Biochemical and Biophysical Research Communications 268, 724 –727 (2000) doi:10.1006/bbrc.2000.2116, available online at http://www.idealibrary.com on
3-Methyladenine–DNA Glycosylase I from Escherichia coli—Computer Modeling and Supporting Experimental Evidence Danuta Płochocka, Andrzej Kierzek, Tomasz Obtułowicz, Barbara Tudek, and Piotr Zielenkiewicz 1 Institute of Biochemistry and Biophysics, Polish Academy of Sciences, Pawin´skiego 5a, 02-106 Warsaw, Poland
Received November 17, 1999
TagA (3-methyladenine–DNA glycosylase I) excises 3-methyadenine and 3-methylguanine from alkylated DNA. The structure of this enzyme has not yet been determined experimentally. We propose a threedimensional model of the TagA protein based on the threading algorithm. The model shows that TagA is a mostly ␣-helical protein, in agreement with circular dichroism measurements. None of the eight cysteines present in the TagA sequence forms a disulfide bridge in the model structure, which has also been experimentally verified with the use of Ellman method. © 2000 Academic Press
Various intra- [1] and extracellular [2] compounds are capable of alkylating DNA. The level of spontaneous, nonenzymatic methylations in human tissues is quite high, ranging from one to a few 7-methylguanine residues per 10 7 nucleotides and increases markedly upon exposure to alkylating agents [3]. That is why all organisms developed defense mechanisms toward deleterious effects of nonenzymatic methylation of DNA, specificity of which is directed by DNA glycosylases and methyltransferases. Escherichia coli possesses two different N-methylpurine–DNA glycosylases: (i) 3-methyladenine–DNA glycosylase I (TagA protein) and 3-methyladenine–DNA glycosylase II (AlkA protein). TagA is expressed constitutively and excises only 3-methyadenine and 3-methylguanine from alkylated DNA [4]. AlkA is induced in E. coli as one of enzymes of adaptive response to alkylating agents [4] and has a broad substrate specificity. It repairs a wide specAbbreviations used: AlkA, E. coli 3-methyladenine–DNA glycosylase II; DTNB, dithionitrobenzene; DTT, dithiothreitol; 3MeAde, 3-methyladenine; 7MeAde, 7-methyladenine; 3-MeGua, 3-methylguanine; 7MeGua, 7-methylguanine; SDS, sodium dodecyl sulfate; TagA, E. coli 3-methyladenine–DNA glycosylase I. 1 To whom correspondence should be addressed. Fax: (48) 39 12 16 23. E-mail:
[email protected]. 0006-291X/00 $35.00 Copyright © 2000 by Academic Press All rights of reproduction in any form reserved.
trum of alkylated bases: 3MeAde, 3MeGua, 7MeGua, 7MeAde, O 2-alkylpyrimidines, together with hypoxanthine [4] as well as cyclic DNA adducts—1,N 6ethenoadenine, N 3,3-ethano- and ethenoguanine [5], oxidative lesions like 5-formyluracil [6] and DNA damages induced by clinically used cancer chemotherapeutics [7]. Although both TagA and AlkA catalyze release of 3MeAde from alkylated DNA, they reveal very poor amino acid sequence homology [8]. AlkA protein was crystallized and protein structure was determined [9]. The active center of AlkA enzyme occurs in a wide cleft between two all-␣-helical domains 2 and 3 [10] and electron-deficient methylated bases are recognized through -donor/acceptor interactions with the electron-rich aromatic cleft. The majority of different DNA damages and in addition unmodified bases fulfill the basic requirement of AlkA for base recognition and this probably explains the broad substrate specificity of the protein and the fact that it is able to eliminate unmodified bases from DNA as well [11]. The threedimensional structure of TagA protein is not yet known in contrast to AlkA, the inhibitors for which have been found. 3-Ethyl-, 3-propyl-, 3-butyl- and 3-benzyladenine inhibit TagA glycosylase in micromolar concentrations [12]. It was postulated that hydrophobic interactions are engaged in inhibitor/substrate binding by TagA, since an inhibitory effect of 3-alkyladenine derivatives increased with the size of the alkyl group and was abolished by changing the polarity of the analog: 3-hydroxyethyladenine, at variance to 3-ethyladenine, did not inhibit the activity of Tag A protein [12]. Thus, TagA might belong to the group of DNA-glycosylases represented by the uracil– DNA glycosylase, which possess a tight-fitting pocket for substrate binding [13]. Here we propose the threedimensional structure of the TagA protein as predicted from the threading method. The CD measurements as well the measured number of free sulfhydryl groups in the protein are in agreement with the predicted TagA structure.
724
Vol. 268, No. 3, 2000
BIOCHEMICAL AND BIOPHYSICAL RESEARCH COMMUNICATIONS
FIG. 1.
TagA– eryrthrocruorin sequence alignment according to MATCHMAKER.
MATERIALS AND METHODS Purification of TagA protein. Plasmid pTag10 bearing a tagA gene under the lac promoter was a kind gift of Dr. T. R. O’Connor (Beckman Research Institute, City of Hope National Medical Centre, Duarte, CA). TagA protein was purified from overproducing strain JM105 bearing a plasmid pTag10 as previously described [12]. Estimation of the number of disulfide bridges in TagA protein. The number of free sulfhydryl groups was measured according to Ellman [14]. Standard reaction mixture in the total volume of 1.1 ml contained: 10 g of protein, 50 mM Tris–HCl, pH 8,00, 5 mM EDTA, 1% SDS and 0.04 mg/ml dithionitrobenzene (DTNB). Control samples contained all the reagents except the protein. Absorbance at 412 nm was measured immediately and the concentration of free ⫺SH groups was estimated from the standard curve prepared in the same conditions using reduced glutathione as ⫺SH group donor. Protein concentration was measured according to Bradford [15]. The results are mean from at least 3 independent experiments in which each sample was measured in duplicate. The potential presence of disulfide bridges in TagA protein was investigated by the measurement of free ⫺SH groups after disruption of ⫺SOSO bonds by their reduction with dithiothreitol (DTT) and comparing to the value obtained for nonreduced, but denatured protein. For reduction, 100 g of purified TagA protein was incu-
FIG. 2.
bated with 1% (final concentration) SDS and 0.1 M DTT (final concentration) for 60 min at 37°C. Subsequently, DTT was dialyzed from the protein preparation by several changes of buffer containing 10 mM Tris–HCl, pH 7.6, 2 mM EDTA, 1 M NaCl. The efficiency of dialysis was verified by parallel dialyzing of DTT solution in the same conditions and measuring the concentration of ⫺SH groups by Ellman method after completion of the process. We were also investigating the distribution of ⫺SH groups in relation to their position to the protein surface. The Ellman reaction was performed with the native protein in the absence of SDS. Absorbance at 412 nm was then measured immediately after addition of DTNB and monitored with time. The groups that were closer to the protein surface were expected to react quicker with DTNB, while the groups hidden in the hydrophobic core of the enzyme were presumed to react slower and thus the absorbance at 412 nm should increase with time if ⫺SH groups occurred in central, hydrophobic part of the protein. Molecular modeling. The TagA protein sequence shows no significant homology to any of the proteins having known threedimensional structure as judged from running FASTA on the sequences from the Protein Data bank. This excludes the possibility of model building by homology. The TagA sequence was therefore subjected to a threading algorithm [16] as implemented in the MATCHMAKER module of SYBYL (Tripos Assoc., U.S.A.). The results revealed that the TagA sequence fits the structure of erythrocruorin
Structural model of TagA protein. 725
Vol. 268, No. 3, 2000
BIOCHEMICAL AND BIOPHYSICAL RESEARCH COMMUNICATIONS TABLE 1
Number of Free Sulfhydryl Groups/Molecule Measured by Elman Method in the Native and Reduced TagA Protein Number of ⫺SH groups/molecule Native TagA Time of Ellman reaction (min)
0
30
Denatured TagA
Reduced TagA
0
0
3.86 ⫾ 1.44 7.86 ⫾ 0.44 7.53 ⫾ 1.70 7.74 ⫾ 0.50
FIG. 3. SDS–PAGE electrophoresis of the TagA protein obtained in the final step of purification. Lane 1, molecular weight markers; lane 2, TagA protein.
(Protein Data Bank entry 1ECA) significantly better than any other fold (with the Z-score at ⫺4.4). Sequence alignment (resulting from threading, i.e., fitting TagA sequence into erythrocruorin structure) is shown in Fig. 1. This alignment was the basis for further modeling steps, executed within the HOMOLOGY module of Biosym/MSI software suite. The number, as well as the length of insertions and deletions in the alignment made modeling a hard, but doable task. First, the TagA residues having alignment counterparts in erythrocruorin sequence were mapped into the template (1ECA) structure. The insertions and deletions were modeled using a loop search, i.e., searching (in the PDB database) the fragments of correct sequence length and end-to-end distance. No manual modeling was done at any of these stages. The obtained model structure was subjected to energy minimization using CVFF force field until the maximum derivative was less than 0.001 kcal/mol Å. The energy minimized model was checked by the WHATCHECK (http://www.emblheidelberg.de) program to check the protein geometry. Several small corrections were manually introduced, mainly in aromatic side chain angles, to improve the geometry. The final model was obtained by subsequent energy minimization.
Four of the cysteines, namely Cys36, 59, 179, 180 appear on the protein surface. However, the attempts to dock the substrate/inhibitors with the use of LIGIN program into the protein were unable to discriminate between the power of different inhibitors. Probably the model structure is still too crude for these studies. Measurement of Disulfide Bridges within the TagA Protein Tag A protein was purified to homogeneity, as judged by SDS–PAGE (Fig. 3) according to procedures previously described [12]. Nucleotide sequence of tagA gene shows that the protein contains 8 cysteine residues [8], potentially capable of forming disulfide bridges. We found that none of 8 cysteines was engaged in the formation of disulfide bond, since 8 free ⫺SH groups per molecule were found both in non-reduced, but SDS-denatured TagA and reduced SDS-denatured enzyme (Table 1). Four of sulfhydryl groups in the native protein reacted immediately with DTNB after Ellman reagent addition and remaining four groups only after 30 min of further incubation of protein with DTNB in 37°C (Table 1).
CD spectrum. Circular dichroism spectrum of TagA protein (22 g/ml) was measured on AVIV Model 202 CD spectrometer in 3 mM HEPES-KOH and 1 mM EDTA buffer.
RESULTS Model Structure The model structure of TagA protein is shown in Fig. 2. As results from the model the protein is mostly ␣-helical. There are 8 helices in the model structure comprising the following residues: Asp4 –Trp15, Glu19 –Leu25, Met28 –Gly33, Val43–Lys46, Glu67– Ala93, Ser136 –Arg146, Gly152–Ser158, Met160 – Val171. In total 50% residues are in the helical conformation. The only -structural fragment is the antiparallel -sheet formed between residues Pro105– Phe109 and Val120 –Thr124. None of the cysteines present in protein sequence form disulfide bridges. 726
FIG. 4.
The circular dichroism spectrum of TagA protein.
Vol. 268, No. 3, 2000
BIOCHEMICAL AND BIOPHYSICAL RESEARCH COMMUNICATIONS
This could suggest that 4 cysteine residues occurred closer to the TagA protein surface, while the other 4 residues were rather situated more centrally and needed time for protein unfolding to make access for DTNB. CD Spectrum The spectrum is presented in Fig. 4. The k2d program (http://kal-el.ugr.es/k2d.html) [17, 18] predicts 56% of residues to be in alpha helices and 10% in the -structures for TagA protein. CONCLUSIONS We have used the threading algorithm followed by molecular modeling to predict the structure of 3-methyladenine–DNA glycosylase I from Escherichia coli. According to our results TagA is mostly an ␣-helical protein. The predicted model structure qualitatively agrees with circular dichroism measurements (50 vs 56% of residues in ␣-helical conformation). The model also correctly predicts that none of the eight cysteines form a disulfide bond. REFERENCES 1. Rydberg, B., and Lindhal, T. (1982) EMBO J. 1, 211–216. 2. Lawley, P. D. (1984) in Chemical Carcinogens (Searle, C. E., Ed.), Vol. 1, ASC Monograph 182, pp. 325– 484, American Chemical Society, Washington, DC.
3. Zhao, C., Tyndyk, M., Eide, I., and Hemminki, K. (1999) Mutat. Res. 424, 117–125. 4. Krokan, E. H., Standal, R., and Slupphaug, G. (1997) Biochem. J. 325, 1–16. 5. Saparbaev, M., Kleibl, K., and Laval, J. (1995). Nucleic Acids Res. 23, 3750 –3755. 6. Bjelland, S., Birkeland, N. K., Benneche, T., Volden, G., and Seeberg, E. (1994) J. Biol. Chem. 269, 30489 –30495. 7. Mattes, W. B., Lee, C. S., Laval, J., and O’Connor, T. R. (1996) Carcinogenesis 17, 643– 648. 8. Steinum, A. M., and Seeberg, E., (1986) Nucleic Acids Res. 14, 3763–3772. 9. Yamagata, Y., Kato, M., Odawara, K., Tokumo, Y., Nakashima, Y., Matsushima, N., Yasumura, K., Tomita, K., Ikahara, K., Fujii, Y., Nakabeppu, Y., and Sekiguchi, M. (1996) Cell 86, 311– 319. 10. Labahn, J., Scharer, O. D., Long, A., Ezaz-Nikpay, K., Verdine, G. L., and Ellenberger, T. E. (1996) Cell 86, 321–329. 11. Berdal, K. G., Johansen, R. F., and Seeberg, E. (1998) EMBO J. 17, 363–367. 12. Tudek, B., VanZeeland, A. A., Kus´mierek, J. T., and Laval, J. (1998) Mutat. Res. 407, 169 –176. 13. Savva, R., McAuley-Hecht, K., Brown, T., and Pearl, L. (1995) Nature 373, 487– 493. 14. Ellman, G. L. (1959) Arch. Biochem. Biophys. 82, 70 –77. 15. Bradford, M. M. (1976) Anal. Biochem. 72, 248 –254. 16. Godzik, A., Kolin˜ski, A., and Skolnick, J. (1993) J. Comput. Aided Mol. Des. 7, 397– 438. 17. Andrade, M. A.,Chaco´n,P., Merelo, J. J., and Mora´n, F. (1993) Protein Eng. 6, 383–390. 18. Merelo, J. J., Andrade, M. A., Prieto, A., and Mora´n, F. (1994) Neurocomputing 6, 443– 454.
727
