Handling the 3-methylcytosine lesion by six humanDNA polymerases members of the B- X- andY-familiesAntonia Furrer and Barbara van Loon

Institute of Veterinary Biochemistry and Molecular Biology University of Zurich Winterthurerstrasse 190 8057Zurich Switzerland

Received July 2 2013 Revised August 26 2013 Accepted September 10 2013

ABSTRACT

Alkylating agents often generate 3-methylcytosine(3meC) lesions that are efficiently repaired by AlkBhomologues If AlkB homologue proteins are notfunctional or the number of 3meC lesions exceedsthe cellular repair capacity the damage will persistin the genome and become substrate of DNA poly-merases (Pols) Though alkylating agents arepresent in our environment and used in the clinicscurrently nothing is known about the impact of3meC on the accuracy and efficiency of humanPols Here we compared the 3meC bypassproperties of six human Pols belonging to thethree families B (Pol d) X (Pols b and j) and Y(Pols i i and g) We show that under replicative con-ditions 3meC impairs B-family blocks X-family butnot Y-family Pols in particular Pols g and i ThesePols successfully synthesize opposite 3meC Pol ipreferentially misincorporates dTTP and Pol gdATP The most efficient extenders from 3meCbase-paired primers are Pols i and g Finally usingxeroderma pigmentosum variant patient cellextracts we provide evidence that the presence offunctional Pol g is mandatory to efficiently overcome3meC by mediating complete bypass or extensionOur data suggest that Pol g is crucial for efficient3meC bypass

INTRODUCTION

Living organisms are exposed to alkylating agents presentin shape of endogenous sources such as cellular metabol-ism exogenously in the environment and in clinics as oneof the most frequently used cancer chemotherapeuticdrugs (12) Exposure of genomic DNA to alkylating

agents often results in cytotoxic andor carcinogenic mu-tations (3) Consequently alkylation chemotherapy wasshown to potentially induce secondary cancerscharacterized by genome instability and mutation accu-mulation (4) Alkylation DNA lesions induced by SN2alkylating agents include 3-methylcytosine (3meC)3-methylthymine 1-methyladenine and 1-methylguanine(5) 3meC lesions are predominantly generated in single-stranded (ss) DNA during replication repair and tran-scription In double-stranded (ds) DNA the reactive siteis involved in base pairing and thus protected from modi-fication (6) The 3meC lesions are efficiently repaired viadirect oxidative reversal mediated by iron(II) and2ketoglutarate (2KG)dependent dioxygenases alkylationprotein B (AlkB) in prokaryotes (7) and the AlkB homo-logues (ALKBH) 2 and 3 in eukaryotes (89) AlkB defi-cient Escherichia coli cells have elevated amount ofalkylated lesions display increased mutation frequencyand are hypersensitive to alkylating agents (1011)Similarly exposure of mammalian cells lackingALKBH2 to the alkylating agent methylmethane sulfon-ate (MMS) results in increased cytotoxicity andmutagenicity (1213) Compared with ALKBH2 profi-cient the deficient cells are characterized with largeincrease in CA and to certain extent CT mutationswhen treated with MMS (13) Additionally ALKBH2 hasbeen implicated in cancerogenesis by influencing cancerprogression as well as sensitivity to cisplatin treatment(1415) Efficient removal of alkylation DNA damage isthus critical for cellular function preservation of genomicintegrity and survivalIncreased cytotoxicity and mutation frequency due to

loss of functional AlkB or ALKBH2 suggest an impairedaction of DNA polymerases (Pols) at the site ofunrepaired alkylation damage Besides a malfunctionaldirect reversal pathway Pols can face 3meC in case (i)when the amount of DNA damage exceeds the cellularrepair capacity as well as (ii) if alkylation damage of

To whom correspondence should be addressed Tel +41 44 635 54 67 Fax +41 44 635 68 40 Email bvanloonvetbiouzhch

Published online 4 October 2013 Nucleic Acids Research 2014 Vol 42 No 1 553ndash566doi101093nargkt889

The Author(s) 2013 Published by Oxford University PressThis is an Open Access article distributed under the terms of the Creative Commons Attribution Non-Commercial License (httpcreativecommonsorglicensesby-nc30) which permits non-commercial re-use distribution and reproduction in any medium provided the original work is properly cited For commercialre-use please contact journalspermissionsoupcom

ssDNA occurs during DNA replication or repair Uponencountering the DNA lesion Pols can either (i) be com-pletely blocked (ii) incorporate a nucleotide opposite thedamage and then be blocked due to inability to extend or(iii) fully bypass by base pairing and extending past thelesion Understanding the role(s) of the different Pols inthe bypass of alkylation damage is complicated by thenumber of different Pols present in every organism witha broad range of catalytic properties For example theprokaryotic organism E coli contains five different Pols(16) whereas in eukaryotes sixteen Pols have beenidentified and based on the sequence homologygrouped in five familiesmdashA B X Y and reverse tran-scriptase (17) Addition of a methyl group to endocyclicN-atoms and subsequent generation of 3meC 3-methylthymine 1-methyladenine and 1-methylguaninelesions leads to DNA replication block in E coli poten-tially due to impairment of B-family replicative Pol III(10) In addition the E coli A-family gap-filling Pol I isinhibited by 3meC damage and besides certain incorpor-ation of dGTP it can misincorporate dATP and dTTPopposite the lesion (1819) While the replicative Pols arecharacterized with a high fidelity enabled by a tight-fittingactive site and 3050 proofreading activity the active sitesof X- and Y-family translesion synthesis (TLS) Polswhich also lack a 3050 proofreading exonucleaseactivity can readily accommodate and efficiently bypassdifferent DNA damages At the same time this results inan increased error-rate on undamaged DNA (20ndash22)Accumulation of the Y-family TLS Pols as result ofSOS-response activation in E coli increases bypass effi-ciency of the 3-methylpirimidine and 1-methylpurinelesions by 3ndash4-fold (1011) A recent structural studyprovided valuable insight into the features of prokaryoticY-family Dpo4-mediated bypass of 3meC lesion (23)Presence of 3meC results in inhibition of Dpo4 and gen-eration of 60 mutagenic products during remainingbypass (23) Crystal structure of Dpo4 ternary extensioncomplex with correct and mismatched 30-terminal primerbase opposite the 3meC lesion indicates that the lesionremains positioned within the DNA templateprimerhelix whereas the correct as well as incorrect nucleotidespaired to 3meC are misaligned and exhibit significantprimer terminal distortion (23) Unfortunately similarstructural or mechanistic studies in the presence of theblocking and mutagenic 3meC lesion with eukaryoticPols are currently lackingIn this work we characterized the impact of the 3meC

lesion on six human Pols belonging to the three families B(Pol d) X (Pols b and ) and Y (Pols k i and Z) Theproperties of human Pols were compared on (i) running-and standing-start templates mimicking replicative condi-tions as well as (ii) 29- 7- and 1-nucleotide (nt) gap con-structs resembling nucleotide excision (NER) and baseexcision repair (BER) intermediates respectively Byusing recombinant proteins we show that under replica-tive conditions 3meC impairs human B-family representsa block for X-family enzymes but can be successfullyutilized by Y-family Pols especially Pols i and Z Singlenucleotide incorporation experiments reveal that Pols iand Z successfully base-pair 3meC Pol i by preferentially

misincorporating dTTP and Pol Z dATP The most effi-cient extenders of base-paired 3meC are Pols k and Ztheir activity depending on the DNA structure Finallyby using human cell extracts we provide evidence thatin the context of the cell the Pol crucial for efficient3meC bypass is Pol Z as its lack results in an inabilityto successfully overcome the lesion These results suggestthe existence of a novel Pol Z-dependent pathway thatcould reduce the mutagenic effects of alkylation treatmentand potentially facilitate cellular survival especially underconditions when the functional ALKBH proteins respon-sible for direct reversal repair are missing

MATERIALS AND METHODS

Labeled g[32P] ATP was purchased from HartmannAnalytic Deoxyribonucleoside triphosphates were fromSigma All other reagents were of analytic grade andpurchased from Fluka Sigma or Merck The 40 39 3532 26 and 17mer and oligonucleotides were obtained fromMicrosynth whereas 72mer containing 3meC lesion orundamaged C were purchased were from Purimex

Preparation of DNA substrates

The sequences of the oligonucleotides used were asfollows 72mer template 50GTATTAGATATTCGGGAGGTTGGGCGCCGGCGXTGTGAATTCGGCACTGGCCGTCGTATGCTCTTGGTTGTA30 40mer primer50TACAACCAAGAGCATACGACGGCCAGTGCCGAATTCACAN30 39mer primer 50TACAACCAAGAGCATACGACGGCCAGTGCCGAATTCACA30 17merprimer 50TACAACCAAGAGCATAC30 35mer 50AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA30 32mer 50CGCCGGCGCCCAACCTCCCGAATATCTAATAC30 and 26mer 50CGCCCAACCTCCCGAATATCTAATAC30 The letter X corresponds to theposition of 3meC and C in the damaged and undamagedDNA template respectively The letter N in 40mer indi-cates four possible nucleotides at the primer 30 terminusG A T or C These four different 40mers were annealedto template strand and used in the extension assay experi-ments (Figure 5 and Supplementary Figure S6) Theprimertemplate resembling running-start conditions wasgenerated by annealing 17mer primer and 72mer template(Figure 1 and Supplementary Figure S1) whereas anneal-ing of 39mer primer and the same template resulted instanding-start situation (Figures 2 6 and SupplementaryFigure S2) The 29-nt gap construct was created by an-nealing 17mer primer and 26mer downstream oligonucleo-tide to 72mer template (Figure 3 and SupplementaryFigure S3) Annealing of 39mer primer and downstream32mer to 72mer template yielded 1-nt gap construct(Figure 4) When instead of 32mer a 26mer was used 7-nt gap was generated (Supplementary Figures S4 and S5)

The 17 39 and 40mer primers were labeled with T4polynucleotide kinase (New England Biolabs) in thepresence of [32P] ATP Each labeled oligonucleotide wasmixed with the 72mer template oligonucletide at 11(MM) ratio in the presence of 20mM TrisndashHCl(pH 74) and 150mM NaCl heated at 95C for 10 min

554 Nucleic Acids Research 2014 Vol 42 No 1

and then slowly cooled down at room temperatureSimilarly to obtain 29- 7- and 1-nt gap constructslabeled primer template and downstream oligonucleotidewere mixed in ration 111 (MMM)

Expression and purification of recombinant proteins

Human recombinant Pol d and PCNA were expressedand purified as described in (2425) Human recombinantPols and b were purified as noted previously in (26) Thebacterial clones for the expression of human Pols Z and iwere a gift from Roger Woodgate (NIH USA) Humanrecombinant Pols Z and i were expressed and purified asdescribed (27) Pol k was purchased from Enzymax USA

Cell lines and culture conditions

XP30R0 fibroblasts are simian virus 40-transformed cellsderived from a patient suffering from the variant form ofxeroderma pigmentosum (XPV) The cells harbor a 13-bpdeletion resulting in the expression of a truncated andnon-functional Pol Z XP30R0 cell line was a gift fromAlan Lehman (University of Sussex Brighton UK) Cellswere grown in a humidified 5 CO2 atmosphere inDulbeccorsquos modified Eaglersquos medium containingGlutaMAX-I supplemented with 10 fetal bovineserum and 100Uml penicillinndashstreptomycin (allobtained from Gibco Invitrogen) To overexpresshuman Pols Z and i mammalian expression vectorspJRM56 and pJRM46 [provided by Roger Woodgate(NIH USA)] were used Empty vector pcDNA 31(Invitrogen) pJRM46 or pJRM56 expression constructswere transiently transfected into XPV cells usinglipofectamine (Invitrogen) according to manufacturersprotocol Forty-eight hours after the transfection cellswere harvested and whole-cell extract (WCE) prepared

Preparation of WCE

Cells were trypsinized washed twice with cold PBS andresuspended in three packed cell volumes of hypotoniclysis buffer [10mM TrisndashHCl (pH 8) 1mM EDTA10(wv) sucrose 10 (vv) glycerol 01mM PMSF1mM DTT 1 mgml pepstatine 1 mgml bestatine and1 mgml leupeptin] After swelling for 20min at 4C cellswere lysed with a Dounce homogenizer (tight pastille)NaCl was added next to a final concentration 035Mand incubated 10min at 4C Samples were diluted withseven packed cell volumes of lysis buffer sonicated andcentrifuged 20min at 15 000 g The supernatants were flashfrozen in liquid N2 or analyzed directly

Protein analysis

WCEs were separated on 10 Bis-Tris poly-acrylamidegels followed by transfer to Immobilon-FL membrane(Millipore) for immunoblotting analysis Primaryantibodies against Pol Z (Abcam) Pol i [custom madeantibody provided by Roger Woodgate (NIH USA)]and Tubulin (Sigma) were detected using infrared Dye-conjugated secondary antibodies (LI-COR Biosciences)The signal was visualized using direct infrared fluores-cence via the Odyssey Scanner (LI-COR Biosciences)

Enzymatic assays

Recombinant human enzyme assaysFor denaturing gel analysis of DNA synthesis productsthe reaction mixture (10 ml) contained 40mM TrisndashHCl(pH 8) 10mM DTT and 250 mgml bovine serumalbumin In case of Pol d reactions 5mM Mg2+ waspresent whereas for all other Pols 1mM Mg2+ wasused Concentrations of Pols and the 50 32P-labeledDNA in the reaction were as indicated in the Figures aswell as Figure Legends The amounts of dNTPs were asfollows 10 mM for Pol d 1 mM in case of Pol and 5 mMfor all other Pols All reactions were performed at 37Cfor 15min and the reaction products analyzed by 7Murea125 polyacrylamide gel electrophoresis

Assay with WCEsFor denaturing gel analysis of DNA synthesis productsthe 10 ml of reactions containing 40mM TrisndashHCl (pH 8)10mM DTT and 250 mgml bovine serum albumin 1mMMg2+ 2 nM 50 32P-labeled DNA 10 mM unlabeled 35mertrap DNA 100 mM dNTPs and 3 ml of (133 mgml) WCEwere incubated at 37C for 30min and reaction productswere separated by 7M urea125 polyacrylamide gelelectrophoresis

Steady state kinetic analysis

Reactions were performed as described earlier in the texteach reaction containing 20 nM of DNA template and15 nM of either Pol i or Z Quantification was performedby densitometry with a PhosphorImager (Typhoon TrioGE Healthcare) The initial velocities of the reaction werecalculated from the values of integrated gel bandintensities with the programs GelEval 135 scientificimaging software (FrogDance Software UK) andGraphPad Prism 50

I T=IT 1

where T is the target site the template position of interestIT is the sum of the integrated intensities at positions TT+1 T+n All of the intensity values were normalizedto the total intensity of the corresponding lane to correctfor differences in gel loading The apparent kM and kcatvalues were calculated by plotting the initial velocities independence of the nucleotide [dNTP] or primer [30-OH]substrate concentrations and fitting the data accordingto MichaelisndashMenten equation

kcatfrac12E0=eth1+kM=frac12STHORN

where [E]0 was the input enzyme concentration and[S] was the variable substrate Substrate incorporationefficiencies were defined as the kcatkM ratio Undersingle nucleotide incorporation conditionskcat=kpolkoff(kpol+koff) and kM=kdkoff(kpol+koff)with kpol being the true polymerization rate koff the dis-sociation rate of the enzymendashprimer complex and kd thetrue Michelis constant for nucleotide binding ThuskcatkM values are equal to kpolkd Substrate concentra-tions used for dNTPs were 01ndash10 mM

Nucleic Acids Research 2014 Vol 42 No 1 555

Time-dependent accumulation of products was fitted tothe mixed exponential equation

frac12P frac14 Aeth1 ekobstTHORN+ksst

where [P] designates the products formed A the burstamplitude t time kobs the apparent burst rate and kssthe steady-state rate which corresponds to koffFree energy change (G) values were estimated

according to the relationship

G frac14 RT lnethkTHORN

where R is the gas constant (19872 calmol1K1) Tis the absolute temperature (K) and k is the ratio betweenthe kinetic constants considered

RESULTS

The Y-family DNA polymerase g successfullybypasses 3meC

3meC is predominantly generated in ssDNA which islargely present during DNA replication To address theimpact of 3meC damage occurring directly in front ofthe replication machinery bypass efficiency of sixhuman Pols was tested under running-start conditions(Figure 1A) using a primertemplate containing thelesion at position +23 from the primer For efficientlesion bypass Pols need to elongate the primer up to thelesion incorporate a nucleotide opposite the damage andextend from it The activity of the major lagging-strandreplicative Pol d was partially impaired in the presenceof 3meC damage (Figure 1B) Pol d reached the lesionand incorporated a nucleotide opposite 3meC howeverits further extension was reduced resulting in very limitedamount of full-length product (Figure 1B lanes 2ndash4) Thissuggested that upon encountering the lesion the replica-tion fork could potentially be significantly slowed-down orstalled and for its successful continuation another morespecialized Pol would be required The repair X-familyPols b and are known to bypass small base lesions(1720) but they were fully blocked by the 3meCdamage (Figure 1B lanes 6ndash12) The three differentmembers of TLS Y-family Pols k i and Z were testednext on 3meC (Figure 1C) As a template DNA in itssequence contains all four bases including undamaged Twhich Pol i favorably mispairs with lsquowooblersquo G (28) Pol iwas unable to efficiently elongate the running-start primerand reach the lesion (Figure 1C lanes 6ndash8) Pol k in con-trast very efficiently elongated the primer but exhibitedstrong pausing at the position immediately preceding thelesion It nonetheless was able to partially bypass 3meCby generating products one or two nucleotides shorterthan the full-length (Figure 1C lanes 2ndash4) This suggestedthat Pol k either loops out 3meC lesion and continues thesynthesis or generates deletions during elongation afterpairing of 3meC To distinguish between the two thelength of Pol k products was compared in the presenceof undamaged and damaged running-start templates(Supplementary Figure S1A) In both cases the productsgenerated by Pol k were one or two nucleotides shorter

than the full-length (Supplementary Figure S1B comparelanes 1ndash8 with lanes 9 and 10) thus indicating that undertested conditions Pol k most likely induces deletionsduring elongation Pol Z finally was the most efficient inbypassing 3meC under running-start conditions Pol Z infact elongated the primer and efficiently generated full-length product showing only a modest pausing at thelesion site (Figure 1C lanes 11 and 12) Taken togetherthese first results indicated that under running-start con-ditions 3meC (i) blocks human X-family Pols (ii) impairsthe activity of B-family Pol d and Y-family Pol k but (iii)can successfully be processed by TLS Pol Z

DNA polymerases i and g efficiently but inaccuratelybase-pair 3meC

While considering the bypass efficiency of a certain DNAlesion it is additionally essential to determine its impacton Pol fidelity To assess the accuracy of 3meC bypass theprimer was annealed to template strand directly in frontof the lesion resulting in standing-start conditions(Figure 2A) The amount of dNTPs in the reaction wasmodulated according to the known affinity of each PolReplicative Pol d under standing-start conditions prefer-entially misincorporated dATP followed by dTTPopposite 3meC (Figure 2B lanes 4 and 5) Additionallyin the presence of all 4 dNTPs Pol d incorporated a singlenucleotide opposite the lesion and in comparison withcontrol template exhibited dramatically reduced extensioncapacity (Figure 2B lane 2 Supplementary Figure S2)thus confirming the observation from Figure 1B Thepresence of 3meC on standing-start primertemplatestrongly inhibited the activity of Pols b and as well ask (Figure 2CndashE) Interestingly Pol k weakly inserted twodCTPs (Figure 2E lane 6) suggesting that understanding-start conditions this Pol either (i) potentiallyskips 3meC and copies the next template base (G) twiceor (ii) incorporates first dCTP opposite the lesion and thesecond one across subsequent template base In compari-son with other human Pols tested the Y-family Pols i andZ efficiently base paired the lesion (Figure 2F and G)Interestingly Pol i more efficiently incorporated dGTPopposite 3meC than across undamaged C (Figure 2Fcompare lane 1 with lane 3) While Pol i successfullymispaired 3meC with dTTP Pol Z with comparable effi-ciency incorporated the inaccurate dATP as well as thecorrect dGTP opposite 3meC (Figure 2F and GTable 1) In addition it also showed a low discriminationagainst the incorporation of dTTP and dCTP (Table 1)Similar to Pol k in the presence of dCTP incorporation oftwo nucleotides was observed (Figure 2G lane 6)Suggesting that Pol Z after copying the lesion incorporatessecond dCTP opposite the next template base or that itskips 3meC and copies subsequent template G twiceOverall the substrate affinity and conversion efficiencywere several fold higher for Pol Z than i (Table 1) PolZ further completed the bypass by generating full-lengthproducts in the presence of all 4 dNTPs (Figure 2G lane2) These findings suggested that Y-family Pols i and Zefficiently misincorporate nucleotides opposite 3meC withPol Z being the more accurate of the two

556 Nucleic Acids Research 2014 Vol 42 No 1

DNA polymerase g is the most proficient in filling 3meCcontaining DNA gaps mimicking NER and BERintermediates

In addition to DNA replication ssDNA regions fre-quently exist in the shape of gap intermediates producedduring repair processes such as NER and BERGeneration of alkylation damage while the NER orBER machineries are still processing these ssDNA gapswould require lesion bypass for efficient gap closure Todetermine the ability of Pols to close NER gap intermedi-ate containing a lesion a 29-nt gap construct with 3meC atposition +23 was generated As the presence of a phos-phate group can dramatically affect bypass efficiency ofcertain Pols (29) the 50-terminus of the gap was eitherphosphorylated (Figure 3A) or not (Supplementary

Figure S3A) As expected titration of Pol d revealedthat this Pol efficiently reached the 3meC was able toincorporate a nucleotide opposite the lesion but waspoor in subsequent extension resulting in inefficient gapfilling independently of the gap 50-end phosphorylationstatus (Figure 3B and Supplementary Figure S3B)While Pol b was once more blocked by 3meC Pol another X-family member efficiently closed the 29-ntgap when the 50-end of the gap was phosphorylated butnot when the phosphate group was lacking (Figure 3Blane 12 compare with Supplementary Figure S3B lane12) Interestingly Pol predominately generatedproducts one nucleotide shorter than the gap size(Figure 3B lane 12) indicating that it either stops justbefore the 50-end of the gap or it generates single

Figure 1 3meC bypass by six human DNA polymerases under running-start conditions (A) Schematic representation of running-startprimer template containing 3meC lesion designated as X Numbers above the template indicate the positions of template bases with respectto the lesion FL stands for full-length (B C) Titration of Pols d b (B) and Pols k i Z (C) under running-start conditions inthe presence of 2 nM DNA and indicated PCNA amounts The experiments were done under conditions as described in lsquoMaterials and Methodsrsquosection

Nucleic Acids Research 2014 Vol 42 No 1 557

nucleotide deletions potentially by looping out the lesionSimilarly to running start conditions Pol i was not able toreach the lesion (Figure 3C and Supplementary FigureS3C) Pol k showed a strong pausing site immediatelybefore the lesion but was eventually able to close the

gap containing both non- and phosphorylated 50-end(Figure 3C and Supplementary Figure S3C) as well ascontinuing up to the end of the template displacing thedownstream strand Pol Z on the other hand efficientlybypassed the lesion and completely filled the gap

Figure 2 Efficiency and fidelity of six human DNA polymerases on the 3meC template under standing-start conditions (A) Standing-start primertemplate with undamaged C or 3 meC lesion designated as X Numbers above the template indicate the position of template bases with respect to thelesion at+1 FL is full-length product (BndashG) Extension of the 2 nM standing-start primer template by 2 nM Pol d in the presence of 100 nM PCNA(B) 2 nM Pol b (C) 2 nM Pol (D) 2 nM Pol k (E) 2 nM Pol i (F) and 2 nM Pol Z (G) Reactions in lanes 1 and 3ndash6 contain dGTP dATP dTTPand dCTP respectively In lane 2 all four dNTPs (N) were added The experiments were performed under conditions as specified in lsquoMaterials andMethodsrsquo section

558 Nucleic Acids Research 2014 Vol 42 No 1

(Figure 3C and Supplementary Figure S3C) It alsoshowed robust strand displacement of bothphosphorylated and non-phosphorylated 50-gap oligo-nucleotide (Figure 3C and Supplementary Figure S3C)

Next the efficiency and accuracy of 3meC bypass wereaddressed on a shorter 7-nt gap template mimickinga long-patch BER intermediate with phosphorylated(Supplementary Figure S4A) or non-phosphorylated

Figure 3 Synthesis by six human DNA polymerases past a 3meC lesion in the context of a NER intermediate (A) Schematic representation of 29-ntgap template containing a 3meC lesion designated as X Numbers indicate the position of template bases with respect to the lesion FL stands forfull-length product generated after strand displacement (B C) Titration of Pols d b (B) and Pols k i Z (C) in the presence of 2 nM DNA andindicated PCNA amounts The experiments were done under conditions as described in lsquoMaterials and Methodsrsquo section

Table 1 Kinetic parameters for nucleotide incorporation by DNA polymerases i and Z opposite 3meC in the template

Pol i Pol Z

dNTP kM (mM) kcat (ms1) kcatkM (M1s1) Fidelitya kM (mM) kcat (ms1) kcatkM (M1s1) Fidelitya

dGTP 6 03 005 103 2 06 031 103

dATP na na na na 15 07 046 103 07dTTP 1264 09 007 103 07 184 07 004 103 89dCTP 35 02 005 103 09 24 03 014 103 22

Not accessible values designated as naaCalculated as VmaxkM(dGTP)VmaxkM(d ATCTP)

The experiments were done under conditions as specified in lsquoMaterials and Methodsrsquo section

Nucleic Acids Research 2014 Vol 42 No 1 559

(Supplementary Figure S5A) 50-gap terminus and bearingthe lesion at position+1 B-family Pol d was impaired by3meC lesion irrespectively of the 50-gap status and onlyweakly misincorporated dATP and dTTP being thenunable to continue synthesis (Supplementary FiguresS4B and S5B) Even though Pol b is suited to fill shortgaps the presence of 3meC entirely blocked its activity(Supplementary Figures S4C and S5C) Pol k was com-pletely blocked with 50-phosphorylated gap(Supplementary Figure S4E) and showed only a veryweak activity on the unphosphorylated substrate(Supplementary Figure S5E) In contrast Pol Z again ef-ficiently closed the gap incorporating mainly the incorrectdATP and correct dGTP opposite the lesion as well as toa lesser extent dCTP and dTTP (Supplementary FiguresS4G and S5G) Pols and i mispaired 3meC with dCTP(Supplementary Figure S4D lane 6) and dTTP(Supplementary Figure S4F lane 5 and SupplementaryFigure S5F lane 5) respectively In comparison tostanding-start conditions (Figure 2F) the presence of3meC did not stimulate Pol i activity While Pol i wasunable to close the gap (Supplementary Figures S4F andS5F lane 2) Pol after dCTP misincorporation reachedthe end of the gap albeit with low efficiency and onlywhen 50 gap terminus was phosphorylated (compareSupplementary Figure S4D lanes 2 and 6 toSupplementary Figure S5D lanes 2 and 6) As understanding-start conditions Pol could not bypass 3meC(Figure 2D) the present observation raised the possibilitythat in the context of short-gap 3meC is not directlymispaired with dCTP but instead that Pol uses itsability to scrunch the template (30) and loop out thelesion thus incorporating dCTP opposite template G atposition+1 However it is also possible that the presenceof the 50-phosphate which is known to increase the effi-ciency of Pol may allow direct incorporation oppositethe lesion

DNA polymerases i and g can fill a single-nucleotide gappositioned opposite 3meC

While long-patch BER is more common during replicationthe majority of BER reactions in G1G2 phases give raiseto ssDNA in shape of a 1-nt gap intermediate as a conse-quence of the combined glycosylase and AP endonucleaseaction Thus we next tested the fidelity on a template con-taining a 1-nt gap positioned opposite 3meC (Figure 4A)and bearing a 50-phosphorylated end Replicative Pol d wasas expected impaired at this template and extremely poorlymisincorporated dATP and dTTP opposite the lesion(Figure 4B) Pol b was unable to act at the site of 3meCdamage even under these classical BER conditions(Figure 4C) Surprisingly Pol that is known to efficientlyperform TLS on 1-nt gap (2631) was blocked in thepresence of 3meC This suggested that during copying ofthe larger 7- and 29-nt gaps (Figure 3 and SupplementaryFigure S4) Pol scrunched the template looped out thedamaged nucleotide elongated and consequently inducedsingle nucleotide deletions From the Y-family enzymesPol k did not act on the lesion template (Figure 4E)whereas Pol i and Z efficiently filled the gap Pol i

misincorporated dTTP (Figure 4F lane 5) and itsactivity similarly to 7-nt gap (Supplementary Figures S4and S5) was not stimulated by 3meC Thus suggesting thatpresence of the lesion stimulates Pol i activity only in rep-licative context (Figure 2F) Pol Z on 1-nt gap templateincorporated the correct dGTP as well as the incorrectdATP (Figure 4G lanes 3 and 4) Interestingly on thistemplate no incorporation of dTTP or dCTP by Pol Zwas noted contrary to what was observed understanding-start conditions and with the longer gapsFurther implying that the previously detected dCTP anddTTP incorporation by Pol Z (Figures 2GSupplementary Figure S4G and S5G) as well as dCTPby Pol k (Figure 2E) resulted from incorporationopposite alternative templating bases which cannot takeplace in a 1-nt gap When all 4 dNTPs were present inthe reaction Pol Z generated a+1 product with electrophor-etic mobility more similar to that of dATP than dGTPindicating potentially inaccurate bypass of 3meC in a 1-ntgap (Figure 4G compare lanes 2 3 and 4) Taken togetheras in the case of the previously used standing-start andgapped templates Pols i and Z were the most proficientPols in performing TLS over a 3meC lesions

DNA polymerases i and g are best extenders from 3meC

In many cases TLS is a two-step process involving incorp-oration of a nucleotide opposite a lesion by a certain Poland subsequent extension of the primer by a specializedextender Pol To test the extension properties of differentPols a primertemplate ending with either the correctG3meC or the three incorrect A3meC T3meC orC3meC base pairs was created (Figure 5A) All sixtested human Pols successfully utilized the control CGsubstrate however with different efficiencies (Figure 5Blanes 2ndash7) In presence of the lesion Pol d only weaklyextended T3meC and C3meC containing constructs(Figure 5B lanes 23 and 30) In contrast Pols k and Zutilized both correctly as well as all three incorrectlyprimed lesion templates (Figure 5B lanes 12 14 19 2126 28 33 and 35) and from all the six Pols tested were thetwo most efficient in extending all the substrates WhilePol Z generated full-length products Pol k in contrarycreated products one or two nucleotides shorter than thefull-length (Figure 5B lanes 1219 26 and 33) in agree-ment with its tendency to introduce 1 deletions (32)

To address whether the observed extension capacity wasconstruct dependent we next generated a 6-nt gap witheither correct G3meC or incorrect A3meC T3meC andC3meC base pair located at the 30-end of the gap(Supplementary Figure S6A) From these substrates Pold only poorly processed A3meC C3meC and T3meCgaps (Supplementary Figure S6B lanes 16 23 and 30)Pol on the other hand was able to extend all lesion con-structs and close 6-nt gap (Supplementary Figure S6B)Similarly Pols k and Z acted on all the templates and aswith the previous substrates were the most efficient exten-ders Interestingly under the 6-nt gap condition Pol k wasthe most active especially if gap contained T3meC orC3meC (Supplementary Figure S6B lanes 26 and 33)Pol Z as well as Pol k but only on the T3meC substrate

560 Nucleic Acids Research 2014 Vol 42 No 1

displayed a robust strand displacement activity In case of aC3meC template both Pols and k predominantlygenerated five nucleotides long product a patch one nu-cleotide shorter than the gap itself thus indicating thatboth Pols potentially re-prime the template by repositioningthe 30 C and base-pairing it with G at position+1 from thelesion In summary these findings indicated that Pols k andZ are the most efficient extenders of 3meC and that theiractivity is influenced by the structure of the DNA construct

Efficient bypass of 3meC in human WCEs requires thepresence of a functional DNA polymerase g

So far in the experiments testing recombinant humanenzymes we demonstrated that Pol i successfullyincorporated a nucleotide opposite 3meC whereas Pol Zwas more efficient in extending a mismatched primer con-taining 3meC Uniquely among all Pols tested Pol Z wasefficient in both (i) incorporation opposite the lesion aswell as (ii) subsequent extension (Figures 1ndash5) To verify

whether Pol Z was indeed playing a major role in 3meCbypass even in the presence of other Pols such as in thecontext of a whole cell we tested the fidelity and bypasscapacity of various cellular extracts under standing-startconditions (Figure 6A) WCEs were prepared from XPVpatient cells lacking endogenous functional Pol Z XPVcells were complemented with either (i) empty vector (ii)human Pol Z or (iii) Pol i expression constructs(Figure 6B) Interestingly although all three WCEs werecomparably active on a control template containing un-damaged C (Figure 6C lanes 1 7 and 13) only WCEprepared from XPV cells that had been complementedwith Pol Z showed robust incorporation opposite 3meC(Figure 6C lanes 8ndash12) As with the recombinant enzyme(Figure 2G) Pol Z complemented WCE predominantlyincorporated dATP and dGTP opposite the 3meC(Figure 6C lanes 9 and 10) In the presence of 4 dNTPs+1 product with electrophoretic mobility similar to theone of A incorporated opposite the lesion was obtained

Figure 4 Ability of six human DNA polymerases to close a single nucleotide gap generated opposite a 3meC damage (A) A 1-nt gap template withundamaged C or 3meC lesion (X) positioned directly opposite the gap (BndashG) Filling of the 2 nM gapped DNA template depicted in (A) by 2 nM Pold in the presence of 100 nM PCNA (B) 2 nM Pol b (C) 2 nM Pol (D) 2 nM Pol k (E) 2 nM Pol i (F) and 5 nM Pol Z (G) Reactions in lanes 1 and3ndash6 contain dGTP dATP dTTP and dCTP respectively In lane 2 all four dNTPs (N) were added The experiments were done under conditions asspecified in lsquoMaterials and Methodsrsquo section

Nucleic Acids Research 2014 Vol 42 No 1 561

(Figure 6C compare lanes 8 and 10) This suggested thatPol Z potentially bypasses 3meC in an inaccurate mannerand further coincided with the results of reactions usingonly the purified protein (Figure 4G) Pol Z mediatedbypass in context of WCE was partially efficient due tolow amount of extension product generated (Figure 6Clane 8) We thus hypothesized that the simultaneousincrease in the amounts of (i) Pol i as inserter and (ii)Pol Z as both inserter and extender of 3meC couldimprove the bypass efficiency Expression of both Polsin XPV cells (Figure 6B) resulted in promoted incorpor-ation of all 4 nucleotides (Figure 6C lanes 21ndash24) as wellas an increased amount of extension product (Figure 6Ccompare lanes 8 and 20) overall leading to improved3meC bypass efficiency Taken together these findings

clearly indicated that the presence of functional Pol Z inhuman cells is crucial for 3meC bypass which can bemediated by either Pol Z alone or as a two-step processinvolving an additional Pol such as Pol i

DISCUSSION

Exposure of genomic DNA to alkylating agents results ingeneration of numerous DNA lesions While the pathwaysto repair alkylating DNA lesions are well characterizedmuch less is known about the influence of alkylationdamage on different Pols Several studies described theproperties of prokaryotic Pols on the alkylation lesion3meC however it is currently unknown how thisdamage affects eukaryotic Pols In this work we

Figure 5 Extension of accurately base-paired and mispaired 3meC primertemplates by six human DNA polymerases (A) Representation ofincorrectly (Y=A T or C) as well as correctly (Y=G) paired primer termini on control (X=C) and lesion (X=3meC) DNA Numbersabove the template indicate the position of template bases with respect to the lesion at+1 (B) Extension of the 2 nM DNA template schematicallyrepresented above the lanes by 5 nM Pol d in the presence of 100 nM PCNA 4 nM Pol b 7 nM Pol 4 nM Pol k 30 nM Pol i and 20 nM Pol ZThe experiments were performed under conditions as described in lsquoMaterials and Methodsrsquo section

562 Nucleic Acids Research 2014 Vol 42 No 1

characterized the properties of 3meC bypass by the sixhuman Pols d b k i and Z members of theB- X- and Y-families Our findings strongly suggest thatamong the six tested enzymes PolZ most efficientlybypassed 3meC On open or long gapped substrates PolZ-mediated 3meC bypass was error-prone leading to theincorporation of almost all four nucleotides with selectiv-ity indexes ranging from 07 (for dATP) to 9 (for dTTP)

with respect to the correct dGTP (Table 1) However onsingle nucleotide gapped substrates incorporation was re-stricted to dATP and dGTP (Figure 4G) Besides itsability to incorporate the incorrect dATP as well as thecorrect dGTP opposite 3meC Pol Z efficiently extendedboth correctly and incorrectly paired primer termini onlesion DNA Pol Z was unique among all the tested Polsin combining the properties of both an inserter and

Figure 6 Only cells expressing DNA polymerase Z are able to successfully overcome 3meC damage (A) Schematic representation of standing-startprimer template with undamaged C or 3meC lesion designated as X Numbers above the template indicate the position of template bases with respectto the lesion at+1 FL is full-length product (B) Immunoblot analysis of XPV WCE complemented with empty vector Pol Z Pol i or Pols Z and iexpression vectors (C) Bypass capacity of 3meC lesion by XPV XPV Pol Z XPV Pol i as well as XPV Pols Z and i complemented extractsThe experiments were done under the conditions specified in lsquoMaterials and Methodsrsquo section

Nucleic Acids Research 2014 Vol 42 No 1 563

extender We further confirmed its importance in thecontext of a whole cell by showing that lack of Pol Z inhuman cell extracts results in a severely reduced ability toact at the site of 3meC (Figure 6C) The importance offunctional Pol Z in response to monoalkylating agentshas been already suggested by in vivo experimentsshowing that the Saccharomyces cerevisiae rad30D strainlacking Pol Z is somewhat sensitive to MMS (3334)Furthermore MMS treatment of human cells wasshown to result in formation of Pol Z foci at the site ofreplication forks blocked by alkylation lesions (35) one ofwhich is potentially 3meC In addition to Pol Z anotherY-family member Pol i efficiently bypassed 3meC by pref-erentially misincorporating dTTP opposite the lesion butwas unable to extend it In the context of WCEs anincrease in the level of Pol i as inserter resulted inpromoted Pol Z-mediated 3meC bypass (Figure 6C)Interestingly another Y-family enzyme Pol k was apoor inserter but proved to be the most efficient in extend-ing from a T3meC mismatch Although the generation ofG3meC pair by Pol Z would create another chance forALKBH2 to repair and thus remove lesion from thegenome it is currently not clear which consequences thepersistence of A3meC as well as T3meC created by PolZ and i respectively would have on genome stability Ifrepair of A3meC and T3meC would be possible byALKBH2 proteins it would result in harmful mismatchesand could potentially lead to mutations in the next roundof DNA replication Generation of mismatches would beespecially detrimental in tissues lacking functionalmismatch repair as was observed for various tumourtypes (3637) To prevent genomic instability in precancer-ous tissues it would thus be crucial to activate a Pol thatcan rather faithful bypass 3meC such as Pol ZUnder replicative conditions 3meC impaired the

activity of the B-family replicative Pol d and blocked the

repair Pols b and (Figure 1B) A similar finding was seenupon 3meC extension condition under which Pol d onlyvery weakly generated full-length product Thus suggest-ing that upon encountering the lesion the replication forkcould potentially be stalled and for its successful continu-ation another more specialized Pol like Pol Z would berequired These results correlate well with findingsobtained in prokaryotes demonstrating impeded activityof Pols I and III at the site of 3meC (101819)Interestingly though Pol was unable to directly incorp-orate nucleotides opposite the lesion in the context ofgapped DNA larger than 1 nt this Pol successfullyscrunched the template and continued the synthesis bypairing the incoming nucleotide with the template baseat+1 position from the lesion The property of Pol toscrunch the template was directly shown by the crystalstructure published previously by Garcia-Diaz et al (30)Also when the terminal nucleotide of the primer was com-plementary to the nucleotide immediately downstream thelesion Pol was able to loop out 3meC and use the basedownstream to re-align the primer This slippage ability ofPol has been already observed for other lesions (38) Inboth cases the result was a 1-nt deletion

Based on our findings we propose a model for 3meCbypass in context of DNA replication (Figure 7) Whenthe lesion is not repaired by ALKBHs and remains presentin the template strand Pols might handle it in one of thefollowing ways (1) after copying 3meC by incorporatinginaccurate dATP or dTTP the replicative Pol d is stalledand TLS Pol Z is recruited to extend the mispair(Figure 7-1) (2) the replication fork is stalled in frontof 3meC leading to Pol Z recruitment predominant inser-tion of dATP opposite the lesion and the extension(Figure 7-2) (3) the replication fork is stalled in front of3meC Pol i is recruited to tolerate the lesion bymisincorporating dTTP and Pol Z to extend the newly

Figure 7 Model of 3meC bypass during DNA replication If 3meC lesion (designated as X) is not repaired by ALKBHs and it remains present inthe template strand Pols might handle it in one of the following ways (1) after copying 3meC by predominantly incorporating inaccurate dATP ordTTP replicative Pol d is stalled and TLS Pol Z is recruited to extend the mispair (2) the replication fork is stalled in front of 3meC leading to Pol Zrecruitment insertion of dATP opposite the lesion and the extension (3) the replication fork is stalled in front of 3meC Pol i is recruited to toleratethe lesion by incorporating dTTP and Pol Z to extend the newly formed T3meC base-pair For simplicity only nucleotides preferentiallyincorporated by each of the Pols are depicted

564 Nucleic Acids Research 2014 Vol 42 No 1

formed T3meC base-pair (Figure 7-3) In summaryour findings suggest the potential existence of novel PolZ-dependent pathway that would enable cells to overcomecytotoxic effects of alkylating agents even at the costof potentially causing mutations especially in theabsence of functional direct reversal proteins

SUPPLEMENTARY DATA

Supplementary Data are available at NAR Online

ACKNOWLEDGEMENTS

The authors thank Ulrich Hubscher and Giovanni Magafor critical reading of the manuscript and their supportThey are very grateful to researches for kindly sharing andproviding the material Roger Woodgate for Pol i and Zexpression clones and Pol i antibody Allan Lehman forXP30R0 cells as well as to Elena Ferrari for purifyingPol d

FUNDING

University of Zurich (to AF and BvL) Funding foropen access charge University of Zurich

Conflict of interest statement None declared

REFERENCES

1 FriedbergEC WalkerGC SiedeW WoodRD SchultzRAand EllenburgerT (2006) DNA Repair and Mutagenesis ASMPress Washington DC USA

2 FuD CalvoJA and SamsonLD (2012) Balancing repair andtolerance of DNA damage caused by alkylating agents Nat RevCancer 12 104ndash120

3 SedgwickB (2004) Repairing DNA-methylation damage NatRev Mol Cell Biol 5 148ndash157

4 AllanJM and TravisLB (2005) Mechanisms of therapy-relatedcarcinogenesis Nat Rev Cancer 5 943ndash955

5 DrablosF FeyziE AasPA VaagboCB KavliBBratlieMS Pena-DiazJ OtterleiM SlupphaugG andKrokanHE (2004) Alkylation damage in DNA and RNAmdashrepair mechanisms and medical significance DNA Repair(Amst) 3 1389ndash1407

6 BodellWJ and SingerB (1979) Influence of hydrogen bondingin DNA and polynucleotides on reaction of nitrogens andoxygens toward ethylnitrosourea Biochemistry 18 2860ndash2863

7 TrewickSC HenshawTF HausingerRP LindahlT andSedgwickB (2002) Oxidative demethylation by Escherichia coliAlkB directly reverts DNA base damage Nature 419 174ndash178

8 DuncanT TrewickSC KoivistoP BatesPA LindahlT andSedgwickB (2002) Reversal of DNA alkylation damage by twohuman dioxygenases Proc Natl Acad Sci USA 9916660ndash16665

9 AasPA OtterleiM FalnesPO VagboCB SkorpenFAkbariM SundheimO BjorasM SlupphaugG SeebergEet al (2003) Human and bacterial oxidative demethylases repairalkylation damage in both RNA and DNA Nature 421859ndash863

10 DelaneyJC and EssigmannJM (2004) Mutagenesisgenotoxicity and repair of 1-methyladenine 3-alkylcytosines1-methylguanine and 3-methylthymine in alkB Escherichia coliProc Natl Acad Sci USA 101 14051ndash14056

11 SikoraA MieleckiD ChojnackaA NieminuszczyJWrzesinskiM and GrzesiukE (2010) Lethal and mutagenicproperties of MMS-generated DNA lesions in Escherichia coli

cells deficient in BER and AlkB-directed DNA repairMutagenesis 25 139ndash147

12 RingvollJ NordstrandLM VagboCB TalstadV ReiteKAasPA LauritzenKH LiabakkNB BjorkA DoughtyRWet al (2006) Repair deficient mice reveal mABH2 as the primaryoxidative demethylase for repairing 1meA and 3meC lesions inDNA EMBO J 25 2189ndash2198

13 NaySL LeeDH BatesSE and OrsquoConnorTR (2012) Alkbh2protects against lethality and mutation in primary mouseembryonic fibroblasts DNA Repair (Amst) 11 502ndash510

14 WuSS XuW LiuS ChenB WangXL WangY LiuSFand WuJQ (2011) Down-regulation of ALKBH2 increasescisplatin sensitivity in H1299 lung cancer cells Acta PharmacolSin 32 393ndash398

15 FujiiT ShimadaK AnaiS FujimotoK and KonishiN (2013)ALKBH2 a novel AlkB homologue contributes to humanbladder cancer progression by regulating MUC1 expressionCancer Sci 104 321ndash327

16 GoodmanMF (2002) Error-prone repair DNA polymerases inprokaryotes and eukaryotes Annu Rev Biochem 71 17ndash50

17 HubscherU SpadariS VillaniG and MagaG (2010) DNAPolymerases Discovery Characterization and Functions inCellular DNA Transactions World Scientific Co Pte Ltd NewJersey

18 BoiteuxS and LavalJ (1982) Mutagenesis by alkylating agentscoding properties for DNA polymerase of poly (dC) templatecontaining 3-methylcytosine Biochimie 64 637ndash641

19 SaffhillR (1984) Differences in the promutagenic nature of3-methylcytosine as revealed by DNA and RNA polymerisingenzymes Carcinogenesis 5 691ndash693

20 HubscherU and MagaG (2011) DNA replication and repairbypass machines Curr Opin Chem Biol 15 627ndash635

21 LangeSS TakataK and WoodRD (2011) DNA polymerasesand cancer Nat Rev Cancer 11 96ndash110

22 YangW and WoodgateR (2007) What a difference a decademakes insights into translesion DNA synthesis Proc Natl AcadSci USA 104 15591ndash15598

23 RechkoblitO DelaneyJC EssigmannJM and PatelDJ(2011) Implications for damage recognition during Dpo4-mediatedmutagenic bypass of m1G and m3C lesions Structure 19821ndash832

24 JonssonZO HindgesR and HubscherU (1998) Regulation ofDNA replication and repair proteins through interaction with thefront side of proliferating cell nuclear antigen EMBO J 172412ndash2425

25 PodustVN ChangLS OttR DianovGL and FanningE(2002) Reconstitution of human DNA polymerase delta usingrecombinant baculoviruses the p12 subunit potentiates DNApolymerizing activity of the four-subunit enzyme J Biol Chem277 3894ndash3901

26 van LoonB and HubscherU (2009) An 8-oxo-guanine repairpathway coordinated by MUTYH glycosylase and DNApolymerase lambda Proc Natl Acad Sci USA 10618201ndash18206

27 FrankEG McDonaldJP KarataK HustonD andWoodgateR (2012) A strategy for the expression ofrecombinant proteins traditionally hard to purify Anal Biochem429 132ndash139

28 TissierA McDonaldJP FrankEG and WoodgateR (2000)poliota a remarkably error-prone human DNA polymeraseGenes Dev 14 1642ndash1650

29 VillaniG HubscherU GironisN ParkkinenS PospiechHShevelevI di CiccoG MarkkanenE SyvaojaJE and TanguyLe GacN (2011) In vitro gap-directed translesion DNA synthesisof an abasic site involving human DNA polymerases epsilonlambda and beta J Biol Chem 286 32094ndash32104

30 Garcia-DiazM BebenekK LarreaAA HavenerJMPereraL KrahnJM PedersenLC RamsdenDA andKunkelTA (2009) Template strand scrunching during DNA gaprepair synthesis by human polymerase lambda Nat Struct MolBiol 16 967ndash972

31 MagaG CrespanE WimmerU van LoonB AmorosoAMondelloC BelgiovineC FerrariE LocatelliG VillaniGet al (2008) Replication protein A and proliferating cell nuclear

Nucleic Acids Research 2014 Vol 42 No 1 565

antigen coordinate DNA polymerase selection in 8-oxo-guaninerepair Proc Natl Acad Sci USA 105 20689ndash20694

32 MukherjeeP LahiriI and PataJD (2013) Human polymerasekappa uses a template-slippage deletion mechanism but canrealign the slipped strands to favour base substitution mutationsover deletions Nucleic Acids Res 41 5024ndash5035

33 RoushAA SuarezM FriedbergEC RadmanM andSiedeW (1998) Deletion of the Saccharomyces cerevisiae geneRAD30 encoding an Escherichia coli DinB homolog confers UVradiation sensitivity and altered mutability Mol Gen Genet 257686ndash692

34 BegleyTJ RosenbachAS IdekerT and SamsonLD (2002)Damage recovery pathways in Saccharomyces cerevisiae revealedby genomic phenotyping and interactome mapping Mol CancerRes 1 103ndash112

35 KannoucheP BroughtonBC VolkerM HanaokaFMullendersLH and LehmannAR (2001) Domain structurelocalization and function of DNA polymerase eta defective inxeroderma pigmentosum variant cells Genes Dev 15 158ndash172

36 MarraG and JiricnyJ (2005) DNA mismatch repair and coloncancer Adv Exp Med Biol 570 85ndash123

37 PoulogiannisG FraylingIM and ArendsMJ (2010) DNAmismatch repair deficiency in sporadic colorectal cancer andLynch syndrome Histopathology 56 167ndash179

38 BlancaG VillaniG ShevelevI RamadanK SpadariSHubscherU and MagaG (2004) Human DNA polymeraseslambda and beta show different efficiencies of translesionDNA synthesis past abasic sites and alternativemechanisms for frameshift generation Biochemistry 4311605ndash11615

566 Nucleic Acids Research 2014 Vol 42 No 1