A Role for DNA Mismatch Repair in Sensing and Responding to Fluoropyrimidine Damage

A role for DNA mismatch repair in sensing and responding to ﬂuoropyrimidine damage

Mark Meyers1, Arlene Hwang1, Mark W Wagner1, Andrew J Bruening1, Martina L Veigl2, W David Sedwick3 and David A Boothman*,1

1Laboratory of Molecular Stress Responses, Department of Radiation Oncology, Case Western Reserve University, Biomedical Research Building 326-East, 2109 Adelbert Road, Cleveland, OH 44106-4942, USA; 2Department of General Medical Sciences (Oncology), Case Western Reserve University, Cleveland, Cleveland, OH 44106-4942, USA; 3Department of Medicine, 11001 Cedar Road, Ireland Cancer Center, University Hospitals of Cleveland, Cleveland, OH 44106, USA

The phenomenon of damage tolerance, whereby cells incur antimetabolite was devised in which a fluorine atom was DNA lesions that are nonlethal, largely ignored, but substituted for the hydrogen atom at the fifth position of highly mutagenic, appears to play a key role in uracil; it was known that fluorine-substituted organic carcinogenesis. Typically, these lesions are generated by compounds showed increased toxicity compared to their alkylation of DNA or incorporation of base analogues. normal counterparts. Since the atomic radius of fluorine This tolerance is usually a result of the loss of specific is similar to hydrogen (1.35 A˚ compared to 1.2 A˚ ), it was DNA repair processes, most often DNA mismatch repair anticipated and indeed confirmed that 5-FU was (MMR). The availability of genetically matched MMR- metabolized in a similar manner as uracil. Once 5-FU deficient and -corrected cell systems allows dissection of was found to have chemotherapeutic activity, its the consequences of this unrepaired damage in carcino- corresponding deoxyribonucleoside derivative 5-fluoro- genesis as well as the elucidation of cell cycle checkpoint 20-deoxyuridine (FdUrd) was developed (Heidelberger responses and cell death consequences. Recent data et al., 1958). 5-FU and FdUrd, used in combination indicate that MMRplays an important role in detecting withotheragents, are now themost widely used drugs in damage caused by fluorinated pyrimidines (FPs) and the treatment of advanced colorectal cancer (CRC) as represents a repair system that is probably not the well as many other cancers. primary system for detecting damage caused by these Fluoropyrimidines (FPs) require uptake and conver- agents, but may be an important system for correcting key sion to their active forms before exerting their cytotoxic mutagenic lesions that could initiate carcinogenesis. In effects. 5-FU shares the same facilitated transport fact, clinical studies have shown that there is no benefit of system as uracil, adenine, and hypoxanthine, whereas FP-based adjuvant chemotherapy in colon cancer patients FdUrd enters the cell by a distinct facilitated membrane exhibiting microsatellite instability, a hallmark of MMR transport mechanism used by purine and pyrimidine deficiency. MMR-mediated damage tolerance and futile nucleosides (Domin et al., 1993; Johnston et al., 1996). cycle repair processes are discussed, as well as possible FPs are then converted into fluorinated ribonucleotides strategies using FPs to exploit these systems for improved and deoxyribonuclotides using the same pathways as anticancer therapy. those used by uracil and thymine (Figure 1). FPs have Oncogene (2003) 22, 7376–7388. doi:10.1038/sj.onc.1206941 three possible mechanisms of action that are exerted by three different metabolites: FdUrd-50-monophosphate Keywords: 5-fluorouracil; 5-fluoro-20-deoxyuridine; (FdUMP) inhibits DNA synthesis by blocking the 5-fluoro-20-deoxycytidine; futile cycles of repair; DNA activity of thymidylate synthase (TS), 5-fluorouridine- damage tolerance; colon cancer treatment 50-triphosphate (FUTP) is incorporated into RNA, and FdUrd-50-triphosphate (FdUTP) is incorporated into DNA. The relative contribution of each mechanism to cytotoxicity depends on the FP used, the concentration Fluoropyrimidine metabolism and duration of exposure, and the system being studied. These mechanisms will be briefly discussed below. 5-Fluorouracil (5-FU) was logically designed by Hei- delberger et al. (1957), as a potential tumor-inhibitory drug. There are many excellent reviews about 5-FU FP effects on TS activity and nucleotide pools available (Weckbecker, 1991; van Laar et al., 1998). The TS catalyses the central reaction in the de novo synthesis rationale for its synthesis was based on the enhanced of thymine nucleosides and nucleotides (Friedkin, 1973). utilization of uracil as a precursor of DNA pyrimidines It directs the synthesis of thymidylate (dTMP) from in a series of transplantable tumors. In this strategy, an deoxyuridylate (dUMP) by transferring a methyl group from its cofactor (5,10-methylene tetrahydrofolate; *Correspondence: DA Boothman; E-mail: [email protected] THF) to the carbon-5 position of dUMP. Normally, a Role of MMR in response to fluoropyrimidines M Meyers et al 7377 UK RNAP controlled by dCMP deaminase (dCMPD) or dCTP FUrd FUMP FUDP FUTP F·RNA 1

RR deaminase; the activity of these deaminases is stimulated by dCTP and inhibited by dTTP (Reichard, 1988). DNAP FdUDP FdUTP F·DNA 2 However, the overall consequence of these interacting UP OPRT processes usually results in the blockage of dTTP dUMP synthesis by FPs, which generally leads to dTTP and THF dGTP depletion and expansion of dCTP, dATP, and TP TK 5-FU FdUrd FdUMP TS 3 dUTP pools (Maybaum et al., 1981; Sedwick et al., 1981). For example, treatment withtheantifolate DHF dTMP methotrexate (an inhibitor of dihydrofolate reductase, the enzyme that generates the THF needed by TS) leads to a decrease in intracellular dTTP of almost two orders dTTP of magnitude witha concomitant increase in intracellular dUTP of approximately 1000-fold (Goulian et al., DNA 1980a). Most significantly, the relative and absolute concentrations of dNTPs are critical for determining the Figure 1 The FP metabolic pathway showing its three major pathways of cytotoxicity. Thymidine phosphorylase (TP) catalyses fidelity of DNA replication (Kunz et al., 1994). the reversible conversion of 5-FU to FdUrd. Pathway 1: In this Substitution errors by DNA polymerase increase as RNA-directed pathway, 5-FU is converted by uridine phosphor- the concentration of the correct dNTP decreases and ylase (UP) to 5-fluorouridine (FUrd), which is then converted to that of an incorrect dNTP increases (Roberts and FUrd 50-monophosphate (FUMP) by uridine kinase (UK). Alternatively, FUMP is formed directly from 5-FU by orotic acid Kunkel, 1988). Additionally, dNTP perturbations can phosphoribosyl transferase (OPRT). Further phosphorylations of cause template–primer misalignments that lead to FUMP result in the formation of FUrd 50-diphosphate (FUDP) frameshift mutations (Bebenek et al., 1992). and FUrd 50-triphosphate (FUTP); FUTP is a substrate for RNA polymerase (RNAP). Pathway 2: In this DNA-directed pathway, FdUrd is phosphorylated by thymidine kinase to form FdUMP, FP effects on RNA metabolism and processing which is further phosphorylated to form FdUrd 50-diphosphate (FdUDP). In an indirect pathway, FdUDP can also be formed FPs can cause general cytotoxicity after conversion to from FUDP by the action of ribonucleotide reductase (RR). A FUTP by their incorporation into RNA in a reaction further phosphorylation of FdUDP yields FdUTP, which is a catalysed by RNA polymerase. FPs can be heavily substrate for DNA polymerase (DNAP). Pathway 3: Thymidylate synthase (TS) normally catalyses the methylation of 20-deoxyur- incorporated into bothnuclear and cytoplasmic RNA idine 50-monophosphate (dUMP) to form 20-deoxythymidine species and can interfere withnormal RNA processing 50-monophosphate (dTMP) using the reduced folate THF as a and function (Johnston et al., 1996; Grem, 1997). Net methyl donor. dTMP undergoes sequential phosphorylations to RNA synthesis is inhibited during and following FP 0 0 form 2 -deoxythymidine 5 -triphosphate (dTTP), which is needed exposure and can be accompanied by alterations in for DNA synthesis. In this DNA-directed pathway, FdUMP is able to stabilize the covalent ternary complex consisting of TS, FdUMP, protein levels (Grem, 1996). The detrimental effect of FP and THF and thus prevent TS from completing its reaction. As a incorporation is different among individual RNA result, dTTP pools become depleted and DNA synthesis is halted species. The synthesis of ribosomal RNA appears to be the most susceptible, followed by polyadenylated RNA and then transfer RNA (Glazer and Hartman, 1981; Weckbecker, 1991). Specifically, FPs inhibit the short-lived, covalent ternary complex consisting of TS, conversion of high Mr nuclear RNA species to lower Mr dUMP, and THF is formed during this reaction. ribosomal RNA (Kanamaru et al., 1986), decrease the FdUMP, which is formed from the phosphorylation of stability of messenger RNA by their inhibition of FdUrd by thymidine kinase, is a suicide inhibitor of TS. polyadenylation (Grem, 1996), interfere withnormal The presence of a fluorine atom instead of a hydrogen at splicing by incorporating into uracil-richsmall nuclear the carbon-5 position of the uracil ring in FdUMP RNA species (Doong and Dolnick, 1988), and inhibit obstructs this reaction (TS cannot break the carbon– transfer RNA function by forming covalent complexes fluorine bond to allow the methylation), thereby withenzymes involved in thepost-translational mod- hindering the elimination of THF from the ternary ification of uracil residues (Santi and Hardy, 1987). In complex and greatly stabilizing this complex (Santi and some tissue culture and in vivo models, the extent of FP McHenry, 1972). As a result, dTMP and ultimately incorporation correlates withcytotoxicity (Kufe and dTTP pools are depleted, resulting in inhibition of DNA Major, 1981; Glazer and Lloyd, 1982). synthesis (Fernandes and Cranford, 1986). The depletion of dTTP has other effects on the cell Effects of FP incorporation into DNA (Kunz, 1996). Through metabolic interconversions and allosteric regulatory mechanisms, fluctuations in any Historically, the incorporation of FdUTP into DNA had specific deoxyribonucleoside triphosphate (dNTP) may been difficult to demonstrate. The availability of tritiated trigger alterations in nucleoside or nucleotide pools. For FPs of high specific activity allowed detection of the low example, dTTP inhibits CDP and UDP reduction, but levels of FPs in DNA. Kufe et al. (1981), exposed mouse stimulates reduction of GDP (Reichard, 1985). Addi- L1210 leukemia cells to 0.1–10 mm [3H]FdUrd for 4–12 h, tionally, the relative dCTP and dTTP levels are separated the nucleic acids by cesium sulfate density-

Oncogene Role of MMR in response to fluoropyrimidines M Meyers et al 7378 gradient centrifugation, digested the DNA to nucleoside growth inhibition has not been characterized in DNA monophosphates and nucleosides, and subjected them repair-deficient E. coli. However, it is clear that to high-performance liquid chromatography separation misincorporated dUTP can directly interfere withthe to show that FdUrd moieties were present in DNA. activity of specific regulatory proteins that bind DNA, Shortly thereafter, similar levels of [3H]5-FU and as exemplified directly by the abrogation of function due [3H]FdUrd incorporation into the DNA of human to a single dUTP substitution in a critical binding site at MCF-7 breast cancer cells was demonstrated (Major position 13 in the lac repressor (Fisher and Caruthers, et al., 1982). Likewise, FP incorporation into mouse 1979). In spite of the relatively low toxic impact of Lewis lung carcinoma and adenocarcinoma-755 tumor dUTP substitution for dTTP in some organisms, dUTP cells was noted withan estimated 100 FP moieties misincorporation has the potential to kill mammalian incorporated per genome (Boothman et al., 1987a, b). cells if delivered at high concentrations. Two mechanisms that limit the extent of FP incorpora- 5-FU differs from uracil in that it has a fluorine atom tion into DNA are discussed below. at the carbon-5 position instead of hydrogen; analo- gously, thymine differs from uracil in that it has a FP-mediated DNA damage methyl group at this position. In standard Watson– Crick hydrogen-bonding patterns between adenine and Two of the three mechanisms of FP-mediated cytotoxi- thymine, the keto group at the fourth position of city act at the level of DNA. Since both the disruption of thymine and the hydrogen of the amino group at the dNTP pools as a result of TS inhibition and the direct third position of thymine are involved in base pairing; incorporation of FPs into DNA occur concomitantly, it the presence of a fluorine or hydrogen atom at carbon-5 is often difficult to distinguishtherelative contribution does not disrupt this hydrogen bonding. Thus, incor- of one mechanism from the other. In any case, DNA- poration of the dTTP analogue FdUTP into DNA directed effects of FPs can arise from the following typically results in A:5-FU base pairs. In fact, it has situations: been shown by ultraviolet spectroscopic melting experiments that an A:5-FU base pair is slightly more stable (a) Direct incorporation of the deoxyribonucleotide than an A:T base pair; this is presumably due to the analogue FdUTP into DNA. unique properties (electronegativity, hydrophobicity, et al (b) Direct incorporation of deoxyribonucleotide analo- and small size) of the fluorine atom (Habener ., et al gue dUTP into DNA due to increased dUMP (and 1988; Coll ., 1989). However, the presence of the hence dUTP) pools following TS inhibition. electron-withdrawing fluorine atom on the pyrimidine (c) Enhanced incorporation of the natural DNA deox- ring could be expected to result in the presence of yribonucleotides (dATP, dCTP, dGTP, dTTP) as ionized 5-FU in DNA at physiological pH, which could mispairs due to decreased fidelity of DNA polymer- lead to mispairing withguanine during replication et al ase brought about by dNTP pool imbalances. (Freese, 1959; Kremer ., 1987). In fact, just as (d) Inhibition of DNA synthesis due to DNA poly- A:5-FU base pairs were found to form pH independent et al merase stalling as a result of dNTP pool imbalances. Watson–Crick structures (Fazakerley ., 1987; et al et al (e) Interference of the activity of DNA repair enzymes Kremer ., 1987; Sowers ., 1987), ionization of due to dNTP pool imbalances. fluorine also allows 5-FU to adopt a Watson–Crick structure when forming a mispair with guanine (Sowers et al., 1988, 1989). Since FdUrd moities are capable of The two most important consequences of these events Watson–Crick mispairing withguanine whenincorpo- are the potential mutagenic effects of base analogues/ rated into DNA, treatment of cells withFdUrd hasbeen mispairs in DNA and the fragmentation of DNA shown to be both mutagenic (Aebersold, 1979) and created in the cell’s attempts to repair these lesions. oncogenic (Jones et al., 1976) as expected. The first consequence of FP treatment is mutagenesis. The combined effects of dNTP pool imbalance and Misincorporation of dUTP in place of dTTP occurs at FdUTP incorporation into DNA also have several high frequency in many organisms and can reach almost consequences on the structural integrity of DNA (Grem, full substitution without inhibiting replication of some 1997). Treatment of cells withFPs results in thegeneration viruses (Takahashi and Marmur, 1963). dUTP incorpo- of DNA single-strand breaks (SSBs) and DNA double- rated in place of dTTP is not necessarily miscoding; for strand breaks (DSBs) (Lo¨ nn and Lo¨ nn, 1984; Schuetz example, some bacterial viruses suchas PBS1 and PBS2 et al., 1984; Yoshioka et al., 1987). Thus, DNA appears to from B. subtilis (Takahashi and Marmur, 1963) have become fragmented when it contains non-DNA bases like evolved that tolerate full substitution of dUTP for dTTP uracil and 5-FU (Weckbecker, 1991). The active removal in DNA. Furthermore, up to 90% substitution of dUTP of these bases from DNA by the normal DNA repair for dTTP was demonstrated under conditions of limited mechanisms (see below) is thought to lead to this growthin multiple mutants of Escherichia coli (el-Hajj fragmentation and contribute to the cytotoxicity of FPs et al., 1992). E. coli normally maintains as muchas one (Caradonna and Cheng, 1980; Ingraham et al., 1980, 1982; dUTP molecule per 200 nucleotides, but the steady-state Grem et al., 1989). In fact, cotreatment of human CRC level of dUTP in DNA of mammalian cells is cells with5-FU and a DNA polymerase a inhibitor approximately 10 000 times lower (Goulian et al., (aphidicolin) blocked incorporation-dependent DNA frag- 1980b; Nilsson et al., 1980). Specific damage leading to mentation, suggesting that the formation of these DNA

Oncogene Role of MMR in response to fluoropyrimidines M Meyers et al 7379 breaks was a function of normal DNA repair processes DNA BER (Lo¨ nn et al., 1989). Additionally, fragmentation may occur via compromised repair processes; proper DNA BER consists of several classes of DNA glycosylases repair requires the availability of all dNTPs at sufficient that recognize abnormal and some mismatched concentrations (Parker et al., 1987). However, DNA DNA bases that cause minor structural changes in et al fragmentation following FP exposure can also occur in DNA (Krokan ., 2000; Norbury and Hickson, the absence of detectable FdUTP incorporation into DNA 2001). In the primary pathway of BER, a glycosylase N (Parker et al., 1987), and one group reported a correlation catalyses the hydrolytic cleavage of the -glycosyl bond between the depletion of dNTP pools and DNA damage linking the base to the sugar. This enzymatic cleavage (Yoshioka et al., 1987). generates an apurinic/apyrimidinic (AP) site and releases the free base from DNA. AP endonucleases and phosphodiesterases then generate a single nucleotide gap containing 30-hydoxyl- and 50-phosphate- Removal of uridine and FP moieties from DNA termini that permit DNA polymerase to fill the gap. Finally, a DNA ligase seals the remaining nick to Mechanisms to detect and remove uridine from DNA complete the repair event. Enzymes that remove uracil from DNA are collec- Uracil (as dUTP) commonly occurs in DNA. tively called UDGs (Krokan et al., 2001). These enzymes Two distinct mechanisms are responsible for dUTP apparently detect changes in DNA tertiary structure incorporation into DNA, namely the deamination caused by the increased potential for base stacking of of cytosine to uracil and the direct incorporation of thymine versus uracil. These structural differences, dUTP in place of dTTP as a result of dramatic reflected in higher thermal stability of dA-dT versus nucleotide pool imbalances. First, spontaneous deami- dA-dU copolymers (Gill et al., 1974), appear to be read nation is a major problem facing the cell. It has been estimated that approximately 100–500 cytosine residues by the glycosylases in the major and minor grooves of DNA as they scan double-stranded DNA (dsDNA) for undergo spontaneous deamination to become uracil per uracil moieties. UDG encoded by the UNG gene day in each human diploid cell (Lindahl, 1993). The accounts for the majority of the total UDG activity. conversion of cytosine into uracil, which is a fully UDG is found in bacteria, yeast, plants, and mamma- competent base-pairing partner for adenine, causes a lian cells, and is highly conserved in evolution. It is G:C to A:T transition in half of the progeny on believed that its primary function is to remove uracil replication. Second, the abundance of UTP in the cell from G:U mispairs resulting from cytosine deamination, required for RNA synthesis as well as modulation of although it also removes uracil from A:U base pairs UTP and dUTP pools through exogenous treatments resulting from the misincorporation of dUMP during causes DNA polymerase to misincorporate dUTP into replication. It is also very effective at removing uracil DNA. In fact, misincorporation of dUTP in place of from single-stranded DNA (ssDNA). UDG is highly dTTP is a muchmore common occurrence than specific for uracil in DNA (probably because of an deamination of cytosine to uracil (Richards et al., 1984). These events presumably place human cells under evolutionary requirement to distinguishbetween uracil and thymine moieties in DNA, which are very similar in high mutational pressure and may have provided the structure). In this regard, it shows negligible activity evolutionary impetus for the development of over- towards the natural DNA bases cytosine or thymine in lapping mechanisms that can detect uracil moieties DNA. However, it does not distinguishbetween the and remove them from DNA. structural similarities of uracil and 5-FU, and conse- As a first line of defense to keep dUTP out of DNA, quently UDG is also effective at removing 5-FU from cells maintain an abundant level of dUTP dipho- DNA (see Table 1). sphohydrolase (dUTPase), an enzyme that effectively Bacterial and mammalian cells also have DNA prevents accumulation of dUTP pools available to DNA glycosylases specific for the repair of different types of polymerase, unless over-run by very high substrate levels (Caradonna and Adamkiewicz, 1984). Furthermore, single-base mismatches. G:T mismatch-specific thymine- DNA base excision repair (BER) mechanisms appear to be primarily responsible for maintenance of uracil- free DNA. BER is normally an error-free mechanism. Table 1 Human DNA glycosylases that recognize 5-FU moieties The initial recognition/excision enzymes, uracil–DNA in DNA glycosylases (UDGs), responsible for this process are Namea Chromosome Known substratesb among the most studied DNA repair enzymes at the location mechanistic and structural levels (Parikh et al., 2000; in humans Pearl, 2000) and will only be briefly described below. UNG1 12q23–q24.1 ssU, U:G, 5-FU:G, U:A, 5-FU:A Predictably, bothdUTPase and many UDGs are active TDG 12q24.1 ss5-FU, 5-FU:G, 5-FU:A, U:G, against 5-FU moieties in DNA as well. While the U:C, U:T, T:G, T:C, T:T MBD4/MED1 3q21–q22 U, 5-FU, or T in U/5-FU/ enzyme-catalysed hydrolysis of dUTP and FdUTP are TpG:CpG/5-meCpG equally efficient, the affinity of human UDG for 5-FU SMUG1 12q13.1–q14 ssU, U:A, U:G moieties in DNA is 17-fold lower than for uracil (Mauro et al., 1993). ass ¼ single stranded

Oncogene Role of MMR in response to fluoropyrimidines M Meyers et al 7380 DNA glycosylase (TDG) removes thymine from DNA MMR G:T mispairs in a CpG context, although G:T mispairs are found in other sequence contexts and thymine Another important DNA repair pathway in the cell is opposite O6-methylguanine, cytosine, and thymine MMR. It eliminates incorrect base pairs and insertion/ are also substrates (Hardeland et al., 2001). Interest- deletion loops that arise during DNA replication, ingly, TDG excises uracil from G:U mispairs thereby maintaining the integrity of the genome (Fishel, more efficiently than it excises thymine from G:T 1998; Kolodner and Marsischky, 1999). The MMR mispairs, whereas neither uracil nor thymine in ssDNA pathway of E. coli is well characterized and dependent or A:U are substrates. BothTDG and MUG (mismatch on MutH, MutL, and MutS proteins (Modrichand uracil-DNA glycosylase, the homologous protein in Lahue, 1996; Modrich, 1997). It uses the methylation bacteria) have been shown to remove 5-FU from DNA status of the DNA strands to correctly repair only to the (Hardeland et al., 2001; Liu et al., 2002). The influence nascent (i.e. transiently unmethylated) strand. The of human TDG on 5-FU incorporation has not yet been proteins involved in mammalian DNA MMR share examined. significant structural similarities to bacterial MMR A human DNA repair protein named methyl- proteins. There are many comprehensive reviews on CpG-binding domain 4 (MBD4) was first identified bacterial MMR and its comparisons withthemore as an MBD-containing protein witha region of complicated MMR systems of eucaryotes (Hsieh, 2001; similarity to bacterial DNA repair enzymes. It was Peltomaki, 2001). Here, we will focus not on general also independently cloned as methyl-CpG-binding MMR processes, but rather on damage detection, endonuclease 1 by its interaction in a two-hybrid tolerance, and cell responses to damage initiated by screen withtheDNA mismatchrepair (MMR) protein MMR processes. MLH1 (although this endonuclease activity now The process of MMR occurs in several principal appears to be insignificant) (Bellacosa et al., 1999; steps: mismatch/loop recognition and assembly of Hendrich et al., 1999). MBD4 acts as a G:T and MMR proteins, degradation of the error-containing G:U mismatch-specific thymine and uracil glycosylase. strand in a strand-specific manner, unscheduled The uracil glycosylase activity of MBD4 is limited DNA synthesis, and finally ligation. A heterodimer of to G:U mismatches and does not remove uracil MSH2-MSH6 or MSH2-MSH3 (MutS homologues) present in ssDNA. MBD4 prefers G:T and G:U is responsible for mismatchrecognition (Modrich mismatches located in the context of methylated and Lahue, 1996). MSH2-MSH6 heterodimers or unmethylated CpG sites; since, these mismatches detect mispairs and small loops; MSH2-MSH3 hetero- can originate via spontaneous hydrolytic deamination dimers primarily detect small loops. ADP to ATP of 5-methylcytosine and cytosine to thymine and nucleotide exchange occurs and induces a conforma- uracil, respectively, it appears that MBD4 is involved tional change in these MutS heterodimers that reduces in the repair of deaminated 5-methylcytosine its affinity for the damage site and allows the complex to and cytosine at CpG sites. It was also shown diffuse freely along the DNA in an ATP hydrolysis- that MBD4 efficiently removes 5-FU in the context independent manner (Fishel, 1999; Gradia et al., 1999; of G:5-FU mismatches (Petronzelli et al., 2000). Wilson et al., 1999). This step is referred to as the In fact, transfection of an MBD4 mutant lacking ‘sliding clamp’. Following lesion detection, DNA its MBD domain into cells was found to be lesion excision then requires the interaction of the associated withmicrosatellite instability (MSI), a MLH1-PMS2 heterodimer (MutL-related proteins) with hallmark of MMR deficiency (Bellacosa et al., 1999) the MSH2-MSH3/6 complexes. The details of (see below). this mechanism are still poorly understood. The Single-strand-selective monofunctional UDG MLH1-containing complex presumably acts as a ‘mo- (SMUG1) was isolated recently (Haushalter et al., lecular matchmaker’ (Sancar and Hearst, 1993) 1999). It is able to excise uracil from ssDNA and from and allows for the interaction of numerous other dsDNA in bothG:U and A:U base pairs, but showedno proteins to form a higher-order complex that is involved activity against G:T mismatches or any range of other in excision of a large fragment of DNA (up to 1000 possible substrates. Its strong specificity for uracil and base pairs) that contains the mismatch. The other its preference for ssDNA substrates led to its designa- factors that comprise this large MMR complex include tion. SMUG1 was found to have activity against a exonucleases and endonucleases (exonuclease-1 and double-stranded oligonucleotide containing 5-FU:A flap endonuclease-1, respectively) and replication factors (Kavli et al., 2002). (DNA polymerase d, proliferating cell nuclear antigen, For the present discussion, it is most important to replication protein A, and replication factor C) note that each of these processes, if interrupted by either (Bellacosa, 2001). DNA synthesis and ligation complete depletion of necessary components or futile repair, may the repair process. The exact manner by which all lead to a variety of DNA lesions manifested as of these steps occur in terms of the order of events unprocessed AP sites or DNA single-stranded gaps of and the required level of association of protein various sizes. Additionally, DSBs and deletions could be complexes is at present poorly understood, but currently generated by the repair by UDG of closely spaced uracil under intense investigation. A discussion of current residues on opposite strands of DNA (Dianov et al., theories regarding these mechanisms is beyond the scope 1991). of this review.

Oncogene Role of MMR in response to fluoropyrimidines M Meyers et al 7381 MMR and drug resistance 1982; Au et al., 1992). Other models that attempt to explain increased cell deathdue to MMR activity Loss of functional MMR is associated withcancer and suggest that repeated rounds of MMR action due to drug resistance. Thus, understanding the cellular me- futile cycles of repair lead to a collision of MMR chanisms of loss of MMR activity, developing ways to enzymes withBER or DNA replication enzymes (Davis restore MMR, and/or developing treatment regimens et al., 1998); however, evidence in support of this theory that target MMR-deficient cells are important. The first is lacking. Finally, the proteins of the MMR system may indication of the importance of MMR in protecting recognize damaged DNA and directly trigger a signal against human disease was found during the investiga- transduction cascade that activates the apoptotic or tion of hereditary nonpolyposis CRC. These studies necrotic cell deathresponses in cells. All of theabove have convincingly demonstrated a direct cause and models attempt to explain how cells with MMR defects effect relationship between mutations in MMR genes (in acquire selective growthadvantages, acquire mutations, particular, hMLH1 and hMSH2) and the development and thus become tumorigenic, whereas MMR-compe- of MSI. MSI is an indication of faulty MMR and tent cells respond to the same DNA lesions with lethal genetic instability. The proteins required for a functional consequences. These MMR-dependent responses appear MMR system are responsible for detecting DNA to support the theory that cell death, as opposed to damage and mediating induction of apoptosis, therefore survival withhighmutagenic consequences, is evolutio- controlling lethality. narily preferable in multicellular organisms (Boothman Cells deficient in MMR are more resistant to the et al., 1988). cytotoxic effects of several DNA-modifying agents than cells that are MMR proficient (Fink et al., 1998; Lage and Dietel, 1999; Jiricny and Nystrom-Lahti, 2000; Jacob et al., 2001). This is especially true of methylating MMR-mediated signaling agents (suchas N-methyl-N0-nitro-N-nitrosoguanidine (MNNG), N-methyl-N-nitrosourea, procarbazine, and Cell cycle arrests are postulated to allow cells greater temozolomide) and the modified purine antimetabolite time to repair DNA damage and/or allow the elimina- 6-thioguanine (6-TG), and much more modestly true of tion of severely damaged cells to prevent tumor agents suchas cisplatin and doxorubicin (Karran, 2001). formation in mammalian cells (Boothman et al., 1988; In fact, this was first observed in E. coli strains treated Hartwell and Kastan, 1994). The responses of MMR- withMNNG (Karran and Marinus, 1982), and later competent cells to specific types of DNA-damaging extended to eucaryotic cells. In MMR-deficient cells, agents appear to highlight both of these responses, and exposure to certain alkylating agents result in the offer the best ‘proof of principle’ of these hypotheses in accumulation of DNA damage, but not cell death. This human cells. Boland and his co-workers first reported phenomenon is often referred to as ‘damage tolerance’ differential cell cycle G2 checkpoint arrest between and could allow boththeuncheckeddevelopment of matched MMR-proficient and -deficient cells treated carcinomas and their subsequent resistance to therapies. withMNNG and later 6-TG (Koi et al., 1994; Hawn et al., 1995). These investigators proposed that G2 arrest would allow cells time to repair lesions created during MMR-mediated futile cycling replication prior to entering mitosis, resulting in greater mutational avoidance. This fits the notion that MMR is A model of ‘futile cycles of repair’ has been proposed to involved in postreplicative DNA repair (Muller and explain the relationship between MMR and induction of Fishel, 2002). However, further examination revealed apoptosis (Karran and Bignami, 1994; Karran and that MMR status only impacts cell cycle progression Hampson, 1996; Karran, 2001). In this theory, MMR- following exposure to certain DNA-damaging agents. proficient cells treated withsome alkylating agents or One pertinent question concerns the origin of the signal antimetabolites (suchas 6-TG or FPs) respond by for this cell cycle checkpoint arrest. Many theories have removing mismatched bases opposite damaged bases. been postulated to explain the MMR-dependent differ- However, the continued presence of antimetabolites or ential cell cycle signaling that has been noted after an abundance of damaged bases leads to further exposure to some, but not all, DNA-damaging agents; misincorporation events and repeated rounds of these are discussed below. MMR. This process is greatly exacerbated by the natural mechanism of MMR, whereby a large tract of Role of p53 protein family members one DNA strand (approximately 1 kb) is typically removed eachtime even a single mispair is repaired by It has been suggested that MMR mediates the stabiliza- MMR. These additional, now futile, cycles of MMR tion of the p53 tumor suppressor protein, and that this is ultimately lead to the generation of DNA strand breaks responsible for G2 arrest checkpoint responses (Stewart and cell death. As a consequence of repeated MMR and Pietenpol, 2001). Some theories hold that both p53 activities in E. coli, it was suggested that the MutH and p73 are involved (Strano et al., 2001). Bothp53 and endonuclease itself might be responsible for producing a its homologue, p73, are activated after DNA damage DSB that ultimately results in cell death by making a and act as transcription factors involved in regulating second incision on the opposite strand of heteroduplex genes involved in cell cycle arrest or apoptosis (Strano DNA at an unmethylated d(GATC) site (Glickman, et al., 2001). The role of post-translational modifications

Oncogene Role of MMR in response to fluoropyrimidines M Meyers et al 7382 involved in p53 (and p73) stabilization and transactiva- indicative of upstream activation of ATM (mutated in tion are complex. p53 can be acetylated and/or ataxia telangiectasia) and/or ATR (ATM and Rad3 phosphorylated at many amino-acid residues, and the related), which are phosphatidylinositol 3-kinase-related coordinated actions of these various modifications allow (PI-3-K) protein kinases (Abraham, 2001; Durocher and for differential regulation of downstream cell cycle and Jackson, 2001; Bernstein et al., 2002). In fact, a recent apoptotic responses. Interestingly, p53 has been shown study has shown that MNNG activates ATM, and to be selectively phosphorylated in an MMR-dependent wortmannin (a PI-3-K inhibitor) effectively blocks manner after MNU or MNNG treatments at two phosphorylation of serine-15 of p53 (Adamson et al., residues, serine-15 and serine-393 (Duckett et al., 2002). Our own lab and others (Carethers et al., 1996) 1999); we have found similar results (Wagner et al., in have found that caffeine over-rides MMR-dependent G2 preparation). These phosphorylation events lead to arrest mediated by MNNG in the HCT116 model cell stabilization of p53 by disrupting its interaction with system (Wagner et al., in preparation). Although the the human homologue of the mouse double minute-2 mechanism by which caffeine abrogates G2 arrest protein, which normally targets p53 for degradation responses is not entirely clear, inhibition of ATM by (Stewart and Pietenpol, 2001). However, our lab (Davis this agent has been proposed. It is also entirely possible et al., 1998) and others (Hickman and Samson, 1999; that these responses are independent of ATM or ATR. Lin et al., 2000) have shown that p53 is dispensable for Since p53 may not influence MMR-dependent cell G2 arrest, apoptosis, and loss of colony-forming ability cycle arrest and apoptosis after exposure to specific after treatment withotherDNA-damaging agents. Loss DNA-damaging agents, a role for p73 may be indicated. of p53 by stable transfection of bothmatchedMMR- p73 may be stabilized in an MMR-dependent manner competent and -deficient cells with the human papillo- after treatment withcisplatin, a bifunctional alkylating mavirus E6 protein, which binds to and destabilizes p53 chemotherapeutic agent that produces platinum–DNA by enhancing its degradation, had little effect on adducts (Gong et al., 1999). This stabilization seems survival (Davis et al., 1998). These experiments demon- dependent on acetylation of key residues by the c-Abl strate that MMR-dependent cell cycle arrest and nonreceptor tyrosine kinase, whose activation also apoptosis probably occur independently of p53 status. seems dependent on ATM (Baskaran et al., 1997; However, contradictory results have been reported by Shafman et al., 1997; Agami et al., 1999; Yuan et al., other groups (Bunz et al., 1999; Vikhanskaya et al., 1999; Costanzo et al., 2002). The MMR and c-Abl 1999). pathway also may be linked to the JNK1, c-jun- Although p53 status does not appear to be required activated kinase pathway (Nehme et al., 1997, 1999). for G2 arrest responses, its activation in an MMR- However, a clear demonstration of a role for p73 in dependent manner can result from intracellular signal- MMR-dependent G2 arrest responses has yet to be ing responses that originate from DNA lesions detected established. (or created) by MMR (Figure 2). The selective One candidate protein speculated by our laboratory phosphorylation of serine-15 on p53 appears to be to be involved in the MMR-dependent G2 arrest is the breast cancer susceptibility protein 1, BRCA1. BRCA1 is involved in G2 cell cycle arrest and is thought to interact downstream of ATM/ATR, playing a role in DNA Mismatch bothcell cycle arrest and homologous DSB repair. In a manner similar to p53, BRCA1 may be hyperpho- sphorylated and thus activated by ATM/ATR, allowing MMR DSB it to act as a transcription factor for several downstream cell cycle arrest proteins, suchas p21 WAF1/CIP1,the growth-arrest and DNA-damage-inducible 45 gene product (GADD45), and 14-3-3s (Wang et al., 2000; ATM/ Caffeine & Wortmannin Venkitaraman, 2002). However, only GADD45 seems ATR to be differentially induced in an MMR-dependent fashion in response to FP exposures (Meyers et al., 2001). Interestingly, BRCA1 also appears to form a E6 p53 BRCA1 c-Abl stable inactive complex withc-Abl thatbecomes activated by ATM following the formation of DNA damage, specifically DSBs. The BRCA1–c-Abl complex Cell cycle arrest & Apoptosis p73 is believed to dissociate due to the activation of c-Abl by (Gadd45, p21, 14-3-3σ) damage and its phosphorylation, releasing c-Abl and Figure 2 A proposed signaling pathway originating from DNA allowing for downstream signaling (Foray et al., 2002). MMR. Upon recognition of a lesion, which may generate a DNA A direct interaction between the MMR protein hMSH2 DSB, the ATM (mutated in ataxia telangectasia) and/or ATR and BRCA1 has been observed and both are proposed (ATM- and Rad3-related) protein kinases are activated, which parts in a multiprotein complex involved in DNA have p53, BRCA1 (breast cancer susceptibility gene 1), and c-Abl as substrates. These proteins directly or indirectly modulate damage recognition and repair known as the BRCA1- proteins (such as the growth-arrest and DNA-damage inducible associated genome surveillance complex (Wang et al., gene 45, p21, 14-3-3s) involved in cell cycle arrest and apoptosis 2000, 2001). Our laboratory is currently investigating

Oncogene Role of MMR in response to fluoropyrimidines M Meyers et al 7383 these hypotheses. However, it is not known how MMR- low serum. Arrested cells were released by low-density dependent detection of specific lesions can activate the replating and treated withFPs immediately before entry ATM/ATR kinases. into S phase (i.e. at 16 h after release from growth Two theories have been proposed to explain the arrest). Only HCT116 3-6 cells showed extensive G2 MMR-dependent cellular responses to specific DNA arrest responses after treatment with0.25 mm FdUrd. At damage (Figure 2). In the first theory, MMR detection higher doses, the G2 arrest was equivalent in bothcell of DNA lesions leads to the creation of DSBs caused by lines and persisted for at least 72 h. Likewise, MMR- futile cycles of repair or repair–replication fork colli- proficient ME-10 cells and MSH2-expressing cells sions (Davis et al., 1998). The other theory states that demonstrated greatly enhanced G2 arrest responses MMR can act as a DNA damage sensor and directly following similar FP exposures (Meyers et al., unpub- signal cell cycle arrest, although evidence for a direct lished observations). interaction between MMR and ATM or ATR has not Finally, HCT116 and HCT116 3-6 cells were treated been established. The aggregate of data seems to with various doses of FdUrd continuously for 72 h, then support the theory that MMR detection of DNA lesions DSB formation was assayed using pulsed-field gel leads to the formation of DSBs after FP exposure electrophoresis. In both cell lines, treatment with as (Meyers et al., 2001), although current technology little as 2.5 mm FdUrd resulted in the appearance of allows detection of DSBs only at lethal drug concentra- DSBs. However, the degree of fragmentation was more tions. However, we speculate that MMR detects DNA pronounced in MMR-proficient HCT116 3-6 cells. lesions and allows repetitive rounds of futile cycle repair Interestingly, the DNA fragmentation did not appear via the formation of multiple sliding clamp complexes in to reflect apoptosis simply, since bothHCT116 and the vicinity of the originally detected DNA lesion. These HCT116 3-6 cells treated in this manner had identical, repetitive MMR events result in fragile DNA, creating low levels of apoptosis (7.570.5%) as quantified by a DSBs. The formation of DSBs then leads to the modified terminal deoxynucleotidyltransferase-mediated activation of ATM and/or ATR or other signal dUTP nick end labeling assay. Based on our published transduction responses leading to G2 arrest. Activation data, we propose the model shown in Figure 3 to of these signal transduction processes then results in G2 represent how MMR processes may mediate cellular arrest responses, withp53 (p73) activation possibly responses to FP treatment. required for apoptotic, but not overall survival re- Although both BER and MMR are clearly involved sponses (Davis et al., 1998). in responses to FP-induced DNA damage, direct interfaces between BER and MMR have not been Role of MMR in detecting and responding to FP moieties demonstrated, other than the two-hybrid interaction in DNA reported between MLH1 and MBD4 previously men- tioned (Bellacosa et al., 1999). BER is clearly involved in We investigated the role of MMR in the cellular initial responses to FP-induced lesions, since bothuracil responses to FPs using various genetically matched cells and 5-FU moities are recognized and excised by UDGs. containing altered MMR proficiency (Meyers et al., Our own data indicate that neither the hMSH2-3 or 2001). Clonogenic assays were performed to determine MSH2-6 complex recognizes an A:5-FU base pair in a the survival of HCT116 human CRC (MMR-deficient) DNA oligonucleotide by band shift and ATPase activity compared to HCT116 3-6 (MMR-proficient) cells assays (Meyers et al., in preparation). However, G:5-FU following continuous treatments withvarious doses of moieties are recognized. These data suggest that MMR 5-FU or FdUrd. HCT116 cells were 18-fold more does not process A:5-FU lesions; they are probably resistant to 7.5 mm 5-FU and 17-fold more resistant to handled by BER. In contrast, MMR appears to process 7.5 mm FdUrd compared to HCT116 3-6 cells. Similarly, the potentially more mutagenic G:5-FU lesions. Alter- in a mouse system (Buermeyer et al., 1999) in which natively, MMR could further play a major role spontaneously immortalized embryonic fibroblasts from following (and in conjunction with) BER’s response to the MLH1À/À mouse were transfected with hMLH1 cDNA (ME-10) or empty vector (CT-5), ME-10 cells were threefold more sensitive to 10 mm FdUrd following 5-FU a 2-hexposure compared to CT-5 cells. In bothsystems, FdUrd FP incorporation subsequent incubation withthymidine was able to rescue BER this cytotoxicity, whereas incubation with uridine did MMR not, indicating that the FP-mediated cytotoxicity was DNA Breaks Death DNA directed. The same phenomenon was observed in both human and mouse cells deficient in the other major polymerase dNTP pool MMR protein, MSH2 (Meyers et al., in preparation). errors G arrest imbalance 2 To examine the influence of MMR pathways on cell cycle checkpoint responses, changes in cell cycle Figure 3 A model for the role of DNA MMR in FP-mediated distribution of MMR-proficient and -deficient cells cytotoxicity. DNA lesions result from direct incorporation of FPs into DNA and from DNA polymerase errors. The activities of following exposure to FdUrd were examined (Meyers MMR, in concert withthoseof DNA BER, generate breaks in et al., 2001). HCT116 or HCT116 3-6 cells were DNA; a G2 arrest is signaled directly through MMR or as a result synchronized by confluence arrest and treatment with of the DNA breaks. Ultimately, these breaks result in cell death

Oncogene Role of MMR in response to fluoropyrimidines M Meyers et al 7384 lesions containing high densities of dUMP and FdUrd. increasing the production of FdUMP and/or the Although the range of specific lesions recognized by stability of FdUMP-THF-TS) withtheanticipation, MMR following FP exposure have not been identified, and realization to some extent, of increased tumor either mispaired FdUTP or mispaired loops at SSB sites selectivity. We suggest that such DNA-targeted cyto- left by incomplete BER may provide suchsubstrates. toxicity may be contraindicated when treating patients For example, in the latter case, one or more unrepaired withMSI tumors. In thesecircumstances, treatment AP sites might destabilize DNA ends at break sites, may result in a cytotoxic response in normal cells with leading to transient formation of mispaired DNA loops. increased resistance to drug treatment due to damage In suchcases, MMR could convert relatively small tolerance in tumors that lack MMR activity. Further- incompletely repaired patches created by BER to large more, analogous to studies observed in bacteria repair patches, thus enhancing the phenomenon of futile (LeClerc et al., 1996; Mao et al., 1997) MMR-deficient cycling by greatly increasing the probability that sites of cells could, in fact, be selected for by suchDNA-directed initial dUMP or FdUMP incorporation are not FP treatments. In addition, treatment of MMR-deficient correctly repaired. tumors may harbor an elevated level of mutation due to The probability that the second scenario may occur in FP-induced DNA lesions. Survival of cells withaccu- mammalian cells is consistent withobservations of mulated mutations could result in increased tumor Duker et al. (1982) showing that AP sites antagonize heterogeneity and selection for more malignant and their own repair if closely spaced in DNA. Thus, under invasive tumor cells. Indeed, two recent clinical reports conditions leading to high dUMP and/or FdUMP have found that standard, 5-FU-based chemotherapy incorporation densities, mispaired loops involving AP given to colon cancer patients withhighlevelsof MSI sites could be more common than anticipated at SSBs, did not result in a significant survival advantage (Goel providing lesion sites that are recognized by MMR and et al., 2003; Ribic et al., 2003). leading to a cascade of events resulting in progressively larger repair patches in DNA until completion of normal repair processes becomes impossible. The latter FdCyd: a new therapeutic approach for treatment of model is attractive because it might be more likely to MMRtumors result in accumulation of SSBs at high enough densities on bothDNA strands to lead to DSBs and apoptotic Although not yet tested in humans, 5-fluoro-2-deoxycy- signaling (Taverna et al., 2001). In bothmodels, tidine (FdCyd) has exhibited promise for tumor therapy however, abrogation of checkpoint responses in in animal models (Mekras et al., 1984; Boothman et al., MMR-defective cells could permit lesion bypass to 1987a). FdCyd antimetabolites can be acted upon by occur before unacceptable lesion densities accumulate deaminases specifically elevated in CRCs, thereby from redundant attempts to repair lesions under effectively delivering FdUrd to DNA in tumors (Mekras conditions where they continuously reform. This could et al., 1984; Kaysen et al., 1986; Boothman et al., then explain the higher tolerance of MMR-defective 1987b). In fact, Mekras et al. (1984, 1985) showed that cells for FPs. Regardless of the mechanism, MMR FdCyd was more effective on a molar basis in vivo than appears to be involved in a major cell cycle checkpoint 5-FU or FdUrd and exhibited more tumor cell-directed response specifically at G2, where it appears to recognize cytotoxicity. This higher efficiency was partially some sort of DNA damage and then prevent cell cycle progression until FP-induced lesions are repaired or cell hMLH1 Expression deathoccurs (Meyers et al., 2001). MMR Detection/ Cytotoxicity Hypomethylation

Clinical applications FdCyd Moieties in DNA FdUrd Moieties in DNA The responses of MMR-competent compared to MMR- deficient cells to FPs have direct clinical relevance, since nearly 15% of all CRC may be attributed to the loss of dCMP MMR and FP antimetabolites remain standard drugs Deaminase FdCMP FdUMP used for the treatment of CRC. Cell culture studies dH4UMP using other 5-substituted halogenated thymidine analo- dCyd dThd gues (suchas 5-bromo-2 0-deoxyuridine and 5-iodo-20- Kinase Kinase dCyd deoxyuridine) do not indicate that there is any survival Deaminase difference between MMR-proficient and -deficient cells; FdCyd FdUrd H Urd or dH Urd however, the greater degree of incorporation of these 4 4 Phosphorylases antimetabolites into the DNA of MMR-deficient cells make these drugs attractive agents for selective radio- FdCyd is not a substrate RNA-level Cytotoxicity sensitization following treatment withionizing radiation for Phosphorylases (Berry et al., 1999, 2000). Likewise, many contemporary Figure 4 FdCyd metabolism and its effects on the cell. See text for combined treatment modalities were designed to in- details dH4Urd can inhibit dCydD directly, as well as dCMPD crease the DNA-level cytotoxicity of 5-FU (e.g. by upon its conversion to dH4UMP

Oncogene Role of MMR in response to fluoropyrimidines M Meyers et al 7385 attributable to the resistance of FdCyd to nucleoside Unresolved issues and future directions phosphorylases (Mekras et al., 1985; Boothman et al., 1987a). Subsequent studies showed that fluori- Since the cell systems used in our laboratory vary only nated cytidine compounds, when modulated with the in specific MMR protein expression, it seems plausible dCyd deaminase (dCydD) inhibitor tetrahydrouridine that DNA lesions account for the survival difference (H4Urd), could deliver as muchas threeto four orders between MMR-proficient and the more resistant MMR- of magnitude higher levels of FdUMP to tumor tissue deficient cells following FP exposure. Unfortunately, it versus normal tissue (Mekras et al., 1984; Boothman has not yet been established whether FdUrd-induced et al., 1987a, b). This occurs because H4Urd can inhibit incorporation of dUTP is required alone or in combina- the conversion of FdCyd to FdUrd and instead channel tion withincorporation of FdUTP for manifestation its conversion to FdCMP, whereby (following deamina- of FdUrd-induced cytotoxicity. In this regard, it is tion) it is converted to FdUMP at high levels, causing important to note that several approaches that greater TS inhibition and cytotoxicity than FdCyd alone might answer this important question are available. (see Figure 4). In addition, administration of H4Urd For example, in mammalian cells, dUMP levels can withFdCyd can cause a five-fold increase in FdCyd be potentially downregulated by concentrating inhibi- incorporation into DNA. Conversely, administration of tion strategies on three critical enzyme activities FdCyd in combination withthedCydD and dCMPD that determine dUMP levels in cells: dUTPase (McIn- inhibitor, deoxytetrahydrouridine (dH4Urd), shunts toshand Haynes, 1997), RR (Kashlan et al., 2002), FdCyd into DNA withminimal conversion to FdUMP and dCydD (Bianchi et al., 1987). Unfortunately, (and thus much lower cytotoxicity). These alterations of genetic antisense RNA or other strategies that might FdCyd metabolism withH 4Urd or dH4Urd were seen in be used to inhibit dUTPase would increase the size of tissue culture and animal studies (Mekras et al., 1984, dUTP pools that result in dUMP incorporation into 1985; Boothman et al., 1985; Kaysen et al., 1986). DNA, and also would increase the probability that In addition to being able to modulate FdCyd bothFdUTP and dUTP would be incorporated into metabolism in tumor tissue selectively, FdCyd is a DNA (Caradonna and Cheng, 1980; Ingraham powerful hypomethylating agent. As shown in Figure 4, and Goulian, 1982). It may also be possible to interdict after its incorporation into DNA, FdCyd has the de novo synthesis of dUTP precursors by manipulating potential to modulate promoter CpG methylation intracellular dNTP levels that regulate the ability through its inhibition of cytidine methylases (Newman of RR to produce deoxynucleotide precursors of and Santi, 1982). In this regard, the FdCyd moiety may dUMP. However, these strategies are complicated form covalent bonds withDNA methylase following (Kashlan et al., 2002). Finally, dUMP and consequent incorporation into DNA and can exhibit higher stability dUTP levels can also be directly downregulated in DNA than the more extensively studied DNA by inhibiting dCMPD and dCydD (Bianchi et al., methylase inhibitors, 5-azacytidine (AzaC) and 5-aza- 1987). For example, it should be possible to 20-deoxycytidine (AzadC) (Newman and Santi, 1982; reduce dUTP misincorporation relative to that of Osterman et al., 1988). This dual activity potential of FdUTP by treating cells witha combination of FdUrd FdCyd makes it of special interest to explore the efficacy and H4dUrd (Boothman et al., 1987a, b). Such of FdCyd for treatment of a significant subset of cancers an approach should have two consequences. that are defective in MMR due to aberrant methylation First, FdUTP incorporation into DNA will be augmen- of the hMLH1 promoter. This subgroup of sporadic ted through the ability of FdUMP to inhibit TS CRCs represents up to 15% of total CRCs (Herman and consequently facilitate the efficiency of its et al., 1998) and a significant proportion of gastric own incorporation through reduction of dTTP pools. cancers (Fleisher et al., 2001). A number of studies have Second, dUTP misincorporation should be reduced demonstrated that CpG methylation-dependent MMR since dCydD activity makes important contributions defects in colon and endometrial tumor cells can be to intracellular dUMP pools. Therefore, the levels transiently reversed by exposure to AzaC and AzadC of uracil moieties in DNA should be reduced in (Kane et al., 1997; Herman et al., 1998; Veigl et al., cells treated with5-FdUrd and dCydD inhibi- 1998). These agents, like FdCyd, work by disrupting tors relative to levels achieved in cells treated promoter methylation patterns seen in CpG sequences with FdUrd alone. This approach should be useful residing in the hMLH1 promoter and transiently in dissecting the relative importance of dUTP mis- restoring the expression of functional hMLH1. Since incorporation to the toxic effects of FdUrd. Since all FdCyd can be modulated by dCMPD and dCydD components of the pathways leading to FdUrd-induced inhibitors to incorporate FdCTP as well as FdUTP into toxicity at the DNA level are interactive, results DNA, it may be of special interest to explore the efficacy of various drug combinations must be carried out of FdCyd for treatment of MMR-defective tumors in the context of careful analysis of nucleotide pool aberrantly silenced for hMLH1 expression. Specifically, levels. However, scrutiny of these inter-relation- FdCyd has the potential to both induce hMLH1 ships should be rewarded with a much clearer expression through its incorporation as FdCTP into understanding of the mechanism leading to FP-induced DNA and to kill MMR-competent cells by TS inhibition cytotoxicity, as well as the promise to improve and effectively inducing incorporation of FdUTP and therapy of patients afflicted with MMR-deficient dUTP into DNA. tumors.

Oncogene Role of MMR in response to ﬂuoropyrimidines M Meyers et al 7386 Acknowledgements NIH/NCI Grant R01-CA67409 to MLV and WDS (co-PIs). This work was supported by NCI/NIH Grant CA-83196-05 to This work was also supported, in part, by a DOD breast DAB, as well as NIH/NCI Grant CA-70788 to MLV and cancer predoctoral fellowship to MWW.

References

Abraham RT. (2001). Genes Dev., 15, 2177–2196. Duckett DR, Bronstein SM, Taya Y and ModrichP. (1999). Adamson AW, Kim W-J, Shangary S, Baskaran R and Brown Proc. Natl. Acad. Sci. USA, 96, 12384–12388. KD. (2002). J. Biol. Chem. 277, 38222–38229. Duker NJ, Jensen DE, Hart DM and Fishbein DE. (1982). Aebersold PM. (1979). Cancer Res., 39, 808–810. Proc. Natl. Acad. Sci. USA, 79, 4878–4882. Agami R, Blandino G, Oren M and Shaul Y. (1999). Nature Durocher D and Jackson SP. (2001). Curr. Opin. Cell Biol., 13, (Lond.), 399, 809–813. 225–231. Au KG, WelshK and ModrichP. (1992). J. Biol. Chem., 267, el-Hajj HH, Wang L and Weiss B. (1992). J. Bacteriol., 174, 12142–12148. 4450–4456. Baskaran R, Wood LD, Whitaker LL, Canman CE, Morgan Fazakerley GV, Sowers LC, Eritja R, Kaplan BE and SE, Xu Y, Barlow C, Baltimore D, Wynshaw-Boris A, Kastan Goodman MF. (1987). J. Biomol. Struct. Dyn., 5, 639–650. MB and Wang JYJ. (1997). Nature (Lond.), 387, 516–519. Fernandes DJ and Cranford SK. (1986). Cancer Res., 46, Bebenek K, Roberts JD and Kunkel TA. (1992). J. Biol. 1741–1747. Chem., 267, 3589–3596. Fink D, Aebi S and Howell SB. (1998). Clin. Cancer Res., 4, Bellacosa A. (2001). Cell Death Differ., 8, 1076–1092. 1–6. Bellacosa A, Cicchillitti L, Schepis F, Riccio A, Yeung AT, Fishel R. (1998). Genes Dev., 12, 2096–2101. Matsumoto Y, Golemis EA, Genuardi M and Neri G. Fishel R. (1999). Nat. Med., 5, 1239–1241. (1999). Proc. Natl. Acad. Sci. USA, 96, 3969–3974. Fisher EF and Caruthers MH. (1979). Nucleic Acids Res., 7, Bernstein C, Bernstein H, Payne CM and Garewal H. (2002). 401–416. Mutat. Res., 511, 145–178. Fleisher AS, Esteller M, Tamura G, Rashid A, Stine OC, Yin Berry SE, Davis TW, Schupp JE, Hwang HS, de Wind N and J, Zou TT, Abraham JM, Kong D, Nishizuka S, James SP, Kinsella TJ. (2000). Cancer Res., 60, 5773–5780. Wilson KT, Herman JG and Meltzer SJ. (2001). Oncogene, Berry SE, Garces C, Hwang HS, Kunugi K, Meyers M, Davis 20, 329–335. TW, Boothman DA and Kinsella TJ. (1999). Cancer Res., Foray N, Marot D, Randrianarison V, Venezia ND, Picard D, 59, 1840–1845. Perricaudet M, Favaudon V and Jeggo P. (2002). Mol. Cell. Bianchi V, Pontis E and Reichard P. (1987). Mol. Cell. Biol., 7, Biol., 22, 4020–4032. 4218–4224. Freese E. (1959). J. Mol. Biol., 1, 87–105. Boothman DA, Briggle TV and Greer S. (1985). Mol. Friedkin M. (1973). Adv. Enzymol., 38, 235–292. Pharmacol., 27, 584–594. Gill JE, Mazrimas JA and Bishop Jr CC. (1974). Biochim. Boothman DA, Briggle TV and Greer S. (1987a). Cancer Res., Biophys. Acta, 335, 330–348. 47, 2344–2353. Glazer RI and Hartman KD. (1981). Mol. Pharmacol., 19, Boothman DA, Briggle TV and Greer S. (1987b). Cancer Res., 117–121. 47, 2354–2362. Glazer RI and Lloyd L. (1982). Mol. Pharmacol., 21, 468–473. Boothman DA, Schlegel R and Pardee AB. (1988). Mutat. Glickman BW. (1982). Molecular and Cellular Mechanisms of Res., 202, 393–411. Mutagenesis Generoso WM (ed). Plenum Press: New York, Buermeyer AB, Wilson-Van Patten C, Baker SM and Liskay NY, pp. 65–87. RM. (1999). Cancer Res., 59, 538–541. Goel A, Arnold CN, Niedzwiecki D, Chang DK, Ricciardiello Bunz F, Hwang PM, Torrance C, Waldman T, Zhang Y, L, Carethers JM, Dowell JM, Wasserman L, Compton C, Dillehay L, Williams J, Lengauer C, Kinzler KW and Mayer RJ, Bertagnolli MM and Boland CR. (2003). Cancer Vogelstein B. (1999). J. Clin. Invest., 104, 263–269. Res., 63, 1608–1614. Caradonna SJ and Adamkiewicz DM. (1984). J. Biol. Chem., Gong JG, Costanzo A, Yang H-Q, Melino G, Kaelin Jr WG, 259, 5459–5464. Levrero M and Wang JYJ. (1999). Nature (Lond.), 399, Caradonna SJ and Cheng YC. (1980). Mol. Pharmacol., 18, 806–809. 513–520. Goulian M, Bleile B and Tseng BY. (1980a). J. Biol. Chem., Carethers JM, Hawn MT, Chauhan DP, Luce MC, Marra G, 255, 10630–10637. Koi M and Boland CR. (1996). J. Clin. Invest., 98, 199–206. Goulian M, Bleile B and Tseng BY. (1980b). Proc. Natl. Acad. Coll M, Saal D, Frederick CA, Aymami J, RichA and Wang Sci. USA, 77, 1956–1960. AH-J. (1989). Nucleic Acids Res., 17, 911–923. Gradia S, Subramanian D, Wilson T, Acharya S, Makhov A, Costanzo A, Merlo P, Pediconi N, Fulco M, Sartorelli V, Cole GrifﬁthJ and FishelR. (1999). Mol. Cell, 3, 255–261. PA, Fontemaggi G, Fanciulli M, Schiltz L, Blandino G, Grem JL. (1996). Cancer Chemotherapy and Biotherapy: Balsano C and Levrero M. (2002). Mol. Cell, 9, 175–186. Principles and Practice Longo DL (ed). Lippincott: Phila- Davis TW, Wilson-Van Patten C, Meyers M, Kunugi KA, delphia, PA, pp. 149–211. Cuthill S, Reznikoff C, Garces C, Boland CR, Kinsella TJ, Grem JL. (1997). Semin. Radiat. Oncol., 7, 249–259. Fishel R and Boothman DA. (1998). Cancer Res., 58, 767–778. Grem JL, Mulcahy RT, Miller EM, Allegra CJ and Fischer Dianov GL, Timchenko TV, Sinitsina OI, Kuzminov AV, PH. (1989). Biochem. Pharmacol., 38, 51–59. Medvedev OA and Salganik RI. (1991). Mol. Gen. Genet., Habener JF, Vo CD, Le DB, Gryan GP, Ercolani L and 225, 448–452. Wang AH-J. (1988). Proc. Natl. Acad. Sci. USA, 85, Domin BA, Mahony WB and Zimmerman TP. (1993). 1735–1739. Biochem. Pharmacol., 46, 503–510. Hardeland U, Bentele M, Lettieri T, Steinacher R, Jiricny J Doong S-L and Dolnick BJ. (1988). J. Biol. Chem., 263, and Schar P. (2001). Prog. Nucleic Acid Res. Mol. Biol., 68, 4467–4473. 235–253.

Oncogene Role of MMR in response to ﬂuoropyrimidines M Meyers et al 7387 Hartwell LH and Kastan MB. (1994). Science (Wash. DC), Kufe DW, Major PP, Egan EM and LohE. (1981). J. Biol. 266, 1821–1828. Chem., 256, 8885–8888. Haushalter KA, Stukenberg PT, Kirschner MW and Verdine Kunz BA. (1996). Mutat. Res., 355, 129–140. GL. (1999). Curr. Biol., 174–185. Kunz BA, Kohalmi SE, Kunkel TA, Mathews CK, McIntosh Hawn MT, Umar A, Carethers JM, Marra G, Kunkel TA, EM and Reidy JA. (1994). Mutat. Res., 318, 1–64. Boland CR and Koi M. (1995). Cancer Res., 55, 3721–3725. Lage H and Dietel M. (1999). J. Cancer Res. Clin. Oncol., 125, Heidelberger C, Chaudhuri NK, Danneberg P, Mooren D, 156–165. Griesbach L, Duschinsky R, Schnitzer RJ, Pleven E and LeClerc JE, Li BG, Payne WL and Cebula TA. (1996). Science Scheiner J. (1957). Nature (Lond.), 179, 663–666. (Wash. DC), 274, 1208–1211. Heidelberger C, GriesbachL, Cruz D, SchnitzerRJ and Lin X, Ramamurthi K, Mishima M, Kondo A and Howell SB. Grunberg E. (1958). Proc. Soc. Exp. Biol. Med., 97, 470–475. (2000). Mol. Pharmacol., 58, 1222–1229. HendrichB, Hardeland U, Ng H-H, Jiricny J and Bird A. Lindahl T. (1993). Nature (Lond.), 362, 709–715. (1999). Nature (Lond.), 401, 301–304. Liu P, Burdzy A and Sowers LC. (2002). Chem. Res. Toxicol., Herman JG, Umar A, Polyak K, Graff JR, Ahuja N, Issa JP, 15, 1001–1009. Markowitz S, Willson JK, Hamilton SR, Kinzler KW, Kane Lo¨ nn U and Lo¨ nn S. (1984). Cancer Res., 44, 3414–3418. MF, Kolodner RD, Vogelstein B, Kunkel TA and Baylin Lo¨ nn U, Lo¨ nn S, Nylen U and Winblad G. (1989). Cancer SB. (1998). Proc. Natl. Acad. Sci. USA, 95, 6870–6875. Res., 49, 1085–1089. Hickman MJ and Samson LD. (1999). Proc. Natl. Acad. Sci. Major PP, Egan E, Herrick D and Kufe DW. (1982). Cancer USA, 96, 10764–10769. Res., 42, 3005–3009. HsiehP. (2001). Mutat. Res., 486, 71–87. Mao EF, Lane L, Lee J and Miller JH. (1997). J. Bacteriol., Ingraham HA and Goulian M. (1982). Biochem. Biophys. Res. 179, 417–422. Commun., 109, 746–752. Mauro DJ, de Riel JK, Tallarida RJ and Sirover MA. (1993). Ingraham HA, Tseng BY and Goulian M. (1980). Cancer Res., Mol. Pharmacol., 43, 854–857. 40, 998–1001. Maybaum J, Cohen MB and Sadee W. (1981). J. Biol. Chem., Ingraham HA, Tseng BY and Goulian M. (1982). Mol. 256, 2126–2130. Pharmacol., 21, 211–216. McIntoshEM and Haynes RH. (1997). Acta Biochim. Pol., 44, Jacob S, Aguado M, Fallik D and Praz F. (2001). Cancer Res., 159–171. 61, 6555–6562. Mekras JA, Boothman DA and Greer SB. (1985). Cancer Res., Jiricny J and Nystrom-Lahti M. (2000). Curr. Opin. Genet. 45, 5270–5280. Dev., 10, 157–161. Mekras JA, Boothman DA, Perez LM and Greer S. (1984). Johnston PG, Takimoto CH, Grem JL, Chabner BA, Allegra Cancer Res., 44, 2551–2560. CJ and Chu E. (1996). Cancer Chemother. Biol. Response Meyers M, Wagner MW, Hwang H-S, Kinsella TJ and Modif., 16, 1–27. Boothman DA. (2001). Cancer Res., 61, 5193–5201. Jones PA, Benedict WF, Baker MS, Mondal S, Rapp U and ModrichP. (1997). J. Biol. Chem., 272, 24727–24730. Heidelberger C. (1976). Cancer Res., 36, 101–107. ModrichP and LahueR. (1996). Annu. Rev. Biochem., 65, Kanamaru R, Kakuta H, Sato T, Ishioka C and Wakui A. 101–133. (1986). Cancer Chemother. Pharmacol., 17, 43–46. Muller A and Fishel R. (2002). Cancer Invest., 20, 102–109. Kane MF, Loda M, Gaida GM, Lipman J, Mishra R, Nehme A, Baskaran R, Aebi S, Fink D, Nebel S, Cenni B, Goldman H, Jessup JM and Kolodner R. (1997). Cancer Wang JYJ, Howell SB and Christen RD. (1997). Cancer Res., 57, 808–811. Res., 57, 3253–3257. Karran P. (2001). Carcinogenesis, 22, 1931–1937. Nehme A, Baskaran R, Nebel S, Fink D, Howell SB, Karran P and Bignami M. (1994). BioEssays, 16, 833–839. Wang JYJ and Christen RD. (1999). Br. J. Cancer, 79, Karran P and Hampson R. (1996). Cancer Surv., 28, 69–85. 1104–1110. Karran P and Marinus MG. (1982). Nature (Lond.), 296, Newman EM and Santi DV. (1982). Proc. Natl. Acad. Sci. 868–869. USA, 79, 6419–6423. Kashlan OB, Scott CP, Lear JD and Cooperman BS. (2002). Nilsson S, Reichard P and Skoog L. (1980). J. Biol. Chem., Biochemistry, 41, 462–474. 255, 9552–9555. Kavli B, Sundheim O, Akbari M, Otterlei M, Nilsen H, Norbury CJ and Hickson ID. (2001). Annu. Rev. Pharmacol. Skorpen F, Aas PA, Hagen L, Krokan HE and Slupphaug Toxicol., 41, 367–401. G. (2002). J. Biol. Chem., 277, 39926–39936. Osterman DG, DePillis GD, Wu JC, Matsuda A and Santi Kaysen J, Spriggs D and Kufe D. (1986). Cancer Res., 46, DV. (1988). Biochemistry, 27, 5204–5210. 4534–4538. ParikhSS, Putnam CD and Tainer JA. (2000). Mutat. Res., Koi M, Umar A, Chauhan DP, Cherian SP, Carethers JM, 460, 183–199. Kunkel TA and Boland CR. (1994). Cancer Res., 54, Parker WB, Kennedy KA and Klubes P. (1987). Cancer Res., 4308–4312. 47, 979–982. Kolodner RD and Marsischky GT. (1999). Curr. Opin. Genet. Pearl LH. (2000). Mutat. Res., 460, 165–181. Dev., 9, 89–96. Peltomaki P. (2001). Mutat. Res., 488, 77–85. Kremer AB, Mikita T and Beardsley GP. (1987). Biochemistry, Petronzelli F, Riccio A, Markham GD, Seeholzer SH, 26, 391–397. Stoerker J, Genuardi M, Yeung AT, Matsumoto Y Krokan HE, Nilsen H, Skorpen F, Otterlei M and Slupphaug and Bellacosa A. (2000). J. Biol. Chem., 275, 32422–32429. G. (2000). FEBS Lett., 476, 73–77. Reichard P. (1985). Genetic Consequences of Nucleotide Pool Krokan HE, Otterlei M, Nilsen H, Kavli B, Skorpen F, Imbalance de Serres FJ (ed). Plenum Press: New York, NY, Andersen S, Skjelbred C, Akbari M, Aas PA and Slupphaug pp. 33–46. G. (2001). Prog. Nucleic Acid Res. Mol. Biol., 68, 365–386. Reichard P. (1988). Annu. Rev. Biochem., 57, 349–374. Kufe DW and Major PP. (1981). J. Biol. Chem., 256, Ribic CM, Sargent DJ, Moore MJ, Thibodeau SN, French AJ, 9802–9805. Goldberg RM, Hamilton SR, Laurent-Puig P, Gryfe R,

Oncogene Role of MMR in response to ﬂuoropyrimidines M Meyers et al 7388 Shepherd LE, Tu D, Redston M and Gallinger S. (2003). N. Takahashi I and Marmur J. (1963). Nature (Lond.), 197, Engl. J. Med., 349, 247–257. 794–795. Richards RG, Sowers LC, Laszlo J and Sedwick WD. (1984). Taverna P, Liu L, Hwang HS, Hanson AJ, Kinsella TJ and Adv. Enzyme Regul., 22, 157–185. Gerson SL. (2001). Mutat. Res., 485, 269–281. Roberts JD and Kunkel TA. (1988). Proc. Natl. Acad. Sci. van Laar JA, Rustum YM, Ackland SP, van Groeningen CJ USA, 85, 7064–7068. and Peters GJ. (1998). Eur. J. Cancer, 34, Sancar A and Hearst JE. (1993). Science (Wash. DC), 259, 296–306. 1415–1420. Veigl ML, Kasturi L, Olechnowicz J, Ma AH, Lutterbaugh Santi DV and Hardy LW. (1987). Biochemistry, 26, 8599–8606. JD, Periyasamy S, Li GM, Drummond J, ModrichPL, Santi DV and McHenry CS. (1972). Proc. Natl. Acad. Sci. Sedwick WD and Markowitz SD. (1998). Proc. Natl. Acad. USA, 69, 1855–1857. Sci. USA, 95, 8698–8702. Schuetz JD, Wallace HJ and Diasio RB. (1984). Cancer Res., Venkitaraman AR. (2002). Cell, 108, 171–182. 44, 1358–1363. Vikhanskaya F, Colella G, Valenti M, Parodi S, Sedwick WD, Kutler M and Brown OE. (1981). Proc. Natl. D’Incalci M and Broggini M. (1999). Clin. Cancer Res., 5, Acad. Sci. USA, 78, 917–921. 937–941. Shafman T, Khanna KK, Kedar P, Spring K, Kozlov S, Yen Wang Y, Cortez D, Yazdi P, Neff N, Elledge SJ and Qin J. T, Hobson K, Gatei M, Zhang N, Watters D, Egerton M, (2000). Genes Dev., 14, 927–939. Shiloh Y, Kharbanda S, Kufe D and Lavin MF. (1997). Wang Q, Zhang H, Guerrette S, Chen J, Mazurek A, Wilson Nature (Lond.), 387, 520–523. T, Slupianek A, Skorski T, Fishel R and Greene MI. (2001). Sowers LC, Eritja R, Kaplan BE, Goodman MF and Oncogene, 20, 4640–4649. Fazakerley GV. (1987). J. Biol. Chem., 262, 15436–15442. Weckbecker G. (1991). Pharmacol. Ther., 50, 367–424. Sowers LC, Eritja R, Kaplan B, Goodman MF and Fazakerly Wilson T, Guerrette S and Fishel R. (1999). J. Biol. Chem., GV. (1988). J. Biol. Chem., 263, 14794–14801. 274, 21659–21664. Sowers LC, Goodman MF, Eritja R, Kaplan B and Yoshioka A, Tanaka S, Hiraoka O, Koyama Y, Hirota Y, Fazakerley GV. (1989). J. Mol. Biol., 205, 437–447. Ayusawa D, Seno T, Garrett C and Wataya Y. (1987). J. Stewart ZA and Pietenpol JA. (2001). Chem. Res. Toxicol., 14, Biol. Chem., 262, 8235–8241. 243–263. Yuan Z-M, Shioya H, Ishiko T, Sun X, Gu J, Huang YY, Lu Strano S, Rossi M, Fontemaggi G, Munarriz E, Soddu S, H, Kharbanda S, Weichselbaum R and Kufe D. (1999). Sacchi A and Blandino G. (2001). FEBS Lett., 490, 163–170. Nature (Lond.), 399, 814–817.

Oncogene