The sequence specificity of alkylation for a series of benzoic acid mustard and imidazole-containing distamycin analogues: the importance of local sequence conformation
The sequence specificity of alkylation for a series of benzoic acid mustard and imidazole-containing distamycin analogues: the importance of local sequence conformation
Michael D. Wyatt+, Moses Lee1, John A. Hartley*
CRC Drug-DNA Interactions Research Group, Department of Oncology, University College London Medical School, London W1P 8BT, UK and 1Department of Chemistry, Furman University, Greenville, SC 29613, USA
Received March 14, 1997;Revised and Accepted May 2, 1997
ABSTRACT
The covalent sequence specificity of a series of nitrogen mustard and imidazole-containing analogues of distamycin was determined using modified sequencing techniques. The analogues tether benzoic acid mustard (BAM) and possess either one, two or three imidazole units. Examination of the alkylation specificity revealed that BAM produced guanine-N7 lesions in a pattern similar to conventional nitrogen mustards. The monoimidazole-BAM conjugate also produced guanine-N7 alkylation in a similar pattern to BAM, but at a 100-fold lower dose. The diimidazole and triimidazole conjugates did not produce detectable guanine-N7 alkylation but only alkylated at selected sites in the minor groove. Unexpectedly, the alkylation specificity at equivalent doses was nearly identical to that found for the previously reported pyrrole-BAM conjugates. The consensus sequence, 5[prime]-TTTTGPu was strongly alkylated by the triimidazole conjugate in preference to other similar sites including three occurrences of 5[prime]-TTTTAA. Footprinting studies were carried out to examine the non-covalent DNA binding interactions. These studies revealed that the tripyrrole- BAM conjugate bound non-covalently to the same AT-rich sites as distamycin. In contrast, whereas the Im3 lexitropsin bound non-covalently to GC-rich sequences, the triimidazole-BAM conjugate did not detectably footprint to either GC- or AT-rich regions at equivalent doses. The results indicate that the alkylation event is not solely dictated by the non-covalent binding and might be influenced by a unique sequence dependent conformational feature of the consensus sequence 5[prime]-TTTTGPu.
INTRODUCTION
The design of sequence-specific DNA-binding agents has been aided by studying naturally occurring molecules that possess a high affinity for DNA. Netropsin and distamycin are two such examples that have been extensively studied for their DNA binding properties, and have been shown to bind in the minor groove of AT-rich sequences (for reviews, see 1-3). The AT sequence specificity is assumed to be derived from an avoidance of GC base pairs due to the steric clash that would occur between the pyrrole hydrogens that protrude from the concave face of the molecules and the minor groove guanine-2-NH2 amino groups (4). The replacement of the pyrroles with imidazoles removes the steric clash and potentially introduces a hydrogen bond between the inward facing imidazole nitrogen and a guanine-2-NH2 hydrogen. Imidazole-containing `lexitropsins', or information reading oligopeptides, have been shown to bind in the minor groove of GC-rich sequences (5-7).
Figure 1 Structures of distamycin, Im3 lexitropsin, pyrrole-BAM and imidazole-BAM conjugates.
Agents that utilize the pyrrole framework of distamycin as a DNA binding vector that tethers a DNA alkylating species have gained attention recently as DNA structural probes and experimental anti-cancer agents. Tallimustine (FCE 24517) is a benzoic acid mustard derivative of distamycin that is undergoing clinical evaluation (8). Studies on its mechanism of action revealed alkylation in the minor groove at selected adenines while showing no major groove guanine-N7 alkylation, which is characteristic of conventional nitrogen mustards (9). Determination of the sequence specificity of alkylation revealed that tallimustine only alkylated the 3[prime]-adenine within the sequence 5[prime]-TTTTPuA and a single base pair change in the consensus sequence prevented alkylation (10).
A series of lexitropsins that conjugate an aromatic nitrogen-mustard were designed in order to test the parameters involved in tethering alkylating groups to minor groove binders (Fig. 1) (11,12). Specifically, the influence of the lexitropsin on the reactivity and sequence specificity of alkylation by the nitrogen mustard and the influence of the nitrogen mustard on the DNA binding of the lexitropsin could be studied in detail. The series of pyrrole- and imidazole-containing analogues of one, two or three heterocyclic units conjugate benzoic acid mustard (BAM) (Fig. 1). Initial characterization revealed that the cytotoxicity increased for each increase in the number of heterocyclic units, such that the triheterocyclic conjugates were the most cytotoxic in each family. The pyrrole- and imidazole-BAM conjugates cross-linked DNA poorly and cross-linking did not correlate with cytotoxicity (12,13).
A study of the sequence specificity of alkylation for the pyrrole-BAM conjugates revealed a clear enhancement of the alkylation specificity for each increase in the number of pyrrole units compared to BAM (14). The monopyrrole-BAM conjugate alkylated guanines in similar pattern to that seen for BAM in a GC-rich region of DNA. However, minor groove sites were preferentially alkylated in an AT-rich region. The di- and tripyrrole-BAM conjugates only alkylated at selected sites in the minor groove of AT-rich sequences, and the tripyrrole-BAM conjugate showed a strong preference for the sequence 5[prime]-TTTTGPu. Alkylation was demonstrated to be at the purine N3 position in the minor groove (14). MPE footprinting studies showed that the tripyrrole-BAM conjugate non-covalently bound DNA similarly to distamycin and that not all sites non-covalently bound were alkylated.
In the present study the sequence specificity of alkylation for the imidazole-BAM conjugates was determined using a Taq polymerase stop assay. The results demonstrated that the di- and triimidazole-BAM conjugates alkylated selected sites in the minor groove in a similar pattern to that seen for the corresponding di- and tripyrrole-BAM conjugates. MPE footprinting studies were performed to directly compare the non-covalent sequence specificity of the Im3 lexitropsin with the triimidazole-BAM conjugate. The results show that the Im3 lexitropsin non-covalently binds to GC-rich sequences while the triimidazole-BAM conjugate did not detectably footprint to either GC- or AT-rich sequences. The role of non-covalent interactions and local conformation of the consensus sequence in the alkylation event is discussed.
MATERIALS AND METHODS
The synthesis of the imidazole-BAM conjugates has been reported (15). BAM was a gift from Dr Philip Burke, CRC Department of Medical Oncology, Charing Cross Hospital. The regions of pBR322 DNA are referred to by sequence number as listed (15). The GC-rich BamHI-SalI region (375-640) is 265 bp long with a GC content of 65% and 18 tracts of four or more GC base pairs. The 213 bp region defined between base pair number 3090 and 3303 has a GC content of 36% and 11 tracts of four or more AT base pairs. MPE was a gift from Dr Peter Dervan, California Institute of Technology.
Drug-DNA reactions
All drug-DNA reactions were performed in 25 mM triethanolamine, 1 mM EDTA, pH 7.2, at 37°C for 5 h and terminated by addition of an equal volume of stop solution (0.6 M sodium acetate, 20 mM EDTA, 100 [mu]g/ml tRNA). The samples were precipitated with 3 vol ethanol and washed with 70% ethanol. Each drug was examined over a wide dose range initially to determine a drug dose which would ensure single hit kinetics, i.e., that the proportion of the DNA remained undamaged.
Taq polymerase stop assay
The procedure employed was an application of a previously described method (16). Prior to drug-DNA incubation, plasmid pBR322 DNA was linearized with an appropriate restriction enzyme to provide a stop for the Taq polymerase downstream from the primer. This was BamHI for the GC-rich region and PvuII for the AT-rich region. The oligodeoxynucleotide primers were 5[prime]-end labeled prior to amplification using T4 polynucleotide kinase and [[gamma]-32P]ATP (5000 Ci/mmol, Amersham). The labeled primers were purified by elution through Bio-Rad BioSpin Columns. The synthetic primers used for amplification of the BamHI-SalI GC-rich region (16) and the AT-rich region (14) have been reported. Linear amplification of DNA was carried out in a total volume of 100 [mu]l containing 0.5 [mu]g template DNA, 5 pmol labeled primer, 125 [mu]M of each dNTP, 1 U Taq polymerase, 20 mM (NH4)2SO4, 75 mM Tris-HCl pH 9.0, 0.01% Tween, 2.5 mM MgCl2 and 0.05% gelatine. After an initial denaturation at 94°C for 4 min, the cycling conditions were 94°C for 1 min, 60°C for 1 min, 72°C for 1 min, for a total of 30 cycles. After amplification the samples were ethanol precipitated and washed with 70% ethanol.
Methidiumpropyl-EDTA footprinting
5[prime]-Singly end-labeled fragments were prepared as described previously (14). Drugs were incubated with singly end-labeled fragments at room temperature for 2 h in footprinting buffer with 3 [mu]g calf thymus DNA, in a total volume of 30 [mu]l. MPE cleavage conditions were identical to those described previously (7).
Acrylamide gel electrophoresis
Samples were dissolved in formamide loading dye, heated for 2 min at 90°C, cooled on ice and electrophoresed at 2500-3000 V for [sim]3 h on a 80 cm × 20 cm × 0.4 mm, 6% acrylamide denaturing sequencing gel (Sequagel, National Diagnostics). The gels were dried and X-ray film exposed to the gels (Hyperfilm, Amersham). Densitometry was carried out on a BioRad GS-670 imaging densitometer.
RESULTS
Sequence specificity of alkylation
Initially the imidazole-BAM conjugates were examined on a 265 bp GC-rich region of pBR322 DNA (375-640) using the Taq polymerase stop assay. BAM, at a concentration of 500 [mu]M, its IC50 value, produced a pattern of alkylation consistent with that seen for conventional nitrogen mustards such as chlorambucil and melphalan (16). The predominant sites of alkylation were guanines within runs of guanines and in the sequence 5[prime]-TGG (Fig. 2, lane b). These lesions were confirmed as major groove guanine-N7 lesions using a piperidine cleavage assay (data not shown). The monoimidazole-BAM conjugate (Fig. 2, lane c) produced an alkylation pattern that was qualitatively similar to that seen for BAM, but at a 100-fold lower dose (lane b). One exception was an additional band within the sequence 5[prime]-AGC (495) for the monoimidazole conjugate. The di- and triimidazole-BAM conjugates did not detectably alkylate in this region at equivalent doses (Fig. 2, lanes d and e).
Figure 2Taq polymerase gel examining damage to the top strand of the GC-rich region caused by BAM and the imidazole-BAM conjugates. Lane a, control; lane b, BAM, 500 [mu]M; lane c, monoimidazole-BAM, 5 [mu]M; lane d, diimidazole-BAM, 5 [mu]M; lane e, triimidazole-BAM, 5 [mu]M. Drug-DNA incubations were for 5 h at 37°C. The arrow indicates the direction of reading on the template strand.
The 213 bp AT-rich region of pBR322 DNA defined from 3090 to 3303 was examined in order to directly compare the imidazole-BAM conjugates with the previously reported pyrrole-BAM conjugates (14). In the AT-rich region, BAM alkylated guanines with a specificity consistent with the results found in the GC-rich region (Fig. 3, lane b). The monoimidazole-BAM conjugate alkylated in a similar pattern to that seen for BAM again at a 100-fold lower dose, but the intensity of some of the bands differed (Fig. 3, lane d). The sites of alkylation that were common for BAM and the monoimidazole-BAM conjugate are listed on the left side of Figure 3, and include GG sites (3212-13, 3192-93), a GGG site (3184-86), and a G4 site (3147-50). Alkylation at adenine sites in the minor groove were also detectable, but only faintly; the strongest was in the sequence 5[prime]-TTAA (3235).
Figure 3Taq polymerase gel examining damage to the top strand of the AT-rich region. Lane a, control; lane b, BAM, 500 [mu]M; lane c, monoimidazole-BAM, 1 [mu]M; lane d, monoimidazole-BAM, 10 [mu]M; lane e, diimidazole-BAM, 1 [mu]M; lane f, diimidazole-BAM, 5 [mu]M; lane g, diimidazole-BAM, 10 [mu]M; lane h, triimidazole-BAM, 1 [mu]M; lane i, triimidazole-BAM, 5 [mu]M; lane j, triimidazole-BAM, 10 [mu]M. Drug-DNA incubations were for 5 h at 37°C. The arrow indicates the direction of reading on the template strand.
The di- and triimidazole-BAM conjugates alkylate with a markedly different specificity from that seen for BAM or the monoimidazole-BAM conjugate. The diimidazole-BAM conjugate recognizes fewer sites than BAM or the monoimidazole-BAM conjugate and does not strongly alkylate at guanines alkylated by BAM (Fig. 3, lane g). The sites of strongest alkylation for the diimidazole-BAM conjugate were 5[prime]-TTTTGG (3193), 5[prime]-AATGA (3246), and two occurrences of 5[prime]-TTTTAA (3235, 3254, lane g). Additionally, alkylation was evident at 5[prime]-ATTAAAA (3240, 3242) and AAAAGG (3212, lane g). The strongest site of alkylation for the triimidazole-BAM conjugate was the 5[prime]-TTTTGG site (3193), while weaker alkylation was detected at 5[prime]-TTTTAA (3254) and 5[prime]-AATGA (3246) (Fig. 3, lane j). Interestingly, the triimidazole-BAM conjugate showed significantly stronger alkylation at one occurrence of the sequence 5[prime]-TTTTAA compared with another on this strand (3254 but not 3235, lane j).
MPE footprinting studies
The non-covalent DNA binding specificity of the Im3 lexitropsin and the triimidazole-BAM conjugate were compared by MPE footprinting studies. In the GC-rich region of pBR322 DNA described above, the Im3 lexitropsin strongly footprinted two occurrences of the sequence 5[prime]-GGGCT, as shown previously (7), but the triimidazole-BAM conjugate did not detectably footprint at doses up to 100 [mu]M (data not shown). The AT-rich region was examined and Figure 4 shows the results for the top strand. Densitometric scans of selected lanes taken from the gel are represented in Figure 5. It is clear from the densitometric scan of the control MPE-cleaved lane (Fig. 5a) that MPE does not cleave with the same efficiency throughout the strand. The stretch of DNA that showed the greatest variability in cleavage was the 43 bp run from 3230-3273 which contains only five GC base pairs. The cleavage at A or T tracts of four or more base pairs, including 3139-3142, 3170-3174, 3207-3211 and 3188-3191 was affected, but to a lesser degree. The implications of this will be considered in the discussion.
Figure 4 MPE Footprinting gel examining the top strand of the AT-rich region. Lane a, control uncleaved; lane b, G+A marker lane (formic acid); lane c, MPE cleaved (no drug); lane d, distamycin, 100 [mu]M; lane e, tripyrrole-BAM, 50 [mu]M; lane f, tripyrrole-BAM, 100 [mu]M; lane g, triimidazole-BAM, 50 [mu]M; lane h, triimidazole-BAM, 100 [mu]M; lane i, Im3 lexitropsin, 100 [mu]M. Drug-DNA incubations were for 2 h at room temperature.
Figure 5 Densitometeric scans from the gel in Figure 4. (a) Scan from control lane c; (b) scan from Im3 lexitropsin lane i; (c) scan from triimidazole-BAM lane h.
Distamycin produced multiple strong footprints in this AT-rich region (Fig. 4, lane d). The tripyrrole-BAM conjugate (Fig. 4, lanes e and f) inhibited cleavage less efficiently than distamycin at an equivalent dose but the strong footprints produced were at the same AT-rich sequences. The Im3 lexitropsin (Fig. 4, lane i, Fig. 5b) produced three distinct footprints at the sequences 5[prime]-AGGGA (3184), 5[prime]-AAGGA (3210) and 5[prime]-ATCCT (3227), while the remainder of the cleavage pattern was similar to that seen in the control MPE-cleaved lane. The footprints obtained were at sites distinct from those produced by distamycin and the tripyrrole-BAM conjugate and no footprints were observed at the sequence 5[prime]-TTTTGG (3193) and 5[prime]-TTTTAA (3254) sequences which were the sites of alkylation by the triimidazole-BAM conjugate on this fragment. The pattern of cleavage for the triimidazole-BAM conjugate at 100 [mu]M (Fig. 4, lane h; Fig. 5c) was almost identical to that seen for the control lane.
DISCUSSION
The imidazole-BAM conjugates are part of a series of lexitropsins tethering an aromatic nitrogen mustard functionality that were designed and synthesized to study both the effect of a nitrogen mustard on the binding specificity of the lexitropsin, and the effect of the lexitropsin on the reactivity and sequence specificity of alkylation by the nitrogen mustard. There are similarities between the imidazole- and pyrrole-BAM conjugates. The conjugates are significantly more cytotoxic than the parent nitrogen mustard BAM (11,12). There is a substantial increase in cytotoxicity for each increase in the number of heterocyclic units, which is not due to any increase in the efficiency of DNA interstrand cross-linking. The increased cytotoxicity seen for the monoimidazole-BAM compared with the parent mustard might be due to the fact that BAM is uncharged, while the dimethylamino group of the monoimidazole BAM is protonated at physiological pH thus increasing its affinity for DNA. The di- and triheterocyclic-BAM conjugates only alkylated at selected sites in the minor groove, indicating that the sequence specificity of the nitrogen mustard has been substantially altered by the lexitropsin. In agreement with the dipyrrole and tripyrrole conjugates, the triimidazole conjugate strongly alkylated fewer sites than the diimidazole conjugate. Surprisingly, the alkylation specificity of the triimidazole conjugate closely resembled that seen previously for the tripyrrole conjugate (14), in that both the triimidazole and tripyrrole conjugates at equivalent doses show strong alkylation at 5[prime]-TTTTGPu sequences and at one of two occurrences of 5[prime]-TTTTAA. Alkylation by the tripyrrole conjugate was at the purine N3 position in the DNA minor groove (14).
Clearly the conditions for detecting alkylation by the polymerase stop assay (i.e. single hit kinetics) are very different to those employed in the footprinting method (several bound molecules per fragment with a high level of occupancy at each binding site). As a result true quantitative comparisons are difficult. Nevertheless, in the present study at a dose of the tripyrrole conjugate that showed distinct evidence by footprinting of non-covalent binding to AT-rich sequences, no footprinting was observed for the triimidazole conjugate, whereas both agents produce alkylation at 5[prime]-TTTTGPu sequences at equivalent doses.
If the alkylation event was solely dictated by the lexitropsin, then the strongest footprinting sites for the Im2 and Im3 lexitropsins [5[prime]-(G.C)3(A.T) and 5[prime]-GGGCT, respectively (7)] would have been the predicted sites of alkylation for the imidazole-BAM conjugates. The Im3 lexitropsin maintained its non-covalent specificity for GC-rich sequences in the AT-rich fragment and did not footprint at the 5[prime]-TTTTGPu sequence where the BAM conjugates preferentially alkylated. Thus it would not appear that the consensus sequence for alkylation represents a unique site non-covalently bound with high affinity by all minor groove binders of this class. A triimidazole conjugate containing a monofunctional chloroethylbenzamido group alkylated in a different pattern to that seen for the triimidazole-BAM conjugate (17). The `half mustard' conjugate alkylated selected sites in the GC-rich region and did not alkylate at the preferred sequence 5[prime]-TTTTGPu. Thus the inclusion of a benzamido group on the imidazole lexitropsin is not sufficient to alter the specificity of the GC-recognizing lexitropsin back to AT-rich sequences.
The tripyrrole-BAM conjugate footprints at identical sequences to distamycin. Alkylation, however, does not occur at all sites non-covalently bound. The covalent modification of the 3[prime]-purine in 5[prime]-TTTTGPu suggests that the sequence conformation must play a role in the alkylation step. Alkylation, presumably through an aziridinium intermediate, at the nucleophilic adenine-N3 and guanine-N3 groups is not surprising, but, in good agreement with the results found for tallimustine (10), the entire composition of the T4 tract followed by two purines appears crucial. The structure and conformation of a number of synthetic oligodeoxynucleotides containing 5[prime]-T4G, 5[prime]-T4GG and 5[prime]-T4GA sequences have been investigated by high field NMR and X-ray crystallography studies (18-23). Although it is clear that these DNA tracts possess an unusual conformation such as a narrowing of the minor groove, stiffening of the DNA helix, stacking of the base pairs and a disruption of the stacking at the T-G junction, the specific alterations and origins of the conformational changes remain a subject of controversy.
Chemical probes such as hydroxyl radical (24,25), ethidium bromide (26), KMnO4 and DEPC (27,28) have been used to study the local sequence conformation of A-tracts within longer fragments of DNA. There is a clear suppression of cleavage in A tracts by each of the chemical probes and this is indicative of a narrowing of the minor groove. Studies on naturally bent kinetoplast DNA using the same chemical probes found that `a 5[prime]-TG step favors a highly stacked conformation' and the authors proposed that `in the A tract function, 5[prime]-TG-3[prime] acts to accentuate the overall bend of an A tract' (28). It is clear from the MPE control cleavage lane in Figure 5 that cleavage is suppressed at A or T tracts of four or more base pairs, including the site strongly alkylated by the triheterocyclic conjugates, 5[prime]-TTTTGG.
Alkylation by the tripyrrole-BAM conjugate does not occur at all non-covalent binding sites. The Im3 lexitropsin and triimidazole-BAM conjugate do not footprint to the 5[prime]-T4GG site at which alkylation by the triimidazole-BAM conjugate occurs. Therefore, non-covalent interactions do not seem to be the primary determinant of the sequence specificity of alkylation for such BAM-containing conjugates. The local DNA conformation may play a role in catalyzing the alkylation event. Detailed structural studies would be necessary to confirm this hypothesis. If, however, the local sequence conformation of the DNA is playing a role in catalyzing the alkylation event, then this represents an additional consideration in the design of small molecules that bind to and alkylate the minor groove of DNA. The design of molecules that bind to increasingly longer sequences of DNA must take into account the conformational shape and width of the target sequence resulting from its base pair composition.
ACKNOWLEDGEMENTS
M.D.W. acknowledges the U.C.L. postgraduate fund for a predoctoral fellowship. This work was supported in part by the Cancer Research Campaign (J.A.H., SP2000/0401) and the National Cancer Institute (M.L., no.1 R15 CA56901-01).
CiteULike 
Connotea 
Del.icio.us What's this?
