AT-rich Islands in Genomic DNA as a Novel Target for AT-specific DNA-reactive Antitumor Drugs
2001; Elsevier BV; Volume: 276; Issue: 44 Linguagem: Inglês
10.1074/jbc.m103390200
ISSN1083-351X
AutoresJan M. Woynarowski, Alex V. Trevino, Karl A. Rodriguez, Stephen C. Hardies, Craig J. Benham,
Tópico(s)RNA and protein synthesis mechanisms
ResumoInterstrand cross-links at T(A/T)4A sites in cellular DNA are associated with hypercytotoxicity of an anticancer drug, bizelesin. Here we evaluated whether these lethal effects reflect targeting critical genomic regions. An in silico analysis of human sequences showed that T(A/T)4A motifs are on average scarce and scattered. However, significantly higher local motif densities were identified in distinct minisatellite regions (200–1000 base pairs of ∼85–100% AT), herein referred to as “AT islands.” Experimentally detected bizelesin lesions agree with these in silico predictions. Actual bizelesin adducts clustered within the model AT island naked DNA, whereas motif-poor sequences were only sparsely adducted. In cancer cells, bizelesin produced high levels of lesions (∼4.7–7.1 lesions/kilobase pair/µm drug) in several prominent AT islands, compared with markedly lower lesion levels in several motif-poor loci and in bulk cellular DNA (∼0.8–1.3 and ∼0.9 lesions/kilobase pair/µm drug, respectively). The identified AT islands exhibit sequence attributes of matrix attachment regions (MARs), domains that organize DNA loops on the nuclear matrix. The computed “MAR potential” and propensity for supercoiling-induced duplex destabilization (both predictive of strong MARs) correlate with the total number of bizelesin binding sites. Hence, MAR-like AT-rich non-coding domains can be regarded as a novel class of critical targets for anticancer drugs. Interstrand cross-links at T(A/T)4A sites in cellular DNA are associated with hypercytotoxicity of an anticancer drug, bizelesin. Here we evaluated whether these lethal effects reflect targeting critical genomic regions. An in silico analysis of human sequences showed that T(A/T)4A motifs are on average scarce and scattered. However, significantly higher local motif densities were identified in distinct minisatellite regions (200–1000 base pairs of ∼85–100% AT), herein referred to as “AT islands.” Experimentally detected bizelesin lesions agree with these in silico predictions. Actual bizelesin adducts clustered within the model AT island naked DNA, whereas motif-poor sequences were only sparsely adducted. In cancer cells, bizelesin produced high levels of lesions (∼4.7–7.1 lesions/kilobase pair/µm drug) in several prominent AT islands, compared with markedly lower lesion levels in several motif-poor loci and in bulk cellular DNA (∼0.8–1.3 and ∼0.9 lesions/kilobase pair/µm drug, respectively). The identified AT islands exhibit sequence attributes of matrix attachment regions (MARs), domains that organize DNA loops on the nuclear matrix. The computed “MAR potential” and propensity for supercoiling-induced duplex destabilization (both predictive of strong MARs) correlate with the total number of bizelesin binding sites. Hence, MAR-like AT-rich non-coding domains can be regarded as a novel class of critical targets for anticancer drugs. minor groove binding drug matrix-associated region stress-induced duplex destabilization quantitative polymerase chain reaction polymerase chain reaction base pair(s) kilobase pair(s) Cellular DNA is not a uniform target for DNA-reactive drugs. At the nucleotide level, drugs recognize and bind short motif(s) of a few base pairs. The location of drug adducts at the genomic level, however, depends on how these short motifs are distributed in larger domains (hundreds of base pairs) that may have distinct structural and functional properties. This aspect, referred to as region specificity, may be critical for the biological outcome of drug action (1Kohn K.W. Hartley J.A. Mattes W.B. Biochem. Pharmacol. 1988; 37: 1799-1800Crossref PubMed Scopus (18) Google Scholar, 2Hartley J.A. Lown J.W. Mattes W.B. Kohn K.W. Acta Oncol. 1988; 27: 503-510Crossref PubMed Scopus (38) Google Scholar). The classical antitumor DNA-reactive drugs currently in the clinic display low sequence specificity, binding virtually indiscriminately to cellular DNA. Not only are such drugs non-region-specific (3Woynarowski J.M. Chapman W.G. Napier C. Herzig M. Juniewicz P. Mol.. Pharmacol. 1998; 54: 770-777Crossref PubMed Scopus (204) Google Scholar, 4Herzig M.C. Trevino A.V. Arnett B. Woynarowski J.M. Biochemistry. 1999; 38: 14045-14055Crossref PubMed Scopus (19) Google Scholar), but they also form most of their adducts with macromolecules other than DNA (5Lawley P.D. Phillips D.H. Mutat. Res. 1996; 355: 13-40Crossref PubMed Scopus (245) Google Scholar). In contrast, certain minor groove-binding agents (MGBs)1 combine a high sequence specificity for AT motifs with a lack of reactivity with non-DNA targets. Bizelesin and adozelesin of the cyclopropylpyrroleindole family are two novel anticancer drugs with these properties (4Herzig M.C. Trevino A.V. Arnett B. Woynarowski J.M. Biochemistry. 1999; 38: 14045-14055Crossref PubMed Scopus (19) Google Scholar, 6Lee C.S. Pfeifer G.P. Gibson N.W. Biochemistry. 1994; 33: 6024-6030Crossref PubMed Scopus (41) Google Scholar, 7Weiland K.L. Dooley T.P. Biochemistry. 1991; 30: 7559-7565Crossref PubMed Scopus (47) Google Scholar, 8Woynarowski J.M. Chapman W.G. Napier C. Herzig M.C.S. Biochim. Biophys. Acta. 1999; 1444: 201-217Crossref PubMed Scopus (22) Google Scholar, 9Woynarowski J.M. McHugh M. Gawron L.S. Beerman T.A. Biochemistry. 1995; 34: 13042-13050Crossref PubMed Scopus (29) Google Scholar, 10Woynarowski J.M. Napier C. Trevino A.V. Arnett B. Biochemistry. 2000; 39: 9917-9927Crossref PubMed Scopus (24) Google Scholar). These sequence-specific cyclopropylpyrroleindole drugs, as well as several other AT-specific MGBs, are currently in phase I clinical trials (11Punt C.J. Humblet Y. Roca E. Dirix L.Y. Wainstein R. Polli A. Corradino I. Br. J. Cancer. 1996; 73: 803-804Crossref PubMed Scopus (32) Google Scholar, 12Weiss G.R. Poggesi I. Rocchetti M. DeMaria D. Mooneyham T. Reilly D. Vitek L.V. Whaley F. Patricia E. Von Hoff D.D. O'Dwyer P. Clin. Cancer Res. 1998; 4: 53-59PubMed Google Scholar, 13Foster B.J. LoRusso P.M. Poplin E. Zalupski M. Valdivieso M. Wozniak A. Flaherty L. Kasunic D.A. Earhart R.H. Baker L.H. Invest. New Drugs. 1995; 13: 321-326Crossref Scopus (29) Google Scholar, 14Fleming G.F. Ratain M.J. O'Brian S.M. Schilsky R.L. Hoffman P.C. Richards J.M. Vogelzang N.J. Kasunic D.A. Earhart R.H. J. Natl. Cancer Inst. 1994; 86: 368-372Crossref PubMed Scopus (41) Google Scholar, 15Shamdas G.J. Alberts D.S. Modiano M. Wiggins C. Power J. Kasunic D.A. Elfring G.L. Earhart R.H. Anticancer Drugs. 1994; 5: 10-14Crossref PubMed Scopus (30) Google Scholar, 16Schwartz G.H. Aylesworth C. Stephenson J. Johnson T. Campbell E. Hammond L. Von Hoff D.D. Rowinsky E.K. Proc. Amer. Soc. Clin. Oncol. 2000; 19: 235aGoogle Scholar, 17Pavlidis N. Aamdal S. Awada A. Calvert H. Fumoleau P. Sorio R. Punt C. Verweij J. Van Oosterom A. Morant R. Wanders J. Hanauske A.R. Cancer Chemother. Pharmacol. 2000; 46: 167-171Crossref PubMed Scopus (30) Google Scholar). Despite the growing interest in novel sequence-specific small molecules, their potential for producing region-specific DNA damage remains largely unexplored. Our recent studies of AT-specific MGBs demonstrated that some, but not all, AT-specific MGBs are able to preferentially damage specific regions of genomic DNA (4Herzig M.C. Trevino A.V. Arnett B. Woynarowski J.M. Biochemistry. 1999; 38: 14045-14055Crossref PubMed Scopus (19) Google Scholar, 8Woynarowski J.M. Chapman W.G. Napier C. Herzig M.C.S. Biochim. Biophys. Acta. 1999; 1444: 201-217Crossref PubMed Scopus (22) Google Scholar, 9Woynarowski J.M. McHugh M. Gawron L.S. Beerman T.A. Biochemistry. 1995; 34: 13042-13050Crossref PubMed Scopus (29) Google Scholar, 10Woynarowski J.M. Napier C. Trevino A.V. Arnett B. Biochemistry. 2000; 39: 9917-9927Crossref PubMed Scopus (24) Google Scholar). Among the small molecules tested, the cyclopropylpyrroleindole drug bizelesin was found to have the greatest potential for producing region-specific damage. In model sequences analyzed, bizelesin adducts were non-randomly distributed and reflected mainly interstrand cross-links at T(A/T)4A sites, although some monoadducts at A(A/T)4A sites are also observed (Fig. 1) (8Woynarowski J.M. Chapman W.G. Napier C. Herzig M.C.S. Biochim. Biophys. Acta. 1999; 1444: 201-217Crossref PubMed Scopus (22) Google Scholar, 9Woynarowski J.M. McHugh M. Gawron L.S. Beerman T.A. Biochemistry. 1995; 34: 13042-13050Crossref PubMed Scopus (29) Google Scholar, 10Woynarowski J.M. Napier C. Trevino A.V. Arnett B. Biochemistry. 2000; 39: 9917-9927Crossref PubMed Scopus (24) Google Scholar). Bizelesin is one of the most cytotoxic compounds ever identified, requiring only a few (<101) drug adducts per cell for cell growth inhibition (4Herzig M.C. Trevino A.V. Arnett B. Woynarowski J.M. Biochemistry. 1999; 38: 14045-14055Crossref PubMed Scopus (19) Google Scholar, 10Woynarowski J.M. Napier C. Trevino A.V. Arnett B. Biochemistry. 2000; 39: 9917-9927Crossref PubMed Scopus (24) Google Scholar). Since typical DNA-reactive drugs must form several thousand lesions/cell to achieve similar effects (4Herzig M.C. Trevino A.V. Arnett B. Woynarowski J.M. Biochemistry. 1999; 38: 14045-14055Crossref PubMed Scopus (19) Google Scholar), one can infer that bizelesin must damage DNA specifically within regions that are crucial for continued cell growth. Based on various observations, including a potent and selective inhibition of DNA replication, we previously hypothesized (10Woynarowski J.M. Napier C. Trevino A.V. Arnett B. Biochemistry. 2000; 39: 9917-9927Crossref PubMed Scopus (24) Google Scholar) that the basis of bizelesin's exceptional potency may be its targeting of AT-rich matrix-associated regions (MARs), domains of critical importance for replication. To investigate the possibility that bizelesin could kill cancer cells by targeting such specific, potentially critical regions, we have developed a new approach that combines bioinformatics and pharmacogenomics with molecular pharmacology. The results demonstrate that bizelesin preferentially targets AT-rich minisatellite loci, in which bizelesin binding sites are highly clustered. The identified domains have a variety of properties known to be associated with MAR function. The long range distribution of drug binding motifs was analyzed using the custom Msearch program (developed by S. C. Hardies) written in Fortran to run under MS DOS, which enables batch processing of multiple GenBank™ entries to tabulate the positions of the exact matches to specified binding motifs in each sequence. The output files were further processed using custom scripts for Excel (Microsoft, Redmond, WA) to generate distribution histograms showing the number of hits in each 250-bp region along the sequence, and to catalog the analyzed sequences and the results of analysis. The majority of the analyzed sequences were selected at random. To ensure that both coding and non-coding regions were adequately represented, contiguous sequences of 30–300 kbp were favored. Entries shorter than 10 kbp were analyzed only when warranted by specific information, such as the reported presence of an AT-rich region. Depending on the motif examined, the total length of human DNA sequences evaluated ranged from 3.1 × 107 bp to 4.3 × 107 bp as indicated in Table I. The distribution parameters (Table I) remained virtually unchanged after the number of analyzed hits exceeded ∼105. For the bizelesin cross-linking motif T(A/T)4A), this number of hits was reached after analyzing ∼107 bp of DNA sequences.Table ILong range in silico sequence analysis for the distribution of possible bizelesin binding sitesDrugMotif searchedPredicted1-aPredicted frequency of specific motifs was calculated based on probability of random occurrence of such motifs (27) assuming the overall DNA composition of 60% AT and 40% GC.Average“Hottest” loci1-bDefined as loci in which peak hits/0.25 kbp exceeded the average peak hits/0.25 kbp in all the sequences analyzed by more than 2.5 × S.D. (rounded).No. of hottest lociRatio hottest loci to averageExpected region specificityhits/0.25kbphits/0.25kbphits/0.25kbploci/MbpBizelesinT(A/T)4A (cross-links)2.922.8 ± 3.640–991.014–34YesA(A/T)4A (mono-adducts)5.838.6 ± 2.660–1290.610–19YesHuman DNA sequences covering 43.0 and 31.4 × 106 bp were analyzed for T(A/T)4A and A(A/T)4A motifs, respectively. The “hits” recorded are exact matches to these motifs on both DNA strands and are given as average values per 0.25 kbp sequence sections (“bins,” cf. Fig. 2).1-a Predicted frequency of specific motifs was calculated based on probability of random occurrence of such motifs (27Kramer J.A. Singh G.B. Krawetz S.A. Genomics. 1996; 33: 305-308Crossref PubMed Scopus (36) Google Scholar) assuming the overall DNA composition of 60% AT and 40% GC.1-b Defined as loci in which peak hits/0.25 kbp exceeded the average peak hits/0.25 kbp in all the sequences analyzed by more than 2.5 × S.D. (rounded). Open table in a new tab Human DNA sequences covering 43.0 and 31.4 × 106 bp were analyzed for T(A/T)4A and A(A/T)4A motifs, respectively. The “hits” recorded are exact matches to these motifs on both DNA strands and are given as average values per 0.25 kbp sequence sections (“bins,” cf. Fig. 2). For some sequences, the distribution of drug binding sites was compared with the distribution of known AT elements such as TATA boxes and polyadenylation signals. In such cases, the sites of potential TATA boxes and polyadenylation signals were determined using an Internet tool (www.itba.mi.cnr.it/webgene/). Sites of drug adducts in naked DNA were determined as described previously using PCR-generated, uniquely end-labeled DNA (9Woynarowski J.M. McHugh M. Gawron L.S. Beerman T.A. Biochemistry. 1995; 34: 13042-13050Crossref PubMed Scopus (29) Google Scholar), except that DNA from untreated CEM cells was used as template (at 6000 cell eq/reaction) and primers designed for the Z79699 AT island (4Herzig M.C. Trevino A.V. Arnett B. Woynarowski J.M. Biochemistry. 1999; 38: 14045-14055Crossref PubMed Scopus (19) Google Scholar). PCR conditions were as described (9Woynarowski J.M. McHugh M. Gawron L.S. Beerman T.A. Biochemistry. 1995; 34: 13042-13050Crossref PubMed Scopus (29) Google Scholar), except that 28 cycles were used and one of the primers was 5′-end-labeled with [γ-32P]dATP (9Woynarowski J.M. McHugh M. Gawron L.S. Beerman T.A. Biochemistry. 1995; 34: 13042-13050Crossref PubMed Scopus (29) Google Scholar). Following the removal of unincorporated primers using Spin Column 200 (Sigma), the labeled DNA (∼0.1 µg/50 µl in 10 mm Tris-HCl, 1 mm EDTA, 100 mm NaCl) was incubated for 4 h with the indicated bizelesin concentrations. After ethanol precipitation to remove unreacted drug, DNA samples were heated for 15 min at 95 °C to convert adducts to breaks. Drug-treated DNA samples were analyzed by sequencing polyacrylamide gel electrophoresis followed by phosphorimaging on a Storm system (Molecular Dynamics, San Jose, CA) . To determine precise adduct positions in the sequence, sequencing reactions using the same primers as the primers for the generation of 5′-end-labeled PCR products were run in parallel on the same gels (9Woynarowski J.M. McHugh M. Gawron L.S. Beerman T.A. Biochemistry. 1995; 34: 13042-13050Crossref PubMed Scopus (29) Google Scholar). The PCR system for the Z79699 AT island used in these experiments generates two related products. The main product (∼85%) is an 859-bp band. A minor product is a 1025-bp band, consistent with the sequence of Z79699 in GenBank™. Cloning and sequencing of the 859-bp product (GenBank™ accession no. AF385609) confirmed that it differs from the 1025-bp Z79699 GenBank™ sequence in the number of repeats in the central segment, with the complete overlaps of extensive flanking areas from both 5′ and 3′ ends (see Fig. 3). Thus, drug adducts in either product would generate identical end-labeled subfragments in those areas of the gel where precise positions of drug adducts can be assessed. Agarose electrophoresis was used to quantitate DNA adducts (following thermal conversion to strand breaks) as described elsewhere (9Woynarowski J.M. McHugh M. Gawron L.S. Beerman T.A. Biochemistry. 1995; 34: 13042-13050Crossref PubMed Scopus (29) Google Scholar). Labeled DNA for these experiments was generated by PCR and included (i) model AT island DNA made as described above for the sites of drug adducts and (ii) a model non-AT island DNA (a 536-bp β-globin fragment generated using the previously described primer system and PCR conditions (3Woynarowski J.M. Chapman W.G. Napier C. Herzig M. Juniewicz P. Mol.. Pharmacol. 1998; 54: 770-777Crossref PubMed Scopus (204) Google Scholar)). These DNAs were either32P-end-labeled exactly as described for the sites of drug adducts or uniformly labeled with [32P]dGTP during the PCR reaction. Following drug treatment and processing as for the sites of drug adducts, samples were analyzed by agarose electrophoresis. After phosphorimaging, the disappearance of full-length materials was quantitated using ImageQuant software (Molecular Dynamics). Based on these data, adduct frequencies were estimated as described (9Woynarowski J.M. McHugh M. Gawron L.S. Beerman T.A. Biochemistry. 1995; 34: 13042-13050Crossref PubMed Scopus (29) Google Scholar). The results from both labeling protocols were pooled for lesion quantitation. In some experiments, unlabeled competitor DNA was added as indicated in Fig. 4C. Human leukemia CEM cells were cultured as described elsewhere (4Herzig M.C. Trevino A.V. Arnett B. Woynarowski J.M. Biochemistry. 1999; 38: 14045-14055Crossref PubMed Scopus (19) Google Scholar). Human colon carcinoma COLO320DM cells (purchased from the ATTC, Manassas, VA) were maintained in RPMI 1640 medium supplemented with 10% fetal calf serum (Life Technologies, Inc.). Prior to drug treatment, cells were prelabeled with [14C]thymidine and the incorporated radioactivity was used to determine cpm/cell ratio (3Woynarowski J.M. Chapman W.G. Napier C. Herzig M. Juniewicz P. Mol.. Pharmacol. 1998; 54: 770-777Crossref PubMed Scopus (204) Google Scholar, 4Herzig M.C. Trevino A.V. Arnett B. Woynarowski J.M. Biochemistry. 1999; 38: 14045-14055Crossref PubMed Scopus (19) Google Scholar, 10Woynarowski J.M. Napier C. Trevino A.V. Arnett B. Biochemistry. 2000; 39: 9917-9927Crossref PubMed Scopus (24) Google Scholar). Following drug treatment for 4 h, cellular DNA was purified using either PureGene or Quiagen kits (3Woynarowski J.M. Chapman W.G. Napier C. Herzig M. Juniewicz P. Mol.. Pharmacol. 1998; 54: 770-777Crossref PubMed Scopus (204) Google Scholar, 4Herzig M.C. Trevino A.V. Arnett B. Woynarowski J.M. Biochemistry. 1999; 38: 14045-14055Crossref PubMed Scopus (19) Google Scholar, 10Woynarowski J.M. Napier C. Trevino A.V. Arnett B. Biochemistry. 2000; 39: 9917-9927Crossref PubMed Scopus (24) Google Scholar). The details of the QPCR assay for region-specific DNA damage have been given elsewhere (3Woynarowski J.M. Chapman W.G. Napier C. Herzig M. Juniewicz P. Mol.. Pharmacol. 1998; 54: 770-777Crossref PubMed Scopus (204) Google Scholar, 4Herzig M.C. Trevino A.V. Arnett B. Woynarowski J.M. Biochemistry. 1999; 38: 14045-14055Crossref PubMed Scopus (19) Google Scholar, 10Woynarowski J.M. Napier C. Trevino A.V. Arnett B. Biochemistry. 2000; 39: 9917-9927Crossref PubMed Scopus (24) Google Scholar). The assay was run under conditions (cycle number, template amounts) that ensure the linearity of the signal as a function of the amount of undamaged template. The amounts of template DNA used in these experiments are expressed as cell eq, based on the14C cpm/cell ratio. Specific PCR primers for various regions and cycling conditions used were as described (4Herzig M.C. Trevino A.V. Arnett B. Woynarowski J.M. Biochemistry. 1999; 38: 14045-14055Crossref PubMed Scopus (19) Google Scholar, 10Woynarowski J.M. Napier C. Trevino A.V. Arnett B. Biochemistry. 2000; 39: 9917-9927Crossref PubMed Scopus (24) Google Scholar), except for the AT island in Z80771. The Z80771 system used 5′-TTCCATTTTATAGTAGAACATGCGTAGA as the upper primer and 5′-AAATGCTGTTGGTATTGTGTTGATAC as the lower primer to generate a product consisting of 928 bp. PCR cycling conditions for the Z80771 system were identical to those described previously for the Z79699/AF385609 system (4Herzig M.C. Trevino A.V. Arnett B. Woynarowski J.M. Biochemistry. 1999; 38: 14045-14055Crossref PubMed Scopus (19) Google Scholar), except that typically 500 and 1000 cell eq of template DNA and 22 cycles were used. Following agarose electrophoresis, autoradiography, and/or phosphorimaging, the signals of the amplified PCR products were quantitated, normalized to signals for DNA from untreated control cells, and converted to lesion frequencies as described previously (3Woynarowski J.M. Chapman W.G. Napier C. Herzig M. Juniewicz P. Mol.. Pharmacol. 1998; 54: 770-777Crossref PubMed Scopus (204) Google Scholar,4Herzig M.C. Trevino A.V. Arnett B. Woynarowski J.M. Biochemistry. 1999; 38: 14045-14055Crossref PubMed Scopus (19) Google Scholar, 10Woynarowski J.M. Napier C. Trevino A.V. Arnett B. Biochemistry. 2000; 39: 9917-9927Crossref PubMed Scopus (24) Google Scholar). For the Z79699 system, which generates two products (see above), the quantitation was based on the data for the shorter, 859-bp variant (main product). The results reported here are from two to four independent experiments, each carried out in triplicate, typically at two different amounts of DNA template. The frequency of lesions was estimated based on the Poisson distribution and normalized per unit length of DNA (3Woynarowski J.M. Chapman W.G. Napier C. Herzig M. Juniewicz P. Mol.. Pharmacol. 1998; 54: 770-777Crossref PubMed Scopus (204) Google Scholar, 4Herzig M.C. Trevino A.V. Arnett B. Woynarowski J.M. Biochemistry. 1999; 38: 14045-14055Crossref PubMed Scopus (19) Google Scholar, 10Woynarowski J.M. Napier C. Trevino A.V. Arnett B. Biochemistry. 2000; 39: 9917-9927Crossref PubMed Scopus (24) Google Scholar). To compare drug effects on diverse regions, lesion frequencies were further normalized by drug concentration to yield lesions/kbp/µm (Fig. 6). Only those drug concentrations producing between 15 and 85% inhibition were included in the second normalization. The consensus repeats for the AT islands with the clusters of bizelesin binding sites were determined using a Tandem Repeats Finder version 2.02 program (18Benson G. Nucleic Acids Res. 1999; 27: 573-580Crossref PubMed Scopus (5550) Google Scholar) made available by its author, Dr. Benson. The thermodynamic stability of the DNA duplex is calculated at each nucleotide position in a given sequence as the melting temperature of the 25-bp oligonucleotide contained within a window of that length that begins at a given position. This calculation is based on a nearest-neighbor algorithm (19Rychlik W. Rhoads R.E. Nucleic Acids Res. 1989; 17: 8543-8551Crossref PubMed Scopus (609) Google Scholar), as implemented in the Oligo program (Molecular Biology Insights, Inc., Cascade, CO), and assumes 120 mm NaCl and 10 mm MgCl2concentrations. Computational techniques to assess the destabilization properties of a DNA sequence that is subjected to a superhelical stress have been described elsewhere (20Benham C.J. J. Mol. Biol. 1992; 225: 835-847Crossref PubMed Scopus (93) Google Scholar, 21Fye R.M. Benham C.J. Phys. Rev. E. 1999; 59: 3408-3426Crossref Scopus (74) Google Scholar). For any user-specified level of superhelicity, the algorithms calculate: (i) the equilibrium probability of denaturation of each base pair along the DNA sequence, and (ii) the incremental free energy G(x) needed to force the base pair at position x to always be separated (22Benham C.J. Proc. Natl. Acad. Sci. U. S. A. 1993; 90: 2999-3003Crossref PubMed Scopus (93) Google Scholar, 23Benham C.J. J. Mol. Biol. 1996; 255: 425-434Crossref PubMed Scopus (92) Google Scholar). The results of such calculations agree precisely with the experimental determinations of the sites that denature and the magnitude of transition at each site at varying levels of superhelicity (20Benham C.J. J. Mol. Biol. 1992; 225: 835-847Crossref PubMed Scopus (93) Google Scholar, 24Benham C. Kohwi-Shigematsu T. Bode J. J. Mol. Biol. 1997; 274: 181-196Crossref PubMed Scopus (120) Google Scholar). Calculations in this study assumed superhelical densities varying from −0.04 up to −0.065, which correspond to moderately low physiological values. Most of the results reported here are for a superhelical density of −0.04 to center on regions that are first to become destabilized. In one case of less destabilized sequence, the calculation used a density of −0.045. To compare various destabilized regions, the integrated SIDD potential was calculated for entire regions within sequence positions stated in Table IIIby integrating the area under the G(x)versus x curve. The integration positions were defined by the cut-off level of G(x) < 7. Based on the previous studies (24Benham C. Kohwi-Shigematsu T. Bode J. J. Mol. Biol. 1997; 274: 181-196Crossref PubMed Scopus (120) Google Scholar, 25Bode J. Kohwi Y. Dickinson L. Joh T. Klehr D. Mielke C. Kohwi-Shigematsu T. Science. 1992; 255: 195-197Crossref PubMed Scopus (393) Google Scholar), distinct MAR domains are expected to exhibit integrated SIDD potential of at least 1500.Table IIIProperties of AT islands targeted by bizelesinGenBank™ entryConsensus repeat sequence and perfect repeat length3-aCopy no. (n) and % match are given in parentheses.AT-rich core sequenceDuplex stability Tm (minimum positions)Integrated SIDD potential (site positions)DNA flexibility twist angle (peak positions)Integrated MAR potential (peak positions)% match°Cdegrees× 10−3Z79699ATATATATTTATATATA TATTTATATTT(A/T)578(98)42.46819.716.419528-mer (n = 19.5, 97)(37360..7880)(37330..37931)(37126..37826)(37300.38000)Z80771TTATATATAAGTATATA TTTATATAATT(A/T)804(85) or [(A/T)23G/C(A/T)3]27(86)45.06591.115.47528-mer (n = 21.6, 91)(3100.3694)(3060..3744)(2876..4050)(3100.3700)X04682TTTTATAATTAAAATAT TTATAATTAAATA(A/T)556(92)49.65002.113.57330-mer (n = 18.2, 93)(14817..15336)(14798..15372)14676..15226)(14700..15100)AC005195ATATATATATATTCC(A/T)646 (87) or [(A/T)13 (G/C)2]43(92)47.86622.616.612815-mer (n = 417.7, 92)(63264..63850)(63205..63888)(63051..64050)(63200..63900)Z72519TA(A/T)318(88)42.82709.516.8512-mer (n = 161.5, 78) (A/T)14(G/C)(A/T)3(33680..33975)(33670..34007)(33501..34201)(33500..34200)18-mer (n = 28.6, 81)Z82900ATATACTAATATATTAA TATAAATAATATATTAAT(A/T)560(92)43.65176.615.91235-mer (n = 11.6, 84)(76163–76678)(76150..76739)(75476..76925)(76100..76800)AF217490TA(A/T)191(91)38.81966.313.7622-mer (78)(147255..147390)(147225..147418)(38951..39026)(146900..14780)X00364No consensus identified[(A/T)8(G/C) (A/T)2(G/C)]24 (73)55.82883.311.57(7418..7572)(7362..7709)(at 7451)(7200..7500)Details on the determination of specific attributes listed in the table are given under “Experimental Procedures.”3-a Copy no. (n) and % match are given in parentheses. Open table in a new tab Details on the determination of specific attributes listed in the table are given under “Experimental Procedures.” The flexibility of DNA in the identified AT islands along with the adjacent sequences was analyzed using the method of Sarai et al. (26Sarai A. Mazur J. Nussinov R. Jernigan R.L. Biochemistry. 1989; 28: 7842-7849Crossref PubMed Scopus (128) Google Scholar), as implemented in the FlexStab program (available at leonardo.ls.huji.ac.il/departments/genesite/faculty/bkerem.htm). This program calculates the flexibility of a given DNA sequence by considering a series of overlapping windows. Within each window, the flexibilities for each dinucleotide step are analyzed to give a window-averaged flexibility. The full GenBank™ entries that contain the indicated AT islands were analyzed using a window length of 250 bp and a shift of 25 bp. To assess the potential of bizelesin-targeted AT islands to function as MARs, an Internet tool (MAR Finder; www.ncgr.org/MarFinder/) developed by Krawetz and co-workers (27Kramer J.A. Singh G.B. Krawetz S.A. Genomics. 1996; 33: 305-308Crossref PubMed Scopus (36) Google Scholar,28Singh G.B. Kramer J.A. Krawetz S.A. Nucleic Acids Res. 1997; 25: 1419-1425Crossref PubMed Scopus (181) Google Scholar) was used. MAR Finder assesses several known properties of MAR sites and then displays their cumulative statistically weighted distribution. Unless noted otherwise, the analysis was run at the standard, default settings (28Singh G.B. Kramer J.A. Krawetz S.A. Nucleic Acids Res. 1997; 25: 1419-1425Crossref PubMed Scopus (181) Google Scholar). MAR potential versus sequence position was plotted after the highest peak in the analyzed sequence is normalized to 1 (i.e. Fig. 7E). MAR Finder tool reports also the peak value used for normalization and the integrated (normalized) MAR potential. To compare MAR potentials of AT islands in various sequences (Table III), the integrated MAR potentials were un-normalized by multiplying these values by the peak value. These data were obtained by analyzing 25-kbp segments of the indicated sequences containing a specific AT island. If the entire sequence was shorter than 25 kbp, then the complete sequence was analyzed. The study reported here consists of three components. First, regions that could potentially be targeted by bizelesin were identified at the genome level by a long range in silico analysis. Next, the in silico predictions were verified by experimentally determining drug-induced lesions, both in naked DNA and in drug-treated cells. Finally, the identified targeted regions were evaluated for attributes that could provide insight into their possible functional roles in the genome. To identify domains likely to be targeted by bizelesin at the genomic level, we examined the distribution of drug binding motifs w
Referência(s)