58305845 Nucleic Acids Research, 2014, Vol. 42, No. 9 Published online 20 March 2014doi: 10.1093/nar/gku214

Molecular dissection of the domain architecture andcatalytic activities of human PrimPolBenjamin A. Keen, Stanislaw K. Jozwiakowski, Laura J. Bailey, Julie Bianchi and AidanJ. Doherty*

Genome Damage and Stability Centre, University of Sussex, Brighton, BN1 9RQ, UK

Received November 8, 2013; Revised February 28, 2014; Accepted February 28, 2014

ABSTRACTPrimPol is a primasepolymerase involved in nu-clear and mitochondrial DNA replication in eukary-otic cells. Although PrimPol is predicted to possessan archaeo-eukaryotic primase and a UL52-like zincfinger domain, the role of these domains has notbeen established. Here, we report that the proposedzinc finger domain of human PrimPol binds zinc ionsand is essential for maintaining primase activity. Al-though apparently dispensable for its polymerase ac-tivity, the zinc finger also regulates the processivityand fidelity of PrimPols extension activities. Whenthe zinc finger is disrupted, PrimPol becomes morepromutagenic, has an altered translesion synthesisspectrum and is capable of faithfully bypassing cy-clobutane pyrimidine dimer photolesions. PrimPolspolymerase domain binds to both single- and double-stranded DNA, whilst the zinc finger domain bindsonly to single-stranded DNA. We additionally reportthat although PrimPols primase activity is requiredto restore wild-type replication fork rates in irradi-ated PrimPol/ cells, polymerase activity is suffi-cient to maintain regular replisome progression inunperturbed cells. Together, these findings providethe first analysis of the molecular architecture ofPrimPol, describing the activities associated with,and interplay between, its functional domains anddefining the requirement for its primase and poly-merase activities during nuclear DNA replication.

INTRODUCTIONDNA replication is an essential biological process, indis-pensable for the existence of life. DNA replication sys-tems rely on a semi-conservative mode of replication where

the initiation of DNA synthesis requires a free 3 hydroxylgroup to which additional nucleotides are subsequentlyadded by replicative polymerases (1). Genome replicationstarts with DNA template-dependent synthesis of shortRNA primers that are further extended with deoxynu-cleotides by the replication machinery. This initial step inDNA replication is often defined as de novo primer synthesisand is catalysed by specialised DNA polymerases known asprimases. Based on their structural topology, these enzymescan be classified into archaeo-eukaryotic primases (AEPs)or DnaG-like prokaryotic primases (2,3).Until recently, the Pol -associated DNA primase small

subunit (PriS) that is responsible for de novo polymer syn-thesis through the production of RNA primers was consid-ered to be the sole AEP superfamily member present in eu-karyotes (1). However, bioinformatic analysis identified theexistence of an additional uncharacterized DNA primase ineukaryotes called PrimPol (CCDC111 or FLJ33167) (37),which belongs to the NCLDV-herpesvirus clade of viralAEPs.Recent studies have reported that PrimPol is a DNA

primase (47), with the ability to synthesise primers us-ing either ribonucleotides (NTPs) or deoxyribonucleotides(dNTPs), preferring tomakeDNAprimers. In addition, theenzyme possesses robust template-dependent DNA poly-merase activity (47). PrimPol is present in both the nu-cleus (46) and mitochondria (7) of eukaryotic cells. Theenzyme localises to nuclear chromatin during replication(46) and this recruitment is more pronounced after treat-ment with damaging agents (e.g. ultraviolet light; UV) orreplication stalling drugs (e.g. hydroxyurea) (4). UV irra-diation can induce the covalent linkage of adjacent pyrim-idines leading to the formation of cyclobutane pyrimidinedimers (CPDs) and pyrimidine (64) pyrimidone photo-products ((64)PPs). These helix-distorting lesions interferewith major biological processes, including DNA replicationand transcription (8). PrimPol can perform translesion syn-thesis (TLS) bypass of the highly distorting (64)PPs but

*To whom correspondence should be addressed. Tel: +44 1723 877500; Email: ajd21@sussex.ac.ukContributed equally to this work.Present address:Julie Bianchi, Department of Oncology-Pathology, Cancer Center Karolinska (CCK), Karolinska Institutet, SE-171 76, Stockholm, Sweden.The authors wish it to be known that, in their opinion, the first two authors should be regarded as Joint First Authors.

C The Author(s) 2014. Published by Oxford University Press on behalf of Nucleic Acids Research.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/3.0/), whichpermits unrestricted reuse, distribution, and reproduction in any medium, provided the original work is properly cited.

Nucleic Acids Research, 2014, Vol. 42, No. 9 5831

is also involved in replication through oxidative lesions, in-cluding 8-oxoguanine (8-oxoG) (4,5,7). Deletion of PrimPol(PrimPol/) induced replication fork slowing, which wasmuch more pronounced when cells were UV irradiated (4).Knockout cells also exhibited increased formation of chro-mosomal breaks, particularly after aphidicolin treatment.This study also reported that PrimPol is not epistatic withthe Pol -dependant CPD-bypass pathway and thereforeappears to form an independent pathway required for thebypass of UV, and other lesions, during replication (4,5).Comparative analysis of the amino acid sequence of eu-

karyotic PrimPols identified the presence of two distinctivedomains (3), an enzymatic AEP polymerase and a UL52-like zinc finger (Zfn) domain (Figure 1A). Mutation of theconserved zinc-chelating cysteine residues in the UL52 do-main of herpes simplex virus type 1 resulted in a severe re-duction in DNA binding (8,9). The AEP polymerase do-main contains the three signature catalytic motifs. A highlyconserved DxE motif (motif I), together with an asparticacid residue from motif III, forms the divalent metal bind-ing site.Motif II (SxH) forms part of the putative nucleotidebinding motif. Mutagenesis of these motifs abolishes thecatalytic activities of these polymerases (4,5,7,10,11).In this study, we have dissected the molecular architec-

ture of human PrimPol to define the activities associatedwith its two major functional domains. We demonstratethat the zinc finger domain is crucial for its primase activ-ity. PrimPol also has DNA template-dependent DNA poly-merase activities and this bifunctional enzymatic activityis reminiscent of archaeal replicative primases (1114). Al-though PrimPols polymerase activities appeared, initially,to be largely independent of the zinc finger domain, ourdata suggests that the zinc finger domain plays importantroles in the processivity and fidelity of DNA synthesis. Ad-ditionally, we demonstrate that the zinc finger domain hasregulatory roles for both DNA/RNA primase and TLSactivities. We also report that a catalytically active frag-ment of human PrimPol, containing only the AEP domain(PrimPol1354), catalyzes TLS bypass of both major UV-induced DNA lesions, CPDs and (64)PPs. Analysis of theDNA binding affinities of catalytically active PrimPol1354and the C-terminal UL52-like domains established that theenzymatic domain can bind both single-stranded (ss) anddouble-stranded (ds) DNA, whilst the zinc finger domaincan only bind to ss DNA. Finally, we report that althoughPrimPols polymerase activity is sufficient to maintain wild-type replication fork rates in unperturbed PrimPol/ cells,primase activity is requisite for normal replication in UV-treated cells. Together, these findings provide the firstmolec-ular insights into the domain architecture of human Prim-Pol, defining the activities associated with its major func-tional modules and delineating the requirement for its spe-cific catalytic activities during perturbed and unperturbedDNA replication.

MATERIALS AND METHODSConstruction of human PrimPol mutants

Human PrimPol was cloned as described previously (4).A number of PrimPol mutants were constructed by poly-merase chain reaction (Figure 1B) and the primers used can

Figure 1. Domain architecture of eukaryotic PrimPol. (A) PrimPol is com-posed of an AEP primasepolymerase and a Zn2+ finger (Zfn) domain.The AEP domain contains three conserved catalytic motifs (IIII). Thefirst motif (motif I, the first two red stars) is the DxE motif that, alongwith the conserved D in the third motif (motif III, third red star), formsthe catalytic triad that coordinates divalent metal ions essential in the syn-thesis of oligonucleotide chains. The second motif (motif II, two light-orange stars) is a conserved SxH motif that is required for the coordina-tion of the incoming nucleotide. The zinc finger domain contains a canon-ical CHCC motif that coordinates a zinc ion and stabilises the anti-parallel -sheet and helix structure. The numbers below the conservedresidues denote their positions in human PrimPol. (B) A number of hu-man PrimPol constructs were produced to analyse the domain architec-ture of PrimPol. A Zfn knockout mutant was produced by mutating thefirst conserved cysteine and histidine residues that coordinate the zinc ion(C419A and H426A, respectively). We constructed a number of trunca-tion mutants based on secondary structure predictions. PrimPol1487 lacksthe C-terminus region downstream of the Zfn domain. PrimPol1354 dele-tion mutant contains the polymerase domain but lacks the Zfn domain.PrimPol372560 possesses the zinc finger domain but lacks the polymerasedomain. PrimPol1354/ZF-KO contains only the zinc finger domain but theC419A andH426A aremutated. A 1334 deletionmutant inXenopus trop-icalis (XPrimPol1334) was additionally constructed that is equivalent toPrimPol1354.

be found in the supplementary data (Supplementary TableS1). Xenopus tropicalis PrimPol cDNA was sub-cloned intopET28a (Novagen) using BamHI andXhoI restriction sites.Subsequently, a 1334 C-terminal deletion (XPrimPol1334)mutant was also constructed (primers in Supplementary Ta-ble S1). Additionally, we constructed a PrimPolAXA (cat-

5832 Nucleic Acids Research, 2014, Vol. 42, No. 9

alytically null) mutant that was cloned and purified, as pre-viously described (4).

Expression and purification of recombinant PrimPol proteins

Wild-type PrimPol and PrimPol1487 were expressed inEscherichia coli BL21(pLysS) cells overnight at 16C, asdescribed previously (4). PrimPol1354, PrimPol372560 andXPrimPol1334 were expressed overnight at 25C. All of theproteins were then purified by the same protocol, exceptwhere otherwise noted. Cells were pelleted and resuspendedin buffer A (50 mM TrisHCl (pH 7.5), 200 mM NaCl,30 mM imidazole, 10% (v/v) glycerol, 17 g/ml PMSF,34 g/ml benzamidine) supplemented with 0.5% IGEPAL.Cells were disrupted by sonication and proteins isolated bycentrifugation. All proteins were purified by affinity chro-matography using a 25 ml Ni2+-NTA agarose (Qiagen) col-umn equilibrated with buffer A and eluted with buffer B (asA, with 300 mM imidazole). The eluted protein-containingfractions were then diluted 1 in 10 in buffer C (50 mMTrisHCl (pH 7.5), 10% (v/v) glycerol) and separated bycharge by affinity exchange chromatography on a 5 ml Hi-Trap Heparin HP column (GE Healthcare) equilibratedwith buffer C. PrimPol372560 was purified on a 5 ml HiTrapQ HP (GE Healthcare) column equilibrated with buffer C.All proteins were subject to gradient elution with buffer D(as C, with 2MNaCl) and protein-containing fractions werepurified by size exclusion chromatography on a SuperdexS-75 analytical gel-filtration column (GE Healthcare) equi-librated in buffer E (50 mM TrisHCl (pH 7.5), 300 mMNaCl, 10% (v/v) glycerol). Protein concentrations were de-termined by absorption spectra at 280 nm.

DNA primase assays

The sequences of the oligonucleotides used in the primaseassays are sequences 14 in Supplementary Table S2. Thenon-radioactive primase assay was performed in three sub-sequent steps. Typically detection of primase activity wasstarted from incubation of 1M of the enzyme to be testedin 20 l reaction volume containing 500 nM homopoly-meric ss DNA templates with a biotin modification at the 5end (see sequences 14 in Supplementary Table S2), 500MrNTPs (Invitrogen) or 500MdNTPs (Roche), 10mMBis-Tris-Propane-HCl (pH 7.0), 10 mM MgCl2, 50 mM NaCl.This primer synthesis reaction was carried out for 2 h at37C. Following the primer synthesis reaction, the reactionmixture was supplemented with 0.2 U of Klenow Taq (pu-rified as in Engelke et al. (15)) and 15 M FAM-6-dATP(Jena-Biosciences), incubated at 37C for 45 min to allowfluorescent labelling of de novo synthesised primers. Theprimer synthesis/labelling enzymatic reactions were termi-nated by adding 450l of binding-washing (B-W) buffer (10mMTris-HCl (pH 8.0), 500mMNaCl, 10mMEDTA). Thequenched reactions were subsequently added to 30 l ofstreptavidin-coated beads (Invitrogen) andmixed on a spin-ning wheel for 1 h at 4C. After binding the ss DNA tem-plates, the suspensions were spun down briefly to sedimentthe beads. The supernatant was removed and the beads werewashed three times with 1 ml B-W buffer. The beads werethen suspended in 20 l of the B-W buffer supplemented

with equal volume of loading buffer (8 M Urea, 10 mMEDTA). The resulting sample was incubated at 95C for3 min in order to liberate the primers synthesised de novo.Subsequently, the samples were spun down briefly and 20l volume of each reaction was resolved by standard de-naturing electrophoresis using a 15% (v/v) polyacrylamidegel matrix containing 7Murea and 1TBE buffer (100mMTris, 100mM Boric Acid, 2mM EDTA). Typically the elec-trophoretic separation of primase assay products was per-formed at 850 V for 2 h. After electrophoresis gels werescanned for fluorescent signal detected by a Fujifilm FLA-5100 image reader.

DNA primer extension assays

Hex-labelled DNA primers were annealed to oligomer tem-plates (sequences in Supplementary Table S2). 34 nM ofprotein was incubated with 20 nM DNA, 10 mM Bis-Tris-Propane-HCl (pH 7.0), 10 mMMgCl2, 1 mMDTT and 200M dNTPs (Roche) to a final volume of 20 l. In the caseof single nucleotide incorporations 200 Mof each individ-ual nucleotide adenosine, cytosine, guanosine or thymidinetriphosphate (dATP, dCTP, dGTP or dTTP respectively)were added in lieu of dNTPs. The reactions were terminatedby the addition of 2 stop buffer (95% formamide, 0.09%xylene cyanol, 0.05% bromophenol blue, 200 nM competi-tor oligonucleotide) and boiled at 95C for 5 min. Sampleswere resolved by electrophoresis on a 15% (v/v) polyacry-lamide gel containing 7 M urea and 1 TBE buffer at 850V for 2.5 h in 1 TBE buffer. Fluorescently labelled DNAoligomers were detected by scanning using a Fujifilm FLA-5100 image reader.

Terminal transferase assays

The terminal transferase capability of wild-type PrimPoland its variants was studied using three types of syntheticDNA substrates: dsDNA (sequence 6 annealed to sequence7 from Supplementary Table S2) and a primer template con-taining both single-stranded and double-stranded DNA in-terfaces (sequence 5 annealed to sequence 6; SupplementaryTable S2). Typical reactions were performed in 20 l vol-ume containing 10 mM Bis-Tris-Propane-HCl (pH 7.0), 10mM NaCl, 10 mM MgCl2, 1 mM MnCl2, 1 mM DTT, 20nMDNA substrate, 200 M dNTPs (Roche), with 100 nMrecombinant human PrimPol or its variants. All reactionswere incubated at 37C for 30min, reactions were quenchedwith addition of 2 stop buffer (95% formamide, 0.09% xy-lene cyanol, 0.05% bromophenol blue, 200 nM competitoroligonucleotide) and boiled at 95C for 5min. Samples wereresolved by electrophoresis as described for the primer ex-tension assays.

DNA extension processivity assays

Full-length and PrimPol1354 constructs were analysed forprocessivity as described previously (16). PrimPol (100 nM)was preincubated at 37C with 60 nM substrate DNA (se-quence 5 annealed to sequence 6 from Supplementary Ta-ble S2), 10 mM Bis-Tris-Propane-HCl (pH 7.0), 10 mMMgCl2 and 1 mM DTT for 30 min. Reactions were ini-tiated by adding dNTPs and excess of sonicated herring

Nucleic Acids Research, 2014, Vol. 42, No. 9 5833

sperm DNA (1 mg/ml) as an enzyme trap. Reactions werequenched after time points of 15, 30, 60, 120 and 360swith addition of 2 stop buffer (95% formamide, 0.09%xylene cyanol, 0.05% bromophenolblue, 200 nM competi-tor oligonucleotide) and boiled at 95C for 5 min. Sampleswere resolved by electrophoresis as described for the primerextension assays. To test the effectiveness of the trap, each ofthe enzymes were also preincubated with excess of herringsperm DNA (1 mg/ml), as well as 40 nM DNA, 10 mMBis-Tris-Propane-HCl (pH 7.0), 10 mM MgCl2 and 1 mMDTT. To ensure the trap did not disrupt the processivity ofthe enzyme, the experiment was also carried out with theKlenow fragment of Taq polymerase.The band intensities at 360 s were measured using Image-

Quant software (GE Healthcare). The percentage of activepolymerases at a given position is given by Equation (1):

% active polymerases at n

= (In + In+1 + In+2...) 100%/(I1 + I2 + I3 . . .) (1)Where I1 is the intensity at position 1, In is the intensity atposition n and so on.

Electrophoretic mobility shift assays

Electrophoretic mobility shift assays (EMSAs) were car-ried out on two synthetic substrates: ss DNA (sequence7 from Supplementary Table S2) and ds DNA (sequence6 annealed to sequence 7 from Supplementary Table S2).Varying concentrations of the human PrimPol1354 andPrimPol372560 constructs were added to 40 nM DNA, 10mMBis-Tris-Propane-HCl (pH 7.0), 10mMMgCl2 and 1.0mMDTT to a final volume of 20 l and incubated at 25Cfor 60 min. The reactions were supplemented with 2 l 25%(w/v) ficoll. Samples were resolved by electrophoresis on a5% (v/v) polyacrylamide gel containing 0.5 TBE buffer at150V for 0.5 h, then 300V for 2.5 h in 0.5TBEbuffer. Flu-orescently labelled DNA oligomers were detected by scan-ning using a Fujifilm FLA-5100 image reader.

Inductively coupled plasma mass spectrometry

The concentrations of zinc ions in protein samples were de-termined using inductively coupled plasma mass spectrom-etry (ICP-MS). For this analysis of humanPrimPol, the pro-teins were gel filtered in zinc-free solutions to minimise theconcentration of zinc ions that were not chelated by protein.Quantitative calibrations were made using standards pre-pared from zinc solutions containing 0, 100, 500 and 1000ng/ml. Each sample was measured in triplicate and back-ground measurements of buffer without protein were deter-mined and subtracted from the appropriate readings. A 2%nitric acidwashwas performed between standards and sam-ples.

PrimPol protein denaturation assays

The full-length and PrimPol1487 constructs were both dial-ysed into a buffer of 200 mM NaF, 15 mM Tris-HCl (pH7.5). 45 l of 1 M protein solution was added to 15 l ofSYPRO Orange, resulting in a final protein concentration

of 0.75 M. A control of 45 l buffer was also added to 15l of SYPRO Orange. 20 l of each sample was aliquotedin triplicate into a 96-well plate. Protein melting experi-ments were carried out using the LightCycler 480 SystemII (Roche). The instrument was configured with a detectionformat of 465 nm as the wavelength of excitation and 580nm as the emission wavelength to detect SYPRO Orange-specific signal. Denaturation curve fluorescent signal wasacquired within a range of 2080C using a ramping rateof 0.03C s1 and an acquisition of 20 data points per de-gree celsius. Melting temperatures (Tm) were determinedthrough themeasurement of the lowest point of the negativedifferential of the denaturation curve. Data was correctedfor the background signal of the buffer conditions and pre-sented as units of fluorescence with respect to temperature 1 SD.

Complementation and survival assays in PrimPol/ aviancells

DT40 cells were grown at 39C in RPMI 1640 mediumsupplemented with 10 M -mercaptoethanol, penicillin,streptomycin, 10% foetal calf serum and 1% chicken serum(Sigma). DT40 cells deleted of PrimPol made previously (4)were stably complemented with wild-type and truncated ormutated forms of human PrimPol cloned into the pCI-neovector by electroporation as described previously (4). Stableexpressing cells were selected using 2 mg/ml G418 (Sigma)and confirmed by western blot. DNA fibre analysis was car-ried out as described previously (4). Fibre analysis was car-ried out in triplicate (n = 3).

RESULTSZinc finger domain of PrimPol coordinates a zinc ion

The C-terminal domain of human PrimPol contains ahighly conserved UL52-like Cys-His-Cys-Cys (CHC2) mo-tif, predicted to fold into a functional zinc finger (Zfn)motif.(Figure 1A) To establish if this is a bona fide zinc-bindingmotif, we produced C-terminal truncations in PrimPolcontaining either an intact (PrimPol372560) or mutated(PrimPol372560/ZF-KO) Zfn motif. PrimPol372560/ZF-KO hasthe first two residues of the Zfn motif mutated to alanine.The proteins were gel filtered in the absence of zinc and con-centrations of zinc were thenmeasured using ICP-MS (Sup-plementary Table S3). The ICP-MS analysis revealed thatthe concentration of the zinc in the UL52-like domain rep-resented an occupancy of 80.38%, suggesting that a smallfraction of the protein is either unfolded or has another di-valentmetal ion chelated. In contrast, the zinc finger knock-out had an occupancy of 3.90%, suggesting that the first twocysteine and histidine residues of this motif are required forthe coordination of zinc. We also measured the zinc occu-pancy in the polymerase domain (PrimPol1354) after pu-rification in the absence of zinc and the occupancy in thiscase was 1.57%. The detection of low levels of zinc in thepolymerase domain is to be expected as it contains divalentcation-binding motifs. These motifs generally bind magne-sium or manganese, which have respective ionic radii of 86and 81 pm, but could conceivably bind zinc, with a compa-rable ionic radius of 88 pm (17).

5834 Nucleic Acids Research, 2014, Vol. 42, No. 9

Figure 2. Primase activity of human PrimPol. (A) Human PrimPol has pri-mase activity and can produce de novo primers using rNTPs and dNTPsopposite a poly(dT) template. (B) PrimPolZF-KO lacks de novo primer syn-thesis activity, suggesting that an intact zinc finger is required for primaseactivity. (C) PrimPol1487 also has primase activity similar to the wild-typePrimPol. The unstructured region that is downstream of the zinc fingeris therefore not required for primase activity. (D) PrimPol1354 has no pri-mase activity, which indicates that PrimPol requires a functional zinc fingerfor primer synthesis.

PrimPols zinc finger domain is essential for primase activity

Previously, we and others have shown that PrimPolhas DNA primase activity in vitro (47), preferentiallycatalysing synthesis of primers composed of DNA. Previ-ous studies on the viral HSV1 helicase/primase complexshowed that mutation of key residues in the UL52 zinc-binding domain of this enzyme resulted in loss of primaseactivity (8,9). Alignment of human PrimPol with othereukaryotic orthologues identified the conserved residues(C419, H426, C446 and C451; Figure 1A) predicted tocoordinate a zinc ion. We prepared a zinc finger knock-out enzyme carrying C419A and H426A point mutations(PrimPolZF-KO) (Figure 1B).As previously observed (4,5), wild-type PrimPol exhib-

ited primase activity on a 60-mer poly-dT ss DNA tem-plate, however, PrimPolZF-KO showed no detectable pri-mase activity (Figure 2). Furthermore, we analysed the pri-mase activity of two deletion mutants, PrimPol1487 andPrimPol1354, which were constructed to determine whetherelimination of primase activity resulted from a structuralchange brought about by the point mutations or whetherthe zinc-finger domain is necessary for primase activity.As expected, a deletion mutant lacking the entire zinc fin-ger (PrimPol1354) showed no detectable primase activity(Figure 2). However, PrimPol1487, containing the entire

Figure 3. Polymerase activity and fidelity of human PrimPol. (A) HumanPrimPol was incubatedwith dNTPs and substrate at 1, 3, 5 and 30min timepoints. PrimPol was proficient at extending an undamaged oligonucleotidetemplate using dNTPs. Human PrimPol did not require an intact zinc fin-ger in order to carry out primer extension, as evidenced by the extensionof primers by PrimPolZF-KO and PrimPol1354. PrimPol1487 that lackedthe unstructured C-terminus of the protein was also polymerase proficient.PrimPol1354 exhibited a higher rate of polymerase activity compared tothe other constructs. (B) Incorporation of nucleotides opposite two tem-plating cytosine bases. PrimPol was incubated for 5 min with the DNAsubstrate and each of the dNTPs. All four of these PrimPol constructs in-serted two guanine nucleotides opposite two cytosines in WatsonCrickbase-pairing manner. PrimPol1354 could additionally incorporate a singleadenine opposite the first cytosine.

zinc-finger domain but lacking the smaller C-terminal por-tion, predicted to be disordered, was an active DNA pri-mase (Figure 2). Notably, the observed primase activity ofPrimPol1487 was lower than the activity observed for theparental enzyme, possibly as this truncation results in par-tial loss of UL52 domain functionality. Together, our dataprovide experimental evidence that the zinc-finger domain,but not the disordered region proximal to this domain, isessential for the primase activity of human PrimPol.

PrimPols DNA polymerase activity is modulated by the zincfinger domain

As the zinc-finger domain is required for the priming ac-tivity of human PrimPol, we also evaluated the impor-tance of this functional module in the context of Prim-Pols DNA polymerase activity. To address this, we em-ployed standard primer extension assays using unmodifiedsynthetic primer templates and dNTPs. We compared therelative DNA polymerase activities of four different vari-ants of human PrimPol including the wild-type enzyme,PrimPolZF-KO, PrimPol1487 and PrimPol1354 (Figure 1B;Table 1A).Although all tested variants of PrimPolwere pro-ficient at DNA template-dependent DNA synthesis, we ob-served variation in the relative specific polymerase activity,processivity and fidelity among these enzymes (Figure 3A).PrimPolZF-KO showed a pronounced decrease in specific

Nucleic Acids Research, 2014, Vol. 42, No. 9 5835

DNA polymerase activity. In contrast, PrimPol1354 exhib-ited both increased relativeDNAprocessivity and extensionactivities. Despite variations in specific DNA polymeraseactivities across the variants of human PrimPol tested, thezinc-finger module does not appear to be essential for DNApolymerase activity of this enzyme. However, a regulatoryrole in catalysis for this functional module cannot be ex-cluded.Consequently, we next examined the potential regulatory

role of this functional module. In particular, we tested thefidelity of incorporation of each of the four single deoxynu-cleotides opposite two templating cytosines (Figure 3B).Wild-type PrimPol, PrimPolZF-KO and PrimPol1487 all in-corporated two correct incoming guanine nucleotides op-posite the templating cytosines at the N+1 and N+2 po-sitions. In contrast, PrimPol1354 that lacks the entire zincfinger domain showed a significantly reduced ability to ac-curately select the incoming nucleotide. In particular, we ob-served significant mis-incorporation of adenine nucleotidesopposite templating cytosine at the N+1 position and mis-incorporation of another guanine nucleotide opposite atemplating adenine at N+3. This observation suggests thepotential importance of the Zfn domain in influencing thefidelity of PrimPols polymerase activity. PrimPol cannot in-corporate NTPs during primer extension in vitro and muta-tion of the Zfn domain did not alter this preference (unpub-lished data). This is in contrast to the AEP domain of LigD,which can extend primers using both dNTPs and NTPs(18,19).

PrimPol exhibits template-independent primer extension ac-tivity in the presence of manganese

The flexible extension activities reported previously forNonhomologous end-joining (NHEJ) AEP polymerases includelimited template-independent terminal transferase activityon both ss DNA and blunt-ended ds DNA (18,19). Suchactivities are often stimulated by the presence of manganeseions, which can accelerate the rate of DNA synthesis and, asa result, lower the overall fidelity of this process. We there-fore tested if human PrimPol also possesses terminal trans-ferase activity in the presence of manganese. Wild type, aswell as all three variants, of human PrimPol showed no de-tectable terminal transferase activity when tested on a bluntended ds DNA substrate (Figure 4, left panels). However,PrimPols DNA synthesis activity appeared to be signifi-cantly increased when primer extension assays were per-formed in the presence of dNTPs (all) and manganese (Fig-ure 4A, right panel). To investigate if this increased activityresulted from altered fidelity, we repeated the assays witheach of the dNTPs and observed that PrimPol catalysedmultiple extension of the primer even in the presence ofa single nucleotide pool. This promiscuous DNA synthe-sis was most efficient when enzyme was utilising dATP ordTTP. In both cases, extension of primers by up to 20nucleotides was observed. Notably, this phenomenon wasnot observed when magnesium was used instead of man-ganese. When a single incorporation of dCTP was investi-gated, we observed significantly slower extension rates butthe enzyme was able to extend the primer to produce pre-dicted full-length product. In contrast, incorporation of the

Figure 4. Template-independent extension in the presence of manganese.Human PrimPol was incubated for 30 min with DNA substrate and eachof the dNTPs in the presence of manganese. (A) Wild-type PrimPol wasunable to extend from a ds DNA template with a blunt end, but couldextend from a primer annealed to an overhanging template, even synthe-sising long tracts of homopolymers. (B) PrimPolZF-KO could extend froman overhanging DNA template in the presence of manganese and, con-sistent with the wild-type, did not extend from a ds DNA substrate. Theincorporation of 1 or 2 nucleotides of guanine or adenine opposite an over-hanging template suggests that PrimPolZF-KO incorporates in a low-fidelitytemplate-dependent manner when incubated with overhanging DNA. (C)PrimPol1487 exhibited a highly similar terminal transferase activity spec-trum to the PrimPolZF-KO. In the presence of manganese, it incorporatedbases opposite an overhang in a low fidelity, template-dependent manner.(D) PrimPol1354 also exhibited low fidelity extension of a primer annealedto an overhanging template in the presence of manganese.

correct incoming nucleotide (dGTP) resulted in least pro-nounced primer extensions up to 16 nucleotides.When this experimental strategy was repeated to study

potential promiscuous DNA polymerase activities in thePrimPolZF-KO and PrimPol1487 variants, we observedminimal mis-incorporation of all four tested single nu-cleotides. Interestingly, when a catalytically active frag-ment (PrimPol1354) was tested, we observed more effi-cient mis-incorporation compared to PrimPolZF-KO andPrimPol1487 variants. However, PrimPol1354 fragment was

5836 Nucleic Acids Research, 2014, Vol. 42, No. 9

Table 1. Extension, TLS and fidelity activities associated with full-length and mutated forms of human PrimPol

Summary of the DNA synthesis activities of wild-type and mutant PrimPol. (A) PrimPol exhibits primase activity, polymerase activity and, in the presence ofmanganese, template-independent primer extension. Whilst the polymerase activity and template-independent primer extension activity appear to residein the AEP polymerase domain alone, the primase activity requires the associated zinc finger. (B) PrimPol can read through a number of common DNAlesions and we have determined some of the lesions PrimPol can read through and which bases are incorporated opposite these lesions. As the UV lesionsspan two adjacent bases, a is used to separate the bases incorporated opposite the first and second damaged bases.

able to mis-incorporate between 1 and 2 incorrect incomingnucleotides. We are unable to conclude if this unique enzy-matic activity of human PrimPol is significant for the bio-logical functions of this enzyme in vivo. We speculate that,under the experimental conditions tested, the Zfn moduleallowed manganese-dependent primer extension resultingin the synthesis of DNA that is non-complementary to thetemplate strand. This hypothetical scenario is only possiblewhen the enzyme was able to bind to a single-stranded tem-plate strand of the DNA primer-template substrate. Thisinteresting observation has raised questions regarding themechanism by which PrimPol binds to and coordinates theprimer-template substrate during DNA polymerisation.

DNA binding activities of human PrimPol

Next, we analysed the binding of the AEP polymerase andZfn domains of PrimPol to a number of DNA substrates todetermine whether they bind specifically to ss or ds DNA.The Zfn domain is presumed to be a DNA-recognition do-main, in addition to contributing to its primase activity, asprevious studies demonstrated altered DNA binding activ-ity associated with mutations in a related UL52 Zfn do-main (8,20). EMSAs were performed to analyse the bind-ing capabilities of the AEP polymerase (PrimPol1354) andZfn (PrimPol372560) domains. PrimPol1354 showed a simi-lar binding capacity to ds and ss DNA, suggesting that thepolymerase can bind both substrates (Figure 5A). The ob-served DNA binding specificity reflects the likely presenceof two closely clustered binding sites allowing recognitionof a primer-template junction, which is an essential require-ment for all DNApolymerases. Observed binding values for

both ss DNA and ds DNA were evident at a minimum of500 nM DNA and a complete shift of the DNA was ev-ident at 5 M DNA. This relatively weak coordination ofDNA substrates is consistent with previously postulated bi-ological functions of human PrimPol (47).Analogous EMSA experiments were also performed

to analyse the DNA binding capacity of the Zfn do-main (PrimPol372560). In contrast with the AEP do-main, PrimPols Zfn domain was unable to bind ds DNAbut did show specific binding to ss DNA (Figure 5B).This intriguing result provides experimental evidence sup-porting the potential importance of the Zfn module inrecognition/coordination of the single-stranded templateDNAdownstreamof the primer-template junction. The sig-nificance of this interaction will be discussed below. To de-termine if DNA binding was performed by the Zfn, we as-sayed a Zfn knockout domain (PrimPol372560/ZF-KO) for ssDNA binding activity (Figure 5C) and observed an absenceof binding with this mutant, confirming the importance ofthe Zfn motif for nucleic acid recognition.

PrimPols AEP domain possesses TLS activities

PrimPol has the capacity to bypass a number of replicationstalling lesions, including UV ((64)PP) (4) and oxidativelesions (4,6). To address whether the TLS activities of Prim-Pol are independent of the zinc finger domain, we assayedthe TLS activities of PrimPol1354 on a range of templatescontaining a variety of DNA modifications (Table 1B). Asfull-length PrimPol appears unable to read throughCPDs invitro, we assayed PrimPol1354 for TLS bypass of this com-mon photolesion. Notably, although PrimPol1354 slowed

Nucleic Acids Research, 2014, Vol. 42, No. 9 5837

Figure 5. PrimPols polymerase domain binds ss and ds DNA but the zinc finger domain binds only ss DNA. (A) PrimPol1354 was tested for its abilityto bind DNA by shift assays. PrimPol polymerase domain was incubated with 40 nM DNA at various protein concentrations (0.05, 0.1, 0.5, 1.0, 5.0, 10.0and 20.0 M) for 60 min at 25C. The polymerase domain binds to ds DNA and ss DNA with approximately equal proficiency. Binding was evident atconcentrations 0.5 M but a complete shift was observed at 5.0 M for each DNA substrate tested. (B) PrimPol372560 (Zfn domain) was also assayedfor binding to DNA activity. It was incubated with 40 nMDNA at various protein concentrations (0.05, 0.1, 0.5, 1.0, 5.0, 10.0 and 20.0 M). A DNA shiftwas not observed with ds DNA substrate but a clear shift was evident with ss DNA, suggesting that the zinc finger binds single-stranded regions of DNA.(C) PrimPol372560/ZF-KO, lacking a functional Zfn, was incubated with 40 nM DNA at the same protein concentrations and ss DNA binding activity wasnot observed.

5838 Nucleic Acids Research, 2014, Vol. 42, No. 9

as it reached a CPD lesion, it showed the capacity to readthrough this bulky lesion (Figure 6A). In contrast, wild-typePrimPol cannot performTLS opposite this lesion (4,5). Sin-gle nucleotide incorporation assays were employed to de-termine which bases are incorporated opposite CPDs, re-vealing that PrimPol performs error-free bypass of this pho-tolesion (Figure 6A). A similar TLS activity was also con-firmed with Xenopus PrimPol1334, which was also capableof reading through CPDs in an error-free manner (Supple-mentary Figure S1A). This TLS activity is similar to Pol ,a specialised TLS polymerase known to bypass CPDs in anerror-free manner (21,22). As this read through of a CPDis only evident in the absence of the Zfn domain, it may bethe case that PrimPol requires a conformational change orpartner to assist in promoting this lesion bypass activity invivo.(64)PPs are a much more highly distorting UV lesion

than CPDs. Pol can incorporate bases opposite (64)PPsbut incorporates an incorrect dG opposite the 3 dT ofthe lesion and is unable to subsequently extend from thisproduct (23). However, we previously reported that Prim-Pol exhibits the ability to incorporate opposite and extendthrough (64)PPs (4). Although PrimPols TLS activity wasretained when the Zfn was deleted, the nucleotide incor-poration signature was altered (Figure 6B), again consis-tent with the proposed regulatory role of the Zfn domain.Wild-type PrimPol incorrectly incorporates thymine oppo-site the first dT of the (64)PP, whereas PrimPol1354 in-corporates dA or dC opposite this base. If dA nucleotideis incorporated opposite the first thymine of the (64)PP,PrimPol1354 then incorporates either dC or dT. If the firstnucleotide added is dC, it will then incorporate dA, dC ordT. To confirmwhether this altered fidelity of nucleotide in-corporation was orthologue-specific, we assayed the equiv-alent Xenopus PrimPol domain (XPrimPol1334) and ob-served showed a similar fidelity to the human PrimPol poly-merase domain (Supplementary Figure S1B).Next, we examined TLS bypass of an oxidative lesion,

8-oxoG. In common with wild-type enzyme, PrimPol1354can read through 8-oxoG lesions with minimal stalling andexhibited the ability to insert A and C opposite 8-oxoGlesions (Figure 7A), as previously observed for wild-typePrimPol (4). A similar TLS activity and fidelity for this le-sion was also evident with XPrimPol1334 (SupplementaryFigure S2A). Cytosine in DNA is susceptible to sponta-neous deamination to deoxyuracil (dU) in vivo, which canbase pair to adenine, resulting in CT transitionmutations(10,11,24). Deoxyuracil is not recognised by, nor does itstall, eukaryotic TLS polymerases (10,13,14,25). Similarly,the polymerase domain of PrimPol does not stall when con-fronted with a dU base (Figure 7B). PrimPol incorporatesan adenine opposite dU therefore, as one might expect, itreads the dU base as a thymine. This is in common withXPrimPol1334 (Supplementary Figure S2B) and wild-typePrimPol (Supplementary Figure S3).Finally, we examined TLS bypass of

apurinic/apyrimidinic (AP sites) and thymine glycol(TG) lesions. PrimPol1354 was unable to extend througheither AP (Supplementary Figure S4A) or TG (Supple-mentary Figure S4B) lesions but there was some singlenucleotide incorporated opposite the TG lesion. This

insertion activity could potentially promote extension byanother polymerase that is capable of extending followingnucleotide insertion opposite the TG. XPrimPol1334 wasalso unable to read through either of these lesions (Sup-plementary Figure S4). The inability of PrimPol1354 toread through AP sites and TG lesions is also evident inwild-type PrimPol (4). This inability to extend through APand TG sites, as well as other lesions, does not preclude thepossibility that PrimPol reprimes following these lesions,allowing replicative polymerases to resume synthesis in apost-lesion fashion, see below.

Human PrimPol is a distributive DNA polymerase

We next analysed the number of nucleotides incorporatedby human PrimPol in a single association-dissociation eventto measure the processivity of this polymerase. In orderto determine the processivity of PrimPol, we performedprimer extension assays in the presence of an excess ofDNA(herring sperm), which acted as a substrate trap for poly-merases that dissociate from the primer-template substrate.Using this approach, we identified that PrimPol inserts upto 4 nucleotides opposite an undamaged template, rarelyinserting more than this (Figure 8A). When the Zfn do-main was removed from PrimPol, the maximum numberof nucleotides incorporated remained at 4 but the averagenumber of nucleotides incorporated was higher, with fewerpolymerases inserting 13 nucleotides (Figure 8B). We alsocarried out processivity assays in the presence of Taq poly-merase to ensure that it was not the trap DNA that keptthe polymerase from inserting more than 4 nucleotides. Weobserved that Taq polymerase possessed high processivityand incorporated nucleotides to the end of the templateDNA (Supplementary Figure S5). Therefore, the Zfn ap-pears to play a role in regulating the number of nucleotidesthat PrimPol inserts, possibly by stabilizing the interactionof the polymerase domain with the template, see the discus-sion.

PrimPols zinc finger domain is required for bypass of lesionsin vivo

We previously reported that avian PrimPol/ cells showedsignificantly decreased rates of replication fork progression(4). To evaluate the requirement for the Zfn domain of hu-man PrimPol in vivo, we produced cell lines, in a DT40PrimPol/ knockout background (4), stably expressinghuman PrimPol, PrimPolZF-KO, PrimPolAXA (catalyticallynull) or PrimPol1354 (Figure 9A). Using DNA fibre anal-ysis, we confirmed that normal replication fork rates wasrestored by adding back humanPrimPol, however this repli-cation fork defect was not corrected by adding back thecatalytically dead PrimPolAXA mutant (Figure 9B). Next,we studied the replication fork speed in PrimPol1354 andPrimPolZF-KO complemented clones and observed that bothproteins were able to restore replication fork rates to nearwild-type levels (Figure 9B). This suggests that the poly-merase activity of PrimPol is largely responsible for themaintenance of fork progression in unperturbed replica-tion. Indeed, PrimPol1354 complemented cells showed evengreater average replication fork rates compared to wild-type

Nucleic Acids Research, 2014, Vol. 42, No. 9 5839

Figure 6. PrimPol1354 can replicate through CPD and (64(PP)) lesions. (A) PrimPol1354 was incubated with a primer-template substrate in which thetemplate contained a CPD lesion downstream of the primer-template junction in the presence of dNTPs. Time points were taken between 0.5 and 60 min.PrimPol1354 extends from the primer up to the CPD, before stalling, it will then added a base opposite the first thymine of the CPD and continue to extenduntil the end of the template (left panel). PrimPol1354 was then incubated with each dNTP to test which nucleotide it incorporates opposite a CPD (rightpanels). PrimPol1354 incorporated adenine opposite the first and second thymine of the CPD. (B) The polymerase domain of PrimPol can also performTLS bypass of a 64(PP) lesion immediately downstream of the primer-template junction (left panel). PrimPol1354 incorporates either an adenine orcytosine nucleotide opposite the first thymine of the 64(PP) (right panels). If an adenine is incorporated opposite the first thymine, a cytosine or thymineis then incorporated opposite the second. If a cytosine is incorporated opposite the first thymine, an adenine, cytosine or thymine will be incorporatedopposite the second thymine of the 64(PP).

cells (Figure 9B), fitting with the superior polymerase activ-ity observed in vitro (Figure 3A).To examine the ability of the different PrimPol variants

to support bypass of bulky lesions during replication invivo, a large pulse of UV irradiation (20 J/m2) was intro-duced to cells between the two labelling steps to investigatefork stalling induced at UV lesions. PrimPol knockout cellsshow a pronounced stalling phenotype, indicated by a largeincrease in the ratio of the CldU to IdU ratios (4). Con-

sistent with these previous findings, we observed comple-mentation by adding back the wild-type enzyme. However,PrimPolAXA, PrimPol1354 or PrimPolZF-KO were unable tocorrect this fork-stalling defect (Figure 9C). Therefore, al-though the polymerase extension activity of PrimPol is suffi-cient formaintenance of unperturbed replication, these datasuggest that its primase activity is requisite for timely repli-cation fork progression following the introduction of signif-icant DNA damage.

5840 Nucleic Acids Research, 2014, Vol. 42, No. 9

Figure 7. PrimPol1354 can replicate through an 8-oxoguanine (8oxoG) lesion and a deoxyuracil (dU) base. (A) The polymerase domain exhibited minimalstalling opposite an 8oxoG lesion and extended fully to the end of the template. PrimPol incorporated an adenine or cytosine opposite the 8oxoG, whichare the expected bases to be incorporated by Hoogsteen base pairing or WatsonCrick base pairing, respectively. (B) PrimPol1354 efficiently synthesisedthrough a templating dU site. PrimPol incorporates an adenine nucleotide opposite the dU lesion through WatsonCrick base pairing.

DISCUSSIONRecent studies have identified that PrimPol is a novel eu-karyotic DNA polymerase capable of performing bothpriming and primer extension synthesis (47). It is likelythat PrimPol was acquired early in eukaryotic evolution toplay a significant role in lesion bypass during DNA replica-tion. PrimPol-mediated damage tolerance appears to pro-vide additional mechanisms that facilitate replication forkprogression following stalling of replisomes at sites of DNAdamage or secondary structure.In this study, we have dissected themolecular activities as-

sociated with the major structural elements of human Prim-Pol (Table 1A) and determined the cellular requirementsfor these activities in vivo. PrimPol is both an active pri-

mase and polymerase with the capacity to perform TLS(Table 1B). An intact zinc finger is essential for PrimPolsprimase activity in vitro (Figure 2), consistent with reportsfor a similar requirement for many diverse DNA primases(8,9,26,27). It has been proposed that this structural ele-ment is required for template recognition and regulationof primer length synthesis (28). Although the DNA poly-merase activity of PrimPol appeared initially to be inde-pendent of the zinc finger domain, as a catalytically ac-tive domain lacking the zinc finger domain (PrimPol1354)showed enhanced relative polymerase and processivity ac-tivities compared to the wild-type enzyme. Further studiesrevealed that the zinc finger modulates the extension activ-ities and fidelity of PrimPol and therefore it should also

Nucleic Acids Research, 2014, Vol. 42, No. 9 5841

Figure 8. Processivity of the polymerase activity of PrimPol. PrimPol was pre-incubated for 30 min at 37C with an undamaged DNA primer-templatesubstrate to allow PrimPol to bind to the DNA. The reaction was initiated by the addition of dNTPs and an excess of sonicated herring sperm DNA(trap) and time points taken at 15, 30, 60, 120 and 360 s. (A) After 360 s, wild-type PrimPol incorporated up to 4 nucleotides opposite the template but asignificant fraction of enzyme incorporated only 1, 2 or 3 nucleotides (left panel). To confirm that the trap prevents polymerase extending from a secondtemplate, the trap was also added into the pre-incubation mix with PrimPol and the DNA substrate. This reaction was supplemented with dNTPs and thereis no extension (right panel), thus successfully exhibiting the effectiveness of the trap. (B) PrimPol1354 also predominantly incorporates up to 4 nucleotidesbut there were fewer polymerases incorporating only 1, 2 or 3 nucleotides (left panel). The effectiveness of the trap was also successfully confirmed (rightpanel). (C) Percentage of PrimPol molecules incorporating at least n dNTPs for either full-length PrimPol or PrimPol1354, calculated using Equation (1).

be considered to be an important regulator of PrimPolspolymerase functions. PrimPol is a very distributive en-zyme, catalysing up to four nucleotide incorporations dur-ing a single DNA substrate binding/polymerisation cyclein vitro. This distributive character of PrimPol is reminis-cent of the DNA polymerase processivity reported for theY-family TLS polymerases, such as Pol and , which pre-dominantly incorporate up to 6 or 7 nucleotides (16,29).Although removal of the last 80 amino acids at the C-

terminal region of human PrimPol (PrimPol1487) did notabolish primase activity, which suggests that the C-terminusis not an absolute requirement for activity, it did reduce ac-tivity to some extent. Thermal denaturation studies haveshown that there is a considerable increase in the stability

of PrimPol upon removal of amino acids 488560 (Supple-mentary Figure S6), suggesting that this region lacks dis-cernible structure that would contribute to protein stabil-ity. Disordered regions of proteins often serve as scaffoldsinvolved in proteinprotein interactions and we speculatethat this portion of human PrimPol might also act as aninteraction interface. Indeed, it has been reported that thecarboxyl-terminus of PrimPol mediates an interaction withthe large subunit of human replication protein A (6).In the presence of manganese, PrimPol appeared to be

much more active than with magnesium. However, uponcloser inspection, we discovered that the enzyme can pro-duce polymeric primer extension products, even when onlya single form of dNTP is present. Under these manganese-

5842 Nucleic Acids Research, 2014, Vol. 42, No. 9

Figure 9. PrimPols zinc finger domain is required for UV lesion bypass in vivo. (A) PrimPol/ DT40 cells were complemented with variants of PrimPoland their in vivo expression confirmed by western blot. (B) Unperturbed replication fork rates were analysed by DNA fibre analysis using CldU and IdUlabels for all the complemented cell types. The data represent the mean of three or more experiments with error bars showing the mean standard deviation.(C) DNA fork rate analysis was also carried out with after a 20 J/m2 pulse of UV, between CldU and IdU labels, to look for fork-stalling. Data is shownas the ratio of the two labels, Cldu, pre-UV : IdU, post UV label (n = 3). The black lines indicate the Cldu:IdU ratio of the two labels in the wild-type cellsand the red lines indicate the ratios in cells complemented with the indicated PrimPol variants.

Nucleic Acids Research, 2014, Vol. 42, No. 9 5843

dependent conditions, PrimPol extends from a primer toform homopolymeric strands that are non-complementaryand therefore unlikely to be annealed to the template strand.This suggests that caution should be taken when inter-preting the results from studies using manganese as theresulting products may represent aberrant synthesis prod-ucts rather than the expected template-dependent ones.Manganese has the effect of decreasing fidelity amongstpolymerases (30,31). Whilst magnesium enforces tetrahe-dral geometry in the arrangement of its ligands, man-ganese will accommodate square, planar, tetrahedral andoctahedral coordinations thus increasing the ability to ac-celerate the rate of reaction in substrates that are per-haps misaligned (32). The physiological implications of thistemplate-independent primer extension activity of PrimPolin the presence of manganese remain unclear, as the cellularconcentrations of manganese are low, suggesting this activ-ity may be irrelevant in vivo or it may be limited to certaincellular compartments (e.g. mitochondria).The translesion bypass activities associated with wild-

type PrimPol and PrimPol1354 are compared in Table 1B.As PrimPol can incorporate bases opposite and extend fromhighly distorting lesions, it is likely that the active site in-volved in production of phosphodiester bond is large andflexible to accommodate a range of lesions. The (64) pho-toproducts, in particular, are highly mutagenic due to thehigh degree of distortion they introduce into the DNAbackbone. The helical bending at these lesions is estimatedto be 44 and the distance between phosphorus atoms inthe paired bases in the (64)PP is6 A, compared to the 7 Aaverage corresponding distance at other base-pairing steps(33). Therefore, it is probably unsurprising that PrimPol ex-hibits low fidelity when inserting bases opposite a (64)PP,as a polymerase that can accommodate this lesion wouldrequire a flexible active site. Unexpectedly, the polymerasedomain of PrimPol, like Pol, is capable of bypassing CPDswith high fidelity. This contrasts with full-length PrimPol,which is not capable of bypassingCPDs unless two adeninesare inserted opposite the CPD (4,5). This result establishesthat the active site of PrimPol has the capacity to bypassthese lesions but this is constrained in the full-length en-zyme. A conformation change may be required, potentiallyinduced by binding partners, such as Replication Protein A(RPA), that alters the activity of PrimPol to allow it to bindto and bypass CPDs in vivo.PrimPol is also confronted with a number of lesions as a

result of oxidative stress in cells. These include AP sites and8-oxoG lesions. Pol , the mitochondrial replicative poly-merase, has the ability to bypass an 8-oxoG site but at a re-duced rate, and cannot bypass AP sites (34,35). Conversely,PrimPol shows no hindrance to incorporate bases opposite8-oxoG lesions and, although it cannot incorporate basesopposite an AP site, it has the capacity to re-prime post le-sion. 8-oxoG is expected to pair with C through WatsonCrick base pairing in its regular anti conformation, but in itssyn conformation it forms Hoogsteen base pairs with ade-nine (36). This ability to insert A, as well as C, opposite 8-oxoG is apparent in a number of polymerases but the ad-dition of auxiliary proteins (e.g. RPA and PCNA) can dra-matically increase the preference for adenine over cytosine,

as is the case with Pol and (36,37). It remains to be seenwhether this is also the case with PrimPol. This ability toaccommodate bulky lesions comes at the expense of fidelityand, therefore, PrimPol is a low fidelity polymerase that islikely to make limited, and perhaps, nonspecific contactswith the replicating base pair. Deoxyuracil is not recognisedas a lesion by replicative polymerases and results in C:G T:A transition mutations. Deoxyuracil is recognised by ar-chaeal family B polymerases as a lesion and stalls replica-tion to preventmutation (38) but PrimPol does not have thisrecognition capacity and incorporates an adenine oppositea deoxyuracil lesion.Canonical DNA polymerases share a common right-

handed architecture containing fingers, palm and thumbsubdomains. All of the three structural elements are clus-tered together providing accurate, highly structurally re-strained positioning of the polymerase at the primer-template junction in order to promote the highest fidelityand processivity of DNA synthesis. Although this tight andspecific binding cleft is perfectly suited for replication oflong stretches of DNA, in some instances such structuralarrangements can be disadvantageous. In particular, it lim-its the capacity of the replicative enzymes to tolerate ab-normalities in the topology of DNA templates, such as sec-ondary structures orDNAdamage-induced distortions.Thethumb subdomain of replicative polymerases interacts withthe sugar-phosphate backbone of the DNA to coordinatethe DNA. This subdomain is optimised for the highly ac-curate replication of DNA and is sensitive for structuralanomalies in the in the DNA template. The predicted struc-ture of PrimPol, in addition to the known structures of pri-mases, suggests that this polymerase lacks structural ele-ments corresponding to the thumb domain that is presentin the replicative polymerases that is implicated in proces-sivity (39). The open planar predicted structure of this poly-merase indicates a transient binding of the polymerase do-main and this is consistent with other low fidelity poly-merases, such as those of the Y-family of polymerases (40).The processivity data presented here suggests that the zincfinger domain of PrimPol regulates the fidelity of these poly-merases. Whilst there is no reduction in the activity of thepolymerase domain lacking the zinc finger, it actually ap-pears more active, we propose that the presence of the zincfinger reduces the processivity of the enzyme to allow aslower, higher fidelity incorporation of complementary nu-cleotides. We postulate that whilst the AEP polymerase do-main provides a platform to recognise the primer-templatejunction of the DNA substrate, the zinc finger binds thesingle-stranded region of the template strand to modulatebinding and fidelity by the AEP domain (Figure 10A). Y-family polymerases have a highly truncated thumb domainand, notably, contain an additional functional module re-ferred to as the polymerase associated domain (PAD). Inter-estingly, studies on TLS polymerases Dpo4 and Dbh pointto a role for PAD in regulation of processivity and fidelity ofthese enzymes (41). This structural element not only bindsss DNA and provides an extended flexible platform coordi-nating the template strand during DNA synthesis on dam-aged DNA templates, but also modulates the activity ofthese TLS DNA polymerases. As part of an adaptation toperform TLS, we suggest that the zinc finger is functionally

5844 Nucleic Acids Research, 2014, Vol. 42, No. 9

Figure 10. Models of the mechanisms of action of PrimPol. (A) Prim-Pol has both DNA primase and polymerase activities. The zinc fingerof PrimPol stabilises the binding of the polymerase domain on single-stranded DNA. The AEP polymerase domain coordinates two adjacentdeoxynucleotides opposite the single-stranded DNA template and the sta-bility afforded to this complex by the Zfn binding allows synthesis of adinucleotide. This dinucleotide can then be extended in a more processiveway to form longer products. PrimPol can also bind primer-template junc-tions to catalyse more canonical primer extension reactions. Both the poly-merase and primase activities are intrinsic to the AEP domain. Althoughthe zinc finger is essential for maintaining less stable primer synthesis, itis also important for modulating the fidelity and processivity of the poly-merase activities. This stabilizing influence of the Zfn may be particularlyimportant for allowing productive synthesis during TLS bypass of distort-ing lesions and structures. We postulate that the primase and polymeraseactivities of PrimPol are not discrete activities but rather represent differentsynthesis modes performed under different DNA binding conditions. (B)PrimPol uses these activities to bypassDNAdamage that blocks replicativepolymerases. When replicative polymerases encounter a blocking lesion,they are displaced or regress from the site of damage. PrimPol can thenaccess the template upstream of the lesion, where it can reprime followingthe lesion (left) or if it can access the primer-template junction it can carryout TLS (right). Once PrimPol-mediated bypass of the lesion has occurred,the replicative polymerase can then resume replication at the fork.

analogous to the PAD subdomain, but is spatially separatedfrom the polymerase domain to allow greater freedom ofbinding of the enzyme to distorted DNA templates, such asthose containing UV-induced lesions.The activity most closely associated with the primase

superfamily, namely primer synthesis, is considered to besomehow different from primer extension activities associ-ated with canonical polymerases. However, primases shouldbe considered to be polymerases that have evolved to catal-yse a poorly processive extension reaction involving the syn-thesis of a dinucleotide. To do this, they have acquired the

capacity to bind a single nucleotide as the primer, whichprovides the 3OH to initiate attack on a second template-bound nucleotide to form the first phosphodiester linkage.As Zfn binds single-stranded DNA and its absence negatesprimer synthesis, we propose that the Zfn acts in concertwith the polymerase domain to stabilise the binding andconfiguration of the template, polymerase and nucleotideslong enough to allow the production of the initial dinu-cleotide primer (Figure 10A). This product is subsequentlyextended in a more processive polymerisation mode. Inreplicative primases, the primase large subunit probablyprovides a similar role (12). In addition to primer synthe-sis, the zinc finger domain of PrimPol also plays additionalroles in regulating the fidelity of these enzymes. By loweringthe processivity of PrimPol, it appears to increase its fidelity,thus reducing its potential to be pro-mutagenic.Finally, regarding the bypass of lesions during genome

replication in vivo, this and previous work (4) suggest thatboth of PrimPols priming and extension activities are requi-site for lesion bypass and replication restart. When replica-tive DNA polymerases encounter a blocking lesion, theystall and PrimPol is recruited, possibly by RPA, to performTLS or re-prime replication downstream of the lesion (Fig-ure 10B). After production of a primer or polymerisationthrough the lesion, the replicative polymerase can then re-sume replication. PrimPols extension activities appear to bePCNA-independent as no stimulation was observable in thepresence of this protein (unpublished data). Although it hasbeen reported that RPA can bind to and potentially recruitPrimPol toDNA, themechanism bywhich it interfaces withstalled replisomes to restart fork progression remains to beestablished.

SUPPLEMENTARY DATASupplementary Data are available at NAR Online.

ACKNOWLEDGEMENTSThe authors would like to kindly thank Chris Dadswell andSarah Parry-Morris for their technical assistance.Wewouldalso like to thank Prof. Shigenori Iwai who kindly providedDNA containing a 64 photoproduct.FUNDINGBiotechnology and Biological Sciences Research Council(BBSRC) [BB/H019723/1]; MRC [G080130]; BBSRC PhDstudentship (BB/F01645X/1 toB.A.K.). Source of open ac-cess funding: RCUK.Conflict of interest statement. None declared.

REFERENCES1. Frick,D.N. and Richardson,C.C. (2001) DNA primases. Annu. Rev.Biochem., 70, 3980.

2. Aravind,L., Leipe,D.D. and Koonin,E.V. (1998) Toprim-a conservedcatalytic domain in type IA and II topoisomerases, DnaG-typeprimases, OLD family nucleases and RecR proteins. Nucleic AcidsRes., 26, 42054213.

3. Iyer,L.M., Koonin,E.V., Leipe,D.D. and Aravind,L. (2005) Originand evolution of the archaeo-eukaryotic primase superfamily andrelated palm-domain proteins: structural insights and new members.Nucleic Acids Res., 33, 38753896.

Nucleic Acids Research, 2014, Vol. 42, No. 9 5845

4. Bianchi,J., Rudd,S.G., Jozwiakowski,S.K., Bailey,L.J., Soura,V.,Taylor,E., Stevanovic,I., Green,A.J., Stacker,T.H., Lindsay,H.D.et al. (2013) Eukaryotic PrimPol bypasses UV photoproducts duringchromosomal DNA replication.Mol. Cell, 52, 566573.

5. Rudd,S.G., Glover,L., Jozwiakowski,S.K., Horn,D. and Doherty,A.J.(2013) PPL2 translesion polymerase is essential for the completion ofchromosomal DNA replication in the African Trypanosome.Mol.Cell, 52, 554565.

6. Wan,L., Lou,J., Xia,Y., Su,B., Liu,T., Cui,J., Sun,Y., Lou,H. andHuang,J. (2013) hPrimpol1/CCDC111 is a human DNAprimase-polymerase required for the maintenance of genomeintegrity. EMBO Rep., 14, 11041112.

7. Garcia-Gomez,S., Reyes,A., Martnez-Jimenez,M.I., Chocron,E.S.,Mouron,S., Terrados,G., Powell,C., Salido,E., Mendez,J., Holt,I.J.et al. (2013) PrimPol, an Archaic Primase/Polymerase Operating inHuman Cells.Mol. Cell, 52, 541553.

8. Biswas,N. and Weller,S.K. (1999) A mutation in the C-terminalputative Zn2+ finger motif of UL52 severely affects the biochemicalactivities of the HSV-1 helicase-primase subcomplex. J. Biol. Chem.,274, 80688076.

9. Chen,Y., Carrington-Lawrence,S.D., Bai,P. and Weller,S.K. (2005)Mutations in the putative zinc-binding motif of UL52 demonstrate acomplex interdependence between the UL5 and UL52 subunits of thehuman herpes simplex virus type 1 helicase/primase complex. J.Virol., 79, 90889096.

10. Lao-Sirieix,S.-H. and Bell,S.D. (2004) The heterodimeric primase ofthe hyperthermophilic archaeon Sulfolobus solfataricus possessesDNA and RNA primase, polymerase and 3-terminal nucleotidyltransferase activities. J. Mol. Biol., 344, 12511263.

11. Lipps,G., Weinzierl,A.O., Scheven,von, G., Buchen,C. and Cramer,P.(2004) Structure of a bifunctional DNA primase-polymerase. Nat.Struct. Mol. Biol., 11, 157162.

12. Lao-Sirieix,S.-H., Nookala,R.K., Roversi,P., Bell,S.D. andPellegrini,L. (2005) Structure of the heterodimeric core primase. Nat.Struct. Mol. Biol., 12, 11371144.

13. Bocquier,A.A., Liu,L., Cann,I.K.O., Komori,K., Kohda,D. andIshino,Y. (2001) Archaeal primase: bridging the gap between RNAand DNA polymerases. Curr. Biol., 11, 452456.

14. Liu,L., Komori,K., Sonoko,I., Bocquier,A.A., Cann,I.K.O.,Kohda,D. and Ishino,Y. (2001) The Archaeal DNA Primase.Biochemical Characterization of the P41-P46 Complex FromPyroccocus Furiosis. J. Biol. Chem., 276, 4548445490.

15. Engelke,D.R., Krikos,A., Bruck,M.E. and Ginsburg,D. (1990)Purification of Thermus aquaticus DNA polymerase expressed inEscherichia coli. Anal. Biochem., 191, 396400.

16. Washington,M.T., Johnson,R.E., Prakash,S. and Prakash,L. (1999)Fidelity and processivity of Saccharomyces cerevisiae DNApolymerase eta. J. Biol. Chem., 274, 3683536838.

17. Shannon,R.D. (1976) Revised effective ionic radii and systematicstudies of interatomic distances in halides and chalcogenides. ActaCrystallogr. A, 32, 751767.

18. Della,M., Palmbos,P.L., Tseng,H.-M., Tonkin,L.M., Daley,J.M.,Topper,L.M., Pitcher,R.S., Tomkinson,A.E., Wilson,T.E. andDoherty,A.J. (2004) Mycobacterial Ku and ligase proteins constitutea two-component NHEJ repair machine. Science, 306, 683685.

19. Pitcher,R.S., Tonkin,L.M., Green,A.J. and Doherty,A.J. (2005)Domain structure of a NHEJ DNA repair ligase fromMycobacterium tuberculosis. J. Mol. Biol., 351, 531544.

20. Chen,Y., Livingston,C.M., Carrington-Lawrence,S.D., Bai,P. andWeller,S.K. (2007) A mutation in the human herpes simplex virustype 1 UL52 zinc finger motif results in defective primase activity butcan recruit viral polymerase and support viral replication efficiently.J. Virol., 81, 87428751.

21. Masutani,C., Kusumoto,R., Yamada,A., Dohmae,N., Yokoi,M.,Yuasa,M., Araki,M., Iwai,S., Takio,K. and Hanaoka,F. (1999) TheXPV (xeroderma pigmentosum variant) gene encodes human DNApolymerase . Nature, 399, 700704.

22. Biertumpfel,C., Zhao,Y., Kondo,Y., Ramon-Maiques,S.,Gregory,M., Lee,J.Y., Masutani,C., Lehmann,A.R., Hanaoka,F. andYang,W. (2010) Structure and mechanism of human DNApolymerase . Nature, 465, 10441048.

23. Johnson,R.E., Haracska,L., Prakash,S. and Prakash,L. (2001) Roleof DNA polymerase in the bypass of a (64) TT photoproduct.Mol. Cell. Biol., 21, 35583563.

24. Duncan,B.K. and Miller,J.H. (1980) Mutagenic deamination ofcytosine residues in DNA. Nature, 287, 560561.

25. Connolly,B.A., Fogg,M.J., Shuttleworth,G. and Wilson,B.T. (2003)Uracil recognition by archaeal family B DNA polymerases. Biochem.Soc. Trans., 31, 699702.

26. Mendelman,L.V., Beauchamp,B.B. and Richardson,C.C. (1994)Requirement for a zinc motif for template recognition by thebacteriophage T7 primase. EMBO J., 13, 3909.

27. Griep,M.A. and Lokey,E.R. (1996) The role of zinc and the reactivityof cysteines in Escherichia coli primase. Biochemistry, 35, 82608267.

28. Corn,J.E., Pease,P.J., Hura,G.L. and Berger,J.M. (2005) Crosstalkbetween primase subunits can act to regulate primer synthesis intrans.Mol. Cell, 20, 391401.

29. Ohashi,E., Bebenek,K., Matsuda,T., Feaver,W.J., Gerlach,V.L.,Friedberg,E.C., Ohmori,H. and Kunkel,T.A. (2000) Fidelity andprocessivity of DNA synthesis by DNA polymerase , the product ofthe human DINB1 gene. J. Biol. Chem., 275, 3967839684.

30. Tabor,S. and Richardson,C.C. (1989) Effect of manganese ions on theincorporation of dideoxynucleotides by bacteriophage T7 DNApolymerase and Escherichia coli DNA polymerase I. Proc. Natl.Acad. Sci. U.S.A., 86, 40764080.

31. Hays,H. and Berdis,A.J. (2002) Manganese substantially alters thedynamics of translesion DNA synthesis. Biochemistry, 41, 47714778.

32. Pelletier,H., Sawaya,M.R., Wolfle,W., Wilson,S.H. and Kraut,J.(1996) A structural basis for metal ion mutagenicity and nucleotideselectivity in human DNA polymerase beta. Biochemistry, 35,1276212777.

33. Kim,J.-K. and Choi,B.-S. (1995) The solution structure of DNAduplex-decamer containing the (6-4) photoproduct of thymidylyl(3 5) thymidine by NMR and relaxation matrix refinement. Eur. J.Biochem., 228, 849854.

34. Pinz,K.G., Shibutani,S. and Bogenhagen,D.F. (1995) Action ofmitochondrial DNA polymerase gamma at sites of base loss oroxidative damage. J. Biol. Chem., 270, 92029206.

35. Graziewicz,M.A., Bienstock,R.J. and Copeland,W.C. (2007) TheDNA polymerase gamma Y955C disease variant associated withPEO and parkinsonism mediates the incorporation and translesionsynthesis opposite 7,8-dihydro-8-oxo-2-deoxyguanosine.Hum. Mol.Genet., 16, 27292739.

36. van Loon,B., Markkanen,E. and Hubscher,U. (2010) Oxygen as afriend and enemy: How to combat the mutational potential of8-oxo-guanine. DNA Repair, 9, 604616.

37. Maga,G., Villani,G., Crespan,E., Wimmer,U., Ferrari,E., Bertocci,B.and Hubscher,U. (2007) 8-oxo-guanine bypass by human DNApolymerases in the presence of auxiliary proteins. Nature, 447,606608.

38. Fogg,M.J., Pearl,L.H. and Connolly,B.A. (2002) Structural basis foruracil recognition by archaeal family B DNA polymerases. Nat.Struct. Biol., 9, 922927.

39. Brautigam,C.A. and Steitz,T.A. (1998) Structural and functionalinsights provided by crystal structures of DNA polymerases and theirsubstrate complexes. Curr. Opin. Struct. Biol., 8, 5463.

40. Lehmann,A.R., Niimi,A., Ogi,T., Brown,S., Sabbioneda,S.,Wing,J.F., Kannouche,P.L. and Green,C.M. (2007) Translesionsynthesis: Y-family polymerases and the polymerase switch. DNARepair, 6, 891899.

41. Boudsocq,F., Kokoska,R.J., Plosky,B.S., Vaisman,A., Ling,H.,Kunkel,T.A., Yang,W. and Woodgate,R. (2004) Investigating the roleof the little finger domain of Y-family DNA polymerases in lowfidelity synthesis and translesion replication. J. Biol. Chem., 279,3293232940.