedthat 3AP inhibited β through iron chelation, either from the βactive site170 or from the labile iron pool.171 However, this theorywas questioned with observations that metal-bound 3AP,particularly Fe(II)-(3AP) complex, under aerobic conditions, iscapable of generating ROS172 that could inhibit β.173 Recentstudies indicate that the Fe(II)-(3AP) complex is the active inhibitorin vitro and can reduce Y• at a rate faster than iron chelation at theβ active site.119 Cultured K562 cells and HU-resistant TA3 cellstreated with 3AP showed no change in iron content within β butunderwent a rapid loss of Y•. No oxidation of β residues oraccumulation of oxidized cellular proteins could be detected,suggesting that ROS is not important for β inhibition by 3AP.These data collectively imply that human β is highly susceptible toradical-targeting drugs, a finding that opens a range of prospectsfor inhibitor design and optimization.

GENETIC STRATEGIES FOR RNR INHIBITIONAlthough the most effective and widely used approaches for RNRinhibition center on small-molecule inhibitors, targeted knock-down of RNR subunits using small interfering RNA (siRNA) also hasbeen developed. RRM2 knockdown impairs tumor cell prolifera-tion174 and sensitizes cancer cells to DNA-damaging agents aswell as the RNR inhibitors HU and 3AP.122 RRM2 knockdown alsoovercomes cisplatin resistance in cultured cells.54 Similarly, siRNAsdesigned to downregulate RRM1 expression in tumors have beendeveloped to overcome gemcitabine resistance.175 Although theapplication of these siRNA-based strategies in patients iscomplicated by challenges related to delivery176 and stability inplasma,177 encouraging results have been observed. A 20-merphosphorothioate oligodeoxynucleotide, GTI-2040,178 that signifi-cantly reduces RRM2 mRNA levels, is presently in clinical trials forvarious solid tumors.174–176 Delivery of siRNA against RRM2 by

nanoparticles or retroviruses has been shown to suppress cancersof the head, neck and pancreas.179,180

SUMMARY AND FUTURE PERSPECTIVESSince the discovery of RNR activity by Reichard et al.181 in 1961,there have been tremendous advances in understanding thestructure, function and biological significance of this essentialfamily of enzymes.17,19,41 The next stage in understanding RNRfunctions in disease-related contexts likely will require a concertedinterdisciplinary effort that merges biochemical and geneticanalyses and couples in vitro enzymology and structural studieswith relevant cell culture and animal models. Many mechanisticdetails remain to be resolved for mammalian RNR, including thedynamics and functional importance of oligomerization andsubcellular localization, as well as the intricacies of cofactorassembly and metalloenzyme regulation. RNR, encompassingboth conventional reductase and possible moonlighting (non-reductase) functions, clearly has important, subunit-specific rolesin cancer biology, influencing tumor initiation, progression andtherapeutic sensitivity, while also serving as a target for anti-cancer drugs. The extent to which the oncogenic impact of RNRrelates to RNR-mediated mutagenesis, suppression of senescence,ROS production or modulation of apoptosis remains a keyquestion for future studies. Finally, given that RNR is well-established as an effective therapeutic target, cell-based high-throughput phenotypic screening assays represent a promisingavenue for future drug development. Beginning with theidentification of a novel radical-based catalytic mechanism, thestudy of RNR has revealed many intriguing biochemical mechan-isms over the years, and we anticipate that more surprises are instore with the continued analysis of the role of RNR in cancer.

CONFLICT OF INTERESTThe authors declare no conflict of interest.

ACKNOWLEDGEMENTSWe thank Professors Rick Cerione and Jennifer Surtees for helpful discussion andcomments on the manuscript. YA acknowledges a faculty development grant fromthe ACCEL program supported by NSF (SBE-0547373), an Affinito-Stewart grant fromthe President's Council of Cornell Women and a Milstein sesquicentennial juniorfaculty fellowship. MJCL acknowledges an HHMI international student predoctoralfellowship.

REFERENCES1 Mathews CK. DNA precursor metabolism and genomic stability. FASEB J 2006; 20:

1300–1314.2 Rampazzo C, Miazzi C, Franzolin E, Pontarin G, Ferraro P, Frangini M et al.

Regulation by degradation, a cellular defense against deoxyribonucleotide poolimbalances. Mutat Res 2010; 703: 2–10.

3 Anglana M, Apiou F, Bensimon A, Debatisse M. Dynamics of DNA replicationin mammalian somatic cells: nucleotide pool modulates origin choice andinterorigin spacing. Cell 2003; 114: 385–394.

4 Courbet S, Gay S, Arnoult N, Wronka G, Anglana M, Brison O et al. Replicationfork movement sets chromatin loop size and origin choice in mammalian cells.Nature 2008; 455: 557–560.

5 Ge XQ, Jackson DA, Blow JJ. Dormant origins licensed by excess Mcm2-7 arerequired for human cells to survive replicative stress. Genes Dev 2007; 21:3331–3341.

6 Reichard P. Interactions between deoxyribonucleotide and DNA synthesis.Annu Rev Biochem 1988; 57: 349–374.

7 Bester AC, Roniger M, Oren YS, Im MM, Sarni D, Chaoat M et al. Nucleotidedeficiency promotes genomic instability in early stages of cancer development.Cell 2011; 145: 435–446.

8 Burrell RA, McClelland SE, Endesfelder D, Groth P, Weller MC, Shaikh N et al.Replication stress links structural and numerical cancer chromosomal instability.Nature 2013; 494: 492–496.

Ribonucleotide reductase and cancerY Aye et al

2017

© 2015 Macmillan Publishers Limited Oncogene (2015) 2011 – 2021

9 Wheeler LJ, Rajagopal I, Mathews CK. Stimulation of mutagenesis by propor-tional deoxyribonucleoside triphosphate accumulation in Escherichia coli. DNARepair (Amst) 2005; 4: 1450–1456.

10 Chabes A, Georgieva B, Domkin V, Zhao X, Rothstein R, Thelander L. Survival ofDNA damage in yeast directly depends on increased dNTP levels allowed byrelaxed feedback inhibition of ribonucleotide reductase. Cell 2003; 112: 391–401.

11 Fersht AR. Fidelity of replication of phage phi X174 DNA by DNA polymerase IIIholoenzyme: spontaneous mutation by misincorporation. Proc Natl Acad Sci USA1979; 76: 4946–4950.

12 Petruska J, Goodman MF, Boosalis MS, Sowers LC, Cheong C, Tinoco I Jr.Comparison between DNA melting thermodynamics and DNA polymerasefidelity. Proc Natl Acad Sci USA 1988; 85: 6252–6256.

13 Perrino FW, Loeb LA. Differential extension of 3' mispairs is a major contributionto the high fidelity of calf thymus DNA polymerase-alpha. J Biol Chem 1989; 264:2898–2905.

14 Mendelman LV, Petruska J, Goodman MF. Base mispair extension kinetics.Comparison of DNA polymerase alpha and reverse transcriptase. J Biol Chem1990; 265: 2338–2346.

15 Bebenek K, Roberts JD, Kunkel TA. The effects of dNTP pool imbalances onframeshift fidelity during DNA replication. J Biol Chem 1992; 267: 3589–3596.

16 Kunz BA, Kohalmi SE. Modulation of mutagenesis by deoxyribonucleotide levels.Annu Rev Genet 1991; 25: 339–359.

17 Cotruvo JA, Stubbe J. Class I ribonucleotide reductases: metallocofactorassembly and repair in vitro and in vivo. Annu Rev Biochem 2011; 80: 733–767.

18 Uhlin U, Eklund H. Structure of ribonucleotide reductase protein R1. Nature 1994;370: 533–539.

19 Minnihan EC, Nocera DG, Stubbe J. Reversible, long-range radical transfer inE. coli class Ia ribonucleotide reductase. Acc Chem Res 2013; 46: 2524–2535.

20 Fu Y, Long MJ, Rigney M, Parvez S, Blessing WA, Aye Y. Uncoupling of allostericand oligomeric regulation in a functional hybrid enzyme constructed fromEscherichia coli and human ribonucleotide reductase. Biochemistry 2013; 52:7050–7059.

21 Fairman JW, Wijerathna SR, Ahmad MF, Xu H, Nakano R, Jha S et al. Structuralbasis for allosteric regulation of human ribonucleotide reductase by nucleotide-induced oligomerization. Nat Struct Mol Biol 2011; 18: 316–322.

22 Aye Y, Stubbe J. Clofarabine 5'-di and -triphosphates inhibit humanribonucleotide reductase by altering the quaternary structure of its large sub-unit. Proc Natl Acad Sci USA 2011; 108: 9815–9820.

23 Aye Y, Brignole EJ, Long MJ, Chittuluru J, Drennan CL, Asturias FJ et al.Clofarabine targets the large subunit (alpha) of human ribonucleotidereductase in live cells by assembly into persistent hexamers. Chem Biol 2012; 19:799–805.

24 Kashlan OB, Cooperman BS. Comprehensive model for allosteric regulation ofmammalian ribonucleotide reductase: refinements and consequences. Bio-chemistry 2003; 42: 1696–1706.

25 Meyer Y, Buchanan BB, Vignols F, Reichheld JP. Thioredoxins and glutaredoxins:unifying elements in redox biology. Annu Rev Genet 2009; 43: 335–367.

26 Tanaka H, Arakawa H, Yamaguchi T, Shiraishi K, Fukuda S, Matsui K et al.A ribonucleotide reductase gene involved in a p53-dependent cell-cyclecheckpoint for DNA damage. Nature 2000; 404: 42–49.

27 Chabes AL, Pfleger CM, Kirschner MW, Thelander L. Mouse ribonucleotidereductase R2 protein: a new target for anaphase-promoting complex-Cdh1-mediated proteolysis. Proc Natl Acad Sci USA 2003; 100: 3925–3929.

28 D'Angiolella V, Donato V, Forrester FM, Jeong YT, Pellacani C, Kudo Y et al. CyclinF-mediated degradation of ribonucleotide reductase M2 controls genomeintegrity and DNA repair. Cell 2012; 149: 1023–1034.

29 Nordlund P, Reichard P. Ribonucleotide reductases. Annu Rev Biochem 2006; 75:681–706.

30 Bourdon A, Minai L, Serre V, Jais JP, Sarzi E, Aubert S et al. Mutation of RRM2B,encoding p53-controlled ribonucleotide reductase (p53R2), causes severemitochondrial DNA depletion. Nat Genet 2007; 39: 776–780.

31 Pontarin G, Ferraro P, Hakansson P, Thelander L, Reichard P, Bianchi V.p53R2-dependent ribonucleotide reduction provides deoxyribonucleotides inquiescent human fibroblasts in the absence of induced DNA damage. J BiolChem 2007; 282: 16820–16828.

32 Pontarin G, Ferraro P, Bee L, Reichard P, Bianchi V. Mammalian ribonucleotidereductase subunit p53R2 is required for mitochondrial DNA replication and DNArepair in quiescent cells. Proc Natl Acad Sci USA 2012; 109: 13302–13307.

33 Guittet O, Hakansson P, Voevodskaya N, Fridd S, Graslund A, Arakawa H et al.Mammalian p53R2 protein forms an active ribonucleotide reductase in vitro withthe R1 protein, which is expressed both in resting cells in response to DNAdamage and in proliferating cells. J Biol Chem 2001; 276: 40647–40651.

34 Mandel H, Szargel R, Labay V, Elpeleg O, Saada A, Shalata A et al. Thedeoxyguanosine kinase gene is mutated in individuals with depleted hepato-cerebral mitochondrial DNA. Nat Genet 2001; 29: 337–341.

35 Saada A, Shaag A, Mandel H, Nevo Y, Eriksson S, Elpeleg O. Mutant mitochondrialthymidine kinase in mitochondrial DNA depletion myopathy. Nat Genet 2001;29: 342–344.

36 Chimploy K, Song S, Wheeler LJ, Mathews CK. Ribonucleotide reductaseassociation with Mammalian liver mitochondria. J Biol Chem 2013; 288:13145–13155.

37 Kollberg G, Darin N, Benan K, Moslemi AR, Lindal S, Tulinius M et al. A novelhomozygous RRM2B missense mutation in association with severe mtDNAdepletion. Neuromuscul Disord 2009; 19: 147–150.

38 Shaibani A, Shchelochkov OA, Zhang S, Katsonis P, Lichtarge O, Wong LJ et al.Mitochondrial neurogastrointestinal encephalopathy due to mutationsin RRM2B. Arch Neurol 2009; 66: 1028–1032.

39 Tyynismaa H, Ylikallio E, Patel M, Molnar MJ, Haller RG, Suomalainen A.A heterozygous truncating mutation in RRM2B causes autosomal-dominantprogressive external ophthalmoplegia with multiple mtDNA deletions. Am J HumGenet 2009; 85: 290–295.

40 Kimura T, Takeda S, Sagiya Y, Gotoh M, Nakamura Y, Arakawa H. Impairedfunction of p53R2 in Rrm2b-null mice causes severe renal failure throughattenuation of dNTP pools. Nat Genet 2003; 34: 440–445.

41 Hofer A, Crona M, Logan DT, Sjoberg BM. DNA building blocks: keeping controlof manufacture. Crit Rev Biochem Mol Biol 2012; 47: 50–63.

42 Engström Y, Eriksson S, Jildevik I, Skog S, Thelander L, Tribukait B. Cell cycle-dependent expression of mammalian ribonucleotide reductase. Differentialregulation of the two subunits. J Biol Chem 1985; 260: 9114–9116.

43 Eriksson S, Graslund A, Skog S, Thelander L, Tribukait B. Cell cycle-dependentregulation of mammalian ribonucleotide reductase. The S phase-correlatedincrease in subunit M2 is regulated by de novo protein synthesis. J Biol Chem1984; 259: 11695–11700.

44 Chabes A, Thelander L. Controlled protein degradation regulates ribonucleotidereductase activity in proliferating mammalian cells during the normal cell cycleand in response to DNA damage and replication blocks. J Biol Chem 2000; 275:17747–17753.

45 Johansson E, Hjortsberg K, Thelander L. Two YY-1-binding proximal elementsregulate the promoter strength of the TATA-less mouse ribonucleotidereductase R1 gene. J Biol Chem 1998; 273: 29816–29821.

46 Engström Y, Rozell B, Hansson HA, Stemme S, Thelander L. Localization ofribonucleotide reductase in mammalian cells. Embo J 1984; 3: 863–867.

47 Hakansson P, Hofer A, Thelander L. Regulation of mammalian ribonucleotidereduction and dNTP pools after DNA damage and in resting cells. J Biol Chem2006; 281: 7834–7841.

48 Chabes AL, Bjorklund S, Thelander L. S Phase-specific transcription of themouse ribonucleotide reductase R2 gene requires both a proximal repressiveE2F-binding site and an upstream promoter activating region. J Biol Chem 2004;279: 10796–10807.

49 Engström Y, Rozell B. Immunocytochemical evidence for the cytoplasmiclocalization and differential expression during the cell cycle of the M1 and M2subunits of mammalian ribonucleotide reductase. EMBO J 1988; 7: 1615–1620.

50 Pontarin G, Fijolek A, Pizzo P, Ferraro P, Rampazzo C, Pozzan T et al.Ribonucleotide reduction is a cytosolic process in mammalian cells indepen-dently of DNA damage. Proc Natl Acad Sci USA 2008; 105: 17801–17806.

51 Niida H, Katsuno Y, Sengoku M, Shimada M, Yukawa M, Ikura M et al. Essentialrole of Tip60-dependent recruitment of ribonucleotide reductase at DNAdamage sites in DNA repair during G1 phase. Genes Dev 2010; 24: 333–338.

52 Hu CM, Yeh MT, Tsao N, Chen CW, Gao QZ, Chang CY et al. Tumor cells requirethymidylate kinase to prevent dUTP incorporation during DNA repair. Cancer Cell2012; 22: 36–50.

53 Zhang YW, Jones TL, Martin SE, Caplen NJ, Pommier Y. Implication of checkpointkinase-dependent up-regulation of ribonucleotide reductase R2 in DNA damageresponse. J Biol Chem 2009; 284: 18085–18095.

54 Anderson DD, Woeller CF, Chiang EP, Shane B, Stover PJ. Serine hydro-xymethyltransferase anchors de novo thymidylate synthesis pathway to nuclearlamina for DNA synthesis. J Biol Chem 2012; 287: 7051–7062.

55 Elledge SJ, Zhou Z, Allen JB, Navas TA. DNA damage and cell cycle regulation ofribonucleotide reductase. Bioessays 1993; 15: 333–339.

56 Hurta RA, Wright JA. Alterations in the activity and regulation of mammalianribonucleotide reductase by chlorambucil, a DNA damaging agent. J Biol Chem1992; 267: 7066–7071.

57 Filatov D, Bjorklund S, Johansson E, Thelander L. Induction of the mouseribonucleotide reductase R1 and R2 genes in response to DNA damage byUV light. J Biol Chem 1996; 271: 23698–23704.

58 Nakano K, Balint E, Ashcroft M, Vousden KH. A ribonucleotide reductase gene is atranscriptional target of p53 and p73. Oncogene 2000; 19: 4283–4289.

59 Chang L, Zhou B, Hu S, Guo R, Liu X, Jones SN et al. ATM-mediated serine 72phosphorylation stabilizes ribonucleotide reductase small subunit p53R2 proteinagainst MDM2 to DNA damage. Proc Natl Acad Sci USA 2008; 105: 18519–18524.

Ribonucleotide reductase and cancerY Aye et al

2018

Oncogene (2015) 2011 – 2021 © 2015 Macmillan Publishers Limited

60 Yamaguchi T, Matsuda K, Sagiya Y, Iwadate M, Fujino MA, Nakamura Y et al.p53R2-dependent pathway for DNA synthesis in a p53-regulated cell cyclecheckpoint. Cancer Res 2001; 61: 8256–8262.

61 Moss J, Tinline-Purvis H, Walker CA, Folkes LK, Stratford MR, Hayles J et al. Break-induced ATR and Ddb1-Cul4(Cdt)(2) ubiquitin ligase-dependent nucleotidesynthesis promotes homologous recombination repair in fission yeast. Genes Dev2010; 24: 2705–2716.

62 Burkhalter MD, Roberts SA, Havener JM, Ramsden DA. Activity of ribonucleotidereductase helps determine how cells repair DNA double strand breaks. DNARepair (Amst) 2009; 8: 1258–1263.

63 Gon S, Napolitano R, Rocha W, Coulon S, Fuchs RP. Increase in dNTP pool sizeduring the DNA damage response plays a key role in spontaneous and induced-mutagenesis in Escherichia coli. Proc Natl Acad Sci USA 2011; 108: 19311–19316.

64 Lis ET, O'Neill BM, Gil-Lamaignere C, Chin JK, Romesberg FE. Identificationof pathways controlling DNA damage induced mutation in Saccharomycescerevisiae. DNA Repair (Amst) 2008; 7: 801–810.

65 Sabouri N, Viberg J, Goyal DK, Johansson E, Chabes A. Evidence for lesion bypassby yeast replicative DNA polymerases during DNA damage. Nucleic Acids Res2008; 36: 5660–5667.

66 Weinberg G, Ullman B, Martin DW Jr. Mutator phenotypes in mammalian cellmutants with distinct biochemical defects and abnormal deoxyribonucleosidetriphosphate pools. Proc Natl Acad Sci USA 1981; 78: 2447–2451.

67 Caras IW, Martin DW Jr. Molecular cloning of the cDNA for a mutant mouseribonucleotide reductase M1 that produces a dominant mutator phenotype inmammalian cells. Mol Cell Biol 1988; 8: 2698–2704.

68 Ullman B, Clift SM, Gudas LJ, Levinson BB, Wormsted MA, Martin DW Jr.Alterations in deoxyribonucleotide metabolism in cultured cells withribonucleotide reductase activities refractory to feedback inhibition by2'-deoxyadenosine triphosphate. J Biol Chem 1980; 255: 8308–8314.

69 Akerblom L, Ehrenberg A, Graslund A, Lankinen H, Reichard P, Thelander L.Overproduction of the free radical of ribonucleotide reductase in hydroxyurea-resistant mouse fibroblast 3T6 cells. Proc Natl Acad Sci USA 1981; 78: 2159–2163.

70 Ahluwalia D, Schaaper RM. Hypermutability and error catastrophe due to defectsin ribonucleotide reductase. Proc Natl Acad Sci USA 2013; 110: 18596–18601.

71 Ward PS, Thompson CB Metabolic reprogramming: a cancer hallmark evenwarburg did not anticipate. Cancer Cell 2012; 21: 297–308.

72 Elford HL, Freese M, Passamani E, Morris HP. Ribonucleotide reductase and cellproliferation. I. Variations of ribonucleotide reductase activity with tumor growthrate in a series of rat hepatomas. J Biol Chem 1970; 245: 5228–5233.

73 Morikawa T, Maeda D, Kume H, Homma Y, Fukayama M. Ribonucleotidereductase M2 subunit is a novel diagnostic marker and a potential therapeutictarget in bladder cancer. Histopathology 2010; 57: 885–892.

74 Wang LM, Lu FF, Zhang SY, Yao RY, Xing XM, Wei ZM. Overexpression of catalyticsubunit M2 in patients with ovarian cancer. Chin Med J (Engl) 2012; 125:2151–2156.

75 Morikawa T, Hino R, Uozaki H, Maeda D, Ushiku T, Shinozaki A et al. Expression ofribonucleotide reductase M2 subunit in gastric cancer and effects of RRM2inhibition in vitro. Hum Pathol 2010; 41: 1742–1748.

76 Lu AG, Feng H, Wang PX, Han DP, Chen XH, Zheng MH. Emerging roles of theribonucleotide reductase M2 in colorectal cancer and ultraviolet-induced DNAdamage repair. World J Gastroenterol 2012; 18: 4704–4713.

77 Liu X, Zhang H, Lai L, Wang X, Loera S, Xue L et al. Ribonucleotide reductasesmall subunit M2 serves as a prognostic biomarker and predicts poor survival ofcolorectal cancers. Clin Sci (Lond) 2013; 124: 567–578.

78 Ma XJ, Salunga R, Tuggle JT, Gaudet J, Enright E, McQuary P et al. Geneexpression profiles of human breast cancer progression. Proc Natl Acad Sci USA2003; 100: 5974–5979.

79 Aird KM, Li H, Xin F, Konstantinopoulos PA, Zhang R. Identification ofribonucleotide reductase M2 as a potential target for pro-senescence therapy inepithelial ovarian cancer. Cell Cycle 2013; 13: 199–207.

80 Aird KM, Zhang G, Li H, Tu Z, Bitler BG, Garipov A et al. Suppression of nucleotidemetabolism underlies the establishment and maintenance of oncogene-inducedsenescence. Cell Rep 2013; 3: 1252–1265.

81 Matsushita S, Ikeda R, Fukushige T, Tajitsu Y, Gunshin K, Okumura H et al.p53R2 is a prognostic factor of melanoma and regulates proliferation andchemosensitivity of melanoma cells. J Dermatol Sci 2012; 68: 19–24.

82 Yanamoto S, Kawasaki G, Yamada S, Yoshitomi I, Yoshida H, Mizuno A.Ribonucleotide reductase small subunit p53R2 promotes oral cancer invasion viathe E-cadherin/beta-catenin pathway. Oral Oncol 2009; 45: 521–525.

83 Yanamoto S, Kawasaki G, Yoshitomi I, Mizuno A. Expression of p53R2, newlyp53 target in oral normal epithelium, epithelial dysplasia and squamous cellcarcinoma. Cancer Lett 2003; 190: 233–243.

84 Okumura H, Natsugoe S, Yokomakura N, Kita Y, Matsumoto M, Uchikado Y et al.Expression of p53R2 is related to prognosis in patients with esophagealsquamous cell carcinoma. Clin Cancer Res 2006; 12: 3740–3745.

85 Uramoto H, Sugio K, Oyama T, Hanagiri T, Yasumoto K. P53R2, p53 inducibleribonucleotide reductase gene, correlated with tumor progression of non-smallcell lung cancer. Anticancer Res 2006; 26: 983–988.

86 Jorgensen CL, Ejlertsen B, Bjerre KD, Balslev E, Nielsen DL, Nielsen KV. Geneaberrations of RRM1 and RRM2B and outcome of advanced breast cancer aftertreatment with docetaxel with or without gemcitabine. BMC Cancer 2013;13: 541.

87 Fan H, Huang A, Villegas C, Wright JA. The R1 component of mammalianribonucleotide reductase has malignancy-suppressing activity as demonstratedby gene transfer experiments. Proc Natl Acad Sci USA 1997; 94: 13181–13186.

88 Gautam A, Li ZR, Bepler G. RRM1-induced metastasis suppression throughPTEN-regulated pathways. Oncogene 2003; 22: 2135–2142.

89 Gautam A, Bepler G. Suppression of lung tumor formation by the regulatorysubunit of ribonucleotide reductase. Cancer Res 2006; 66: 6497–6502.

90 Xu X, Page JL, Surtees JA, Liu H, Lagedrost S, Lu Y et al. Broad overexpression ofribonucleotide reductase genes in mice specifically induces lung neoplasms.Cancer Res 2008; 68: 2652–2660.

91 Bepler G, Sharma S, Cantor A, Gautam A, Haura E, Simon G et al. RRM1 and PTENas prognostic parameters for overall and disease-free survival in patients withnon-small-cell lung cancer. J Clin Oncol 2004; 22: 1878–1885.

92 Zheng Z, Chen T, Li X, Haura E, Sharma A, Bepler G. DNA synthesis and repairgenes RRM1 and ERCC1 in lung cancer. N Engl J Med 2007; 356: 800–808.

93 Gong W, Zhang X, Wu J, Chen L, Li L, Sun J et al. RRM1 expression and clinicaloutcome of gemcitabine-containing chemotherapy for advanced non-small-celllung cancer: a meta-analysis. Lung Cancer 2012; 75: 374–380.

94 Nakahira S, Nakamori S, Tsujie M, Takahashi Y, Okami J, Yoshioka S et al. Invol-vement of ribonucleotide reductase M1 subunit overexpression in gemcitabineresistance of human pancreatic cancer. Int J Cancer 2007; 120: 1355–1363.

95 Ohtaka K, Kohya N, Sato K, Kitajima Y, Ide T, Mitsuno M et al. Ribonucleotidereductase subunit M1 is a possible chemoresistance marker to gemcitabine inbiliary tract carcinoma. Oncol Rep 2008; 20: 279–286.

96 Reynolds C, Obasaju C, Schell MJ, Li X, Zheng Z, Boulware D et al. Randomizedphase III trial of gemcitabine-based chemotherapy with in situ RRM1 and ERCC1protein levels for response prediction in non-small-cell lung cancer. J Clin Oncol2009; 27: 5808–5815.

97 Akita H, Zheng Z, Takeda Y, Kim C, Kittaka N, Kobayashi S et al. Significanceof RRM1 and ERCC1 expression in resectable pancreatic adenocarcinoma.Oncogene 2009; 28: 2903–2909.

98 Rodriguez J, Boni V, Hernandez A, Bitarte N, Zarate R, Ponz-Sarvise M et al.Association of RRM1 -37A>C polymorphism with clinical outcome in colorectalcancer patients treated with gemcitabine-based chemotherapy. Eur J Cancer2011; 47: 839–847.

99 Jordheim LP, Seve P, Tredan O, Dumontet C. The ribonucleotide reductase largesubunit (RRM1) as a predictive factor in patients with cancer. Lancet Oncol 2011;12: 693–702.

100 Ceppi P, Volante M, Novello S, Rapa I, Danenberg KD, Danenberg PV et al. ERCC1and RRM1 gene expressions but not EGFR are predictive of shorter survival inadvanced non-small-cell lung cancer treated with cisplatin and gemcitabine.Ann Oncol 2006; 17: 1818–1825.

101 Dong X, Hao Y, Wei Y, Yin Q, Du J, Zhao X. Response to first-line chemotherapy inpatients with non-small cell lung cancer according to RRM1 expression. PLoSOne 2014; 9: e92320.

102 Bepler G, Sommers KE, Cantor A, Li X, Sharma A, Williams C et al. Clinical efficacyand predictive molecular markers of neoadjuvant gemcitabine and pemetrexedin resectable non-small cell lung cancer. J Thorac Oncol 2008; 3: 1112–1118.

103 Jordheim LP, Guittet O, Lepoivre M, Galmarini CM, Dumontet C. Increasedexpression of the large subunit of ribonucleotide reductase is involved inresistance to gemcitabine in human mammary adenocarcinoma cells. MolCancer Ther 2005; 4: 1268–1276.

104 Davidson JD, Ma L, Flagella M, Geeganage S, Gelbert LM, Slapak CA. An increasein the expression of ribonucleotide reductase large subunit 1 is associated withgemcitabine resistance in non-small cell lung cancer cell lines. Cancer Res 2004;64: 3761–3766.

105 Bergman AM, Eijk PP, Ruiz van Haperen VW, Smid K, Veerman G, Hubeek I et al.In vivo induction of resistance to gemcitabine results in increased expression ofribonucleotide reductase subunit M1 as the major determinant. Cancer Res 2005;65: 9510–9516.

106 Bepler G, Kusmartseva I, Sharma S, Gautam A, Cantor A, Sharma A et al.RRM1 modulated in vitro and in vivo efficacy of gemcitabine and platinum innon-small-cell lung cancer. J Clin Oncol 2006; 24: 4731–4737.

107 Simon G, Sharma A, Li X, Hazelton T, Walsh F, Williams C et al. Feasibilityand efficacy of molecular analysis-directed individualized therapy in advancednon-small-cell lung cancer. J Clin Oncol 2007; 25: 2741–2746.

108 Fan H, Villegas C, Wright JA. Ribonucleotide reductase R2 component is a novelmalignancy determinant that cooperates with activated oncogenes to

Ribonucleotide reductase and cancerY Aye et al

2019

© 2015 Macmillan Publishers Limited Oncogene (2015) 2011 – 2021

determine transformation and malignant potential. Proc Natl Acad Sci USA 1996;93: 14036–14040.

109 Fan H, Villegas C, Huang A, Wright JA. The mammalian ribonucleotide reductaseR2 component cooperates with a variety of oncogenes in mechanisms of cellulartransformation. Cancer Res 1998; 58: 1650–1653.

110 Duxbury MS, Ito H, Zinner MJ, Ashley SW, Whang EE. RNA interferencetargeting the M2 subunit of ribonucleotide reductase enhances pancreaticadenocarcinoma chemosensitivity to gemcitabine. Oncogene 2004; 23:1539–1548.

111 Zhang K, Hu S, Wu J, Chen L, Lu J, Wang X et al. Overexpression of RRM2decreases thrombspondin-1 and increases VEGF production in human cancercells in vitro and in vivo: implication of RRM2 in angiogenesis. Mol Cancer 2009;8: 11.

112 Xue L, Zhou B, Liu X, Heung Y, Chau J, Chu E et al. Ribonucleotide reductasesmall subunit p53R2 facilitates p21 induction of G1 arrest under UV irradiation.Cancer Res 2007; 67: 16–21.

113 Liu X, Zhou B, Xue L, Shih J, Tye K, Lin W et al. Metastasis-suppressing potentialof ribonucleotide reductase small subunit p53R2 in human cancer cells. ClinCancer Res 2006; 12: 6337–6344.

114 Liu X, Lai L, Wang X, Xue L, Leora S, Wu J et al. Ribonucleotide reductase smallsubunit M2B prognoses better survival in colorectal cancer. Cancer Res 2011; 71:3202–3213.

115 Mannava S, Moparthy KC, Wheeler LJ, Natarajan V, Zucker SN, Fink EE et al.Depletion of deoxyribonucleotide pools is an endogenous source of DNAdamage in cells undergoing oncogene-induced senescence. Am J Pathol 2013;182: 142–151.

116 Mannava S, Moparthy KC, Wheeler LJ, Leonova KI, Wawrzyniak JA, Bianchi-Smiraglia A et al. Ribonucleotide reductase and thymidylate synthase or exo-genous deoxyribonucleosides reduce DNA damage and senescence caused byC-MYC depletion. Aging (Albany NY) 2012; 4: 917–922.

117 Reubold TF, Eschenburg S. A molecular view on signal transduction by theapoptosome. Cell Signal 2012; 24: 1420–1425.

118 Chandra D, Bratton SB, Person MD, Tian Y, Martin AG, Ayres M et al. Intracellularnucleotides act as critical prosurvival factors by binding to cytochrome C andinhibiting apoptosome. Cell 2006; 125: 1333–1346.

119 Aye Y, Long MJ, Stubbe J. Mechanistic studies of semicarbazone triapinetargeting human ribonucleotide reductase in vitro and in mammalian cells:tyrosyl radical quenching not involving reactive oxygen species. J Biol Chem2012; 287: 35768–35778.

120 Martin KR, Barrett JC. Reactive oxygen species as double-edged swords incellular processes: low-dose cell signaling versus high-dose toxicity. Hum ExpToxicol 2002; 21: 71–75.

121 Feig DI, Reid TM, Loeb LA. Reactive oxygen species in tumorigenesis. Cancer Res1994; 54: 1890s–1894s.

122 Lin ZP, Belcourt MF, Cory JG, Sartorelli AC. Stable suppression of the R2 subunitof ribonucleotide reductase by R2-targeted short interference RNA sensitizesp53(-/-) HCT-116 colon cancer cells to DNA-damaging agents and ribonucleotidereductase inhibitors. J Biol Chem 2004; 279: 27030–27038.

123 Yanamoto S, Iwamoto T, Kawasaki G, Yoshitomi I, Baba N, Mizuno A. Silencingof the p53R2 gene by RNA interference inhibits growth and enhances5-fluorouracil sensitivity of oral cancer cells. Cancer Lett 2005; 223: 67–76.

124 Yokomakura N, Natsugoe S, Okumura H, Ikeda R, Uchikado Y, Mataki Y et al.Improvement in radiosensitivity using small interfering RNA targeting p53R2 inesophageal squamous cell carcinoma. Oncol Rep 2007; 18: 561–567.

125 Guittet O, Tebbi A, Cottet MH, Vesin F, Lepoivre M. Upregulation of the p53R2ribonucleotide reductase subunit by nitric oxide. Nitric Oxide 2008; 19: 84–94.

126 Mansson E, Flordal E, Liliemark J, Spasokoukotskaja T, Elford H, Lagercrantz Set al. Down-regulation of deoxycytidine kinase in human leukemic cell linesresistant to cladribine and clofarabine and increased ribonucleotide reductaseactivity contributes to fludarabine resistance. Biochem Pharmacol 2003; 65:237–247.

127 Tang H, Xiao G, Behrens C, Schiller J, Allen J, Chow CW et al. A 12-gene setpredicts survival benefits from adjuvant chemotherapy in non-small cell lungcancer patients. Clin Cancer Res 2013; 19: 1577–1586.

128 Ferrandina G, Mey V, Nannizzi S, Ricciardi S, Petrillo M, Ferlini C et al. Expressionof nucleoside transporters, deoxycitidine kinase, ribonucleotide reductaseregulatory subunits, and gemcitabine catabolic enzymes in primaryovarian cancer. Cancer Chemother Pharmacol 2010; 65: 679–686.

129 Itoi T, Sofuni A, Fukushima N, Itokawa F, Tsuchiya T, Kurihara T et al.Ribonucleotide reductase subunit M2 mRNA expression in pretreatmentbiopsies obtained from unresectable pancreatic carcinomas. J Gastroenterol2007; 42: 389–394.

130 Fisher SB, Patel SH, Bagci P, Kooby DA, El-Rayes BF, Staley CA 3rd et al. Ananalysis of human equilibrative nucleoside transporter-1, ribonucleosidereductase subunit M1, ribonucleoside reductase subunit M2, and excision repair

cross-complementing gene-1 expression in patients with resected pancreasadenocarcinoma: implications for adjuvant treatment. Cancer 2013; 119:445–453.

131 Okumura H, Natsugoe S, Matsumoto M, Mataki Y, Takatori H, Ishigami S et al. Thepredictive value of p53, p53R2, and p21 for the effect of chemoradiation therapyon oesophageal squamous cell carcinoma. Br J Cancer 2005; 92: 284–289.

132 Stover PJ, Weiss RS. Sensitizing cancer cells: is it really all about U? Cancer Cell2012; 22: 3–4.

133 Zhou BB, Elledge SJ. The DNA damage response: putting checkpoints inperspective. Nature 2000; 408: 433–439.

134 Shao J, Zhou B, Chu B, Yen Y. Ribonucleotide reductase inhibitors and futuredrug design. Curr Cancer Drug Targets 2006; 6: 409–431.

135 Nocentini G. Ribonucleotide reductase inhibitors: new strategies for cancerchemotherapy. Crit Rev Oncol Hematol 1996; 22: 89–126.

136 Stubbe J, van Der Donk WA. Protein radicals in enzyme catalysis. Chem Rev 1998;98: 705–762.

137 Manegold C, Zatloukal P, Krejcy K, Blatter J. Gemcitabine in non-small cell lungcancer (NSCLC). Invest New Drugs 2000; 18: 29–42.

138 Stubbe J, van der Donk WA. Ribonucleotide reductases: radical enzymes withsuicidal tendencies. Chem Biol 1995; 2: 793–801.

139 Wang J, Lohman GJ, Stubbe J. Enhanced subunit interactions with gemcitabine-5'-diphosphate inhibit ribonucleotide reductases. Proc Natl Acad Sci USA 2007;104: 14324–14329.

140 Bonate PL, Arthaud L, Cantrell WR Jr, Stephenson K, Secrist JA 3rd, Weitman S.Discovery and development of clofarabine: a nucleoside analogue fortreating cancer. Nat Rev Drug Discov 2006; 5: 855–863.

141 Saven A, Burian C, Koziol JA, Piro LD. Long-term follow-up of patients with hairycell leukemia after cladribine treatment. Blood 1998; 92: 1918–1926.

142 Walker S, Palmer S, Erhorn S, Brent S, Dyker A, Ferrie L et al. Fludarabinephosphate for the first-line treatment of chronic lymphocytic leukaemia. HealthTechnol Assess 2009; 13 (Suppl 1): 35–40.

143 Cohen A, Hirschhorn R, Horowitz SD, Rubinstein A, Polmar SH, Hong R et al.Deoxyadenosine triphosphate as a potentially toxic metabolite in adenosinedeaminase deficiency. Proc Natl Acad Sci USA 1978; 75: 472–476.

144 Sigal DS, Miller HJ, Schram ED, Saven A. Beyond hairy cell: the activity ofcladribine in other hematologic malignancies. Blood 2010; 116: 2884–2896.

145 Avramis VI, Plunkett W. 2-Fluoro-ATP: a toxic metabolite of 9-beta-D-arabinosyl-2-fluoroadenine. Biochem Biophys Res Commun 1983; 113: 35–43.

146 Rofougaran R, Vodnala M, Hofer A. Enzymatically active mammalian ribonu-cleotide reductase exists primarily as an α6β2 octamer. J Biol Chem 2006; 281:27705–27711.

147 Alvarez-Salas LM. Nucleic acids as therapeutic agents. Curr Top Med Chem 2008;8: 1379–1404.

148 Spasokoukotskaja T, Sasvari-Szekely M, Hullan L, Albertioni F, Eriksson S,Staub M. Activation of deoxycytidine kinase by various nucleoside analogues.Adv Exp Med Biol 1998; 431: 641–645.

149 Gonzalez H, Leblond V, Azar N, Sutton L, Gabarre J, Binet JL et al. Severe auto-immune hemolytic anemia in eight patients treated with fludarabine. HematolCell Ther 1998; 40: 113–118.

150 King KM, Damaraju VL, Vickers MF, Yao SY, Lang T, Tackaberry TE et al.A comparison of the transportability, and its role in cytotoxicity, of clofarabine,cladribine, and fludarabine by recombinant human nucleoside transportersproduced in three model expression systems. Mol Pharmacol 2006; 69: 346–353.

151 Lindemalm S, Liliemark J, Juliusson G, Larsson R, Albertioni F. Cytotoxicity andpharmacokinetics of cladribine metabolite, 2-chloroadenine in patients withleukemia. Cancer Lett 2004; 210: 171–177.

152 Ferracin M, Zagatti B, Rizzotto L, Cavazzini F, Veronese A, Ciccone M et al.MicroRNAs involvement in fludarabine refractory chronic lymphocytic leukemia.Mol Cancer 2010; 9: 123.

153 Pfisterer J, Vergote I, Du Bois A, Eisenhauer E. Combination therapy withgemcitabine and carboplatin in recurrent ovarian cancer. Int J Gynecol Cancer2005; 15 (Suppl 1): 36–41.

154 Sandler A, Ettinger DS. Gemcitabine: single-agent and combination therapy innon-small cell lung cancer. Oncologist 1999; 4: 241–251.

155 Plunkett W, Liliemark JO, Adams TM, Nowak B, Estey E, Kantarjian H et al.Saturation of 1-beta-D-arabinofuranosylcytosine 5'-triphosphate accumulation inleukemia cells during high-dose 1-beta-D-arabinofuranosylcytosine therapy.Cancer Res 1987; 47: 3005–3011.

156 Spasokoukotskaja T, Sasvari-Szekely M, Keszler G, Albertioni F, Eriksson S,Staub M. Treatment of normal and malignant cells with nucleoside analoguesand etoposide enhances deoxycytidine kinase activity. Eur J Cancer 1999; 35:1862–1867.

157 Faderl S, Gandhi V, O'Brien S, Bonate P, Cortes J, Estey E et al. Results of a phase1-2 study of clofarabine in combination with cytarabine (ara-C) in relapsed andrefractory acute leukemias. Blood 2005; 105: 940–947.

Ribonucleotide reductase and cancerY Aye et al

2020

Oncogene (2015) 2011 – 2021 © 2015 Macmillan Publishers Limited

158 Sigmond J, Bergman AM, Leon LG, Loves WJ, Hoebe EK, Peters GJ Staurosporineincreases toxicity of gemcitabine in non-small cell lung cancer cells: role ofprotein kinase C, deoxycytidine kinase and ribonucleotide reductase. AnticancerDrugs 2010; 21: 591–599.

159 Hehlmann R, Heimpel H, Hasford J, Kolb HJ, Pralle H, Hossfeld DK et al.Randomized comparison of busulfan and hydroxyurea in chronic myelogenousleukemia: prolongation of survival by hydroxyurea. The German CMLStudy Group. Blood 1993; 82: 398–407.

160 Sterkers Y, Preudhomme C, Lai JL, Demory JL, Caulier MT, Wattel E et al.Acute myeloid leukemia and myelodysplastic syndromes following essentialthrombocythemia treated with hydroxyurea: high proportion of cases with 17pdeletion. Blood 1998; 91: 616–622.

161 Levin VA. The place of hydroxyurea in the treatment of primary brain tumors.Semin Oncol 1992; 19: 34–39.

162 Graslund A, Ehrenberg A, Thelander L. Characterization of the free radical ofmammalian ribonucleotide reductase. J Biol Chem 1982; 257: 5711–5715.

163 Wright JA, Alam TG, McClarty GA, Tagger AY, Thelander L. Altered expression ofribonucleotide reductase and role of M2 gene amplification in hydroxyurea-resistant hamster, mouse, rat, and human cell lines. Somat Cell Mol Genet 1987;13: 155–165.

164 McClarty GA, Chan AK, Engström Y, Wright JA, Thelander L. Elevated expressionof M1 and M2 components and drug-induced posttranscriptional modulation ofribonucleotide reductase in a hydroxyurea-resistant mouse cell line. Biochemistry1987; 26: 8004–8011.

165 Temperini C, Innocenti A, Scozzafava A, Supuran CT. N-hydroxyurea—a versatilezinc binding function in the design of metalloenzyme inhibitors. Bioorg MedChem Lett 2006; 16: 4316–4320.

166 Nyholm S, Thelander L, Graslund A. Reduction and loss of the iron center in thereaction of the small subunit of mouse ribonucleotide reductase with hydro-xyurea. Biochemistry 1993; 32: 11569–11574.

167 Sahlin M, Graslund A, Petersson L, Ehrenberg A, Sjoberg BM. Reduced forms ofthe iron-containing small subunit of ribonucleotide reductase from Escherichiacoli. Biochemistry 1989; 28: 2618–2625.

168 Nutting CM, van Herpen CM, Miah AB, Bhide SA, Machiels JP, Buter J et al. PhaseII study of 3-AP Triapine in patients with recurrent or metastatic head and necksquamous cell carcinoma. Ann Oncol 2009; 20: 1275–1279.

169 Finch RA, Liu M, Grill SP, Rose WC, Loomis R, Vasquez KM et al. Triapine(3-aminopyridine-2-carboxaldehyde- thiosemicarbazone): a potent inhibitorof ribonucleotide reductase activity with broad spectrum antitumor activity.Biochem Pharmacol 2000; 59: 983–991.

170 Cory JG, Cory AH, Rappa G, Lorico A, Liu MC, Lin TS et al. Inhibitors of ribonu-cleotide reductase. Comparative effects of amino- and hydroxy-substitutedpyridine-2-carboxaldehyde thiosemicarbazones. Biochem Pharmacol 1994; 48:335–344.

171 Chaston TB, Lovejoy DB, Watts RN, Richardson DR. Examination of the anti-proliferative activity of iron chelators: multiple cellular targets and the differentmechanism of action of triapine compared with desferrioxamine and the potentpyridoxal isonicotinoyl hydrazone analogue 311. Clin Cancer Res 2003; 9:402–414.

172 Popovic-Bijelic A, Kowol CR, Lind ME, Luo J, Himo F, Enyedy EA et al.Ribonucleotide reductase inhibition by metal complexes of Triapine(3-aminopyridine-2-carboxaldehyde thiosemicarbazone): a combined experi-mental and theoretical study. J Inorg Biochem 2011; 105: 1422–1431.

173 Zhu L, Zhou B, Chen X, Jiang H, Shao J, Yen Y. Inhibitory mechanisms ofheterocyclic carboxaldehyde thiosemicabazones for two forms of humanribonucleotide reductase. Biochem Pharmacol 2009; 78: 1178–1185.

174 Avolio TM, Lee Y, Feng N, Xiong K, Jin H, Wang M et al. RNA interferencetargeting the R2 subunit of ribonucleotide reductase inhibits growth of tumorcells in vitro and in vivo. Anticancer Drugs 2007; 18: 377–388.

175 Wonganan P, Chung WG, Zhu S, Kiguchi K, Digiovanni J, Cui Z. Silencing ofribonucleotide reductase subunit M1 potentiates the antitumor activity ofgemcitabine in resistant cancer cells. Cancer Biol Ther 2012; 13: 908–914.

176 Oh YK, Park TG. siRNA delivery systems for cancer treatment. Adv Drug Deliv Rev2009; 61: 850–862.

177 Reischl D, Zimmer A. Drug delivery of siRNA therapeutics: potentials and limits ofnanosystems. Nanomedicine 2009; 5: 8–20.

178 Lee Y, Vassilakos A, Feng N, Lam V, Xie H, Wang M et al. GTI-2040, an antisenseagent targeting the small subunit component (R2) of human ribonucleotidereductase, shows potent antitumor activity against a variety of tumors. CancerRes 2003; 63: 2802–2811.

179 Rahman MA, Amin AR, Wang X, Zuckerman JE, Choi CH, Zhou B et al. Systemicdelivery of siRNA nanoparticles targeting RRM2 suppresses head and necktumor growth. J Control Release 2012; 159: 384–392.

180 Duxbury MS, Ito H, Benoit E, Zinner MJ, Ashley SW, Whang EE. Retrovirallymediated RNA interference targeting the M2 subunit of ribonucleotidereductase: a novel therapeutic strategy in pancreatic cancer. Surgery 2004; 136:261–269.

181 Reichard P, Baldesten A, Rutberg L. Formation of deoxycytidine phosphates fromcytidine phosphates in extracts from Escherichia coli. J Biol Chem 1961; 236:1150–1157.

Supplementary Information accompanies this paper on the Oncogene website (http://www.nature.com/onc)

Ribonucleotide reductase and cancerY Aye et al

2021

© 2015 Macmillan Publishers Limited Oncogene (2015) 2011 – 2021