Regiospecific Solid-Phase Synthesis of Branched Oligonucleotides

Symbols: ribose (r); deoxyribose (d); base stacking (▫); hydrogen bonds (- - -). ..... Ruskin, B., Krainer, A. R., Maniatis, T., and Green, M. R. (1...
0 downloads 0 Views 178KB Size
Download PDF
Bioconjugate Chem. 1997, 8, 370−377

370

Regiospecific Solid-Phase Synthesis of Branched Oligonucleotides. Effect of Vicinal 2′,5′- (or 2′,3′-) and 3′,5′-Phosphodiester Linkages on the Formation of Hairpin DNA Ravinderjit S. Braich and Masad J. Damha* Department of Chemistry, Otto Maass Chemistry Building, McGill University, 801 Sherbrooke Street West, Montreal, Quebec, Canada H3A 2K6. Received January 13, 1997X

A general procedure for the solid-phase regiospecific synthesis of branched oligonucleotides (bNA) analogues using readily available phosphoramidite reagents has been developed. The key feature of this method is use of the solid-phase phosphoramidite procedure to assemble linear oligonucleotide sequences and sequential removal of the phosphate (β-cyanoethyl or methyl) and silyl protecting groups without detaching the nascent oligonucleotide from the solid support. Conversion of the phosphate backbone into the more stable phosphodiester linkages allows for removal of the 2′-O-tert-butyldimethylsilyl protecting group without cleavage or isomerization at the branch point. This method allows for the formation of branched oligonucleotides with sequences of arbitrary base composition, length, and orientation around the branch point junction, including a “Y”-shaped octadecamer d(TACTA)-rA[2′,5′d(GTATGT)]3′,5′d(CAAGTT). Studies to explore structural effects in the use of a branched adenosine as replacement for nucleotide loops in duplex and triplex DNA are also described. Branched oligonucleotides of the type rA[2′,5′dCndA10-5′]3′,5′dCndT10-3′ and rA[2′,5′dCn3′,3′dA10-5′]3′,5′dCnT103′ form hairpin duplexes with thermal stability comparable to or better than that of one with a natural deoxynucleotide loop.

INTRODUCTION

Branched nucleic acids (bNAs) have been the subject of a number of scientific studies since the discovery of RNA “forks” and “lariats”(1-4) and multicopy singlestranded DNA (msDNA) in prokaryotic and eukaryotic cells (5, 6). msDNA is a chimera of DNA and RNA found in some prokaryotes having a branched or forked structure wherein the 5′-nucleotide of a single-stranded DNA chain is esterified to the 2′-hydroxyl of an internal residue of the RNA chain. The novelty of these structures has raised interest as to their possible role in regulating RNA splicing and debranching. In addition, there has been increasing recent interest in synthetic branched DNA and RNA for use in diagnostic applications (7), as “molecular anchors” for inducing the formation of novel triple-helical DNA (8-13), and as tools for studying branched RNA/ RNA complexes (9) and the substrate specificity of debranching enzymes (14, 15). We have been describing in detail a methodology for the solid-phase synthesis of open-chain forked or “Y” RNA and DNAs such as a branched 18-mer UACUAA(2′GUAUGU)3′-GUAUGU, in which the branchpoint adenosine nucleotide (A) is linked to identical GUAUGU “tails” via vicinal 2′,5′- and 3′,5′-phosphodiester bonds (16). The branched “V” 21-mer rA(2′dT10)3′dT10 (10) and a “dendritic” 87-mer structure have also been synthesized from this laboratory (17, 18). Our synthetic “convergent” strategy is based on our discovery that nucleosides 2′,3′O-bisphosphoramidite synthons react with adjacent solidsupport bound chains, yielding symmetric “V”-like molecules. Synthesis is then continued in the 3′-to-5′ direction from the apex of the “V” to yield forked or “Y” structures. More recently, we developed a convergent method that * Author to whom correspondence should be addressed [telephone (514) 398-7552; fax (514) 398-3797; e-mail damha@ omc.lan.mcgill.ca]. X Abstract published in Advance ACS Abstracts, April 15, 1997.

S1043-1802(97)00030-X CCC: $14.00

generates a mixture of bRNA’s with 2′- and 3′-chains of different base composition, e.g., A(2′-GUAUGU)3′-CAAGUU (19). Although this is the easiest way to prepare an array of branched oligoribonucleotides having different 2′- and 3′-chain sequences (requiring commercially available building blocks), this method necessitates separation of very similar branched molecules. It is therefore only practical for “combinatorial” or biochemical investigations in which the amount of material needed is very small. All other synthetic strategies reported so far for the regiospecific assembly of branched oligonucleotides mimicking the natural lariat structures are based upon solution-phase phosphotriester methods, except for the recent and elegant work by Sproat and co-workers, who use a “divergent” solid-phase phosphoramidite strategy (20). This strategy has permitted the synthesis of medium size branched oligoribonucleotides; however, it requires the use of elaborate phosphoramidites and branchpoint synthons, which limits the ease and speed of a potential synthesis. Given this limitation, and those of our current methods, we directed our attention toward alternative strategies for the regiospecific synthesis of branched oligonucleotides. As a starting point, we have chosen to regiospecifically synthesize branched (“Y”shaped) oligonucleotides consisting of three DNA chains joined to the 2′-, 3′-, and 5′-positions of an adenosine branchpoint. The synthesis of branched oligonucleotides in which 2′-deoxyribose is substituted for ribose sugars would be useful for studying the substrate specificity of debranching enzymes and for evaluating the role of 2′OH groups in enzyme-bRNA interactions. Furthermore, the conformational rigidity imparted to the DNA chains, by the branchpoint ribose, could be exploited to preorganize and induce the formation of duplex and triplex DNA. Here we describe a facile method for the regiospecific synthesis of branched oligodeoxynucleotides. Some sequences were synthesized only on the basis of mimicking the structure of naturally occurring branched introns © 1997 American Chemical Society

Synthesis and Association of Branched Oligonucleotides Table 1. Branched Sequences Synthesized for This Study

(e.g., 1 and 2, Table 1). Others were prepared to explore structural effects in the use of a branched adenosine as a replacement for nucleotide loops in a hairpin (duplex) DNA (6-9). EXPERIMENTAL PROCEDURES

Reversed-phase C18 Sep-Pak cartridges were obtained from Waters (Milford, MA). Polyacrylamide gel electrophoresis (PAGE) reagents were purchased from Bio-Rad (Toronto). Fused silica capillaries, for capillary electrophoresis (CE), were obtained from Polymicro Technologies (Phoenix, AZ). Long-chain alkylamine controlled pore glass (CPG) bearing deoxyribonucleosides were prepared using our protocols (21). Tetra-n-butylammonium fluoride (1 M) in tetrahydrofuran (THF; desilylating reagent) and (methacryloxypropyl)trimethoxysilane were purchased from Aldrich. Enzymes were purchased from Boehringer Mannheim (Quebec). Incubation buffers for enzyme digestions were prepared using sterilized deionized water, filtered through a sterile 0.2 µm-pore filter and stored at -20 °C prior to use. Snake-venom phosphodiesterase (SVPDE)/alkaline phosphatase (AP) incubations were performed in 50 mM Tris-HCl/10 mM MgCl2, pH 8, while Nuclease P1/AP incubations were performed in 0.1 M Tris-HCl/1 mM ZnCl2, pH 7.2. Synthesis of Branched Oligonucleotides. Branched oligomers, the structures of which are illustrated in Table 1, were synthesized on an Applied Biosystems DNA synthesizer (Model 381A) via phosphoramidite chemistry.

Bioconjugate Chem., Vol. 8, No. 3, 1997 371

The procedure can be illustrated by describing the synthesis of 1. The linear sequence 3′,5′-linked 5′-HOd(TACTA)-rA-d(CAAGTT)-3′ was synthesized by the normal synthesis cycle (22), and the free 5′-hydroxyl group was capped by running the capping (Ac2O) cycle. The synthesis column was removed from the synthesizer and the solid support dried by flushing argon through the column. All phosphate protecting groups were removed by treating the support with a solution of NEt3/ dioxane/thiophenol (10 mL, 2:2:1 v/v/v) (when rA methylamidite was used as branching synthon) or NEt3/ CH3CN (10 mL, 4:6 v/v) (when rA β-cyanoethylamidite was used as branching synthon). This step is done via a syringe by pushing the deprotection solution through the column over a 90 min period (room temperature). The solid support was then washed extensively with EtOH (30 mL, for methyl protection only) and CH3CN (30 mL), followed by THF (30 mL). Removal of the 2′-silyl protecting group was carried out by pushing a solution of 1 M TBAF/THF (1 mL) through the synthesis column over a period of 10 min. After the removal of silyl protecting group, the support was washed with THF (50 mL), followed by CH3CN (50 mL). The column was then reinstalled on the synthesizer and synthesis continued by the normal cycle for the addition of 5′-phosphoramidites except that the first addition required the use of more concentrated phosphoramidite solution (0.3 M) and longer coupling times (30 min) (see Table 2). Average coupling yields (trityl assay method) for 5′-phosphoramidites couplings were in the range of 95-98%. CPG-bound oligomers were deprotected under standard conditions (concentrated NH4OH/ethanol; 4:1, room temperature, 48 h). Following deprotection, the branched oligomers (17) were purified by polyacrylamide gel electrophoresis (PAGE) and desalted by reversed-phase chromatography on C18 Sep-Pak cartridges (22). Alternatively, 5′-tritylated oligonucleotides can be purified by reversed-phase chromatography on OPC cartridges (as in the case of 8 and 9) using supplier’s specification (Dalton Chemical Co., Toronto). Analysis and Characterization of Oligonucleotides. For enzymatic digestions, 0.5 A260 unit of the purified lyophilized branched oligomer 1 was dissolved in the appropriate buffer (20 µL), and to this was added either SVPDE (1 µL, 0.002 U) and AP (1 µL, 9 U) or nuclease P1 (3 µL, 0.9 U) and AP (1 µL, 9 U). After incubation was complete (37 °C, 24 h), the samples were lyophilized to dryness, redissolved in 15 µL of sterilized water, and analyzed by HPLC as described below. HPLC analysis of enzymatic digests was carried out on a Waters Max 480 system (Millipore) equipped with dual 501 pumps, UK6 injector, and a 480 tunable UV detector, with the gradient being controlled by a 600E gradient controller and solvent delivery system. Analyses were conducted at 254 nm under the following conditions: reversed-phase Whatman Partisil ODS-2 column (10 µm, 4.6 × 250 mm, Chromatographic Specialties); mobile phase solvent A, 20 mM KH2PO4 (pH 5.5); solvent B, methanol, gradient 0-50% solvent B in 25 min. Peak areas and the previously reported extinction coefficient values at 254 nm (23) were used to calculate relative concentrations of monomers. Capillary electrophoresis analyses were carried out on a CE system constructed at McGill (24). Capillary, 75 µm i.d., contained 9% (w/v) acrylamide and had a total length of 55 cm and separation length of 35 cm. Samples were injected electrokinetically by applying a voltage of 9 kV for 30 s to 3 min depending on sample concentration. Melting Experiments. Melting experiments were carried out in 50 mM magnesium chloride/10 mM Tris-

372 Bioconjugate Chem., Vol. 8, No. 3, 1997

Braich and Damha

Scheme 1. Regiospecific, Divergent Synthesis of Branched Oligonucleotides

HCl, pH 7.3. The solution was loaded into a cuvette, and the absorbance versus temperature profile was recorded using a Cary 3E UV-vis spectrophotometer fitted with a thermostated cell block and temperature controller. The cell block was continuously purged with dry nitrogen to prevent moisture condensation at low temperatures. Solutions containing the hairpin duplexes (1.1-110 µM range) or triplex 6/dT10 (2.2 µM) were transferred to the cuvette and heated to 90 °C for 5 min and then cooled to 5.0 °C for 1-12 h. The solutions were heated at a rate

of 0.5 °C/min, and the absorbance at 260 nm was measured. The melting curve was then obtained as described above. Tm values were calculated from the first derivative of the melting curve. Precision in Tm values, determined from variance of repeated experiments, is no greater than (0.5 °C. Molecular Modeling. These studies were performed in the AMBER force field developed for nucleic acids (Hypercube’s HyperChem 3.0). The cutoff function (switched: inner ) 10 Å and outer ) 14 Å) and the RMS

Bioconjugate Chem., Vol. 8, No. 3, 1997 373

Synthesis and Association of Branched Oligonucleotides Table 2. Conditions and Yields of Branched Oligonucleotide Synthesis a c scale % coupling yield crudeb purification (A260 U) d sequence (µmol) rA 2′5′pX next A260 U loaded recovered

1 2e 3 3 4 5 6 7 8 9

1.0 0.6 1.0 1.0 1.0 1.0 1.0 1.0 0.1 0.1

95 61e 100 97g 87 97 98 89 97 97

91 60f 97 99 115 89 81 91 124 111

99 78 101 111 86 93 83 92 102 135

87 29 131 64 45 85 43 123 10 9

25 20 25 25 20 25 25 20 8 7

7.4 3.1 2.8 6.2 2.1 6.0 2.6 3.1 1.4 1.9

a Coupling yield of the 2′,5′-pX residue was calculated relative to coupling of branchpoint A. b Total oligomer recovered after cleavage and deprotection. c PAGE followed by desalting with C18 Sep-Pak cartridges. Sequences 8 and 9 were purified by the “tritylon” reversed-phase procedure (OPC purification matrix). d Unless otherwise indicated, rABz-5′-O-DMT-2′-O-silyl-3′-O-cyanoethylphosphoramidite was used as the branching synthon. e rABz-5′-O-DMT3′-O-silyl-2′-O-cyanoethylphosphoramidite. f Lower coupling reflects the inadvertent use of lower concentration (0.1 M) of the phosphoramidite solution after the branchpoint. g rABz-5′-O-DMT2′-O-silyl-3′-O-methylphosphoramidite used.

Figure 2. Capillary gel electrophoresis of branched oligomer 3: (a) crude sample obtained from the convergent method (16); (b) crude sample obtained by the present divergent method. See Experimental Procedures for conditions.

Figure 1. Electrophoresis of branched oligonucleotides on a 24% polyacrylamide/7 M urea gel: (A) lane1, oligomer 3 prepared by the present divergent method; lane 2, authentic sample of oligomer 3 prepared by a convergent method (16); lane 3, crude sample of 3 prepared by the present divergent method; lane 4, marker dyes bromophenol blue (fast) and xylene cyanol (slow); (B) lanes 1 and 5, marker dyes; lane 2, branched 18-mer TACTAA(2′-GTATGT)3′-GTATGT (16); lane 3, branched 18-mer 1; lane 4, linear 12-mer TACTArA2′5′GTATGT.

gradient were kept constant at 10-5 kcal/mol Å. The branched pentaloops for 6, 8, and 9 were first minimized individually. These were then capped to A/T duplexes built in the classical B-form. Three preliminary structures were minimized, and short simulations (10-70 ps) were performed in vacuum on unconstrained helices. A stereoview of the most energetically favored structure of 9 is shown in Figure 5. RESULTS AND DISCUSSION

Synthesis of Branched Oligodeoxynucleotides. Our methodology is illustrated in Scheme 1 for the synthesis of branched 18-mer 1. This oligonucleotide has all conserved nucleobases and mimics the naturally occurring sequence found in Staphylococcus cerevisiae lariat introns (Table 1). The commercially available N6benzoyl-5′-dimethoxytrityl-2′-O-tert-butyldimethylsilyladenosine is the key building block in our strategy because

it allows sequential extension of the chain from the branchpoint rA in all three directions. The other synthons required are 2′-deoxynucleoside-3′-O- or 5′-Ocyanoethylphosphoramidite monomers, which are also commercially available. Long-chain alkylamine CPG served as the solid support and was derivatized with the 3′-terminal nucleoside as previously described (21). The synthesis began by assembly of the linear tridecanucleotide 5′-d(TACTA)-rA-d(CAA GTT)-3′ in the normal 3′to-5′ fashion. Following acetylation of the terminal 5′hydroxyl group, the support was treated with triethylamine/acetonitrile solution (4:6 v/v, 1.5 h, room temperature) to affect the removal of all cyanoethyl phosphate protecting groups, thus providing an oligomer with a intrinsically more stable phosphodiester backbone (25). This is necessary because deblocking of a 2′-OH group vicinal to a 3′,5′-phosphotriester linkage under neutral, acidic, or basic conditions has been shown to lead to phosphoryl migration and/or chain cleavage (26-29). An important feature was the use of fluoride ions to effect the removal of the 2′-O-TBDMSi without “dissolving” the solid support (SiO2) and/or detaching the oligomer from its surface. Thus, the silyl group was removed from riboadenosine by treatment of the support with 1 M tetra-n-butylammonium fluoride/THF (10 min, room temperature), followed by washing with THF and acetonitrile. Model experiments with DMT-dT-(LCAA-CPG) showed that extended treatment with fluoride (15, 30, and 60 min) leads to significant cleavage of the nucleoside from the solid support (10%, 25%, and >65%, respectively). At this point, the 2′-chain (5′-GTA TGT-3′) was synthesized in the 5′-to-3′ direction using commercially available 5′-phosphoramidite derivatives (30). To force branching at the sterically hindered 2′-hydroxyl group, both the concentration and the coupling time of the first 5′-amidite (dG) were tripled to 0.3 M and 30 min, respectively (12, 31). Under these conditions coupling proceeded with 91% efficiency. Lower amidite concentrations (0.10 M) resulted in significantly lower coupling yields (ca. 60%, e.g., 2, Table 2). Synthesis under

374 Bioconjugate Chem., Vol. 8, No. 3, 1997

Figure 3. Melting curves of branched and linear hairpin structures in 50 mM MgCl2/10 mM Tris-HCl buffer, pH 7.3: (a) 6, 7, and I; (b) 8, 9, and I. Oligomer concentration was 1.1 µM.

standard conditions was then continued until the desired “Y”-mer 1 was assembled. The successful completion of 1 was followed by the synthesis of eight other “V”- and “Y”-shaped oligonucleotides 2-9 (Table 1). Sequence 2 mimics a lariat structure in which the conserved 2′-guanine base has been replaced by hypoxanthine. Apart from the use of deoxyinosine as the 2′-residue, the other major difference in the preparation of 1 and that of 2 was the use of adenosine 2′-O-cyanoethylphosphoramidite instead of the regioisomeric 3′-O-cyanoethylphosphoramidite. Thus, 5′d(TACTA)-rA2′,5′d(ITATGT)-3′ was first assembled in the

Braich and Damha

conventional 3′-to-5′ direction and capped (Ac2O). This was followed by removal of the cyanoethyl and 3′-silyl groups and assembly of the 3′-branch, namely d(CAA GTT). All other aspects of branch assembly, deprotection procedures, and handling remained invariant. Under the conditions used (0.5 M tetrazole, room temperature, 30 min), the branch rA-3′-amidite isomer coupled (95%) with greater efficiency than the 2′-amidite isomer (61%). However, if 5-ethylthio-1H-tetrazole is used instead of tetrazole as the acid catalyst, both 2’ and 3’-amidite couple with similar (97-99%) efficiency (Wasner and Damha, unpublished results). For the preparation of sequence 3, both rA-3’-O-methyl and cyanoethyl protected phosphoramidites were used as the branching synthon (Table 2). In the case of methyl protection, a thiophenoxide step was used to cleave the methyl phosphate protecting group at the branchpoint rA prior to 2′desilylation and 2′-chain assembly. Presumably, the thiophenoxide step also removes the cyanoethyl groups attached to the deoxynucleotide residues. Coupling yields at various stages of these syntheses (as determined by the trityl assay method) as well as isolated yields of oligomers are given in Table 2. The branched oligonucleotides were removed from the support and deblocked by treatment with 15 M aqueous ammonia/ethanol 4:1 v/v (room temperature, 48 h). Removal of the ammoniacal solution furnished the crude products. Initial characterization and purification of oligomers was accomplished by electrophoresis by comparison of mobilities to linear oligomers and to authentic samples of branched oligonucleotides. For example, a sample of 3 produced via the convergent approach (10, 16) was the same as a sample produced by the present method, as shown by PAGE and CE (Figures 1A and 2) and by thermal dissociation of antiparallel T/A:T triplexes formed with dA10 (10). Also, a sample of 1 (18-mer) exhibited similar electrophoretic mobility to a “Y”-18-mer prepared via the convergent approach (Figure 1B). The oligonucleotides were purified by either preparative gel electrophoresis or reversed-phase (trityl-on) chromatography. In the former case, the major band was cut out under UV shadow, extracted with water at 37 °C overnight, and desalted by reversed-phase chromatography (Sep-Pak cartridges). Isolated yields of crude and purified oligonucleotides are reported in Table 2. As a further check on the branched structure and nucleotide composition, a small sample of 1 and 2 was subjected to enzymatic hydrolysis and the resulting products were identified by reversed-phase HPLC. Treatment with snake venom phosphodiesterase (Crotalus durissus) and alkaline phosphatase (AP) (from calf intestine) gave the expected nucleoside composition. Furthermore, treatment with nuclease P1 (Penicillium citrinum)/AP gave the branched core trinucleoside diphosphate A(2′dG)3′dC (32) and nucleosides in the expected ratios (data not shown).

Figure 4. Melting temperatures and structures of branched and linear hairpin DNA (50 mM MgCl2/10 mM Tris-HCl buffer, pH 7.3).

Synthesis and Association of Branched Oligonucleotides

Bioconjugate Chem., Vol. 8, No. 3, 1997 375

Figure 5. Stereoview of an energy-minimized structure of branched hairpin 9. Bases in the pentaloop structure are shown in boldface; only two A:T base pairs (bottom) are shown for clarity.

Figure 6. Schematic representation of the structural features of stacking interactions of the branched pentaloop structure of hairpin 9. Symbols: ribose (r); deoxyribose (d); base stacking (9); hydrogen bonds (- - -).

Figure 7. Melting curves of (a) 6/dT10 and (b) 6 in 50 mM MgCl2/10 mM Tris-HCl buffer, pH 7.3.

Branched Nucleic Acid Hairpins. Compounds 6-9 were designed to test the effect of the branchpoint “linker” in systems potentially capable of folding into a hairpin duplex structure. Compounds 6-9 exhibit single cooperative melting transitions (Tm) that are independent of oligonucleotide concentrations over a 10-100-fold range (50 mM MgCl2/10 mM Tris, pH 7.3) (Figure 3). In addition, their Tm values are >28 °C higher than for double-stranded complexes formed from two independent strands e.g., dA10 + dT10, Tm ) 32 °C. These results

support the view that branched oligomers can form hydrogen-bonded, base-stacked structures reversibly in solution by folding intramolecularly to a hairpin conformation. Molecular modeling performed on hairpins 6-8 suggests that their structure is energetically most favorable when the “loop” is composed of a heptanucleotide (5′. . . AXX-3′,2′rA-3′,5′YYT .. . 3′) and the “stem” has seven A/T base-paired residues. In agreement with the view that 6-8 have similar folded structures, the temperatures of dissociation of these complexes are comparable (58-61 °C, Figure 4). By contrast, the energetically optimized structure of compound 9 (the constitutional isomer of 8) can accommodate an additional A/T base pair, leaving a loop that is only five unpaired bases (5′. . .CC-5′,2′rA-3′,5′CC . . . 3’) that is constrained sterically by the requirements of loop closing. It is noteworthy that the structural features of the dC-5′,2′rA-3′,5′dC residues in this loop are in excellent agreement with the NMR-derived structure of small branched RNA fragments (33, 34), the most significant being the extensive base-base stacking interactions between the central rA and the 2′,5′-dC residue (. . . dC5′T2′rA . . .) (Figures 5 and 6). The remaining three dC residues in 9 also basebase stack in a way that is reminiscent of stable nucleic acid hairpins (35). Such stacking interactions are less important in the minimized structure of 8, and this is due to the structural constraints of the vicinal 2′,3′/3′,5′ linkages. Since, in fact, the Tm value is 6 °C greater for compound 9 than for 8, or an unmodified hairpin with identical stem and loop sequences (I, Figure 4), we conclude that the pentaloop contributes significantly to the stabilization of 9. The favorable stabilization derived from the branched pentaloop structure in 9 is also evident when its thermal dissociation (64.1 °C) is compared to that of a model hairpin with an identical stem but a fourunit (CCCC) loop (60.7 °C), the optimum loop length for DNA hairpins (35). We speculate that extensive base stacking interactions within the branched pentaloop provide, at least in part, the structural basis for the gain of thermal stability observed in 9 relative to I (35). Additional studies on various branched pentaloops containing various sequence combinations will be necessary to test this hypothesis. Triplex Formation. Association of a Branched A:T Hairpin with dT10. We have also conducted preliminary studies on the interaction of branched hairpins with single-stranded DNA, namely, dT10 with hairpin 6. dT10 forms a stable complex with 6 in Tris buffer (pH 7.3, 50 mM MgCl2) as indicated by the presence of two well-separated transitions (Figure 7). The lower temperature transition (21.5 °C) is nearly coincident with the first transition given by the classical dA10:2dT10 triplex (Tm ) 19 °C) and corresponds to the process triplex

376 Bioconjugate Chem., Vol. 8, No. 3, 1997

f dT10 + duplex (36). The lower temperature transition (21.5 °C) is therefore attributed to dissociation of dT10 from the triplex dT10:6, whereas the high-temperature transition (61.5 °C) corresponds to the helix-to-coil transition of duplex 6 (Figure 3). This result demonstrates that a unimolecular branched duplex (hairpin) can interact with a pyrimidine sequence to form a triplex structure. We note that the unique folded structure of branched single-stranded loops may confer desirable resistance against degradative nucleases (37). These characteristics combined may be significant in the development of antisense therapeutic agents that target single-stranded RNA via triple-helix formation (38-40). In summary, an efficient and regiospecific synthesis of “V”- and “Y”-shaped oligonucleotides has been developed. We have also demonstrated that ribonucleoside branchpoints can be used as linkers to stabilize hybridization and to direct folding of oligonucleotide strands into duplex structures. In one example, insertion of a branchpoint ribonucleoside in the loop of a DNA hairpin has increased the stability of the hairpin duplex. A branched hairpin structure was found to bind singlestranded DNA via triplex formation. These properties combined make branched oligonucleotides appealing choices for use as novel probes for studying nucleic acid structure and as sequence selective oligonucleotides that can interact with single-stranded nucleic acids. Branched oligonucleotides of the type described here may also serve as probes for studying RNA processing in the cell. ACKNOWLEDGMENT

We thank Dr. W. C. Purdy and Dr. T. Tadey for assistance in capillary electrophoresis analysis of oligonucleotides and Dr. D. N. Harpp for continued support. This research was supported by grants from NSERC (Collaborative Program) and FCAR (Quebec). LITERATURE CITED (1) Wallace, J. C., and Edmonds, M. (1983) Polyadenylated nuclear RNA contains branches. Proc. Natl. Acad. Sci. U.S.A. 80, 950-954. (2) Domedy, H., Apostol, B., Lin, R.-J., Newman, A., Brody, E., and Abelson, J. (1984) Lariat structures are in vivo intermediates in yeast pre-mRNA splicing. Cell 39, 611-621. (3) Ruskin, B., Krainer, A. R., Maniatis, T., and Green, M. R. (1984) Excision of an intact intron as a novel lariat structure during pre-mRNA splicing in vitro. Cell 38, 317-331. (4) Padgett, R. A., Konarska, M. M., Grabowski, P. J., Hardy, S. F., and Sharp, P. A. (1984) Lariat RNA’s as intermediates and products in the splicing of messanger precursors. Science 225, 898-903. (5) Inouye, S., Furuichi, T., Dhundle, A., and Inouye, M. (1987) Molecular Biology of RNA: New Perspectives (M. Inouye and B. S. Dudock, Eds.) pp 271, Academic Press, San Diego, CA. (6) Yee, T., Furuichi, T., Inouye, S., and Inouye, M. (1984) Multicopy single-stranded DNA isolated from a Gram-negative bacterium, Myxococcus xanthus. Cell 38, 203-209. (7) Urdea, M., Horn, T., Fultz, T., Anderson, M., Runnings, J., Hamren, S., Ahle, D., and Chang, C.-A. (1991) Branched DNA amplification multimers for the sensitive, direct detection of human hepatitis viruses. Nucleic Acids Res. Symp. Ser. 24, 197-200. (8) Hudson, R. H. E., and Damha, M. J. (1993) Association of branched nucleic acids. Nucleic Acids Res. Symp. Ser. 29, 9799. (9) Hudson, R. H. E., Ganeshan, K., and Damha, M. J. (1994) Branched nucleic acids: synthesis and biological applications. Carbohydrate Modifications in Antisense Research (Y. S. Sanghvi and P. D. Cook, Eds.) ACS Symposium Series 580, pp 133-152, American Chemical Society, Washington, DC. (10) Hudson, R. H. E., Uddin, A. H., and Damha, M. J. (1995) Association of branched nucleic acids: structural and phys-

Braich and Damha icochemical analysis of antiparallel T.AT triple-helical DNA. J. Am. Chem. Soc. 117, 12470-12477. (11) Azhayeva, E., Azhayev, A., Guzaev, A., Hovinen, J., and Lo¨nnberg, H. (1995) Looped oligonucleotides form stable hybrid complexes with a single-stranded DNA. Nucleic Acids Res. 23, 1170-1176. (12) Brandenburg, G., Petersen, G. V., Rasmussen, K., and Wengel, J. (1995) Branched oligodeoxynucleotides: a new synthetic strategy and formation of strong intra- and intermolecular triple helical complexes. Bioorg. Med. Chem. Lett. 5, 791-794. (13) von Bu¨ren, M., Petersen, G. V., Rasmussen, K., Brandenburg, G., Wengel, J., and Kirpekar, F. (1995) Branched oligodeoxynucleotides: automated synthesis and triple helical hybridization studies. Tetrahedron 51, 8491-8501. (14) Nam, K., Hudson, R. H. E., Chapman, K. B., Ganeshan, K., Damha, M. J., and Boeke, J. D. (1994) Yeast lariat debranching enzyme: substrate and sequence specificity. J. Biol. Chem. 269, 20613-20621. (15) Sproat, B. S., Beijer, B., Grøtli, M., Ryder, U., Morand, K. L., and Lamond, A. I. (1994) Novel solid-phase snthesis of branched oligoribonucleotides, including a substrate for RNA debranching enzyme. J. Chem. Soc., Perkin Trans. 1, 419430. (16) Damha, M. J., Ganeshan, K., Hudson, R. H. E., and Zabarylo, S. V. (1992) Solid-phase synthesis of branched oligoribonucleotides related to messenger RNA splicing intermediates. Nucleic Acids Res. 200, 6565-6573. (17) Hudson, R. H. E., and Damha, M. J. (1993) Nucleic acid dendrimers: novel bioploymer structures. J. Am. Chem. Soc. 115, 2119-2124. (18) Amato, I. (1993) How to drive nucleic acids up a tree. Science 260, 491. (19) Ganeshan, K., Tadey, T., Nam, K., Braich, R., Purdy, W. C., Boeke, J. D., and Damha, M. J. (1995) Novel approaches to the synthesis and analysis of branched RNA. Nucleosides Nucleotides 14, 1009-1013. (20) Grøtli, M., and Sproat, B. S. (1995) A universal solid-phase synthesis of branched oligoribonucleotides. J. Chem. Soc., Chem Commun., 495-497. (21) Damha, M. J., Ginnaris, P. A., and Zabarylo, S. V. (1990) An improved procedure for derivatization of controlled-pore glass beads for solid-phase oligonucleotide synthesis. Nucleic Acids Res. 13, 3813-3821. (22) Damha, M. J., and Ogilvie, K. K. (1993) Oligoribonucleotide synthesis. Protocols for Oligonucleotide and Analogues: Synthesis and Properties, Methods in Molecular Biology (S. Agrawal, Ed.) pp 81-114, Humana Press, Totowa, NJ. (23) Pon, R. T. (1984) Ph.D. Thesis, McGill University, Montreal, PQ, Canada. (24) Tanya, T., and Purdy, W. C. (1994) Capillary electrophoretic separation of nucleotide isomers via compelexation with cyclodextrin and borate. J. Chromatogr. B 657, 365-372. (25) Kierzek, R., Kopp, D. W., Edmons, M., and Caruthers, M. H. (1986) Chemical synthesis of branched RNA. Nucleic Acids Res. 14, 4751-4764. (26) Brown, D. M., Magrath, D. I., and Todd, A. R. (1955) J. Chem. Soc., 4496. (27) de Rooij, J. F. M., Willie-Hazeleger, G., Burgers, P. M. J., and van Boom, J. H. (1979) Neighbouring group participation in the unblocking of phosphotriesters of nucleic acids. Nucleic Acids Res. 6, 2237-2259. (28) Damha, M. J., Pon, R. T., and Ogilvie, K. K. (1985) Chemical synthesis of branched RNA: novel trinucleotide diphosphates containing vicinal 2′-5′ and 3′-5′ phosphodiester linkages. Tetrahedron Lett. 26, 4839-4842. (29) Reese, C. B., and Skone, P. A. (1985) Action of acid on oligoribonucleotide phosphotriester intermediates. Effect of released vicinal hydroxy functions. Nucleic Acids Res. 13, 5215-5231. (30) van de Sande, J. H., Ramsing, N. B., Germann, M. W., Elhorst, W., Kalisch, B. W., Kitzing, E. V., Pon, R. T., Clegg, R. C., and Jovin, T. M. (1988) Parallel stranded DNA. Science 241, 551-557. (31) Damha, M. J., Usman, N., and Ogilvie, K. K. (1989) Solution and solid phase chemical synthesis of arabinonucleotides. Can. J. Chem. 67, 831-839.

Bioconjugate Chem., Vol. 8, No. 3, 1997 377

Synthesis and Association of Branched Oligonucleotides (32) Damha, M. J., and Ogilvie, K. K. (1988) Synthesis and spectroscopic analysis of branched RNA fragments: messenger RNA splicing intermediates. J. Org. Chem. 53, 37103722. (33) Damha, M. J., and Ogilvie, K. K. (1988) Conformational properties of branched RNA fragments in aqueous solution. Biochemistry 27, 9403-6416. (34) Sund, C., Agback, P., Koole, L. H., Sandstro¨m, A., and Chattopadhyaya, J. (1992) Assessment of competing 2′-5′ versus 3′-5′ stackings in solution structure of branched-RNA by 1H- and 31P-NMR spectroscopy. Tetrahedron 48, 695-718. (35) Varani, G. (1995) Exceptionally stable nucleic acid hairpins. Annu. Rev. Biophys. Biomol. Struct. 24, 379-404. (36) Pilch, D. S., Levenson, C., and Shafer, R. H. (1990) Structural analysis of the (dA)10‚2(dT)10 triple helix. Proc. Natl. Acad. Sci. U.S.A. 87, 1942-1946.

(37) Tang, J. Y., Temsamani, J., and Agrawal, S. (1993) Selfstabilized antisense oligodeoxynucleotide phosphorothioates: properties and anti-HIV activity. Nucleic Acids Res. 21, 2729-2735. (38) Giovannangeli, C., Monteray-Garestier, T., Rouge´e, M., Chassignol, M., Thuong, N. T., and He´le`ne, C. (1991) Singlestranded DNA as a target for triple-helix formation. J. Am. Chem. Soc. 113, 7775-7777. (39) Salunkhe, M., Wu, T., and Letsinger, R. L. (1992) Control of folding and binding of oligonucleotides by use of a nonnucleotide linker. J. Am. Chem. Soc. 114, 8768-8772. (40) Wang, S., and Kool, E. T. (1994) Recognition of singlestranded nucleic acids by triplex formation: the binding of pyrimidine-rich sequences. J. Am. Chem. Soc. 116, 8857-8858.

BC9700300