| Nucleic Acids Research | Pages |
A search for base analogs to enhance third-strand binding to `inverted' target base pairs of triplexes in the pyrimidine/parallel motif
Introduction
Materials And Methods
Deoxyoligonucleotides
Triplex mixtures
UV melting
Results And Discussion
Effect of the Z:A·T test triplet
Effect of the Z:G·C test triplet
Effect of the test triplets with the `inverted' base pairs Z:T·A and Z:C·G
Specificity of triplex stabilization by some analogs
Conclusions
Acknowledgements
References
A search for base analogs to enhance third-strand binding to `inverted' target base pairs of triplexes in the pyrimidine/parallel motif
Received July 14, 1999; Revised and Accepted October 4, 1999
ABSTRACT Eight base analogs were tested as third strand residues in otherwise homopyrimidine strands opposite each of the `direct' (A·T and G·C) and `inverted' (T·A and C·G) Watson-Crick base pairs, using UV melting profiles to assess triplex stability. The target duplexes contained 20 A·T base pairs and a central test base pair X·Y, while the third strand contained 20 T residues and a central Z test base. Z included 5-bromo-uracil, 5-propynyluracil, 5-propynylcytosine, 5-methyl-cytosine, 5-bromocytosine, hypoxanthine, 2-amino-purine and 2,6-diaminopurine. Some of the base analogs enhanced third strand binding to the target duplex with one or other `inverted' central base pair relative to the binding afforded by any of the canonical bases. Other analogs did the same for the duplexes with the `direct' target pairs. The increasing order of triplex stabilization by these base analogs is: opposite the `inverted' base pairs, for T·A, A < C < 5-pC < 5-pU < T < 5-BrC < 5-meC < 5-BrU < 2-AP < 2,6-DAP < Hy < G, for C·G, 2-AP < A < Hy < G < 5-pC < 5-BrC < 5-meC < C < 2,6-DAP < T < 5-BrU < 5-pU; opposite the `direct' base pairs, for A·T, 2-AP < A < 5-meC < C < G < Hy < 2,6-DAP < 5-pU < T = 5-BrU < 5-BrC < 5-pC, for G·C, G < 2,6-DAP < 2-AP < A < Hy < T < 5-BrU < 5-pU < 5-pC < 5-BrC < C < 5-meC.
INTRODUCTION
Intermolecular DNA triplex formation by sequence-specific triplex-forming oligonucleotides is of considerable interest because of its great potential for a variety of analytical and therapeutic applications, including among others selective control of gene expression (1-3), site-directed mutagenesis (4,5) and gene repair (6), and in situ chromosome and other molecular diagnostics (7). Currently, the major limitation to the application of third strand binding is the requirement for the target sequence to be homopurine·homopyrimidine (8), as interruptions of such a sequence pattern by `inverted' base pairs can drastically reduce binding affinity (9,10). One of the ways to circumvent this problem is to discover synthetic base analogs that bind strongly to inverted base pairs, or at least minimize triplex destabilization by them.
In the present work, we have systematically compared eight base analogs (Fig. 1) in an otherwise homopyrimidine third strand with respect to their affinity and specificity for all four canonical target base pairs in an A·T target host duplex. The analogs were selected in the hope of finding bases better disposed to bind to inverted base pairs than the canonical ones already known, and developing insights regarding the alternative ways such interactions might be favored. For this purpose, the deoxyoligomers A10XA10 and T10YT10 were used to form a target DNA duplex with each of the four possible X·Y central target base pairs A·T, T·A, G·C or C·G, and T10ZT10 was used as a third strand, where Z included 5-bromouracil (5-BrU), 5-propynyl-uracil (5-pU), 5-propynylcytosine (5-pC), 5-methylcytosine (5-meC), 5-bromocytosine (5-BrC), hypoxanthine (Hy), 2-amino-purine (2-AP) and 2,6-diaminopurine (2,6-DAP).
Figure 1. Structures of base analogs tested as third strand residues opposite each of the four possible target base pairs (A·T, T·A, G·C and C·G).
MATERIALS AND METHODS
Deoxyoligonucleotides
Phosphoramidites of all the analogs were obtained from Glenn Research (Sterling, VA). All oligomers were synthesized by the phosphoramidite method, deprotected and purified by denaturing PAGE (16%). Bands were visualized by UV light, eluted with 10 mM Tris-HCl, 2 mM EDTA, pH 7.0, and desalted by C18 reverse phase chromatography. Purity was ascertained by denaturing PAGE of 32P-end-labeled oligomers. The concentrations of all strands were estimated using either the molar extinction coefficients for poly(dA) at 25°C in double distilled H2O of [epsiv]257 = 8600 or for poly(dT) of [epsiv]265 = 8700.
Triplex mixtures
Equimolar amounts of duplex strands were mixed in the standard solvent, 0.15 M NaCl, 0.005 M MgCl2, 0.01 M cacodylate (Na+), pH 7.0, heated to 75°C and slowly cooled to room temperature, at which point a stoichiometric amount of the third strand was added and allowed to anneal overnight at 4°C.
UV melting
Absorbance-temperature profiles were obtained and the triplex Tm values determined as described in Fossella et al. (11). Absorption spectra were recorded every 2°C between 0 and 70°C, with samples equilibrated for 10 min at each temperature prior to recording the spectrum. Profiles were plotted at multiple wavelengths, and left uncorrected for dilution due to thermal expansion at elevated temperatures. The largest percentage hyperchromic change for the triplex->duplex + third strand transition is at 280 nm, so the data at that wavelength are more accurate, and were used to obtain Tm values. Tm values were from both the midpoint of the triplex melting transition and the maximum of dA280/dT (from the first derivative profile); the values from duplicate experiments invariably coincided within experimental error (±0.1°C). This error was estimated from the Tm values of the same duplex for a series of melting profiles (data not shown). If the triplex melting transition was already in progress at 0°C, it was extrapolated to lower temperatures by using the characteristic shape of these melting profiles and assuming the same hyperchromic change for all triplex transitions. The absence of significant variation in duplex Tm values confirms that no strand exchange occurred between the duplex strand T10YT10 and the third strand T10ZT10.
RESULTS AND DISCUSSION
Melting for all triplex mixtures in the standard solvent displayed two major transitions, that at lower temperature representing the dissociation of the triplex to duplex + single (third) strand, and the second transition at higher temperature corresponding to dissociation of the duplex (Fig. 2) (10). Table 1 lists the Tm values for the transitions displayed by each triplex mixture, denoted by its test Z:X·Y combination, and the transition breadths ([Delta]T).
Figure 2. Melting of triplex with the 5-meC:G·C test triplet. (Top) UV melting profile at 260 (--) and 280 nm (- - -). (Bottom) Derivative profile at 260 (--) and 280 nm (- - -). Note that the small transition with Tm ~17°C probably represents the melting of a small amount of excess homopyrimidine strand forming a homoduplex in the Mg2+-containing solvent.
Table 1. Tm values (°C) and transition breadths ([Delta]T) for biphasic melting of triplexes with various test triplets
Transition breadth is the width at the half-height of the first derivative of the transition.
Effect of the Z:A·T test triplet
The effect of different Z:A·T test triplets on triplex stability varies widely, with Tm values ranging from -1.5 to 29.4°C. With the canonical test triplet T:A·T resulting in a triplex with a Tm value of 24.7°C, it is interesting that two pyrimidine analogs, 5-pC > 5-BrC, result in greater stabilization, while another, 5-BrU, stabilizes comparably. On the other hand, 5-pU << 2,6-DAP < Hy << 5-meC < 2-AP are increasingly destabilizing. Not surprisingly, 2,6-DAP, Hy and 2-AP are destabilizing, probably for steric reasons, but the effect of 5-meC, in contrast to significant stabilization caused by the similar 5-pC, is puzzling. So is the fact that the 5-substituted C analogs 5-pC and 5-BrC are more stabilizing than the U analogs. It can also be seen that triplex destabilization correlates with a decrease in cooperativity of the melting transition, reflected by a larger width at the half-height of the melting profile (Table 1).
Effect of the Z:G·C test triplet
The same range of Z residues results in Z:G·C combinations with a rather narrower range of stability. Interestingly, C:G·C is the most stabilizing canonical triplet in the same third strand environment (11), even though the third strand C requires protonation nearly 3 pH units above its intrinsic pK value (12). It is not surprising, therefore, that the three most stabilizing analogs are 5-meC > 5-BrC > 5-pC, all of which have higher pK values than C. The U derivatives 5-BrU and 5-pU cause similar moderate destabilization, probably because they are comparably somewhat sterically disruptive. The purine analogs 2-AP and 2,6-DAP, on the other hand, destabilize rather more, probably because their stacking tendency is not enough to compensate for their steric disruption of the pyrimidine triplex motif. No cooperativity trends are apparent in this case. However, it is interesting that G·C is a more `accommodating' target base pair than A·T, as none of the Z:G·C triplets drastically destabilize the triplex as do some of the Z:A·T triplets (Table 1).
Effect of the test triplets with the `inverted' base pairs Z:T·A and Z:C·G
The inverted target base pair T·A remains difficult to stabilize in a triplex. Thus, none of the tested analogs proved to be better than G (9-11), with Hy a close second: G > Hy >> 2,6-DAP > 2-AP > 5-BrU > 5-meC > 5-BrC > 5-pU > 5-pC. The rank order of stabilizing effectiveness gives no real insight into its origin. Also, there is no correlation of Tm with the cooperativity of the transition.
For the C·G base pair, the most effective test residue is 5-pU, followed closely by 5-BrU and T: 5-pU > 5-BrU > T >> 2,6-DAP >> 5-meC > 5-BrC > 5-pC >> Hy >> 2-AP. Again, there is no correlation between stability and cooperativity. In this case, then, two analogs prove to be more effective than the best canonical base.
Specificity of triplex stabilization by some analogs
Base analogs can stabilize third strand binding either by direct improvement of binding to an inverted base pair or, alternatively, by increasing the stability of triplets with `direct' targets, thereby counteracting the destabilizing effect of the inverted base pairs. For example, the G:T·A test triplet results in a Tm of 17.6°C, which is only 7°C lower than the Tm for the `conventional' T:A·T triplet. This 7°C destabilization can be counteracted by using 5-pC in place of T in a 5-pC:A·T triplet (relative stabilization 5°C). However, there is a price to pay in terms of specificity for such stabilization, as U and C analogs bind almost equally well to A·T and G·C base pairs, whereas C is highly specific for the G·C base pair, as is T for the A·T base pair (Table 1).
Conclusions
Efforts to overcome sequence limitations of triplex-based approaches have ranged from the synthesis of new base analogs and sequence-specific intercalating agents (13-15) to the tethering of stabilizing ligands to a triplex-forming oligonucleotide (16). However, for therapeutic third strand binding applications in vivo, simple purine and pyrimidine derivatives are much less likely to engage in unforeseen interactions with cell components. In that respect, the use of a combination of the most stabilizing (or least destabilizing) simple base analogs can enable sufficiently stable third strand binding even to duplex target sequences with inverted base pairs (6,17).
We had hoped that the relative tendencies of the various base analogs to stabilize or destabilize the host triplex when opposite particular target base pairs in the test position would provide some obvious insight regarding the interactions that are critical for that binding. But no obvious rationale has emerged. Thus, for example, it is no more apparent why the 5-propynyl substituent on C opposite A·T or on U opposite C·G enhances third strand binding than why this substituent on C or T substantially enhances antisense DNA strand binding to mRNA sequences via duplex formation (18,19). While it is disappointing that the observed effects, which are highly reproducible, cannot be ascribed in a simple way to some combination of hydrogen bonding, van der Waals contacts and nearest neighbor stacking interactions, it is at least apparent that steric interference with regular triplex continuity is a destabilizing effect. Moreover, the knowledge of these effects is a requisite for uncovering their rationale.
ACKNOWLEDGEMENTS
We are grateful to Jay George for providing the oligomers with base analogs. We also thank Dmitry Klimov for assistance with figure preparation. This work was supported by grants from the NIH (GM42936) and the DOE (DE-FG02-96ER62202) and by a fellowship from Codon Pharmaceuticals to O.A.
REFERENCES
This is paper no. 28 in the series entitled `Polynucleotides', of which the last was Amosova,O., George,J. and Fresco,J.R. (1997) Nucleic Acids Res., 25, 1930-1934. MEDLINE Abstract
*To whom correspondence should be addressed. Tel: +1 609 258 3927; Fax: +1 609 258 1028; Email: jrfresco{at}princeton.edu
This page is run by Oxford University Press, Great Clarendon Street, Oxford OX2 6DP, as part of the OUP Journals
Comments and feedback: jnl.info{at}oup.co.uk
Last modification:
Copyright© Oxford University Press, 1999.
CiteULike 
Connotea 
Del.icio.us What's this?
This article has been cited by other articles:
|
|
 |
|
  S. P. Parel and C. J. Leumann Triple-helix formation in the antiparallel binding motif of oligodeoxynucleotides containing N9- and N7-2-aminopurine deoxynucleosides Nucleic Acids Res., June 1, 2001; 29(11): 2260 - 2267. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||