5-Fluorouracil Incorporated Into DNA Is Excised by the Smug1 DNA Glycosylase to Reduce Drug Cytotoxicity Qian An, Peter Robins, Tomas Lindahl, and Deborah E

Research Article

5-Fluorouracil Incorporated into DNA Is Excised by the Smug1 DNA Glycosylase to Reduce Drug Cytotoxicity Qian An, Peter Robins, Tomas Lindahl, and Deborah E. Barnes

Cancer Research UK London Research Institute, Clare Hall Laboratories, South Mimms, Hertfordshire, United Kingdom

Abstract tumor response (5), and the effects of modulating cellular levels of 5-Fluorouracil (FU) has been widely used for more than four dUTPase (Fig. 1) would seem to be consistent with the model (6, 7). decades in the treatment of a range of common cancers. However, mammalian cells lacking the major replication-associat- The fluorine-substituted uracil analogue is converted to ed uracil-DNA glycosylase, Ung (8), do not show altered resistance several active metabolites but the mechanism of cytotoxicity to FU (9), and Ung overexpression does not affect the sensitivity to has remained unclear. In a widely cited but unsubstantiated specific TS inhibitors (10). Thus, the model that FU killing is model, FU is thought to kill cells via the inhibition of mediated through inhibition of TS and increased incorporation of thymidylate synthase and increased use of dUTP in place of uracil in DNA remains unsubstantiated. TTP during DNA replication, with subsequent excision of high FU is converted to several active metabolites (Fig. 1). In addition levels of uracil causing the fragmentation of newly synthesized to inhibiting TS, FU may exert its effects through incorporation of DNA. Using gene-targeted cell lines defective in one or both of the fluoronucleotide into RNA and disruption of RNA processing (reviewed in ref. 1). FU is also incorporated into DNA following its the two mammalian uracil-DNA glycosylase repair enzymes, we were able to test this model of FU cytotoxicity. Here, we conversion to FdUTP, either via FdUMP or by reduction of the show that incorporation of FU itself into DNA has been ribonucleoside diphosphate. The introduction of FU into DNA via previously underestimated and is a predominant cause of the latter pathway may be unaffected by FdUMP-mediated cytotoxicity. FU readily becomes incorporated into the DNA of inhibition of TS (Fig. 1). dUTPase hydrolyses FdUTP as well as drug-treated cells, and accumulation of FU in the genome, dUTP (11), and the significance of FU incorporation in DNA has rather than uracil excision, is correlated with FU cytotoxicity been difficult to evaluate from the low levels measurable in drug- in mammalian cells. Furthermore, the Smug1, but not the Ung, treated normal or tumor cells (12–15). Interestingly, both the uracil-DNA glycosylase excises FU from DNA and protects Ung and Smug1 uracil-DNA glycosylases have been reported to against cell killing. The data provides a clearer understanding excise FU from DNA in vitro (16). Smug1 is not found in lower of the action of FU, suggesting predictive biomarkers of drug eukaryotes such as Saccharomyces cerevisiae and has only evolved response and a mechanism for acquired resistance in tumors. as a second uracil-DNA glycosylase in vertebrates (17), in which it [Cancer Res 2007;67(3):940–5] is nonredundant with the Ung enzyme in excising uracil or its derivatives from the mammalian genome (8, 18, 19). Thus, FU toxicity in S. cerevisiae would not be a relevant model for cancer Introduction chemotherapy (20). Studies in mammalian cells show that FU 5-Fluorouracil (FU) is widely used in the treatment of a range of cytotoxicity is not mediated by Ung-initiated excision of uracil cancers, particularly colorectal cancer but also cancers of the from DNA (9) but have not addressed the role of Smug1. Here, breast, head-and-neck, and aerodigestive tract (1). After four using gene-targeted cell lines defective in one or both of the decades in clinical use, FU or an oral prodrug (2) is presently used mammalian uracil-DNA glycosylases, for the first time, we were in the treatment of some two million individuals a year (3). able to show the contribution of FU incorporation in DNA to FU The fluorine-substituted uracil analogue is an antimetabolite that cytotoxicity and the relevance of the Smug1 uracil-DNA glycosylase was designed to inhibit DNA synthesis in an early example of in protecting against cell killing (Fig. 1). the rational development of a cytotoxic drug (4). Inhibition of thymidylate synthase (TS) by FdUMP prevents the conversion Materials and Methods of dUMP to the sole de novo source of dTMP (Fig. 1), depleting Cell lines. The Ung / mouse embryo fibroblast (MEF) cell line, isogenic the cell of dTTP for use in DNA replication and repair. This leads wild-type (Ung+/+) control, and stable small interfering RNA–mediated to deoxynucleoside triphosphate precursor pool imbalances and knockdown of Smug1 in both a Ung+/+ (Smug1# cell line) and a Ung / increased levels of dUTP. In a frequently proposed model, background (Smug1#, Ung / cell line) have been described (19). Cells were FU-mediated cell death is attributed to the use of dUTP in place cultured in DMEM/Ham’s F12 (3:1) with 10% fetal bovine serum. of dTTP during DNA synthesis, and subsequent DNA fragmenta- High-performance liquid chromatography analysis of bases tion due to extensive uracil excision in newly synthesized DNA or released from the DNA of FU-treated cells by recombinant Smug1 and Ung. Exponentially growing cells (f107) were treated with FU-2-14C repeated futile repair attempts in the presence of a high dUTP/ (53 mCi mmol 1;Sigma, St. Louis, MO), treated with nonradiolabeled FU and dTTP ratio (reviewed in ref. 1). TS status has been correlated with labeled with uracil-2-14C (52 mCi mmol 1;Sigma), or treated with nonradiolabeled FU. The concentration of FU (and uracil) used was 10 Amol/L and cells were exposed to FU for 48 h (Vtwo cell divisions). Requests for reprints: Tomas Lindahl, Cancer Research UK London Research Genomic DNA was isolated from the cells (Wizard Genomic DNA Institute, Clare Hall Laboratories, South Mimms, Hertfordshire, EN6 3LD, United purification kit;Promega, Madison, WI) and 10 Ag 14C-labeled DNA or Kingdom. Phone: 44-17-0762-5993/5995;Fax: 44-20-7269-3803;E-mail: tomas.lindahl@ 15 Ag nonradiolabeled DNA was incubated with recombinant human cancer.org.uk. I2007 American Association for Cancer Research. SMUG1 and UNG in standard enzyme reactions (see below). After incubation doi:10.1158/0008-5472.CAN-06-2960 at 37jC for 1 h, DNA was precipitated with ethanol and the supernatant

Cancer Res 2007; 67: (3). February 1, 2007 940 www.aacrjournals.org

Figure 1. FU metabolism and the role of Smug1. FU is converted to the active precursors FUTP, FdUTP, and FdUMP (boldface); these can be incorporated into RNA, DNA, or inhibit TS, respectively. The key steps associated with FU sensitivity (red) or FU resistance (green). DHFU, dihydrofluorouracil. Dashed lines indicate our main findings here: (a) incorporation of FU into DNA is cytotoxic; (b) Smug1 excises FU from DNA in vivo and protects cells from FU killing.

analyzed by high-performance liquid chromatography (HPLC), as described line (8) and stable small interfering RNA–mediated knockdown of previously (19). Fractions were collected at 0.5 min intervals, and released Smug1 (Smug1#) in both the Ung / and Ung+/+ background (19). DNA bases detected by scintillation counting or UV absorbance at 254 nm; In order to determine whether FU is incorporated and persists in reference compounds were detected by UV absorbance. the DNA of drug-treated uracil-DNA glycosylase-deficient cells, the In vitro DNA glycosylase assays. Standard uracil-DNA glycosylase assays Ung / cell line, Smug1# and Smug1#, Ung / cell lines were were carried out using recombinant human SMUG1 or UNG proteins in 20 14 mmol/L of Tris-HCl (pH 8.0), 50 mmol/L of NaCl, 1 mmol/L of DTT, exposed to C-labeled FU, genomic DNA was isolated from live 1 mmol/L of EDTA, 100 AgmL1 bovine serum albumin, 1 ng of AP cells after 48 h of FU exposure, and samples were digested with endonuclease, and 0.24 pmol of DNA substrate at 37jC for 1 h, as described recombinant Smug1 and Ung. Radiolabeled ethanol-soluble prod- previously (19). The DNA substrate contained a centrally placed U:A or FU:A ucts released by the recombinant proteins were then resolved by base pair in the 5¶-32P-end–labeled strand of a 19-mer double-stranded HPLC and identified by comparison with reference compounds oligonucleotide. After the addition of 1 mol/L of piperidine and incubation at (Fig. 2A). HPLC analysis resolved a small discrete peak of 90jC for 20 min, samples were dried under vacuum, denatured in 95% radioactive material in DNA from wild-type cells that was formamide, and analyzed by 10% PAGE and phosphorimaging. Oligonucleo- coincident with a FU marker;a similar small peak of 14C-FU was 1 tides were synthesized by MWG-Biotech; the FU-containing oligonucleotide detected in Ung / cells. This corresponded to 1.5 Â 105 FU was HPLC-purified and certified by MALDI-TOF mass spectrometry. residues per diploid genome in both wild-type and Ung / cells FU cell survival assays. Exponentially growing cells plated in triplicate 14 in 100 mm dishes were treated at a range of drug doses with FU or (Table 1). In marked contrast, a large peak of C-FU was detected # # / fluorodeoxyuridine (FUdR), and surviving colonies were scored after 10 to in DNA from both the Smug1 and Smug1 , Ung cell lines, Â 6 Â 6 12 days in culture (as a percentage of the untreated control and after corresponding to 4.3 10 and 3.2 10 FU residues per diploid adjustment for plating efficiency), essentially as described (19). Where cells genome, respectively (Table 1);a slightly lower level of FU were supplemented with thymidine, 100 Amol/L of thymidine was given incorporated in the double mutant may reflect the slower doubling along with 1 Amol/L of deoxycytidine (21). FU, FUdR, and thymidine were time of this cell line (data not shown) and the limited labeling obtained from Sigma. period. The data shows that there is substantial incorporation of Generation of stable Smug1-overexpressing MEF cell lines. The FU in the DNA of drug-treated cells, with no effect of Ung murine Smug1 open reading frame (837 bp;ref. 17) was cloned under the deficiency on FU levels but a 20- to 30-fold increase in the level of control of the cytomegalovirus promoter in the pTARGET Mammalian FU in the genome of Smug1-deficient cells (Fig. 2A;Table 1);this is Expression System (Promega). The construct was linearized with ScaI and f transfected into the wild-type MEF cell line using Polyfect Transfection equivalent to the substitution of 1 in 500 thymine residues. Reagent (Qiagen, Chatsworth, CA). G418-resistant colonies were isolated Similar experiments were conducted in which cells were treated following culture in 1.4 mg mL 1 of G418 and the level of Smug1 expression with the same molar concentration of FU as above but this was 14 was assayed by quantitative real-time PCR of reverse-transcribed total RNA, nonradiolabeled;cells were instead labeled with C-uracil in order as described previously (19). Colonies showing elevated Smug1 mRNA levels to examine the levels of uracil incorporated into the DNA. These were subcultured for three to four passages in G418, and Smug1 expression experiments showed that uracil accumulates in cellular DNA of FU- was re-assessed and confirmed by Western blotting with antibodies against treated cells, but to a lesser degree than FU (Fig. 2B), and in wild- human SMUG1 (17). type cells, there were 8.2 Â 104 uracil residues per diploid genome (Table 1). However, unlike FU in DNA, uracil levels were affected by Results the loss of either Ung or Smug1. Thus, there was an increase of Accumulation of FU in the DNA of FU-treated Smug1- uracil in DNA in Ung / as well as Smug1# cells, and an additive deficient cells. We have previously generated a Ung / MEF cell 10-fold increase (8.2 Â 105 uracil residues per diploid genome) in the double mutant. The accumulated level of uracil was 3- to 4-fold less than that of FU in the Smug1#, Ung / cell line (Table 1), 1 http://www.mwg-biotech.com/html/all/index.php. representing the substitution of f1 in 2,000 thymine residues. Both www.aacrjournals.org 941 Cancer Res 2007; 67: (3). February 1, 2007

FU and U were also detected, by UV absorbance in comparison Table 1. Incorporation of FU and U in the genome of # / with known quantities of reference standards, in Smug1 , Ung uracil-DNA glycosylase-deficient cell lines cells treated with nonradiolabeled FU, and were present at levels equivalent to those observed in radiolabeling experiments (Fig. 2C). Cell line Base residues per diploid genome When total RNA as well as genomic DNA was isolated from 14 duplicate samples of FU-treated cells, the ratio of C-FU FU U (cpm Ag 1) incorporated into RNA/DNA in the wild-type and Ung / cell lines was f11:1, in agreement with other studies (15), Wild-type 1.5 Â 105 8.2 Â 104 and this ratio was reduced in the Smug1# and Smug1#, Ung / cell Ung / 1.5 Â 105 2.1 Â 105 lines, consistent with increased levels of FU in the DNA of Smug1- Smug1# 4.3 Â 106 4.5 Â 105 / Â 6 Â 5 deficient cells (data not shown). Our data indicate that whereas Smug1#, Ung 3.2 10 8.2 10 both Smug1 and Ung excise uracil incorporated in U:A base pairs in vivo, the Smug1 uracil-DNA glycosylase—but not Ung—excises NOTE: Values were calculated for cells treated for 48 h with 10 Amol/L FU arising through the use of FdUTP instead of dTTP during DNA of 14C-FU, or 10 Amol/L of nonradiolabeled FU plus 14C-uracil, as synthesis (Fig. 1). shown in Fig. 2A and B. In vitro DNA repair assays. In vitro, the Ung and Smug1 uracil- DNA glycosylases efficiently excise uracil from U:A base pairs in

nucleosomes as well as in naked DNA (22). Both enzymes have also been reported to excise FU from DNA in vitro (16). In light of our data demonstrating the accumulation of FU in the DNA of Smug1- but not Ung-deficient cells, repair assays were conducted to examine the FU-excision activity of Smug1 and Ung in vitro.We assayed the relative activity of each enzyme on a double-stranded oligonucleotide substrate containing a centrally placed FU:A or a U:A base pair. Both enzymes are considerably less active on a FU:A–containing versus a U:A–containing substrate but, whereas Ung activity is z100Â reduced, Smug1 activity is only f10Â reduced on a FU:A versus U:A base pair (Fig. 3). FU:A excision activity could not be detected in wild-type cell-free extracts (data not shown) so it was not possible to assess the contribution of Smug1 and Ung to the total cellular FU-excision activity. However, the greatly reduced activity of recombinant Ung on FU:A versus U:A in vitro (Fig. 3) correlates with the failure of Ung to contribute to FU repair in vivo (Fig. 2A;Table 1). Conversely, the much more robust activity of recombinant Smug1 on FU:A base pairs is consistent with the accumulation of FU in the DNA of Smug1- deficient FU-treated cells and the conclusion that Smug1 excises FU incorporated into FU:A DNA base pairs in vivo. Sensitivity of Smug1-deficient cell lines to FU killing. In order to determine whether the observed accumulation of FU in the genome of Smug1-deficient cells had any effect on cell viability, we compared the sensitivity of the Smug1# and Smug1#, Ung / MEF cell lines to FU killing, with that of the Ung / cell line and wild-type control. Smug1-deficiency significantly altered the cellular response to FU (Fig. 4A) and the Smug1# cell line was f A 2-fold more sensitive to FU (D37 =3.2 mol/L) than the wild- type control (D37 =6.8Amol/L; D37 is the drug dose that results in 37% cell survival). In agreement with previous data (9), there was no significant difference in survival of the Ung / cell line treated with FU in comparison with the wild-type control. Furthermore, there was no additive effect due to loss of Ung, with Smug1#, Ung / cells showing the same sensitivity as the Smug1# cell line. The relative sensitivity of the four cell lines to killing Figure 2. Accumulation of FU and/or uracil in the genome of FU-treated cells. by FU correlates with the levels of FU measured in cellular DNA; Cells were treated for 48 h with 10 Amol/L of FU (and uracil) as follows: 14C-FU it does not correlate with the levels of uracil in DNA (Fig. 2A; (A), nonradiolabeled FU and 14C-uracil (B), and nonradiolabeled FU (C). Genomic DNA was isolated and incubated with recombinant human SMUG1 and Table 1). FU in DNA was measured after 48 h treatment of cell 14 UNG proteins. Ethanol-soluble material was analyzed by HPLC, and released lines with 10 Amol/L of C-FU (Fig. 2A);this same drug dose DNA bases detected by scintillation counting (A and B) or UV absorbance at resulted in 13.5% versus 4.2% survival of wild-type and Smug1# 254 nm (C) in comparison with reference compounds. A and B, wild-type (n), Ung/ (E), Smug1# (5), and Smug1#, Ung/ (4) cell lines; C, Smug1#, cells, respectively (Fig. 4A). Thus, the persistence of FU in DNA, Ung/ cell line. rather than the excision of uracil, is the predominant means of

Cancer Res 2007; 67: (3). February 1, 2007 942 www.aacrjournals.org

real-time PCR and verified by Western blotting with Smug1-specific antibodies. Two sublines expressing f1.5Â and f2.5Â the wild- type level of Smug1 protein were assayed for cellular sensitivity to FU in comparison with the parental wild-type line (Fig. 5). There was a significantly increased resistance to FU in the cell line expressing f2.5Â Smug1 protein relative to the wild-type parental line (D37 ratio 1.7:1).

Discussion We have shown that the Smug1 DNA glycosylase excises FU from DNA in vivo;extensive accumulation of FU in the genome of Figure 3. In vitro activity of the Smug1 and Ung enzymes on FU:A DNA base pairs. DNA glycosylase activity of recombinant human SMUG1 or UNG Smug1-deficient cells sensitizes them to FU killing, providing direct proteins was assayed on a 19-mer double-stranded oligonucleotide substrate proof that FU incorporated in DNA is cytotoxic (Fig. 1). The level of containing a centrally placed U:A or FU:A base pair in the 5¶-32P-end–labeled FU in DNA measured here in wild-type cells is comparable to early strand, as indicated. After cleavage of the resultant abasic site, the 9-mer 32P-labeled product was resolved by PAGE. reports for normal or tumor cell lines in which the occurrence of FU in DNA was first described (12–15). However, the extent to which FU becomes incorporated in DNA and its relevance to FU FU-mediated cell killing. Our data show that accumulation of FU cytotoxicity is only revealed here in a Smug1-deficient cell line in DNA is cytotoxic and base excision repair of FU initiated by (Fig. 2A). There is an appreciable level of FU in DNA even in the Smug1 protects cells from FU killing. presence of Smug1 (Table 1), and it seems likely that this would As with FU, Smug1# and Smug1#, Ung / cells treated with also contribute to the toxicity of FU in wild-type cells. Consistent FUdR showed significantly increased sensitivity to the fluoropyr- with this, a modest f2.5-fold increase in the level of Smug1 protein imidine, whereas there was again no marked difference in the survival of Ung / versus wild-type cells (Fig. 4B). Thus, Smug1- deficient cells (D37 = 11.6 Amol/L) were f2-fold more sensitive to FUdR than wild-type cells (D37 = 20.8 Amol/L). This indicates that FU is introduced into the DNA by treatment with FUdR as well as FU, and hypersensitivity of Smug1# cells to FUdR is, as with FU treatment, due to the deficiency in Smug1-initiated base excision repair of FU incorporated in place of thymine in DNA. The cell is able to synthesize dTMP de novo via a nonessential salvage pathway catalyzed by thymidine kinase, bypassing the effects of TS inhibition and restoring the dTTP pool;due to the allosteric regulation of deoxynucleoside triphosphate synthesis, cells were supplemented with thymidine as well as a trace of deoxycytidine (21). Thymidine supplementation may not affect the use of FdUTP in DNA synthesis (Fig. 1). Consistent with this, supplementing FUdR-treated cells with thymidine decreased the sensitivity of all four cell lines but Smug1# cells still remained more sensitive than Ung / and wild-type cells (Fig. 4B). Both wild-type and Smug1-deficient cell lines were 3- to 4-fold more sensitive to killing by FU than FUdR (note different dose ranges on the X-axes in Fig. 4A and B), and although thymidine supplementation also proportionately decreased the sensitivity of both cell lines to FU, it was less effective in rescuing FU-treated compared with FUdR-treated cells (data not shown). These data indicate that FU-mediated killing of wild-type cells is predominantly due to active metabolites that are preferentially produced from FU rather than FUdR;these are likely to be the RNA and DNA precursors, FUTP and FdUTP, rather than the TS-inhibitory FdUMP (Fig. 1). Overexpression of Smug1 protects cells against FU-mediated cell killing. We have shown that FU incorporation in DNA is a predominant means of FU-mediated cell killing and that cells deficient in DNA base excision repair of FU by Smug1 are sensitized to cell killing. We therefore investigated whether overexpression Figure 4. Smug1-deficient cells were hypersensitive to killing by FU. Wild-type (n), Ung/ (E), Smug1# (5), and Smug1#, Ung/ (4) MEF cell lines were of Smug1 in wild-type cells might increase cellular resistance to FU treated with FU (A) or FUdR( B) at the doses indicated and surviving colonies killing. A construct expressing murine Smug1 under the control of scored after 10 to 12 days in culture (as a percentage of the untreated control). the cytomegalovirus promoter was transfected into the wild-type Points, mean from three experiments; bars, SE. B, FUdR-treated cells were grown either with (- - -) or without (—) supplementation with 100 Amol/L of MEF cell line. Sublines were established that stably expressed thymidine and 1 Amol/L of deoxycytidine. Note the different dose ranges on the moderately elevated levels of Smug1, as ascertained by quantitative X-axes in (A)and(B). www.aacrjournals.org 943 Cancer Res 2007; 67: (3). February 1, 2007

overcome TS-inhibition and restore the dTTP pool (Fig. 1) only partially rescued FU-treated wild-type cells and did not reduce the hypersensitivity of Smug1-deficient cells to either FU or FUdR, as the use of FdUTP in DNA synthesis remained unaltered. It is presently unclear why FU in DNA is cytotoxic. We cannot formally exclude the possibility that FU bases themselves are benign substitutions in DNA but that instead a small proportion are converted to a highly toxic lesion in situ. However, massive substitution of f1 in 500 thymine residues with FU in the genome of Smug1-deficient cells (Table 1) would itself be expected to affect fundamental processes, such as the transcription of essential genes or the establishment of the chromatin structure, by altering DNA dynamics in A:T–rich regions (25, 26) or interfering with protein- DNA interactions (27). Such processes might also be compromised if there were nonproductive binding to FU residues in DNA by another DNA repair/processing enzyme. Mismatch repair (28), and Figure 5. Overexpression of SMUG1 increases the FU-resistance of the MBD4 and TDG mismatch-specific DNA glycosylases (29, 30), wild-type cells. A wild-type MEF cell line (n), and stable transfected clones have also been reported to act on FU in vitro but in the context of a overexpressing recombinant SMUG1 protein, at f2.5-fold (.)orf1.5-fold (o) wild-type levels, were treated with FU at the doses indicated and surviving FU:G mispair;use of FdUTP during DNA synthesis typically gives colonies scored (as a percentage of the untreated control), as in Fig. 4. rise to an FU:A base pair (25) but rare mutagenic FU:G mispairs could theoretically arise in vivo (26, 31). In contrast to Smug1- significantly increased the resistance of FU-treated cells to killing. deficient cells, and by an as yet unexplained mechanism, Mbd4 / Thus, the excision of FU from DNA by Smug1 may be relatively MEFs (29), and mismatch repair–defective cell lines and tumors inefficient, in agreement with the limited FU excision capacity (28) are apparently resistant to FU. Thus, rare FU:G base pairs observed in normal mammalian cells (23). We have shown in might arise upon replication of FU:A–containing DNA and be in vitro assays that Ung has greatly reduced activity on FU:A versus abortively processed by mismatch repair or a mismatch-specific U:A base pairs, which would explain why Ung does not contribute DNA glycosylase. Elevated levels of FU in DNA did not seem to be to FU repair in vivo. Furthermore, as Ung but not Smug1 is sorted associated with increased DNA fragmentation in FU-treated to the mitochondria as well as the nucleus (24), this would leave Smug1-deficient cells (data not shown). TDG, which is a very the mitochondrial genome vulnerable to incorporated FU and a minor uracil-excision activity compared with UNG or SMUG1 (8), possible target of FU toxicity in wild-type cells. SMUG1 is able to has been reported to act on FU:A in vitro (30);if this is relevant excise uracil derivatives with substitutions at the C5 position of the in vivo and whether ablation of Tdg might modify the effect of pyrimidine ring, as the aromatic residue that forms a steric barrier Smug1-deficiency remains to be determined. to binding of C5-modified bases in UNG is not conserved in the Our data show that the introduction of FU into DNA and its active site structure of SMUG1 (18). Our data are consistent with contribution to FU-mediated cell killing has been previously this, demonstrating considerably less discrimination by Smug1 underestimated. This is particularly relevant as formation of than Ung against FU:A versus U:A base pairs in vitro, and the FdUTP and its use in DNA synthesis seems to occur irrespective contribution of Smug1 but not Ung to FU repair in vivo (Fig. 4). of TS inhibition and the level of the TTP pool in FU-treated cells Smug1-deficient cells are sensitized to killing by FU, with no (Fig. 4), being more clearly correlated with ribonucleotide effect of Ung on FU survival. Thus, FU cytotoxicity is directly metabolism and dependent on the activity of ribonucleotide correlated with the loss of Smug1 FU-excision activity and greatly reductase converting FUDP to FdUDP (Fig. 1). The relationship increased levels of FU in DNA. The accumulation of uracil in the of FU in DNA to both the strategy and efficacy of FU treatment in DNA of Smug1# cells as well as Ung / cells, with an up to 10-fold patients should be further evaluated (2, 3, 32). Our data shows that increase in the Smug1#, Ung / double mutant, was not correlated even a 2- to 3-fold increase in Smug1 protein significantly increases with FU survival. As Smug1#, Ung / cells lacking uracil-DNA the resistance of FU-treated cells to killing, such that Smug1 up- glycosylase activity are sensitized to killing, the data clearly oppose regulation could provide a previously unrecognized mechanism of a model in which FU toxicity is due to the excision of uracil from acquired drug resistance in tumors. This may be particularly DNA. Our data do not directly address whether specific inhibition relevant where FU is used in combination with radiotherapy (3), as of TS kills cells due to elevation of the dUTP precursor pool Smug1 also helps protect cells against killing by low-level ionizing and associated base excision repair of uracil in DNA, but the FU radiation (19). Gene products identified to date as markers of FU hypersensitivity of Smug1# cells versus unaltered resistance of resistance have invariably been involved in metabolizing the drug Ung / cells shows that use of FdUTP in DNA synthesis rather than or modulating TS inhibition, rather than directly in repairing FU FdUMP-mediated inhibition of TS is the predominant mechanism DNA damage (Fig. 1), but it would now seem pertinent to analyze by which FU kills cells. Consistent with this, all MEF lines here were the Smug1 enzyme during prolonged FU exposure and in gene considerably less sensitive to FUdR than to FU, as has been expression profiles of FU-resistant tumors (33, 34). described for other murine and human cell lines (15). FUdR would be expected to exert its effects primarily through TS inhibition, as Acknowledgments FUdR is converted directly to FdUMP, whereas the conversion of FU Received 8/9/2006;accepted 11/20/2006. to FUMP allows incorporation in RNA but also DNA via reduction Grant support: Cancer Research UK. The costs of publication of this article were defrayed in part by the payment of page of the ribonucleoside diphosphate, which would be unaffected by charges. This article must therefore be hereby marked advertisement in accordance TS inhibition (Fig. 1). Thus, supplementing cells with thymidine to with 18 U.S.C. Section 1734 solely to indicate this fact.

Cancer Res 2007; 67: (3). February 1, 2007 944 www.aacrjournals.org

References 12. Kufe DW, Major PP, Egan EM, Loh E. 5-Fluoro-2¶- alternative splicing and transcription from different deoxyuridine incorporation in L1210 DNA. J Biol Chem positions in the UNG gene. Nucleic Acids Res 1997;25: 1. Longley DB, Harkin DP, Johnston PG. 5-Fluorouracil: 1981;256:8885–8. 750–5. mechanisms of action and clinical strategies. Nat Rev 13. Ingraham HA, Tseng BY, Goulian M. Nucleotide levels 25. SowersLC,EritjaR,KaplanB,GoodmanMF, Cancer 2003;3:330–8. and incorporation of 5-fluorouracil and uracil into DNA Fazarkerley GV. Structural and dynamic properties of 2. Miwa M, Ura M, Nishida M, et al. Design of a novel of cells treated with 5-fluorodeoxyuridine. Mol Pharma- a fluorouracil-adenine base pair in DNA studied by oral fluoropyrimidine carbamate, capecitabine, which col 1982;21:211–6. proton NMR. J Biol Chem 1987;262:15436–42. generates 5-fluorouracil selectively in tumours by 14. Schuetz JD, Wallace HJ, Diasio RB. 5-Fluorouracil 26. Habener JF, Vo CD, Le DB, Gryan GP, Ercolani L, enzymes concentrated in human liver and cancer tissue. incorporation into DNA of CF-1 mouse bone marrow Wang AH-J. 5-Flurodeoxyuridine as an alternative to the Eur J Cancer 1998;34:1274–81. cells as a possible mechanism of toxicity. Cancer Res synthesis of mixed hybridization probes for the 3. Rich TA, Shepard RC, Mosley ST. Four decades of 1984;44:1358–63. detection of specific gene sequences. Proc Natl Acad continuing innovation with fluorouracil: current and 15. Peters GJ, Laurensse E, Leyva A, Lankelma J, Pinedo Sci U S A 1988;85:1735–9. future approaches to fluorouracil chemoradiation ther- HM. Sensitivity of human, murine, and rat cells to 5- 27. Pu WT, Struhl K. Uracil interference, a rapid and apy. J Clin Oncol 2004;22:2214–32. fluorouracil and 5¶-deoxy-5-fluorouridine in relation to general method for defining protein-DNA interactions 4. Heidelberger C, Chaudhari NK, Danenberg P. Fluori- drug-metabolizing enzymes. Cancer Res 1986;46:20–8. involving the 5-methyl group of thymines: the GCN4- nated pyrimidines: a new class of tumor inhibitory 16. Kavli B, Sundheim O, Akbari M, et al. hUNG2 is the DNA complex. Nucleic Acids Res 1992;20:771–5. compounds. Nature 1957;179:663–6. major repair enzyme for removal of uracil from U:A 28. Meyers M, Hwang A, Wagner MW, et al. A role for 5. Peters GJ, Backus HHJ, Freemantle S, et al. Induction matches, U:G mismatches, and U in single-stranded DNA mismatch repair in sensing and responding to of thymidylate synthase as a 5-fluorouracil resistance DNA, with hSMUG1 as a broad specificity backup. J Biol fluoropyrimidine damage. Oncogene 2003;22:7376–88. mechanism. Biochim Biophys Acta 2002;1587:194–205. Chem 2002;277:39926–36. 29. Cortellino S, Turner D, Masciullo V, et al. Base 6. Canman CE, Radany EH, Parsels LA, Davis MA, 17. Nilsen H, Haushalter KA, Robins P, Barnes DE, excision repair enzyme MED1 mediates DNA damage Lawrence TS, Maybaum J. Induction of resistance to Verdine GL, Lindahl T. Excision of deaminated cytosine response to antitumor drugs and is associated with fluorodeoxyuridine cytotoxicity and DNA damage in from the vertebrate genome: role of the SMUG1 uracil- mismatch repair system integrity. Proc Natl Acad Sci human tumor cells by expression of Escherichia coli DNA glycosylase. EMBO J 2001;20:4278–86. U S A 2003;100:15071–6. deoxyuridinetriphosphatase. Cancer Res 1994;54:2296–8. 18. Wibley JE, Waters TR, Haushalter K, Verdine GL, 30. Hardeland U, Bentele M, Jiricny J, Scha¨r P. The 7. Koehler SE, Ladner RD. Small interfering RNA- Pearl LH. Structure and specificity of the vertebrate anti- versatile thymine DNA-glycosylase: a comparative char- mediated suppression of dUTPase sensitizes cancer mutator uracil-DNA glycosylase SMUG1. Mol Cell 2003; acterisation of the human, Drosophila and fission yeast cells to thymidylate synthase inhibition. Mol Pharmacol 11:1647–59. orthologs. Nucleic Acids Res 2003;31:2261–71. 2004;66:620–6. 19. An Q, Robins P, Lindahl T, Barnes DE. C!T 31. SowersLC,EritjaR,KaplanB,GoodmanMF, 8. Nilsen H, Rosewell I, Robins P, et al. Uracil-DNA mutagenesis and g-radiation sensitivity due to deficien- Fazarkerley GV. Equilibrium between a wobble and glycosylase (UNG)-deficient mice reveal a primary role cy in the Smug1 and Ung DNA glycosylases. EMBO J ionized base pair formed between flurouracil and of the enzyme during DNA replication. Mol Cell 2000;5: 2005;24:2205–13. guanine in DNA as studied by proton and fluorine 1059–65. 20. Seiple L, Jaruga P, Dizdaroglu M, Stivers JT. Linking NMR. J Biol Chem 1988;263:14794–801. 9. Andersen S, Heine T, Sneve R, et al. Incorporation of uracil base excision repair and 5-fluorouracil toxicity in 32. Noordhuis P, Holwerda U, van der Wilt CL, et al. dUMP into DNA is a major source of spontaneous DNA yeast. Nucleic Acids Res 2006;34:140–51. 5-Fluorouracil incorporation into RNA and DNA in damage, while excision of uracil is not required for 21. Reichard P. Interactions between deoxyribonucleotide relation to thymidylate synthase inhibition of human cytotoxicity of fluoropyrimidines in mouse embryonic and DNA synthesis. Annu Rev Biochem 1988;57:349–74. colorectal cancers. Ann Oncol 2004;15:1025–32. fibroblasts. Carcinogenesis 2005;26:547–55. 22. Nilsen H, Lindahl T, Verreault A. DNA base excision 33. SalongaD,DanenbergKD,JohnsonM,etal. 10. Welsh J, Hobbs S, Aherne GW. Expression of uracil repair of uracil in reconstituted nucleosome core Colorectal tumours responding to 5-fluorouracil have DNA glycosylase does not affect cellular sensitivity to particles. EMBO J 2002;21:5943–52. low gene expression levels of dihydropyrimidine dehy- thymidylate synthase (TS) inhibition. Eur J Cancer 2003; 23. Mauro DJ, De Riel JK, Tallarida RJ, Sirover MA. drogenase, thymidylate synthase, and thymidine phos- 39:378–87. Mechanisms of excision of 5-fluorouracil by uracil DNA phorylase. Clin Cancer Res 2000;6:1322–7. 11. Caradonna SJ, Cheng Y-C. The role of deoxyuridine glycosylase in normal human cells. Mol Pharmacol 1993; 34. Wang W, Cassidy J, O’Brien V, Ryan KM, Collie- triphosphate nucleotidohydrolase, uracil-DNA glycosy- 43:854–7. Duguid E. Mechanistic and predictive profiling of lase, and DNA polymerase a in the metabolism of FUdR 24. Nilsen H, Otterlei M, Haug T, et al. Nuclear and 5-fluorouracil resistance in human cancer cells. Cancer in human cells. Mol Pharmacol 1980;18:513–20. mitochondrial uracil-DNA glycosylases are generated by Res 2004;64:8167–76.

www.aacrjournals.org 945 Cancer Res 2007; 67: (3). February 1, 2007

Downloaded from cancerres.aacrjournals.org on September 28, 2021. © 2007 American Association for Cancer Research. 5-Fluorouracil Incorporated into DNA Is Excised by the Smug1 DNA Glycosylase to Reduce Drug Cytotoxicity

Qian An, Peter Robins, Tomas Lindahl, et al.

Cancer Res 2007;67:940-945.

Updated version Access the most recent version of this article at: http://cancerres.aacrjournals.org/content/67/3/940

Cited articles This article cites 34 articles, 19 of which you can access for free at: http://cancerres.aacrjournals.org/content/67/3/940.full#ref-list-1

Citing articles This article has been cited by 12 HighWire-hosted articles. Access the articles at: http://cancerres.aacrjournals.org/content/67/3/940.full#related-urls

E-mail alerts Sign up to receive free email-alerts related to this article or journal.

Reprints and To order reprints of this article or to subscribe to the journal, contact the AACR Publications Subscriptions Department at [email protected].

Permissions To request permission to re-use all or part of this article, use this link http://cancerres.aacrjournals.org/content/67/3/940. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC) Rightslink site.