Báo cáo khoa học: Sulfoquinovosylmonoacylglycerol inhibitory mode analysis of rat DNA polymerase b

Sulfoquinovosylmonoacylglycerol inhibitory mode analysis of rat DNA polymerase b Nobuyuki Kasai1, Yoshiyuki Mizushina2, Hiroshi Murata1, Takayuki Yamazaki1, Tadayasu Ohkubo3, Kengo Sakaguchi1 and Fumio Sugawara1

reported that

1 Department of Applied Biological Science, Tokyo University of Science, Noda, Chiba, Japan 2 Department of Nutritional Science, Kobe-Gakuin University, Kobe, Hyogo, Japan 3 Department of Pharmaceutical Sciences, Osaka University, Suita, Osaka, Japan

Keywords binding site; DNA polymerase b; inhibitor; NMR chemical shift mapping; sulfoquinovosylmonoacylglycerol

Correspondence F. Sugawara, Department of Applied Biological Science, Tokyo University of Science, Noda, Chiba 278–8510, Japan Fax: +81 4 7123 9767 Tel: +81 4 7124 1501 (ext. 3400) E-mail: sugawara@rs.noda.tus.ac.jp

(Received 11 May 2005, revised 29 June 2005, accepted 6 July 2005)

We have previously sulfoquinovosylmonoacylglycerol (SQMG) is a potent inhibitor of mammalian DNA polymerases. DNA polymerase b (pol b) is one of the most important enzymes protecting the cell against DNA damage by base excision repair. In this study, we charac- terized the inhibitory action of SQMG against rat pol b. SQMG competed with both the substrate and the template-primer for binding to pol b. A gel mobility shift assay and a polymerase activity assay showed that SQMG competed with DNA for a binding site on the N-terminal 8-kDa domain of pol b, subsequently inhibiting its catalytic activity. Fragments of SQMG such as sulfoquinovosylglycerol (SQG) and fatty acid (myristoleic acid, MA) weakly inhibited pol b activity and the inhibitory effect of a mixture of SQG and MA was stronger than that of SQG or MA. To characterize this inhibition more precisely, we attempted to identify the interaction interface between SQMG and the 8-kDa domain by NMR chemical shift mapping. Firstly, we determined the binding site on a fragment of SQMG, the SQG moiety. We observed chemical shift changes primarily at two sites, the residues comprising the C-terminus of helix-1 and the N-terminus of helix-2, and residues in helix-4. Finally, based on our present results and our previously reported study of the interaction interface of fatty acids, we constructed two three-dimensional models of a complex between the 8-kDa domain and SQMG and evaluated them by the mutational analysis. The models show a SQMG interaction interface that is consistent with the data.

screened many DNA polymerase

[1–6].

polymerase I [7,8]. SQMG showed potent antitumor activities in vivo in nude mice transplanted with human adenocarcinoma cells [9,10] and suppressed tumor cell proliferation in vitro [11]. We have already reported a pathway of total chemical synthesis of SQMG for bio- chemical and medicinal experiments [12].

We inhibitors obtained from natural sources, such as long chain unsaturated fatty acids, bile acids, terpenoids, and Sulfoquinovosylmonoacylglycerol sulfolipids (SQMG) (Fig. 1A,B), which was isolated from sea algae, has been shown to be a potent inhibitor of euk- aryotic DNA polymerases (pol) a, pol b, pol d, pol e, pol g, pol j, pol k, terminal deoxynucleotidyl trans- ferase (TdT) and HIV-1 reverse transcriptase, but not of prokaryotic polymerases such as E. coli DNA

Pol b is a key enzyme that protects the cell against DNA damage by base excision repair. Eukaryotic DNA polymerases are classiﬁed into four group: A, B, X and Y [13]. Pol b is a member of the polymerase X (pol X)

doi:10.1111/j.1742-4658.2005.04848.x

FEBS Journal 272 (2005) 4349–4361 ª 2005 FEBS

4349

Abbreviations HSQC, heteronuclear single quantum coherence; LA, lithocholic acid; MA, myristoleic acid; NA, nervonic acid; oligo(dT), oligo deoxyribothymidylic acid; Pol, DNA polymerase; SQG, sulfoquinovosylglycerol; SQMG, sulfoquinovosylmonoacylglycerol; TdT, terminal deoxynucleotidyl transferase.

N. Kasai et al. Interaction mode between SQMG and DNA Pol b

A

SO3H H

OH O

B

SO3H H

OH O

mammals [15]. Pol b has two domains with apparent ﬂexibility at a protease-sensitive region between residues 82–86. Trypsin treatment produced an N-terminal 8-kDa domain fragment, which retained binding afﬁnity for ssDNA, and a C-terminal 31-kDa domain fragment with reduced DNA polymerase activity. The crystal structure of the full-length pol b [16] and the solution structure of the 8-kDa domain of pol b have been repor- ted [17]. The crystal structure of the pol b-DNA com- plex has also been determined, and it reveals important structure-function relationships governing the processes of DNA polymerization and DNA repair [18,19]. Pol b is one of the most intensively investigated polymerases, particularly among those present in eukaryotic cells.

C

SO3H H

OH O

D

We have determined the binding sites for two types of pol b inhibitors, nervonic acid (NA) (Fig. 1E) and litho- cholic acid (LA) by NMR experiments [20,21]. These inhibitors bound to the 8-kDa domain of pol b and dis- turbed its binding to the template-primer DNA. In this study, we examine the structural interactions of SQMG with rat pol b and discuss the inhibitory action of SQMG against pol b, comparing this with mechanisms of other inhibitors. It is hoped that these studies will aid efforts to design more effective inhibitors of pol b.

Results and Discussion

E

Effects of two SQMGs and NA on the activity of rat DNA polymerase b

C O

F

C O

family, which is composed of pol b, pol k, pol l and TdT. Pol X family members share regions that are similar to the full-length pol b (two helix-hairpin-helix motifs and a pol X domain) [14]. Pol b is the smallest known DNA polymerase in mammalian cells, contain- ing 335 amino-acid residues with a molecular mass of 39 kDa, and its structure is highly conserved among

In this study, we examine two types of SQMG, whose fatty acid moieties occur at C14 and C18, respectively. As shown in Fig. 1, SQMG(C14:1) bears a myristoleic (Fig. 1F) on the glycerol moiety, and acid (MA) SQMG(C18:1) bears an oleic acid on the glycerol moiety. Figure 2A shows the inhibitory dose–response curves of SQMG(C18:1), SQMG(C14:1) and NA against pol b. We measured the DNA polymerization activity under the same condition in order to make precise comparisons between these inhibitors. IC50 values of SQMG(C18:1), SQMG(C14:1) and NA were determined to be 0.8, 1.8 and 5 lm, respectively. SQMG(C18:1) inhibited pol b activity more strongly than SQMG(C14:1). The inhibi- tory effect of SQMG showed a similar tendency to that of fatty acid [1]. The hydrophobic interaction is import- ant for binding, as the difference of SQMG(C14:1) and SQMG(C18:1) is only in the length of the fatty acid moi- ety. The inhibitory effect of SQMG was greater than that of NA. The molecular lengths of SQMG(C18:1), SQMG(C14:1) and NA are about 32.0, 27.0 and 28.4 A˚ , respectively, as derived from computer models. The molecular size of SQMG(C14:1) was very similar to that of NA. The difference in the inhibitory potency of

FEBS Journal 272 (2005) 4349–4361 ª 2005 FEBS

4350

[SQMG(C14:1)], the compounds (A) sulfoquino- (B) sulfoquinovosylmono- (D) (C) sulfoquinovosylglycerol (SQG), [SQMG(C18:1)], Fig. 1. Chemical structures of vosylmonoacylglycerol acylglycerol D-glucose, (E) nervonic acid (NA) and (F) myristoleic acid (MA).

N. Kasai et al. Interaction mode between SQMG and DNA Pol b

C

A

full-length pol β 0.5 pmol

31-kDa domain 1.5 pmol

100

)

SQMG(C14:1) conc.

( y t i v i t c a β e s a r e m y o p A N D

Compound (µM)

20 mer

17 mer

Lane I/P

1 100

2 10

3 1

4 0

5 100

6 0

B

8-kDa domain 0.15 nmol

full-length pol β 0.15 nmol

SQMG(C14:1) conc.

start

DNA + pol β complex

DNA + 8-kDa domain complex

M13 ssDNA

3 4 5 6 7

Lane 1 2 I/P

0 0.1 1 10

9 10 0 0.1 1 10

SQMG and NA can be attributed to the relative hydro- phobicity of the sulfoquinovosyl moiety vs. the hydroxyl moiety.

Mode of DNA polymerase b inhibition by SQMG and NA

In order to elucidate the inhibition mechanism, the extent of inhibition as a function of DNA template- primer or dNTP substrate concentrations was studied

(Table 1). SQMG(C18:1) inﬂuenced the activities more strongly than did SQMG(C14:1); Table 1 shows a kinetic analysis of the inhibitors. In this analysis, poly(dA) ⁄ oligo(dT)12)18 and dTTP were used as the DNA template-primer and dNTP substrate, respect- ively. Double reciprocal plots of the results show that all of the inhibitors tested for pol b activity competed with the DNA template and the substrate (Table 1). In the case of the DNA template-primer, the apparent maximum velocity (Vmax) was unchanged at 111 pmolÆ

FEBS Journal 272 (2005) 4349–4361 ª 2005 FEBS

4351

Fig. 2. (A) Dose–response curves of SQMG(C14:1) and SQMG(C18:1) and nervonic acid. Rat DNA polymerase b (0.05 units) was preincubated with the indicated concentrations (0–10 lM) of SQMG(C14:1) (j), SQMG(C18:1) (d) or NA (n). DNA polymerase activity in the absence of added compounds was taken to be 100%. (B) Gel mobility shift analysis. Gel mobility shift analysis of binding between M13 ssDNA and DNA polymerase b. M13 plasmid ssDNA (2.2 nmol; nucleotide, single strand and singly primed) was mixed with puriﬁed proteins and SQMG(C14:1). Lanes 2–5 contained the full-length DNA polymerase b at a concentration of 7.5 lM; lanes 7–10 contained the 8-kDa domain at a concentration of 7.5 lM; lanes 1 and 6 contained no protein. Lanes 2, 3, 4, 5, 7, 8, 9 and 10 were mixed with various concentrations of SQMG(C14:1). The concentrations were as follows: lanes 2 and 7, lanes 3 and 8, lanes 4 and 9, and lanes 5 and 10 were zero, 0.75, 7.5 and 75 lM, respectively. (C) Analysis of the poly(dA) ⁄ oligo(dT)16 template ⁄ primer synthetic products. DNA synthetic reactions were carried out with 20 lM poly(dA) ⁄ oligo(dT)16 (¼ 2 ⁄ 1) and 20 lM [32P]dTTP[aP] (60 CiÆmmol)1), and the products were examined by gel electrophoresis and imaging analysis as described in the Experimental procedures section. The protein concentrations were as follows: lanes 1–4, 25 nM of the full-length DNA polymerase b; lanes 5 and 6, 75 nM of the 31-kDa domain. SQMG(C14:1) concentrations were as follows: lanes 1–6 were 2500, 250, 25, 0, 7500 and 0 nM, respectively. Markers indicate the positions of the extended primer.

N. Kasai et al. Interaction mode between SQMG and DNA Pol b

Table 1. Kinetic analysis of the inhibitory effects of sulfoquinovosylmonoacylglycerols (SQMG(C14:1), SQMG(C18:1)) and NA on the activities of DNA polymerase b, as a function of the DNA template-primer and the nucleotide substrate concentrations. Rat DNA polymerase b was 0.05 units.

a (pmoÆh)1)

b (lM)

Compound Inhibitory modea Substrate conc. (lM) Compound (lM) Km Vmax Ki

SQMG(C14:1) DNA template-primerc 111 0.89 Competitive 6.74

0 1 2 16.3 33.2

Nucleotide substrated 62.5 2.8 Competitive 3.05 4.95 0 1 2 20.0 SQMG(C18:1) DNA template-primer 111 0.42 Competitive 6.74

0 0.5 1 18.4 34.8

Nucleotide substrate 62.5 1.44 Competitive 3.05 6.34 0 0.5 1 23.5 NA DNA template-primer 111 4.0 Competitive 6.74

0 4 6 17.2 31.0

a From Lineweaver–Burke plot. b From Dixon plot. c Poly (dA) ⁄ oligo(dT)12)18. d dTTP.

h)1, whereas 242% and 493% increases in the Michael- is constant (Km) were observed in the presence of 1 and 2 lm SQMG(C14:1), respectively (Table 1). The Vmax for the dNTP substrate was 62.5 pmolÆh)1, and the Km for the substrate increased from 3.05 to 20.0 lm in the presence of 2 lm SQMG(C14:1) (Table 1). The inhibitor constant (Ki) values, obtained from Dixon plots, were found to be 0.89 lm and 2.8 lm in the presence of 2 mm for the DNA template-primer and the dNTP sub- strate, respectively (Table 1). Similarly, the Ki values of SQMG(C18:1) were found to be 0.42 lm and 1.44 lm for the DNA template-primer and dNTP substrates, respectively, and the Ki values of NA were found to be 4.0 lm and 3.5 lm for the DNA template-primer and dNTP substrates, respectively. All of the pol b inhibi- tors examined competed with both the DNA template- primer and the dNTP substrate.

Binding analysis comparing SQMG and the N-terminal 8-kDa domain of pol b by a gel mobility shift assay

mobility shift assay. Figure 2B shows results of a gel mobility shift assay demonstrating M13 single stranded DNA (ssDNA) binding to the full-length pol b (lane 2), as well as to the 8-kDa domain (lane 7). The full-length pol b and the 8-kDa domain formed complexes with the M13 ssDNA, leading to changes in the DNA mobility that appeared as shifts in its position. However, the 31-kDa domain, the polymerization domain without a DNA-binding site, was not shifted [23]. SQMG(C14:1) interfered with complex formation between M13 ssDNA and pol b (left panel) and between M13 ssDNA and the 8-kDa (right panel) to the same extent. The molecular ratios of SQMG(C14:1) (I) and the proteins (P) are repre- interference by sented by I ⁄ P in Fig. 2B. The SQMG(C14:1) is shown with the I ⁄ P ratios in lanes 2, 3, 4 and 5, and in lanes 7, 8, 9 and 10 of 0, 0.1, 1 and 10, respectively. The interference by SQMG(C14:1) was nearly complete at an I ⁄ P ratio of 1, and it disappeared at the ratio of 0.1, suggesting that one molecule of SQMG(C14:1) competed with one molecule of M13 ssDNA and subsequently interfered with the binding of DNA to the full-length pol b or to the 8-kDa domain. the gel mobility shift assay using The results of SQMG(C18:1) instead of SQMG(C14:1) were similar (data not shown).

We investigated the interaction between the 8-kDa domain of pol b and SQMG in detail. The DNA binding activity of the 8-kDa domain was analyzed using a gel

FEBS Journal 272 (2005) 4349–4361 ª 2005 FEBS

4352

Nucleotide substrate 62.5 3.5 Competitive 3.05 4.80 0 4 6 18.7

Product analysis after DNA synthesis on poly(dA) ⁄ oligo(dT)

the IC50 value was 120 lm. When SQG was present in the polymerase reaction mixture, the MA inhibitory effect on pol b was approximately 2.6-fold stronger. The pol b inhibitory effect of SQMG(C14:1) was stron- ger than that of a mixture of SQG and MA (Fig. 3B).

An excessive amount of SQG or MA (i.e. I ⁄ P ¼ 10) did not inhibit the ssDNA binding activity of the 8-kDa domain of pol b (Lanes 3 and 4 of Fig. 3C). On the other hand, a mixture of SQG and MA inhibited the activity (Lane 5 of Fig. 3C). As the mode of the pol b inhibition by SQG and MA was competitive against both DNA template-primer and dNTP sub- strate (data not shown), it was suggested that a mix- ture of SQG and MA could also competitively inhibit the binding activity of DNA template-primer. These results suggested that the SQG moiety could enhance the inhibitory activities of the DNA polymerization and ssDNA binding by MA.

We examined whether the catalytic activity on the 31-kDa domain was inhibited by SQMG. The 31-kDa domain can bind to the DNA template-primer (although weakly), and it retains the DNA polymeriza- tion activity. We used poly(dA)oligo(dT)16 as the tem- plate-primer, and analyzed newly synthesized DNA fragments produced by the 31-kDa domain (Fig. 2C). The reaction products in vitro were investigated by using denaturing polyacrylamide gel electrophoresis. Figure 2C shows the products formed by the full- length pol b (lanes 1–4) or the 31-kDa domain (lanes 5–6). It is known that DNA polymerase b is a distribu- tive enzyme [24], which adds a single nucleotide and then dissociates from the template-product complex. The 31-kDa domain can replicate DNA in a similar manner to the full-length pol b.

NMR experiment to determine the interaction interface between SQMG(C14:1) and the 8-kDa domain

the

Within a 10-minute incubation period, most of the primers were elongated (lane 4). With 1.5 pmol of the 31-kDa domain, DNA replication was observed (lane 6). The 8-kDa domain fragment was incapable of repli- cating DNA [23]. At an I ⁄ P ratio of more than 10, SQMG(C14:1) (lanes 1–2) completely suppressed DNA polymerization by the full-length pol b. At an I ⁄ P ratio of 1 for the protein (lane3), DNA synthesis hardly occurred. However, the 31-kDa domain synthesized DNA without interference from SQMG(C14:1) (lane 5). At the range of the SQMG(C14:1) concentrations that inﬂuence the template-primer-binding site on the 8-kDa domain, SQMG(C14:1) is thus thought to indirectly inhi- bit DNA polymerization at the 31-kDa catalytic site because the site lacks a template-primer, and it is also thought to compete with the substrate. The results of the products analysis using SQMG(C18:1) instead of SQMG(C14:1) were similar (data not shown).

Biochemical characterization of fragments of SQMG

fragments

SQMG(C14:1), the

separated two sulfoquinovosylglycerol

A titration experiment using the 8-kDa domain and a 1 mm stock solution of SQMG(C14:1) was performed as follows. Two-dimensional 1H-15N HSQC spectra of the 8-kDa domain-SQMG(C14:1) complex at SQMG(C14:1) concentrations of 0.05, 0.1, 0.15 and 0.2 mm were concentration of SQMG(C14:1) recorded. As increased, the cross-peaks of the 8-kDa domain broad- ened. At an SQMG(C14:1) concentration of 0.1 mm, most of the cross-peaks disappeared and some broad cross-peaks appeared at 7.8–8.5 p.p.m. SQMG(C14:1) may aggregate at the millimolar concentration required for NMR experiments, and the 8-kDa domain may interact with micelle-like forms of SQMG(C14:1) [25]. If the experiment could be carried out at micromolar concentrations, SQMG(C14:1) would not aggregate, as SQMG(C14:1) inhibited the DNA polymerization activ- ity of the full-length pol b but not the 31-kDa domain. This ﬁnding indicated that pol b was not denatured by surface-active effects of SQMG(C14:1). Consequently, the NMR relaxation time shortening was due to the increase of the apparent molecular weight, leading to the appearance of cross-peaks of unstructured residues. For this reason, we could not directly identify the the 8-kDa interaction interface of SQMG(C14:1) domain. To avoid the aggregation of SQMG, we used chemically synthesized SQG, which did not bind the fatty acid moiety.

To determine the inhibitory mechanism of pol b by of SQMG(C14:1), (SQG) (Fig. 1C) moiety and the myristoleic acid (MA) moiety, were prepared. SQG weakly inhibited the DNA poly- merization activity of pol b with the IC50 value of 7.95 mm (Fig. 3A). The inhibition dose–response curves of SQG and MA against pol b were shown in Fig. 3B. In the range of 0–1 mm, SQG did not inﬂuence pol b activity, although MA inhibited it with the IC50 value of 375 lm. The inhibitory effect of a mixture of SQG and MA was stronger than that of SQG or MA, and

The fragment linking method is commonly used in the NMR-based drug design process [26]. A strongly binding compound can be synthesized by combining

FEBS Journal 272 (2005) 4349–4361 ª 2005 FEBS

4353

N. Kasai et al. Interaction mode between SQMG and DNA Pol b

A ) 100 %

B ) 100 %

( y t i v i t c a β e s a r e m y o p A N D

0 0.2 0.4 0.6 0.8 1 Compound (mM)

0 20 40 60 80 100 SQG (mM)

C

8-kDa domain 7.5 µM

A M + G Q S

G Q S

A M

start

DNA+8-kDa domain complex

M13 ssDNA

Lane 1 2 3 4 5 I/P - 0 10 10 10

several low afﬁnity compounds with different binding sites for a target protein. By applying this fragment linking method inversely, we attempted to identify the interaction interface of SQMG with the 8-kDa domain. Firstly, we determined the interaction interfaces of the SQG and fatty acid separately. We then combined these two interaction interfaces and identiﬁed the SQMG interaction interface of the 8-kDa domain.

Analysis of the SQG interface with the 8-kDa domain by NMR chemical shift changes

spectra were recorded for the 8-kDa domain-SQG complex at SQG concentrations of 10, 30, 60 and 100 mm. The pol b-SQG complex was in the fast exchange region on the NMR time scale, permitting us to follow the chemical shift changes of the backbone NH and 15N signals of the 8-kDa domain upon complex formation. This was achieved by recording a series of 1H-15N HSQC spectra of the uniformly 15N- labeled 8-kDa domain in the presence of increasing amounts of SQG. Of the 80 amides in residues 6–87 of the 8-kDa domain, 76 amides were assigned in the SQG complex using the CBCA(CO)NH and HNCACB spectra to conﬁrm the reported assignments [17]. NH and 15N chemical shift differences along the amino-acid sequence of the 8-kDa domain in the pres- ence of 100 mm SQG are indicated in Fig. 4.

The solution structure of the 8-kDa domain has been determined by Mullen et al. [17]. According to their results, the 8-kDa domain (residues 1–87) formed four a-helices packed as two antiparallel pairs. The pairs of a-helices crossed each other at 60(cid:1), producing a V-like shape. The 8-kDa domain contains a helix-hairpin-helix motif that is classiﬁed as a DNA binding domain. There is a hydrophobic region between helix-1 and helix-2.

The 8-kDa domain was titrated with a 1 m stock 1H-15N HSQC

The residues displaying chemical shift changes upon binding to SQG in the structure of the 8-kDa domain with or without SQG are shown in Fig. 5A. The surfa- ces of residues with NH chemical shift changes in the range of 0.02–0.03 p.p.m and 15N chemical shift chan- ges of 0.15–0.25 p.p.m. (A6, T10, L11, G13, V20, L22,

solution of SQG. Two-dimensional

FEBS Journal 272 (2005) 4349–4361 ª 2005 FEBS

4354

Fig. 3. (A) Dose–response curve of SQG. Rat DNA polymerase b (0.05 units) was pre- incubated with the indicated concentrations (0–100 mM) of SQG. DNA polymerase activ- ity in the absence of added compounds was taken to be 100%. (B) Dose–response curves of SQG, MA, and a mixture of SQG and MA.Rat DNA polymerase b (0.05 units) was preincubated with the indicated con- centrations (0–1 mM) of SQG (d), MA (n) or a mixture of SQG and MA (s). The DNA polymerase activity in the absence of added compounds was taken to be 100%. (C) Gel mobility shift analysis. Gel mobility shift ana- lysis of binding between M13 ssDNA and the 8-kDa domain of pol b. M13 plasmid ssDNA (2.2 nmol; nucleotide, single strand and singly primed) was mixed with puriﬁed proteins and SQMG(C14:1). Lanes 2–5 con- tained the 8-kDa domain at a concentration of 7.5 lM; lanes 1 contained no protein. The compounds (75 lM each) were as follows: lanes 3, 4, and 5 were SQG, MA, and a mix- ture of SQG and MA, respectively.

N. Kasai et al. Interaction mode between SQMG and DNA Pol b

A

(N24, V29, S30, I33, Y39 and I69) are colored orange. NH chemical shift changes exceeding 0.04 p.p.m and 15N chemical shift changes exceeding 0.35 p.p.m. (E9, A23, E26, K35, H51, G66, A70, R83 and L85) are colored red.

B

C

In the presence of SQG, the cross-peaks were shifted as follows: A6, E9 and L11 were in the unstructured segment. G13, V20, L22, A23, N24, F25, E26, K27 and N28 were in helix-1; V29, S30 and Q31 were in the loop between helix-1 and helix-2; I33, K35, Y36 and N37 were in helix-2; H51 was in the loop between helix-2 and helix-3; K60 and L62 were in helix-3; G64 and G66 were in the loop between helix-3 and helix-4; I69, A70, D74 and L77 were in helix-4; L82, R83, K84 and L85 were in the unstructured linker segment that connected to the 31-kDa catalytic domain. The N- (residues 1–10) and C-termini (residues 83–87) were disordered, as judged from the heteronuclear 15N-{1H} NOE data (values < 0.4) [17]. As the chemical shifts of the residues in the disordered regions are changed easily by minor changes in the environment (buffer, in the disordered etc.), we excluded the residues regions from our analysis. These chemical shift chan- ges could be explained in terms of SQG contact and perturbations of the electrostatic charge distribution at the surface. Surface residues displaying chemical shift changes were predominantly, although not entirely, clustered at two sites of the 8-kDa domain (Fig. 5A), e.g. Site I: L22, F25, E26, N28, I33, K35, N37 and Y39, and Site II: K60, L62, G64, G66 and A70.

the inhibitor constant

F25, K27, N28, Q31, Y36, N37, V45, K60, L62, G64, D74, L77, L82 and K84) are colored yellow. Those with NH chemical shift changes of 0.03–0.04 p.p.m and 15N chemical shift changes of 0.25–0.35 p.p.m.

The cross-peak for K35 at Site I was sufﬁciently resolved during the titration to determine the mole fraction of protein bound to SQG. The backbone amide proton of K35 displayed a long chemical shift change upon complex formation. The change in the chemical shift of the K35 resonance was interpreted as resulting from an average (dav) of the chemical shifts for the free and the bound forms (db) of the K35 resonance. Similarly, the KD value was determined by the chemical shift change of G66 at Site II. Assuming that SQG binds to the 8-kDa domain as a 2 : 1 com- plex with each site having the same afﬁnity, the KD values determined by K35 and G66 were 59 and 79 mm, respectively (Fig. 6). The average KD value was 69 mm. We have reported that the KD value of NA was 1.02 mm [20]. Linking of two moieties that each has a millimolar afﬁnity has been reported to cre- ate a compound with a micromolar afﬁnity [26]. Thus, (Ki) it was reasonable that values of SQMG(C14:1) and SQMG(C18:1) were found to be 0.89 lm and 0.42 lm, respectively (Table 1). In order to determine whether or not the chemical shift the 8-kDa change induced by SQG was

speciﬁc,

FEBS Journal 272 (2005) 4349–4361 ª 2005 FEBS

4355

Fig. 4. Chemical shift changes of HN and 15N for the pol b 8-kDa (A) Overlay of the domain upon complex formation with SQG. 1H-15N HSQC spectra of the 15N-labeled N-terminal 8-kDa domain of pol b (0.1 mM) in the absence (blue) and presence (red) of 100 mM SQG. (B and C) Chemical shift changes, |D1H| (panel B) and |D15N| (panel C), are plotted vs. residue number, where D1H and D15N are the differences in p.p.m. between the free and bound chemical shifts.

0.06

)

N. Kasai et al. Interaction mode between SQMG and DNA Pol b

L77L77

H51H51

0.05

Site II Site II

D74D74

K60K60

A70A70

K60K60

L62L62

0.04

A

90 o

G64G64

G66G66

L22L22

Y39Y39

Site I Site I

N24N24

0.03

E26E26

F25F25

Site I Site I

E26E26 K35K35

K35K35

V29V29

F25F25 V29V29 N28N28

I33I33

K27K27

0.02

Q31Q31

S30S30

0.01

K52K52

Site III Site III

L77L77

Site III Site III

T79T79

H51H51

m p p ( e c n e r e f f i d t f i h s l a c m e h c H N

L77L77

H51H51

T79T79

D74D74

G80G80

D74D74

100

F76F76

B

SQG (mM)

Site I Site I

L22L22

Y39Y39

E26E26

E26E26 K35K35

Site I Site I

K35K35

V29V29

I33I33

V29V29

S30S30

Fig. 6. Determination of KD for SQG binding to the 8-kDa domain of DNA polymerase b. Titration of SQG was performed to measure the chemical shift change at the nondegenerate K35 (diamonds) and G66 (triangles) NH in 1H-15N HSQC spectra at 750 MHz (25 (cid:1)C). The average KD value of SQG was 69 mM.

K52K52

L77L77

H51H51

T79T79

H51H51

T79T79

G80G80

D74D74

F76F76

K60K60

A70A70

K60K60

C

L62L62

G64G64

G66G66

L22L22

Y39Y39

N24N24

E26E26

F25F25

K35K35

E26E26 K35K35

F25F25 V29V29 N28N28

V29V29

I33I33

K27K27

Q31Q31

S30S30

domain was titrated with d-glucose (Fig. 1D) at con- centrations of 10, 50 and 100 mm. d-glucose is a pyra- nose, as is SQG, but it possesses neither a sulfonyl nor a glycerol moiety. No NH or 15N chemical shift chan- ges were observed upon addition of d-glucose. There- fore, the interaction of SQG with the 8-kDa domain was speciﬁc.

I73I73

K60K60

Analysis of the SQMG binding site on the pol b 8-kDa domain

E71E71

L62L62

K72K72

D

I69I69

G66G66

K72K72

G64G64

G66G66

K68K68

F25F25

K35K35

We determined the interaction interface between fatty acids and the 8-kDa domain of pol b in the previous report [20]. We analyzed the binding site of SQMG based on the results of the NMR chemical shift map- pings of SQG (Fig. 5A) and the fatty acid (Fig. 5B) [20]. We propose two possible models of the SQMG- pol b complex. We constructed both models of the complex between the 8-kDa domain and SQMG based on the following analysis (Fig. 7).

(1–10) and C-terminal

In the ﬁrst model (hereafter referred to as Model A), the sulfoquinovosylglycerol moiety of SQMG interacts with the 8-kDa domain at Site I and the alkyl moiety of SQMG interacts with the C-terminus of helix-4 (Site III). In the model of the fatty acid complex with the 8-kDa domain, the carboxyl moiety of the fatty acid interacts with Site I and the alkyl moiety interacts at Site III (Fig. 7A). There is a hydrophobic region between helix-1 and helix-2 (Fig. 5C). There is an over- lap at Site I in the interaction interfaces of SQG and the fatty acid. In Model A, the sulfoquinovosylglycerol moiety of SQMG is bound to Site I instead of the carboxyl moiety of the fatty acid, and the alkyl moiety binds to Site III.

FEBS Journal 272 (2005) 4349–4361 ª 2005 FEBS

4356

Fig. 5. Interaction interfaces between DNA polymerase b and SQG, fatty acid and ssDNA, and hydrophobicity representation; the N-ter- minal (81–87) unstructured regions were removed for clarity. (A) Interaction interface between SQG and the 8-kDa domain. The amino-acid residues of the major shifted cross- peaks from the 1H-15N HSQC spectra are indicated. NH chemical shift changes of 0.02–0.03 p.p.m and 15N chemical shift changes of 0.15–0.25 p.p.m. are depicted in yellow. NH chemical shift changes of 0.03–0.04 p.p.m and 15N chemical shift changes of 0.25– 0.35 p.p.m. are indicated in orange. NH chemical shift changes of more than 0.04 p.p.m and 15N chemical shift changes of more than 0.35 p.p.m. are indicated in red. (B) Interaction interface between fatty acids and the 8-kDa domain. The amino-acid residues of the major shifted cross-peaks from the 1H-15N HSQC spectra are indica- ted in red. the molecular surfaces (i.e. blue is hydrophilic and red is hydrophobic). These images were prepared using the computer program INSIGHT II. (D) Interaction interface between ssDNA and the 8-kDa domain. The amino acid residues related to DNA binding are depicted in cyan.

N. Kasai et al. Interaction mode between SQMG and DNA Pol b

A

that

structures sulfate moiety. We analyzed 35 crystal deposited in the PDB, which were collected based on the criteria listed in a previous report [27]. The sulfonyl moieties interacted with the sidechain of arginine and lysine in 12 and 10 crystal structures, respectively. This implies the sulfoquinovosylglycerol moiety of SQMG would interact with residues in Site I. K35 is the only basic amino-acid residue in Site I and the NH chemical shift of K35 was greatly changed by addition of SQG. Thus, the sulfonyl moiety of SQMG may form a salt bridge to the amino moiety of the side- chain of K35. The hydroxyl moieties of the sugar of SQMG might interact with the sidechain carboxyl moi- ety of E26. The NH chemical shift of E26 was also changed greatly by addition of SQG.

F25

K35

B

In the second model (hereafter referred to as Model B), the sulfoquinovosylglycerol moiety of SQMG inter- acted with the 8-kDa domain at Site II and the alkyl moiety of SQMG interacted at Site III (Fig. 7B). At Site II, the residues in which NH or 15N chemical shift were greatly changed were G66, I69 and A70, which does not possess any amino moiety. The survey of crystal structures showed that the sulfonyl moieties interacted with the backbone amide in 10 out of 35 crystal structures. Thus, the sulfonyl moiety of SQMG might bind to the backbones of these residues, as shown in this model.

A70

G66

We examined the general binding mode of the sulfo- nyl moiety by analysis of crystal structures of com- plexes between proteins and sugars containing the

To examine which is a more reasonable model, we performed a mutational analysis of pol b. We altered four residues whose chemical shifts were greatly chan- ged by addition of SQG. In Site I, we mutated E26 and K35 to alanine to remove the charged moieties of the sidechains. In Site II, we altered G66 and A70 to proline to remove the backbone amide protons. All the mutants of pol b retained the DNA polymerization activity. We measured the SQMG(C14:1) inhibitory the DNA polymerization activity against effects of (Table 2). The IC50 value of these four mutants SQMG(C14:1) against the wild type pol b protein was 1.8 lm. On the other hand, the IC50 values against the E26A, G66P and A70P mutants were determined to be 10.6, 89.2 and 11.8 lm, respectively, whereas that against the K35A mutant was more than 200 lm. As the inhibitory effects of SQMG(C14:1) on all the mutants decreased signiﬁcantly, these four residues may be involved in the interaction with SQMG(C14:1). inhibitory effect on the G66P The SQMG(C14:1) mutant was approximately 50-fold weaker compared to that of the wild type pol b protein. Moreover, the IC50 value against the K35A mutant was more than two times that against the G66P mutant. The K35A mutant was inﬂuenced most weakly among the four mutants. Therefore, these results suggested that Model

FEBS Journal 272 (2005) 4349–4361 ª 2005 FEBS

4357

Fig. 7. Possible structures of the complex formed between the 8-kDa domain and SQMG(C14:1).The sulfurs, carbons, oxygens, and hydrogens in the inhibitor structures are indicated in orange, green, red, and white, respectively. (A) Model A. SQMG(C14:1) binds to the 8-kDa domain of pol b at Site I and Site III. The molecular orien- tation of pol b-SQMG(C14:1) is almost the same as that in Fig. 5 in the left column image. (B) Model B. SQMG(C14:1) binds to the 8-kDa domain at Site II and Site III. The molecular orientation of pol b-SQMG(C14:1) is almost the same as that in Fig. 5 in the right col- umn image. These images were prepared using PYMOL (DeLano Sci- entiﬁc, CA, USA).

N. Kasai et al. Interaction mode between SQMG and DNA Pol b

Table 2. IC50 values of SQMG(C14:1) against the DNA polymeriza- tion activity of mutants of DNA polymerase b. SQMG(C14:1) was incubated with each enzyme (0.05 units). The enzymatic activity was measured as described under Experimental procedures. Enzyme activity in the absence of the compound was taken as 100%.

Pol b IC50 values of SQMG(C14:1) (lM)

1.8

10.6 > 200

would interact with DNA binding site and compete with DNA in a similar fashion, whereas the hydropho- bic part of SQMG would bind to and then anchor at Site III. Both hydrophobic and hydrophilic types of afﬁnity contribute to the formation of the SQMG-pol b complex. SQMG(C18:1) showed a larger inhibitory effect on pol b than did SQMG(C14:1). Their structural difference was just in the length of the fatty acid moi- ety. This suggests that the fatty acid moiety contributes to the binding afﬁnity to some extent. In the case of fatty acids, the inhibitory effect increased in propor- tion to the number of carbons comprising the alkyl chain [1]. These three-dimensional structural models could facilitate the design of more potent inhibitors for DNA pol b.

and the

A may be more reasonable to represent the interaction 8-kDa interface between SQMG(C14:1) domain.

Proposed inhibitory mode of SQMG against pol b

the interaction interface of

and the

SQMG inhibited not only pol b, but also pol a, pol d, pol e, pol g, pol j, pol k and TdT [8]. It was sug- gested that similar binding sites were present in these mammalian polymerases. For example, they might possess hydrophobic cores adjacent to DNA binding sites where SQMG could interact. Their amino-acid sequences differ, but they might have similar three- dimensional structures. The binding site might be essential for their DNA polymerase activity, and such a region might have been conserved evolutionarily among the mammalian polymerases. Low molecular weight organic compounds may prove useful as molecular probes to investigate the structural homo- logy structure-function relationships of enzymes whose three-dimensional structures are as yet unknown.

Wild type Mutants E26A K35A G66P A70P 89.2 11.8

Experimental procedures

Materials

the Figure 5D shows 8-kDa domain with ssDNA. This model is based on site-directed mutagenesis assays [28] and NMR experi- ments [17]. According to the report of Prasad et al. [28], point mutants at F25, K35, K60, and K68 showed impaired ssDNA binding activity. The NMR experiment indicated which residues (K60, L62, G64, G66, I69, E71, K72, I73 and R83) had NH chemical shift changes over 0.2 p.p.m and 15N chemical shift changes over 1.0 p.p.m. upon addition of 5 mer- ssDNA, p(dT)8 or 9 mer-ssDNA [17]. In Model A, SQMG competes with template DNA for binding to Site I, and subsequently inhibits the template DNA binding to the 8-kDa domain. Binding of SQMG to K35 would disrupt its interaction with ssDNA. In Model B, SQMG competes with template DNA for binding to Site II. Subsequently, SQMG blocks bind- ing of template DNA to pol b. In both models, SQMG would prevent template DNA binding to the 8 kDa domain at Site I or Site II. Consequently, SQMG would inhibit the DNA polymerization activ- ity of pol b.

Sulfoquinovosylmonoacylglycerol and sulfoquinovosylglyc- erol were chemically synthesized according to our previ- ously reported method [12]. NA was purchased from Sigma (St Louis, MO, USA), and 15N-NH4Cl was pur- chased from Cambridge Isotope Laboratory (Andover, MA, USA). Nucleotides and chemically synthesized tem- plate-primers such as poly(dA), poly(rA), oligo(dT)12)18, and oligo(dT)16 were purchased from Amersham Bio- science (Uppsala, Sweden). The other reagents of ana- lytical grade were purchased from Junsei Kagaku (Tokyo, Japan).

DNA polymerase assays

Activity of pol b was measured by the methods described previously [1,23,30]. For DNA polymerases, poly(dA) ⁄ oli- go(dT)12)18 and dTTP were used as DNA template-primer

We have previously reported the interaction inter- face of lithocholic acid (LA) with the 8-kDa domain of pol b [21]. LA binds to the 8-kDa domain at helix-3 and helix-4, but not at Site I. Many other inhibitors, such as glycyrrhizic acid, bind to Site II [29]. Gly- cyrrhizic acid would compete with template DNA for binding to Site II of the 8-kDa domain. Most inhibi- tors of pol b, whose interaction interfaces are known thus far, bind competitively to the DNA binding site of the 8-kDa domain. The hydrophilic part of SQMG

FEBS Journal 272 (2005) 4349–4361 ª 2005 FEBS

4358

Expression and puriﬁcation of the mutant proteins

The expression vectors of the mutant proteins were con- structed by QuikChange II (Stratagene, La Jolla, CA, USA). The full-length rat DNA pol b mutant proteins were overexpressed in Escherichia coli strain BL21(DE3) harbor- ing the expression plasmid pET28a. After Ni-NTA column (Qiagen, Valencia, CA, USA) puriﬁcation, following the procedure recommended by the manufacturer, the mutant proteins were puriﬁed on a Hi-Trap SP-Sepharose cation exchange column (Amersham) by elution with a concentra- tion gradient of 0–1 m potassium chloride.

synthetic

(i.e.

Preparation of the isotope labeled 8-kDa domain of pol b

and nucleotide substrate, respectively. Inhibitors were dis- solved in dimethyl sulfoxide (DMSO) at various concentra- tions and sonicated for 30 s. Four lL of sonicated samples were mixed with 16 lL of each enzyme (ﬁnal 0.05 units) in 50 mm Tris ⁄ HCl (pH 7.5) containing 1 mm dithiothreitol, 50% glycerol and 0.1 mm EDTA, and kept at 0 (cid:1)C for 10 min. These inhibitor-enzyme mixtures (8 lL) were added to 16 lL of each of the enzyme standard reaction mixtures, and incubation was carried out at 37 (cid:1)C for 60 min. The activity without the inhibitor was considered to be 100%, and the remaining activities at each concentration of inhib- itor were determined as percentages of this value. One unit of each DNA polymerase activity was deﬁned as the amount of enzyme that catalyzes the incorporation of 1 nmol of deoxyribonucleotide triphosphates (i.e. dTTP) into poly(dA) ⁄ template-primers oligo(dT)12)18, A ⁄ T ¼ 2 ⁄ 1) in 60 min at 37 (cid:1)C under the normal reaction conditions [1,23].

the

Gel mobility shift assay

The gel mobility shift assay was carried out as described by Casas-Finet et al. [31]. The binding mixture (in a ﬁnal vol- ume of 20 lL) contained 20 mm Tris ⁄ HCl, pH 7.5, 40 mm KCl, 50 lgÆmL)1 bovine serum albumin (BSA), 10% DMSO, 2 mm EDTA, M13 plasmid DNA (2.2 nmol; nuc- leotide, single-strand and singly primed), and 0.15 nmol of the full-length pol b or the 8-kDa domain of pol b. Various concentrations of inhibitors were added to the binding followed by incubation at 25 (cid:1)C for 10 min. mixture, in 0.1 m Tris- Samples were run on a 1.0% agarose gel acetate buffer, pH 8.3, containing 5 mm EDTA at 50 V for 2 h.

For the NMR experiment, the 8-kDa domain of rat DNA pol b (residue 1–87) was overexpressed in Escherichia coli strain BL21(DE3) harboring expression plasmid pET21a grown on minimal media containing 15NH4Cl (0.5 gÆL)1) as the sole nitrogen source, to produce the uni- formly 15N-labeled protein. Protein expression was induced at a D600 ¼ 0.6 by addition of IPTG to a ﬁnal concentra- tion of 1 mm. The 8-kDa domain was passed through a DEAE-Sepharose column (Amersham) and then puriﬁed on a Hi-Trap SP-Sepharose column by elution with a concen- tration gradient of 0–1 m sodium chloride, and was then further puriﬁed by a Superdex 75 gel ﬁltration column (Amersham). In preparing the NMR sample, the puriﬁed 8-kDa domain was dialyzed against 10 mm Tris ⁄ HCl buffer (pH 7.0) and concentrated using an Amicon Ultra ﬁlter (Millipore). The sample for the NMR experiment contained 0.1 mm of the 15N-labeled 8-kDa domain.

In vitro DNA synthesis on poly(dA) ⁄ oligo(dT)

NMR Analysis

All NMR experiments were carried out at 25 (cid:1)C on Bruker DMX600 and DMX750 (Rheinstetten, Germany) spectro- meters. Each 1H-15N HSQC spectrum size was 1024 points in the t2 dimension and 256 points in the t1 dimension. SQMG was titrated directly into the NMR samples con- taining the 8-kDa domain of pol b at each titration point. To conﬁrm the chemical shift assignments of the 8-kDa domain, 3D CBCA(CO)NH and HNCACB spectra were recorded [33]. All the acquired data was processed using the nmrpipe and pipp programs [34,35].

(40 · 20 cm, 0.4 mm)

The titration curves were analyzed by nonlinear least-

square ﬁtting to the following equations

dob (cid:1) df ¼ ðdsat=½P(cid:2)tÞfðKD þ ½P(cid:2)t þ n½I(cid:2)tÞ

ð1Þ

(cid:1) ½ðKD þ ½P(cid:2)t þ n½I(cid:2)tÞ2 (cid:1) 4½P(cid:2)tn½I(cid:2)t(cid:2)1=2g=2

For DNA synthesis, the reaction mixture (20 lL) contained 50 mm Tris ⁄ HCl, pH 7.5, 3 mm MgCl2, 5 mm dithiothrei- tol, 10% methanol, 20 lm poly(dA) ⁄ oligo(dT)16 (¼ 2 ⁄ 1), 20 lm [32P]dTTP[aP] (60 CiÆmmol)1), and the full-length pol b or the 31-kDa domain of pol b. Various concentra- tions of SQMG were dissolved in 100% methanol and then added to the above reaction mixture, followed by incuba- tion at 37 (cid:1)C for 10 min. The products were precipitated with 100% ethyl alcohol, and then washed with 70% ethyl alcohol. Bromophenol blue was added to the precipitate, which was then loaded onto a 15% polyacrylamide-7 m urea gel in a buffer containing 6.7 mm Tris-borate, pH 7.5 and 1 mm EDTA [32]. The gel was prerun for 1 h at 2000 V, and electrophoresis was per- formed at 2000 V. After electrophoresis, the gel was dried and then exposed to imaging plates for 30 min and scanned with a Bio Imaging Analyzer BAS 2000 system (Fujiﬁlm, Tokyo, Japan).

ð2Þ

dsat ¼ db (cid:1) df

FEBS Journal 272 (2005) 4349–4361 ª 2005 FEBS

4359

N. Kasai et al. Interaction mode between SQMG and DNA Pol b

inhibitor of DNA polymerase beta and lambda. J Biol Chem 277, 630–638.

where dob is the chemical shift of the protein at each titra- tion point and df is the chemical shift of the protein in the absence of inhibitor, db is the chemical shift of the protein when fully bound by inhibitor, [P]t is the total concentra- tion of the protein, [I]t is the total concentration of inhib- itor, and n is the number of inhibitor binding sites on the protein [36,37].

6 Mizushina Y, Kamisuki S, Mizuno T, Takemura M, Asahara H, Linn S, Yamaguchi T, Matsukage A, Hanaoka F, Yoshida S, Saneyoshi M, Sugawara F & Sakaguchi K (2000) Dehydroaltenusin, a mammalian DNA polymerase alpha inhibitor. J Biol Chem 275, 33957–33961.

Structural models of the complex

7 Ohta K, Mizushina Y, Hirata N, Takemura M, Suga- wara F, Matsukage A, Yoshida S & Sakaguchi K (1998) Sulfoquinovosyldiacylglycerol, KM043, a new potent inhibitor of eukaryotic DNA polymerases and HIV-reverse transcriptase type 1 from a marine red alga, Gigartina tenella. Chem Pharm Bull (Tokyo) 46, 684–686.

8 Mizushina Y, Xu X, Asahara H, Takeuchi R, Oshige

Molecular docking of the 8-kDa domain of pol b and SQMG was performed using the affinity program within the insight ii software (Accelrys, San Diego, CA, USA). The coordinates of the 8-kDa domain of pol b (Protein Data Bank (PDB) ID: 1DK3) were obtained from PDB. The initial position of SQG was determined based on the result of chemical shift mapping. The calculation used an ESFF force ﬁeld in the discovery program and a Simula- ted Annealing method in the affinity program (Accelrys).

M, Shimazaki N, Takemura M, Yamaguchi T, Kuroda K, Linn S, Yoshida H, Koiwai O, Saneyoshi M, Suga- wara F & Sakaguchi K (2003) A sulphoquinovosyldi- acylglycerol is a DNA polymerase epsilon inhibitor. Biochem J 370, 299–305.

N. Kasai et al. Interaction mode between SQMG and DNA Pol b

Acknowledgements

9 Sahara H, Ishikawa M, Takahashi N, Ohtani S, Sato N, Gasa S, Akino T & Kikuchi K (1997) In vivo anti- tumour effect of 3¢-sulphonoquinovosyl 1¢-monoacyl- glyceride isolated from sea urchin (Strongylocentrotus intermedius) intestine. Br J Cancer 75, 324–332. 10 Ohta K, Mizushina Y, Yamazaki T, Hanashima S,

We thank Dr S. Hanashima, Dr S. Kamisuki and Dr K Matsumoto of Tokyo University of Science for helpful discussion. We also thank Dr K. Morikawa, Dr I. Ohki and Dr T. Kodama of BERI for helpful discussion, and Dr C. Shionyu of BERI for helpful support of database search.

References

1 Mizushina Y, Tanaka N, Yagi H, Kurosawa T, Onoue

Sugawara F & Sakaguchi K (2001) Speciﬁc interaction between an oligosaccharide on the tumor cell surface and the novel antitumor agents, sulfoquinovosylacylgly- cerols. Biochem Biophys Res Commun 288, 893–900. 11 Ohta K, Hanashima S, Mizushina Y, Yamazaki T, Saneyoshi M, Sugawara F & Sakaguchi K (2000) Studies on a novel DNA polymerase inhibitor group, synthetic sulfoquinovosylacylglycerols: inhibitory action on cell proliferation. Mutat Res 467, 139–152. 12 Hanashima S, Mizushina Y, Yamazaki T, Ohta K,

M, Seto H, Horie T, Aoyagi N, Yamaoka M, Matsukage A, Yoshida S & Sakaguchi K (1996) Fatty acids selectively inhibit eukaryotic DNA polymerase activities in vitro. Biochim Biophys Acta 1308, 256–262. 2 Ogawa A, Murate T, Suzuki M, Nimura Y & Yoshida S (1998) Lithocholic acid, a putative tumor promoter, inhibits mammalian DNA polymerase beta. Jpn J Cancer Res 89, 1154–1159.

Takahashi S, Sahara H, Sakaguchi K & Sugawara F (2001) Synthesis of sulfoquinovosylacylglycerols, inhibitors of eukaryotic DNA polymerase alpha and beta. Bioorg Med Chem 9, 367–376.

3 Mizushina Y, Kamisuki S, Kasai N, Ishidoh T,

Shimazaki N, Takemura M, Asahara H, Linn S, Yosh- ida S, Koiwai O, Sugawara F, Yoshida H & Sakaguchi K (2002) Petasiphenol: a DNA polymerase lambda inhi- bitor. Biochemistry 41, 14463–14471.

4 Murakami C, Ishijima K, Hirota M, Sakaguchi K,

13 Burgers PM, Koonin EV, Bruford E, Blanco L, Burtis KC, Christman MF, Copeland WC, Friedberg EC, Hanaoka F, Hinkle DC, Lawrence CW, Nakanishi M, Ohmori H, Prakash L, Prakash S, Reynaud CA, Sugino A, Todo T, Wang Z, Weill JC & Woodgate R (2001) Eukaryotic DNA polymerases: proposal for a revised nomenclature. J Biol Chem 276, 43487–43490.

Yoshida H & Mizushina Y (2002) Novel anti-inﬂamma- tory compounds from Rubus sieboldii, triterpenoids, are inhibitors of mammalian DNA polymerases. Biochim Biophys Acta 1596, 193–200.

5 Mizushina Y, Kamisuki S, Kasai N, Shimazaki N,

14 Aravind L & Koonin EV (1999) DNA polymerase beta- like nucleotidyltransferase superfamily: identiﬁcation of three new families, classiﬁcation and evolutionary his- tory. Nucleic Acids Res 27, 1609–1618.

15 Kornberg A & Baker T (1992) DNA Replication, 2nd

edn, pp. 197–225. W.H. Freeman, New York, USA.

Takemura M, Asahara H, Linn S, Yoshida S, Matsuk- age A, Koiwai O, Sugawara F, Yoshida H & Sakaguchi K (2002) A plant phytotoxin, solanapyrone A, is an

FEBS Journal 272 (2005) 4349–4361 ª 2005 FEBS

4360

16 Sawaya MR, Pelletier H, Kumar A, Wilson SH &

26 Lepre CA, Moore JM & Peng JW (2004) Theory and

applications of NMR-based screening in pharmaceutical research. Chem Rev 104, 3641–3676.

Kraut J (1994) Crystal structure of rat DNA polymerase beta: evidence for a common polymerase mechanism. Science 264, 1930–1935.

17 Maciejewski MW, Liu D, Prasad R, Wilson SH &

27 Shionyu Mitsuyama C, Shirai T, Ishida H & Yamane T (2003) An empirical approach for structure-based pre- diction of carbohydrate-binding sites on proteins. Protein Eng 16, 467–478.

28 Prasad R, Beard WA, Chyan JY, Maciejewski MW,

Mullen GP (2000) Backbone dynamics and reﬁned solu- tion structure of the N-terminal domain of DNA poly- merase beta. Correlation with DNA binding and dRP lyase activity. J Mol Biol 296, 229–253.

Mullen GP & Wilson SH (1998) Functional analysis of the amino-terminal 8-kDa domain of DNA polymerase beta as revealed by site-directed mutagenesis. DNA binding and 5¢-deoxyribose phosphate lyase activities. J Biol Chem 273, 11121–11126.

18 Sawaya MR, Prasad R, Wilson SH, Kraut J & Pelletier H (1997) Crystal structures of human DNA polymerase beta complexed with gapped and nicked DNA: evidence for an induced ﬁt mechanism. Biochemistry 36, 11205– 11215.

29 Hu HY, Horton JK, Gryk MR, Prasad R, Naron JM, Sun DA, Hecht SM, Wilson SH & Mullen GP (2004) Identiﬁcation of small molecule synthetic inhibitors of DNA polymerase beta by NMR chemical shift map- ping. J Biol Chem 279, 39736–39744.

19 Pelletier H, Sawaya MR, Wolﬂe W, Wilson SH & Kraut J (1996) Crystal structures of human DNA polymerase beta complexed with DNA: implications for catalytic mechanism, processivity, and ﬁdelity. Biochem- istry 35, 12742–12761.

20 Mizushina Y, Ohkubo T, Date T, Yamaguchi T,

Saneyoshi M, Sugawara F & Sakaguchi K (1999) Mode analysis of a fatty acid molecule binding to the N-term- inal 8-kDa domain of DNA polymerase beta. A 1: 1 complex and binding surface. J Biol Chem 274, 25599– 25607.

30 Mizushina Y, Yagi H, Tanaka N, Kurosawa T, Seto H, Katsumi K, Onoue M, Ishida H, Iseki A, Nara T, Mor- ohashi K, Horie T, Onomura Y, Narusawa M, Aoyagi N, Takami K, Yamaoka M, Inoue Y, Matsukage A, Yoshida S & Sakaguchi K (1996) Screening of inhibitor of eukaryotic DNA polymerases produced by microor- ganisms. J Antibiot (Tokyo) 49, 491–492.

31 Casas Finet JR, Kumar A, Morris G, Wilson SH &

21 Mizushina Y, Ohkubo T, Sugawara F & Sakaguchi K (2000) Structure of lithocholic acid binding to the N-terminal 8-kDa domain of DNA polymerase beta. Biochemistry 39, 12606–12613.

Karpel RL (1991) Spectroscopic studies of the structural domains of mammalian DNA beta-polymerase. J Biol Chem 266, 19618–19625.

22 Hanashima S, Mizushina Y, Ohta K, Yamazaki T,

32 Detera SD & Wilson SH (1982) Studies on the mechan- ism of Escherichia coli DNA polymerase I large frag- ment. Chain termination and modulation by polynucleotides. J Biol Chem 257, 9770–9780.

Sugawara F & Sakaguchi K (2000) Structure-activity relationship of a novel group of mammalian DNA poly- merase inhibitors, synthetic sulfoquinovosylacylglycer- ols. Jpn J Cancer Res 91, 1073–1083.

33 Cavanagh J, Fairbrother WJ, Palmer AG III & Skelton NJ (1996) Protein NMR Spectroscopy: Principles and Practice. Academic Press, Inc, CA, USA.

23 Mizushina Y, Yoshida S, Matsukage A & Sakaguchi K (1997) The inhibitory action of fatty acids on DNA polymerase beta. Biochim Biophys Acta 1336, 509–521.

24 Abbotts J, SenGupta DN, Zmudzka B, Widen SG,

34 Delaglio F, Grzesiek S, Vuister GW, Zhu G, Pfeifer J & Bax A (1995) NMRPipe: a multidimensional spectral processing system based on UNIX pipes. J Biomol NMR 6, 277–293.

35 Garrett DS, Powers R, Gronenborn AM & Clore GM

Notario V & Wilson SH (1988) Expression of human DNA polymerase beta in Escherichia coli and character- ization of the recombinant enzyme. Biochemistry 27, 901–909.

25 Matsumoto K, Takenouchi M, Ohta K, Ohta Y, Imura

(1991) A common sense approach to peak picking two-, three- and four-dimensional spectra using automatic computer analysis of contour diagrams. J Magn Reson 95, 214–220.

36 Cantor CR & Schimmel PR (1980) Biophysical Chemis- try, Part III: the Behavior of Biological Macromolecules. W.H. Freeman, NY, USA.

37 Freire E, Mayorga OL & Straume M (1990) Isothermal

T, Oshige M, Yamamoto Y, Sahara H, Sakai H, Abe M, Sugawara F, Sato N & Sakaguchi K (2004) Design of vesicles of 1,2-di-O-acyl-3-O-(beta-D-sulfoquinovosyl) -glyceride bearing two stearic acids (beta-SQDG-C18), a novel immunosuppressive drug. Biochem Pharmacol 68, 2379–2386.

titration calorimetry. Anal Chem 10, 2359–2367.

FEBS Journal 272 (2005) 4349–4361 ª 2005 FEBS

4361

N. Kasai et al. Interaction mode between SQMG and DNA Pol b