Identification of the Archaeoglobus fulgidus endonuclease III DNA interaction surface using heteronuclear NMR methods

Research Article

919

Identification of the Archaeoglobus fulgidus endonuclease III DNA interaction surface using heteronuclear NMR methods Alexander Shekhtman1, Lynn McNaughton2, Richard P Cunningham3 and Susan M Baxter2* Background: Endonuclease III is the prototype for a family of DNA-repair enzymes that recognize and remove damaged and mismatched bases from DNA via cleavage of the N-glycosidic bond. Crystal structures for endonuclease III, which removes damaged pyrimidines, and MutY, which removes mismatched adenines, show a highly conserved structure. Although there are several models for DNA binding by this family of enzymes, no experimental structures with bound DNA exist for any member of the family. Results: Nuclear magnetic resonance (NMR) spectroscopy chemical-shift perturbation of backbone nuclei (1H, 15N, 13CO) has been used to map the DNA-binding site on Archaeoglobus fulgidus endonuclease III. The experimentally determined interaction surface includes five structural elements: the helix-hairpin-helix (HhH) motif, the iron–sulfur cluster loop (FCL) motif, the pseudo helix-hairpin-helix motif, the helix B–helix C loop, and helix H. The elements form a continuous surface that spans the active site of the enzyme.

Addresses: 1Department of Physics, University at Albany, SUNY, 1400 Washington Avenue, Albany, NY 12222, USA, 2Wadsworth Center and Department of Biomedical Sciences, University at Albany, Empire State Plaza, PO Box 509, Albany, NY 12201-0509, USA and 3Department of Biological Sciences, University at Albany, SUNY, 1400 Washington Avenue, Albany, NY 12222, USA. *Corresponding author. E-mail: [email protected] Key words: endonuclease, NMR, protein–DNA interactions Received: 12 February 1999 Revisions requested: 30 March 1999 Revisions received: 12 April 1999 Accepted: 21 April 1999 Published: 15 July 1999

Conclusions: The enzyme–DNA interaction surface for endonuclease III contains five elements of the protein structure and suggests that DNA damage recognition may require several specific interactions between the enzyme and the DNA substrate. Because the target DNA used in this study contained a generic apurinic/apyrimidinic (AP) site, the binding interactions we observed for A. fulgidus endonuclease III should apply to all members of the endonuclease III family and several interactions could apply to the endonuclease III/AlkA (3-methyladenine DNA glycosylase) superfamily.

Introduction A large number of damaged or mismatched bases in DNA are recognized and removed by the action of DNA N-glycosylases [1–4]. These enzymes cleave the N-glycosidic bond of the target base to create an apurinic/apyrimidinic (AP) site that is subsequently processed by the base excision repair pathway to restore the DNA to its original sequence. Endonuclease III was originally described as an enzyme activity from Escherichia coli that could nick heavily irradiated DNA [5,6]. Subsequently, it was shown that the enzyme removed ring-saturated, ring-opened and ring-rearranged pyrimidines via an N-glycosylase activity [7] and then cleaved the phosphodiester backbone at the resulting AP site via a β-elimination event [8–10]. Such a cleaved AP site can also be processed by the base excision repair pathway [11]. The nth gene of E. coli was cloned [12] and sequenced and endonuclease III was expressed and purified [13] for biochemical and biophysical analysis. Endonuclease III is a monomeric protein that contains a [4Fe–4S] cluster in the +2 oxidation state [14]. The crystal structure of E. coli

Structure August 1999, 7:919–930 http://biomednet.com/elecref/0969212600700919 © Elsevier Science Ltd ISSN 0969-2126

(Eco) endonuclease III has been solved [15,16]. The protein is elongated and contains two domains of equal size that are separated by a deep cleft. Highly conserved residues, including aspartic acid 138, flank the cleft, suggesting that it constitutes the catalytic site of the enzyme [16]. The two defining structural features of this enzyme are a helix-hairpin-helix (HhH) motif and an iron–sulfur cluster loop (FCL) motif [16]. The HhH motif contains a characteristic sequence, LPGVG (single-letter amino acid code), that is highly conserved among family members. Thayer and coworkers [16] have proposed that the HhH motif nonspecifically binds the sugar–phosphate backbone of substrate DNA. The [4Fe–4S] cluster, which is not redox active [17], is not near the active site of the protein. It has been suggested that the cluster serves as an architectural element that positions a loop containing positively charged amino acids near the phosphodiester backbone of a target DNA molecule [16]. Remarkably, a number of other glycosylases have been described that have a very similar amino acid sequence to endonuclease III [2,4]. All these enzymes contain the FCL

920

Structure 1999, Vol 7 No 8

motif, the HhH motif, and an aspartic acid residue at the active site. The endonuclease III family of glycosylases includes MutY protein, which removes adenines from A•G and A•8-oxoguanine mismatches [18–23], thymine DNA mismatch glycosylase (TDG), which removes the pyrimidine from T•G and U•G mismatches [24,25], and the UV endonuclease from Micrococcus luteus, which cleaves the N-glycosidic bond of the 5′ pyrimidine of a cyclobutane dimer [26]. Endonuclease III is broadly distributed in phylogeny from bacteria to archaea to eukarya and has been found in almost every genome sequenced to date. The human homolog of endonuclease III has been described and the gene has been cloned [27]. MutY is also found from bacteria to humans. Other members of the endonuclease III family have been found only in the bacteria and the archaea to date.

Results and discussion

When the crystal structure of 3-methyladenine DNA glycosylase (AlkA) was solved [28], it was found to have considerable structural homology to endonuclease III, including the HhH motif and the active-site aspartic acid, but not the FCL motif. The gene for yeast 8-oxoguanine DNA glycosylase (yOGG1) was cloned [29,30] and it showed sequence similarity to AlkA, including the HhH motif and the active site, suggesting that there is a superfamily of DNA glycosylases including the endonuclease III family and the AlkA family [29,31]. No crystal structure of any member of the endonuclease III/AlkA superfamily bound to target DNA has been reported. The predicted interactions of these enzymes with DNA have been based on adeninebinding studies [32], electrostatics, sequence alignment, and several site-directed mutants [16].

Backbone resonance assignments were made using [U-15N, 13C], [U-15N], and a series of specifically 15N-labeled endonuclease III samples. Sequential assignments were largely achieved using HNCA/HNCOCA [36–38] and HNCO/HNCACO [36,37,39] pairs of experiments so that connectivities could be traced through two independent, through-bond pathways [40]. Additionally, these triple-resonance experiments were chosen to ensure good sensitivity so that the majority of expected correlations were observed. The completeness of these sequential pathways is indicated in a connectivity diagram shown in Figure 2. The HNCACB [41] experiment largely provided verification of amino acid type, based on 13Cα and 13Cβ chemical shifts, as few sequential carbon resonances were detected because of the effective size of the protonated protein. Samples specifically labeled with 15N-lysine, 15N-leucine, and 15N-valine provided anchor points in the sequence. Adding glycine, threonine, serine, and alanine assignments, made on the basis of their unique 13Cα and 13Cβ shifts, 95 amino acids out of 209 were assigned by amino acid type. These assignments strongly facilitated the sequential assignment procedure, eliminating the majority of chemical-shift degeneracies characteristic of α-helical proteins. It was also particularly useful having amino acid type assignments to correctly place assigned sequences, as Afu endonuclease III contains 14 prolines, which interrupt sequential assignments based on amide resonances. Finally, [1H–15N] NOESYHSQC (nuclear Overhauser effect spectroscopy heteronuclear single quantum coherence) [42,43] experiments allowed tracing of sequential amide–amide connectivities through helices and several turn structures.

Previously we used oligonucleotides containing a reduced AP site as a noncleavable substrate for footprinting, electrophoretic mobility shift analysis [33], and fluorescence quenching studies [34]. Endonuclease III binds tightly and specifically to this substrate. We and others [25,35] have shown that TDG and MutY bind tightly to product AP sites as well, confirming that the enzymes in this family recognize oligonucleotides containing AP sites. Here we report heteronuclear nuclear magnetic resonance (NMR) spectroscopy data that describes the DNAbinding face of Archaeoglobus fulgidus (Afu) endonuclease III. Using changes in the NMR chemical shifts of backbone amide nuclei (1H, 15N, and 13CO), we mapped structural elements of endonuclease III perturbed upon binding of a double-stranded oligonucleotide containing a reduced AP site (rAP 13-mer). Because the target DNA used in this study contained a generic AP site, the binding interactions described here should apply to all members of the endonuclease III family, and several interactions could apply to the endonuclease III/AlkA superfamily.

Sequence-specific backbone resonance assignments for free endonuclease III

Afu endonuclease III was overexpressed in E. coli using a T7 expression system. The purified protein has a primary sequence of 209 amino acids, including 14 prolines. Mass spectroscopy confirmed the presence of a covalently bound [4Fe–4S] cluster (molecular weight = 24,018 Da). This relatively large protein was well behaved in solution at pH 6.5 and at a range of temperatures from 10° to 45°C. Endonuclease III samples were long-lived in solution when dithiothreitol (DTT) was included in the buffer and when the sample was concentrated and sealed in the NMR tube under inert conditions. Endonuclease III exhibited good chemical-shift dispersion as shown in an 1H–15N correlation spectrum (Figure 1).

Overall, 91% of the backbone 1H, 13Cα, 13CO, and 15N backbone resonances were sequentially assigned. The incompleteness of the assignments can be traced to resonance-overlap problems as a result of the large number of signals, low signal-to-noise ratios due to amide proton exchange processes in loop structures, and the presence of

Research Article Endonuclease III–DNA interactions Shekhtman et al.

Figure 1

prolines that interrupted sequential connectivities. Despite these gaps in backbone assignments, the majority of backbone resonance assignments are complete for Afu endonuclease III, providing site-specific reporters for DNA recognition and binding events upon complex formation.

(a) K209

S104 G182

921

V114 G118

I78

111

Secondary structure of Afu endonuclease III T155 I28

G80 S41 G120

V81 T140

K59

I200 E111 N125

S123 S103

E72

C195

Q185

R17

D139

V21 D101 L98

Q54

I4

K96 R12

R166 L67

V36

Y102

K115

L69

K79

V208

V206 K156

L94

123

V86

A19

A14

K89

E162

E47

A152 L77 K192 K97 E100 Q33 H34 N55 R90 D65 T49 Y22 L66 K201 E174 V138 T160 E11 K62 I15 E8 V163 A146 K92 R144 R51 R121 L91 E159 A75 E13 L167 A58 M10 I145 A37 E5

E71

K68

Y84

H143

L40

F57 S70

E112

N147

F168

S42 K18

N (ppm)

F32

G207

E27

15

T154

S109

E75

117

G184

I119

E105

L108 K16

A124 R153

A135

F183 L194

129

K175

T186 G150

9.3

8.1 1 H (ppm)

6.9

(b) R144 V6

E8 V127

A26

I74 K87 A38

A95

L56

A26

D46

I136

W151

L39 K25

L39

V9

L35 L24

E73

A52

A48

L116

V50 I133 A88

V176

V126 K189

122.2

15

E161

N (ppm)

V99

A38

121.3

I7

V60

120.4

H23

L26

8.2

8.0 H (ppm)

7.8

1

Structure

Fingerprint HSQC spectrum for free endonuclease III. (a) Complete [1H–15N]-HSQC heteronuclear correlation spectrum of [U-15N] Afu endonuclease III and (b) expanded view of the crowded region in the center of the HSQC spectrum. Gray crosspeaks are resonances folded in the nitrogen dimension. Resonances are labeled with assignments. Gradient-based WATERGATE water suppression [70] was incorporated into the pulse sequence. The spectrum was taken at 30°C on a 1 mM endonuclease III sample dissolved in 95% H2O/5% D2O, 50 mM potassium phosphate buffer, pH 6.6, 400 mM KCl, 1 mM DTT and 0.02% NaN3 under argon atmosphere. The HSQC spectrum resolves 167 backbone resonances out of 194 expected HSQC resonances, excluding prolines and the N-terminal methionine.

The secondary structure of Afu endonuclease III, shown in Figure 2, is based on sequential NOE patterns and 13C chemical-shift indices (CSI) [44,45]. α Helices were identified on the basis of continuous stretches of positive consensus CSI values and amide–amide connectivities. Using these criteria, the secondary structure pattern of Afu endonuclease III closely matched the α-helical structure of Eco endonuclease III, for which a high-resolution crystal structure is available [16,46]. The primary sequences of Afu and Eco endonuclease III, shown in Figure 3, are highly similar, with 37% sequence identity. The Afu enzyme contains the characteristic set of identical amino acids conserved among all endonuclease III family members. The secondary structure data from the NMR studies, in combination with the sequence alignment, leads us to propose that the overall fold of Afu endonuclease III is similar to the Eco endonuclease III structure, retaining all features characteristic of this family of DNA-repair enzymes. Through-bond connectivites were incomplete in two regions, Arg43–Arg45 and Gly82–Phe83, due to broadened and weakened signals. These resonances are contained in predicted turn and loop regions. The turn between helix B and helix C, containing Arg43–Arg45, has been implicated in minor groove contact with substrate DNA [32]. Gly82, contained in a predicted hairpin loop, is highly conserved and is thought to be involved in DNA backbone recognition [32]. The NMR data indicated that these residues experience enhanced exchange with water or exist in several conformations, typical dynamic behavior observed for amides contained in loops and turns. Additionally, flexibility for both regions may imply an inducedfit recognition mechanism for DNA docking. Chemical-shift changes observed upon formation of the endonuclease III–DNA complex

To characterize the endonuclease III–DNA complex, an oligonucleotide duplex containing a reduced AP site was used as a model substrate. Reduced AP sites are not cleaved by endonuclease III because the β-elimination reaction cannot take place if the reactive open-chain aldehyde is reduced to the alcohol. Endonuclease III family members generate product AP sites, suggesting that the AP site is a generic recognition target. Indeed, O’Handley and coworkers [33] used quantitative gel-shift assays to demonstrate that Eco endonuclease III specifically binds rAP-containing (reduced AP site containing) oligonucleotides. The complex between Afu endonclease III and a 13-mer oligonucleotide containing a reduced AP site

922

Structure 1999, Vol 7 No 8

Figure 2

αA

C αi 13 α C i–1 13 COαi 13 COαi–1

αB

αC

M D P I EV I EV M E R E A I K R K A PVY H L K A E I KT P F Q H LV A A L L S S RT R D E ATV R A A Q N L F A KV 10 20 30 40 50 60

13

dNN +1 CSI –1 αD

C αi C αi–1 13 COαi 13 COαi–1

αE

αF

K K P E D L L K L S E E E I A E L I K GV G FY RV K A K R L K E L A K K LV E DY S S EV P L S F E E LV K L P G I G 70 80 90 100 110 120

13

13

dNN +1 CSI –1 αG

Cαi Cαi–1 13 CO αi 13 CO αi–1

αH

αI

αJ

R K S A NVV L AY S D I P A I PV DT HV H R I A N R L GW A RTT K P E ET E EV L K R L F P L E FW E KV N R A M 130 140 150 160 170 180

13

13

Summary of the through-bond connectivities, sequential amide–amide NOEs, and chemicalshift indices (CSI) identified for assignment of backbone resonances for Afu endonuclease III. Along the top, the location of the helices in Eco endonuclease III is indicated above the complete sequence of Afu endonuclease III, shown in one-lettter amino acid code. Four correlations, between the 15Ni nucleus and 13C nuclei in amino acids i and i–1, provided the basis for sequential assignments. Through-bond correlations involving 15Ni were identified in the following experiments: 13Cαi, HNCA; 13Cαi–1, HN(CO)CA and HNCA; 13CO , HN(CA)CO; 13CO , HNCO. Solid i i–1 bars in each row indicate the presence of a crosspeak in the spectra for these correlations. The relative intensity of amide–amide NOE correlations is indicated by the height of the bar connecting residues i and i + 1 in the row labeled dNN. The consensus chemical-shift indices are based on the changes from random-coil values for both 13Cα and 13CO. These values are contained in the bottom row of the figure labeled CSI. A positive value is indicative of a helical backbone conformation.

dNN +1 CSI –1 αJ

Cαi Cαi–1 13 COαi 13 α C i–1

VGFGQTVCKPQKPLCDECPIKGCPRVGVK 190 200

13 13

dNN +1 CSI –1

(rAP 13-mer) purified as a single, sharp peak by gel-filtration chromatography, suggesting a long-lived and tightly associated complex. As predicted from gel-filtration studies, the Afu endonuclease III–rAP DNA complex showed slow-exchange behavior on an NMR timescale. At 1:1 stoichiometry between the protein and the DNA substrate, many peaks in the [1H–15N] HSQC spectrum corresponding to free endonuclease III disappeared. There was no evidence of resonances from the free form of the protein in the complex spectra. The amide-resonance line widths broadened significantly in the spectrum of the complex. Resonance line widths for free endonuclease III were approximately 22 Hz in the proton dimension and 28 Hz in the proton dimension of the bound-protein spectrum, reflecting the expected effective increase in molecular weight upon complex formation between endonuclease

Structure

III (24,290 Da for the [U-15N] protein) and the rAP oligonucleotide (7604 Da). Additionally, the number of resonances observed in the complex [1H–15N] HSQC spectrum was similar to the number of resonances found in the free endonuclease III spectrum, consistent with a single form of the protein–DNA complex in solution. There were three types of resonance changes observed in the [1H–15N] HSQC spectra upon complex formation (Figure 4): resonances that did not shift significantly, but broadened; resonances that underwent significant proton chemical-shift changes (> 0.1 ppm); and peaks that shifted more than 0.5 ppm upon complex formation. Approximately half of all amide resonances belonged to the first type. The remainder shifted significantly upon DNA binding. However, the [1H–15N] HSQC spectra of both free and DNA-bound endonuclease III contained significant overlap in the spectrum due to the large number of

Research Article Endonuclease III–DNA interactions Shekhtman et al.

923

Figure 3 Sequence alignment of four proteins of the endonuclease III family: Afu endonuclease III, Eco endonuclease III, Eco MutY, and Mth thymine DNA mismatch glycosylase (TDG). Single-letter amino acid codes are used. Amino acids completely conserved among endonuclease III protein family members are in red and similar residues are in blue. Diamonds mark well-conserved positions in the sequence that coincide with structural elements and sequence motifs identified in the Eco endonuclease III and MutY crystal structures. These include the helix B–helix C loop (Leu40–Asp46), pseudo helix-hairpinhelix (pHhH) (Ile78–Phe83), helix-hairpin-helix (Leu113–Leu128) motif, helix H (Thr140–Leu49), and the FCL (Cys189–Cys203) motif.

♦

-------MDPIEVIEVMEREAIK-RK-APVYHLKAEIKTPFQHLVAALLSSRTRDEATVRAAQNLFAKVK -------MNKAKRLEILTRLREN-NP-HPTTELN--FSSPFELLIAVLLSAQATDVSVNKATAKLYPVAN ------MQASQFSAQVLDWYDKYGRKTLPWQIDK----TPYKVWLSEVMLQQTQVATVIPYFERFMARFP MDDATNKKRKVFVSTILTFWNTD-RRDFPWRHTR--D--PYVILITEILLRRTTAGHVKKIYDKFFVKYK ♦

♦

♦

KPEDLLKLSEEEIAELIKGVGFYRVKAKRLKELAKKLVEDYSSEVPLSFEELVKLPGIGRKSANVVLAYS TPAAMLELGVEGVKTYIKTIGLYNSKAENIIKTCRILLEQHNGEVPEDRAALEALPGVGRKTANVVLNTA TVTDLANAPLDEVLHLWTGLGYY-ARARNLHKAAQQVATLHGGKFPETFEEVAALPGVGRSTAGAILSLS CFEDILKTPKSEIAKDIKEIGLSNQRAEQLKELARVVINDYGGRVPRNRKAILDLPGVGKYTCAAVMCLA ♦

61 59 60 65

131 129 129 135

♦

♦

-DIPAIPVDTHVHRIANR---LGWARTTKP----EETEEVLKRLFPLEFWEKVNRAMVGFGQTVCKPQKP FGWPTIAVDTHIFRVCNR---TQFAPGKNV----EQVEEKLLKVVPAEFKVDCHHWLILHGRYTCIARKP LGKHFPILDGNVKRVLARCYAVSGWPGKKE—-VENKLWSLSEQVTPAVGVERFNQAMMDLGAMICTRSKP FGKKAAMVDANFVRVINR---YFGGSYENLNYNHKALWELAETLVPGGKCRDFNLGLMDFSAIICAPRKP

LCDECPIK-GCPRVGVK RCGSCIIEDLCEYKEKVDI KCSLCPLQNGCIAAANNSW… KCEKCGMSKLCSYYEKCST

193 192 197 202

Afu Endonuclease III Eco Endonuclease III Eco MutY Mth TDG Structure

peaks and helical nature of the protein. Consequently, analysis of the endonuclease III–DNA complex required [U-15N, 13C] protein and HNCO data sets. Pattern matching and average chemical shifts calculated for bound endonuclease III

The HNCO experiment resolved 190 and 193 resonances (out of 194 expected) for free and DNA-bound endonuclease III, respectively. For each peak in the free HNCO spectrum we used a basic pattern-matching scheme described previously [47,48] to assign resonances in the bound endonuclease III spectrum and to calculate a minimum ppm distance, min(∆ppm), or average shift. The minimum ppm distance was normalized to the proton chemical-shift scale, resulting in a qualitative measure of average chemical-shift changes. We conservatively restricted our search for a corresponding peak in the bound spectrum within a min(∆ppm) value less than 1.0 ppm from the free HNCO resonance. Where large chemical-shift changes prevented conclusive assignments in the bound HNCO spectrum, residues are reported as having min(∆ppm) of 1 ppm. The excellent sensitivity and resolution of the HNCO experiments allowed us to relate 73% of the assigned resonances of the free protein to resonances in the bound spectrum. The remainder of the assigned resonances were perturbed so greatly that conclusive assignments could not be made using only HNCO data. Figure 5 reports the min(∆ppm) calculated for assigned residues in Afu endonuclease III. About 50% of the amino acid residues experienced significant chemical-shift perturbations (> 0.3 min∆ppm) upon DNA binding. The same pattern of significant chemical-shift changes plotted against amino

acid sequence can be seen if changes in 1H, 15N, and 13CO chemical shifts are plotted separately (data not shown). In particular, five stretches of the sequence are strikingly affected by DNA binding: Ala38–Glu47, Tyr84–Ala88, Gly118–Ser123, Val138–Ala146, and Phe183–Lys201. The min(∆ppm) values for the assigned residues in these regions exceeded 0.7 ppm or had shifted so greatly that they could not be assigned in the bound spectrum of endonuclease III. Chemical-shift perturbation can be interpreted in at least two ways [49–53]. The exquisite sensitivity of the chemical-shift may reflect changes in the magnetic environment affected by a conformational change in the protein. Alternatively, chemical-shift changes may highlight the interface between interacting molecules. The chemical-shift changes reported here describe the effects of rAP-DNA binding on the backbone of endoncuclease III — some changes were due to direct DNA contact and some were due to changes in the protein conformation distant from the protein–DNA interface. It is most likely that direct contacts between the protein and DNA are mediated by both backbone and sidechain contacts, so the absence of chemical-shift changes for backbone nuclei does not necessarily rule out involvement of a particular sidechain in DNA contact. Nonetheless, the qualitative interaction map generated on the basis of chemical-shift data was remarkably revealing of interactions between endonuclease III and a generic DNA substrate. To analyze the min(∆ppm) values further, we mapped the changes onto a model Afu endonclease III structure. The endonuclease III structure is highly conserved among DNA glycosylase superfamily members. Using homology modeling and energy minimization, a model of Afu endonuclease III was generated on the basis of the

924

Structure 1999, Vol 7 No 8

Figure 4 (a) S104 K209

V114

G182

I78

I28

111

T155

G80 V81 S123

G184

E75

N125 F168 E27 L98

D101

F57 E174 K201

S70

L91

E47

R12 V36

E11

I145 A14

N55 I15

L77 T49 T160

Q33

A152 K97 D65

Y102 Y22

L66

E8 V163

E13

K92

R51

L167

E5

V208 K79

I119 K156

A124

E105

V86

K16

L108

R153

K115

A58

L39

V206

L94

L69

E159

A26

A75 A37 E161 A19

E71

K68

E162

A146

M10

L67

K89

K96

H34 R90

K62

K59 E112 S109 R166

R17

I4

E72 E100

N147

Q54

123

G207

V21

(ppm)

F32

K18

117

S103

T154

15 N

E111

A135

129

K175

T186

G150

9.3

8.1 1 H (ppm)

6.9

(b) V6

I7

A38

W151 V9

A48 K25

V50 A88

V126

L26

V176

8.2

A52 L116

L24

E161

8.0 H (ppm)

I133

121.3

L35

E73

I136

122.2

A26 L56 A26

A95

N (ppm)

V127

I74

15

H23 K87

120.4

E8 V60 V99

7.8

1

Structure

Fingerprint HSQC spectrum for DNA-bound endonuclease III. (a) Complete [1H–15N]-HSQC heteronuclear correlation spectrum of [U-15N] Afu endonuclease III bound to rAP ds13-mer (1:1 complex) and (b) expanded view of the crowded region in the center of the HSQC spectrum. Gray crosspeaks are resonances folded in the nitrogen dimension. Resonances are labeled with assignments. The spectrum was collected at 30ºC on complex dissolved in 50 mM potassium phosphate buffer, pH 6.6, 400 mM KCl, 1 mM DTT and 0.01% NaN3. Resonances are labeled with assignments. Several types of resonance changes due to complex formation can be observed in the spectrum: K68 (6.9 x 115.8 ppm) did not shift significantly; K89 (6.8 x 116.1 ppm) shifted significantly in the proton dimension ( > 0.1 ppm); and N147 (7.7 x 115.0 ppm) shifted more than 0.5 ppm in the proton dimension upon complex formation.

Eco endonuclease III structure, the alignment of helices determined by NMR methods, and the alignment of conserved amino acids. Figure 6 demonstrates that all amino acids greatly affected by DNA binding (min(∆ppm) > 0.5 ppm) are located on one face of the protein, evidence of a unique, contiguous binding site for DNA substrate. For consistency, the helices are labeled A–J, as denoted for the Eco endonuclease III structure. The interaction map suggests that the chemical-shift perturbations largely reflect a contact surface, not a global conformational change. The resulting chemical-shift map of endonuclease III bound to DNA defines not only an interaction surface but also highlights five regions of the protein that are greatly affected by DNA binding. Residues in the hairpin loop of the HhH motif (Gly118–Ser123), in helix J and the FCL motif (Phe183–Ile200), in the helix B-loop-helix C region (Ala38–Glu47), in the N-terminal portion of helix E (Tyr84–Ala88), and in helix H (Val138–Ala146) show dramatic chemical-shift changes upon endonuclease III–DNA complex formation. Tainer and coworkers pointed out characteristic and conserved structural elements in the Eco endonuclease III [16] and MutY [32] crystal structures that led to a set of testable hypotheses regarding the endonuclease III family–DNA interface. Thayer and coworkers first noted the helix-hairpin-helix (HhH) motif formed by helices F and G in structural studies of Eco endonuclease III. On the basis of high sequence conservation and their crystal structure, they proposed that glycine amide protons in the hairpin loop interact nonspecifically with DNA phosphates. Gly118 and Gly120, the homologous glycines in Afu endonuclease III, each experienced dramatic chemical-shift changes upon DNA binding, further implicating these HhH residues in DNA binding. Additionally, in the free form of the protein, these glycine amide resonances were significantly broadened, suggesting that in solution they are exposed to solvent or contained in a flexible loop. Interestingly, the chemical-shift changes upon DNA binding do not encompass the entire lengths of helix F and helix G, supporting the idea that the helices position loop residues for DNA interactions. Indeed, the chemicalshift perturbation was not symmetric across the HhH element. DNA binding primarily affected residues in the hairpin loop and the N-terminal end of helix G of the HhH element, which contains Lys120, the active-site nucleophile for endonuclease III [16]. Thayer and coworkers [16] also highlighted the FCL sequence motif in the Eco endonuclease III structure. The FCL motif is contained in a solvent-exposed loop between the first two [4Fe–4S] cluster cysteine ligands, and forms a structure similar to zinc-binding motifs found in DNA-binding proteins like Gal4. In Eco endonuclease III this loop contains charged residues that could contact

Research Article Endonuclease III–DNA interactions Shekhtman et al.

925

Figure 5 pHhH

B–C loop

1.0

HhH

αH

FCL

0.9 0.8 0.7 min(∆ppm)

Weighted minimal chemical-shift differences, min(∆ppm), for backbone nuclei of Afu endonuclease III upon DNA binding plotted against residue number. The location of characteristic structural elements and sequence motifs of the endonuclease III protein family are noted at the top of the graph. The average min(∆ppm) value (0.32 ppm) is shown by the dashed line. Gray bars with values of 0.02 ppm represent residues that are not significantly perturbed upon rAP-oligonucleotide binding. Blanks and gaps in the graph represent prolines and unassigned residues in the endonuclease III sequence.

0.6 0.5 0.4 0.3 0.2 0.1 0.0 1

10

20

30

40

50

60

70

80

90 100 110 120 130 140 150 160 170 180 190 200

Residue number

the DNA backbone. Specifically, a conserved lysine at position 191 in Eco endonuclease III, contained in the FCL motif, has been implicated in DNA binding by mutagenesis studies. The chemical shift interaction map shows that a stretch of residues in the C-terminal end of helix J and the FCL loop are affected by DNA binding. The conserved lysine, Lys192 in Afu endonuclease III, and residues flanking it all shift more than 0.5 min(∆ppm) Figure 6 Illustration of the homology modeled structure of Afu endonuclease III showing DNA-induced chemical-shift perturbations along the protein backbone. The backbone tube representation was created using the program SETOR [71]. Two views of the structure are shown, related by a 180° rotation about the vertical axis. (a) The front of endonuclease III, largely encompassing the interaction surface, and (b) the back of the protein. The color ramps have been selected so that residues with minimal chemical-shift perturbation less than the average, (min(∆ppm) ≤ 0.32 ppm, are colored blue; residues with perturbation between min(∆ppm) = 0.32 ppm and one standard deviation from the corresponding mean value (min(∆ppm) = 0.52 ppm) are colored green; and residues with min(∆ppm) ≥ 0.52 ppm are colored red. Gray indicates prolines and unassigned residues. Yellow and red spheres represent sulfur and iron, respectively, in the characteristic [4Fe–4S] cluster.

Structure

upon DNA binding. Surrounding residues in Afu endonuclease III differ significantly from the Eco sequence, however. Arginines at positions 190 and 193 are replaced with glutamines in the Afu structure. Afu endonuclease III contains an arginine at position 194, replacing a leucine in the Eco sequence. However, interactions mediated by sidechain atoms, whether electrostatic or hydrogenbonding in nature, may cause a change in loop structure

926

Structure 1999, Vol 7 No 8

that is transmitted to helix J. The sequence similarities, conservation of the [4Fe–4S] cluster, and NMR data showing large chemical-shift changes, support the importance of the FCL loop in DNA binding and recognition.

occupied by a glutamine with arginines being present in several cases, including Afu endonuclease III. The NMR data highlight the importance of this structural element in DNA binding.

Interestingly, the C-terminal residues (Val206–Lys209) are largely unaffected by DNA binding as judged from the chemical shift perturbation studies presented here. These residues lie on the back surface of the protein (Figure 6b). The C-terminal residues, together with the FCL loop, helix H, and helix J, form the hydrophobic pocket that sequesters the [4Fe–4S] cluster from solvent. Previous studies have determined that the iron–sulfur cluster is catalytically inactive and redox-silent upon DNA binding and cleavage. The NMR data support these findings and reinforce the idea that the [4Fe–4S] cluster is an architectural scaffold that supports structural elements that mediate DNA binding.

In addition to motifs and elements suggested by model studies and crystallographic analysis, the chemical shift interaction map implicates an additional structural element of endonuclease III involved in DNA binding and recognition. Large chemical-shift perturbations were observed for backbone atoms in helix H, encompassing a stretch of residues from Val138 to Arg148. Helix H contains the completely conserved aspartate at position 139, which is thought to be catalytically important. Additionally, two highly conserved arginines at positions 144 and 148 in Afu endonuclease III are located in this helix. The model of Afu endonuclease III suggests that these two arginine sidechains may point into the cleft of the protein. The chemical-shift change for the backbone nuclei of these arginines was significant, and large changes in magnetic environment were seen for the surrounding helical residues, suggesting an entire realignment or movement of this helix upon DNA binding.

Recently Guan and coworkers [32] identified a pseudo HhH formed by helices 4 and 5 in MutY, another member of the endonuclease III family. These researchers proposed that both the HhH and the pseudo HhH loops closely approach the DNA backbone from opposite sides of the catalytic cleft and effect a ‘pincher’ grasp on the substrate backbone. In Afu endonuclease III, the homologous pseudo HhH is formed by helices D and E. The Afu sequence, in fact, contains a GVG sequence in the loop connecting helices D and E, similar to the pseudo HhH sequence identified in Eco MutY. However, the NMR data suggest that the amide proton of Gly80 does not make direct hydrogen-bonding DNA contact — it has a minimal chemical-shift perturbation (min∆ppm = 0.23 ppm) upon DNA binding. Other endonuclease III family members do not have a glycine at position 80. Unfortunately, the completely conserved Gly82 is unassigned in Afu endonuclease III. However, dramatic chemical-shift changes are seen for residues at the N-terminal end of helix E, including two positively charged residues, Arg85 and Lys87. Thus, the NMR data support involvement of the pseudo HhH in DNA binding. Guan and coworkers [32] also proposed that helix 2 and helix 3 in the MutY structure form a minor groove reading patch. Analysis of the complex formed by a catalytically inactive MutY protein and adenine suggested that the helix2/helix3 motif would recognize the minor groove of DNA substrate. The homologous structural element in Afu endonuclease III is formed by helices B and C. The chemical shift interaction map highlights a stretch of the protein backbone, formed by residues from Ala38 to Ala48, encompassing the C-terminal end of helix B, the intervening turn, and the N-terminal end of helix C. The loop between helices B and C contains an arginine at position 43. This position is highly conserved in various members of the endonuclease III family. It is usually

Among residues that showed significant shifts that are not on the continuous DNA-interaction surface, two regions seem important. First, Val21 and Tyr22 experienced significant chemical-shift perturbations upon DNA binding. These residues are contained in a linker segment that joins the two lobes of the endonuclease III structure. Guan and coworkers [32] noted a variability in the interdomain cleft among equivalent domains of Eco endonuclease III, AlkA, and MutY, and suggested that the linkers between the two domains may allow a clamping motion. The NMR data suggest that motion may be accommodated by the linker loop containing Val21 and Tyr22. Secondly, significant chemical-shift changes were noted for a stretch of residues from Phe168 and Glu174 in the loop between helices I and J, which is located at the back of the protein. Presumably DNA binding could cause a propagated change in helix packing around helix H that is accommodated by a conformational change in the loop connecting helices I and J. Apart from these two changes, the back of the protein was remarkably unperturbed, as determined by chemical-shift mapping, suggesting that the opposite face of the endonuclease III structure does not undergo large conformational changes, such as a loss or gain of helical structure, upon DNA binding. In summary, Figure 7 shows a cartoon model of the DNA–Afu endonuclease III complex based on the NMR chemical shift interaction map, the predicted electrostatic surface, previous crystal structures of endonuclease III family members, and site-directed mutagenesis studies. The surface of the model is colored by Coulomic electrostatic potential. The positively charged potential surfaces

Research Article Endonuclease III–DNA interactions Shekhtman et al.

Figure 7

A proposed model of DNA binding to Afu endonuclease III consistent with the interaction surface predicted by the chemical shift perturbation data and the calculated electrostatic surface potential. Electrostatic potential of the model of Afu endonuclease III was calculated using the program GRASP [72]. Regions of positive electrostatic potential are colored blue; negative surface charges are colored red. Positive surfaces span both lobes of the protein and are located in the catalytic cleft of the protein and at the surface formed by helix H. The cartoon model suggests a mode of DNA binding that places the DNA substrate in close proximity to Arg43 in the helix B–helix C loop and Arg144 and Arg148 contained in helix H.

form patches that run across one face of the protein, spanning both lobes of the endonuclease III structure. This positively charged surface is in striking agreement with the NMR-based DNA interaction map shown in Figure 6a. Notably, the solvent-accessible face of Helix H forms a positively charged surface in the cleft of the protein. In contrast, residues at the back of the protein (Figure 6b) experienced only minor chemical-shift perturbation upon DNA binding and the GRASP model suggests that this side of the molecule has very little positive electrostatic surface for DNA docking (data not shown). The model constructed here suggests that the bound substrate might require a slight bend to lie along the cleft of the protein and involve Helix H in DNA docking.

Biological implications DNA-repair enzymes must recognize and remove damaged, inappropriate, and mismatched bases from DNA molecules. The mode of recognition of the target base must be variable for the different substrates that are

927

recognized [54]. In theory, inappropriate or damaged bases can be recognized directly because they are structurally and chemically dissimilar from the four canonical bases. To remove one particular base from a mismatch requires that additional information, in the form of the two bases in the mismatched base pair, be used because the target base is a normal base. A mismatched base pair may be recognized by the overall structure of the mismatch, or by examining both bases separately to see if they constitute a canonical Watson–Crick base pair. Bases can be identified on the basis of shape, size, hydrophobicity, hydrogen-bonding properties, and/or electronic structure. Specific examples reveal how several enzymes recognize target bases in DNA. It has been shown that uracil N-glycosylase recognizes uracil by flipping the base out of the DNA duplex and into a deep binding pocket [55,56]. This pocket contains amino acids that exclude thymine and purines on the basis of steric bulk, and that hydrogen bond to the exocyclic oxygen O4, which differentiates uracil from cytosine. OGG1, which recognizes and removes 8-oxoguanine from DNA, shows an opposite base dependence because 8-oxoguanine is most readily removed when found opposite a cytosine [57,58]. This enzyme must examine both the damaged target base and the normal base-pair partner. TDG (thymine DNA mismatch glycosylase), which removes the thymine from a T•G mismatch, binds equally well to oligonucleotides containing either a T•G mismatch or a G opposite an apyrimidinic (AP) site [25], the product of the enzyme action. These results suggest that the targetbase thymine does not contribute substantial binding energy to this enzyme–DNA recognition complex. It has been shown that the HhaI methyltransferase uses a base-flipping mechanism [59,60]. The enzyme will also flip a uracil or an adenine into its catalytic pocket, which normally converts cytosine to 5-methylcytosine [61]. In this case, the catalytic pocket does not contribute to substrate specificity in a structural sense, but does differentiate between correct and incorrect bases due to its inability to chemically transfer a methyl group to uracil or adenine. Repair glycosylases thus display multiple mechanisms for damage recognition, including targetbase recognition, opposite-base recognition, base-pair recognition, and chemical editing. The recognition of target bases by other DNA-repair enzymes probably involves several of the strategies outlined above. One might expect that the DNA strand containing the damaged base is contacted and possibly compressed to allow the base to flip out of the helix. The opposite strand might also be contacted and the so-called ‘orphan’ base might be examined for identity as well. The target base might be flipped into a catalytic pocket that could contain structural residues that identify the base and catalytic residues that promote N-glycosidic bond cleavage.

928

Structure 1999, Vol 7 No 8

In other cases the pocket might provide only catalytic residues that chemically edit the flipped-out base. We have begun to map the interactions that occur when endonuclease III binds to DNA. We have chosen to use an oligonucleotide containing an AP site because we believe it may reflect many of the interactions made by repair enzymes of different substrate specificities. DNA footprinting [33] and our present work suggest that both DNA strands are contacted and that five elements of the protein structure are involved. The protein contacts at positions in the HhH (helix-hairpin-helix) and pseudo HhH could serve to compress the phosphodiester backbone of the strand containing the damaged base, as suggested by Guan and coworkers [32]. The helix B–helix C loop is in position to interact with the opposite strand and the orphan base. These contacts could help push the damaged base out of the helix and could also be involved in orphan-base recognition and binding. The contacts at the FCL (iron–sulphur cluster loop) motif would allow nonspecific binding of the enzyme to the DNA. The contacts at helix H are near the catalytic pocket and could reflect binding of the AP site or interactions with the phosphodiester backbone adjacent to the AP site. Our results clearly indicate that several interactions occur between endonuclease III and a DNA substrate. We now have the basis for rational site-directed mutagenesis to dissect the functions underlying these multiple interactions, which provide a platform for tailored specificity among different endonuclease III family members.

Materials and Methods Protein expression and purification The gene encoding Afu endonuclease III was found using a Gapped BLAST [62] search with E. coli endonuclease III as the query sequence. The Afu endonuclease III gene was amplified from genomic DNA by PCR. The amplified gene was cloned into the expression vector pET24a (Novagen, Inc.). The expression plasmid was transformed into E. coli BL21 (DE3) and Afu endonuclease III was purified from induced cells. For [U-15N, 13C] endonuclease III, cells were grown at 37°C in M9 medium supplemented with 0.1% of ferric citrate and with 0.5 g/l of 12C-glucose and 1 g/l of 15NH SO as the sole carbon and nitrogen 4 4 sources until the cell culture reached an absorbance of 0.7 OD at 600 nm. It was determined that at this cell density all the glucose in the medium was consumed and the cells stopped growing. At this point, 0.5 g/l of uniformly labeled 13C-glucose was added and protein overexpression was induced by addition of isopropyl-β-D-thiogalactoside (IPTG) to a concentration of 1.0 mM . After 3 h of induction, cells were harvested by centrifugation. After lysis and batch precipitation of DNA, the protein was purified using cation exchange, heparin affinity and size-exclusion chromatography. The final yield was 6 mg/l of uniformly 13C, 15N labeled Afu endonuclease III with 13C isotope enrichment of 93% (100% enriched = 25,335 Da), according to ESI (electrospray ionization) mass spectroscopy. [U-15N] Afu endonuclease III was purified from cells grown in M9 medium with (15NH4)2SO4 at 1 g/l as the sole nitrogen source. Cells were induced at 0.8 OD600 for 4 h and the purification procedure was the same as that used for the double-labeled sample. The final yield of

protein was 13 mg/l. ESI mass spectroscopy confirmed 99% enrichment and proper expression (100% enriched = 24,290 Da).

15N

For specific labeling with 15N lysine, 15N valine, and 15N leucine, cells were grown on defined rich medium with addition of 15N-labeled lysine, leucine, or valine [63]. The cells were induced at 0.4 OD600 for 3 h and then harvested. The overexpressed protein was purified using cationexchange chromatography. Homogeneity of the protein at this point was > 95% as estimated by Coomassie Blue staining of gels from sodium dodecyl sulphate polyacrylamide gel electrophoresis (SDS–PAGE). The sample was exchanged into NMR buffer (50 mM potassium phosphate buffer, pH 6.6, 400 mM KCl, 1 mM DTT and 0.01% NaN3). The final yields were approximately 14 mg/l for the specifically labeled proteins.

Purification of rAP 13-mer Oligonucleotides, 5′-GCGCAGUAGCCG-3′ and 5′-CGGCTGACTGCGC-3′, were purchased from Oligos Etc. The uracil-containing oligonucleotide (1 mM, 2 ml) was incubated for 3 h with uracil N-glycosylase at 50 µg/ml in 30 mM HEPES-KOH pH 8.0, 50 mM KCl, 1 mM EDTA at 37°C, followed by the addition of sodium borohydride to the concentration of 0.6 M. After 10 min incubation, the reduction reaction was terminated by the addition of HCl. The resulting oligonucleotide was purified by anion-exchange and reversed-phase chromatography. The complementary oligonucleotide was purified by the same procedure. Oligonucleotide concentrations were determined spectroscopically using millimolar extinction coefficients of 111.1 mM–1 cm—1 for the for rAP 13-mer and 112.7 mM—1 cm–1 for the complementary strand.

NMR experiments NMR experiments were performed on a Avance Bruker 600 MHz spectrometer. Protein samples, with concentrations ranging from 1 to 2 mM, were dissolved in NMR buffer (50 mM potassium phosphate, pH 6.6, 400 mM KCl, 0.02% NaN3, 1 mM DTT) and sealed in NMR tubes under argon atmosphere. To prepare the protein–DNA complex, a solution of 5 mM rAP 13-mer dissolved in NMR buffer was titrated into a 0.8 mM [U-13C, 15N] endonuclease III solution until the DNA was in slight excess (1:1.2). The complex was degassed and sealed in the NMR tubes under argon atmosphere. All NMR experiments were collected at 30°C. HNCA, HN(CO)CA, HNCO, HN(CA)CO, HNCACB triple-resonance experiments were acquired with previously described pulse sequences [36–39,41,64] and included constant-time nitrogen evolution, WATERGATE [65] water suppression and continuous 1H WALTZ16 [66] decoupling to minimize 13C relaxation. Spectral widths in 1H, 15N, 13C, and 13CO dimensions in all experiments, except HNCACB, were 2155.2, 1317.7, 3800, and 1509 Hz, respectively, with the proton carrier set to the center of the amide proton region. Spectral width in HNCACB experiment was 10,000 Hz in 13C dimension. Acquired data matrices contained 512 × 24 × 32 complex data points. The 15N-filtered NOESY spectrum [42,43] was acquired with 2155.2 and 6540 Hz proton and 1317.7 Hz nitrogen spectral widths and the proton carrier was set to the middle of the amide resonances. The acquired data matrix contained 512 × 64 × 128 complex data points. 15N-HSQC spectra were collected with proton and nitrogen spectral widths of 2155.2 and 1317.7 Hz, respectively. The 15N-HSQC data matrix contained 512 × 128 complex data points. All spectra were processed and analyzed using Felix 97.0 (Molecular Simulations, Inc.) software. Chemical shifts were referenced to DSS (sodium 2,2-dimethyl-2-silapentane-5-sulfonate). For the triple-resonance experiments an exponential window function with a linewidth of 2 Hz was applied to all data sets in proton dimensions. In carbon dimension, indirect FIDs (free induction decays) were apodized with a skewed sinebell window function, zero filled to twice the size and Fourier transformed. 15N dimensions were apodized using an exponential window function and zero filled to 62 complex points, Fourier transformed and phased. Three-dimensional data sets from the HNCA, HN(CO)CA, HNCO, and HN(CA)CO and HNCACB experiments consisted of 512 × 64 × 64 real points. For 15N-NOESY-HSQC spectra

Research Article Endonuclease III–DNA interactions Shekhtman et al.

apodization and zero filling were employed similarly to yield final spectra of 512 × 128 × 64 real points. Minimal ppm distance [47] for the pattern matching in the bound HNCO spectrum was calculated by: Minimum (∆ppm) = min{[(HN∆ppm)2 + (N∆ppm × σN)2 + (CO∆ppm × σCO)2]1/2} where HN∆ppm, N∆ppm, and CO∆ppm were the 1H, 15N, and 13CO chemical-shift differences between HNCO correlations for substrate-free and DNA-bound endonuclease III. Scaling factors, σN = 0.17 and σCO = 0.40, were ratios of chemical-shift ranges, based on the atomspecific chemical-shift dispersion observed in the endonuclease III spectra: 4 ppm for 1HN; 24 ppm for 15N; and 10 ppm for 13C. The scaling factors allowed normalization of the minimum ppm to the proton chemical-shift scale [47]. A peak in the bound HNCO spectrum was correlated to that of free HNCO if the min (∆ppm) between those peaks was less than 1 ppm, otherwise the min (∆ppm) was reported as 1 ppm.

Homology modeling Homology modeling was performed using the Swiss-Model optimization mode prepared by Swiss-PDB Viewer [67,68]. Sequence alignment between Eco endonuclease III and Afu endonuclease III was based on alignment of conserved residues and the homology of the secondary structures as revealed by the NMR experiments. Energy minimization was performed using GROMOS96 with successive application of 200 cycles of steepest descent and 300 cycles of conjugate gradient minimizers. The relative position of the four conserved cysteines ligating the [4Fe–4S] cluster was restricted during energy minimization. The resultant structure was further refined with the DISCOVER program (Molecular Simulations, Inc.) employing the all atom version of AMBER as a force field. Energy minimization was performed until the maximum derivative of energy fell below 0.5 kcal/mol (2.1 kJ/mol). The quality of the resultant structure was estimated by the PROCHECK program [69]. 86.5% of the Afu endonuclease III residues fall into the most favorable region of the Ramachandran plot, 13% to the allowed region and one amino acid, Ser42, into the disallowed region. A backbone atom rms deviation of 1.04 Å was calculated by comparing the model to the Eco endonuclease III crystal structure.

Acknowledgements We acknowledge the contributions of the Biological Mass Spectroscopy and Structural Biology NMR core facilities at the Wadsworth Center. This work was supported by a SUNY Faculty Research Award (DA 15681) to RPC and SMB.

References 1. Tainer, J.A., Thayer, M.M. & Cunningham, R.P. (1995). DNA repair proteins. Curr. Opin. Struct. Biol. 5, 20-26. 2. Cunningham, R.P. (1997). DNA glycosylases. Mutat. Res. 383, 189-196. 3. Cunningham, R.P. (1997). DNA repair: caretakers of the genome? Curr. Biol. 7, R576-579. 4. Mol, C.D., Parikh, S.S., Lo, T.P. & Tainer, J.A. (1998). Structural phylogenetics of DNA base excision repair. In DNA Repair. (Eckstein, F. & Lilley, D.M.J., eds), pp 29-69, Springer-Verlag, Berlin. 5. Strniste, G.F. & Wallace, S.S. (1975). Endonucleolytic incision of X-irradiated deoxyribonucleic acid by extract of Escherichia coli. Proc. Natl Acad. Sci. USA 72, 1997-2001. 6. Radman, M. (1976). An endonuclease from Escherichia coli that introduces single polynucleotide chain scissions in ultravioletirradiated DNA. J. Biol. Chem. 251, 1438-1445. 7. Boorstein, R.J., Hilbert, T.P., Cadet, J., Cunningham, R.P. & Teebor, G.W. (1989). UV-induced pyrimidine hydrates in DNA are repaired by bacterial and mammalian DNA glycosylase activities. Biochemistry 28, 6164-6170. 8. Mazumder, A., et al., & Bolton, P.H. (1991). Stereochemical studies of the beta-elimination reactions at aldehydic abasic sites in DNA: endonuclease III from Escherichia coli, sodium hydroxide, and Lys-TrpLys. Biochemistry 30, 1119-1126. 9. Bailly, V. & Verly, W.G. (1987). Escherichia coli endonuclease III is not an endonuclease but a β-elimination catalyst. Biochem. J. 242, 565-572.

929

10. Kim, J. & Linn, S. (1988). The mechanisms of action of E. coli endonuclease III and T4 UV endonuclease (endonuclease V) at AP sites. Nucleic Acids Res. 16, 1135-1141. 11. Doetsch, P.W. & Cunningham, R.P. (1990). The enzymology of apurinic/apyrimidinic endonucleases. Mutat. Res. 236, 173-201. 12. Cunningham, R.P. & Weiss, B. (1985). Endonuclease III (nth) mutants of Escherichia coli. Proc. Natl. Acad. Sci. USA 82, 474-478. 13. Asahara, H., Wistort, P.M., Bank, J.F., Bakerian, R.H. & Cunningham, R.P. (1989). Purification and characterization of Escherichia coli endonuclease III from the cloned nth gene. Biochemistry 28, 4444-4449. 14. Cunningham, R.P., et al., & Emptage, M.H. (1989). Endonuclease III is an iron-sulfur protein. Biochemistry 28, 4450-4455. 15. Cunningham, R.P., Ahern, H., Xing, D., Thayer, M.M. & Tainer, J.A. (1994). Structure and function of Escherichia coli endonuclease III. Ann. N. Y. Acad. Sci. 726, 215-222. 16. Thayer, M.M., Ahern, H., Xing, D., Cunningham, R.P. & Tainer, J.A. (1995). Novel DNA binding motifs in the DNA repair enzyme endonuclease III crystal structure. EMBO J. 14, 4108-4120. 17. Fu, W., O'Handley, S., Cunningham, R.P. & Johnson, M.K. (1992). The role of the iron-sulfur cluster in Escherichia coli endonuclease III. A resonance Raman study. J. Biol. Chem. 267, 16135-16137. 18. Au, K.G., Clark, S., Miller, J.H. & Modrich, P. (1989). Escherichia coli mutY gene encodes an adenine glycosylase active on G-A mispairs. Proc. Natl. Acad. Sci. USA 86, 8877-8881. 19. Michaels, M.L., Pham, L., Nghiem, Y., Cruz, C. & Miller, J.H. (1990). MutY, an adenine glycosylase active on G-A mispairs, has homology to endonuclease III. Nucleic Acids Res. 18, 3841-3845. 20. Tsai-Wu, J.J., Liu, H.F. & Lu, A.L. (1992). Escherichia coli MutY protein has both N-glycosylase and apurinic/apyrimidinic endonuclease activities on A•C and A•G mispairs. Proc. Natl Acad. Sci. USA 89, 8779-8783. 21. Slupska, M.M., Baikalov, C., Luther, W.M., Chiang, J.H., Wei, Y.F. & Miller, J.H. (1996). Cloning and sequencing a human homolog (hMYH) of the Escherichia coli mutY gene whose function is required for the repair of oxidative DNA damage. J. Bacteriol. 178, 3885-3892. 22. Bulychev, N.V., et al., & Johnson, F. (1996). Substrate specificity of Escherichia coli MutY protein. Biochemistry 35, 13147-13156. 23. Zharkov, D.O. & Grollman, A.P. (1998). MutY DNA glycosylase: base release and intermediate complex formation . Biochemistry 37, 12384-12394. 24. Horst, J.P. & Fritz, H.J. (1996). Counteracting the mutagenic effect of hydrolytic deamination of DNA 5-methylcytosine residues at high temperature: DNA mismatch N-glycosylase Mig.Mth of the thermophilic archaeon Methanobacterium thermoautotrophicum THF. EMBO J. 15, 5459-5469. 25. Begley, T. & Cunningham, R.P. (1999). Methanobacterium thermoformicicium thymine DNA mismatch glycosylase: conversion of an N-glycosylase to an AP lyase. Prot. Eng. 12, 333-340 26. Piersen, C.E., Prince, M.A., Augustine, M.L., Dodson, M.L. & Lloyd, R.S. (1995). Purification and cloning of Micrococcus luteus ultraviolet endonuclease, an N-glycosylase/abasic lyase that proceeds via an imino enzyme-DNA intermediate. J. Biol. Chem. 270, 23475-23484. 27. Hilbert, T.P., Chaung, W., Boorstein, R.J., Cunningham, R.P. & Teebor, G.W. (1997). Cloning and expression of the cDNA encoding the human homologue of the DNA repair enzyme, Escherichia coli endonuclease III. J. Biol. Chem. 272, 6733-6740. 28. Labahn, J., Scharer, O.D., Long, A., Ezaz-Nikpay, K., Verdine, G.L. & Ellenberger, T.E. (1996). Structural basis for the excision repair of alkylation-damaged DNA. Cell 86, 321-329. 29. Nash, H.M., et al., & Verdine, G.L. (1996). Cloning of a yeast 8-oxoguanine DNA glycosylase reveals the existence of a baseexcision DNA-repair protein superfamily. Curr. Biol. 6, 968-980. 30. van der Kemp, P.A., Thomas, D., Barbey, R., de Oliveira, R. & Boiteux, S. (1996). Cloning and expression in Escherichia coli of the OGG1 gene of Saccharomyces cerevisiae, which codes for a DNA glycosylase that excises 7,8-dihydro-8-oxoguanine and 2,6-diamino-4hydroxy-5-N- methylformamidopyrimidine. Proc. Natl Acad. Sci. USA 93, 5197-5202. 31. Doherty, A.J., Serpell, L.C. & Ponting, C.P. (1996). The helix-hairpinhelix DNA-binding motif: a structural basis for non-sequence-specific recognition of DNA. Nucleic Acids Res. 24, 2488-2497. 32. Guan, Y., et al., & Tainer, J.A. (1998). MutY catalytic core, mutant and bound adenine structures define specificity for DNA repair enzyme superfamily. Nat. Struct. Biol. 5, 1058-1064. 33. O'Handley, S., Scholes, C.P. & Cunningham, R.P. (1995). Endonuclease III interactions with DNA substrates. I. Binding and footprinting studies with oligonucleotides containing a reduced apyrimidinic site. Biochemistry 34, 2528-2536.

930

Structure 1999, Vol 7 No 8

34. Xing, D., Dorr, R., Cunningham, R.P. & Scholes, C.P. (1995). Endonuclease III interactions with DNA substrates. II. The DNA repair enzyme endonuclease III binds differently to intact DNA and to apyrimidinic/apurinic DNA substrates as shown by tryptophan fluorescence quenching. Biochemistry 34, 2537-2544. 35. Porello, S.L., Cannon, M.J. & David, S.S. (1998). A substrate recognition role for the [4Fe-4S]2+ cluster of the DNA repair glycosylase MutY. Biochemistry 37, 6465-6475. 36. Kay, L.E., Ikura, M., Tschudin, R. & Bax, A. (1990). Three-dimensional triple resonance NMR spectroscopy of isotropically enriched proteins. J. Magn. Reson. 89, 496-514. 37. Grzesiek, S. & Bax, A. (1992). Improved 3D triple-resonance NMR techniques applied to a 31 kDa protein. J. Magn. Reson. 96, 432-440. 38. Farmer II, B.T., Venters, L.D., Spicer, L.D., Wittekind, M.G. & Mueller, L. (1992). A refocussed and optimized HNCA: increased sensitivity and resolution in large macromolecules. J. Biomol. NMR 2, 195-202. 39. Clubb, R.T., Thanabal, V. & Wagner, G. (1992). A constant-time threedimensional triple-resonance pulse scheme to correlate intraresidue 1H, 15N, and 13C chemical shifts in 15N-13C labeled proteins. J. Magn. Reson. 97, 213-217. 40. Cavanagh, J., Fairbrother, W.J., Palmer, A.G. & Skelton, N.J. (1996). Protein NMR Spectroscopy: Principles and Practice. Academic Press, San Diego. 41. Wittekind, M.G. & Mueller, L. (1993). HNCACB, a high sensitivity 3D NMR experiment to correlate amide-proton and nitrogen resonances with alpha and beta carbon resonances in proteins. J. Magn. Reson. B 101, 201-295. 42. Marion, D., et al., & Clore, G.M. (1989). Overcoming the overlap problem in the assignment of 1H NMR spectra of larger proteins by use of three-dimensional heteronuclear 1H-15N Hartmann-Hahnmultiple quantum coherence and nuclear Overhauser- multiple quantum coherence spectroscopy: application to interleukin 1 beta. Biochemistry 28, 6150-6156. 43. Zuiderweg, E.R. & Fesik, S.W. (1989). Heteronuclear threedimensional NMR spectroscopy of the inflammatory protein C5a. Biochemistry 28, 2387-2391. 44. Wishart, D.S. & Sykes, B.D. (1994). The 13C chemical-shift index: a simple method for the identification of protein secondary structure using 13C chemical-shift data. J. Biomol. NMR 4, 171-180. 45. Wishart, D.S., Bigam, C.G., Holm, A., Hodges, R.S. & Sykes, B.D. (1995). 1H, 13C and 15N random coil NMR chemical shifts of the common amino acids. I. Investigations of nearest-neighbor effects. J. Biomol. NMR 5, 67-81. 46. Kuo, C.F., McRee, D.E., Fisher, C.L., O'Handley, S.F., Cunningham, R.P. & Tainer, J.A. (1992). Atomic structure of the DNA repair [4Fe–4S] enzyme endonuclease III. Science 258, 434-440. 47. Farmer, B.T., II, et al., & Mueller, L. (1996). Localizing the NADP+ binding site on the MurB enzyme by NMR. Nat. Struct. Biol. 3, 995-997. 48. Constantine, K.L., et al., & Farmer, B.T., II (1997). Characterization of NADP+ binding to perdeuterated MurB: backbone atom NMR assignments and chemical-shift changes. J. Mol. Biol. 267, 1223-1246. 49. Williamson, R.A., Carr, M.D., Frenkiel, T.A., Feeney, J. & Freedman, R.B. (1997). Mapping the binding site for matrix metalloproteinase on the N-terminal domain of the tissue inhibitor of metalloproteinases-2 by NMR chemical shift perturbation. Biochemistry 36, 13882-13889. 50. Lee, A.L., et al., & Wemmer, D.E. (1997). Chemical shift mapping of the RNA-binding interface of the multiple-RBD protein sex-lethal. Biochemistry 36, 14306-14317. 51. Schmiedeskamp, M., Rajagopal, P. & Klevit, R.E. (1997). NMR chemical shift perturbation mapping of DNA binding by a zinc-finger domain from the yeast transcription factor ADR1. Protein Sci. 6, 1835-1848. 52. Rajagopal, P., Waygood, E.B., Reizer, J., Saier, M.H., Jr. & Klevit, R.E. (1997). Demonstration of protein-protein interaction specificity by NMR chemical shift mapping. Protein Sci. 6, 2624-2627. 53. Foster, M.P., et al., & Wright, P.E. (1998). Chemical shift as a probe of molecular interfaces: NMR studies of DNA binding by the three aminoterminal zinc finger domains from transcription factor IIIA. J. Biomol. NMR 12, 51-71. 54. Verdine, G.L. & Bruner, S.D. (1997). How do DNA repair proteins locate damaged bases in the genome? Chem. Biol. 4, 329-334. 55. Slupphaug, G., Mol, C.D., Kavli, B., Arvai, A.S., Krokan, H.E. & Tainer, J.A. (1996). A nucleotide-flipping mechanism from the structure of human uracil-DNA glycosylase bound to DNA. Nature 384, 87-92.

56. Parikh, S.S., Mol, C.D., Slupphaug, G., Bharati, S., Krokan, H.E. & Tainer, J.A. (1998). Base excision repair initiation revealed by crystal structures and binding kinetics of human uracil-DNA glycosylase with DNA. EMBO J. 17, 5214-5226. 57. Bjoras, M., et al., & Seeberg, E. (1997). Opposite base-dependent reactions of a human base excision repair enzyme on DNA containing 7,8-dihydro-8-oxoguanine and abasic sites. EMBO J. 16, 6314-6322. 58. Girard, P.M., D'Ham, C., Cadet, J. & Boiteux, S. (1998). Opposite base-dependent excision of 7,8-dihydro-8-oxoadenine by the Ogg1 protein of Saccharomyces cerevisiae. Carcinogenesis 19, 1299-1305. 59. Klimasauskas, S., Kumar, S., Roberts, R.J. & Cheng, X. (1994). HhaI methyltransferase flips its target base out of the DNA helix. Cell 76, 357-369. 60. Reinisch, K.M., Chen, L., Verdine, G.L. & Lipscomb, W.N. (1995). The crystal structure of HaeIII methyltransferase convalently complexed to DNA: an extrahelical cytosine and rearranged base pairing. Cell 82, 143-153. 61. O'Gara, M., Horton, J.R., Roberts, R.J. & Cheng, X. (1998). Structures of HhaI methyltransferase complexed with substrates containing mismatches at the target base. Nat. Struct. Biol. 5, 872-877. 62. Altschul, S.F., et al., & Lipman, D.J. (1997). Gapped BLAST and PSIBLAST: a new generation of protein database search programs. Nucleic Acids Res. 25, 3389-3402. 63. Muchmore, D.C., McIntosh, L.P., Russell, C.B., Anderson, D.E. & Dahlquist, F.W. (1989). Expression and nitrogen-15 labeling of proteins for proton and nitrogen-15 nuclear magnetic resonance. Methods Enzymol. 177, 44-73. 64. Munhandriham, D.R. & Kay, L.E. (1994). Gradient-enhanced tripleresonance three dimensional NMR experiments with improved sensitivity. J. Magn. Res. B 103, 203-216. 65. Sklenar, V., Piotto, M., Leppik, R. & Saudek, V. (1993). Gradienttailored water suppression for 1H-15N HSQC experiment optimized to retain full sensitivity. J. Magn. Res. A 102, 241-245. 66. Shaka, A.J., Keeler, J. & Freeman, R. (1983). Evaluation of a new broadband decoupling sequence: WALTZ-16. J. Magn. Res. 53, 313-340. 67. Peitsch, M.C. (1996). ProMod and Swiss-Model internet based tool for automated comparative protein modelling. Biochem. Soc. Trans. 24, 274-279. 68. Guex, N. & Peitsch, M.C. (1997). Swiss-Model and the SwissPdbviewer: an environment for comparative protein modelling. Electrophoresis 18, 2714. 69. Laskowski, R.A., MacArthur, M.W., Moss, D.S. & Thornton, J.M. (1993). PROCHECK: a program to check the stereochemical quality of a protein structure. J. Appl. Crystallogr. 26, 283-291. 70. Piotto, M., Saudek, V. & Sklenar, V. (1992). Gradient-tailored excitation for single-quantum NMR spectroscopy of aqueous solutions. J. Biomol. NMR 2, 661-665. 71. Evans, S.V. (1993). SETOR: hardware lighted three-dimensional solid model representations of macromolecules. J. Mol. Graphics 11, 134-138. 72. Nicholls, A., Bharadwaj, R. & Honig, B. (1993). GRASP: a graphical representation and analysis of surface properties. Biophys. J. 64, A166.

Because Structure with Folding & Design operates a ‘Continuous Publication System’ for Research Papers, this paper has been published on the internet before being printed (accessed from http://biomednet.com/cbiology/str). For further information, see the explanation on the contents page.