Báo cáo khoa học: DmGEN shows a flap endonuclease activity, cleaving the blocked-flap structure and model replication fork docx

In a preliminary report, we described the presence of a new nuclease, Drosophila melanogaster XPG-like endo-nuclease DmGEN which belongs to the RAD2⁄ XPG nuclease family, shows unique ac

the blocked-flap structure and model replication fork

Yoshihiro Kanai, Gen Ishikawa, Ryo Takeuchi, Tatsushi Ruike, Ryo-ichi Nakamura, Ayumi Ihara, Tetsuyuki Ohashi, Kei-ichi Takata, Seisuke Kimura and Kengo Sakaguchi

Department of Applied Biological Science, Tokyo University of Science, Chiba, Japan

DNA replication, recombination and repair are key

processes in maintaining genome integrity Nucleases

are necessary for their nucleolytic activities They act on

a variety of structural frameworks, ranging from

site-specific (e.g AP endonuclease) to structure-site-specific (e.g

RAD2⁄ XPG nuclease family) and nonspecific (e.g

DNase I) nucleases In particular, members of the

RAD2⁄ XPG nuclease family have unique nuclease

activities and play critical roles in genome stability [1–6]

In a preliminary report, we described the presence of a

new nuclease, Drosophila melanogaster XPG-like

endo-nuclease (DmGEN) which belongs to the RAD2⁄ XPG

nuclease family, shows unique activity and possibly

plays a critical role in genome stability [7] The ORF of

the DmGEN gene encoded a predicted protein of 726

amino acid residues with a molecular mass of 82.5 kDa

The gene was located at 64C9 on the left arm of

Drosophilapolytene chromosome 3 as a single site

The RAD2⁄ XPG family of nucleases, which have two conserved nuclease domains (the N domain and the I domain), are currently separated into three clas-ses (XPG⁄ class I, FEN-1 ⁄ class II and EXO-1 ⁄ class III) based on the types of nuclease activity and sequence homology [8,9] In Drosophila, mus201 protein (class I), FEN-1 homologue protein (class II), and Tosca protein (class III) have been reported as RAD2 family proteins The DmGEN protein showed a relatively high degree of sequence homology with RAD2 nuc-leases, particularly XPG, although the locations of the

N and I domains were similar to those of FEN-1 and EXO-1, and the molecular mass of DmGEN was found to be close to that of EXO-1 Therefore, we pro-posed a new class (class IV) to categorize DmGEN and SEND-1, which we also found in higher plants [8] Recently, a new member of the class IV nucleases, OsGEN-like, has been reported in rice; RNA-mediated

Keywords

active-site mutant; DmGEN; novel flap

endonuclease; nuclease activity;

site-directed mutagenesis

Correspondence

K Sakaguchi, Department of Applied

Biological Science, Faculty of Science and

Technology, Tokyo University of Science,

2641 Yamazaki, Noda-shi, Chiba-ken

278-8510, Japan

Fax: +81 4 7123 9767

Tel +81 4 7124 1501 (Ex 3409)

E-mail: kengo@rs.noda.tus.ac.jp

(Received 5 April 2007, revised 24 May

2007, accepted 7 June 2007)

doi:10.1111/j.1742-4658.2007.05924.x

Drosophila melanogaster XPG-like endonuclease (DmGEN) is a new cate-gory of nuclease belonging to the RAD2⁄ XPG family The DmGEN pro-tein has two nuclease domains (N and I domains) similar to XPG⁄ class I nucleases; however, unlike class I nucleases, in DmGEN these two nuclease domains are positioned close to each other as in FEN-1⁄ class II and EXO-1 ⁄ class III nucleases To confirm the properties of DmGEN, we characterized the active-site mutant protein (E143A E145A) and found that DmGEN had flap endonuclease activity DmGEN possessed weak nick-dependent 5¢)3¢ exonuclease activity Unlike XPG, DmGEN could not incise the bub-ble structure Interestingly, based on characterization of flap endonuclease activity, DmGEN preferred the blocked-flap structure as a substrate This feature is distinctly different from FEN-1 Furthermore, DmGEN cleaved the lagging strand of the model replication fork Immunostaining revealed that DmGEN was present in the nucleus of actively proliferating Dro-sophila embryos Thus, our studies revealed that DmGEN belongs to a new class (class IV) of the RAD2⁄ XPG nuclease family The biochemical properties of DmGEN and its possible role are also discussed

Abbreviations

BF, blocked flap; DmGEN, Drosophila melanogaster XPG-like endonuclease; dsDNS, double stranded DNA; RF, replication fork; ssDNA, single-stranded DNA.

silencing of the OsGEN-like caused male sterility due

to a defect in microspore development [9] Although

DmGEN homologues are found widely in mammals

and higher plants [7,9], knowledge about their

bio-chemical properties is limited In this study, we

deter-mined the biochemical properties of native and an

active-site mutant DmGEN to more deeply understand

the nature of this new class of nucleases

As for the biochemical features, class I consists of

XPG homologues, which cleave at the 3¢ side of the

bubble structure formed during nucleotide excision

repair [10,11] Class II comprises the FEN-1

homo-logues, which show 5¢-flap endonuclease, 5¢)3¢

exon-uclease and gap endonexon-uclease activities, and play

important roles in RNA primer removal, base excision

repair and apoptotic DNA fragmentation [12–14]

Clas-s III iClas-s made up of the EXO-1 homologueClas-s, which have

5¢)3¢ exonuclease activity and are involved in DNA

recombination, mismatch repair and DNA replication

[15–18] The function for class IV, however, remains

unclear In relation to the studies, we must correct

some mistakes in our previous study We reported

pre-viously that DmGEN has not only 3¢)5¢- and nick- and

gap-dependent 5¢-3¢ exonuclease activities, but also

endonuclease activity at a site 3 or 4 bp from the 5¢-end

[7] However, such activities were not found when

DmGEN was purified more carefully, although

nick-dependent 5¢-3¢ exonuclease activity was present Thus,

it is important to re-characterize this novel enzyme

Here, we report that the DmGEN protein is a new

flap endonuclease, which is different in nature from

FEN-1 Based on our studies we have confirmed that

DmGEN belongs to a new class of the RAD2⁄ XPG

family In addition, we have characterized the

bio-chemical properties of DmGEN

Results

Design of substrates

All DNA substrates were designed as shown in Fig 1,

and were assembled using the oligonucleotides

des-cribed in Table 1

Comparison of the nuclease domain

of the RAD2/XPG family

The RAD2⁄ XPG family of proteins have two

con-served nuclease domains (N and I domains), and these

are essential for nuclease activity and substrate

specif-icity DmGEN also has these conserved nuclease

domains The N and I domains of DmGEN were

similar to those of XPG⁄ class I (Fig 2A,B) The

N domain of DmGEN showed 35.1, 25.0 and 10.9% homology (% identity) with the N domains of HsXPG, HsFEN-1 and HsEXO-1, respectively The I domain of DmGEN showed 44.2, 38.5 and 38.5% homology (% identity) with the I domains of HsXPG, HsFEN-1 and HsEXO-1, respectively The spacer region between the N and I domains is not required for nuclease activity, but contributes to substrate spe-cificity [19] The spacer region of DmGEN is very short, similar to that of FEN-1⁄ class II and EXO-1 ⁄ class III, but not XPG⁄ class I (Fig 2A) Therefore, DmGEN cannot be categorized into class I, II, or III Like other members of the RAD2⁄ XPG family, DmGEN also contains several acidic residues coordi-nating two Mg2+at the active center for catalysis; one

of these, which is an aspartic acid residue in other members of the RAD2⁄ XPG family, is a glutamic acid residue in DmGEN (Fig 2B, asterisk 1) In addition

to the nuclease domains, the X-ray crystal structures

Fig 1 The three categories of DNA substrates used in this study Names shown (A, B, C, A-Flap, etc.) correspond to the oligonucleo-tides summarized in Table 1.

of the archaeal FEN-1 homologues have assisted in

the identification of critical structural elements (helical

clamp, H3TH motif and several loop regions) for

sub-strate binding [14] Recently, Qui et al [20] identified

18 positively charged amino acids that are important

in the FEN)1–DNA interaction DmGEN contains a

number of positively charged residues; however, most

of the positively charged amino acid residues forming

the DNA-biding domain of HsFEN-1 are not

con-served in DmGEN (Fig 2C, Table 2)

Expression, purification and characterization

of DmGEN

DmGEN was expressed in Escherichia coli, tagged

with six His residues at the N-terminus DmGEN

expression in E coli was increased dramatically over

the previously reported amount [7] by using the pCold I expression vector carrying the cold-shock promoter and inducing overexpression of DmGEN at

15C The recombinant protein was sequentially puri-fied by chromatography using a Ni-NTA resin col-umn, SP Sepharose beads, and then fractionated on a Superdex-200 gel-filtration column In the gel-filtra-tion column, the protein (expected molecular mass

 82.5 kDa) migrated between the expected molecular mass markers 75 and 100 kDa (Fig 3A) Gel-filtra-tion chromatography was crucial to completely purify the protein

Next, to characterize DmGEN nuclease activity more precisely, we constructed an active-site double mutant (E143A E145A) of DmGEN as described in Experimental procedures As shown in Fig 2B, these active-site residues (asterisks 2 and 3 in Fig 2B) are highly conserved in the I domain of the RAD2⁄ XPG family, and are important in coordinating divalent metal ions to interact with an incoming nucleotide [21] We previously reported that DmGEN has both 5¢)3¢ and 3¢)5¢ exonuclease activity and endonuclease activity at a site 3 or 4 bp from the 5¢-end in double-stranded DNA (dsDNA) [7] The mutants were used

to confirm these activities

We analyzed the nuclease activities of wild-type and E143A E145A double-mutant DmGEN Wild-type DmGEN did not show any detectable 3¢)5¢ exonuc-lease and endonucexonuc-lease activities towards either single-stranded DNA (ssDNA) or dsDNA (Fig 3D) These results differed from our previous report [7] The previ-ously reported nuclease activities of DmGEN [7] may have resulted from other contaminating nucleases Indeed, when we checked other fractions from the gel-filtration column, we found these activities in a frac-tion obtained from the shoulder area of the elufrac-tion profile In our previous study, we were not able to use gel-filtration for purification because of the low yield

of DmGEN

Subsequently, on further careful characterization, we found that purified wild-type DmGEN shows flap endonuclease activity, and the E143A E145A double mutant lacks such activity (Fig 3B) Thus, DmGEN cleaves the flap structure substrate at the junction between the ssDNA and dsDNA, and subsequently generates a product of 20 nucleotides This activity of DmGEN was confirmed using 3¢-end-labeled flap sub-strate (Fig 3C) As shown, the 3¢-end-labeled 30-nuc-leotide flap substrate was cleaved by wild-type DmGEN, but not by mutant DmGEN Neither wild-type nor mutant DmGEN cleaved a 30-nucleotide ssDNA substrate or blunt-ended dsDNA substrate (Fig 3D)

Table 1 Oligonucleotides used to construct various DNA

sub-strates shown in Fig 1.

Oligo

name Sequences (5¢- to 3¢)

GAC

AGCA

TTTTTTTTTTTTTTTTTTTTTTTTTTTTTCAG

TGCCACGTTGTATGCCCACGTTGACCG

TTTTTTTTTTTTTTTTTTTTTTTTTTTTTAT

GTCCTAGCAAAGCGTATGTGATCACTGG

CTTTGCCCACGTTGACCCG

TGTAATCGTCTATGACGTC

TGCTGTCTAGAGACTATCGC

GCCAGAATTCGGCAGCGTC

X1half GGACATCTTTGCCCACGTTGACCCG

X1half-g4 ATCTTTGCCCACGTTGACCCG

X1half-g8 TTGCCCACGTTGACCCG

X4half GCGATAGTCTCTAGACAGCATGTCC

Fig 2 Comparison of the amino acid sequences of the RAD2 ⁄ XPG family (A) Schematic representation of the conserved N and I domains

of some members of the RAD2 ⁄ XPG family The total number of amino acids in each protein and homology (% identity) between DmGEN and other members are indicated (B) The conserved sequences encompassing the nuclease active site are aligned Asterisk 1 indicates the nonconserved aspartic residue of DmGEN Asterisks 2 and 3 indicate the active residues of DmGEN substituted by site-directed mutagen-esis (C) Comparison of positively charged amino acid residues essential for the FEN )1–DNA interaction with those of DmGEN In total, 18 amino acid residues are compared in gray boxes.

Ability of DmGEN to cleave other structures

To analyze the substrate specificity of DmGEN, we

examined various test substrates, which were expected

to be cleaved by the nuclease These are shown

schemat-ically in Fig 1 First, we tested whether DmGEN

pos-sesses nick- or gap-dependent 5¢)3¢ exonuclease activity,

and produced gapped and nicked double-stranded

sub-strates as reported previously [7] As shown in Fig 4A,

DmGEN exhibited weak nick-dependent 5¢)3¢

exonuc-lease activity, but showed little or no gap-dependent

5¢)3¢ exonuclease activity We confirmed that DmGEN

cut only one nucleotide from the 3¢-end-labeled nicked

substrate (Fig 4B) Because we had to use a large

amount of DmGEN to cleave the nicked substrate, the

cleaving rate of the nick-dependent 5¢)3¢ exonuclease

activity of DmGEN is obviously lower than the flap

endonuclease activity Next, we tested whether DmGEN

would cleave the bubble-like and the Holliday junction

substrates, which are known to exist in vivo Although

XPG (class I nuclease) incised the target strand 3¢ to

the bubble-like and damage-containing structures [10],

DmGEN was unable to cleave the bubble-like structure

substrate (Fig 4C, left) Nor was the Holliday junction

substrate cleaved by DmGEN (Fig 4C, right)

Biochemical properties of DmGEN

To characterize the difference between the flap endo-nuclease activity of DmGEN and that of FEN-1 (class II), optimal reaction conditions for DmGEN were first determined using the flap structure substrate (Fig 5) (Details of the reaction conditions are given

in the legend to Fig 5.) The optimal pH for DmGEN flap activity was 8, which was same as for FEN-1 (class II) DmGEN required divalent metal ions (such

as Mg2+ and Mn2+), and the concentration of Mg2+

or Mn2+ ions required for optimal flap activity was

5 mm However, the cleavage product of DmGEN in the presence of 5 mm Mn2+ was only 43.3% that in the presence of 5 mm Mg2+ Ca2+ and Zn2+ could not substitute for Mg2+ or Mn2+ DmGEN nuclease activity was highest in reaction mixtures containing

25 mm KCl, and further increasing the concentration

of KCl inhibited the activity These biochemical prop-erties of DmGEN differed from data reported previ-ously [7] The requirement for divalent metal ions and low ionic strength for DmGEN optimal activity were like those of EXO-1 (class III) The above-described biochemical properties of DmGEN differed from those

of other members (class I, II, and III) of the RAD2⁄ XPG family

Effect of DmGEN on the various flap structures

To determine the flap nuclease activity of DmGEN,

we tested the action of DmGEN on several derivatives

of the flap structure substrates We prepared pseudo

Y, gapped-flap, blocked-flap, double-flap and 3¢-flap substrates, as shown in Fig 1 In agreement with pre-vious studies [14,22], FEN-1 cleaved the flap,

pseu-do Y, gapped-flap, blocked-flap and double-flap structure substrates (Fig 6) Previously, it was repor-ted that the blocked-flap substrate was hardly cleaved

by the flap endonuclease activity of FEN-1 [23] How-ever, according to a recent report, the blocked-flap was cleaved by the gap endonuclease activity of FEN-1 [14] Unlike FEN-1, DmGEN did not cleave the

pseu-do Y, gapped-flap and 3¢-flap, but did cleave the blocked-flap and double-flap structures In contrast to FEN-1, DmGEN preferred the blocked-flap structure substrate (Fig 6)

DmGEN cleaves blocked-flap structures and model replication fork substrates

To characterize DmGEN nuclease activity on the blocked-flap structure, we prepared blocked-flap struc-ture substrates with various sizes of oligonucleotides

Table 2 Conservation of essential positively charged amino acid

residue.

HsFEN-1 DmGEN Conserved Functional motif Binding site

loops

Upstream

loops

Upstream

loops

Upstream

loops

Upstream

loops

Upstream

loops

Downstream

loops

Upstream

loops

Upstream

bound to the 5¢-tail of the single-stranded flap The

sizes of the blocking oligonucleotides in the

blocked-flap (BF) substrates – BF1, BF2 and BF3 – were 15,

17 and 19 bp, respectively (Fig 7A) The

single-stran-ded region of the BF structure was 19 bp, and the gap

sizes of the blocked strand of BF1, BF2 and BF3 were

4, 2 and 0 nucleotides, respectively In agreement with a previous study [24], the cleavage efficiency of HsFEN-1 decreased with the narrower gapped substrate (Fig 7A) FEN-1 cleaved the blocked-flap substrate at much slower rate than the free flap struc-ture substrate However, DmGEN cleaved both BF1

Fig 3 Flap endnuclease activity of purified

wild-type and mutant recombinant DmGEN

(labeled as GEN) (A) Silver-stained gel

showing the molecular mass markers and

290 ng of purified DmGEN and DmGEN

(E143A E145A) Proteins were separated by

electrophoresis on a 10%

SDS-polyacryla-mide gel (B) Flap endonuclease activity at

different concentrations of DmGEN

5¢-End-labeled flap structure substrate (25 n M ) was

incubated with different amounts of

DmGEN (24, 48, 96 and 192 n M ) or DmGEN

E143A E145A double mutant (24, 48, 96

and 192 n M ) at 37 C for 90 min in a 20 lL

reaction volume (C) 3¢-End-labeled flap

structure substrate (25 n M ) was incubated

with DmGEN (24 and 48 n M ) or DmGEN

E143A E145A double mutant (96 n M ) at

37 C for 90 min in a 20 lL reaction volume.

(D) ssDNA and dsDNA substrate (25 n M )

was incubated with DmGEN (192 n M ) or

DmGEN E143A E145A double mutant

(192 n M ) at 37 C for 60 min in a 20 lL

reaction volume Asterisk indicates the

posi-tion of the radiolabel Substrate and

clea-vage product sizes were as indicated.

Electrophoresis was carried out on 10%

polyacrylamide ⁄ 7 M urea gels The amounts

of nuclease products were calculated with

the aid of an image analyzer ( IMAGE J 1.36b,

National Institutes of Health).

and BF2 to a similar extent as the nonblocked flap

substrate (Fig 7B) There was also some cleavage of

BF3, the blocked-flap substrate without gap (Fig 7B)

Because the free 5¢ ssDNA end of the flap is important

for FEN-1 cleavage efficiency [14,23,24], we also

exam-ined flap endonuclease activity on the hairpinned-flap

structure with no free 5¢-end We prepared the

hair-pinned-flap structure substrates with the same sequence

as the blocked-flap substrate (Table S1) In agreement

with a previous study [24], the cleavage efficiency of

HsFEN-1 on the hairpinned-flap substrates decreased

considerably with the narrower gapped substrate

(Fig S1A) However, DmGEN cleaved

hairpinned-flap, although the activity was weaker than on the free

flap substrate (Fig S1B)

Because DmGEN cleaved the blocked-flap structure,

we examined whether DmGEN cleaves model

repli-cation fork (RF) substrates Model replirepli-cation fork

substrates, in which the junction branch migrates,

were made as shown in Fig 7C We prepared four

derivatives (RF1–RF4) of the model replication fork

RF1 resembles the replication fork that lacks the

pro-gressing lagging strand RF2, RF3 and RF4 are the

forms of the normal replication fork differing in the

gap sizes of the lagging strand The gaps in the RF2,

RF3 and RF4 are 8, 4 and 0 bp, respectively As

shown in Fig 7C, DmGEN cleaved the lagging strand

of the model normal replication forks with gaps (RF2 and RF3) to the similar extent as the RF1 However, DmGEN poorly cleaved RF4, the substrate without any gap both at the lagging strand and the leading strand (Fig 7C)

Localization of DmGEN in Drosophila embryos

To confirm the relationship between DmGEN and DNA replication in vivo, immunostaining of Drosophila embryos was performed In Drosophila, the embryonic stages were separated into 17 steps [25] The first 13 nuclear divisions occurred in stage 1–4 embryos (0:00–2:10 h embryos) The first seven rounds take place within the interior of the embryo, the majority of nuclei then migrate to the cortex during cycles 8 and 9, leaving behind a small number of yolk nuclei [26] Polyclonal anti-DmGEN serum used for the immunocytochemical study reacted specifically with the DmGEN protein (expected molecular mass 82.5 kDa) in a crude extract of 0–3 h Drosophila embryos (Fig 8, left) As a result of immunostaining, DmGEN was localized in the nucleus throughout the

13 nuclear division cycles The nuclear localization of DmGEN was seen in the interior of the embryo at the

A

B

C

Fig 4 (A) Nuclease activity of DmGEN pro-tein (192 n M ) on the 5¢-end-labeled nicked and gapped substrates (25 n M ) The reaction condition is described in Experimental pro-cedures Time-course experiments were performed Substrates are depicted sche-matically in each panel The asterisk indicates the position of the radiolabel Substrate and cleavage product sizes were

as indicated (B) Nuclease activity of DmGEN protein (192 n M ) on the 3¢-end-labe-led nicked substrates (25 n M ) Time-course experiments were performed (C) Nuclease activity of DmGEN (192 n M ) on bubble and Holliday junction structure substrates (5 and

25 n M , respectively) Incubation was carried out at 37 C for 60 min in a 20 lL reaction volume Substrates are depicted schemati-cally in each panel Asterisk indicates the position of the radiolabel Substrate sizes were as indicated (A,B) Electrophoresis car-ried out on 20% polyacrylamide ⁄ 8 M urea gels (C) Electrophoresis carried out on 10% polyacrylamide ⁄ 7 M urea gels wt: DmGEN wild-type; mut: DmGEN E143A E145A double mutant.

stage 2 (Fig 8A–C, right) However, nuclear

localiza-tion of DmGEN was observed in a wide range of

embryo at the stage 3 (Fig 8D–F, right)

Discussion

The purpose of this study was to precisely characterize

a newly found member (class IV) of the RAD2⁄ XPG

family of nucleases, DmGEN, from Drosophila

melano-gaster The biochemical properties of class IV

nucleas-es are largely unknown in various animals and plants For this purpose, we created an active-site mutant, and used this mutant to confirm the biochemical prop-erties of DmGEN We purified wild-type and mutant DmGEN protein using an improved purification pro-tocol, and analyzed the nuclease activities of the puri-fied proteins Thus, we showed that DmGEN was a new type of flap endonuclease

Fig 5 Biochemical properties of the DmGEN protein Purified DmGEN (39 n M ) was incubated with 5¢-end-labeled flap structure substrate (25 n M ) at 37 C for 90 min in a 20 lL reaction volume To test the effect of divalent metal ions, the reaction was carried out in 1 m M dithio-threitol, 10% glycerol, 50 m M Tris (pH 8) supplemented with 50 m M KCl and various concentrations of a given divalent metal ion, as indica-ted in the figure To test the effect of salt, the reaction was carried out in 5 m M MgCl2, 1 m M dithiothreitol, 10% glycerol and 50 m M Tris (pH 8) and a given concentration of KCl, as indicated in the figure To test the effect of pH, the reaction was carried out in 5 m M MgCl 2 ,

1 m M dithiothreitol, 50 m M KCl, 10% glycerol and 50 m M Tris (pH 6.5–9.5) Following the reaction, products were resolved on 20% polyacryl-amide ⁄ 8 M urea gels and quantified using the IMAGE J 1.36b image analyzer.

The amino acid sequence of DmGEN protein has

three principal features (Fig 2) First, one of the acidic

residues at the active center for catalysis was not

con-served in DmGEN (Fig 2B, asterisk 1) Regarding this

nonconserved aspartic acid residue, Constantinou et al

[27] reported that the D77E active-site mutant of XPG

protein showed considerably lower nuclease activity

than wild-type XPG protein Second, most of the

posi-tively charged amino acids residues, which are essential

for binding FEN-1 to DNA [14,20], were not

con-served in DmGEN protein (Fig 2C, Table 2) These

two features contribute to the low nuclease activity of

DmGEN Lastly, DmGEN shows high homology

between the N and I regions and XPG (class I), but

the spacing of these regions is similar to in FEN-1 and

EXO-1 (class I and III, respectively) We confirmed

how this feature contributes the nuclease activity of

DmGEN (class IV) DmGEN had a flap endonuclease

activity, like FEN-1, but was not able to cleave the

bubble structure, unlike XPG (Figs 3, 4) Because

DmGEN had no 5¢)3¢ exonuclease activity on the

dsDNA substrate (Fig 3D), DmGEN is distinctly

dif-ferent from EXO-1 (class III) Recently, it was

sugges-ted that the activity of the RAD2⁄ XPG nuclease

family is determined by the properties and positions of

the two nuclease domains [19,28] The adjacent

posi-tion of the two domains may be responsible for not

cleaving the bubble structure (Fig 4C), because a

XPG mutant with a deletion in the spacer region was shown to prefer the pseudo Y structure to the bubble structure [19]

The flap endonuclease activity of DmGEN is more accurate and weaker than that of FEN-1 (class II nuc-lease) For example, FEN-1 cleaves many DNA struc-tures such as 5¢-single-strand overhang including flap, pseudo Y, gapped-flap and 5¢-overhang double-strand [14] In contrast, as shown Fig 6, DmGEN cleaves the normal flap substrate and a special flap structure: the blocked-flap substrate in which the 5¢-single-strand overhang of the flap is double-stranded; DmGEN cleaves just at the ssDNA⁄ dsDNA junction point We found very little cleavage of gapped-flap and pseudo Y substrates by DmGEN Therefore, the DNA structure

at the junction seems to be important for DmGEN-mediated cleavage Unlike pseudo Y and gapped-flap, DmGEN preferred a substrate in which the 5¢-upstream of the flap is completely double-stranded This idea is supported by the fact that DmGEN cleaved the double-flap substrate (Fig 6) The interest-ing feature of DmGEN is that this nuclease cleaves the blocked-flap structure, and this activity is slightly stronger than the normal flap structure cleaving activ-ity, a feature that is distinctly different from that of FEN-1 In agreement with the previous report [24], we also found that the activity of FEN-1 decreases consid-erably when the flap substrate is double-stranded (Fig 7A) On the hairpinned-flap substrates having no free 5¢-end, the nuclease activity of both FEN-1 and DmGEN are weaker than that on the normal flap substrate (Fig S1) Because FEN-1 prefers a free 5¢ ssDNA end of flap [14,23,24], the nuclease activity

on the hairpinned-flap substrate is weak, like for the blocked-flap substrate [24] Therefore, in contrast to FEN-1, DmGEN prefers a free 5¢-end of flap, which is either single- or double-stranded, this is deduced from the fact that DmGEN preferred the blocked-flap struc-ture, but not the hairpinned-flap structure These results suggest that binding of the substrate to DmGEN might differ from that of FEN-1 This is also suggested by the fact that most of the positively charged amino acids residues, which are essential for binding of FEN-1 to DNA [14], were not conserved in the DmGEN protein (Fig 2C)

The most interesting activity of DmGEN is that it cleaves the blocked-flap structure and the hairpinned-flap structure substrate (Fig 7B, Fig S1) The blocked-flap structure can be regarded as a model for the normal replication fork Interestingly, DmGEN cleaved the lagging strand of the model replication fork with gaps (Fig 7C) Furthermore, DmGEN was localized in the nucleus of Drosophila

Fig 6 Nuclease activities of DmGEN (51 n M ) and HsFEN-1

(4.7 n M ) on various flap structure substrates (5 n M ) Incubation was

carried out at 37 C for 60 min for DmGEN and 15 min for

HsFEN-1 in a 20 lL reaction volume Electrophoresis was carried out on a

20% polyacrylamide ⁄ 8 M urea gel Substrates are depicted

sche-matically in each panel Asterisk indicates the position of the

radio-label Substrate and cleavage product sizes were as indicated.

Amounts of nuclease products were calculated with the aid of the

IMAGE J 1.36b image analyzer wt: DmGEN wild-type; mut: DmGEN

E143A E145A double mutant; hFEN-1: HsFEN-1 wild-type.

B

C

Fig 7 (A) Nuclease activity of HsFEN-1

(4.7 n M ) on the flap structure substrate and

blocked-flap structure substrates (5 n M ) The

reaction condition is described in

Experi-mental procedures Time-course

experi-ments were performed Substrates are

depicted schematically in each panel

Aster-isk indicates the position of the radiolabel.

Substrate and cleavage product sizes were

as indicated Electrophoresis was carried

out on a 20% polyacrylamide ⁄ 8 M urea gel.

hFEN-1: HsFEN-1 (B) Nuclease activity of

DmGEN (48 n M ) on the flap structure

sub-strate and blocked-flap structure subsub-strates

(25 n M ) (C) Nuclease activity of DmGEN

protein (97 n M ) on the model replication fork

substrates (25 n M ) (B,C) Electrophoresis

was carried out on 10% polyacrylamide⁄ 7 M

urea gels Amounts of nuclease products

were calculated with the aid of the IMAGE J

1.36b image analyzer wt: DmGEN

wild-type; mut: DmGEN E143A E145A mutant.