Probing Nucleic Acid Geometries: Oligonucleotides Containing 2'-O

Bioconjugate Chem. 1996, 7, 108−120

108

Probing Nucleic Acid Geometries: Oligonucleotides Containing 2′-O-Phenethyladenosine at Specific Sites Hemant M. Deshmukh and Arthur D. Broom* Department of Medicinal Chemistry, College of Pharmacy, University of Utah, Salt Lake City, Utah 84112. Received August 10, 1995X

The synthesis of 2′-O-phenethyladenosine and its incorporation into an oligodeoxyribonucleotide and a chimeric oligodeoxy/oligoribonucleotide are described. The study was designed to determine the consequences of site-specific induction of such a small, flexible hydrophobic ligand into a nucleic acid. Through the use of optical (Tm, circular dichroism) spectroscopy and nuclear magnetic resonance, it was discovered that this substitution destabilized the duplex and that the benzene ring of the phenethyl group lies in the major groove of both the DNA analog and the DNA-RNA duplex structure. The plane of the benzene ring lies perpendicular to the stacked bases of the oligonucleotide which, in other respects, exhibits relatively normal B-type geometry for the oligodeoxynucleotide and a modified A-type geometry in the DNA-RNA duplex. This finding has implications for the synthesis of oligodeoxynucleotides containing major groove binders as ligands.

INTRODUCTION

The specificity inherent in Watson-Crick and Hoogsteen hydrogen bonding among the purines and pyrimidines which comprise interacting single-stranded nucleic acids has stimulated great interest in both academic and industrial laboratories. A major focus of this interest has been the design and synthesis of antisense oligodeoxynucleotides and their utilization in the inhibition of gene expression by duplex formation with complementary single-stranded regions of messenger RNA (1-7). Additionally, substantial work has been carried out using Hoogsteen hydrogen bonding in the design and synthesis of triplex-forming oligonucleotides (1, 8-11). A number of investigators have recognized that attachment of intercalating (12, 13), alkylating (14), cleaving (15-17), or photo cross-linking (18, 19) functional groups to the 3′- or 5′-end of antisense oligonucleotides should afford enhanced affinity and/or cleavage specificity for the oligonucleotide-directed interruption of gene expression. In very few cases, however, has any exploration of the fate of ligands attached to the 2′-hydroxyl of a ribonucleotide incorporated into an oligonucleotide been undertaken. While a substantial number of studies have been published on 2′-O-alkyl oligos (20-22), the primary motivation for these efforts was enhanced stability toward ribonuclease hydrolysis of synthetic RNA molecules. It has been reported that anthraquinone moieties may be linked to the 2′-hydroxyl of uridine via a methylene linker (23) or to U, C, and A via a more complex and considerably longer linker (24). In both cases, the anthraquinone was said to intercalate on the basis of observed duplex stabilization. A similar observation was made for substitution of an anthraquinone moiety at the 1′-position of 2′-deoxyuridine on a 10-atom tether (25). On the other hand, attachment of the larger pyrene group to the 2′-OH of uridine led to destabilization of the duplex when the modified nucleoside was placed at an internal position (26). Recently, we have synthesized 2′-O-(anthraquinon2-ylmethyl)adenosine, incorporated it into oligodeoxynucleotides, and demonstrated unequivocally using NMR X Abstract published in Advance ACS Abstracts, December 1, 1995.

1043-1802/96/2907-0108$12.00/0

and other techniques that intercalation does occur, even with only a one-carbon tether (27). However, no study has been reported that describes the consequences of site-specific introduction of a small, flexible hydrophobic ligand, such as phenethyl, at the 2′position of a ribonucleotide incorporated into a nucleic acid. In principle, the benzene ring in such a construct could assume one of four orientations: (1) it could intercalate between the base of the nucleoside to which it is attached and its 3′-neighboring base, (2) it could lie along the minor groove of the duplex, (3) it could reside in the major groove of the duplex, or (4) it could extend into the aqueous solution. In the present paper, we will demonstrate a result that was to us counterintuitive. That is, in both DNA duplexes and DNA-RNA hybrids, the benzene ring resides in the major groove of the duplex in an orientation with the plane of the phenyl ring essentially perpendicular to that of the planes of the stacked bases. EXPERIMENTAL PROCEDURES

General Methods. 1H-NMR spectra were recorded on either an IBM AF 200 MHz FT-NMR spectrometer or a Varian Unity 500 MHz spectrometer. NMR samples of nucleosides were prepared using (CD3)2SO or CDCl3, with tetramethylsilane (TMS) or solvent peak as internal standard. Electrospray mass spectra were recorded on a Vestec 201 ionization instrument with a quadrupole mass analyzer. Thin-layer chromatography (TLC) was performed on Kieselgel 60 F254 aluminum-backed silica gel sheets. CD spectra were recorded on a Jasco J-720 spectropolarimeter. Elemental analyses were performed by M-H-W Laboratories in Phoenix, AZ. Mass spectra were recorded on either a Finnegan MAT 95 or a VG 7050E mass spectrometer. The oligonucleotides were synthesized on an Applied Biosystems 380B instrument. Materials. All chemicals were purchased from Aldrich Chemical Co. unless otherwise specified. The enzymes, snake venom phosphodiesterase (SVP) and bacterial alkaline phosphatase (BAP), were purchased from Sigma. Preparation of Oligonucleotides. Four oligonucleotides were prepared using standard machine chemistry and commercially available reagents except for the 2′O-phenethyladenosine derivative described below. For initial studies, including Tm measurements, syntheses © 1996 American Chemical Society

Bioconjugate Chem., Vol. 7, No. 1, 1996 109

Probing Nucleic Acid Geometries Table 1. Sequences of Various Oligonucleotides and Their Tms (°C) 5 6 7 8

sequence

name

Tm

5′d(CGC ACA TGT GCG)3′ 5′d(CGC A*CA TGT GCG)3′ A* ) 2′-O-phenethyladenosine 5′d(CGC ACA) r(UGU GCG)3′ 5′d(CGC A*CA) r(UGU GCG)3′ A* ) 2′-O-phenethyladenosine

standard 12-mer phenethyl 12-mer

41.6 29.4

standard hybrid phenethyl hybrid

36 31

were carried out on a 1 µmol scale. Subsequently, 10 µmol syntheses were performed in order to obtain sufficient material for NMR studies. As expected, yields at steps corresponding to ribonucleotide introduction were somewhat lower than those for standard DNA oligonucleotide synthesis. The four self-complementary oligonucleotides used to study the role of 2′-O-phenethyl substitution are given in Table 1, along with their Tm values. Tm Determination. All Tm values were determined on a Hewlett-Packard diode array spectrophotometer equipped with an electronic temperature-controlling device. Absorbance at 260 nm was measured at 1 °C intervals, and the cuvette was equilibrated at each temperature for 2 min. For duplex formation, the solution was heated to ∼70 °C and then cooled slowly. Duplexes were kept at 5 °C overnight before Tm measurements were carried out. Duplexes were formed at 1.58 × 10-4 M concentration (stock solution) in 100 mM NaCl containing 20 mM phosphate buffer (pH 7). The stock solution was diluted 50-fold at 4 °C with the same buffer for Tm measurement. High-Performance Liquid Chromatography (HPLC) Purification and Analysis of Oligomers. All HPLC purifications and analyses were performed after detritylation on a Hitachi D-6200 HPLC system equipped with an L-3000 diode array spectrophotometer. Oligonucleotides synthesized on a 1 µmol scale were purified on a preparative Whatman Partisil ODS reverse-phase column. Oligomers prepared on a 10 µmol scale were purified on a Sephadex G25 column (100 × 2.5 cm). Elution was performed with distilled water at a flow rate of 20 mL/h. Purified oligomers were analyzed on a Rainin Microsorb C18 analytical column (25 cm × 4.4 mm). Solvents used for elution were as follows: solvent A, 95% 50 mM ammonium bicarbonate buffer (pH 7)/5% acetonitrile; and solvent B: 80% 50 mM ammonium bicarbonate buffer (pH 7)/20% acetonitrile. The following gradient was used for purification of oligomers: 100% A (0-4 min, linear 100% A to 100% B (4-15 min), 100% B (15-25 min), linear 100% B to 100% A (25-32 min), re-equilibration with 100% A (32-40 min); flow rate, 2 mL/min. The following gradient was used for analysis of oligomers: 100% A (0-4 min), linear 100% A to 100% B (410 min, 100% B (10-15 min), linear 100% B to 100% A (15-20 min), re-equilibration with 100% A (20-30 min); flow rate, 1 mL/min. Nuclease Digestion and Analysis of Modified and Unmodified Oligomers. A mixture of oligonucleotide (0.5 OD units) with 50 mM Tris-HCl (pH 8), 10 mM MgCl2, SVP (3 units), and BAP (3 units in 100 µL of distilled water) was incubated at 37 °C for 5 h; 15 µL of 5 M sodium acetate was added to the hydrolysis mixture, followed by the addition of 230 µL of absolute ethanol. This mixture was chilled at -70 °C for 30 min and centrifuged at 30000g for 5 min. The supernatant was diluted with 1 mL of 95% ethanol, chilled, centrifuged, decanted, and evaporated to dryness. The residue was redissolved in distilled water for HPLC analysis.

HPLC analysis for unmodified oligonucleotide hydrolysates was performed on a C18 column; a C8 column was used for the analysis of modified oligonucleotide hydrolysates. Elution was performed isocratically with a 50 mM phosphate buffer (pH 4)/methanol linear gradient. NMR Experiments on Oligonucleotides. One- and two-dimensional (1D and 2D) proton data sets were collected on a Varian Unity 500 MHz spectrometer. NMR samples were prepared to a concentration of oligonucleotide of ∼1.6 mM. One-dimensional temperature-dependent 500 MHz 1H data sets were collected in 90% H2O buffer/10% D2O. One-dimensional protondecoupled 500 MHz 31P data sets were collected in 100% buffered D2O solution. The oligomers were lyophilized twice from 99.996% D2O and then dissolved in 99.996% D2O containing 100 mM sodium chloride and 20 mM phosphate buffer (pH 7). Two-dimensional phase-sensitive NOESY spectra of all the oligomers were collected using a 250 or 400 ms mixing time. All data sets were acquired in the hypercomplex mode, with 256 increments in the t1 dimension, 32 or 64 scans per free induction decay (fid), and 2048 complex points in t2. The t2 increments were zero-filled to 2048 points and transformed with a Gaussian apodization function. Two-dimensional proton phase-sensitive double-quantum filtered COSY (DQF-COSY) experiments were collected using the standard pulse sequence with a 2 µs pulse repetition time and homospoil of 1 ms. Data sets were collected in hypercomplex mode with 256 t1 increments, 32 or 64 scans per fid, and 2048 complex points in T2. The t2 dimension was processed with a Gaussian filter; the t1 increments were zero filled to 2048 points and transformed with a Gaussian apodization function. Molecular Modeling. Molecular modeling studies were performed using Silicon Graphics IRIS or Indigo workstations. Molecules were visualized using BIOSYM software; minimizations and dynamics were performed using the DISCOVER program. For simulation of nucleosides, cvff potentials were used. All nucleosides were first minimized using the steepest gradient for 500 iterations and then a biconjugate gradient for 2000 iterations. To obtain a global minimum, the molecular dynamic simulations were used. All dynamics were performed at 300 K for 50 ps, including a 20 ps equilibration period. The low-energy conformers obtained in the molecular dynamics simulation were further minimized to obtain a family of low-energy conformers. The protocol used for initial minimization was used for these minimizations also. Oligonucleotides were visualized using the bipolymer module of the InsightII software. AMBER potentials were used during minimization of oligonucleotide conformations. When the phenethyl groups were attached to the 2′-OH, the software was unable to assign a potential for C5 of adenosine. Potential “CB” was assigned to this carbon manually. Only minimizations were performed on the oligonucleotides. The phenethyl group was placed in the major groove by torsioning the bonds between the benzene ring and the 2′-hydroxyl. The torsions were removed, and the oligonucleotides were minimized first by 500 iterations of the steepest gradient minimizer and then by 2000 iterations of the conjugate gradient minimizer. Partial charges on the backbone phosphorus nuclei were reduced to 0.8, while those on oxygen were reduced to -0.5. In some instances, all the hyrogen-bonded atoms of the first three base pairs were fixed in space before minimization. 2′-O-Phenethyladenosine (1). Adenosine (10 g, 37.4 mmol) was dissolved in anhydrous DMF (250 mL), and

110 Bioconjugate Chem., Vol. 7, No. 1, 1996

the solution was cooled to 0 °C. Sodium hydride (1.1 g, 45 mmol, prewashed with hexanes) was added to the adenosine solution. The reaction mixture was vigorously stirred for 90 min. Phenethyl tosylate (10 g, 36 mmol) in anhydrous DMF (100 mL) was added in portions over a period of 2 h, during which time the reaction mixture was allowed to warm to room temperature. After 2 h at room temperature, the mixture was heated to 60 °C and another 8.0 g of phenethyl tosylate was added in portions over a 2 h period. The reaction mixture was then allowed to cool to room temperature, and 1.0 g of NaH (60% dispersion in mineral oil) was added. The reaction mixture was again heated to 60 °C, a final aliquot of phenethyl tosylate (8 g) was added, and the mixture was stirred overnight at 60 °C. Solvent was evaporated in vacuo to yield a residue to which water (75 mL) was added. The aqueous layer was washed with hexanes (3 × 50 mL) to remove styrene produced from phenethyl tosylate and then with ethyl acetate. The ethyl acetate was dried over sodium sulfate and evaporated in vacuo to afford an oil. The product was crystallized from ethanol to provide 2.15 g (15.2%): UV (H2O) λmax 254 nM ( ) 21 880); 1H-NMR [(CD3)2SO] δ 8.37 (s, 1H, H8), 8.14 (s, 1H, H2), 7.39 (s, 2H, NH2), 7.08 (m, 5H, Ar), 6.00 (d, 1H, 1′-H, J1′,2′ ) 6.2 Hz), 5.45 (m, 1H, 5′-OH), 5.22 (d, 1H, 3′-OH), 4.52 (m, 1H, 2′-H), 4.33 (m, 1H, 3′-H), 4.00 (m, 1H, 4′-H), 3.85-3.38 (complex m, 4H, 5′-H, 5′′-H, OCH2), 2.72 (t, 2H, CH2Ph); MS m/z 372 (M + 1)+. Anal. Calcd for C18H21N15O4: C, 58.20; H, 5.70; N, 18.86. Found: C, 58.06; H, 5.74; N, 18.79. N6-Benzoyl-2′-O-Phenethyladenosine (2). The title compound was synthesized using the transient protection procedure of Jones (28). 2′-O-Phenethyladenosine (2.70 g, 5.6 mmol) was dissolved in pyridine (50 mL). Trimethylsilyl chloride (5.05 mL, 39.6 mmol) was added, and the solution was stirred for 45 min. Benzoyl chloride (3.24 mL, 38 mmol) was added, and the reaction mixture was allowed to stir for 3.5 h. The mixture was cooled in an ice bath, and 10 mL of cold water was added. After 5 min, 20 mL of 30% aqueous ammonia was added, and the mixture was stirred at room temperature for 30 min. The solvent was removed under vacuum, and the residue was suspended in water and extracted with ethyl acetate (3 × 50 mL). The combined organic layers were washed with water (2 × 50 mL), followed by brine, dried over sodium sulfate, and evaporated to dryness. The compound was not isolated but used directly in the next reaction. Judging from the NMR spectrum, the yield of the reaction was about 75%: 1H-NMR [(CD3)2SO] δ 11.22 (s, 1H, NH), 8.74 and 8.70 (2s, 2H, H8, H2), 6.12 (d, 1H, 1′-H, J1′2′ ) 6.0 Hz); MS m/z 476 (M + 1)+. N6-Benzoyl-5′-(dimethoxytrityl)-2′-O-phenethyladenosine (3). Dimethoxytritylation of 2 was carried out according to the procedure of Wu et al. (29). Compound 2 (1.95 g, 5.4 mmol) was dried by coevaporation with pyridine and dissolved in pyridine (20 mL). Dimethoxytrityl chloride (2.92 g, 8.6 mmol) was added to the solution, and the reaction mixture was allowed to stir for 12 h at 4 °C. Thin-layer chromatography indicated the presence of starting material; therefore, an additional 500 mg of dimethoxytrityl chloride was added, and the reaction continued for an additional 12 h. Methanol (5 mL) was added to quench the reaction; the solvent was evaporated in vacuo, and the residue was dissolved in ethyl acetate (50 mL). The ethyl acetate layer was washed with 5% sodium bicarbonate (3 × 75 mL), water (75 mL), and brine and dried over sodium sulfate. The product was purified by silica gel chromatography with a gradient of hexane in chloroform (20 to 0%). Triethylamine (0.1%) was added to the eluting solvent to prevent

Deshmukh and Broom

degradation of the tritylated product: yield 2.4 g (75%); UV (H2O) λmax 234 nm ( 27 175), 282 ( 16 736); 1H-NMR [(CD3)2SO] δ 11.30 (s, 1H, NH), 8.62 (s, 1H, H8), 8.55 (s, 1H, H2), 6.20 (d, 1H, 1′-H), 5.34 (d, 1H, 3′-OH), 4.74 (t, 1H, 2′-H), 4.48 (m, 1H, 3′-H), 4.18 (m, 1H, 4′-H), 3.953.57 (2m, 2H, 5′,5′′-H), 3.70 (s, 6H, OCH3); MS m/z 777.3141 (M + 1)+, calcd 777.3162 (difference ) 2.1 mmu). The purity of the compound was confirmed by normal-phase HPLC. N6-Benzoyl-5′-(dimethoxytrityl)-2′-O-phenethyladenosine 3′-O-(Cyanoethyl)-N,N-diisopropylphosphoramidite (4). The title compound was synthesized according to the general procedure of Kierzek et al. (30). Acetonitrile (15 mL) and β-(cyanoethyl)-N,N,N′,N′-tetraisopropyl phosphordiamidite (0.47 mL, 1.4 mmol) were added to 2 (1 g, 1.28 mmol) and diisopropylammonium tetrazolide (109 mg, 0.64 mmol, dried in vacuo for 3 h). After 12 h, thin-layer chromatography indicated the presence of some starting material; hence, an additional 0.1 mL of phosphordiamidite was added, and the reaction was continued for an additional 12 h. The reaction mixture was diluted with saturated sodium bicarbonate (50 mL) and extracted with dichloromethane (2 × 50 mL). Combined organic extracts were washed three times with brine, dried over sodium sulfate, and concentrated in vacuo to a dry foam. The foam was purified by silica gel chromatography using 20% hexane in chloroform and 1% triethylamine: yield 700 mg (56%); 31P-NMR δ 150.86, 150.22. RESULTS

Chemistry. Synthesis of the nucleoside required for this study, 2′-O-phenethyladenosine, was rather difficult. The 3′,5′-O-tetraisopropyldisiloxane protection/strong organic base-mediated alkylation methodology of Sproat (31) was unsuccessful in this case, as was the 2′,3′-Ostannylene chemistry devised by Moffatt (32) and successfully applied to the synthesis of 2′-O-(anthraquinon2-yl methyl)adenosine by Yamana et al. (23). The simple approach of Ts’o (33) succeeded, albeit giving rise to 2′O-phenethyladenosine in a modest 15% yield. The obvious difficulty leading to poor yields in this reaction is the favored elimination giving rise to styrene under the strongly basic (sodium hydride) conditions of this reaction. A number of leaving groups were examined, most of which were either too unreactive (Br) or eliminated too readily (OMs,I). The tosyloxy leaving group was found to be the best compromise and permitted us to obtain adequate amounts of 2′-O-phenethyladenosine. Since the 3′-O-phenethyladenosine is also a product of the reaction, it was necessary to establish unequivocally the site of alkylation. This was accomplished by means of 2D COSY and NOESY experiments. Briefly, both NOESY and COSY spectra revealed that the signal at 6.00 δ corresponding to H1′ was coupled to H2′ at 4.52 which, in turn, was coupled to H3′ at 4.33. H3′ showed strong NOESY and COSY cross-peaks to the OH signal at 5.22 δ, whereas the H2′ signal showed no connectivity to exchangeable OH protons. This unequivocally established the position of substitution of the phenethyl group at 2′. Protection of 2′-O-phenethyladenosine using standard chemistry provided N6-benzoyl-5′-O-(dimethoxytrityl)-2′O-phenethyladenosine (cyanoethyl)-N,N-diisopropylphosphoramidite. This synthon was used in a standard machine synthesis to give the DNA 12-mer 6 and the DNA-RNA hybrid 8. Predictably, yields at the step incorporating 2′-O-phenethyladenosine and yields of the RNA components of the hybrids were modest, but reac-

Probing Nucleic Acid Geometries


Figure 1. Circular dichroism spectra for the standard 12-mer (A) and the phenethyl 12-mer (B) at various temperatures.

tions carried out on a 10 µM scale afforded sufficient material for all the physicochemical measurements undertaken. Melting temperatures were obtained for both standard oligomers 5 and 7 and the modified oligomers 6 and 8, and the data are presented in Table 1. Substitution of phenethyloxy for proton (6 vs 5) led to a significant destabilization of the duplex. In order to determine whether this was a result arising largely from inclusion of a ribonucleoside unit as opposed to a deoxyribonucleoside unit, melting curves were also obtained on oligomers in which the modified nucleoside was adenosine (Tm ) 37.4 °C), 2′-O-methyladenosine (Tm ) 34.6 °C), and 2′O-benzyladenosine (Tm ) 29.0 °C). Clearly, the data reveal that the observed destabilization seems to result largely from the steric bulk of the substituent rather than from conformational alterations imposed by the 2′-OH. It is worth noting in this regard that the anthraquinon2-ylmethyl substitution previously reported (23, 27) led to significant stabilization of the duplex resulting from intercalation of the anthraquinone moiety on the 3′-side of the adenosine residue. Circular Dichroism Spectropolarimetry. The CD spectra for the standard 12-mer and the phenethyl 12mer at various temperatures are shown in Figure 1. It is readily apparent that, although subtle differences exist in the spectra of compounds 5 and 6, both spectra at all temperatures are highly characteristic of B-DNA and show little evidence of significant deviations from normal DNA duplex formation (34). NMR Spectroscopy of Oligonucleotides. Solutions for 1D NMR in which imino proton signals were to be observed were prepared in H2O solutions containing 10% D2O. The water peak was suppressed using a water suppression pulse sequence. The imino proton spectra for the standard 12-mer at various temperatures are shown in Figure 2. The self-complementary structure of the

Figure 2. Imino proton spectra for the standard 12-mer at various temperatures.

oligonucleotide divides the duplex into identical halves; consequently, only six imino proton resonances should be observed when all bases of the duplex are paired. At 5 °C, all six imino protons were observed. The two downfield resonances correspond to AT base pairs, and the four upfield resonances correspond to GC base pairs (35). The downfield signal of the central pair of resonances is the first to melt out between 5 and 15 °C, indicative of fraying of the terminal GC base pair. Peaks sharpen with increasing temperature and the consequent increased tumbling motion of the molecules to about 45 °C, at which time melting begins and peaks again broaden


Figure 3. Imino proton spectra for the phenethyl 12-mer at various temperatures.

and disappear. The remaining central resonance is the next to disappear as predicted for the penultimate GC base pair. The Tm plot derived from this information reveals that the melting temperature is between 50 and 55 °C. This value is somewhat higher than that from the UV melting experiments, a typical result considering the much higher concentration of oligomer driving the equilibrium toward the duplex form (27). The imino proton spectra for the phenethyl 12-mer at various temperatures are shown in Figure 3. Although the spectra are generally similar in appearance to those obtained for oligo 5, the differences are significant. At 5 °C, three pairs of broad signals are observed. Again, the most downfield signals correspond to AT base pairs. One-

Deshmukh and Broom

dimensional NOE difference spectroscopy was used to assign these two resonances (data not shown). The hydrogen-bonded N3H of T is close to the adenine H2 proton. Consequently, irradiation of N3H leads to a change of intensity in the H2 proton signal. Irradiation of the upfield resonance of the pair gave rise to change in the intensity of a peak at 7.26 δ. This peak was identified (vide infra) by two-dimensional NMR spectroscopy as H2 of A6, identifying the upfield resonance as T7-N3H. Selective irradiation of the more downfield imino resonance could not be accomplished because of peak overlap. Irradiation at 13.4 δ led to a change in the intensities of two peaks, one of which was H2 of A6. The second peak was later identified as H2 of A4 by 2D NOESY spectroscopy. Thus, the most downfield proton corresponds to T9-N3H which hydrogen bonds to A4. As the temperature is increased, the resonances become sharper until 25 °C, at which time peak broadening indicates the onset of melting. Careful inspection of the spectra reveals that the peaks corresponding to the G2C11 imino proton and the T9-A4 imino proton are broadening simultaneously. The former was anticipated since it forms the penultimate base pair. Broadening of the A4-T9 imino proton signal indicates local denaturation of the helix, which must be resulting from the influence of the phenethyl group. The NMR-derived melting temperature is between 35 and 40 °C, also higher than that observed by UV spectroscopy. For two-dimensional spectra, oligomers 5 and 6 were dissolved in D2O containing 0.1 M NaCl and 20 mM phosphate buffer (pH 7) and lyophilized several times from 99.996% D2O. The oligomers were annealed by heating each sample to 65 °C and slowly cooling to room temperature. Spectra for oligomer 5 were collected at 5 °C, and because of significant overlap at lower temperatures, spectra for oligomer 6 were collected at 15 °C. A mixing time of 250 ms was used in DNA oligomer NOESY experiments. Figure 4 shows the expanded contour plot of the NOESY spectrum for the standard oligo 5, showing

Figure 4. Expanded contour plot (aromatic H1′ region) of the NOESY spectrum for the standard 12-mer.



Figure 5. Expanded contour plot (aromatic H1′ region) of the NOESY spectrum for the phenethyl 12-mer.

connectivities between aromatic protons and H1′ protons. One may easily observe that the peaks are of rather uniform intensity, characteristic of normal B-DNA geometry with each aromatic (H6/H8) proton almost equidistant from its own H1′ and the previous H1′ (36). This region of the spectrum was used to assign all the base and H1′ protons. The aromatic methyls of T7 and T9 provide convenient starting points for the assignments. Each methyl shows cross-peaks with the H6 proton of its own base and the aromatic proton of the preceding base. This resulted in quick identification of A6, T7, G8, and T9. The rest of the nucleosides were identified by performing a “NOESY-walk” in this region, as previously described (36), giving rise to the assignments illustrated in Figure 4. The assignments were confirmed using the aromatic H2′ region of the NOESY spectrum and by using DQF-COSY spectroscopy (data not shown). In Figure 5, similar data for the aromatic H1′ region of the NOESY spectrum of oligo 6, the phenethyl 12-mer, are presented. Assignments were made in the same way as for oligo 5 and, in general, are consistent with overall B-DNA geometry. Examination of the NOESY data reveals the cross-peak between H8 of A4 and its own H1′ is considerably weaker than the rest of the peaks. Since the intensity of the cross-peak depends upon the distance between the protons, the distance between H8 and H1′ of A4 must be longer than normal. This could result from a change in sugar pucker or from a change in the glycosyl bond angle. The second factor is regarded as less likely since a change in bond angle would alter the distance between A4-H8 and C3-H1′. No such change was observed. The sugar pucker of A4 was qualitatively analyzed in a DQF-COSY experiment. The intensity of a crosspeak between two protons in this experiment depends on the scalar coupling constant between them which, in turn, depends upon the dihedral angle. Figure 6 shows the expanded H1′-H2′ region of the DQF-COSY spectrum for the phenethyl 12-mer. The arrow in the figure points at the cross-peak for H1′-H2′ of A4, whereas the other peaks correspond to the remain-

ing H1′-H2′ cross-peaks. Most studies involving DNARNA hybrids have found the ribo sugar pucker to be quite typical of A-type or 3′-endo conformation (37-39) (vide infra). If the sugar pucker for A4 were close to C3′-endo, the signal corresponding to the H1′-H2′ cross-peak (Figure 6) would have been absent. The appearance of this peak, albeit weaker than the others, suggests that A4 has a conformation significantly different from C3′-endo. Molecular modeling (vide infra) is in accord with this conclusion, showing an envelope conformation which has more C2′-endo than C3′-endo character. The information available from the NOESY spectra also addresses the issue of possible intercalation of the benzene ring. If intercalation were to occur, as was recently shown to be the case for the anthraquinon-2ylmethyl derivative (27), cross-peaks would be observed between the benzene aromatic protons and the H1′ protons of A4 and C5. Furthermore, cross-peaks should appear between the benzene aromatic protons and the aromatic protons of A4 and C5. Neither was observed in the NOESY spectrum. Also, because intercalation would increase the distance between A4 and C5, the loss of cross-peaks between H1′ of A4 and H6 of C5 would be observed. However, that cross-peak was clearly visible in the aromatic H1′ region of the NOESY spectrum (Figure 5). Further evidence regarding the conformational preference of the phenethyl group may be derived from the aromatic H2′ region of the NOESY spectrum (Figure 7). The arrow in Figure 7 points toward two peaks. The upper cross-peak shows the through-space connectivity between benzylic protons of the phenethyl group and H6 of C5. The lower peak reveals connectivity between benzylic protons and the aromatic protons of the benzene ring. Molecular modeling suggests these NOE conditions can be satisfied only when the benzene ring lies outside the helix. The DQF-COSY spectrum confirms the location of the benzene ring as being inside the major groove. Figure 8A shows the cross-peaks between the H5 and H6 protons of the cytidines of the standard 12-mer. The


Deshmukh and Broom

Figure 6. Expanded contour plot (H1′-H2′ region) of the DQF-COSY spectrum for the phenethyl 12-mer. The arrow indicates the H1′-H2′ cross-peak for A4.

Figure 7. Expanded contour plot (aromatic H2′ region) of the NOESY spectrum for the phenethyl 12-mer. The arrow indicates connectivities of the benzylic proton.

cross-peaks are uniform in intensity, which is expected since the distance between H5 and H6 is invariant. Differences in the chemical shifts arise from differential shielding and can be predicted from the location of cytidines within the duplex. Figure 8B reveals that the signal for C5 is substantially shielded relative to that in the standard oligomer and, most importantly, is of greatly reduced intensity. This may be accounted for by locating the benzene ring in the

major groove in such a way that it is perpendicular to the planes of the stacked bases and places H5 of C5 within its shielding cone. This results in an upfield chemical shift, and the motion of the benzene ring causes C5 to experience a variable magnetic field, such that the H5 resonance is considerably broadened. This broadening results in the apparent weak intensity of the H5H6 cross-peak. The stacked plot clearly reveals this broadening relative to the other cytidine peaks in the set



Figure 8. Expanded contour plots (H5-H6 region) of the DQF-COSY spectrum (A) for the standard 12-mer and (B) for the phenethyl 12-mer.

(Figure 9). The molecular model shown as a stereoplot in Figure 10 in which the arrow points at the benzene ring is fully consistent with these data. The NMR studies on the chimeric DNA-RNA hybrid structures 7 and 8 were performed in much the same way as described above for the DNA oligomers. However, it has been demonstrated that the increased relaxation time of RNA-type protons may give spurious results if mixing times in NOESY experiments are too short. All NOESY experiments on the hybrids, therefore, were performed using three mixing times: 250, 400, and 600 ms. The NOESY experiments with 400 ms mixing times showed all the required connectivities, and no additional connectivities were observed in the 600 ms experiments. All discussions presented here will be taken from NOESY experiments performed with a 400 ms mixing time.

The expanded contour plot on the NOESY spectrum for the standard hybrid 7 is shown in Figure 11 in the aromatic H1′ region. The presence of ribonucleotides has spread the aromatic H1′ region over a large chemical shift range. Peak intensities are not uniform, and the assignment of all peaks reveals that peaks corresponding to RNA connectivities are considerably weaker than those for DNA connectivities, as expected. The chemical shifts for DNA and RNA H1′ differ significantly, and one can divide the aromatic H1′ region at 5.69 ppm. All ribonucleotide resonances are upfield from 5.69 ppm, whereas all DNA H1′, except C5-H1′, are downfield. Evaluation of the standard 12-mer hybrid 7 reveals that the structure generally resembles the modified A-form geometry described by Reid and his colleagues (37). The A-form character of the RNA component is


Figure 9. Stacked plot (H5-H6 region) of the DQF-COSY spectrum for the phenethyl 12-mer. The arrow indicates H5 of C5.

confirmed by the presence of strong cross-peaks from aromatic H8/H6 to H2′ of the (N1) sugar and the lack of similar intranucleoside cross-peaks (data not shown). As

Deshmukh and Broom

noted by asterisks in Figure 11, the H2 protons of A4 and A6 demonstrate connectivity with H1′ of G10 and G8, respectively. Such intrastrand NOEs are not associated with B-form DNA but arise with narrowing of the minor groove (40). H2 of A6 and A4 also show weak connectivities with their own H1′, another observation which is not found in B-DNA geometry. On the other hand, H2 of A6 shows clear connectivity with H1′ of U7, a finding more typical of DNA duplex geometry. This alteration of the A-type pattern at the junction may be responsible for the modest destabilization of the hybrid noted by melting studies as compared to standard DNA. The expanded contour plot for the H1′-H2′ region of the DQF-COSY spectrum for the standard hybrid is shown in Figure 12. In general, the C3′-endo sugar pucker of RNA strands results in very small J1′,2′ values, and in fact, those cross-peaks are missing in this spectrum, suggesting that most of the ribo sugars assume the expected C3′-endo sugar pucker. The exceptions are G12, the terminal nucleoside in which rapid equilibration of conformers gives rise to strong 1′,2′-couplings and, in-

Figure 10. Stereoview of the model for the phenethyl 12-mer showing proximity of the benzene ring to C5.

Figure 11. Expanded contour plot (aromatic H1′ region) of the NOESY spectrum for the standard hybrid.



Figure 12. Expanded contour plot (H1′-H2′ region) of the DQF-COSY spectrum for the standard hybrid.

Figure 13. Expanded contour plot (aromatic H1′ region) of the NOESY spectrum for the phenethyl hybrid.

terestingly, U7 which apparently assumes much more of a C2′-endo conformation, presumably because of its position at the junction. The expanded contour plot for the aromatic H1′ region of the NOESY spectrum for phenethyl hybrid 8 is shown in Figure 13. The assignments are illustrated on the figure. Data are entirely consistent with the conformation observed for the standard hybrid. Thus, the A-H2 protons show the intrastrand connectivities as marked by asterisks, and H2 of A6 also shows the previously identified intrastrand connectivity with H1′ of U7. The close similarity between 7 and 8 conformations is also

apparent from the DQF-COSY spectrum of the 1′,2′region shown in Figure 14, leaving the final issue to be resolved as the location of the benzene ring from the phenethyl nucleoside. This was determined using NOESY and DQF-COSY experiments, as described above, for the phenethyl DNA 6. In Figure 15, the aromatic H2′ cross-peaks may be seen. The arrow points to the peaks corresponding to connectivity between the benzylic proton of the phenethyl group and H6 of C5 (upper signal) and between the benzylic protons and the aromatic protons of the phenethyl group (lower signal).


Deshmukh and Broom

Figure 14. Expanded contour plot (H1′-H2′ region) of the DQF-COSY spectrum for the phenethyl hybrid.

Figure 15. Expanded contour plot (aromatic H2′ region) of the NOESY spectrum for the phenethyl hybrid. The arrow indicates connectivities of the benzylic proton.

Finally, in Figure 16, comparison of the data for the standard hybrid with that for the phenethyl hybrid reveals, just as seen previously, a marked reduction in signal intensity for the H5-H6 coupling of C5, again consistent with the benzene ring residing in the major groove and exerting an anisotropic effect on the H6 proton of C5. The only significant difference noted between substitution of 2′-O-phenethyladenosine into DNA as opposed to

the DNA-RNA hybrid lies in the observation of an H1′H2′ cross-peak for A4 in the former but not in the latter. While one might normally consider the lack of coupling in the latter case to be associated with C3′-endo sugar geometry, a pure C3′-endo is not consistent with the demonstration that the benzene ring is introduced into the major groove; rather, such a conformation would place the benzene ring firmly in the minor groove. Therefore, one must assume that the conformations



Figure 16. Expanded contour plot (H5-H6 region) of the DQF-COSY spectrum for the standard hybrid (A) and the phenethyl hybrid (B).

available for 2′-O-phenethyladenosine in the hybrid 4 lie between O4′-exo and C2′-exo, or in the “northwest” quadrant of the pseudorotational cycle (41).

Center Grant IP30 CA42014. The authors are very grateful to Darrell R. Davis for many helpful discussions. LITERATURE CITED

DISCUSSION

As noted in the Introduction, the introduction of a small, flexible hydrophobic group into the 2′-position of a ribonucleoside and inclusion of that nucleoside into an oligodeoxynucleotide or a chimeric DNA-RNA hybrid could, in principle, give rise to four locations of the benzene ring in the complex: (1) it could extend into the aqueous solution, (2) it could lie in the minor groove, (3) it could intercalate, or (4) it could reside in the major groove. Although initially we regarded the last option as the least likely, it turns out that in both oligonucleotide duplexes this is precisely what takes place. Although there are slight differences in the conformation of the 2′O-phenethylribose in the two circumstances as reflected by the existence of weak H1′-H2′ coupling in the DNA and lack of such coupling in the hybrid, it remains a fact that the benzene ring must reside in the major groove because of its profound impact on the H6 proton of C5. It is interesting that the substitution of the phenethyloxy group for the proton in DNA, although it leads to some destabilization of the duplex, does not create gross distortions in the oligonucleotide backbones. That is, the NMR data are fully consistent with B-type geometry in the DNA oligomers and a modified A-type geometry, as previously described (37-40), in the DNA-RNA duplex. This markedly unusual introduction of an aromatic substituent into the major groove has obvious implications in the design of oligonucleotides containing DNA major groove binding molecules. This area is under intense evaluation in this laboratory at the present time. ACKNOWLEDGMENT

These studies were supported in part by NIH Grant RO1 AI27692 and by a research fellowship (H.M.D.) from the University of Utah Research Committee. Facilities support (mass spectrometry, NMR, and DNA synthesis) was partially provided through NIH Instrumentation Grant 1S1 ORRO6262 for partial purchase of the Varian Unity 500 MHz NMR spectrometer and through Cancer

(1) Uhlmann, E., and Peyman, A. (1990) Antisense Oligonucleotides: A New Therapeutic Principle. Chem. Rev. 90, 543-584. (2) Beaucage, S. L., and Iyer, R. P. (1993) The Synthesis of Modified Oligonucleotides by the Phosphoramidite Approach and Their Applications. Tetrahedron 49, 6123-6194. (3) Beaucage, S. L., and Iyer, R. P. (1993) The Synthesis of Specific Ribonucleotides and Unrelated Phosphorylated Biomolecules by the Phosphoramidite Method. Tetrahedron 49, 10441-10488. (4) Tonkinson, J. L., and Stein, C. A. (1993) Antisense Nucleic AcidssProspects for Antiviral Intervention. Antiviral Chem. Chemother. 4, 193-200. (5) Stein, C. A. and Cheng, Y.-C. (1993) Antisense Oligonucleotides as Therapeutic AgentssIs the Bullet Really Magical? Science 261, 1004-1012. (6) Beaucage, S. L., and Iyer, R. P. (1992) Advances in the Synthesis of Oligonucleotides by the Phosphoramidite Approach. Tetrahedron 48, 2223-2311. (7) Milligan, J. F., Matteucci, M. D., and Martin, J. C. (1993) Current Concepts in Antisense Drug Design. J. Med. Chem. 36, 1923-1937. (8) Cheng, Y.-K., and Pettitt, B. M. (1992) Hoogsteen versus Reversed-Hoogsteen Base Pairing: DNA Triple Helices. J. Am. Chem. Soc. 114, 4465-4474. (9) Thuong, N. T., and He´le`ne, C. (1993) Sequence-Specific Recognition and Modification of Double-Helical DNA by Oligonucleotides. Angew. Chem., Int. Ed. Engl. 32, 666-690. (10) Hobbs, C. A., and Yoon, K. (1994) Differential Regulation of Gene Expression in vivo by Triple Helix-Forming Oligonucleotides as Detected by a Reporter Enzyme. Antisense Res. Dev. 4, 1-8. (11) Clarenc, J.-P.; Lebleu, B., and Leónetti, J.-P. (1994) Base Changes and Triple-Helix Hybridization Properties of GT containing Third Strands: A Systematic Study. Nucleosides Nucleotides 13, 799-809. (12) Sun, J.-S., Franç ois, J.-C., Montenay-Garestier, T., SaisonBehmoaras, T., Roig, V., Thuong, N. T., and He´le`ne, C. (1989) Sequence-Specific Intercalating Agents: Intercalation at Specific Sequences on Duplex DNA via Major Groove Recognition by Oligonucleotide-Intercalator Conjugates. Proc. Natl. Acad. Sci. U.S.A. 86, 9198-9202.

120 Bioconjugate Chem., Vol. 7, No. 1, 1996 (13) Mann, J. S., Shibata, Y., and Meehan, T. (1992) Synthesis and Properties of an Oligodeoxynucleotide Modified with a Pyrene Derivative at the 5′-Phosphate. Bioconjugate Chem. 3, 554-558. (14) Vlassov, V. V., Gaidamakov, S. A., Zarytova, V. F., Knorre, D. G., Levina, A. S., Nikonova, A. A., Podust, L. M., and Fedorova, O. S. (1988) Sequence-Specific Chemical Modification of Double-Stranded DNA with Alkylating Oligodeoxyribonucleotide Derivatives. Gene 72, 313-322. (15) Boidot-Forget, M., Chassignol, M., Takasugi, M., Thuong, N. T., and He´le`ne, C. (1988). Site-Specific Cleavage of SingleStranded and Double-Stranded DNA Sequences by Oligodeoxyribonucleotides Covalently Linked to an Intercalating Agent and an EDTA-Fe Chelate. Gene 72, 361-371. (16) Lin, S.-B., Blake, K. R., Miller, P. S., and Ts’o, P. O. P. (1989) Use of EDTA Derivatization To Characterize Interactions between Oligodeoxyribonucleoside Methylphosphonates and Nucleic Acids. Biochemistry 28, 1054-1061. (17) Iverson, B. L., and Dervan, P. B. (1987) Nonenzymatic Sequence-Specific Cleavage of Single-Stranded DNA to Nucleotide Resolution. DNA Methyl Thioether Probes. J. Am. Chem. Soc. 109, 1241-1243. (18) Levis, J. T., and Miller, P. S. (1994) Properties of Exonuclease-Resistant, Psoralen-Conjugated Oligodeoxyribonucleotides in vitro and in Cell Culture. Antisense Res. Dev. 4, 231-241. (19) Godard, G., Franç ois, J.-C., Duroux, I., Asseline, U., Chassignol, M., Thuong, N., He´le`ne, C., and Saison-Behmoaras, T. (1994) Photochemically and Chemically Activatable Antisense Oligonucleotides: Comparison of Their Reactivities towards DNA and RNA Targets. Nucleic Acids Res. 22, 4789-4795. (20) Morvan, F., Porumb, H., Degols, G., Lefebvre, I., Pompon, A., Sproat, B. S., Rayner, B., Malvy, C., Lebleu, B., and Imbach, J.-L. (1993) Comparative Evaluation of Seven Oligonucleotide Analogues as Potential Antisense Agents. J. Med. Chem. 36, 280-287. (21) Lesnik, E. A., Guinosso, C. J., Kawasaki, A. M., Sasmor, H., Zounes, M., Cummins, L. L., Ecker, D. J., Cook, P. D., and Freier, S. M. (1993) Oligodeoxynucleotides Containing 2′-O-Modified Adenosine: Synthesis and Effects on Stability of DNA:RNA Duplexes. Biochemistry 32, 7832-7838. (22) Lamond, A. I., and Sproat, B. S (1993) Antisense Oligonucleotides Made of 2′-O-AlkylRNA: Their Properties and Applications in RNA Biochemistry. FEBS Lett. 325, 123127. (23) Yamana, K., Nishijima, Y., Ikeda, T., Gokota, T., Ozaki, H., Nakano, H., Sangen, O., and Shimidzu, T. (1990) Synthesis and Interactive Properties of an Oligonucleotide with Anthraquinone at the Sugar Fragment. Bioconjugate Chem. 1, 319-324. (24) Keller, T. H., and Ha¨ner, R. (1993) Synthesis and Hybridization Properties of Oligonucleotides Containing 2′-O-Modified Ribonucleotides. Nucleic Acids Res. 21, 4499-4505. (25) Dan, A., Yoshimura, Y., Ono, A., and Matsuda, A. (1993) Nucleosides & Nucleotides. 118. Synthesis of Oligonucleotides Containing a Novel 2′-Deoxyuridine Analogue that Carries an Aminoalkyl Tether at 1′-Position; Stabilization of Duplex Formation by an Intercalating Group Accommodated in the Minor Groove. Bioorg. Med. Chem. Lett. 3, 615-618. (26) Yamana, K., Gokota, T., Ozaki, H., Nakano, H., Sangen, O., and Shimidzu, T. (1992) Enhanced Fluorescence in the

Deshmukh and Broom Binding of Oligonucleotides with a Pyrene Group in the Sugar Fragment to Complementary Polynucleotides. Nucleosides Nucleotides 11, 383-390. (27) Deshmukh, H. M., Joglekar, S. P., and Broom, A. D. (1995) Self-Complementary Oligodeoxyribonucleotides Containing 2′-O-(Anthraquinone-2-methyl)adenosine. Bioconjugate Chem. 6, 578-586. (28) Ti, G. S., Gaffney, B. L., and Jones, R. A. (1982) Transient Protection: Efficient One-Flask Syntheses of Protected Deoxynucleosides. J. Am. Chem. Soc. 104, 1316-1319. (29) Wu, T., Ogilvie, K. K., and Pon, R. T. (1989) Prevention of Chain Cleavage in the Chemical Synthesis of 2′-Silylated Oligoribonucleotides. Nucleic Acids Res. 17, 3501-3517. (30) Kierzek, R., Caruthers, M. H., Longfellow, C. E., Swinton, D., Turner, D. H., and Freier, S. M. (1986) Polymer-Supported RNA Synthesis and Its Application To Test the NearestNeighbor Model for Duplex Stability. Biochemistry 25, 78407846. (31) Sproat, B. S., Beijer, B., and Iribarren, A. (1990) New Synthetic Routes to Protected Purine 2′-O-Methylriboside3′-O-phosphoramidites Using a Novel Alkylation Procedure. Nucleic Acids Res. 18, 41-49. (32) Wagner, D., Verheyden, J. P. H., and Moffatt, J. G. (1974) Preparation and Synthetic Utility of Some Organotin Derivatives of Nucleosides. J. Org. Chem. 39, 24-30. (33) Yano, J., Kan, L. S., and Ts’o, P. O. P. (1980) A Simple Method of the Preparation of 2′-O-Methyladenosine. Methylation of Adenosine with Methyl Iodide in Anhydrous Alkaline Medium. Biochim. Biophys. Acta 629, 178-183. (34) Gray, D. M., Ratliff, R. L., and Vaughan, M. R. (1992) Circular Dichroism Spectroscopy of DNA. Methods Enzymol. 211, 389-405. (35) Tonelli, A. E., and Patel, D. J. (1974) Proton Nuclear Magnetic Resonance Investigations and Ring Current Calculations of Guanine N-1 and Thymine N-3 Hydrogen-Bonded Protons in Double-Helical Deoxyribonucleotides in Aqueous Solution. Proc. Natl. Acad. Sci. U.S.A. 71, 1945-1948. (36) Reid, B. R. (1987) Sequence-Specific Assignments and Their Use in NMR Studies of DNA Structure. Q. Rev. Biophys. 20, 1-34. (37) Salazar M., Fedoroff, O. Y., Miller, J. M., Ribeiro, N. S., and Reid, B. R. (1993). The DNA Strand in DNA‚RNA Hybrid Duplexes is Neither B-Form nor A-Form in Solution. Biochemistry 32, 4207-4215. (38) Wang, A. H. J., Fujii, S., van Boom, J. H., van der Marel, G. A., van Boeckel, S. A. A., and Rich, A. (1982) Molecular Structure of r(GCG)d(TATACGC):DNA-RNA Hybrid Helix Joined to Double Helical DNA. Nature 299, 601-604. (39) Happ, C. S., Happ, E., Clore, G. M., and Gronenborn, A. M. (1988) Refinement of the Structure of RNA-DNA Hybrid 5′-[r(GCA)d(TGC)]2. FEBS Lett. 236, 62-70. (40) Fedoroff, O. Y., Salazar, M., and Reid, B. R. (1993) Structure of a DNA:RNA Hybrid Duplex. Why RNase H Does Not Cleave Pure RNA. J. Mol. Biol. 233, 509-523. (41) Hosur, R. V., Ravikumar, M., Chary, K. V., Sheth, A., Govil, G., Zu-Kun, T., and Miles, H. T. (1986) Solution Structure of d-GAATTCGAATTC by 2D-NMR. A New Approach to Determination of Sugar Geometries in DNA Segments. FEBS Lett. 205, 71-76.

BC9500812

Probing Nucleic Acid Geometries: Oligonucleotides Containing 2'-O

Recommend Documents