J. Mol.

Biol.

(1992) 227, 347-351

CRYSTALLIZATION NOTES

Crystallization and Crystallographic Characterization of the Iron-Sulfur-containing DNA-repair Enzyme Endonuclease III from Escherichia coli he-Fu Kuo’, Duncan E. McRee’, Richard P. Cunningham’ ‘Department

of Molecular Biology, The Scripps La Jolla, CA 92037, U.S.A.

and John A. Tainer’

Research Institute

2Department of Biological Sciences, Center for Biochemistry and Biophysics State University of New York at Albany, Albany, NY 12222, U.S.A. (Received 27 February

1992; accepted 27 April

1992)

Endonuclease III from Escherichia coli is an iron-sulfur enzyme possessing both DNA M-glycosylase and apurinic/apyrimidinic lyase activities. It could serve to repair damaged thymine residues in DNA via base excision-repair. We have crystallized endonuclease III by a combination of dialysis and seeding techniques after exploration of a wide variety of precipitants which failed to yield macroscopic crystals. Important features of the optimized crystallization include: the use of 5 to 10% glycerol, a temperature of 15”C, controlled dialysis to decrease ionic strength and macroseeding using a 200 m&r-NaCl transfer buffer to dissolve microcrystalline contamination. The crystals belong to space group P2,2,2, with unit cell dimensions of a = 48.5 A, b = 65.8 A, c = 86.8 A, a = b = y = 90”, have one 23 kDa monomer per asymmetric unit, and diffract to 1.84 A. A native anomalous Patterson map located the iron-sulfur cluster and reaffirmed its existence. The reported crystallization procedures ensure an ample supply of crystals for the extensive heavy-atom derivative search necessary for this labile iron-sulfur enzyme. The elucidation of endonuclease III structure will facilitate not only the understanding of glycosylase and lyase mechanisms but also the structure and function of this new class of iron-sulfur proteins. Keywords:

endonuclease

III;

iron-sulfur X-ray

protein; DNA diffraction

excision-repair;

crystallization;

Endonuclease III from Escherichia coli has both Dn’A Zglycosylase and apurinic/apyrimidinic lyase activity and can repair damaged thymine residues NA (Breimer & Lindahl, 1980, 1984; Radman, 1976). Thymine residues are known to be damaged by oxidative agents and oxygen free radicals generated by normal aerobic metabolism or by ionizing radiation (Breimer & Lindahl, 1985; Denq & Fridovich, 1989). The rV-glycosylase activity of endonuclease III releases damaged thymine residues. The associated lyase activity of endonuclease III incises the DNA sugar-phosphate backbone 3’ to the abasic site once the damaged thymine ring is removed by the glycosylase activity (Katcher & Wallace, 1983; Demple & Linn, 1980; Kow & Wallace, 1987). Further processing by endonueleases: DNA polymerase and DNA ligase completes the repair of damaged DNA (Doetsch &

product overproduced and purified in large quantity (Cunningham & Weiss, 1985; Asahara et al. i 1989). Endonuclease III is a monomeric protein of 23 kDa, containing 211 amino acid residues. Mijssbauer and electron paramagnetic resonance (EPR,?) spectroscopies identify a (4Fee4S)+’ iron-sulfur cluster within this protein (Cunningham et al., 1989). This iron-sulfur cluster within endonuclease III broadens the scope of roles iron-sulfur clusters may play in biological systems, in addition to its well documented role as an electron carrier in ferredoxins and its catalytic role in aconitase and dehydratases (Beinert, 1990; Cammack, 1991). Although endonuclease III cleaves the DP\‘A backbone at a position 3’ to the abasic sites, it does not do so by classic phosphodiesterase activity which will leave behind a 3’.OH end. Instead, endonnclease III

Cunningham,

t Abbreviations used: EPR, electron paramagnetic resonance; Hepes, 4-(2-hydroxyethyl)-l-piperazineethanesulfonic acid.

1990).

The nth gene

coding for endonuclease III in E. coli has been cloned and sequenced and its gene OQ2%2836/92/190347+5

$08.00/O

347

0 1992 Academic Press Limited

348

C.-F.

Kuo et al

cleaves the DNA backbone 3’ to the abasic sites via a p-elimination mechanism (Bailly & Verly, 1987; Mazumder et aZ., 1991). The exact mechanism of this p-elimination is unknown and it may be a fortuitous activity gained during the evolution of this enzyme as a glycosylase. Elucidation of this p-elimination mechanism will be greatly facilitated by a detailed knowledge of the atomic structure of endonuclease HT. Here, we report, the crystallization and characterization of endonuclease III: as a step toward the elucidation of the atomic structure of E. coli endonuclease III by X-ray crystallograph,y. Large quantities of overproduced recombinant protein, together with development of novel methods for controlling crystallization, allowed us to obtain large-diffraction quality crystals, suitable for collection of 2 A native data (1 A = 0.1 nm) for the atomic structure determination. Endonuclease ITT was purified from an overproducing strain of E. coli as described previously (Asahara et al., 1989). Purified enzyme stored in 100 mr\l-potassium phosphate (pH 6.6), 50 y0 glycerol was stable at -20°C for several months. Protein used for crystallization exhibited a A,,,/A,s, ratio of 0.39 and this value indicates that the iron-sulfur clusters of these enzymes are mostly intact, based upon preparations characterized by EPR (Cunningham et al., 1989). For crystallization, a sample of protein solution was removed from storage at - 20°C and dialyzed overnight at 4°C against 20 mM-potassium phosphate (pH 6.6), 200 mM-KCl, 3 mM-NaN,, 5% glycerol to remove excess glycerol and to exchange buffers. The dialyzed protein solution was then concentrated to a final concentration of 5 mg/ml using a Schleicher and Schuell vacuum concentration device. Concentrated protein solution (25 ~1) was placed inside the well of a dialysis button and covered with dialysis membrane. The dialysis button was then placed inside a 50 ml flask containing 50 ml of crystallization buffer (5 mM4-(2-hydroxyethyl)l-piperazine-ethanesulfonic acid (Hepes)/Na+ (pH 7-O), 100 mM-NaCl, 3 miv-NaN,, 5% glycerol), and left without perturbation inside an incubator kept at 15°C. Large crystals appeared within several hours. For seeding, protein solution was concentrated as described above, then dialyzed against 5 mM-Hepes/Na+ (pH 7.0), 140 miv-NaCl, 3 mM5 y0 glycerol. Dialyzed protein solution NaN,, set up against a reservoir (20 ~1) was then containing crystallization buffer using the sitting drop vapor diffusion configuration. Protein solution with this slightly higher ionic strength (i.e. 140 miv-NaCl) was unable to initiate de nova crystallization, but was able to sustain growth of crystals or nuclei when introduced into the protein drop. The protein drop was immediately streaked with a clean human hair probe which had just touched the surface of a crystal grown de nova by the dialysis method. A string of crystals emerged along the track of streaking within several hours. These

microcrystals were allowed to grow until protein was consumed, as indicated by the bleaching of the brownish protein drop. Microcrystals with suitable morphology and clean surfaces were then washed in buffer containing 200 mM-NaC1 and transferred into a fresh protein drop set up as described above. Single large untwinned crystals could be obtained reproducibly in this manner. Precession photographs at 15” intervals were taken using a Charles Supper precession camera mounted on a Rigaku RU 200 rotating copper anode X-ray generator operating at 40 kV and 150 mA. A complete native dat’a set was collected on a Siemens area detector equipped with a P4RA 4-axis goniometer using 1.54 A Cu K, iadiation. Data were collected using the FRAMBO dat’a acquisition program and processed using the Xengen version 1.3 data processing program (Howard et al., 1987). Data collection was achieved rxsing 0.1” oscillation steps around the omega axis and 180 seconds exposure per frame, at a crystai-todetector distance of 12 cm and at a swing angle of 30”. Our initial attempts to crystallize endonuclease III with numerous precipitants including salts; polyalcohol or organic solvents were unsuccessful. The observat,ion that glistening precipitate formed during dialvsis of concentrated endonuclease ITT solution agkd low ionic strength buffer suggested dialysis as a means of crystallization, which then proved to be successful. The final ionic strength of dialysate is critical in determining the outcome of crystallization. In our case: 100 to 120 rnM-NaCP and 4 to 5 mg protein/ml concentration were optimal. Higher ionic strength prevented de nova crystallization and lower ionic strength resulted in precipitation rather than crystallization. Although de nouo crystallization by dialysis is fast, we had difficulty in controlling the resulting quality and quantity of crystals. Varying the temperature, ionic st’rength and protein concentration as well as using a stepwise gradient failed to improve the result of dialysis crystallization. Microseeding and macroseeding techniques were therefore used to obtain reproducibly large quantities of diffraction quality crystals. The success of this unusual combination of seeding and dialysis techniques is demonstrated in Figure 1. Using higher ionic strength dissolving buffer (200 miv-NaCl) as the transferring solution for handling crystals during macroseeding was eritical to prevent microcrystalline contamination, Glycerol, which was thought to be detrimental to the crystallization process and hence an uncommon additive in crystallization (Gilliland, 1988), was used successfully in our crystallization procedure, The addition of glycerol and decreasing the temperature to 15°C delayed protein precipitation to prolong the viability of the protein solution thus contributing to the success and reproducibility of crystallization. Examination of precession photographs revealed mmm symmetry and the major axes had systematic absence at every odd reflection indicating Z-fold

Crystallization

(b) ‘P

(cj Figure 1. Crystallization of endonuclease III by dialysis; microseeding, and macroseeding methods. (a) De nowo crystallization by dialysis: the dialysis was carried out as described in the text. The crystals emerged within several hours. These crystals tended to nucleate along the side and bot,tom of the well and are hollow at the ends (0.7 mm x 0.2 mm x OS! mm). (b) Microseeding: crystals appeared along the tract of streaking and had solid rather than hollow ends (0.15 mm x 0.1 mm x 0.1 mm). Crystals with clean surfaces and suitable morphology were chosen as seeds for macroseeding. (c) Macroseeding: microcrystals obtained from microseeding were used to seed a fresh protein drop as described in the text. The use of higher ionic strength buffer (200 mw-NaCl) as the transferring solution is critical to avoid microcrystalline contamination and obtain large single crystals (1.2 mm x 0.5 mm x 0.3 mm).

screw axes. This, and the lack of higher allowed unit

the assignment cell constants:

c = 86.8 A,

of space

group

symmetry,

P2,2,2,

with

a = 48.5 A, b = 65.8 A, cx = fi = y = 90”, with one monomeric

Notes

349

Figure 2. Native anomalous difference Patterson map of an endonuclease III crystal. The anomalous difference Patterson was calculated using data between 20 and 5 a resolution and was contoured at 1 sigma per level. In the Harker section at u = 0.5 shown here, the iron-sulfur cluster peak appears at (0.5, 0, 0.38) wit’h a height of 7 sigma, as compared to a maximum of 2 sigma for other peaks. The peak at the right lower corner (0.5, 0.5, @5) is an overlapping peak from another Harker se&on. The coordinates of the cluster were determined to be I = 0.025, y =025, 5 =@09.

protein molecule per asymmetric unit. Matthew’s (1968) coefficient was calculated to he 3.01 A’/ dalton. Crystal density was measured using Ficoll density gradient centrifugation to be 1.124 g/cm3, and solvent content of the crystal was 55% (Westbrook, 1985). The crystals diffracted to a maximal resolution of 1.8 .& and survived X-ray irradiation well, diffracting to 2.5 ,& for 48 h at 18 "C. A complete native data set was collected to 1.98 a with statistics shown in Table 1. A native anomalous difference Patterson was calculated using AIFl’ between Uijvoet mates to locate the 4Fe-413 cluster. A section of this Patterson map at Ha&r’s section u = 0.5 is shown in Figure 2. Although the data collection strategy was optimized for the collection of high resolution data rather than the anomalous signals, the anomalous Patterson was clear and confirmed both the space group determinaltion and the existence of the 4Fe-4X cluster. Oxidation damages to DNA are well-documented inevitable complications of aerobic metabolism. Among the four nucleotides that compose DKA, thymine is the most susceptible to damage by ultraviolet light as well as by oxidation, due to the nature of Cc5)-Cc6) double bond and the accessibility of this bond from the major groove of DNA. Oxidation products of thymine have been identified in humans and rats, and are derived from endo-

350

C.-F.

KILO et al.

Table 1 Data collection

statistics for

E. coli endonuclease

III

crystals

Resolution limit (8)

Number of observations

Number unique

Percent complete

%ymt

3.60 2.86 2.50 2.27 2.10 1.98 Total

29,010 30,191 26,702 20,186 18,167 16.940 141,196

3461 3311 3296 3269 3254 3242 19,833

99.2 100.0 100.0 99.6 %+8 PO00 99.8

2.7 45 8.0 9.4 12.9 18.3 64

Average I/sigma(Z) 164.9 797 38.4 25.3 15.5 91 56.7

t %,m is the unweighted absolute value R-factor on I between symmetry mates x 100. Statistics of data collection on endonuclease III crystals. The data were collected on a Xentronics area detector at an ambient temperature of I8”C, and were processed as described in the text. The data sets from 3 separate crystals were combined to obtain the complete data set with statistics shown.

origins (Cathcart et al., 1984). Hence it is not surprising that several elaborate systems have evolved to specifically deal with damaged thymine among which photolyase, T4 residues in DNA; UV-endonuclease and E. coli endonuclease III are most representative (Sancar & Sancar, 1988). The endonuclease III structure will provide information regarding: (1) the liganding and environment of the 4Fe-4S cluster; (2) the structural basis for the proposed B-elimination and glycosylase mechanisms; and (3) the substrate specificity for damaged thymine. Comparison of the endonuclease III sequence with homologous MutY glycosylase (Michaels et al., 1990) suggests that MutY is also a 4Fe-4S cluster containing enzyme, and the four conserved cysteine residues identify the probable ligands for both enzymes. The predicted endonuclease III Cys ligand spacing (Cys-X6-Cys-X,-Cys-X5-Cys), which places all four ligands in a continuous and unusually short sequence is distinct from those in all other 4Fe-4S proteins: ferredoxin (distal Cys + Cys-X,-Cys-X,Cys); high-potential iron protein (Cys-X,-CysX,,Cys-X1,-Cys); and trimethylamine dehydrogenase (Cys-X,-Cys-X,-Cys-X,,-Cys) (Beinert, 1990; Cammack, 1991). Thus, the endonuclease III structure will be representative of a novel class of 4Fe-4S cluster proteins. Glycerol is commonly believed to both hinder crystallization and reduce crystalline order. For the growth of highly ordered endonuclease III crystals, glycerol had two major positive effects. First, glycerol improved the reproducibility of crystallization by stabilizing the protein, as evidenced by delayed protein precipitation. Second, glycerol improved crystal quality by slowing the rate of crystal growth while reducing twinning and cuspidend formation. These results on endonuclease III crystallization suggest that glycerol should be included as a variable parameter in crystallization scans, particularly where its presence improves protein stability under crystallization conditions. The unusual combination of dialysis and seeding, together with the lower temperature and glycerol techniques presented here, ensures an ample supply genous

of diffraction quality crystals. This allows tbe extensive search for heavy-atom derivative necessary for this labile iron-sulfur enzyme. The 2 -4 resolution native data set is of sufficiently high quality to promise an accurate atomic structure. The native anomalous difference Patterson demonstrates the existence of the metal cluster within the crystals will allow crystallographic which therefore characterization for this new class of 4Fe-4S proteins. The solution of the endonuclease III structure by multiple isomorphous replacement method is currently underway. This work was supported by National Tnstit,utes Health Grant GM 46312. We thank Dr Hans Parge help with precession photography.

of for

References Asahara,

H., Wistort;

P. -li., Bank,

& Cunningham,

R.

P.

J. F., Bakerian,

(1989).

Purification

R. 1%.

and

characterization of Escherichia coli endonuclease III from the cloned nth gene. Biochemistry, 28, 4444-4449. Bailly, V. & Verly. W. G. (1987). E. coli endonuclease III is not an endonuclease but a p-elimination catalyst. Biochcm. .J. 242, 565-572. Beinert), H. (1990). Recent developments in the field of iron-sulfur proteins. FAXEB J. 4. 2483-2491. Breimer, L. & Lindahl, T. (1980). A D1;A glycosylase from Escherichia coli that releases free urea from a polydeoxyribonucleotide containing fragments of base residues. Nucl. Acids Res. 8, 6199-6211. Breimer, 1,. H. & Lindahl, T. (1984). DNA glycosylase activities for thymine residues damaged by ring saturation, fragmentation, or ring contraction are functions of endonuclease III in Escherichia co&. J.

Riol.

Chem.

259,

5543-5548.

Breimer. L. hi. & Lindahl, T. (1985). Thymine lesions produced by ionizing radiation on double stranded DKA. Biochemistry, 24, 401%4022. Cammack, R. (1991). Iron-sulfur clusters in enzymesthemes and variations. Advan. dnorg. Chem. 38; l-18. Cathcart, R., Schwiers, E., Saul; R. L. & Ames, B. PI’. (1984). Thymine glycol and thymidine glycol in human and rat urine: a possible assay for oxidative DNA damage. Proc. Nat. Acad. Sci., U.S.A. 81, 5633-5637.

Crystallization

Cunningham, R. P. & Weiss, B. (1985). Endonuclease III mutants of E. coli. Proc. Nat. Acad. Sci., U.S.A. 82, 474-478. Cunningham, R. I’.; Asahara, H., Bank, J. F., Scholes, C. P.; Salerno, J. C., Surerus, K., Miinck, E., McCracken, J., Peisach, J. & Emptage, M. H. (1989). Endonuclease III is an iron-sulfur protein. Biochemistry, 28, 4450-4455. Demple. B. & Linn, S. (1980). DNA N-glycosylases and UV repair. Nature (London), 287, 203-208. Deny, R.-Y. & Fridovich, I. (1989). Formation of endonuclease III sensitive sites as a consequence of oxygen attack on DNA. Free Radical Biol. Med. 6. 123-l 29. Doetsch, I”. W. & Cunningham, R. P. (1990). The enzymology of apurinic/apyrimidinic endonucleases. Mutat. Res. 236, 173-201. Gilliland, G. L. (1988). A biological macromolecule crystallization database: a base for a crystallization strategy. .P. Crystal Growth, 90, 51-59. Howard, A. J., Gilliland, G. L., Finzel, B. C. & Poulos, T. L. (1987). The use of an imaging proportional counter in macromolecular crystallography. J. A&. Cryatallogr. 20; 383-387. Kateher, L. H. & Wallace, S. S. (1983). Characterization of the Escherichia coli X-ray endonuclease, endo-

Edited

Notes

351

nuclease HI. Biochemistry, 22, 4071b408H. Y. W. & Wallace, S. S. (1987). Mechanism of action of Escherichia coli endonuclease III. Biochemistry, 26, 8200-8206. Matthews, B. W. (1968). Solvent content of protein crystals. J. Mol. Biol. 33, 491-497. Mazumder, A., Geralt, J. A., Abalson, M. J., Stubbe, J., Cunningham, R. P., Withka, J. & Bolton, P. H. (1991). Stereochemical studies of the p-elimination reactions at aldehydic abasic sites in DNA: endonuclease III from Escherichia coli; sodium hydroxide, and Lys-Trp-Lys. Biochemistry, 30. 1 E19.-1126. Michaels, M. L., Pham, L., Nghiem, Y., Cruz, C. & Miller, J. H. (1990). MutY, an adenine glucosylase active on G-A mispairs, has homology to endonuclease III. Nucl. Acids Res. 18, 3841-3845. Radman, M. (1976). An endonuclease from Escherichia co& that introduces single polynucleotide chain scission in ultraviolet-irradiated DNA. *I. Biol. Chem. 251, 1438-1445. Sancar, A. & Sancar, G. 15. (1988). DNA repair enzymes. Annu. Rev. Biochem. 57, 29-67. Westbrook, E. M. (1985). Crystal density measurements using aqueous Ficoll solutions. Methods Enzymol. 114, 187-196. Kow,

by W. Hendrickson