Characterization of a 12-mer duplex d (GGCGGAGTTAGG). cntdot. d

Dale L. Boger, Donald L. Hertzog, Bernd Bollinger, Douglas S. Johnson, Hui Cai, Joel Goldberg, and Philip Turnbull. Journal of the American Chemical S...
0 downloads 0 Views 3MB Size
Chem. Res. Toxicol. 1992,5,167-182

167

Characterization of a 12-mer Duplex d(GGCGGAGmAGG)*d(CCTAACTCCGCC) Containing a Highly Reactive (+)-CC-1065 Sequence by ‘H and 31P NMR, Hydroxyl-Radical Footprinting, and NOESY Restrained Molecular Dynamics Calculations Chin Hsiung Lin,t G . Craig Hill,$ and Laurence H. Hurley**tJ Drug Dynamics Institute, College of Pharmacy, and Department of Chemistry and Biochemistry, University of Texas at Austin, Austin, Texas 78712-1074 Received September 13, 1991

The solution structure of the GC-rich non-self-complementaryDNA 12-mer duplex (I), which contains a (+)-CC-1065 highly reactive bonding sequence 5’AGTTA* (where * denotes the 5’1

2 3 4 5 6 7 8 9

GGCGGAGTTATGG

lOllG 3‘

CCGCCTCAATCC 24 2322212019 18 17 1615 1413

3’4

5‘

I

covalent modification site), has been examined thoroughly by one- and two-dimensional proton and phosphorus NMR spectroscopy, hydroxyl-radical footprinting, and NOESY restrained molecular mechanics and dynamics calculations. The assignments of the nonexchangeable proton resonances (except some of the H5’ and H5” protons due to severe resonance overlap), phosphorus resonances, and the exchangeable resonances (except amino protons of adenosine and guanosine) of this 12-mer duplex have been made. The results show that this 12-mer duplex maintains an overall B-form DNA with all anti base orientation throughout in aqueous solution a t room temperature. Hydroxyl-radical footprinting experiments on a 21-mer sequence that contains this 12-mer duplex used for NMR studies showed that the minor groove is somewhat narrowed a t the 7G-8T and 17A-18C steps, as indicated by the inhibition of cleavage a t these locations. Although both high-field NMR and hydroxyl-radical footprinting experiments supported a bent-like structure for this 12-mer duplex, nondenaturing gel electrophoresis on the ligated 21-mer sequence that contains this 12-mer duplex did not show the abnormally slow migration characteristic of a bent DNA duplex. Analysis of the NMR data sets reveals several local structural perturbations similar to those found on an (A), tract DNA duplex. For example, the existence of a propeller twist was detected within the A-T-rich region for both the 12-mer and the (A), tract DNA duplexes. The 18CH5 aromatic resonance that is directly adjacent to the 3’ side of the 5’TAA segment was significantly shifted upfield with a chemical shift of 5.10 ppm, which is almost within the region normally associated with sugar H3’ protons. The sugar geometries for 18C and 7G, which are located to the 3’ side of the 5’TAA segment, are proposed to be in the neighborhood of C3‘-endo and 01’-endo C3’-endo, respectively. We propose that this unusually upfield-shifted resonance signal for 18CH5 and the average C3’-endo sugar geometry for 18C nucleotide on the 12-mer duplex is connected with the peculiar conformation, possibly a transient kink, within the 5’AC/GT step. The results of the NOESY restrained molecular mechanics and dynamics calculations on the 12-mer sequence reveal two kinks, which are located on either side of the 18C nucleotide that has an average C3’-endo sugar geometry. The two phosphorus resonance signals that are located at the 7G-8T and the 18C-19T steps, where the minor groove is narrowed as indicated by the hydroxyl-radical footprinting experiments, displayed unusual upfield chemical shifts. Also identified were two unusually broadened base protons of the 16A nucleotide and one imino proton belonging to the 9T-16A base pair within the A-T-rich segment. We proposed that this broadening phenomenon is most likely due to a unique internal motion characterized by a rapid local conformational equilibrium between microstates of the 12-mer duplex in aqueous solution a t room temperature. This local conformational flexibility, a transient kink, and the bent-like structure are proposed to play a critical role in the sequence-specific recognition of the DNA duplex by (+)-CC-1065.

-

Introductlon The recent in chemical synthesis of olii techniques (l-g), and godeoxynucleotid-, Qh-field m

molecular mechanics and dynamics calculations ( 10) have permitted the determination Of the Solution Structures Of short fragments of DNA (typically 6-12 bp)’ and their complexes with drugs or proteins. The combination of these techniques is a powerful tool to elucidate the se* To whom correspondence should be addressed. quence-specific interactions of drugs and proteins with t Department of Chemistry and Biochemistry. DNA at the molecular level. The non-self-complementmy Drug Dynamics Institute, College of Pharmacy. O893-228xJ92/2~O5-O16~$O3.00 JO 0 1992 American Chemical Society

f

168 Chem. Res. Toxicol., Vol. 5, No. 2, 1992 Scheme I. The Sequence and Origin of the 12-mer Duplex Used in This Studyo 21 BP

21 BP

21 BP

5 ' iGGGGCGGAGAATGGGCGGPC~GGGCGGAGTTAGGGGCGGG~TGGGCGGAGTTAGGGGCGGG~3 '

1;(*)

1 2 3 4

-:,I

5 6 7 8 9101112

GGCGGAGTTAGG

Lin et al. tional flexibility (17). We propose that this unique local internal motion meets the sequence-dependent conformational flexibility requirement for (+)-CC-1065bonding, Following the covalent bonding reaction with (+)-CC-1065, this 12-mer duplex DNA becomes bent, and the bending locus is located at the 8T to 9T step (18). A detailed analysis of the similarities in NOE' data between the (A), tract sequence and the 12-mer duplex provides critical insight into one of the molecular mechanisms for the sequence-specific recognition of this DNA duplex by (+)CC-1065.

Materials and Methods

CCGCCTCAATCC 24 23 22 21 20 19 18 17 16 15 14 13

indicates the covalent bonding site of (+)-CC-1065.

12-mer duplex shown in Scheme I is contained within the 21 bp repeat sequence of the early promoter region of simian virus 40 DNA. This 21 bp repeat sequence contains the transcription factor Spl binding site 5'GGGCGG (11) as well as the highly reactive bonding sequence 5'AGlTA* (where * denotes the covalent modification site) of the antitumor antibiotic (+)-CC-1065 (12-15). A consensus sequence analysis of the (+)-CC-1065 bonding sites on DNA demonstrates that there are two sets of DNA sequences, 5'PuNTTAl and 5'AAAAA, that are highly specific (15). As part of an effort to elucidate the molecular mechanisms through which the sequence-dependent recognition of DNA by (+)-CC-1065 can be expressed, we have thoroughly characterized the 12-mer duplex shown in Scheme I by high-field NMR, NOESY' restrained molecular mechanics and dynamics calculations, and hydroxyl-radical footprinting experiments. In addition, a detailed comparison has been carried out of the structural features of this 12-mer duplex and a sequence containing an A-tract. The proton NMR assignments for this 12-mer duplex and the deduced DNA conformation provide the basis for the subsequent NMR studies on the covalent bonding of the antitumor antibiotic (+)-CC-1065 within the minor groove of DNA to this sequence.2 Although the 12-mer duplex (Scheme I) exhibits no intrinsic bending as estimated by nondenaturing gel electrophoresis: the duplex has characteristics reminiscent of an (A), tract and a bent DNA structure. For example, the nucleotide adjacent to the 3'side of the 5'TAA segment on the noncovalently modified strand displays an anomalous upfield-shifted H5 proton resonance and has an average C3'-endo sugar geometry. The A.T base pairs within the B'AG'ITA sequence exhibit a propeller twist between base pairs that is similar to that observed on a bent (A), tract DNA at the corresponding nucleotide units (16). In addition, a unique intemal motion characterized by the rapid local conformational equilibrium between microstates was detected for the single nucleotide unit (16A) located one nucleotide removed in the 3' side on the complementary strand of the covalent modification site. We have previously suggested that the primary molecular basis for sequence selectivity is through a sequence-dependent catalytic activation and/or a sequence-dependent conformal Abbreviations: AMBER, assisted model building with energy refinement; bp, base pair; COSY, two-dimensional correlated spectroscopy; DQF-COSY, phase-sensitive double quantum filtered correlated spectroscopy; FID, free induction decay; N, any base; NMR, nuclear magnetic resonance; NOE, nuclear Overhauser effect; NOESY, two-dimensional NOE correlated spectroscopy; Py, pyrimidine; Pu, purine; ROESY, rotating-frame NOE spectroscopy;TSP, 3-(trimethylailyl)propionate. * C. H. Lin and L. H. Hurley, unpublished results, 1991. D. Sun and L. H. Hurley, unpublished results, 1991.

Chemicals. Chemicals for the preparation of the oligomer and NMR buffer and for HPLC purification of the oligomers have been previously reported (18-20). Preparation and Purification of the Non-Self-Complementary 12-mer Duplex. The non-self-complementary d(GGCGGAG"AGG) and d(CCTAACTCCGCC) dodecanucleotides (Scheme I) for one- and two-dimensional high-field NMR studies were synthesized in house on a 10-pmol scale by using the solid-phase cyanoethyl phosphoramidite approach (21) on an Applied Biosystems automated DNA synthesizer, Model 381A. The general procedures for deprotection, annealing, and HPLC purification of the DNA samples have been previously reported (18-20).

High-Field NMR Spectroscopy. All one- and two-dimensional 500-MHz 'H and 202.44-MHz 31PNMR data seta in 90% H20 and 100% D20 buffered solution were recorded on a General Electric GN-500 FT-NMR spectrometer. Proton resonance chemical shifts were recorded in parts per million (ppm) and referenced relative to external TSP' (1mg/mL) in 99.96% D20 or 90% H20/10% D20. The HOD signal was set to 4.75 ppm for each proton spectrum recorded at room temperature. Phosphorus chemical shifts were referenced relative to external 85% H3P0, in DzO. Two-Dimensional NMR Experiments. Phase-sensitive two-dimensional proton NOESY data seta (1)in D20 in 120-, 250-, and 500-msmixing time were collected with a 1.5s pulse repetition time, a sweep width of 4081 Hz, and a 90° pulse width of 14.90 ps. The carrier frequency was placed on the residual HOD resonance, which was irradiated with the decoupler channel. Thirty-two scans were collected for each of the 512 FIDs' in the dimension to generate 2 X 512 X 2048 data matrices. The phase-sensitive two-dimensional proton NOESY spectrum in 90% H 2 0 (200-m~mixing time) was recorded using the States hypercomplex method (1) with a 1.0-5 pulse repetition delay, a sweep width of loo00 Hz,and a 90' pulse width of 27.5 p.. Carrier frequency was set to water signal, and water presaturation was applied. Sixty-four scans were collected for each of the 512 FIDs in the tl dimension to generate 2 X 512 X 2048 data matrices. Two-dimensionalproton rotating-frame NOE (ROESY)' data seta (5, 7) in D 2 0 at 100-ms spin-lock time were collected with a 1 . 5 s pulse repetition time, a sweep width of 4587.2 Hz, and a 90° pulse width of 14.50 ps. Thirty-two scans were collected for each of the 512 FIDs in the tl dimension to generate 2 X 512 X 1024 data matrices. The carrier frequency was placed on the residual HOD resonance using a mixing sequence (pr),with /3 = 32' = 5.15 ps and a T = 51.50 ps spin-lock field (6.13 kHz, corresponding to 5.15-ps 32' pulse width) to eliminate the Hartmann-Hahn artifacts. Two-dimensional proton phase-sensitive double quantum filtered COSY' (DQF-COSY)' spectra in D20 were collected using the standard pulse sequence ( 3 , 4 ) with a 1.5-5 pulse repetition time, a sweep width of 4202 Hz, and a 90' pulse width of 14.25 ps. The carrier frequency was placed on the residual HOD resonance, which was irradiated with the decoupler channel. One hundred sixty-four scans were collected for each of the 256 FIDs in the t l dimension to generate 2 X 256 X 2048 data matrices. For all proton two-dimensional phase-sensitive experiments, a 45O shifted sine bell function was applied to the data seta in both dimensions prior to transformation, and the data seta were zero filled twice in the tl dimension such that the final frequency domain spectra consisted of 1K X 1K data matrices.

Chem. Res. Toxicol., Vol. 5, No. 2, 1992 169

Characterization of a 12-mer Duplex

- Sf Figure 1. One-dimensional proton NMR spectrum (500 MHz, 0.95-8.4 p p m ) of t h e d(GGCGGAGTTAGG1.d(CCTAACTCCGCC) 12-mer duplex in 0.5 mL of D20 buffer containing 10 mM sodium phosphate and 100 mM sodium chloride, pH 6.85 at 23 OC. Two-dimensionalproton magnitude COSY data sets (I)in D20 were collected with a 1.5-s pulse repetition time, a sweep width of 3921 Hz, and a 90° pulse width of 13.25 ps. The carrier frequency was placed on the residual HOD resonance, which was irradiated with the decoupler channel. One hundred ninety-two scans were collected for each of the 256 FIDs in the tl dimension to generate 2 X 256 X 1024 data matrices. The COSY data sets were zero filled twice in the tl dimension such that the final frequency domain spectra consisted of 1K X 1K data matrices. Two-dimensional heteronuclear phosphorus-detected 31P-1H phase-sensitive COSY experiments were recorded according to the published procedures (6)with a 2.0-s pulse repetition time, a sweep width of 896 Hz for phosphorus and 931.0 Hz for proton, and a 90”pulse width of 28.0 pa. One hundred twenty-eight scans were collected for each of the 512 FIDs in the tl dimension to generate 2 X 512 X 512 data matrices. One-dimensional NOE difference experiments were performed at 23 OC with an irradiation time of 0.25 s. Suppression of the water signal was achieved with the standard 1-3-3-1 selective excitation pulse sequence (2),and the interpulse delay was set to 120 ps. Hydroxyl-RadicalFootprinting Experiments. The detailed procedures for the hydroxyl-radical footprinting experiments on the 21-mer sequence containing the 12-mer duplex have been reported previously (18). Molecular Modeling Studies. Molecular mechanics and dynamics calculations were performed using the AMBER^ program, Version 3.0 A (IO),on a Silicon Graphics 4D-25. The dodecanucleotide duplex was constructed in AMBER usingAmott’s B DNA geometry. The sequence was energy minimized in vacuo using a distance-dependent dielectric constant until the rms gradient was less than 0.1 kcal/(mol& The steepest descent was only used for the first 100 steps and for the first 10 steps after each update of the nonbonded air list. The cutoff distance for the nonbonded pairs was 99 and was updated every 100 cycles. NOE data from the two-dimensional NOESY experiments (75 and 250 ms) were c h i f i e d according to intensity (strong, medium, weak, and very weak) and assigned distances of 2.75, 3.50,4.50, and 5.0 A, respectively. An energy penalty of 1kcal/(mol.A) was used to constrain these atom-atom (proton-proton) connectivities. The unusual sugar pucker at 18C nucleotide (average C3’-endo) was also constrained using an energy penalty of 2.5 kcal/mol for each of the five angles (v0-u4). The duplex was minimized to an rmsvalue CO.01 kcal/(mol& The resulting model was subjected to 5 ps of equilibration dynamics followed by 100 ps of production dynamics (constant temperature 300 f 5 K). A constant pressure of 1 atm and 0.002 ps time step with “shake” on was used throughout. The final 100-ps coordinates of the trajectory were averaged and reminimized without NOE constraints and used for analysis.

1

Results The non-self-complementary 12-mer duplex sequence used in this study is part of the 21 bp repeat region contained in the early promoter region of SV 40 virus DNA

iI I

QB

0

Y

k

7. 4 7. 2 Figure 2. Two-dimensional magnitude COSY expanded contour plot of the 12-mer duplex in D20-buffered solution containing 10 mM sodium phosphate and 100 mM sodium chloride, pH 6.85 at 23 “C, demonstrating the through-bond scalar couplings between the thymine methyl resonances (1.1-1.75 ppm) and the thymine H6 protons (7.1-7.5 ppm) of four thymine nucleotides. Cross-peaks A-D are assigned as follows: (A) 15T(H6)-15T(H5); (B)9T(H6)-9T(H5); (C) 19T(H6)-19T(H5); (D) 8T(H6)-8T(H5).

(Scheme I). This 21 bp repeat sequence contains the S p l protein binding site as well as two identical (+)-CC-1065 highly reactive bonding sequences 5/AGTTA* [where * denotes the covalent modification site of (+)-CC-1065]. The sequence and the numbering scheme of this 12-mer duplex are shown in Scheme I. (A) Assignments of the Nonexchangeable Protons in the 12-mer Duplex. The one-dimensional nonexchangeable proton (0.95-8.4 ppm) spectrum of the 12-mer duplex in D20-buffered solution at 23 OC is plotted in Figure 1. It is apparent that the one-dimensional proton NMR spectrum (Figure 1)of this G-C-rich 12-mer duplex is extremely broad and complex. We suspect this is probably due to the rapid interconversion between two or more conformers of the 12-mer duplex at this temperature (see later). Upon adduct formation with (+)-CC-1065,the proton resonance signals become considerably sharpera2 The nonexchangeable proton resonances (except some of the H5’ and H5” sugar protons, due to severe resonance overlap) have been assigned by utilizing two-dimensional NMR techniques: COSY- and DQF-COSY-through-bond scalar couplings and NOESY- and ROESY-through-space connectivities. The thymine H5 (methyl) and H6 base protons can be assigned by analyzing the expanded contour plots of the 12-mer duplex between the 7.10-7.50 ppm and 1.10-1.75 ppm region in both the COSY (Figure 2) and ROESY expansions (Figure 3). In the DNA duplex, each thymine methyl proton exhibits weak four-bond scalar coupling and through-space connectivity to its own H6 proton (€49,221. For the right-handed DNA helix, through-space connectivities can also be detected between adjacent purine H8 and pyrimidine H5/Me base protons in the 5Tu-Py3’ step and between adjacent pyrimidine H6 and pyrimidine H5/Me base protons in the 5’Py-Py3/ step (22). For example, the thymine H5 methyl signal resonating at 1.16 ppm displays a scalar coupling to the thymine H6 proton at 7.13 ppm (cross-peak D, Figure 2) and an ROE connectivity to the aromatic proton at 7.44 ppm (cross-peak D, Figure 3). The thymine H5 signal resonating at 1.60

Lin et al.

170 Chem. Res. Toxicol., Vol. 5, No. 2, 1992 m

-11 LJl

“@------aG

1

Dl

L

7.8

7.2 Figure 3. Two-dimensional phase-sensitive ROESY (100-ms spin-lock time) expanded contour plot of the 12-mer duplex in D20-bufferedsolution containing 10 mM sodium phosphate and 100 mM sodium chloride, pH 6.85 at 23 “C, demonstrating ROE connectivities for the PuH8/PyH6 (7.1-8.3 ppm) to its own sugar 8. 2

8. 0

7.6

7.4

in P

,,A

31

y U U

I m

,

,

,

Q I

I

,

,

,

7 5

8 0 .

,

7 0

B

*‘cQ

H2’ protons (denoted with nucleotide units and numbers) and to the sugar H2” protons of the 5’ neighbor. For the upper strand from 1G to 12G, the ROES are connected with solid lines, and for the lower strand from 13C to 24C, the ROES are connected with broken lines. Cross-peaks A-H are assigned as follows: (A) 8T(H6)-9T(H5);(B) 9T(H6)-9T(H5);(C) 8T(H6)-8T(H5);(D) 7G(H8)-8T(H5); (E) 15T(H6)-15T(H5);(F) 14C(H6)-15T(H5); (G) 18C(H6)-19T(H5); (H) 19T(H6)-19T(H5). ppm exhibits a J coupling to the thymine H6 proton at 7.33 ppm (cross-peak B, Figure 2) and an ROE connectivity to the adjacent 5’-neighbor thymine H6 proton at 7.13 ppm (cross-peak A, Figure 3). Since only 9T in the 12-mer sequence is adjacent to a thymine (in the 5’ direction), the signals resonating at 1.60 and 7.33 ppm were assigned to 9TH5 and 9TH6 protons, and the signals resonating at 1.16 and 7.13 ppm were assigned to 8TH5 and 8TH6 protons, respectively. The aromatic proton resonating at 7.44 ppm was assigned to the 7GH8 proton because of its ROE connectivity to the 8TH5 proton (cross-peak D, Figure 3). (1) Aromatic (PuHS/PyHG) to H1’ Sugar Proton Connectivities. Once the H6 protons of 8T and 9T and the H8 proton of 7G have been assigned, this information can be used as the starting point for sequential assignment of the aromatic to H1’ sugar protons in the appropriate expansion. The two-dimensional phase-sensitive NOESY (250-ms) expanded contour plots demonstrating the through-space connectivities for the PuH8/PyH6 (6.8-8.4 ppm) to the H1’ sugar protons (4.8-6.3 ppm) of the 12-mer duplex are plotted in Figure 4A for the upper strand from 1G to 12G and in Figure 4B for the lower strand from 13C to 24C. For the right-handed DNA duplex, each purine H8/pyrimidine H6 proton exhibits NOES to its own H1’ sugar proton and to the H1’ sugar proton of its 5’ neighbor, except for the 5’4erminal nucleotide which lacks 5’flanking sugar (8,9,22). Thus,it is possible to walk along the complete strand of a double helix DNA and assign all of the PuH8/PyH6 base protons and the H1’ sugar protons. All of the intra- and internucleotide connectivities between the aromatic and the H1’ sugar proton were detected in the 12-mer duplex at room temperature. One unusual feature in this region is that the internucleotide connectivity for the 18CH6 to the H1’ proton of its 5’

L” P

-11

Nil, U U

I

8 0

, e , , 8. 0

,

Q) I

,

, 7.5

,

I

Q ,

,

7.0

Figure 4. Two-dimensional phase-sensitive NOESY (250-ms) expanded contour plots of the 12-mer duplex in D20-buffered solution, pH 6.85 at 23 O C , demonstrating the NOE connectivities for the PuH8/PyH6 (6.8-8.4 ppm) to their own sugar H1’ protons (4.8-6.3 ppm) (denoted with nucleotide units and numbers corresponding to the positions in the 12-mer duplex sequence) and to the H1’ protons of the 5’ neighbors. (A) Upper strand, 1G to 12G, and (B) lower strand, 13C to 24C. Cross-peaks A and B in (A) are assigned as follows: (A) 7G(H8)-6A(H1’);(B) 7G(H8)7G(H1’). Cross-peaks A and B in (B) are assigned as follows: A.l8C(H6)-17A(Hl’);B.l8C(H6)-18C(Hl’). neighbor 17A (cross-peak A in Figure 4B) is significantly stronger than the intranucleotide connectivity for 18CH6 to ita own H1’ proton (cross-peak B in Figure 4B). In order to avoid effects due to spin diffusion, a 75-ms mixing time experiment was also run to c o n f i i unusual intensities for cross-peaks (data not shown). This intensity reversal is uncharacteristic of a normal right-handed B-form DNA and is the indication of an unusual sugar pucker (possibly in the neighborhood of a C3’-endo geometry) for this nucleotide. For the 7G nucleotide that is the complementary partner of 18C, the internucleotide connectivity for the H8 proton to the H1’ proton of its 5’ neighbor (cross-peak A in Figure 4A) is also stronger than the intranucleotide

Chem. Res. Toxicol., Vol. 5, No. 2,1992 171

Characterization of a 12-mer Duplex

Table 1. Chemical Shift Assignments (in ppm) of the Nonexchangeable Protons of the 12-mer Duplex at Room Temperature’ chemical shifts, ppm H8 H6 H2 H5 H1’ H2’ H2” H3’ H4’ H5’,“’‘ G1 7.83 5.63 2.51 2.67 4.78 4.14 3.67 G2 7.82 5.89 2.59 2.67 4.94 4.35 NAb c3 7.24 5.29 5.65 1.87 2.29 4.79 4.12 NA G4 7.76 5.47 2.57 2.67 4.95 4.27 NA G5 7.68 5.52 2.55 2.70 4.96 4.33 NA A6 8.02 7.58 6.01 2.64 2.88 4.98 4.41 4.14-4.21 G7 7.44 5.78 2.38 2.67 4.83 4.34 4.1814.18 T8 7.13 1.16 5.92 1.99 2.48 4.79 4.19 4.0814.08 T9 7.33 1.60 5.64 2.01 2.34 4.83 4.06 4.01-4.08 8.18 7.44 5.89 2.65 2.78 4.98 4.35 4.04-4.10 A10 G11 7.62 5.57 2.47 2.59 4.89 4.33 4.1514.15 G12 7.64 6.07 2.38 2.30 4.55 4.16 NA C13 7.76 5.89 5.92 2.23 2.52 4.63 4.06 3.76 C14 7.62 5.65 5.94 2.08 2.44 4.78 4.15 NA T15 7.36 1.66 5.50 2.04 2.32 4.82 4.08 NA A16 8.23 7.04‘ 5.89 2.73 2.88 5.01 4.37 4.02-4.10 A17 8.09 7.53 5.97 2.61 2.74 4.96 4.43 4.2114.21 5.10 5.62 1.97 2.39 4.49 4.18 4.1314.13 C18 7.19 T19 7.37 1.45 5.97 2.13 2.51 4.81 4.15 4.02-4.10 c20 7.57 5.55 5.87 2.10 2.38 4.81 4.12 4.0414.04 e21 7.40 5.55 5.51 2.03 2.34 4.81 4.05 NA G22 7.88 5.88 2.64 2.69 4.95 4.34 3.9814.06 C23 7.36 5.41 5.94 2.08 2.42 4.81 4.12 NA C24 7.78 5.65 6.18 2.23 2.23 4.52 3.96 4.11/4.11 Sample was dissolved in 0.5 mL of D20 buffer containing 0.1 M sodium chloride and 10 mM sodium phosphate, pH 6.85 at 23 O C . NA, not assigned. Resonances that are upfield shifted relative to signals of other similar protons are underlined.

-I

Q 0)

A

-

0

H1’

Inul

c&-+ E@-

t

------q C 2 3

__________*c3

_-_________________*c1e (99:

0 L”-

e 7 3 -

u

I

,

l

l

l

‘(I’ 1

1

1

O

.

0

€90

9

w o 1

I

I

I

I

I

I

Figure 5. Two-dimensional phase-sensitivityNOESY (250-ms) expanded contour plot of the 12-mer duplex in D,O buffer solution, p H 6.85 at 23 OC, demonstrating the NOE connectivities for the adenine H2 to the sugar H1’ protons and for the cytosine

H5 to the PuHSJPyH6 of the 5’ neighbor. Cross-peaks A-N are identified as follows: (A) 18C(H5)-17A(H8); (B) 3C(H5)-2G(H8); (C) 23C(H5)-22G(H8); (D)14C(H5)-13C(H6); (E) 24C(H5)23C(H6); (F)6A(H2)-6A(Hl’); (G) 6A(H2)-20C(Hl’); (H) 6A(H2)-7G(Hl’); (I) 17A(H2)-17A(Hl’); (J) 17A(H2)-9T(Hl’); (K) lOA(HS)-lOA(Hl’) and lOA(H2)-16A(Hl’); (L) 10A(H2)-11G(Hl’); (M) 16A(H2)-10A(Hl’) and 16A(H2)-16A(Hl’); (N) 17A(H2)-18C(Hl’). Arrowheads denote the cross-strand NOE connectivities.

is expected that each base proton (H8 or H6) will be in close proximity to the H2’ and the H2” protons of its own nucleotide and to that of ita 5’-neighbor nucleotide (23). In addition, if a relatively short spin-lock time is used in the ROESY experiment, each base proton (H8 or H6) should exhibit ROES to its own sugar H2’ proton and to the H2” proton of its 5’-neighbor nucleotide, since they are within 3.5 A. The two-dimensional ROESY (100-ms spin-lock time) expanded contour plot displaying ROE connectivities for each PuH8/PyH6 to the H2’ sugar proton (labeled with nucleotide unit and number corresponding to the position in the 12-mer duplex) and to the H2” protons of ita 5’ neighbor in the 12-mer duplex in D20 is shown in Figure 3 (for the upper strand 1G12G the ROES are connected with solid lines; for the lower strand 13C-24C the ROES are connected with broken lines). The characteristic ROE conectivities for the right-handed B-form DNA are all observed in the 12-mer duplex, except for the 7G and 18C nucleotides. For the 18C nucleotide, the internucleotide ROE from base to the H2” sugar protons of ita 5’ neighbor and the intranucleotide ROE for

- -

weak intensity (Figure 3). In the case of 7G, the internucleotide ROE for base to the H2” sugar protons of its 5’ neighbor and the intranucleotide ROE for base to ita own H2’ sugar proton are of the same intensity (Figure 3). Through-space connectivities should be detected between adjacent purine H8 and pyrimidine H5/Me base protons in the VPu-Py3’ steps (22). A total of four such connectivities in the 12-mer duplex were expected and found 2G3C (crowpeak B, Figure 5), 7 G 8 T (cross-peak D, Figure 3), 17A-18C (cross-peak A, Figure 5 ) , and 2 2 6 23C (cross-peakC, Figure 5). Through-space connectivitiea are also observed between adjacent pyrimidine H6 and

172 Chem. Res. Toxicol., VoE.5, No.2, 1992

Lin et al.

A

‘AI42

8. 0 7.5 7.0 Figure 6. Two-dimensional phase-sensitive NOESY (250-ms mixing time) expanded contour plot of the 12-mer duplex in D20 buffer solution, pH 6.85 at 23 OC, demonstrating the NOE connectivities for the PuH8/PyH6 (7.0-8.4 ppm) to the PuH8/PyH6 (7.0-8.4 ppm) protons. Cross-peaks A-U are assigned as follows: (A) llG(H8)-10A(H8); (B) 10A(H8)-9T(H6);(C) 9T(H6)-8T(H6); (D)8T(H6)-7G(H8): (E) 7G(H8)-6A(H8): (F)6A(H8)-5G(H8): (G) 5G(H8)-4G(HB)i (H) 4G(H8)-3C(H6); (I) 3C(H6)-2G(H8)f (J) 24C(H6)-23C(H6); (I()23C(H6)-22G(H8); (L)22G(H8)21C(H6); (M)21C(H6)-20C( H6) and 20C(H6)-19T( H6); (N) 19T(H6)-18C(H6);(0)18C(H6)-17A(H8); (P)17A(H8)-16A(H8); (Q) 16A(H8)-15T(H6); (R) 15T(H6)-14C(H6); (S) 14C(H6)13C(H6); (T) 16A(H2)-10A(H2); (U)16A(H2)-17A(H2).

pyrimidine H5/Me base protons in the 5’Py-Py3’ steps (22). Of a total of seven such expected connectivities in the 12-mer duplex, five were identified: 8T-9T (croas-peak B, Figure 3), 13C-14C (cross-peak D, Figure 5),14C-l5T (cross-peak F, Figure 3), 18C-19T (cross-peak G, Figure 3), and 23C-24C (cross-peak E, Figure 5), and the remaining two were overlapped: 19T-20C and 20C-21C (overlapped with the cross-peak for H5 to H6 protons of 21C and 20C, respectively). (3) Adenine H2 to H1’ Sugar Proton Connectivities. Each adenine H2 resonance has been assigned on the basis of the NOE connectivities to the H1’ sugar protons to ita own, to ita 3’ neighbor, and to the nucleotide at the 3’ side one base removed on the complementary strand (9,24). For example, in the 12-mer duplex, we detected NOES between 6AH2 and the H1’ protons of 6A (cross-peak F, Figure 5), 7G (cross-peakHI Figure 5 ) , and 20C (croas-peak G, Figure 5). Similar mnnectivities were detected and used to assign all of the remaining H2 protons of 10A, 16A, and 17A (Figure 5). The adenine H2 protons can usually be identified on the basis of their long spin relaxation times in the DNA duplex. For this 12-mer duplex, the H2 protons of 6A and 17A exist as sharp resonances and the 10AH2 proton is somewhat broadened. However, the adenine H2 proton of 16A is unusually broad and occurs at a chemical shift of 7.04 ppm, which is at least 0.4 ppm more

N

I! H 1’

upfield than that of the other H2 protons in the 12-mer duplex (for comparison, see Table I). We suspect that the broadening is due to a unique local internal motion associated with the 9Te16A bp, and the upfield shift is due to the high propeller twist angle of this bp. A detailed explanation of these features is given in the Discussion. (4) Aromatic (PuH8/PyH6) to Aromatic Proton Connectivities. The two-dimensional phase-sensitive NOESY (250-ms) expanded contour plot of the 12-mer duplex in D20-buffered solution demonstrating the NOE connectivities for the aromatic (PuH8/PyH6) to the aromatic protons is shown in Figure 6. The internucleotide

Characterization of a 12-mer Duplex

73

I

Chem. Res. Toxicol., Vol. 5, No. 2, 1992 173

rapid sugar conformational equilibrium (24-26). For the C2’-endo sugar geometry in a B-form DNA, each H2’ and H2” sugar proton is scalar coupled to the H1’ proton in the same nucleotide unit, and the Hl’-H2’ coupling is normally stronger than that for the Hl’-H2’’. In addition, the Hl’-H2’’ through-space connectivities for each nucleotide are always stronger than that of the Hl’-H2’, due to the shorter distance between H1’ and H2” protons (22, 27). As a relatively short spin-lock time was applied in this ROESY experiment, the ROE connectivity of H1’ to H2’ for each nucleotide unit is too weak to be detected (Figure 7B). For 18C the Hl’-H2’ and Hl’-H2” cross-peaks are of similar and weak intensity in the DQF-COSY expansion (Figure 7A), and the H3’-H4’ scalar coupling in the COSY expansion is quite strong (data not shown); both observations suggest an average C3’-endo sugar geometry for this nucleotide unit (27). For all the remaining nontermind nucleotides the couplings/connectivitiesare in accord with an average C2’-endo sugar conformation (Figure 7). (6) Base to H3’ and H4’ Proton Connectivities. The 250-ms NOESY expanded contour plot displaying NOE connectivities for aromatic (7.1-8.3 ppm) to the H3’ and H4’ sugar protons (4.1-5.1 ppm) in the 12-mer duplex in D20 is shown in Figure 8. The assignments of the H3’ sugar protons were accomplished by analyzing the H2’-H3’ through-bond scalar coupling in the COSY (data not shown) and the base to the H3’ proton through-space connectivities in the NOESY expansion (Figure 8) and confirmed by the H2’-H3’ and the H2”-H3’ NOE connectivities in the NOESY expansion (data not shown). The NOE connectivities between the base and the H3’ protons are depicted by nucleotide units and numbers that correspond to the positions of the nucleotides in the 12-mer sequence. For a C3‘-endo sugar pucker, the distance between the base and its own H3’ sugar protons is shorter than the corresponding distances in the C2’-endo pucker (28). The NOE connectivity for the 18CH6 base proton to its own H3’ sugar proton is stronger than for 18CH6 proton to it’s own H2’ proton (data not shown). This observation is associated with an average C3’-endo sugar geometry for the 18C nucleotide. For the 7G nucleotide, which is the complementary partner of 18C, the base to the H3’ sugar protons of its own and to its 5’ neighbor are of equal intensity, which also suggests an unusual sugar geometry for this nucleotide. For the remaining nucleotides the NOE connectivities are in accord with an average C2’-endo sugar conformation. (7) H1’ to H4’ Sugar Proton Connectivities. The two-dimensionalphase-sensitive ROESY (1oO-m~spin-lock time) expanded contour plot of the 12-mer duplex in D20-buffered solution demonstrating the through-space connectivities between the H1’ and the H4’ protons is shown in Figure 9. The intranucleotide ROE connectivities for the H1’ to the H4’ sugar protons are depicted by numbers that correspond to the positions of the nucleotides in the 12-mer sequence. Each H4’ proton assignment, which is made from analyzing the aromatic to the H4’ through-space coupling and the H3’-H4’ scalar coupling, is further confirmed by detection of the unique Hl’-H4’ ROE cross-peaks in this region. All of the nonexchangeable proton chemical shift assignments (except some of the H5’ and H5” resonances) of the 12-mer duplex in D20-buffered solution at 23 “C are listed in Table I. (B) Assignments of the Exchangeable Protons in the 12-mer Duplex. The one-dimensional exchangeable proton spectrum (12.5-14.0 ppm) of the 12-mer duplex in 90% H 2 0 buffered solution, pH 6.85 at 23 OC, is plotted

1

8. 2 8. 0 7 .0 7.6 7 .4 7 .2 Figure 8. Two-dimensional phase-sensitiveNOESY (250 me) expanded contour plot of the 12-mer duplex in D20 buffered solution, pH 6.85 at 23 O C , demonstratingintranucleotideNOE connectivities for the PuH8/PyH6 (6.8-8.40 ppm) to their own sugar H3’ and H4’ protons (4.5-5.1 ppm). The NOE connectivities between the base and the H3’ protons are depicted by nucleotide Units, and numbers correspond to the positions of the nucleotides in the 12-mer sequence. The NOE connectivities for the base to the H3’ and H4’ protons of each nucleotide are connected with solid linea for the upper strand and with broken lines for the lower strand in the 12-mer sequence.

NOE connectivities between aromatic and aromatic resonances are denoted from A to S. The cross-strand and through-strand NOE connectivities for 16AH2 to the adenine H2 protons of 1OA (cross-peak T, Figure 6) and 17A (cross-peak U, Figure 6) were detected. It is quite unusual that a cross-strand NOE connectivity, such as that found between 16AH2 and 10AH2 (cross-peak T, Figure 6), is found to be very strong. In addition, the 16AH8 aromatic proton is significantly broadened relative to other adenine H8 protons (e.g., 6A, 10A, and 17A) in the 12-mer duplex. (5) H1’ to H2’ and H2” Sugar Proton Couplings/ Connectivities. The two-dimensional phase-sensitive DQF-COSY and ROESY (100-ms spin-lock time) expanded contour plots of the 12-mer duplex in D20 buffer demonstrating the couplings/connectivities between the H1’ (5.4-6.3 ppm) and the H2’ and H2” sugar protons (1.8-3.0 ppm) are shown in panels A and B, respectively, of Figure 7. The Hl’-H2’ and Hl’-H2’’ through-bond scalar couplings and through-space connectivities for each nucleotide are labeled with the nucleotide number in Figure 7. The assignments of all the H2’ and the H2” protons were accomplished by analyzing the two-dimensional DQF-COSY, ROESY, and NOESY data sets of this region. For the 12-mer duplex, except for the terminal nucleotides 12G and 24C, the H2” resonance is consistently downfield from that of the H2’ resonance on each nucleotide unit (Figure 7A). For 24C there is an overlap, and for 12G the H2” resonance is upfield from that of the H2’ resonance. This chemical shift reversal for terminal nucleotides that lack structural constraints at one end is very common in B-form DNA duplex and is an indication of

Lin et al.

174 Chem. Res. Toxicol., Vol. 5, No. 2, 1992

H 1'

16 910

P

Q Q

",

/

/

I

I

l

l

I

l

l

I

l

lI

6.2 6.0 5. a 5 . 6 PPM Figure 9. Two-dimensional phase-sensitive ROESY (100-ms spin-lock time) expanded contour plot of the 12-mer duplex in HzO-buffered solution, pH 6.85 a t 23 OC, demonstrating ROE connectivities for the H1' protons (5.1-6.3 ppm) to the H4' protons (3.9-4.5 ppm). The ROE connectivities between the H1' and the H4' protons are denoted with nucleotide numbers corresponding to the positions of the nucleotides in the 12-mer sequence.

I

!

,

I

,

,

,

,

!

12.5

PPM

A

I

B

1

I

13. 0

13. 5

4G 26

H

t

IJ

K

'M

-

A'

I

B

R

PPM Figure 10. The one-dimensional exchangeable proton NMR spectrum (500MHz, 11.8-14.8 ppm, expanded downfield region) of the 12-mer duplex in 0.5 mL of 90% H20/10% D20 buffered solution containing 10 mM sodium phosphate and 100 mM sodium chloride, pH 6.85 a t 23 "C. The broadening imino proton resonance of the 9Ta16A base pair is denoted with an arrowhead. 14.0

13.5

13.0

c

I'

12.5

in Figure 10. A totalof nine resolved exchangeable proton resonances (2G and 4G imino protons are overlapped) were identified in the region from 12.5 to 13.9 ppm. Five resonances were guanosine H1 imino protons resonating between 12.5 and 13.2 ppm (terminal imino protons 1G and 12G are missing at this temperature), and four were thymine H3 imino protons resonating between 13.5 and 13.9 PPmExchangeable proton resonances of the 12-mer duplex in 90% H 2 0 buffered solution have been assigned by analyzing one-dimensional NOE difference (data not shown) and two-dimensional NOESY spectra (Figure 11). Expanded contour plot of the two-dimensional NOESY spectrum (200 ms) of the 12-mer duplex in 90% H 2 0 displaying NOE connectivities for the DNA imino to imino protons is shown in Figure 114the expanded contour plot

1

a

'

~

~

7

~

6

1 PPM

"

~

Figure 11. Expanded contour plots of the two-dimensional phase-sensitive NOESY spectrum (200 ms) of the 12-mer duplex in 0.45 mL of 90% H20/10% DzO buffered solution containing 10 mM sodium phosphate and 100 mM sodium chloride, pH 6.85 a t 23 OC. (A) NOE cross-peaks demonstrating internucleotide connectivities for DNA imino (12.3-13.9 ppm) to imino protons (12.3-13.9 ppm) and (B)NOE cross-peaks demonstrating connectivities for the imino (12.3-13.9 ppm) to aromatic and amino (5.6-8.6 ppm) protons. The cross-peaks A-C (panel A) and D-D' (panel B) are labeled as follows: (A) 4G(imino)-5G(imino); (B) 5G(imino)-lgT(imino); (C) 7G(imino)-8T(imino) and 7G(imino+lST(imino); (D) 5G(imin0)-2OC(NH4~);(E) 5G(imino)-6A(H2);(F)5G(imino)-20C(NH4,); (G) 5G(imino)-20C(H5); (H) YG(imin0)-18C(NH4~);(I) 7G(imino)-GA(H2); (J) 7G(imino)-17A(H2); (K) 7G(imino)-18C(NH4,); (L) 11G(imino)-184C(NH4b); (M)llG(imino)-14C(NH4,); (N) 22G(imh0)-3C(NH4~) and 2G(imin0)-23C(NH4~);(0)22G(imino)-3C(NH4,); (P) 2G(imino)-23C (NH4,) ; (Q) 4G (imino)-2 1C(NH4,) ; (R) 20C(NH4&21C(NH4b); (S) 4G(imino)-21C(NH4,); (T) 15T(imino)-lOA(H2); (U) lBT(imino)-lGA(H2); (V) ST(imino)-17A(H2); (W)9T(imino)-16A(H2); (X) 8T(imino)-17A(H2); (Y) 8T(imino)-16A(H2); (Z) 19T(imino)-6A(H2);(A') 4G(imino)-21C(H5); (B')9T(imino)-lOA(H2); (C') lgT(imin0)-2OC(NH4~);(D') 19T(imino)-SOC(NH4,).

~

1

Chem. Res. Toxicol., Vol. 5, No. 2, 1992 175

Characterization of a 12-mer Duplex

Table 11. Chemical Shift Assignments (in ppm) of the Exchangeable Protons in the 12-mer Duplex at Room Temperature' chemical shifts, ppm base pair Thy-H3 Gua-H1 Ade-H6 Ade-H6 Cyt-HQ CY~-H~~ GlC24 12.98* NA' NA 8.32 6.52 G2C23 13.02 12.99 C3sG22 8.34 6.35 13.02 G4C21 8.51 6.74 12.64 8.28 6.67 G5C20 13.75 A6*T19 NA NA 12.81 7.75f 6.54 G7C18 13.73 NA NA T8sA17 13.68 NA NA T9eA16 13.61 NA NA AlOeT15 12.85 GllC14 8.42 6.96 13.03* NA NA G 12.C 13 "Sample was dissolved in 0.45 mL of 90% water buffer containing 0.1 M NaCl and 10 mM sodium phosphate, pH 6.85 at 23 OC. Chemical shifts were determined at 10 OC. Hydrogen-bonded amino proton. Exposed (non-hydrogen-bonded) amino proton. e NA, not assigned at this temperature. Wpfield-shifted resonance.

tF

t2-23

'H >

,

5 0

1

1 d 1 2 1-1: 1

1

1

1

4 8

1

1

,

1

4 6

1

I

1

4 4

I

I

23-21'I I

4 2

,

,11, I

,

4 0

Figure 13. Two-dimensional heteronuclear phosphorus-detected 31P-1Hphase-sensitive COSY contour plot of the 12-mer duplex

in 0.5 mL of D20 buffer containing 10 mM sodium phosphate and 100 mM sodium chloride, pH 6.85 a t 23 "C.

I

-1, 0

-1.5

1

-2. 0

Figure 12. One-dimensional proton-decoupling 202.44-MHz phosphorus NMR spectrum (-0.95 to -1.90 ppm) of the 12-mer duplex in 0.5 mL of D20-buffered solution containing 10 mM sodium phosphate and 100 mM sodium chloride, pH 6.85 a t 23

"C.

of the imino to base and amino protons is shown in Figure 11B. The NOE connectivities between the thymine imino and the adenine H2 protons of all four A.T base pairs in the 12-mer duplex were identified (cross-peaks Z, X, W, and T for 19T.6A, 8T017A, 9T.l6A, and 15T*10A,respectively, in Figure 11B). The nonterminal guanosine imino protons were assigned on the basis of their NOE connectivities to cytosine hydrogen-bonded and non-hydrogenbonded amino protons, which again display NOE connectivities to cytosine H5 protons in the G C base pairs. For example, the 5G imino proton exhibited NOE connectivities to the hydrogen-bonded and non-hydrogenbonded amino proton of 20C (cross-peaks D and F, respectively, in Figure 11B). Similar NOE connectivities were identified for the remaining nonterminal imino protons (see Figure 11B). The guanosine imino protons in the duplex can also be assigned by tracing NOE connectivities to the adjacent imino protons in both the 5' and 3' directions (Figure 11A). For example, the imino proton of 5G exhibited NOE connectivities to the imino protons of 4G (cross-peak A, Figure 11A) in the 5' direction and to

19T (cross-peak B, Figure 11A) in the 3' direction. The chemical shifts of the imino and amino exchangeable protons of the 12-mer duplex in 90% H 2 0 buffered solution at 23 OC are given in Table 11. The assignments of the imino proton resonances of the 12-mer duplex are also summarized in Figure 10. Significantly, the imino proton belonging to the 9T.16A base pair is broadened (Figure 10, denoted with arrowhead), and the hydrogenbonded amino proton of 18C is unusually upfield shifted (Table 11, denoted with underline). This signal is at least 0.5 ppm further upfield than that for other hydrogenbonded amino protons of cytosines in the 12-mer duplex, suggestive of a reduction in the hydrogen bonding strength within this bp. (C) Assignments of the 31PNMR Resonance Signals in the 12-mer Duplex. The one-dimensional 202.44-MHz proton-decoupled phosphorus spectrum of the 12-mer duplex in D20-buffered solution at pH 6.85 is plotted in Figure 12. Twenty-two phosphorus resonances are dispersed between -1.11 and -1.71 ppm relative to external 85% H3P04. The phosphorus resonance signals have been assigned by analyzing the two-dimensional heteronuclear phosphorus-detected 31P-1H relay COSY data sets, and the results in contour plot are shown in Figure 13. Each phosphorus resonance is scalar coupled to the H3' protons in the 5' direction and to the H4', H5', and H5" protons in the 3' direction. On the basis of the known assignments of these proton resonances (see Table I), the phosphorus resonance signals can be assigned. The chemical shifts of the phosphorus resonances in the 12-mer duplex are given in Table 111. The phosphorus resonances 2Gp3C and

176 Chem. Res. Toxicol., Vol. 5,No. 2, 1992

Lin et al.

Table 111. Chemical Shift Assignments (in ppm) of the Phosphorus Resonance Signals of the 12-mer Duplex at Room Temperaturea phosphorus phosphorus chemical chemica1 step shifts, ppm step shifts, ppm C13-Cl4 -1.28 G1-G2 -1.10 C14-Tl5 -1.49 G2-C3 -1.39 T15-Al6 -1.26 C3-G4 -1.24 A16-Al7 -1.38 G4-G5 -1.15 A17418 -1.24 G5-A6 -1.26 C18-Tl9 -1.62 A6-G7 -1.36 T19-C20 -1.54 G7-T8 -1.58 c20-c21 -1.28 T8-T9 -1.15 C21-G22 -1.18 T9-Al0 -1.36 G22-C23 -1.71 A10-G11 -1.23 C23-C24 -1.13 Gll-Gl2 -1.11

Phosphorus chemical shifts were referenced relative to external 85% H3P04in D20. Sample was dissolved in 0.5 mL of D20-buffered solution containing 0.1 M sodium chloride and 10 mM sodium phosphate, pH 6.85 at 23 "C.

C

Inhibition of *OHcleavage

ii

5'

GGCGGAGTTAG~ CCGCCTCAATCC? 24

tt

13

Figure 15. Laser densitometer tracines of hvdroxvl-radical " " footprinting of the 21-mer duplex sequence d(5'TOGGCGGAGTTA*GGGGCGGGA3')-d( B'CCATCCCGCCCCTAACTCCGC3') [where * denotes the (+NC-1065 ~~.~ modification site] that contains the 12-mer duplex for high-field NMR studies. (A) The upper strand, (B) the lower strand, and (C) the summary scheme of the hydroxyl-radical cleavage in the 12-mer duplex, where (1) indicates the inhibition of hydroxylradical cleavage. ,

~

~

A-tracts have also been demonstrated to produce an undulation in the intensity of hydroxyl-radical cleavage of the backbone of DNA, which is consistent with a compression of the minor groove at the 3' end of the tract (30). . . . . . . . . . . . . . . . . . . . . . . .

5I-G

G

C

G

G

A'G'T

T

A

G

G3'

Purine-purineSteric Clash

Figure 14. Phosphorus chemical shift vs position in 12-mer diagram for the duplex. Solid l i e conneds the 31Pchemical shifts for the phosphate resonances from the 5' to 3' end of the upper strand (from 1G to 12G). Broken lines connect the 31Pchemical shifts for the complementary strand fram the 3' to 5' direction (from 24C to 13C). Arrows indicate the positions of purine-purine

steric clash, and arrowheads denote the unusual upfield-shifted phosphorus resonances in the 12-mer duplex. 22Gp23C, which are associated with the 5'-end region of the upper strand and the 3'-end region of the lower strand in the 12-mer duplex, exhibited upfield 31Pchemical shifts. This is expected because these two phosphate resonance signals are affected by the nearby purine-purine steric clash (the locations of purine-purine steric clash are denoted with arrows in Figure 14) (29). The two phosphorus resonances, 7Gp8T and 18Cp19T,surrounding the 5TAA and 5'CTC junction also display upfield chemical shifts. These upfield shifts cannot be explained by a purinepurine steric clash and are suggestive of sequence-dependent DNA conformational deviations at these locations (see later). (D) Hydroxyl-Radical Footprinting. The results of hydroxyl-radical footprinting of the 21-mer sequence of SV 40 DNA, which contains the 12-mer duplex used for this high-field NMR study, are shown in Figure 15 and exhibit a pattern of inhibition of cleqvage across the minor groove to the 5'side of the covalent attachment site on both strands of DNA. This result suggests that the A-Trich region within the 12-mer duplex exhibits an unusual conformation, and the minor groove is somewhat narrowed at the junction between the 5'TAA and 5'CTC segments.

Discusslon (A) Choice of the 12-mer Duplex for Detailed Structural Studies. During our initial studies designed to determine the DNA sequence specificity of (+)-CC-1065, we had deliberately chosen to study the early promoter region of SV40 DNA as a biologically interesting and relevant DNA target sequence (11). Within this region there are two equivalent (+)-CC-1065 highly reactive adenine-containing sequences 5'AGTTA* (where * denotes the covalent attachment site) that occur as part of the two perfect 21 bp repeats (Scheme I). These repeats are part of the region of 63 bps containing the six GC boxes that are binding sites for the transcription factor Spl. (B) Origin of the DNA Sequence Specificity of we have (+)-CC-1065.In previous publications (13,14,17), suggested that the primary molecular basis responsible for the sequence selectivity of (+)-CC-1065 is a sequence-dependent catalytic activation and/or a sequence-dependent conformational flexibility. Using 'H NMR in combination with "0-labeled water and phosphates on this same 12-mer duplex covalently modified with (+)-CC-1065,we have recently demonstrated the critical involvement of a bridging water molecule between the phenolic proton on the alkylating subunit of (+)-CC-1065 and the anionic oxygen in the phosphate between the 16A and 17A step on the noncovalently modified strand of DNA (20). In addition, a second ordered water molecule associated with the hydrogen-bonded amino protons on N6 of the covalently modified adenine was also identified. This structure has important implications for catalytic activation of the covalent reaction between (+)-CC-1065 and the DNA duplex and, consequently, the molecular basis for sequence selective recog nition of DNA by the alkylating subunit of (+)-CC-1065. We predict that any DNA sequence that lacks such flexibility and/or ability to position an ordered water molecule

Characterization of a 12-mer Duplex

between a phosphate group on the noncovalently modified strand and the H8 phenolic proton of the A subunit will show low reactivity toward (+)-CC-1065. The sequence of the 12-mer used in this study did not provide any immediate clues as to whether conformational flexibility might also be a component partially responsible for the high reactivity of (+)-CC-1065with this sequence. However, the reactivity of 5’AAAAA* sequences toward (+)-CC-1065 (15)and the experimental finding that DNA bending is a consequence of covalent bonding to DNA (18) were suggestive of the involvement of DNA bending in the sequence recognition by (+)-CC-1065. Further evidence for the importance of an A-tract-like bending in the structure of the (+)-CC-1065-DNA adduct was obtained by high-field NMR, gel electrophoresis, and hydroxylradical footprinting (18). In the 12-mer sequence described here, a bending of 17-22O was found following the covalent adduct formation with (+)-CC-1065, and the locus of bending was determined to be at the 8T to 9T step (18). If a propensity to bend is a component of the DNA sequence recognition by (+)-CC-1065, then the 12-mer sequence described here, which has a very high reactivity toward (+)-CC-1065,might be expected to possess an intrinsic capability to form a bent DNA structure. However, nondenaturing gel electrophoresis did not reveal any bent structure? Therefore, the purpose of this investigation was not only to determine the general conformational features of this 12-mer duplex, but also to determine if any evidence existed for this sequence to have a propensity to form a bent DNA structure and/or intrinsic conformational flexibility. A consensus sequence analysis of the (+)-CC-1065 bonding sites on DNA reveals that 5’PuNTTA* and 5‘AAAAA* are highly specific (15). It has been demonstrated by means of gel electrophoresis and electron microscopy that DNA sequences containing (A), tracts, where n 1 3, are bent (31-36). In addition, two-dimensional NMR experiments on these bent DNA sequences reveal a compression of the minor groove from the 5’ to 3’side of the (A), tract (16,26,37-39). Since both 5’PuNTTA and 5’AAAAA sequences are recognized by (+)-CC-1065, they might share a common sequence recognition mechanism through which sequence selectivity can be achieved by the covalent reaction. The transient bending of the DNA duplex might be expected to stabilize the transition state for the covalent reaction; in another words, the reaction of (+)-CC-1065with DNA would entrap the intrinsic transient bend in DNA to form a drug-DNA covalent adduct and would be at least partially responsible for sequence selectivity of (+)-CC-1065. In this transition state for the covalent reaction, the formation of a noncovalent drug-DNA complex will bring the N3 of adenine within covalent bond distance of the reactive cyclopropyl moiety on the A subunit of (+)-CC-1065, and then covalent bonding will take place. This idea gains credibility as we demonstrate that this 12-mer duplex, which contains the highly reactive 5‘AGTTA sequence, has both internal motion and transient kinks. (C) General Features of the 1%-merDuplex. Generally, the two-dimensional proton NMR data for the 12mer duplex in aqueous solution is consistent with a structure within the broad family of B-form DNA, and the 24 nucleotide units are all in the anti base orientation throughout. The characteristic internucleotide and intranucleotide NOE connectivities and intranucleotide J coupling of a B-type DNA duplex are detected. However, examination of the data sets for this 12-mer duplex also reveals anomalous structural deviations such as unusual sugar geometries, upfield chemical shifts, and propeller

Chem. Res. Toxicol., Vol. 5, No. 2,1992 177

twisting of the A-Tbases, which are reminiscent of an (A), tract DNA. These observations might provide significant insight into why this particular 12-mer duplex is a highly reactive bonding sequence for (+)-CC-1065. The strategies to derive information about the sugar pucker geometry from the two-dimensional COSY spectrum have been previously reported (27). Since the experimental digital resolution along the f l and f2 axes of the 2D COSY experiment is limited, the COSY cross-peak intensities are approximately proportional to the magnibetween two nuclei. tudes of the coupling constants Vas) The sugar geometries for 13C and 24C nucleotides located at the ends of the 12-mer duplex are in equilibration between C2’-endo and C3’-endo sugar geometry. This conclusion is based upon the detection of strong H2”-H3’ (data not shown), Hl’-H2’, and Hl’-H2” (Figure 7A) scalar couplings in the COSY expansions. For the 12G nucleotide, the COSY expansion (Figure 7A) displays equal intensity for Hl’-H2’ and Hl’-H2’’ couplings and a strong H3’-H4’ coupling (data not shown). This suggests that the 12G nucleotide sugar geometry is in the range of 01’-endo geometry. For the 18C nucleotide, both the DQF-COSY (Figure 7A) and COSY (data not shown) exhibit weak Hl’-H2’ and Hl’-H2” scalar couplings, as well as a strong H3’-H4’ coupling (data not shown). These data suggest that the sugar geometry for the 18C nucleotide is in the neighborhood of C3’-endo. This average C3’-endo sugar conformation for 18C is independently confiied by detecting a stronger internucleotide NOE connectivity for the 18CH6 base proton to the H1’ proton of ita 5’ neighbor (17A) than the intranucleotide NOE connectivity to its own H1’ sugar proton in a 75-ms NOESY data set (data not shown). In addition, the H6 proton of 18C displays a stronger intranucleotide NOE to its own H3’ sugar proton than to ita own H2’ proton (data not shown). Indeed, for an A-form DNA with C3’-endo sugar geometry, the PuH8/PyH6 base protons are expected to display much stronger internucleotide NOES to the H1’ sugar protons of their 5’ neighbor than the intranucleotide NOES to their own H1’ protons and stronger intranucleotide NOES to their own H3’ sugar proton than to their own H2’ proton (39). The hydrogen-bonded amino proton of 18C is extremely upfield shifted to 7.75 ppm (Table 11). This resonance is at least 0.53 ppm more upfield than those for other nonterminal cytosine hydrogen-bonded amino protons in the 12-mer duplex. This upfield-shifted resonance is associated with a reduction of the hydrogen-bond strength between the hydrogen-bonded C4-NH2of cytosine and 0 4 of thymine in the 7G-18C base pair, which is again due to the peculiar conformation at the junction between 5’TAA and 5’CTC (Scheme 11). In the case of the 7G nucleotide, we detected a strong scalar coupling for H3’-H4’ in the COSY spectrum (data not shown), and the NOE connectivity for the 7GH8 base proton to the H1’ sugar proton of its 5’ neighbor 6A is stronger than to its own H1’ proton (Figure 4A). In addition, equal NOE intensities for 7GH8 to 7GH3’ and 7GH8 to 6AH3’ were detected (Figure 8). On the basis of the above data, the sugar geometry for 7G nucleotide is proposed to be in between 01’-endo and C3’-endo geometry. This deviation from C2’-endo sugar geometry for the nonterminal nucleotides is also very unusual for a DNA duplex in aqueous solution. The H2 proton of 16A was found to have an unusual upfield chemical shift of 7.04 ppm, which is at least 0.40 ppm further upfield than those for other adenine H2 protons in the 12-mer duplex (see Table I). This upfield chemical shift is most likely due to the ring current

178 Chem. Res. Toxicol., Vol. 5, No. 2, 1992

Lin et al. Scheme 11. Summary Figure of the Similarities in Structural Features of the 12-mer Duplex and an A-Tract-Containing Sequence Propeller Twisted

!, GGCGGAGTTAGG 3 1

5

7u,

12

11)

=proton upfield shifted

(Kink)

-

Internal Motion

GGCTTTTTTGCG 3 3’ cy* AAAAAACGC 5

5

/

II

Ropciicr Twisted

Unusual Conformation

HE&ton upfield shifted

a. o 7 5 7 0 PPM Figure 16. One-dimensional proton NMR spectra (500 MHz, 6.90-8.30 ppm, expanded downfield region) of (A) the (+)-CC1065-12-merduplex adduct in 0.5 mL of DzO-bufferedsolution containing 10 mM sodium phosphate and 100 mM sodium chloride, pH 6.85 at room temperature, and the 12-mer duplex in 0.5 mL of DzO-bufferedsolution containing 10 mM sodium phosphate and 100 mM sodium chloride, pH 6.85 (B) at room temperature, (C) at 20 “C, and (D) at 15 “C. The H2 protons of adenosines are labeled with nucleotide numbers, and the H8 protons are labeled with nucleotide units and numbers corresponding to the positions in the 12-mer duplex. The broadened 16AH8 and 16AH2 resonances at different temperatures are denoted with arrowheads.

shielding effect resulting from base-base stacking of the adjacent nucleotides. The results of the NOESY restrained molecular mechanics and dynamics calculations (see later) on the 12-mer duplex indicate that it is the adenine (10A) located to the 5’ side of 16A on the complementary strand that is the likely candidate for causing this unusual upfield chemical shift of the 16AH2 resonance. (D) Evidence for an Unusual Structure within the (+)-CC-1065Binding Sequence in the 12-merDuplex Which Is Suggestive of a Bent DNA Structure. (1) Hydroxyl-Radical Footprinting. The results of hydroxyl-radical footprinting of the 21-mer sequence that contains the 12-mer duplex for this NMR study are consistent with an unusual conformation within the A-T-rich region and especially at the junction between 5’TAA and 5’CTC (Figure 15). The inhibition of hydroxyl-radical footprinting cleavage observed for this 21-mer sequence indicates the minor groove is narrowed at the junction between 5’TAA and 5’CTC segments. (2) Proton and 31PNMR Experiments. (a) Unique Internal Motion. The one-dimensional proton NMR spectrum expanded downfield region (7.0-8.4 ppm) of the 12-mer duplex in D20displays a broad 16AH8base proton, as well as a broad and upfield-shifted 16AH2 proton (see Figure 16B-D; resonance signals denoted with arrowheads). For the exchangeable protons, the imino proton belonging to the 9Tq16A base pair is also significantly broadened relative to that of the other thymine imino protons in the 12-mer duplex (see Figure 10; resonance signal denoted with arrowhead). We propose that this signal broadening is due to a unique internal motion characterized by a rapid local conformational equilibrium (around 16A) between microstates of the 12-mer duplex

in aqueous solution at room temperature (40). For comparison, the expanded downfield region of the identical 12-mer sample at 20 and 15 “C is plotted in Figure 16, parts C and D, respectively. At lower temperatures, the internal motion is diminished such that the 16AH8proton is of the same line width as that for the other adenine H8 protons in the 12-mer sequence. Although the rapid local conformational equilibrium between microstates will affect the line width of the resonances, the overall pattern of the NOE connectivities for the DNA duplex is not expected to change (39). Indeed, the sugar geometry for 16A is still within the neighborhood of a C2’-endo geometry and thus is not affected by the local conformational equilibrium. We propose that this rapid local conformational equilibrium between microstates, or sequence-dependent conformational flexibility, may be important for the sequence-specific recognition of this 12-mer duplex by (+)-CC-1065. Significantly, upon adduct formation with (+)-CC-1065, the H8 and H2 protons of 16A nucleotide in the (+)-CC1065-12-mer duplex adduct become significantly sharper (Figure 16A), as would be expected if the conformation flexibility propmed to be required for the covalent reaction is lost following the adduct formation.2 (b) Phosphorus Chemical Shifts. It has been previously shown that the chemical shifts of the 31Presonance signals of DNA duplexes vary in response to local sequence-specific distortions in the duplex geometry (41). The arrows in Figure 14 indicate two 5’Py-Pu3’ steps (purine-purine steric clash) in the 12-mer duplex. The chemical shifts of these resonance signals that are associated with these steps in the 12-mer duplex follow the downfield shift direction as predicted on the basis of previous 31PNMR studies on synthetic oligomers (29). In addition, the two 31PNMR resonance signals (7Gp8T and 18Cp19T) that are in proximity to the B’TAA and B’CTC junction (Figure 14) also display unusual upfield chemical shifts, suggesting that local sequence-specific distortions in the duplex geometry exist in this junction region. This result is also consistent with an unusual local structural distortion at the 5’AC/GT step (see later) that is demonstrated independently by hydroxyl-radical footprinting and proton two-dimensional NOESY experiments (see before). (c) Propeller Twisting of the A.T Base Pairs. It has been demonstrated by two-dimensional NMR (16,26)and X-ray diffraction studies (42,43)that the A-T-rich region of an (A),-tract DNA is highly propeller twisted. The A-T-richregion of the 12-mer duplex was discovered to be propeller twisted by monitoring the NOE connectivities

Characterization of a 12-mer Duplex

Chem. Res. Toxicol., Vol. 5, No. 2, 1992 179

between adenosine H2 protons of adjacent base pairs (44). The NOE connectivities between 17AH2 and 16AH2 (cross-peaks U, Figure 6) are significantly weaker than between 16AH2 and 10AH2 (cross-peaks T, Figure 6). This indicates that the distance from the (dT8-dT9). (dA16-dA17) step to the (dT9-dAlO).(dT15-dA16)step is reduced. The reduction of distance reflects the consequences of base-pair propeller twisting for the right-handed DNA duplex (44), in which the distance between H2 protons on adjacent adenosines decreases in the order (dTS-dTS).(dAlS-dAl7) > (dT9-dAlO).(dT15-dA16). (d) Upfield Chemical Shifts. The chemical shifts of the nonexchangeable proton resonance signals of the 12mer duplex in D,O solution at room temperature are listed in Table I. The resonance signal of the H5 base proton of 18C occurs at 5.10 ppm, which is significantly upfield to the expected position and is almost within the region normally associated with H3’ sugar proton^.^ A ring current shielding effect resulting from the base-base stacking of the adjacent nucleotides is the most likely cause for such a large upfield chemical shift (45). Since 19T is located on the 3’ side of 18C and is further away from the H5 base proton of 18C in right-handed DNA duplex, this base is unlikely to be responsible for this upfield shift. This leaves 17A, which is on the 5’ side of 18C, as the remaining candidate that could produce such a strong intrinsic ring current effect, and this base is therefore likely to produce the observed upfield chemical shift of the 18CH5 proton. A strong base-base NOE connectivity between the 18CH6 and 17AH8 protons (cross-peak 0, Figure 6), but only a weak NOE connectivity between H6 base protons of 18C and 19T (cross-peak N, Figure 61, was observed in the 2D NOESY expansion. Therefore, these NOESY data are also consistent with the notion that the 17A nucleotide located on the 5’ direction of 18C produces a greater base-base stacking and thus is responsible for the upfield chemical shift of the 18CH5 proton. We suggest that this unusual upfield-shifted resonance signal for 18CH5 and the proposed average C3’-endo sugar geometry for 18C nucleotide on the 12-mer duplex are connected with the peculiar conformation at the junction between the 5’TAA and 5’CTC segments (see later). A similar upfield-shifted resonance has previously been reported for the guanosine H8 base proton, which is located adjacent to the 3’ side of the (A), tract on a bent 12-mer duplex d(CGCAAAAAAGCG).d(CGCTTTTTTGCG)(Scheme 11). It was suggested that this upfield-shifted resonance was due to a junction between the (A), tract and the sequence to the 3’ side (16). (E) Gel Electrophoresis of the 21-mer Sequence That Contains the 12-mer Duplex. A 21-mer sequence, which contains the 12-mer duplex utilized for one- and two-dimensional proton and phosphorus N M R studies, did not exhibit anomalous gel migration and, therefore, DNA bending when the monomers were ligated into multimers and subjected to nondenaturing gel electrophoresis at room temperat~re.~ It has recently been reported that DNA sequences that contain 5’AC/GT or 5’CA/TG steps exhibit anomalous high gel mobility (low RL values5) as compared to the predicted RLvalues (46). Such an unusually fast migration in the gel electrophoresis is suggestive of a structural change in DNA within the 5’AC/GT or 5’CA/TG steps.

The formation of a “transient kink” at certain weak sequence elements, such as the 5’AC/GT or 5’CA/TG steps, was suggested to be the probable basis for the structural change and associated fast migration (46). Larsen and co-workers have also published a paper (47) on the X-ray crystal structure of a dodecamer duplex containing a 5’CA/TG step in which the twist angle within this step is smaller (i.e., unwound) than that of the other steps on this dodecamer duplex. Furthermore, they have speculated that the 5’CA step could be “bistable” (or even “multistable”), so that the 12-mer environment forces it to be unwound. Similarly, we propose that the unusual average C3’-endo sugar geometry for 18C and the associated conformational and dynamic changes, including narrowing of the minor groove in the AT region and rapid local conformational flexibility at 16A, are related to a transient kink at the 5’AC/GT step in the 12-mer studied here. It has also been suggested that, in a right-handed B-form DNA, a kink of about 40’ is produced by the change of sugar pucker from C2’-endo to C3’-endo, and a -10’ unwinding of adjacent base pairs occurs at the kink site (23). On the basis of the two-dimensional COSY and NOESY results reported here, the 18C nucleotide that is contained within the 5’AC/GT step is proposed to have an average C3’-endo sugar geometry, and the hydrogen-bonded amino proton belonging to the 18C nucleotide of the 7G-18C bp is shifted upfield. The unusual phenomenon observed through the NOESY experiments, hydroxyl-radical footprinting, and gel electrophoresis can best be explained by the occurrence of a “transient kink” within the 5’AC/GT step. Indeed, the NOESY restrained molecular mechanics and dynamics calculations on the 12-mer duplex reveal two major kinks that are located on either side of the 18C nucleotide (see later). (F) Molecular Modeling Studies. A series of NOESY restrained molecular mechanics and dynamics calculations on the 12-mer duplex were carried out using the AMBER program, Version 3.OA (lo),based upon a total number of one hundred twenty-seven NOE constraints (see supplementary material, Table S1) derived from a 250-ms NOESY data set. The result of NOESY constrained molecular mechanics and dynamics calculations on this 12-mer duplex indicates that the 12-mer duplex is bent toward the minor groove with a bending angle of about 20’. In addition, the 12-mer duplex contains two major kinks. One, which is toward the minor groove, is located at the 5’AC/GT step, and the other, which is also toward the minor groove, is located at the 5’CT/GA step on the noncovalently modified strand (Figure 17, top). On either side of the 18C nucleotide that is proposed to have an average C3‘-endo sugar pucker! the backbone is distorted, being more compressed toward the major groove on the 5’ side (Le., between 17A and 18C) and elongated toward the major groove on the 3’ side (i.e., between 18C and 19T), such that the aromatic to aromatic proton distance between the 17A and 18C nucleotide is decreased, and between the 18C and 19T nucleotide, significiantly increased. This is in accord with the NOE data described here (see before). These kinks around the 18C nucleotide bring the 18CH5 proton under the shielding cone of the aromatic ring of 17A (the H5 base proton of 18C is denoted with a red arrow in Figure 17, bottom), explaining the upfield

Following the covalent bonding reaction with (+)-CC-1065 within the minor groove of the 12-mer duplex, this resonance was shifted further upfield from 5.10 to 4.86 ppm,* presumably due to the increasing shielding effect of the purine ring currents of 17A nucleotide. RL value is defined as the apparent size divided by the actual size of each DNA sequence.

Short molecular dynamics runs (5, 10, 20 ps) showed that it was necessary to constrain 18C as C3’-endo in order to maintain the kink and proper sugar pucker. However, at 100 ps molecular dynamics stimulation without using the angle constraints for C3’-endo at 18C, we found a structure similar to that proposed; Le., an O4’-exo for 18C and a kink at the 17A-18Cf7G-8T step.

Lin et al.

180 Chem. Res. Toxicol., Vol. 5, No. 2, 1992

!

Figure 17. Stereoscopic diagrams generated from the restrained molecular dynamics calculation of the 12-mer duplex (for sequence see Scheme I): (top) full length in space-filling model; and (bottom) expanded 6 bp region 5’GAGTTA/5’TAACTC. The arrows in both the top and bottom panels indicate the average C3’-endo sugar pucker of 18C. Two upfield-shifted 31Presonances located a t the 7G-8T step and the 18C-19T step are highlighted in yellow color. Table IV. Summary Table of the Helical Twist (or Repeat) Angles for the d(GGCGGAGTTAGG) d(CCTAACTCCGCC) 12-mer Duplex base stem helical twist angle, deg lG-24C/2G-23C 2G-23C/3C-22G 3C-22G/4G-21C 4G-21C/ 5G-2OC 5G-20C/6A-l9T 6A-l9T/ 7G-18C 7G-18C/8T-l7A 8T-l7A/ 9T-16A 9T-16A/lOA-l5T 10A-15T/llG-l4C llG-14C/12G-I3C

33.0 31.9 31.3 28.8

30.0

25.0” 26.00 28.0 30.3 30.2 32.7

aUnwound base pair steps are underlined.

shift of this proton resonance in the 12-mer duplex. Furthermore, the unusually upfield-shifted phosphorus resonance 7Gp8T on the upper strand (covalently modified strand) and 18Cp19T on the lower strand (noncovalently modified strand) can also be explained by the conformational disruption associated with the kinks at these steps. The unusually upfield-shifted phosphorus resonances 7Gp8T and 18Cp19T are denoted in yellow in Figure 17, top. In this space-filling model, the dislocation of these two phosphorus resonances is evident, especially for 18Cp19T (Figure 17, top). Last, the 18C nucleotide is displaced into the helix axis toward the major groove, resulting in an unwinding at the 5’AC/GT step and a

destacking at the 5’CT/AG step (for helical twist angle between base pair steps, see Table IV). As a result, the base-base NOE connectivities between the 18C and 19T (cross-peak N, Figure 6) are relatively reduced when compared with the adjacent base step between the 17A and 18C (cross-peak 0, Figure 6). Likewise, the upfield shift of the hydrogen-bonded amino proton of 18C, and thus the reduction of the hydrogen bond strength, can be explained by this base displacement. This kind of base displacement toward the major groove has previously been reported on the amsacrine-d(CGCG)2 complex (48). The summary table for the helix twist (or repeat) angles of the 12-mer duplex is given in Table IV.

Summary and Conclusions In summary, the nonexchangeable proton (except some of the H5’ and H5” protons), phosphorus resonance signals, and most of the exchangeable proton resonance signals of the 12-mer duplex have been assigned by one- and twodimensional proton and phosphorus NMR experiments. The 12-mer duplex maintains an overall B-form DNA with all anti base conformation throughout. Detailed examination of the 12-mer duplex by high-field NMR and hydroxyl-radical footprinting experiments reveals several local structural perturbations similar to those associated with an (A),-tract DNA duplex. For example, the propeller twisting between base pairs that was detected within the AOT-rich region in the 12-mer duplex is also found in DNA

Chem. Res. Toxicol., Vol. 5, No. 2, 1992 181

Characterization of a 12-mer Duplex

I RopellerTwisted I

!h i

7 u

1

12

5’

GGCGGAG-TTAGG 3

3’

CGC?t\

TCC 5’

24

I

,

technical assistance, to David M. Bishop for careful editing of the manuscript, and to Daekyu Sun for providing the hydroxyl-radical footprinting results. Registry No. (+)-CC-1065, 69866-21-3; 12-mer duplex, 129895-85-8.

Supplementary Material Available: Table S1, giving the NOESY constrainta used for the molecular mechanics and dynamics calculations on t h e d(GGCGGAGTTAGG).d(CCTAACTCCGCC) 12-mer duplex (2 pages). Ordering information is given on any current masthead page.

i‘ *H8and H2 are Broadened -Unique Internal Motion Sugar Geometry *H5Roton Upfield

*HZ Roton Upfield

*Hydrogen-bonded Amino hoton Upfield Shifted

-31P ’-

Resonances Upfield Shifted

- * Inhibition of Hydroxyl-Radical Cleavage ‘TransientKink”

Figure 18. Summary figure of the unusual structural features in the 12-mer duplex.

duplexes containing (A), tracts. In addition, an unusual upfield-shifted aromatic resonance and compression of the minor groove is found adjacent to the 3‘side of the A-Trich segment for both the 12-mer duplex and (A),-tract DNA duplex. The phosphorus resonances associated with the 7G-8T and 18C-19T steps of the 12-mer duplex, where the minor groove is narrowed, displayed unusual upfield chemical shifts. A unique local internal motion at 16A is detected at the nucleotide level. There is a pronounced local distortion centered around 18C, which is proposed to have an average C3’-endo sugar geometry. The NOESY restrained molecular mechanics and dynamics calculations on the 12-mer sequence showed that, on either side of 18C, the backbone is distorted, being more compressed toward the major groove on the 5’side (i-e., at the 5’AC/GT step) and elongated toward the major groove on the 3’ side (i.e., at the 5’CT/AG step). A summary of these unusual structural features in this 12-mer duplex sequence is provided in Figure 18. We propose that the propensity for a DNA sequence to adopt a bent-type is an important factor for the sequence selectivity of (+)-CC-1065. Furthermore, the high reactivity of this particular sequence with (+)-CC-1065 is at least in part due to the unique internal motion (i.e., conformational flexibility) and pronounced local distortion centered around 18C. Upon adduct formation with (+)-CC-1065,this discontinuity at 18C is even more pronounced, and a second discontinuity is induced around the covalent bonding site. The intrinsic internal motion of the 12-mer duplex is annulled, and the bonding sequence now assumes a bent DNA structure quite similar to an A-tract. Finally, we propose that the high reactivity of this sequence is not only due to this inherent propensity to form a bent DNA structure but is also due to the catalytic activation implicated from the experiment in which a water molecule has been shown to bridge between the A subunit of (+)-CC-1065 and the phosphate between 16A and 17A (20).

Acknowledgment. This research was supported by grants from the U.S. Public Health Service (CA-49751), the Welch Foundation, and the Burroughs Wellcome Scholars Program. We are grateful to Steve D. Sorey for

References (1) States, D. J., Haberkorn, R. A., and Ruben, D. J. (1982) A

two-dimensional Nuclear Overhauser experiment with pure absorption phase in four quadrants. J.Magn. Reson. 48,286-292. (2) Hore, P. J. (1983) Solvent suppression in Fourier transform Nuclear Magnetic Resonance. J. Magn. Reson. 55, 283-300. (3) Shaka, A. J., and Freeman, R. (1983) Simplification of NMR spectra by filtration through multiple-quantum coherence. J. Magn. Reson. 51,169-173. (4) Rance, M., Sorensen, 0. W., Bodenhausen, G., Wagner, G., Ernst, R. R., and Wuthrich, K. (1983) Improved spectral resolution in COSY ‘H NMR spectra of proteins via double quantum filtering. Biochem. Biophys. Res. Commun. 117, 479-485. (5) Bothner-By, A. A., Stephens, R. L., and Lee, J.-M. (1984) Structure determination of a tetrasaccharide: Transient Nuclear Overhauser effects in the rotating frame. J. Am. Chem. SOC. 106, 811-813. (6) Bax, A,, and Sarkar, S. K. (1984) Elimination of refocusing pulsea in NMR experiments. J.Magn. Reson. 60, 170-176. (7) Kessler, H., Griesinger, C., Kerssebaum, R., Wagner, K., and Ernst, R. R. (1987) Separation of cross-relaxation and J crosspeaks in 2D rotating-frame NMR spectroscopy. J. Am. Chem. SOC. 109,607-609. (8) Reid, B. R. (1987) Sequence-specific assignments and their use in NMR studies of DNA structure. Q. Rev. Biophys. 20, 1-34. (9) Patel, D. J., Shapiro, L., and Hare, D. R. (1987) NMR studies of conformations and dynamics in solution. Q.Rev. Biophys. 20, 35-112. (10) Seibel, G. (1989) AMBER (UCSF) Version 3.OA, Department of Pharmaceutical Chemistry, University of California, San Francisco. (11) Kadonaga, J. T., Jones, K. A., and Tijan, R. (1986) Promoterspecific activation of RNA polymerase I1 transcription by Spl. Trends Biochem. Sci. 11,20-23. (12) Hurley, L. H., Reynolds, B. L., Swenson, D. H., and Scahill, T. (1984) Reaction of the antitumor antibiotic CC-1065 with DNA Structure of a DNA adduct with DNA sequence specificity. Science (Washington, D.C.) 226, 843-844. (13) Hurley, L. H., Lee, C.-S., McGovren, J. P., Mitchell, M., Warpehoski, M. A., Kelley, R. C., and Aristoff, P. A. (1988) Molecular basis for the DNA sequence specificity of CC-1065. Biochemistry 27, 3886-3892. (14) Hurley, L. H., Warpehoski, M. A., Lee, C.-S., McGovren, J. P., Scahill, T. A,, Kelly, R. C., Mitchell, M. A., Wicnienski, N. A., Gebhard, I., Johnson, P. D., and Bradford, V. S. (1990) Sequence specificity of DNA alkylation by the unnatural enantiomer of CC-1065 and its synthetic analogs. J. Am. Chem. SOC.112, 4633-4649. (15) Reynolds, V. L., Molineux, I. J., Kaplan, D., Swenson, D. H., and Hurley, L. H. (1985) Reaction of the antitumor antibiotic CC-1065 with DNA, location of the site of thermally induced strand breakage, and analysis of DNA sequence specificity. Biochemistry 24, 6228-6237. (16) Katahira, M., Sugeta, H., Kyogoku, Y., Fujii,.S., Fujisawa, R., and Tomita, K. (1988) A new model for the bending of DNAs containing the oligo(dA) tracts based on NMR observations. Nucleic Acids Res. 16, 8619-8632. (17) Warpehoski, M. A., and Hurley, L. H. (1988) Sequence selectivity of DNA covalent modification. Chem. Res. Toxicol. 1, 315-333. (18) Lin, C. H., Sun, D. K., and Hurley, L. H. (1991) (+)-CC-1065 produces bending of DNA that appears to resemble adenine/ thymine tracts. Chem. Res. Toxicol. 4, 21-26. (19) Lin, C. H., and Hurley, L. H. (1990) Determination of the major tautomeric form of the covalently modified adenine in the (+I-

182 Chem. Res. Toxicol., Vol. 5, No. 2, 1992 CC-1065-DNA adduct by lH- and I5N-NMR studies. Biochemistry 29, 9503-9507. (20) Lin, C. H., Beale, J. M., and Hurley, L. H. (1991) Structure of the (+)-CC-1065-DNA adduct: Critical role of ordered water molecules and implications for involvement of phosphate catalysis in the covalent reaction. Biochemistry 30, 3597-3602. (21) Gait, M. J., Ed. (1984) Oligonucleotide Synthesis-A Practical Approach, IRL, Oxford, England. (22) Hare, D. R., Wemmer, D. E., Chou, S. H., Drobny, G., and Reid, B. R. (1983) Assignment of the non-exchangeable proton resonances of d(C-G-C-G-A-A-T-T-C-G-C-G) using two-dimensional Nuclear Magnetic Resonance methods. J. Mol. Biol. 171,319-336. (23) Saenger, W. (1984) Principles of Nucleic Acid Structure, Chapter 14, pp 324-327, Springer-Verlag, New York, Berlin, Heidelberg, and Tokyo. (24) Weiss, M. A., Patel, D. J., Sauer, R. T., and Karplus, M. (1984) Two-dimensional proton NMR study of the h operator site OJ: A sequential assignment strategy and its application. Proc. Natl. Acad. Sci. U.S.A. 81, 13C+134. (25) Wemmer, D. V., Chou, S.-H., and Reid, B. R. (1984) Sequence-specific recognition of DNA. Nuclear magnetic resonance assignments and structural comparison of wild-type and mutant h OR3 operator DNA. J. Mol. Biol. 180, 41-60. (26) Kintanar, A., Klevit, R. E., and Reid, B. R. (1987) Two-dimensional NMR investigation of a bent DNA fragment: Assignment of the resonances and preliminary structure. Nucleic Acids Res. 15, 5845-5862. (27) Chary, K. V. R., Hosur, R. V., and Govil, G. (1987) Novel solution conformation of DNA observed in d(GAAl"CGAAlTC) by two-dimensional NMR spectroscopy. Biochemistry 26, 1315-1322. (28) Gao, X., and Patel, D. J. (1988) NMR studies of echinomycin bisintercalation complexes with d(Al-C2-G3-T4) and d(T1-CBG3-A4) duplexes in aqueous solution: Sequence-dependent formation of Hoogsteen AlsT4 and Watson-Crick TlvA4 base pairs flanking the bisintercalation site. Biochemistry 27, 1744-1751. (29) Gorenstein, D. G., Schroeder, S. A., Fu, J. M., Metz, J. T., Roongta, V., and Jones, C. R. (1988) Assignment of 31PNMR resonances in oligodeoxyribonucleotides: Origin of sequencespecific variations in the deoxyribose phosphate backbone conformation and the 31Pchemical shifts of double-helical nucleic acids. Biochemistry 27, 7223-7237. (30) Burkoff, A. M., and Tullius, T. D. (1987) The unusual conformation adopted by the adenine tracts in kinetoplast DNA. Cell 48,935-943. (31) Wu, H. M., and Crothers, D. M. (1984) The locus of sequence-directed and protein-induced DNA bending. Nature (London) 308,509-513. (32) Hagerman, P. J. (1985) Sequence dependence of the curvature of DNA: A test of the phasing hypothesis. Biochemistry 24, 7033-7037. (33) Hagerman, P. J. (1986) Sequence-directed curvature of DNA. Nature (London) 321, 449-450. (34) Koo, H. S., Wu, H. M., and Crothers, D. M. (1986) DNA

Lin et al. bending at adenine-thymine tracts. Nature (London) 320, 501-506. (35) Griffith, J., Bleyman, M., Rauch, C. A., Kitchin, P. A., and Englund, P. T. (1986) Visualization of the bent helix in kinetoplast DNA by electron microscopy. Cell 46, 717-724. (36) Haran, T. E., and Crothers, D. M. (1989) Cooperativity in A-tract structure and bending properties of composite T,A, blocks. Biochemistry 28, 2763-2767. (37) Celda, B., Widmer, H., Leupin, W., Chazin, W. J., Denny, W. A,, and Wuthrich, K. (1989) Conformational studies of d(AAAAATTTTT), using constraints from Nuclear Overhauser Effect and from quantitative analysis of the cross-peak fine structures in two-dimensional 'H Nuclear Magnetic Resonance spectra. Biochemistry 28, 1462-1471. (38) Nadeau, J., and Crothers, D. M. (1989) Structural basis for DNA bending. h o c . Natl. Acad. Sci. U.S.A.86, 2622-2626. (39) Haasnoot, C. A. G., Westerink, H. P., van der Marel, G. A., and van Boom, J. H. (1984) Discrimination between A-type and Btype conformation of double helical nucleic acid fragments in solution by means of two-dimension Nuclear Overhauser experiments. J. Biomol. Struct. Dyn. 2, 345-360. (40) Keepers, J. W., and James, T. L. (1982) Methods for DNA backbone motions: An interpretation of NMR relaxation experiments. J. Am. Chem. SOC.104,928-939. (41) Eckstein, F., and Jovin, T. M. (1983) Assignment of resonances in the phosphorus-31 Nucleic Magnetic Resonance spectrum of poly[d(A-T)] from phosphorothioate substitution. Biochemistry 22, 4546-4550. (42) Nelson, H. C. M., Finch, J. T., Luisi, B. F., and Klug, A. (1987) The structure of an oligo(dA).oligo(dT) tract and ita biological implications. Nature (London) 330, 221-226. (43) Coll, M., Frederick, C. A., Wang, A. H.-J., and Rich, A. (1987) A biofurcated hydrogen-bonded conformation in the d(AT) base pairs of the DNA dodecamer d(CGCAAATTTGCG)and its complex with distamycin. Proc. Natl. Acad. Sci. U.S.A. 84, 8385-8389. (44) Patel, D. J., Kozlowski, S. A,, Weiss, M., and Bhatt, R. (1985) Conformation and dynamics of the Pribnow box region of the self-complementary d(C-G-A-T-T-A-T-A-A-T-C-G) duplex in solution. Biochemistry 24, 936-944. (45) Arter, D. B., and Schmidt, P. G. (1976) Ring current shielding effects in nucleic acid and double helixes. Nucleic Acids Res. 3, 1437-1447. (46) Bolshoy, A., McNamara, P., Harrington, R. E., and Trifonov, E. N. (1991) Curved DNA without A-A Experimental estimation of all 16 DNA wedge angles. Proc. Natl. Acad. Sci. U.S.A. 88, 2312-2316. (47) Larsen, T. A., Kopka, M. L., and Dickerson, R. E. (1991) Crystal structure analysis of the B-DNA dodecamer CGTGAATTCACG. Biochemistry 30,4443-4449. (48) Graves, D. E., and Wadkins, R. M. (1990) Molecular Basis of Specificity in Nucleic Acid-Drug Interactions (Pullman, B., and Jortner, J., Eds.) pp 177-189, Kluwer Academic Publishers, Dordrecht, Boston, and London.