DNA triple helix formation has been the focus of much attention recently in terms of its potential use as a method for selective regulation of gene expression(1) . Oligonucleotides can inhibit protein synthesis in several different ways. In the antisense strategy, an oligonucleotide binds to a targeted mRNA molecule in a sequence-specific manner to prevent subsequent translation of the message into protein. Alternatively, binding of an oligonucleotide directly to a gene or gene promoter, via triplex formation, can arrest or block transcription. The process of down-regulating gene expression through triple helix formation is referred to as the antigene strategy. Triple helix formation has also been utilized in the development of artificial nucleases, created by tethering a cleaving agent to a triplex-forming oligonucleotide(2, 3, 4) . This leads to breaks in double-stranded DNA at very specific sites. Several studies have demonstrated the feasibility of using oligonucleotides in gene regulation(5, 6, 7) . Several properties need to be better understood in order to design optimally effective therapeutic agents based on triplex formation. These include sequence specificity of triplex formation, stability of the complex formed, delivery of the oligonucleotide into cells, and resistance of the oligonucleotides to endogenous nucleases.

The primary requirement for triple helix formation is a homopurine-homopyrimidine sequence in the target duplex. Typically, the third strand sequence is also homopurine or homopyrimidine. Depending on the nature of the third strand, there are two main categories of triple helices: pyrimidine* purine*pyrimidine (pyr*purpyr) or purine*purinepyrimidine (pur*purpyr), where * represents Hoogsteen base pair formation between the third strand and the purine strand of the duplex and denotes the Watson-Crick base pair of the duplex. Most studies to date have focused on pyr*purpyr triplexes. This type of triplex is more stable at low pH because the cytosines on the third strand require protonation in order to form Hoogsteen hydrogen bonds with the purine strand of the duplex. In contrast, pur*purpyr triplexes are stable at neutral pH. Thermal denaturation studies on pur*purpyr triplexes often reveal a single transition, suggesting the simultaneous dissociation of all three strands(8) . In most examples of pyr*purpyr triplexes, a biphasic transition is observed, with the third strand dissociating at a lower temperature than the duplex(9, 10, 11, 12) . This suggests that a pyrimidine third strand may be less stable, in general, than a purine third strand.

Unlike the Watson-Crick base-paired double helix, in which the two strands are bound antiparallel to each other, the third strand of a triple helix can be either parallel or antiparallel with respect to the purine strand to which it is hydrogen-bonded. The polarity of the third strand in the triplex is dependent on the sequence as well as the base composition(13, 14) . A homopyrimidine third strand containing cytosine and/or thymine binds to the purine strand in a parallel orientation. The binding of a homopurine oligonucleotide to the purine strand depends on the sequence. Experimental evidence has been reported for both parallel (15) and antiparallel (16) orientations of the third strand in triplexes composed solely of G*GC triplets. But if the third strand contains both Gs and As, the polarity of the third strand will be antiparallel with respect to the purine strand(8, 17) , since As with anti-glycosidic bond conformations can form only reverse-Hoogsteen hydrogen bonds (13) and hence direct the oligonucleotide in an antiparallel orientation.

Hogan and co-workers (18, 19) have reported antigene activity of oligonucleotides composed of G and T bases. However, there have been only a limited number of physical studies on triplexes formed by such sequences. Helene and co-workers (13, 20) have demonstrated that the orientation of a G, T third strand depends on its sequence. Calculations indicate that, for a 10-mer triplex whose third strand is composed of equal number of G and T bases (with corresponding G and A bases in the purine strand of the target duplex), the third strand will be antiparallel to the duplex purine strand if there are three or more GpT or TpG steps; otherwise it will be parallel. According to the calculations of Sun and Helene(13) , this can be explained in terms of a balance between the preference for the parallel orientation (Hoogsteen hydrogen bonding) in an all G third strand and the better tolerance for the lack of isomorphism between G*GC and T*AT triplets in the antiparallel orientation (reverse Hoogsteen hydrogen bonding). Radhakrishnan et al.(21) carried out NMR studies on an intramolecular triplex composed of a third strand containing Gs and one T. These studies provided clear evidence for the distortion induced by insertion of a T*AT triplet within a stack of G*GC triplets.

In earlier studies, we have targeted the d(GAG) d(CTC) for triple helix formation with either d(CTC), to make a pyr*purpyr triplex(9) , or with d(GAG), to make a pur*purpyr triplex(8) . We have recently described the Fourier transform infrared spectrum of the pyr/pur*purpyr intermolecular triplex d(GTG)*d(GAG)d(CTC) (22) . Here we employ UV, CD, gel electrophoresis, and NMR to further characterize this triplex. While results from fluorescence energy transfer experiments are inconclusive regarding the orientation of the third strand relative to the purine strand of the duplex, UV melting analysis of fluorescently labeled triplexes suggests that the preferred orientation of the third strand is parallel to the purine strand of the duplex in the unlabeled triplex.

MATERIALS AND METHODS

Oligodeoxynucleotides were synthesized using standard phosphoramidite chemistry on an automated DNA synthesizer as described earlier(8) . The deprotected oligonucleotides were extensively dialyzed against 1 mM Tris-HCl with several changes of buffer over a period of 2-3 days. The purity of the resulting oligonucleotides was checked by NMR and gel electrophoresis and found to be greater than 95%. Molar extinction coefficients of various oligonucleotides in 1 mM Tris-HCl were determined by phosphate analysis(23) , and the concentration of the stock solutions was determined using the following values: d(GAG): = 11500 cm(mol of base)L; d(CTC): = 8300 cm(mol of base)L; d(GTG): = 9900 cm(mol of base)L; d(TCTCT): = 8500 cm(mol of base)L. All experiments were carried out in buffer containing 10 mM Tris-HCl with 50 mM MgCl at pH 7.4. Duplex and triplex samples were prepared by mixing the oligonucleotides at appropriate ratios in the desired buffer, followed by heating at 80 °C for about 5 min and cooling slowly to room temperature. The samples were equilibrated at 5 °C overnight before use.

Labeled oligonucleotides were synthesized using phosphoramidite chemistry on ABI 394 DNA synthesizers. The 3`-labels were coupled to oligonucleotides prepared using the 3`-DMT-C6 amine-ONTM controlled pore glass (Clontech). The 5`-carboxytetramethlyrhodamine was added to a 5`-aminohexyl-derivatized oligonucleotide (Glen Research). All amino oligonucleotides were ion-exchanged to their lithium salts by precipitation from ethanol-acetone as described previously(24) . The crude, derivatized oligonucleotides were dissolved in sodium carbonate buffer (0.25 ml, 0.1 M, pH 9.0) and treated with 5-carboxytetramethylrhodamine-N-hydroxysuccinimidyl ester (Applied Biosystems, 3 µl in MeSO) or Malachite Green isothiocyanate (1 mg in 50 µl of dry N,N-dimethylformamide, Molecular Probes). After coupling overnight at room temperature, the oligonucleotides were again precipitated as their lithium salts from ethanol-acetone. The oligonucleotides were resuspended in triethylammonium acetate buffer (0.1 M, pH 7.0, buffer A), filtered, and purified by reverse-phase HPLC ()on a Waters 996 Diode Array Detector system with a Hamilton PRP-1 column (300 7 mm) using a gradient of 0-40% acetonitrile in buffer A at a flow rate of 2 ml/min.

The 5`-Malachite Green oligonucleotides were made directly on the DNA synthesizer using a leuco-Malachite Green phosphoramidite, details of which will be published elsewhere. ()The oligonucleotides were coupled to the Malachite Green amidite using normal activation and oxidation steps. On removal from the instrument, the synthesis columns were treated with a solution of freshly prepared iodosobenzene in dichloromethane (0.01 M, 30 s) and washed with more solvent and air-dried. The derivatized controlled pure glass was treated with ammonia for 4 h at 55 °C, then the supernatants were passed through NAP-10 columns (Pharmacia LKB Biotechnol), and the desired oligonucleotides were purified by reverse phase-HPLC as described above.

The UV absorbance and melting studies were carried out on a Gilford 2600 UV/Vis spectrophotometer equipped with a Gilford 2527 thermoprogrammer or on a Cary 3 spectrophotometer. UV melts were done with a heating rate of 0.25 or 0.3 deg/min. CD spectra were recorded on a Jasco J500A spectropolarimeter using cells of 0.1-, 1.0-, or 10-mm optical pathlength, with temperature controlled by an external circulating water bath. CD spectra reported are the average of eight scans. CD melting experiments were done by manually changing the temperature of the bath and letting it equilibrate at the desired temperature for 5 min. The cell temperature was monitored by attachment of a microprobe directly onto the sample cell. The ellipticity data were collected at various wavelengths, and each point on the melting curve represents the average of 100 readings taken over a period of 100 s. Mixing curves were constructed from data obtained from the CD spectra of samples containing varying mole ratios of duplex and the third strand, with the total concentration of the duplex plus the third strand held constant.

Thermodynamic parameters for the formation of the triplex were estimated from the concentration dependence of the thermal melting temperature of the triplex in the concentration range 10 to 850 µM. Since the two transitions were well separated from each other over a wide concentration range, we were able to follow the concentration dependence of each transition separately. Assuming a two-state model for each transition, we analyzed the biphasic melting curves according to the procedure described by Pilch et al.(9) using the equation, 1/T = (R/H°)lnC + (S° - 0.188R)/H° for each individual transition. H° and S° are the enthalpy and entropy changes, respectively, associated with the respective structural transitions, T is the temperature corresponding to the maximum of the dA/d(1/T) plot derived from the melting curves, where A is the absorbance, C is the total concentration of the d(GTG) strand (which is equal to the total concentration of each of the other two strands), and R is the gas constant. H° and S° were calculated from the slope and intercept of the straight lines obtained by fitting the data points on 1/Tversus lnC plots by linear regression. The UV melting curves used for the thermodynamic analysis were obtained at a heating rate of 0.25°/min; a window of ± 2 °C was used for calculating the derivative. The reversibility of the transitions was checked by heating the sample at 1 °C/min up to 90 °C followed by cooling at the same rate. No hysterisis was observed in the melting curves showing that the equilibrium is maintained at each point on the melting curve.

Gel electrophoresis was carried out under non-denaturing conditions in 15% polyacrylamide gels containing acrylamide and bis(acrylamide) in a 29:1 ratio, cast in 90 mM Tris-borate and 50 mM MgCl. Samples for electrophoresis were prepared in 90 mM Tris-borate and 50 mM MgCl and were diluted into loading buffer containing 90 mM Tris-borate, 50 mM MgCl, and 5% Ficoll, so that the final concentration ranged from 50 to 60 µM in single strand, duplex, or triplex. Then, 40 µl of each sample were loaded into the wells, and electrophoresis was carried out in buffer containing 90 mM Tris-borate and 50 mM MgCl, in the cold room (4 °C) or at room temperature (25 °C) at about 7 V/cm for 24 h. Gels were then visualized by UV shadowing on a fluorescent background and photographed.

Proton NMR spectra of the samples were recorded at 500 MHz on a General Electric spectrometer (GN-500) equipped with an Oxford Instruments magnet and a Nicolet 1280 computer. Spectra of the samples were taken in aqueous solutions containing 10 mM Tris-HCl, 0-50 mM MgCl, and 85% HO, 15% DO, at 5 or 20 °C. MgCl titrations were carried out by adding aliquots of MgCl into samples containing the duplex and the third strand in equimolar ratios (1 mM each) dissolved in the buffer. After each addition of MgCl, the samples were heated at 80 °C for 5 min and cooled. The spectra were acquired using a T331 pulse sequence for solvent suppression, with a pulse repetition time of 5 s and interval delay of 109 µs.

RESULTS

We have shown previously that the oligonucleotide d(GTG), used here as the third strand for triplex formation, can, in the presence of monovalent cations such as Na or K, self-associate to form a quadruplex structure(25) . This quadruplex structure can be eliminated by dialysis against 1 mM Tris-HCl buffer. Tris, a bulky cation, does not promote quadruplex formation of d(GTG), nor do the concentrations of Mg used in this study to stabilize the triplex (data not shown).

UV Melting Studies

CD Spectra

Figure 1: CD spectra of d(GTG)*d(GAG)d(CTC) triplex (-),d(GAG)d(CTC) duplex (- -), d(GTG) () and mathematical sum of the spectra of d(GAG)d(CTC) and d(GTG) (- - -), all at 0.5 mM in structure (single strand, duplex, or triplex) in buffer containing 10 mM Tris-HCl (pH = 7.5) and 50 mM MgCl, recorded at 5 °C in cells of 0.1-mm pathlength.

Thermal denaturation of the d(GTG)*d(GAG) d(CTC) triplex was studied by temperature induced changes in the CD spectrum at the same high oligonucleotide concentrations. Fig. 2a shows a typical thermal denaturation profile of the triplex monitored by CD changes at 240 and 277 nm. As the temperature increases, the ellipticity values at the two wavelengths also increase, except for a small decrease at high temperature for the 277 nm band. The change in ellipticity with temperature is biphasic and cooperative for both wavelengths, with the two transitions centered around 38 and 60 °C. It is evident from Fig. 2a that the low temperature transition is most readily followed at 277 nm while the high temperature transition is best monitored at 240 nm. Fig. 2b shows the CD melting profile of the duplex at the same two wavelengths. The duplex shows only one transition around 60 °C. The ellipticity at 277 nm does not show any cooperative change in the lower temperature region, where the triplex shows a large change. As in the case of the triplex, however, the temperature profile shows a relatively small but cooperative decrease in ellipticity around 60 °C. The 240-nm melting curve is also monophasic, with a transition at the same temperature. Hence the transition that occurs around 38 °C for the triplex is due to the dissociation of the third strand from the triplex while that at 60 °C is due to the dissociation of the duplex into single strands.

Figure 2: CD melting curves of the triplex, d(GAG)* d(GAG)d(CTC) (a) and the duplex, d(GAG)d(CTC) (b) in 10 mM Tris-HCl and 50 mM MgCl, monitored by the change in ellipticity at 277 nm (filled circle) or 240 nm (open circle) bands of the CD spectrum. Sample conditions are the same as in Fig. 1.

The differences in the CD spectra elicited by the binding of d(GTG) to the duplex were utilized to construct a mixing curve that provides further evidence for complex formation between the duplex and the third strand and also determines the stoichiometry of the complex formed. Fig. 3presents the mixing curve for the binding of d(GTG) to the duplex monitored by the CD changes associated with complex formation for the 277 nm band. The titration of the third strand into the duplex was carried out by varying the mole ratio of the duplex and the third strand, keeping the total concentration, duplex plus third strand, constant. The discontinuous change in slope in the mixing curve observed when the mixture contains equal amounts of the duplex and the third strand demonstrates that the stoichiometry of the complex formed is 1 duplex/1 strand of d(GTG).

Figure 3: CD mixing curve for the binding of d(GTG) to the d(GAG)d(CTC) duplex monitored by the ellipticity changes for the CD band at 277 nm. Sample conditions are the same as in Fig. 1.

Thermodynamic Analysis

Figure 4: Concentration variation of T, determined from UV melting curves, for both the low temperature (open circle) and high temperature (closed circle) melting transition for the triplex d(GTG)*d(GAG)d(CTC) in the same buffer as in Fig. 1.

Gel Electrophoresis

Figure 5: Polyacrylamide gel electrophoresis pattern of various complexes of d(GTG), d(GAG), and d(CTC) under non-denaturing conditions in the presence of 40 mM MgCl at 4 °C (A) or at 25 °C (B). Lane 1, d(GAG); lane 2, d(CTC); lane 3, d(GTG); lane 4, d(GAG) + d(CTC); and lane 5, d(GAG) + d(CTC) + d(GTG).

NMR Spectra

Figure 6: Imino proton region of the NMR spectrum of d(GAG)d(CTC) duplex + d(GTG) in 85% HO, 15% DO containing 10 mM tris/HCl in the absence of any MgCl (A) or in the presence of 50 mM MgCl (B), at 20 °C. Sample concentration is 1 mM in each strand.

In an effort to sharpen the NMR lines, we designed another triplex in which the pyrimidine strand has two extra thymines at each end, the other two strands being same as in the original triplex. With this ``overhang'' triplex, d(GTG)*d(GAG) d(TTCTCTT), we were able to reduce the temperature and maintain reasonably narrow line widths for the imino proton peaks. Fig. 7provides a comparison of the spectra obtained for this triplex with that of the ``blunt end'' triplex, d(GTG)* d(GAG)d(CTC), at 5 °C. The blunt end spectrum is very broad and featureless whereas the overhang spectrum has fairly well resolved peaks. In this case, the additional hydrogen bonds due to Hoogsteen base pairing of the third strand to the duplex are clearly seen. The improvement in the triplex spectrum due to the presence of the overhang thymines is quite remarkable.

Figure 7: Imino proton spectrum of the d(GTG)*d(GAG) d(TCTCT) overhang triplex (A) and the blunt end triplex, d(GTG)*d(GAG)d(CTC) (B) in the presence of 50 mM MgCl, at 5 °C. Other conditions are the same as in Fig. 5.

The possible hydrogen bonding schemes for the interaction of the third strand with the purine strand of the duplex are depicted in Fig. 8. In either the Hoogsteen or reverse Hoogsteen modes, there are six guanine imino protons and four thymine imino protons involved in base pair formation involving the third strand. This results in a maximum of 20 imino protons potentially detectable by NMR. In the spectrum in Fig. 7there are only approximately 16 observable protons, most likely due to fraying of the triplex ends. While we have not made assignments for these imino protons, evidence for formation of G*GC triplets can be found in both CD spectra (see above) as well as Fourier transform infrared results(22) , while the latter also provides evidence for formation of T*AT triplets.

Figure 8: Hoogsteen (left) and reverse Hoogsteen (right) base pairing patterns for T*AT and G*GC triplets. - - - -, Watson-Crick hydrogen bonding interactions; , Hoogsteen or reverse Hoogsteen interactions.

Studies on Fluor-quencher Oligonucleotide Conjugates

We have attempted to gather information about the orientation of the third strand in the triplex d(GTG)*d(GAG) d(CTC) via energy transfer measurements. Oligonucleotides were labeled with TAMRA (donor) or Malachite Green (acceptor) at the 3` or 5` end (see Fig. 9). TAMRA, a rhodamine derivative, has an absorbance maximum at 560 nm and strong fluorescence emission in the 570-650 nm range. Malachite Green, the acceptor, has an absorbance in the 550-700 nm region and is non-fluorescent. This choice of donor-acceptor pair simplifies the analysis, since only the donor is fluorescent. Hence, the efficiency of energy transfer can be calculated directly from the decrease in donor fluorescence in the presence of the acceptor without interference from acceptor fluorescence. Triplex samples using labeled oligonucleotides were made using the same conditions as that used for the unlabeled samples and at the same high DNA concentrations.

Figure 9: Structure of oligonucleotide conjugates for studies on labeled triplexes.

Somewhat unexpectedly, the fluorescence was efficiently quenched (88% energy transfer efficiency) when TAMRA was at either the 5` or 3` end of the third strand, with Malachite Green fixed at the 3` end of the purine strand of the duplex. Control experiments carried out on duplex samples revealed an efficiency of transfer of 56% with both labels at the 3` end and 87% with one label at the 3` end, the other at the 5` end. These results, expected for antiparallel duplexes, are similar to those reported on other duplexes(29) . Thus the most likely interpretation of our energy transfer results on the triplex samples is that the donor and the acceptor are positioned at the same end of the molecule in both doubly labeled samples. This implies that the orientation of the third strand is determined by the favorable interaction between donor and acceptor molecules tethered to the oligonucleotides. Thus, any inherent free energy difference between the two orientations of the third strand in the absence of labels must be small compared to the magnitude of the donor-acceptor interaction.

If there is an inherent preference for one orientation of the third strand over the other, one may expect some difference in the thermal stability of the two doubly labeled triplexes. Fig. 10shows the first derivative UV melting curves for the completely unlabeled triplex (sample A), the doubly labeled triplex with the third strand labeled either at the 5` end (sample B), or at the 3` end (sample C). As mentioned above, the unlabeled triplex possesses two transitions; as the concentrations used for the UV melts are significantly lower than those used for the CD melts, these transitions occur at somewhat lower temperatures. The first one, centered at about 18 °C, arises from the dissociation of the third strand while the second one, around 55 °C, is due to the melting of the duplex. While samples B and C also melt in a biphasic manner, there is a significant difference in transition temperatures. The first transition of each of these samples is shifted to higher temperature than in the unlabeled sample. In sample B, this transition occurs at 42 °C, while for sample C, it is further shifted to 57 °C. The second transition occurs at 62 °C in both samples, which is somewhat higher than the second transition in the unlabeled triplex.

Figure 10: First derivative plots of the UV melting curves of the unlabeled triplex, d(GTG)*d(GAG)d(CTC) () and fluorescently labeled triplexes with the acceptor at the 3` end of the purine strand of the duplex and the donor at the 5` end (---) or at the 3` end (-) of the third strand. All samples contained 0.033 mM triplex.

The remarkable increase in stability of the third strand observed in samples B and C must arise from the interaction between the two dye molecules in close proximity to each other. This large effect supports the hypothesis that the orientation of the third strand is determined by the favorable donor-acceptor interaction in these samples. As a result of this interaction, the donor and acceptor molecules are positioned at the same end of the triplex in both samples B and C. In order for this to occur, the third strand in sample B will have to bind to the duplex in an antiparallel orientation with respect to the purine strand whereas for sample C, the binding has to be parallel to the purine strand. Thermal denaturation of the two triplexes, B and C, shows that sample C forms a much more stable triplex compared to sample B, implying that the triplex with the third strand parallel to the purine strand of the duplex is more stable than the one with antiparallel orientation. If we assume that the stabilization due to the dye-dye interaction is similar in both cases, we may infer that, in the absence of this interaction, the preferred orientation of the third strand in the triplex, d(GTG)*d(GAG)d(CTC), is parallel to the purine strand of the underlying duplex. This agrees with the predictions based on theoretical calculations that for a triplex forming oligonucleotide with 50% each of Gs and Ts, the preferred orientation will be parallel to the purine strand if the number of GpT or TpG steps is 3 or smaller(13) .

DISCUSSION

The data presented above provide several lines of evidence for triple helix formation by the binding of a mixed purine/pyrimidine oligonucleotide, d(GTG), to a homopurine-homopyrimidine duplex, d(GAG)d(CTC). It was clear from the beginning of these studies that this triplex, d(GTG)* d(GAG)d(CTC), was strikingly less stable than the closely related pur*purpyr triplex, d(GAG)*d(GAG) d(CTC), because its formation required substantially higher oligonucleotide concentrations. Unlike the latter triplex, the G/T third strand triplex exhibits a biphasic melting profile as measured both by CD and UV absorbance.

Circular dichroism, which is inherently more sensitive to conformational changes than absorbance, proved to be a useful tool in studying triplex formation for the G/T third strand considered here. The large negative bands in the triplex CD spectrum, with minima at 277 and 210 nm, are not present in either the duplex or the d(GTG) spectra and hence result from the binding of the third strand to form the triplex. The negative ellipticity near 277 nm has been described before in poly(dG)*poly(dG)poly(dC) triplexes(30, 31) . Any changes in the third strand binding state should, then, primarily affect these bands. The CD melting curve recorded at 277 nm has a large cooperative increase in ellipticity at 38 °C, indicating the dissociation of the third strand in this temperature region. This melting curve also shows a small cooperative decrease in ellipticity in the temperature range corresponding to dissociation of the duplex. In comparison, the melting curve of the duplex shows a small but non-cooperative increase in ellipticity at 277 nm at low temperatures, followed by a larger, cooperative decrease in ellipticity at 60 °C as it denatures. These observations indicate that the 277 nm band primarily tracks the binding of the third strand.

The CD melting curves presented in this study show some similarities with melting curves reported for the premelting transitions of a 45-base pair long duplex composed of blocks of d(A)d(T) distributed between random GC base pairs(32) . For this duplex, the ellipticity at the longer wavelength band of the CD spectrum gradually increased with increasing temperature until the dissociation of the duplex. This premelting transition was attributed to temperature-dependent structural changes in the A tract of the B form double helix. At low temperature the double helix is bent due to the presence of the A tract. As the temperature is increased, the helix straightens out. The ellipticity change at 270 nm due to premelting is several times higher than that associated with the subsequent duplex dissociation. But in the case of d(GAG)d(CTC), the total ellipticity change for the 277 nm band in the temperature region below denaturation of the duplex is actually smaller than that due to the duplex dissociation. In contrast, the triplex melting curve for d(GTG)*d(GAG)d(CTC), recorded at 277 nm, exhibits a much larger change in the low temperature region compared to that corresponding to the duplex dissociation. Unlike the low temperature region of the duplex melting curve, however, this transition is highly cooperative. It is very unlikely that this transition is due to the premelting changes in the A tract of the triple helix. The binding of the third strand in the major groove of the duplex is expected to make the helix more rigid compared to the duplex and hence less bending, if any, may be expected in the triplex.

The thermodynamic stability of the d(GTG)*d(GAG) d(CTC) triplex is considerably lower than that of related triplexes with the same underlying duplex but all purine or all pyrimidine third strands. Thus, at 25 °C, G° for binding of the third strand to the underlying duplex is -5.8 kcal/mol, corresponding to a binding constant of 1.8 10. Both H° and S° are less negative for this triplex than for the other triplexes based on this duplex (8, 9) with the balance overall leading to a lower stability in terms of G°. This decrease in stability arises, at least partially, from the presence of GpT and TpG steps which is destabilizing due to the non-isomorphic nature of the G*GC and T*AT triplets(13) .

Gel retardation experiments carried out at two different temperatures were undertaken to demonstrate formation of the triple helix and also to confirm that the low temperature transition observed in the CD melting studies is indeed due to the dissociation of the third strand from the duplex. The experiments were carried out at two temperatures, one well below the midpoint of the first transition and the second at a temperature above the first transition midpoint but below the starting point of the second transition. The presence of a slowly migrating species at low temperature demonstrates formation of the triplex. A similar gel retardation pattern was reported for the d(GAG)*d(GAG)d(CTC) triplex(8) . The smearing of this band may be due to end-to-end association of the triplex molecules. At the higher temperature, the slowly migrating band, due to the triplex, is not present while the band due to the duplex remains unaffected, indicating dissociation of the third strand from the triplex at this higher temperature. These experiments clearly show that the triplex formed at low temperature dissociates into duplex and single strand in the temperature region well below the dissociation of the duplex, confirming the conclusions derived from the CD studies on the biphasic melting of this triplex.

Additional evidence for triplex formation is obtained from the NMR spectrum of the mixture of the duplex and the third strand in the presence of MgCl. The exchangeable proton region of the NMR spectrum of the mixture in the absence of MgCl shows peaks from imino protons that are in slow exchange with solvent due to the duplex formation through Watson-Crick hydrogen bonding. Appearance of new imino protons in the presence of MgCl results from the formation of hydrogen bonds between the third strand and the duplex. These new resonances arise from the imino protons on the third strand guanines and thymines interacting with the duplex purine strand, as illustrated in Fig. 8, thereby confirming the formation G*GC and T*AT base triplets.

Intermolecular triplexes typically exhibit very broad NMR lines (33, 34, 35) , compared to those of duplexes, despite the fact that their overall dimensions, and hence correlation time, differ only slightly from that of duplexes. In earlier studies involving DNA duplexes, we observed evidence of end-to-end stacking, based on concentration-dependent nuclear Overhauser effect between terminal deoxynucleosides(36) . Because of the larger surface area for stacking in the case of a terminal triplet compared to a terminal base pair, this effect should be enhanced for triplexes. Thus, minimization or elimination of this stacking between the molecules should improve the spectrum. In an effort to diminish the efficiency of intermolecular stacking, we designed a triplex with one strand possessing extra thymines on each end. These extra terminal residues are expected to disrupt the end-to-end stacking between triplex molecules, thereby minimizing line broadening. Spectra shown in the Fig. 7clearly demonstrate the effectiveness of this design. The intermolecular stacking interaction may also be responsible for some effects seen in optical and electrophoretic studies, such as the relatively large slope observed for the low temperature base lines in the UV melting profiles of triplexes and also characteristic smearing of triplex bands in gel electrophoresis experiments(8) .

The orientation of the third strand with respect to the underlying duplex is a critical element in designing triplex-forming oligonucleotides. A homopyrimidine third strand binds to the duplex with an orientation parallel to the purine strand. There is some experimental evidence for the antiparallel orientation in triplexes composed solely of A*AT triplets(37) . Also, there have been reports of both parallel (15) and antiparallel (16, 38) third strand orientations in triplexes composed solely of G*GC triplets. Theoretical calculations, however, indicate greater stability for the parallel orientation in triplexes containing only G*GC triplets(15, 20) . Oligonucleotides composed of Gs and As have been shown to bind antiparallel to the purine strand of the duplex(8, 17) . Theoretical studies have suggested that a third strand containing both Gs and Ts can bind either parallel or antiparallel to the purine strand of the duplex(13, 20) . A parallel orientation is preferred if the number of TpG or GpT steps is less than three for a sequence containing an equal number of Gs and Ts. An increase in the number of purine-pyrimidine steps will result in an antiparallel orientation for the third strand.

We have addressed this question using fluorescence resonance energy transfer experiments. Our results showed extensive quenching of the donor fluorescence regardless of which end of the third strand was labeled. The most likely explanation of this observation is that, in the doubly labeled triplexes, the orientation of the labeled third strand is determined by the favorable interaction between the donor and acceptor labels rather than by the inherent preference for orientation of the third strand in unlabeled triplexes. Thus, we significantly perturbed the triplex by adding the fluorescence labels and were prevented from ascertaining the desired information. UV melting studies, however, did provide suggestive evidence by showing that the triplex with the third strand labeled at the same end as the purine strand was substantially more stable, in agreement with the predictions of Helene and co-workers(13, 20) .

Other reports have appeared concerning the strand polarity in triplexes containing third strands composed of Gs and Ts. Beal and Dervan (17) have studied the binding of a triplex-forming oligonucleotide containing Gs and Ts. This oligonucleotide, containing over 70% Gs, was observed to bind antiparallel to the purine strand of the duplex target. Based on the number of TpG or GpT steps, this is consistent with earlier predictions(13, 20) . However, a recent report on the oligonucleotide d(GGGTTGG), conjugated to an intercalating agent, states that this oligonucleotide, which also has about 70% Gs but only two GpT or TpG steps, binds to the purine strand in an antiparallel orientation(39) . This is unexpected based on our experimental results reported here as well as the calculations of Sun et al. (13, 20). One possible explanation for this discrepancy is the presence of the tethered intercalator, which may have influenced the orientation of the third strand.

In summary, we have demonstrated the formation of the d(GTG)*d(GAG)d(CTC) triplex in the presence of MgCl by various spectrophotometric measurements, imino proton NMR and gel electrophoresis. While both UV and CD melting curves are biphasic, the latter show more pronounced changes for dissociation of the triplex. The third strand of this triplex binds less tightly than that of d(GAG)*d(GAG) d(CTC) and appears to have the opposite orientation. This preference in strand orientation is not very strong, as the presence of a donor-acceptor pair of fluorescence labels can alter the orientation of the third strand.

FOOTNOTES

*

This work was supported by Grant MCB 9218687 awarded by the National Science Foundation and a grant from the University of California Systemwide Biotechnology Research and Training Program. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore by hereby marked ``advertisement'' in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

§

To whom correspondence should be addressed. Tel.: 415-476-2761; Fax: 415-476-0688.

()

The abbreviation used is: HPLC, high performance liquid chromatography.

()

S. G. Will and D. Knowles, manuscript in preparation.

ACKNOWLEDGEMENTS

REFERENCES

1. Helene, C. (1991) Anticancer Drug Des. 6, 569-584 [Medline] [Order article via Infotrieve]
2. Strobel, S. A., and Dervan, P. B. (1990) Science 249, 73-75 [Abstract/Free Full Text]
3. Strobel, S. A., Doucette-Stamm, L. A., Riba, L., Housman, D. E., and Dervan, P. B. (1991) Science 254, 1639-1642 [Abstract/Free Full Text]
4. Strobel, S. A., and Dervan, P. B. (1991) Nature 350, 172-174 [CrossRef][Medline] [Order article via Infotrieve]
5. Duval-Valentin, G., Thuong, N. T., and Helene, C. (1992) Proc. Natl. Acad. Sci. U. S. A. 89, 504-508 [Abstract/Free Full Text]
6. Ing, N. H., Beekman, J. M., Kessler, D. J., Murphy, M., Jayaraman, K., Zendegui, J. G., Hogan, M. E., O'Malley, B. W., and Tsai, M. J. (1993) Nucleic Acids Res. 21, 2789-2796 [Abstract/Free Full Text]
7. Roy, C. (1993) Nucleic Acids Res. 21, 2845-2852 [Abstract/Free Full Text]
8. Pilch, D. S., Levenson, C., and Shafer, R. H. (1991) Biochemistry 30, 6081-6088 [CrossRef][Medline] [Order article via Infotrieve]
9. Pilch, D. S., Brousseau, R., and Shafer, R. H. (1990) Nucleic Acids Res. 18, 5743-5750 [Abstract/Free Full Text]
10. Plum, G. E., Park, Y. W., Singleton, S. F., Dervan, P. B., and Breslauer, K. J. (1990) Proc. Natl. Acad. Sci. U. S. A. 87, 9436-9440 [Abstract/Free Full Text]
11. Scaria, P. V., and Shafer, R. H. (1991) J. Biol. Chem. 266, 5417-5423 [Abstract/Free Full Text]
12. Mergny, J. L., Duval-Valentin, G., Nguyen, C. H., Perrouault, L., Faucon, B., Rougee, M., Montenaygarestier, T., Bisagni, E., and Helene, C. (1992) Science 256, 1681-1684 [Abstract/Free Full Text]
13. Sun, J. S., and Helene, C. (1993) Curr. Opin. Struct. Biol. 3, 345-356
14. Thuong, N. T., and Helene, C. (1993) Angew. Chem. Int. Ed. Engl. 32, 666-690
15. Ouali, M., Letellier, R., Sun, J. S., Akhebat, A., Adnet, F., Liquier, J., and Taillandier, E. (1993) J. Am. Chem. Soc. 115, 4264-4270 [CrossRef]
16. Chen, F. M. (1991) Biochemistry 30, 4472-4479 [CrossRef][Medline] [Order article via Infotrieve]
17. Beal, P. A., and Dervan, P. B. (1991) Science 251, 1360-1363 [Abstract/Free Full Text]
18. Orson, F. M., Thomas, D. W., McShan, W. M., Kessler, D. J., and Hogan, M. E. (1991) Nucleic Acids Res. 19, 3435-3441 [Abstract/Free Full Text]
19. McShan, W. M., Rossen, R. D., Laughter, A. H., Trial, J., Kessler, D. J., Zendegui, J. G., Hogan, M. E., and Orson, F. M. (1992) J. Biol. Chem. 267, 5712-5721 [Abstract/Free Full Text]
20. Sun, J. S., De Bizemont, T., Duval-Valentin, G., Montenay-Garestier, T., and Helene, C. (1991) C. R. Acad. Sci. III 313, 585-590 [Medline] [Order article via Infotrieve]
21. Radhakrishnan, I., de los Santos, C., and Patel, D. J. (1991) J. Mol. Biol. 221, 1403-1418 [Medline] [Order article via Infotrieve]
22. Dagneau, C., Liquier, J., Scaria, P. V., Shafer, R. H., and Taillandier, E. (1994) in Proceedings of the Eighth Conversation, State University of New York, Albany NY (Sarma, R. H., and Sarma, M. H., ed) pp. 103-111, Adenine Press, Schenectady, NY
23. Griswold, B. L., Humoller, F. L., and MacIntyre, A. R. (1951) Anal. Chem. 23, 192-194 [CrossRef]
24. Levenson, C., and Chang, C. (1990) in PCR Protocols: a Guide to Methods and Applications (Innis, M. A., Sinsky, D. H., and White, T. J., ed) pp. 99-112, Academic Press, San Diego
25. Scaria, P. V., Shire, S. J., and Shafer, R. H. (1992) Proc. Natl. Acad. Sci. U. S. A. 89, 10336-10340 [Abstract/Free Full Text]
26. Clegg, R. M. (1992) Methods Enzymol 211, 353-388 [Medline] [Order article via Infotrieve]
27. Clegg, R. M., Murchie, A. I., Zechel, A., and Lilley, D. M. (1993) Proc. Natl. Acad. Sci. U. S. A. 90, 2994-2998 [Abstract/Free Full Text]
28. Mergny, J. L., Boutorine, A. S., Garestier, T., Belloc, F., Rougee, M., Bulychev, N. V., Koshkin, A. A., Bourson, J., Lebedev, A. V., Valeur, B., Thuong, N. T., and Helene, C. (1994) Nucleic Acids Res. 22, 920-928 [Abstract/Free Full Text]
29. Cooper, J. P., and Hagerman, P. J. (1990) Biochemistry 29, 9261-9268 [CrossRef][Medline] [Order article via Infotrieve]
30. Marck, C., Thiele, D., Schneider, C., and Guschlbauer, W. (1978) Nucleic Acids Res. 5, 1979-1996 [Abstract/Free Full Text]
31. Thiele, D., Marck, C., Schneider, C., and Guschlbauer, W. (1978) Nucleic Acids Res. 5, 1997-2012 [Abstract/Free Full Text]
32. Chan, S. S., Breslauer, K. J., Austin, R. H., and Hogan, M. E. (1993) Biochemistry 32, 11776-11784 [CrossRef][Medline] [Order article via Infotrieve]
33. Rajagopal, P., and Feigon, J. (1989) Nature 339, 637-640 [CrossRef][Medline] [Order article via Infotrieve]
34. Umemoto, K., Sarma, M. H., Gupta, G., Luo, J., and Sarma, R. H. (1990) J. Am. Chem. Soc. 112, 4539-4545 [CrossRef]
35. Pilch, D. S., Levenson, C., and Shafer, R. H. (1990) Proc. Natl. Acad. Sci. U. S. A. 87, 1942-1946 [Abstract/Free Full Text]
36. Hu, S. H., Weisz, K., James, T. L., and Shafer, R. H. (1992) Eur. J. Biochem. 204, 31-38
37. Jayasena, S. D., and Johnston, B. H. (1992) Biochemistry 31, 320-327 [CrossRef][Medline] [Order article via Infotrieve]
38. Jayasena, S. D., and Johnston, B. H. (1993) Biochemistry 32, 2800-2807 [CrossRef][Medline] [Order article via Infotrieve]
39. Mouscadet, J. F., Ketterle, C., Goulaouic, H., Carteau, S., Subra, F., Le Bret, M., and Auclair, C. (1994) Biochemistry 33, 4187-4196 [CrossRef][Medline] [Order article via Infotrieve]

CiteULike

Complore

Connotea

Del.icio.us

Digg

Reddit

Technorati

This article has been cited by other articles:

				R. A. J. Darby, M. Sollogoub, C. McKeen, L. Brown, A. Risitano, N. Brown, C. Barton, T. Brown, and K. R. Fox High throughput measurement of duplex, triplex and quadruplex melting curves using molecular beacons and a LightCycler Nucleic Acids Res., May 1, 2002; 30(9): e39 - e39. [Abstract] [Full Text] [PDF]

				E. B. Khomyakova, H. Gousset, J. Liquier, T. Huynh-Dinh, C. Gouyette, M. Takahashi, V. L. Florentiev, and E. Taillandier Parallel intramolecular DNA triple helix with G and T bases in the third strand stabilized by Zn2+ ions Nucleic Acids Res., September 15, 2000; 28(18): 3511 - 3516. [Abstract] [Full Text] [PDF]

This Article Abstract Full Text (PDF) Submit a Letter to Editor Alert me when this article is cited Alert me when eLetters are posted Alert me if a correction is posted Citation Map Services Email this article to a friend Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Request Permissions Citing Articles Citing Articles via HighWire Citing Articles via Google Scholar Google Scholar Articles by Scaria, P. V. Articles by Levenson, C. Search for Related Content PubMed PubMed Citation Articles by Scaria, P. V. Articles by Levenson, C. Social Bookmarking                      What's this?