Proton NMR of an oligodeoxynucleotide containing a

Hai Huang , Hao Wang , Markus W. Voehler , Albena Kozekova , Carmelo J. Rizzo , Amanda K. McCullough , R. Stephen Lloyd , and Michael P. Stone...
0 downloads 0 Views 4MB Size
Chem. Res. Toxicol. 1993,6, 825-836

825

lH NMR of an Oligodeoxynucleotide Containing a Propanodeoxyguanosine Adduct Positioned in a (CG)3 Frameshift Hotspot of Salmonella typhimurium hisD3052 Hoogsteen Base-Pairing at pH 5.8 Udai S. Singh,tJ James G. Mae,§ G. Ramachandra Reddy,? J a s o n P. Weisenseel,$ Lawrence J. Marnett,? and Michael P. Stone*#§ Center in Molecular Toxicology, Department of Chemistry, and A. B. Hancock, Jr., Memorial Laboratory for Cancer Research, Department of Biochemistry, Vanderbilt University, Nashville, Tennessee 37235 Received January 20,1993

The exocyclic DNA adduct 1~-propano-2’-deoxyguanosine(PdG) was inserted into the oligodeoxynucleotide 5’-CGC(PdG)CGGCATG-3’ and annealed to the complementary oligodeoxynucleotide 5’-CATGCCGCGCG-3’. This sequence is derived from a spontaneous revertant of the hisD3052 gene in a frameshift-sensitive tester strain of Salmonella typhimurium and is a hotspot for two-base pair deletions. The solution structure of the modified duplex was examined by lH NMR spectroscopy. The exocyclic lesion resulted in loss of Watson-Crick base-pairing capability. Modification resulted in an -24 “C decrease in T, of the duplex. NMR experiments revealed pH-dependent conformational equilibria, which involved the modified base pair and its 3’-neighbor base pair. At pH 5.8, the lesion resulted in a localized perturbation of the B-form helix. PdG was rotated about the glycosyl bond from the anti to the syn conformation, thus placing the propano moiety into the major groove. This resulted in the observation of a strong NOE between the imidazole proton of PdG and the anomeric proton of the attached deoxyribose. Additional NOES were observed between the methylene protons of the propano moiety and H5 and H6 of the 5’-neighbor cytosine. An imino proton resonance from the cytosine complementary to PdG and protonated at N3, characteristic of a Hoogsteen base pair, was observed a t 15 ppm, but was broadened due to exchange with water. The amino protons of the complementary cytosine were shifted downfield from the other cytosine amino protons, characteristic of a Hoogsteen-like conformation a t the site of modification. A second equilibrium involved the 3’-neighbor base pair, which alternated between Watson-Crick and Hoogsteen pairing, also via rotation of the guanosine glycosyl bond from the anti to the syn conformer. The conformational exchange of the 3’-neighbor base pair was sufficiently slow on the NMR time scale to allow simultaneous observation of resonances from the Watson-Crick and the Hoogsteen conformers. Reactive electrophiles can form adducts, potentially leading to mutations, at many sites in genomic DNA.’ Frameshifts, which result from the addition or deletion of one or more base pairs from DNA, represent one class of

mutationsthat can be induced by DNA adducts. Additions or deletions occur most frequently in DNA sequences containing reiterated bases, palindromes, and tandem repeats. Several mechanisms have been proposed to explain this sequence dependence (1-3). Because frameshift mutations alter the reading frame of the coding t Center in Molecular Toxicology and A. B. Hancock, Jr., Memorial sequence, they are generally expected to inactivate the Laboratory for Cancer Research, Department of Biochemistry. t Present address: Department of Environmental Medicine, School of corresponding gene products. Medicine, New York University, Long Meadow Rd., Tuxedo, NY 10987. The hisD3052 mutation arose from the histidinol I Center in Molecular Toxicology and Department of Chemistry. Abstract published in Advance ACS Abstracts, October 1, 1993. dehydrogenase gene of Salmonella typhimurium by 1 Abbreviations: DNA, deoxyribonucleic acid; EDTA, ethylenedideletion of a cytosine induced by ICR-191 (4-6). It is aminetetraacetic acid; HPLC, high-pressure liquid chromatography; reverted by a series of additions and deletions that restore NMR, nuclear magnetic resonance; NOE, nuclear Overhauser enhancement; ppm, parts per million; MIG, 3-(~-~-ribofuranosyl)pyrimido[1,2- the reading frame but do not necessarily reverse the a]purin-l0(3Zf)-one; MDA, malondialdehyde; PdG, lJP-propano-2’forward mutation (7). The most common reversion event deoxyguanosine; DMT, dimethoxytrityl; DSS,sodium 4,4-dimethyl-4is a CG deletion in the reiterated sequence (CG)4 (8).One silapentanesulfonate; TPPI, time-proportional phase increment; TOCSY, total homonuclearcorrelated spectroscopy;lD, one-dimensional;ZD, twocompound that reverts hisD3052 is malondialdehyde dimensional. The oligonucleotides discussed in this paper do not have (MDA), a mutagen produced endogenously in humans terminal phosphate groups-we abbreviate the nomenclature for oligonucleotides by leaving out the phosphodiester linkage. A, C, G, T, and during lipid peroxidation and prostaglandin biosynthesis X refer to mononucleotideunits, X is the exocyclic 1JVrpropanoguanosine (9-13). In aqueous solutions, MDA exists as its tautomer nucleotide PdC. A right superscript refers to numerical position in the b-hydroxyacrolein and is negatively charged at pH 17. oligonucleotide sequence starting from the 5’-terminus of chain A and proceeding to the 3’-terminus of chain A and then from the 5’-terminus Structure-activity studies (14)suggest that both carbonyl of chain B to the 3’-terminus of chain B. CZ, C5, C6, C8, Cl’, CZ’, CZ”, equivalents must react to generate a premutagenic lesion etc., represent specific carbon nuclei. H2, H5, H6, H8, Hl’, HZ’, HZ”, that induces frameshifts. This is consistent with the etc., represent the protons attached to these carbons. 0893-228x/93/2706-0825$04.00/00 1993 American Chemical Society

826 Chem. Res. Toxicol., Vol. 6, No. 6,1993

formation of a pyrimidopurinone adduct termed M1G (eq 1) (15-18). The instability of M1G to the conditions of

ofH-

guan0 sIne I ,

X

= PdG Adduct

HO

MDA

Non-adducled(lop) and adducled (bonom) ollpodaoxynucleotldesused In this sludy.

PHydroxyacroleln

n

oligodeoxynucleotide synthesis has hampered investigations of its role in malondialdehyde-induced frameshifts. However, the chemically stable analog 1JP-propanodeoxyguanosine (PdG) can be readily incorporated into oligodeoxynucleotides(19).Recently, Benamira et al. (20) incorporated PdG into the (CG)4 repeat of a recombinant M13 phage (M13MB102), by ligating the oligodeoxynucleotide 5’-CGCXCGGCATG-3’ (X = PdG) into a duplex containing a gap between the BssHII and SphI restriction endonuclease cleavage sites. When transformed into Escherichia coli, the mutations arising from the PdGadducted plasmid were deletions, and 70% occurred by deletion of CG. NMR studies of oligodeoxynucleotides containing PdG opposite either adenine or guanine have been reported by Pate1 and co-workers (21,22). Their studies revealed X(syn)*A(anti)pairing at pH 5.8, simultaneous partial intercalation of the complementary X and A bases at pH 8.9, and X(syn).G(anti) pairing which was pH-independent. The solution structures of the pH 5.8 and pH 8.9 X.A adducts were subsequently refined using molecular dynamics calculations which incorporated NOE distance restraints (23,24).A related study established that the exocyclic ring of PdG was inserted into the DNA duplex when positioned opposite an apurinic site (25). Thermodynamic measurements, obtained from UV melting assays by Breslauer and co-workers, revealed that the modified base X reduced the thermal stability, transition enthalpy, and transition free energy of the duplex when positioned opposite cytosine or adenine and that the thermal destabilization of the duplex was not sensitive to whether the base opposite the lesion was adenine or cytosine (26). To better understand how PdG adduct structure may correlate with specific adduct-induced frameshift mutations, we prepared the modified and unmodified duplex oligodeoxynucleotides depicted in Figure 1, where the modified nucleotide X4 was located within a run of 3 CG repeats. The conformation of the PdG adduct opposite cytosine within this frameshift-prone sequence was found to be dependent upon pH. This paper examines the structure of the modified oligodeoxynucleotide at pH 5.8 by high-resolution 1H NMR spectroscopy. At pH 5.8, the adduct-induced perturbation of the oligodeoxynucleotide is localized to the site of adduction and the 3’-neighbor base pair. The modified guanine base is in the syn conformation and forms a protonated Hoogsteen pair with the complementary cytosine. The incorporation of this exocyclic lesion perturbs the 3’-neighbor C-G base pair, which equilibrates between Watson-Crick and Hoogsteen pairing. The NMR analysis of this modified oligodeoxynucleotide, when compared with adduct-directed mu-

Figure 1. The oligodeoxynucleotide examined in this work contains a CG repeat sequence which is a hotspot for two-base pair deletions. The saturated analog 1,W-propanoguanosine (PdG)is used as a model compound for MIG in structural studies.

tagenesis experiments on the same modified sequence (20), suggests potential mechanisms involving adduct-induced slippage of two base pairs, whereby PdG could induce frameshift mutations in a CG repeat sequence. In addition, the observationof PdG(syn)C+(anti)base-pairingprovides a potential mechanism for error-free bypass of this lesion during strand replication.

Materials and Methods Oligonucleotides. The unmodified oligodeoxynucleotide,5’d(CGCGCGGCATG)-3’, and ita complement 5’-d(CATGCCGCGCG)-3’,were obtained from the Midland Certified Reagent Co. (Midland,TX). PdG was synthesized as previouslydescribed by Marinelli and co-workers (19). The 5’-DMT-protected phosphoramidite of PdG was obtained via standard chemistry and characterizedby NMR spectroscopyand normal-phasethinlayer chromatography (0.5% triethylamine in chloroform) on a Partisil column (Alltech Associates, Deerfield, IL, 10 pm; 4.6 X 250 mm). The protected phosphoramiditewas incorporatedinto the oligomer by the Midland Certified Reagent Co. Additional modified oligodeoxynucleotide was prepared in house: PdG phosphoramidite was introduced into the spare port of an oligodeoxynucleotidesynthesizer, and the modified oligodeoxynucleotide 5’-d(CGCXCGGCATG)-3’was prepared using standard protocols on a scale of 10 pmol. Purification of the modified oligodeoxynucleotide was performed by reverse-phase HPLC, using a semipreparative-scale PRP-1column (Hamilton,Inc.,Reno, NV) equilibrated at 60 O C . The purity of the modified oligodeoxynucleotidewas verified by gel electrophoresis and also by capillary gel electrophoresis. The base composition of the modified oligodeoxynucleotide was confirmed by HPLC analysis following enzymaticdigestion with phosphodiesterase and alkaline phosphatase (20). Extinction coefficients for d(CGCXCGGCATG) and d(CATGCCGCGCG) were derived experimentally by phosphate analysis and were determined to be 7.91 X 104and 9.91 x 104M-l cm-l, respectively, at 254 nm. The modified duplexes were prepared by stoichiometric addition of the two strands, followed by hydroxylapatite chromatography,which served to remove any excesssingle strand from the resulting duplex. The samples were then dialyzed against 0.6 M NaCl buffer to exchange the DNA to the sodium salt. Subsequent dialysis against distilled water removed excess counterions. Selectivedeuteration of the purine H8 protons was performed by treatment of the oligodeoxynucleotideunder basic conditions,followed by reannealing with the complementstrand (27).

Spectroscopy. UV melting studies were performed using a CARY 2390 spectrophotometer (Varian Associates, Palo Alto, CA) interfaced with a Neslab (Newington, NH) cryobath and temperature gradient programmer. Melting data were collected at a temperature rise of 1 “C/min. The temperature range was 5-90 “C. The buffer was 1.0 M NaC1,O.Ol M NaHzPO., and 0.05 mM Na2EDTA (pH 5.8). For NMR studies, the unmodified duplex was prepared at a concentration of 4.4 mM. The samples of the modified duplex were prepared at a concentration of 6.4 mM. All samples were prepared in 0.1 M NaC1,O.Ol M NaH2Pod, and 0.05 mM NazEDTA (pH 5.8). lH NMR spectra were recorded at 400.13 and 500.13 MHz. Data processing utilized

Chem. Res. Toxicol., Vol. 6, No. 6, 1993 827

NMR of a PdC Adduct in a Frameshift Hotspot FELIX (Biosym Technologies,Inc., San Diego, CA),running on Iris workstations (Silicon Graphics, Inc., Mountain View, CA). Chemicalshifts were referenced internally to DSS. Temperature was controlled to0.5OC. For observation of exchangeableprotons, oligodeoxynucleotides were dissolved in 9 1 H20/D20. Onedimensional spectra were obtained using either the 1-1 or the 1-3-3-1 (28, 29) binomial pulse sequences. Two-dimensional spectra were obtained by replacing the final 90' pulse of the NOESY experiment by a 1-1 binomial pulse (30,31). In some experiments, a homospoil pulse was applied during the mixing time to remove residual transverse magnetization prior to the final 90° pulse. Convolution difference was used during processing to minimize the residual signal arising from water (32). NOESY experiments utilized the TPPI phase sequence. In short mixing time NOESY experiments, a modified pulse sequence designed to minimize spectral artifacts arising from zz magnetization, and to move artifacts arising from zero quantum coherences away from the peaks of interest, was used (33,341. TOCSY experiments used the standard pulse sequence, with a 120-ms MLEV-17 spin lock (35). Molecular Modeling. Calculations were performed using a Silicon Graphics 4D35TG workstation. SCF calculations were carried out by the MNDO method using MOPAC (36). Energy minimization and molecular dynamics calculations were performed using X-PLOR (37),derived from CHARMM (38))and speciallyadapted for restrained molecular dynamics. INSIGHTII (Biosym Technologies, Inc., San Diego, CA), was used to build the initialstructures,and for visualization of calculatedstructures. Propanodeoxyguanosinewas constructed by bondinga propano group to N1 and N2 of guanine to form the exocyclic ring. The partial charges on PdG and the protonated cytosines were approximated by performing SCF calculations on the free bases (excludingthe sugar-phosphate portion) using a total charge of +1 for the protonated cytosine and a neutral total charge on PdG. The calculated partial charges were incorporated into the models for subsequent energy minimizations and molecular dynamics calculations. Each structure was first energy minimized for 100 iterations by the conjugate gradient method followed by 5 ps of molecular dynamics at 300 K. A force constant of 50 kcal mol-' A-2 was used for base-pairing distance restraints and 20 kcal mol-' kzfor the base-pair planarity restraints. The molecules were weakly coupled to a temperature bath with a target temperature of 300 K and a coupling constant of 0.05 ps (39). The time step of the integrator used in the MD calculations was 1 fs. Structure coordinateswere archived every0.l ps during the MD simulations, and 20 structuresfrom the final 2 ps were averaged. The average molecular dynamics structures were subjected to a final 500 iterations of conjugate gradient energy minimization to correct distortions caused by the averaging procedure and to obtain the final structures.

Results S t a b i l i t y of d(CGCGCGGCATG)-d(CATGCCGCGCG) a n d d (CGCXC GGC ATG)-d(CATGCCGCGCG). Optical melting curves of the unmodified and PdG-modified oligodeoxynucleotide duplexes are shown in Figure 2. Incorporation of PdG decreases the stability of this oligodeoxynucleotide duplex as evidenced by a 24 "C decrease in T, from 65 to 41 "C. The unmodified and the modified oligomers were examined in melting studies at pH 5.8,7.0,and 8.2. No pH dependence was observed in the case of the unmodified oligomer. With the modified oligomer, first derivative plots (inset, Figure 2b) demonstrated that T , was independent of pH. However, the shape of the melting transition varied with pH such that a greater degree of cooperativity appeared to be present both at pH 5.8 and at pH 8.2, as compared to pH 7. 1H N M R Spectral Assignments for d(CGCGCGGCATG)-d(CATGCCGCGCG). Spectral assignment

A. I

I

0.99

d -g W

0.96

g@)

52

z{ 5i

'

0.94

0.91

0.88

0

20

40

60

80

100

Temperature "C

B.

1

1.01)

0.92

0

I

I

20

40

Tmprnture "C

60

80

100

Temperature 'C

Figure 2. UV melting studies. (a) The melting profiles of the modified and unmodified duplexes. (b) UV melting profile of the modified duplex as a function of pH. Inset: First derivative is independent of pH. plots demonstrate that T,,, was challenging due to the presence of 9 COGbase pairs which have closely overlapping chemical shifts. To facilitate assignments for the nonexchangeable protons in the two strands, strand-selective deuteriation (27)was utilized. Figure S1 (supplementary material) details the aromatic-H1' connectivities from C1 to G1'. Elimination of the guanine H8 signals arising from the complementary strand (C12 to G22)allowed unequivocal assignments to be made for nucleotides C1 to GI1. The G4H8 resonance was closely superimposed with those arising from GI8and G20 in the complementary strand. Figure S2 (supplementary material) shows an expansion of the base-to-H1' crosspeaks of the phase-sensitive NOESY spectrum obtained at 15 "C for d(CGCGCGGCATG).d(CATGCCGCGCG). It was observed that within the (CG)3 run of nucleotides, the intensity of the cross-peak between each cytosine H1' and the 3'-neighbor G H8 proton was weaker than the corresponding cross-peak between each guanine H1' and the 3'-neighbor cytosine H6, indicative of a greater H1'aromatic proton distance at the CG steps than at the GC steps in this alternating sequence. Assignments of the remaining deoxyribose protons of the unmodified oligomer were made using standard methods. 1H NMR Spectral Assignments for d(CGCXCGGCATG) .d(CATGCCGCGCG). 0ligodeoxynucleotide Protons. The 1H NMR spectrum of the modified oligodeoxynucleotide was found to be dependent upon pH. At pH 7.0 and at pH 8.2, spectral broadening was

Singh et al.

828 Chem. Res. Toxicol., Vol. 6, No. 6, 1993

A9

i e

QD 6.4

5.8

6.0

6.2

D1

h\ sa

(wm) %T14

5.6

5.4

5.2

1

ComplementSfrandpH 5.8

"1 Cl

e

v

e

0

a5

G20

a5 cu

w a5 6.4

6.2

6.0

5.8

5.6

5.4

5.2

D1 (PPm

Figure 3. Phase-sensitive NOESY spectra with a mixing time of 250 ms for the fully protonated modified hisD3052 oligomer. Panels a and b individually trace through the sequential connectivities of the two strands.

observed and was interpreted to be the result of a conformational equilibrium involving the modified base X4. However, at pH 5.8, the modified oligomer exhibited sharper resonances, albeit with some spectral broadening still observed proximate to X4;under these conditions one conformation predominated. Figure 3 shows an expansion of the base-to-H1' cross-peaks of the phase-sensitive NOESY spectrum obtained at 15 "C (pH 5.8). When compared to the correspondingspectra for the unmodified oligodeoxynucleotide, it was found that the observed changes occur in the immediate vicinity of the modified base pair. For the adducted strand, no cross-peak was observed between C3H1' and X4H2 (the imidazole proton of PdG; note the modified numbering scheme for this nucleotide). Furthermore, X4 H2 was shifted upfield by approximately 0.5 ppm in the spectrum of the modified oligomer as compared to the position of G4 H8 in the unmodified spectrum. The H5, H6, and H1' resonances for C5 were broadened; a very weak cross-peak was observed between X4 H1' and C5 H6 (not observed at the contour level plotted in Figure 3), and between C5H6 and C5 Hl'. The latter cross-peak was partially superimposed with the stronger but also broadened cross-peak between C5 H5 and H6. In the complementary strand, H1' of C17 was partially superimposed with H5 of C16. The resonances of C19, the base complementary to X4, were broadened, but weak cross-peakswere observed between G18 H1' and C19H6, and between C19H6 and C19Hl'. The broadening

of the cross-peaks involving base pairs X4*Cl9and C5=Gl8 was in both cases more pronounced for the cytosineprotons and arose from exchange between multiple conformations of these two base pairs. In this spectral region, B-form oligodeoxynucleotides in which the glycosyl bonds are in the anti conformation exhibit strong NOEs only between cytosine H5 and H6 protons. A comparison of data from a TOCSY experiment (which reveals through-bond connectivities and allows identification of each of the nine cytosine H5-H6 crosspeaks) and a NOESY experiment with a mixing time of 50 ms (which reveals only strong NOEs arising from shortrange dipole-dipole interactions) demonstrated that two strong cross-peaks between purine imidazole and H1' protons were observed in this region, characteristic of glycosyl torsion angles in the syn conformation. These two additional cross-peaks were superimposed along the 0 2 axis and were partially superimposed along D1. This is shown by the arrows in Figure 4b,c. The stronger of the two cross-peakswas the NOE between X4H2 and X4Hl', indicative of a syn conformation of the glycosyl bond at residue X4. The second additional cross-peakwas assigned to the NOE between G18 H8 and G18 Hl', indicating a syn conformation at the glycosyl bond of G18. This crosspeak provided evidence for conformational exchange at base pair C5*G18,because a cross-peak of weaker intensity, corresponding to the anti conformation at the glycosyl bond of G18and occurringfurther downfield, was observed in Figure 3b. The intensity of these two strong crosspeaks was also observed in Figure 3, although at the longer mixing time used in that experiment, intensity from the weaker dipolar interactions between G15H1' and C16 H6, and between G20 H1' and C21 H6, also contributed. Supporting evidence for the assignments of X4 H2 and G18 H8 was derived from examination of the amino and imino portions of the spectrum, described in the following paragraphs. DNA Exchangeable Protons. Figure 5 shows the downfield region of the lH NMR spectrum at 5 "C. In the far downfield region, two broadened resonances were observed at 15 ppm. The broadening of these resonances resulted from rapid exchange with water. They were assigned as arising from protonation at cytosine N3. Sharp resonances arising from thymine and guanine imino protons were observed between 12.5 and 13.5 ppm. The two sharp downfield resonances, located at 13.5 and 13.7 ppm, arose from thymine imino protons, as evidenced by NOEs to the corresponding adenine H2 protons. Several of the guanine imino protons overlapped in the range 12.613.1 ppm. Integration of the imino proton spectral region (and the resonance observed at 10.9 ppm) revealed the presence of 9 protons. Therefore, in addition to the imino proton missing from PdG, one additional guanine imino proton was partially or completely exchange broadened under these conditions. The partial assignment of the imino region of the spectrum was achieved by inspection of NOESY spectra obtained in H20 buffer (Figure 6). Observation of NOES arising between neighboring base pairs revealed that the most upfield signal, located at 12.6 ppm, consists of two resonances, from G7 and G15 N1H. The NOE observed between G7N1H and G6N1H located the latter resonance at 12.8 ppm. G6N1H showed an NOE to G18NlH, located at 12.7 ppm. The pattern of NOE connectivities was interrupted in the 5'-direction at G18 NlH, due to the presence of X4. The imino protons of base pairs

Chem. Res. Toxicol., Vol. 6, No. 6, 1993 829

NMR of a PdG Adduct in a Frameshift Hotspot

x .m c5

4'

GI9

&k1 Clt

"

Figure 4. Comparison of the region of the two-dimensional spectrum containing cross-peaks arising from the aromatic protons and anomeric protons. In (a) the TOCSY spectrum allows identification of the 9 cross-peaks arising from J couplings between cytosine H5 and H6 protons. In (b) 2D NOE spectra run with a mixing time of 50 ms indicate the presence of two additional cross-peaks arising from short-range NOEs. The large NOE between H2 of X4 and H1' of the attached deoxyribose is consistent with a syn orientation of the glycosyl bond. The upfield shift of the H8 proton is also consistent with reorientation of this proton into the helix. Immediately adjacent to the strong NOE assigned to X4 H2 is a second strong NOE, which is assigned to the Hoogsteen conformation of G**in the 3'-neighbor base pair. In (c) the data are shown in the form of a stacked plot. The additional peaks observed in the spectrum arise from the modified pulse sequence which was used (33,34).

I

I

12

I

I

la

0

GI5 N I H m d G I N l H

GI

/a 16

14

12

10

8

ppm

Figure 5. Downfield region of the 'H NMR spectrum recorded in H20 buffer.

G2=C2l,and C3,G2O, were superimposed at 13 ppm. No NOE was observed between base pair T10-A13and G 1 W 2 , suggesting that the guanine imino proton of this terminal base pair was exchanging rapidly with water under these conditions. In the spectral region between 9 and 11ppm, additional resonances were observed in Figure 5. Two resonances were observed at 8.8 and 9.9 ppm, respectively. A resonance was observed at 10.9 ppm, and two less-intense resonances were observed in the upfield shoulder of the resonance located at 9.9 ppm. Additional weak signals were discernible in the shoulders of the strong signal at 8.8 ppm. The intensity of these weaker resonances was dependent upon pH, indicativeof a measurable population of an additional conformer present in slow exchange with the predominant conformation. The observation that the weaker resonances did not integrate in proportion to the stronger signals supported the analysis that they arose from conformational exchange. Inspection of the NOESY spectra run in H20 buffer allowed assignment of each of the signals in the 9-11 ppm spectral region (Figure 7). The two strong signals located at 8.8 and 9.9 ppm could be assigned to the two protons of C19NH2 by observation of an NOE to C19H5 (arrow in top left corner of Figure 7). Additional NOEs were

1

I 13.5

13.0

12.5

12.0

11.5

11.0

01 (PPW

Figure 6. Assignment of imino protons. The NOESY spectrum reveals the presence of an exchange cross-peak for G18 NlH, which equilibrates between Watson-Crick and Hoogsteen conformations a t pH 5.8.

observed in the 5'-direction between C19NH2 and C3NH:! (the latter identified by its NOE to C3 H5),confirming this assignment. Inspection of the small signals arising from the minor conformation revealed the presence of strong exchange cross-peaks between signals located at 9.7 and 8.8 ppm, and the two protons of C5 NH2. The observationof these strong exchange cross-peaks provided evidence for conformational exchange at base pair C5=G18. The two minor signals located at 9.7 and 8.8 ppm were assigned to the two protons of C5NH2, revealing a second conformation in slow exchange with the predominant conformation. In addition, a second set of exchange crosspeaks were observed between C19NH2 and signals located at 9.9 and 8.8 ppm, which indicated that the chemical shift environment of C19 NH2 was dependent upon the conformationalequilibriumat base pair C5*G18.The amino signals observed between 8.8 and 10 ppm and assigned to base pairs X4*C19and C5=G18 were shifted downfield from the spectral region in which cytosine amino protons normally resonate in B-form DNA, into the region where the amino protons of cytosine protonated at N3 and involved in Hoogsteen hydrogen bonding are characteristically observed. Evidence that this chemical shift

Singh et al.

830 Chem. Res. Toxicol., Vol. 6,No. 6,1993

W N H 2 C19?Y2

I

i % 100

95

90

85

80

o n

7.5

3 G

A 7.0

65

D1 (PPW

Figure 7. The amino proton region of the NOESY spectrum recordedin H20 buffer. The predominantfeature is the downfield shift of C19NH2 and the presence of exchange cross-peaksfor C5 NH2, which is equilibratingbetween Watson-Crick and Hoogsteen pairing under these conditions. Note that the low intensity of the diagonal cross-peaks for the Hoogsteen conformation of C5 NH:!contrastswith the higher intensityof the exchangecrosspeaks between the Watson-Crick and Hoogsteen conformations at C5 NH2. The arrows indicate sequential NOE connectivities between C19 NH2 and C3 NH2 (in the 5’-direction)and between C19 NHB and C19 H5.

resulted from protonation at cytosine N3 was provided by the observation of the two broadened resonances at 15 ppm. This observation corroboratedthe above assignment of the two strong aromatic-H1’ NOEs in Figure 4b to X4 H2 and G18H8: Hoogsteen pairing at these two base pairs would place these two guanines in the syn conformation about the glycosyl bond, resulting in strong NOEs between the imidazole proton and H1’ proton. The G18 N1H resonance located at 12.8ppm exhibited an exchange crosspeak to the signal observed at 10.9 ppm, also consistent with the above assignments. The latter signal gave only a very weak diagonal peak in the NOESY spectrum (no diagonal peak was observed in Figure 7), due to exchange with water. The 10.9 ppm chemicalshift was characteristic of a guanine imino proton exposed to solvent. PdG Protons. The PdG exocyclic protons appeared in the upfield region of the spectrum, partially overlapped with the deoxyribose H2’,H2” protons (Figure 8). They were broadened in the spectrum as compared to the H2’,H2” protons of the deoxyribose ring. This was a consequence of exchange broadening arising from the conformational equilibrium involving base pair C5-G18. Since the propano protons of PdG were observed as sharp signals in other modified oligodeoxynucleotides, the broadening was not thought to be a consequence of conformational exchange involving the propano ring itself. Small changes in chemical shift were observed for the bridging protons of the trimethylene moiety of PdG, as compared to their chemical shift in the PdG-modified single strand (20),suggesting an extrahelical environment for these protons.2 H7a and H7b of PdG resonated separately, at 6 1.83 and 1.95 ppm, but were not assigned individually. H6a and H6b were observed at 6 3.36 ppm. One proton at C8 was observed at 6 3.63 ppm, which was assigned as H8a on the basis of an NOE in the 5’-direction, The definitions of the diastereotopic protons at C6, C7, and C8 are based upon the Cahn, Ingold, and Prelog nomenclature. H6a is the pro-S proton at C6, and H6b is the pro-R proton at C6. H7a is the pro-R proton at C7, and H7b is the pro-S proton at C7. H8a is the pro-R proton at C8, and H8b is the p r o 3 proton at C8. Protons H6a, H7a, and H8a face in the 5’-direction, whereas protons H6b, H7b, and H8b face in the 3‘direction.

3: 2

2.8 DI

2.4 (ppm)

2.0

1.6

1.2

Figure 8. NOESY spectrum locating the CH2 protons of the PdG propanyl moiety. HGa,b, H7a,b, and H8a,b are partially superimposed upon the deoxyribose H2’,H2” protons. The methylenic resonances are broader than the deoxyribose resonances, consistent with conformational exchange between Watson-Crick and Hoogsteen pairing at base pair C6*G18.

to C3 H5 and H6. A weaker signal, at 6 3.78 ppm, is tentatively assigned as H8b (not marked on Figure 8). We were unable to locate the amino proton of the PdG exocyclic moiety, which may undergo rapid exchange with H2O under these conditions. Previous workers also failed to observe this proton (22). Molecular Modeling. A canonical B-DNA 11-mer (40, 41) was used as the initial structure for model building. The base X4 was rotated -180’ about the Cl’-N9 bond into the syn conformation to form a Hoogsteen base pair with C19,which was protonated at N3. This was used as the starting structure for the major conformation present at pH 5.8. The starting structure for minor conformation at pH 5.8 was built from the major conformationstructure, with base G18 rotated -180’ into the syn conformation to form a second Hoogsteen base pair with C5protonated at N3. The calculations were based on an energy function approach in which the total energy was the sum of the empirical energy of the molecule and effective energy, comprised of the restraint energy terms. The empirical energy function (42) used was developed especially for nucleic acids and treated all hydrogens explicitly. It consisted of the usual energy terms for bonds, bond angles, torsional angles, tetrahedral and planar geometries, hydrogen bonding, and nonbond interactions including van der Waals and electrostatic forces. The van der Waals energy term was approximated using the Lennard-Jones potential energy function. The electrostatic term used the Coulombic function and was based on a reduced charge set of partial charges (-0.32/residue) and a distancedependent dielectric constant of 4.0, to mimic solvent screeningof charge. The nonbonded pair list was updated if any atom moved more than 0.5 A,and the cutoff radius for nonbonded interactions was 11A. All calculationswere performed in uacuo without explicit counterions. The effective energy function was comprised of two terms and describingdistance and dihedral restraints (E&& The effective distance restraint potential, Edict, and dihedral restraint potential, Edihe, were in the form of a standard square well potential (43). All bond lengths involving hydrogens were fixed with the SHAKE algorithm (44) during molecular dynamics calculations. All base pairs were held together by adding distance restraints between bases at the atoms involved in hydrogen bonding. The distance restraints for all Watson-Crick base pairs were defined as follows. For the

NMR of a PdG Adduct in a Frameshift Hotspot

G-Cbase-pairs rrcytosine N4-guanosine 061 = 2.70 f 0.20 = 2.91 f 0.10 A, and rrcytosine 02-guanosine N21 = 3.01 f 0.20 A. For A-T base pairs r[adenine N6-thymidine 041 = 2.80 f 0.20 A andr[adenine N1-thymine N31= 2.95 f0.lOA. Distance restraints for the Hoogsteen base pairs were defined as r[PdG N7-cytosine N3] = 2.93 f 0.10 A and r[PdG 06cytosine N41 = 2.86 f 0.10 A. To prevent excessive propeller twisting of the base pairs in the molecular dynamics calculations, a torsion angle restraint was added between the X4.Cl9base pair in the major conformer and between the X4.C19 and C6.G18 base pairs with a value of 0 f 100. The final models for both the major and minor conformation at pH 5.8 are shown in Figure ga. The PdG adduct was well accommodated in the helix, but the rise between X4 and C5 was slightly increased. In the minor conformer the increased helical rise was more dramatic than in the major conformer. The alignments of base pairs X4.C19and C5.Gl8are shown in Figure 9b. Comparison of the calculated structures for the major and minor conformers showed that there was an increase in separation between C5and CI8in the minor conformer. The distance from N3 to C6to N3 of C19in the major conformer was 6.81 8, while the same distance in minor conformer was 7.08 A. The final energies for the major and minor conformers were 392 and 398 kcal/mol, respectively.

A, rrcytosine N3-guanosine Nl]

Discussion The present study represents an effort to examine PdG in the context of a frameshift-prone oligodeoxynucleotide for which mutagenesis studies have been carried out in parallel (20).This study also represents the first example of PdG inserted into an alternating CG sequence and opposite cytosine in the complementary strand. PdG lesions located opposite adenine and guanine (21,221,and have previouslybeen studied opposite an apurinic site (25)) by NMR spectroscopy. Initial NMR experiments revealed that the structure is dependent upon pH, suggesting that the modified nucleotide X4 equilibrates between two or more conformations. Each conformation could influence the ability of this adduct to induce frameshifts. At neutral pH, the equilibration involving X4 exhibits intermediate exchange kinetics on the NMR time scale, as evidenced by line broadening in the spectra. However, under mildly acidic conditions at pH 5.8, one conformation predominates. Therefore, we chose to focus our initial work on the pH 5.8 conformation at X4. Localized Perturbation of the Modified Duplex. The presence of PdG in the hisD3052 oligodeoxynucleotide resulted in a localized perturbation of the DNA helix. At low temperature, 9 signals arose from imino protons of the base pairs and were observed between 10.9 and 13.8 ppm. The imino proton resonance arising from the terminal base pair G11.C12was not observed, presumably because of exchange broadening (Figures 5 and 6). PdG disrupted Watson-Crick base-pairing at the site of adduction, where the exocyclicpropano derivative eliminated the imino proton at X4. In addition, it perturbed WatsonCrick base-pairing at the 3’-neighbor C-G base pair. Inspection of the spectral data revealed that for residues G6to Gl1 and C12 to C1’, the pattern of NOE crosspeaks was reproduced, as compared to the corresponding spectrum for the unmodified duplex. Thus we conclude that modification with PdG does not substantially perturb the conformation of these 5 base pairs at pH 5.8. Likewise,

Chem. Res. Toxicol., Vol. 6, No. 6,1993 831 for residues C1 to C3 and G2O to G22,the pattern of NOE cross-peaks in this region was also reproduced, indicating little if any conformational change for these three base pairs. Hoogsteen (PdG)(syn).C+(anti)Base-Pairingat the Site of Adduction. The observation of a strong NOE between X4 H2 (the imidazole proton) and X4 Hl’, indicative of a close contact between these two protons, provided evidence that the glycosyl torsion angle of the propano-modified base had rotated to the syn conformation. This strong cross-peak was clearly observed in the NOESY spectrum obtained at a mixing time of 50 ms (Figure 4b). Under those conditions aromatieH1’ crosspeaks which arise from nucleosideshaving glycosyl torsions in the anti range were not observed. In addition, the H2 proton resonance of X4 shifted upfield as compared to its location in the unmodified oligomer, such that it was located in the same region of the spectrum as most of the cytosine H6 resonances. Thus, the PdG moiety must be oriented into the major groove of the helix, placing X4H2 into the minor groove. This orientation facilitates protonation of Clg N3, allowing formation of a Hoogsteen pair with X4. In this orientation, the exocyclic amino group of Clg donates a hydrogen bond to the keto group of X4. We observed one cytosine amino group, assigned as Cl9 NHz, to be shifted downfield, characteristic of a protonated cytosine base (22, 45). The NOESY spectrum revealed that this downfield-shifted amino group exhibited NOE connectivity to C19 H5 and that it did not exhibit NOE connectivity to a guanine imino proton (Figure 7). The N3 proton of protonated cytosine was observed at 15ppm, but it was exchange broadened, precluding observation of an NOE between N3H+ and the hydrogen-bonded amino proton. Figure 10 summarizes the observed chemical shift perturbations for the aromatic and anomeric protons in going from the unmodified oligodeoxynucleotide to the X4(syn)Og(anti)conformation of the modified oligodeoxynucleotide at pH 5.8. Under these conditions, the chemical shift influence of the modified base was localized to the site of adduction and immediately adjacent base pairs. The greatest changes in chemical shift among the nonexchangeable aromatic protons were observed for the X4.C19base pair and its 3’-neighbor base pair C6-Gl8.The H2 resonance of X4 shifted upfield by 0.49 ppm, whereas H5 and H6 of C5shifted downfield by 0.31 and 0.28 ppm, respectively. On the complementary strand, H5 and H6 of C19,which is opposite X4, shifted downfield by 0.39 and 0.28 ppm, respectively. In addition, the anomeric proton of C3 showed a downfield shift of 0.25 ppm, whereas the anomeric proton of C19 shifted downfield by 0.16 ppm. These chemical shift changes were judged to be the result of changes in base stacking and to electrostatic changes upon reorientation of base pair X4.C19into the Hoogsteen conformation and protonation of C19. Restrained molecular dynamics calculations were performed in uacuo without NOE restraints (see discussion following) (Figure 9) to develop a model for the modified oligonucleotide. The calculations predicted that the methylenic protons H6a and H8a on the top (5’) face the exocyclic PdG moiety should have distances to C3 H6 of 3.5 and 3.4A. The NOESY spectrum revealed the presence of these NOE cross-peaks (Figure 11). The distance between X4 H8a and C3 H5 was estimated to be 3.1 A. A stronger cross-peak corresponding to this NOE was

Singh et al.

832 Chem. Res. Toxicol., Vol. 6, No. 6, 1993

1

Major Groove

Major Groove

%+ Major Groove

Major Conformation

Major Groove

Minor Conformation

Figure 9. Molecular models of the major and minor conformations of the PdG-modified oligomer at pH 5.8. (a) Space-filling representations were derived from restained molecular dynamics simulation, followed by energy minimization (NOE restraints are not included, seetext). In the major conformation(modelon left), X4is colored in red. Base pair X 4 0 9is in the Hoogsteenconformation, and all other base pairs are Watson-Crick. In the minor conformation (modelon right), X4 is colored in blue. Base pairs X409 and CS*G18are in the Hoogsteen conformation. In both conformations,the propano protons of X4face into the major groove,and structural perturbation of the DNA helix is localized at these two base pairs. (b) Base-pairing alignments at base pairs X4-C19and C5.Gl8, for the major (left) and minor (right) conformations, respectively. Note the protrusion of the propano group into the major groove.

observed in Figure 11. These NOES were predicted by the molecular modeling, and their presence confirmed that the PdG moiety was oriented into the major groove of the duplex. Similar NOESwere observed in the PdG(syn)eA(anti) pH 5.8 adduct examined by Pate1 and co-workers (22),in contrast to the chemical shifts observed for a PdG-

(anti)*A(anti)adduct observed at pH 8.9 (21). Observation of a Second Conformation: Equilibration of the 3’-Neighbor Base Pair between Watson-Crick and Hoogsteen Pairing. The presence of a second conformation of this modified oligodeoxynucleotide, characterized by adjacent Hoogsteen base pairs, was

NMR of a PdG Adduct in a Frameshift Hotspot 1.oo

I

n IC

Modified strand DH 5.8

V . I J

I

I

I

I

Chem. Res. Toxicol., Vol. 6, No. 6, 1993 833

-

I

0

0.50 0.25

H H8,H6

0.00

H5 H1‘

-0.25

-

m b

-0.50 -0.75 4

nn

0.75

’ *

-

Complementary strand pH 5.8

t

-

0.50 0.25 0.00

H H8,H6 E4 H5

-0.25

H1‘

H

0

-0.50 -0.75

-1 .oo

1

2

3

4

5

6

7

8

91011

Base Pair Figure 10. Chemical shift perturbations observed in the major conformation of the modified hisD3052 oligomer. Analysis of chemical shift perturbations suggest that the influence of the modified base is localized to the site of adduction and immediately adjacent base pairs. The y-axis monitors A6 values, where A6 = &(modifiedoligomer) - &(unmodifiedoligomer) (ppm).

observed. A model of the minor conformation is shown in Figure 9. Evidence for conformational exchange at pH 5.8 was provided by the observation of spectral broadening for the protons located at base pairs X4*C19and C5-G18 (Figure 3). Furthermore, inspection of the 9-11 ppm region of the spectrum as recorded in H20 revealed additional small signals which arose from a second conformation (Figures 5 and 7). The NOESY spectrum provided evidenceof magnetization transfer between Watson-Crick and Hoogsteen environments for the cytosine amino protons of base pair C5-G18(Figure 7), and as well for the G18imino proton of this base pair (Figure 6). The results demonstrated that, in this modified oligodeoxynucleotide at pH 5.8, there was an observable tendency for the 3‘neighbor base pair to also shift into a Hoogsteen conformation. This reorientation resulted in the G18N1H proton being oriented into the major groove of the helix. It exhibited increased shielding and resonated at 10.9 ppm. We conclude that modification of this alternating (CG)3 sequence by the exocyclic adduct at X4 results in two possible conformational states at pH 5.8: one in which the adducted base pair X 4 0 9is Hoogsteen and all other base pairs remain Watson-Crick (the predominant conformation), and a second (minor) conformation consisting of two adjacent Hoogsteen pairs, located at the modified base pair, and at the 3’-neighbor base pair (Figure 9). The forces that regulate the position of this conformational equilibrium remain to be established. Protonation of C19N3 allows for Hoogsteen hydrogen bonding at base pair X4*C19,and it is likely that the pKa for C19is increased relative to cytosine in an unmodified base pair.3 Although favorable base-stacking interactions and steric considerations could provide a driving force for conversion of both 3 Preliminary results indicate that a significant amount of the major conformation described for this modified oligonucleotide, in which base pair X4.C19is in the Hoogsteen conformation, is present at pH 7, which suggests that this conformation may play a role in adduct-directed mutagenesis by PdG in this sequence.

C3 H6

X4

H6a -->C3H6

x4

H8a-->C3H6

I

5

8.0

Y# 3;

I

7.5

I

7.0

X4 H8a --> C3 H5*

..- . -

1

6.5

I

6.0

0 c

I

5.5

Figure 11. NOES observed between X4 and DNA. Although the methylenic protons of PdG are broadened due to conformational exchange, the NOESY spectrum reveals the presence of a set of cross-peaks between X4 H6 and H8 and C3 H6, and between X4 H8 and C3H5. These NOE connectivities place the exocyclic lesion in the major groove of the DNA and proximate to the 5’-neighbor base. The NOE between X4H7 and C3H6 is superimposed among the cross-peaks between C3 H6 and deoxyribose H2’,H2’’ protons.

base pairs to the Hoogsteen conformation, this would probably be opposed by unfavorable electrostatic interactions between positive charges at adjacent Hoogsteen pairs. Thus, the pK, for C5 N3 might be expected to be lower than for C19 N3, consistent with the observation that at pH 5.8 it is the minor conformation which has the two Hoogsteen pairs. The restrained molecular dynamics calculationsperformed on the two conformationspredicted the energy of the minor conformer to be 5.47 kcal/mol higher than that of the major conformer, in agreement with the experimental observation. In addition, the calculations predicted an increase in separation between the protonated bases C5 and C19 in the minor conformer, which appezrs to be consistent with the expectation of unfavorable electrostatic repulsion terms in the minor conformer. Thermodynamic measurements conducted by Breslauer and co-workers, in which PdG was flanked by G C base

834

Singh et al.

Chem. Res. Toricol., Vol. 6, No. 6, 1993

pairs on both the 5' and 3' sides, demonstrated that T, for the PdG-A duplex did not show a pH-dependent transition (26). This observation was surprising since NMR studies had established the presence of a conformational equilibrium involving a major structural rearrangement at neutral pH (21, 22). Furthermore, the PdG(anti).A and PdG(syn).A structures exhibited only a 0.4 kcal/mol difference in van't Hoff free energies at 25 "C (26). It was concluded that the DNA duplex can accommodate the PdG-Amismatch with only a minimum energetic penalty. In the present study, the T, measurements reported in Figure 2 for the PdGC mismatch in the hisD3052sequence showed T, to be independent of pH, although the shape of the melting transition varied with pH. While a quantitative thermodynamic study remains to be completed for the PdG lesion in the hisD3052 sequence, it may be that the lesion-induced structural perturbation induced in this sequence at pH 5.8 also can be accommodated with minimum energetic penalty. CD measurements on the PdG-Cpair flanked by 5'- and 3'-neighboring G C pairs failed to reveal a pH-dependent structural transition, as was observed for the PdGSA pair (26). That observation was surprising in light of the present results, which demonstrate Hoogsteen-like pairing for PdG-C at pH 5.8. It perhaps suggests that the conformation of a PdG lesion located opposite cytosine depends upon sequence. The mechanism by which conformational interconversion occurs in the hisD3052 sequence also remains to be established. A mechanism recently proposed for the pHdependent conformational interconversion of the PdG-A mismatch, based upon molecular modeling studies which incorporated NMR constraints, postulates an adductinduced kink in the DNA which accommodates rotation of PdG about the glycosyl bond (24). Future experiments designed to examine hydrogen exchange kinetics (46) and electrophoretic mobility (47) may provide insight into the longer-range conformation and rate of base pair opening adjacent to the site of modification. Molecular Modeling. Eisenberg and co-workers presented the results of molecular modeling for the PdG-A mismatch at pH 5.8 and 8.9, which included NOE restraints (23,24). Their calculations incorporated the assumption that NOE data obtained at each of the two pH values defined a closely related set of conformationalmicrostates in rapid exchange on the NMR time scale, which could be refined as a single time-averaged (on the NMR time scale) structure. In the present study, such an assumption was not valid, because the PdGC base pair at pH 5.8 equilibrated between two distinct conformations. This dictated that the two conformations be modeled independently. We were unable to resolve two sets of NOE restraints, corresponding to each of the two distinct conformations, which precluded detailed structural refinement of the two conformations using NOE distance constraints. We decided instead to construct models for each of the two conformations, using molecular dynamics calculations restrained by generic distance constraints for either Watson-Crick or Hoogsteen pairs, as dictated for each of the two conformations, but not incorporating NOE data. The two sets of calculations used canonical B-DNA in which either base pair X4.C19 (the major conformation) or both base pairs X 4 0 9 and C5eG18 (the minor conformation) were in the Hoogsteen alignment, as starting structures. Canonical B-DNA provided a reasonable

5'-CATGCCGC 3'-GTACGGCXCGCGCCT-5' y w a p

Synthesm/

5'-CATGCCGC 3'-GTACGqFGCGCCT-5' C-X

5 -CATGCCG 3'-GTACGGCXCGCGCCT-5'

/5;"thens

5'-CATGCCG 3'-GTACGqFGCGCCT-5'

c-X

Figure 12. Potential mechanisms of CG deletion by PdG. The top pathway represents the Kunkel model (2) whereby slippage occurs subsequent to nucleotide insertion due to pausing of the replication apparatus at the PdG-C pair. The bottom pathway represents the Streisinger model (1) whereby slippage occurs due to pausing of the replication apparatus prior to insertion of a nucleotide opposite PdG.

starting structure since the experimental data were consistent with a B-form duplex (except at the site of adduction and neighboring base pair), and the NMR data demonstrated that adduction did not cause a large perturbation in DNA conformation except at the site of the lesion (e.g., the chemical shift data shown in Figure 10). The calculated model structures are less defined in the absence of experimental NOE constraints than if a full set of NOE Constraintshad been available. Nevertheless,they demonstrate that the propano protons of PdG can be oriented into the major groove in a location which is consistent with the observation of NOESbetween the PdG protons and the H5 and H6 protons of the 5'-neighbor cytosine, shown in Figure 11. The calculated relative energies of the major and minor structures support the experimental observation that the adducted structure having the single Hoogsteen pair is favored over the structure having adjacent Hoogsteen pairs. The calculations predict that the presence of two adjacent protonated Hoogsteen pairs in the latter structure results in an increase in the separation of the two base pairs, presumably as a result of unfavorable electrostatic interactionsbetween the two protonated cytosines. These calculated structures support the working hypothesis that the structural perturbation caused by PdG in this oligodeoxynucleotide is localized to the site of adduction and 3'-neighbor base pair. Correlationwith Mutagenesis Data. Cyclization of the guanine ring at the [1,2-a3positions removes WatsonCrick base-pairing capability. Thus, PdG is expected to be a premutagenic lesion in DNA. The formation of a PdG(syn)C+(anti) Hoogsteen base pair may contribute to the genesis of frameshift mutations. Adduct-directed mutagenesis experiments using the same PdG-modified oligodeoxynucleotide that was used in the present NMR studies revealed that when inserted into M13mp19, followed by transformation into E . coli, PdG induced twobase deletions of CG (20). The studies described herein reveal that, at pH 5.8, modification of this oligodeoxynucleotide results in destabilization of the duplex structure and perturbation of base pairs X4-Clgand CS.G18,which may correlate with the observed two-base pair deletion.3 Figure 12 details two potential mechanisms for PdGinduced strand slippage deletions of CG, from a CG repeat. The classical slipped mispairing mechanism of Streisinger (1)is depicted by the two steps at the bottom of the figure. Pausing of the replication apparatus at the position of the adduct allows slippage to occur, which results in a twobase pair loop and a terminal G C base pair, from which

NMR of a PdG Adduct in a Frameshift Hotspot the growing strand can be extended. The insertionslippage model of Kunkel(2) is depicted at the top of the figure. Incorporation of a C residue opposite PdG, which we hypothesize occurs via a Hoogsteen template at the PdG lesion, precedes slippage to a two-base pair loop, providing a two-base pair terminus from which polymerization can be extended. The insertion-slippage mechanism (2) requires the replication apparatus to pause long enough after insertion of the C opposite PdG for slippage to occur. Shibutani and Grollman recently demonstrated that extension from a PdGC base pair by DNA polymerase I in vitro occurs one thousand times more slowly than extension from a G-C base pair: an observation that is consistent with the insertion-slippage model of Kunkel. Frameshift mutagenesis by (acety1amino)fluorenein the NurI restriction sequence d(GGCGCC) (48,49)also results in deletion of CG (20). In site-specific mutagenesis experiments, these deletions only occurred when the adduct was located at the third guanine residue in the sequence (48,49). This striking sequence dependence led to the proposal that the (acety1amino)fluoreneadduct may induce an unusual DNA structure, perhaps Z-like at this location (48). Deviations from B-form DNA were observed in an X-ray crystallography study of an unmodified oligodeoxynucleotide containing the embedded NurI sequence, suggesting that this sequence could be prone to a specific adduct-induced conformational change (50). However, a more recent crystallographic study failed to observe this perturbation (51). In the present case, the exocyclic ring of PdG would not support base-pairing in the Z conformation. Our observation of the presence of a minor conformation in which the PdG-modified base pair and the 3’-neighbor base pair are both shifted into the Hoogsteen conformation does support the notion that adduct formation within an alternating CG sequence can induce a two-base pair perturbation of the helix, which may be correlated to subsequent two-base pair deletions. Grollman and co-workers reported that PdG induces a low frequency of PdG T mutations (-0.4%) when a plasmid containing it is replicated in E. coli or monkey kidney cells in vivo (52). This low frequency is somewhat surprising because the Watson-Crick base-pairing region of PdG is completely blocked, and because of the tendency toward insertion of adenosine at noninformational lesions in DNA (53, 54). Previous NMR studies by Pate1 and co-workers examined PdG inserted opposite adenosine at both acidic and basic pH (21,221and opposite guanine at neutral pH (22). Those studies detected the presence of stable PdG(syn).A+(anti) base pairs (22). That observation, combined with the present results demonstrating the formation of a PdG(syn)C+(anti)Hoogsteen base pair, provides a potential explanation for the pattern of mutations and their low frequency of occurrence. Incorporation of A opposite PdG during DNA replication would result from formation of a mismatch between PdG in the syn conformation and protonated A in the anti conformation. Replication of the A-containing strand in a subsequent round of replication would place a T in the position of the original PdG, resulting in the observed PdG T mutation. Alternatively, formation of a Hoogsteen base pair between PdG and protonated C would provide a mechanism for accurate bypass of PdG. The low frequency of PdG T mutations correlates with the relative pKa values of adenine N1 vs cytosine N3. The pKa of cytosine N3 is greater than the pKa of adenosine

-

-

-

4

Professor A. P. Grollman, personal communication.

Chem. Res. Toxicol., Vol. 6, No. 6, 1993 835 N1, so a t physiological pH the PdGC+ Hoogsteen pairing may predominate over the PdG-A+pairing. Thus, incorporation of A opposite PdG would be predicted to occur less frequently than incorporation of C, thereby explaining the low frequency of PdG T mutations (52). Summary. This work examines the structure of a PdG adduct located within a frameshift-prone CG repeat sequence from the hisD3052 genome. The results reveal that PdG, when incorporated opposite cytosine at pH 5.8, is oriented such that the propano moiety faces into the major groove. This is a result of rotation about the glycosyl bond to the syn conformation, with formation of a Hoogsteen-like base pair between PdG and N3-protonated cytosine. In this alternating (CG)3 sequence, a second pH-dependent conformational equilibrium is present, in which the 3’aeighbor base pair alternates between Watson-Crick and Hoogsteen bonding. The frameshifts induced by PdG in this sequence are consistent with potential mechanisms whereby the modified guanine and adjacent cytosine loop out of the helix due to pausing of the replication apparatus prior to insertion of a nucleotide opposite PdG or to pausing prior to extension from a PdGC base pair.

-

Acknowledgment. This research was supported by the NIH: CA-55678 (M.P.S.) and CA-47479 (L.J.M.). J.P.W. received support from an NIH predoctoral training grant in molecular biophysics, GM-08320. Partial funding to enable purchase of the AMX-500 NMR spectrometer was received from Shared Instrumentation Grant RR05805 and from the Vanderbilt Center in Molecular Toxicology, ES-00267. Supplementary Material Available: Figures S1 and 52, showing the assignmentsof the unmodified oligodeoxynucleotide sequence through the use of selective deuteriation of one strand (2 pages). Orderinginformation is given on any current masthead page. Coordinates for the calculated structures of the major and

minor conformers are available from M.P.S.upon request.

References (1) Streisinger, G., Okada, Y.,Enrich, J., Newton, J., Tsugita, A., Terzaghi, E., and Inouye, M. (1966)Frameshift mutations and the genetic code. Cold Spring Harbor Symp. Quant. Biol. 31,77-84. (2) Kunkel, T. A. (1990)Misalignment-mediated DNA synthesis errors. Biochemistry 29,8003-8011. (3) Ripley, L. S.(1982)Model for the participation of quasi-palindromic DNA sequences in frameshift mutation. h o c . Natl. Acad. Sci. U.S.A. 79,4128-4132. (4) OHara, S.M., and Marnett, L. J. (1991)DNA sequence analysis of spontaneous and 8-methoxy-acrolein-inducedmutations in salmonella typhimurium hisD3052. Mutat. Res. 247,45-56. (5) Oeschger,N. S., and Hartman, P. E. (1970)ICR-induced frameshift mutations in histidine operon of salmonella. J.Bacteriol. 101,490504. (6) Hartman, P. E., Ames, B. N., Roth, J. R., Barnes, W. M., and Levin, D. E. (1986)Target sequencesfor mutagenesisin salmonellahistidinerequiring mutants. Enuiron. Mutagen. 8, 631-641. (7) McCann, J., Spingain, N. E., Kobori, J., and Ames, B. N. (1975) Detection of carcinogensas mutagens-bacterial tester strains with R-factor plasmids. R o c . Natl. Acad. Sci. U.S.A. 72,979-983. (8) Isono, K.,and Yourno, J. (1974)Chemical carcinogensas frameshift mutagens-salmonella DNA sequence sensitive to mutagenesis by polycycliccarcinogens. Proc.NatZ. Acad. Sci. U.S.A. 71,1612-1617. (9) Mukai, F. H.,and Goldstein, B. D. (1976)Mutagenicity of malonaldehyde,a decompositionproduct of peroxidized polyunsaturated fatty acids. Science 191,868-869. (IO) Marnett, L. J., and Tuttle, M. A. (1980) Comparison of the mutagenicities of malondialdehyde and the side products formed during ita chemical synthesis. Cancer Res. 40,276-282. (11) Basu, A. K., and Marnett, L. J. (1983)Unequivocal demonstration that malondialdehyde is a mutagen. Carcinogenesis 4,331-333.

836 Chem. Res. Toxicol., Vol. 6,No. 6,1993 (12) Crawford, D. L., Sinnhuber, R. O., Stout, F. M., Oldfield, J. E., and Kaufmes, J. (1965)Acute toxicity of malonaldehyde. Toxicol. Appl. Pharmacol. 7, 826-832. (13) Spalding, J. W. (1988) Toxicology and carcinogenesis studies of malondialdehyde sodium salt (3-hydroxy-2-propena11, sodium salt) in F344/N rats and B6C3F1 mice. Natl. Toxicol. Program Tech. Rep. Ser. No. 331, 5-13. (14) Basu, A. K., and Marnett, L. J. (1984) Molecular requirements for the mutagenicity of malondialdehyde and related acroleins. Cancer Res. 44, 2848-2854. (15) Seto, H., Okuda, T., Takesue, T., and Ikemura, T. (1983) Reaction of malonaldehyde with nucleic acid. I. Formation of fluorescent pyrimido [1,2-a]purin-l0(3H)-one nucleosides. Bull. Chem. SOC. Jpn. 56, 1799-1802. (16) Seto, H., Seto, T., Takesue, T., and Ikemura, T. (1986) Reaction of malonaldehyde with nucleic acid. 111. Studies of the fluorescent substances released by enzymatic digestion of nucleicacids modified with malonaldehyde. Chem. Pharm. Bull. 34,5079-5085. (17) Marnett, L. J., Basu, A. K., O’Hara, S. M., Weller, P. E., Rahman, A. F. M. M., and Oliver, J. P. (1986) Reaction of malondialdehyde with guanine nucleosides: formation of adducts containing oxadiazabicyclononeneresidues in the base-pairing region. J.Am. Chem. SOC. 108, 1348-1350. (18) Basu, A. K., OHara, S. M., Valladier, P., Stone, K., Mob, O., and Marnett, L. J. (1988) Identification of adducts formed by reaction of guaninenucleosideswithmalondialdehyde and structurallyrelated aldehydes. Chem. Res. Toxicol. 1,53-59. (19) Marinelli, E. R., Johnson, F., Iden, C. R., and Yu, P. L. (1990) Synthesis of 1,N2-(1,3-propano)-2’-deoxyguanosineand incorporation into oligodeoxynucleotides: a model for exocyclic acroleinDNA adducts. Chem. Res. Toxicol. 3,49-58. (20) Benamira, M., Singh, U., and Marnett, L. J. (1992) Site-specific frameshift mutagenesis by a propanodeoxyguanosine adduct positioned in the (CpG), hot-spot of salmonella typhimurium hisD3052 carried on an M13 vector. J. Biol. Chem. 267, 22392-22400. (21) Kouchakdjian, M., Eisenberg, M., Live, D., Marinelli, E., Grollman, A. P., and Patel, D. J. (1990) NMR studies of an exocyclic 1JPpropanodeoxyguanosineadduct (X) located oppositedeoxyadenosine (A)in DNA duplexes a t basic pH: simultaneous partial intercalation of X and A between stacked bases. Biochemistry 29,4456-4465. (22) Kouchakdjian, M., Marinelli, E., Gao, X. L., Johnson, F., Grollman, A,, and Patel, D. (1989) NMR studies of exocyclic l,N2-propanodeoxyguanosine adducts (X) opposite purines in DNA duplexes: protonated X(syn):A(anti) pairing (acidic pH) and X(syn):G(anti) pairing (neutral pH) a t the lesion site. Biochemistry 28,5647-5657. (23) Huang,P.,andEisenberg,M. (1992)The three-dimensionalstructure in solution (pH 5.8) of a DNA 9-mer duplex containing 1JPpropanodeoxyguanosineopposite deoxyadenosine. Restrained molecular dynamics and NOE-based refinement calculations. Biochemistry 31, 6518-6532. (24) Huang, P., Patel, D. J., and Eisenberg, M. (1993) Solution structure of the exocyclic 1JP-propanodeoxyguanosine adduct opposite deoxyadenosine in a DNA nonamer duplex at pH 8.9. Model of pH-dependent conformational transition. Biochemistry 32,38523866. (25) Kouchakdjian, M., Eisenberg, M., Johnson, F., Grollman, A. P., and Patel, D. J. (1991) Structural features of an exocyclic adduct positioned opposite an abasic site in a DNA duplex. Biochemistry 30,3262-3270. (26) Plum,G. E., Grollman,A. P., Johnson,F., and Breslauer, K. J. (1992) Influence of an exocyclic guanine adduct on the thermal stability, conformation, and melting thermodynamics of a DNA duplex. Biochemistry 31, 12096-12102. (27) Brush, C. K., Stone, M. P., and Harris, T. M. (1988) Selective reversible deuteriation of oligodeoxynucleotides: simplification of two-dimensionalnuclear OverhausereffedNMRspectralassignment of a non-selkomplementary dodecamer duplex. Biochemistry 27, 115-122. (28) Plateau, P., and Gueron, M. (1982) Exchangeable proton NMR without base-line distortion, using new strong-pulse sequences. J. Am. Chem. SOC. 104,7310-7311. (29) Hore, P. J. (1983)Anew method for water suppression in the proton NMR spectra of aqueous solutions. J. Magn. Reson. 64,539-542. (30) Bax, A., Sklenar, V., and Clore, G. M. (1987) Water suppression in two-dimensional spin-locked nuclear magnetic resonance experiments using a novel phase-cycling procedure. J. Am. Chem. SOC. 109,6511-6513. (31) Sklenar, V., Brooks, B. R., Zon, G., and Bax, A. (1987) Absorption mode two-dimensional NOE spectroscopy of exchangeable protons in oligonucleotides. FEBS Lett. 216, 249-252.

Singh et a1. (32) Marion, D., Ikura, M., and Bax, A. (1989) Improved solvent suppression in one- and two-dimensional NMR spectra by convolution of time-domain data. J. Magn. Reson. 84,425-430. (33) Rauce, M., Bodenhausen, G., Wagner, G., Wuthrich, K., and Emst, R. R. (1985) A systematic approach to the suppression of J cross peaks in 2D exchange and 2D NOE spectroscopy. J.Magn. Reson. 62,497-510. (34) Bodenhausen, G., Wagner,G.,Rance,M., Sorensen,0.W., Wuthrich, K., and Emst, R. R. (1984) Longitudinal two-spin order in 2D exchange spectroscopy (NOESY). J. Magn. Reson. 89, 542-550. (35) Bax, A., and Davis, D. G. (1985) MLEV-17-based two-dimensional homonuclear magnetization transfer spectroscopy. J.Magn. Reson. 65,355-360. (36) Stewart, J. J. P. (1983) MOPAC. QCPE 3,43. (37) Brunger, A. T. (1990) Refinement of three-dimensional structures of proteins and nucleic acids. In Molecular Dynamics (Goodfellow, J. M., Ed.) pp 137-178, CRC Press, Boca Raton. (38) Brooks, B. R., Bruccoleri, R. E., Olafson, B. D., States, D. J., Swamianathan, S., and Karplus, M. (1983) CHARMM: a program for macromolecularenergy,minimization,and dynamicscalculations. J. Comput. Chem. 4, 187-217. (39) Berendsen, H. J. C., Postma, J. P. M., van Gunateren, W. F., DiNola, A., and Haak, J. R. (1984) Molecular dynamics with coupling to an external bath. J.Phys. Chem. 81, 3684-3690. (40) Arnott, S., and Hukins, D. W. L. (1972) Optimised parameters for A-DNA and B-DNA. Biochem. Biophys. Res. Commun. 47,15041509. (41) Arnott, S., and Hukins, D. W. L. (1973) Refinement of the structure of B-DNA and implications for the analysis of X-ray diffraction data from fibers of biopolymers. J.Mol. Biol. 81, 93-105. (42) Nilsson, L., and Karplus, M. (1986) Empirical energy functions for energy minimiition and dynamics of nucleic acids. J. Comput. Chem. 7, 591-616. (43) Clore, G. M., Gronenborn, A. M., Carlson, G., and Meyer, E. F. (1986) Stereochemistry of binding of the tetrapeptide acetyl-proala-pro-tyr-NHz to porcine pancreatic elastase. Combined use of two-dimensionaltransferred nuclear overhauser enhancement measurements, restrained molecular dynamics, X-ray crystallography and molecular modelling. J. Mol. Biol. 190, 259-267. (44) Ryckaert, J.-P., Ciccotti,G., and Berendsen,H.J. C. (1977) Numerical integration of the Cartesian equations of motion of a system with constraints: molecular dynamics of n-alkanes. J. Comput. Phys. 23, 327-341. (45) Gao, X. L., and Patel, D. J. (1988) NMR studies of echinomycin bisintercalation complexes with d(AI-C2-G3-T4)and d(Tl-C2-G3A4) duplexes in aqueous solution: sequence-dependent formation of Hoogsteen Al:T4 and Watson-Crick Tl:A4 base pairs flanking the bisintercalation site. Biochemistry 27, 1744-1751. (46) Gueron, M., Kochoyan, M., and Leroy, J. L. (1987) A single mode of DNA base-pair opening drives imino proton exchange. Nature 328,89-92. (47) Koo, H. S., and Crothers, D. M. (1988)Calibration of DNA curvature and aunified description of sequencedirected bending. Proc. Natl. Acad. SC~. U.S.A 85,1763-1767. (48) Burnouf, D., Koehl, P., and Fuchs, R. P. P. (1989) Single adduct mutagenesis: strong effect of the position of a single acetylaminofluorene adduct within a mutation hot spot. Proc. Natl. Acad. Sci. U.S.A. 86,4147-4151. (49) Koehl, P., Burnouf, D., and Fuchs, R. P. P. (1989) Construction of plasmids containing a unique acetylaminofluorene adduct located within amutation hot spot. A new probe for frameshift mutagenesis. J. Mol. Biol. 207, 355-364. (50)Timeit, Y., Westhof,E., Fuchs, R. P. P., andMoraa,D. (1989)Unusual helical packing in crystals of DNA bearing a mutation hot spot. Nature 341,459-462. (51) Heinemann, U., Alings, C., and Bansal, M. (1992) Double helix conformation, groove dimensions and ligand binding potential of a G/C stretch in B-DNA. EMBO 11,1931-1939. (52) Grollman, A. P. (1990) Site specific mutagenesis. Prog. Clin. Biol. Res. 3408, 61-70. (53) Sagher, D., and S t r a w , B. (1983) Insertion of nucleotides opposite apurinic/apyrimidinic sites in deoxyribonucleic acid during in vitro synthesis: uniqueness of adenine nucleotides. Biochemistry 22, 4518-4526. (54) Straws, B. S. (1985) Cellular aspects of DNA repair. Adu. Cancer Res. 46, 45-105.