|Home | About | Journals | Submit | Contact Us | Français|
DNA·RNA hybrid duplexes are functionally important structures in gene expression that are underutilized as potential drug targets. Several tools are described here for the discovery and characterization of small molecules capable of the selective recognition of DNA·RNA hybrid structures. Competition dialysis and thermal denaturation of mixtures of polynucleotide structures can be used to identify small molecules that bind selectively to DNA·RNA hybrids. An assay that measures small molecule inhibition of RNase H can be used to measure a functional response to these ligands.
The DNA·RNA hybrid helix was first discovered in 1960, only seven years after the discovery of the famed DNA double helix (1, 2). The hybrid structure was immediately recognized as a key component of genetic information transfer, even though mRNA and tRNA had yet to be discovered (2). After nearly half a century, DNA·RNA hybrid duplexes have yet to be exploited as a target in drug design, despite their unquestioned functional significance. This stands in contrast to other non-canonical nucleic acid structures (e.g. DNA triplexes and quadruplexes, Z-DNA, RNA secondary structures), all of which have been recognized as potential therapeutic targets. The reasons underlying the failure to target the DNA·RNA hybrid are most likely related to uncertainties and misunderstandings about the precise nature of the hybrid duplex structure, along with the lack of available lead ligands (whether of natural or synthetic origin) and, finally, the dearth of convenient assays to assess structural-selective binding to the hybrid.
Structure- and sequence-selective ligands for DNA·RNA hybrids have a variety of potential pharmaceutical applications. For instance, the first-formed product of HIV reverse transcriptase is a DNA·RNA hybrid duplex. Generation of the viral DNA duplex and its integration into the host genome are dependent on digestion of the RNA strand of this hybrid by RNase H (3). Loading DNA·RNA hybrid duplexes with small-molecule ligands inhibits RNase H (4, 5) and thus provides a new route to the control of HIV infection. The tumour-specific enzyme, telomerase is a specialized reverse transcriptase, with protein structure domains closely-related to the HIV enzyme (6). A DNA·RNA hybrid duplex is generated in the course of the telomerase reaction, so telomerase is also amenable to such a small-molecule targeting strategy (7, 8). Indeed, the thermodynamic stability of the DNA-RNA hybrid appears to regulate telomerase activity (9). A long-term goal of nucleic acid-directed drug discovery is the control of gene expression. This may potentially be achieved through sequence-selective intervention in the RNA polymerase products of normal gene expression or binding to the Okazaki fragment RNA·DNA hybrid formed at the initiation of DNA replication (10).
The consensus view of hybrid duplex structures in solution is that the overall helix conformation lies somewhere between the global B-form adopted by pure DNA and the A-form of pure RNA duplexes. The conformation of each strand is driven by the sugar puckers, which in turn are determined by the β-effect of the 3' and 2' substituents of the furanose rings. Thus in the DNA strand, “North” puckers predominate, driving that strand towards a B-conformation. In contrast, for the RNA strand, more heterogeneous sugar puckers are detected (11) resulting in an overall A-like structure. The duplex formed on combination of such diverse strands differs significantly from both pure DNA·DNA and RNA·RNA duplexes (12). The structures of different DNA·RNA hybrids can differ, depending on whether the DNA strand is purine or pyrimidine (13, 14). DNA·RNA hybrids of differing sequence may thus present different backbone conformations and groove geometries, distinctive features that may be selectively targeted by small molecules.
Intuitive attempts to identify ligands for DNA·RNA hybrids using established DNA duplex groove binders or intercalators are perhaps doomed. The grooves of the hybrid are not recognized by the classic minor groove binders distamycin (for examples, see data in Figure 3). Amongst intercalators, matching the planar surface shape and size of a ligand to the cross-sectional dimensions of triplex and quadruplex DNA proved a fruitful approach to the identification of structure-selective binding (15). However, the search for structure selective intercalators of DNA·RNA hybrids is constrained by the similarity of the Watson-Crick base-pairs, and hence intercalation site dimensions, to those of the pure DNA or RNA duplexes. None-the-less some success has been achieved in identifying ethidium as possessing a marked binding preference for the poly(A)·poly(dT) hybrid duplex (16) and re-designing actinomycin to enhance hybrid selectivity (17, 18). The few hybrid binding ligands so far reported are dominated by ethidium and aminoglycoside structures (4, 8, 19, 20) and have recently been the subject of a comprehensive review (19). All of these investigations focus on measuring binding phenomena using spectrophotometric, fluorimetric and calorimetric techniques but offer little rationalization of the inter-molecular interactions that drive the recognition event.
Herein we describe three assays, used in our laboratories, that are useful in exploring the structure and sequence binding preferences of small molecules that target DNA·RNA hybrid duplexes. Competition equilibrium dialysis (21) delivers a detailed and thermodynamically-rigorous profile of the binding strengths and preferences of an investigational ligand across a broad range of up to 48 different nucleic acid structures and sequences (22–25). The assay in its current form typically includes both poly(dA)·poly(U) and poly(A)·poly(dT) hybrids alongside the counterpart pure DNA and pure RNA duplexes. The melting of mixtures assay (26) is an extension of the well-established DNA thermal denaturation assay (27, 28) that detects the effects of added ligand on thermal denaturation of four duplexes simultaneously. It is valuable for rapidly assessing binding preferences amongst DNA, RNA and hybrid duplexes; for the hybrid systems it also provides evidence of selectivity between dA·U and A·dT sequences. Finally a RNase H assay is included that provides a pharmaceutically-relevant biological endpoint for studies of ligands for DNA·RNA hybrids. Various protocols for the detection and quantitation of telomerase inhibition are described in the literature (29–31).
All three assays described here are UV spectrophotometric methods, directly detecting either ligand or nucleic acid without the use of radio-labels or fluorescent tags. This latter is particularly important since the presence of large aromatic groups, typical of fluorophores and quenchers, will itself alter the nucleic acid structures and thereby ligand binding events.
The competition dialysis assay is a simple extension of equilibrium dialysis binding assays (32, 33). In the competition dialysis assay, an array of nucleic acid structures is dialyzed against a common test ligand ("drug") solution. Each structure in the array is at an identical concentration and is contained within its own dialysis unit, isolated from all other structures. All structures are in contact with the same ligand solution. At equilibrium, the free ligand concentration is identical for all structures in the array, but each structure will bind ligand according to its affinity for the ligand. The difference in the amount of ligand bound to each structure is proportional to the association constant of the ligand for that structure. A simple bar graph reveals at a glance the sequence or structural preferences of the ligand. Figure 1 shows a representative result for the DNA intercalator ethidium. In this case, binding to an array of 13 nucleic acid structures reveals that ethidium binds with a slight, but significance, preference to a DNA·RNA hybrid structure.
Competition dialysis is particularly powerful when examining a library of small molecules. In such a case, competition dialysis difference plots can be used to highlight those ligands that bind with high affinity and selectivity to a particular structure or sequence of interest (34, 35). Figure 2 shows an example in which ligands that bind to a DNA·RNA hybrid structure were sought. In this case, the binding of a library of 126 nucleic acid binding compounds to an array of 13 nucleic acid structures and sequences was studied. Figure 2 shows three difference plots that emphasize difference binding to a DNA·RNA hybrid structure, to a standard DNA duplex, and to an RNA duplex. Four compounds stand out as having higher than average affinity for the DNA·RNA hybrid, while showing clear preferential binding to the hybrid over duplex DNA or RNA. These compounds are ellipticine (III), the topoisomerase poison TAS103 (I), thiazole orange (II), and ethidium (IV). These compounds are all DNA intercalators, and while they show modest preferential binding to the DNA·RNA hybrid, the intercalator motif may not provide optimal lead compounds for more stringent targeting of hybrid structures and sequences. The primary point of Figure 2 is to illustrate the power of competition dialysis for screening libraries to reveal binding preferences.
Detailed, step-by-step protocols for the competition dialysis assay and for the analysis of competition dialysis data have been presented in several publications (21, 22, 25, 34, 35). These protocols will not be repeated here, and the interested investigator should consult the published protocols. Competition dialysis is an important element of the toolbox for the discovery of DNA·RNA hybrid binders, as this brief discussion indicates.
Thermal denaturation is a powerful tool for monitoring binding to nucleic acids (27, 28, 36, 37). Ligands that bind to structured nucleic acids stabilize the ordered structure and elevate the melting temperature for the denaturation of that structure. The magnitude of the change in melting temperature is a complex function of binding affinity and stoichometry (28). Qualitatively, an increase in the melting temperature for a nucleic acids structure provides unambiguous evidence for binding. A recent extension of the thermal denaturation assay resulted from the simple expediency of making a mixture of different nucleic acid structures that could be distinguished by their melting temperatures(26). Addition of low binding ratios of a test ligand altered the melting temperature of the nucleic acid structure to which it bound most avidly, providing a simple, rapid evaluation of its structural preference. A mixture containing DNA, RNA, and two DNA·RNA hybrid forms was described and tested (26), and a protocol for that preparation and use of that mixture will be described here.
Example melting of mixtures data are presented for established DNA minor groove binders (Figure 3) and a family of analogues of the biarylpyrimidine 1a (Figure 4, panel A.) These demonstrate the value of this assay in rapidly and simultaneously assessing thermal stabilization and selectivity amongst the four nucleic acid structures. Thus although analogue 1d showed the greatest ΔTm for the poly(dA)·poly(U) duplex it was the analogue 1b that showed the most selectivity between the structures, with minimal shifting of the DNA duplex.
The data of Figure 4 also show a direct correlation between the ligand-induced ΔTms of the poly(dA)·poly(U) duplex (panel A, inset) and the potency of RNase H inhibition (panel B). The rank order of ΔTms is the same as the rank order as inhibitors of RNase H digestion of that hybrid duplex. Thus there is a strong correlation between strength of the ligand–hybrid interaction and RNase H inhibitor potency; this means that the relatively quick and inexpensive mixed melting assay may be used as a surrogate for the more expensive and time-consuming RNase H assay.
This spectrophotometric assay is based on that previously described by Rasche et al (38). The release of nucleotides from the hybrid duplex substrate is followed as an increase in A260. The overall ΔA260 change detected for complete digestion of the RNA strand of a hybrid duplex approaches that observed following loss of the duplex secondary structure in the UV thermal melting experiment. This is a direct assay that monitors disappearance of substrate: the signal is independent of the inhibitor under investigation. The scale on which the reaction is performed (enzyme and substrate concentrations) is determined by the sensitivity of the UV spectrophotometer (Note 1) and there is no need for either radioactive labelling or the introduction of fluorescent tags.
Solutions of the hybrids 10 µM bp were found to give a final ΔA260 ≈ 0.06, appropriate for monitoring the reaction. Typical time courses for the digestion of poly(dA)·poly(rU) and poly(A)·poly(dT) are shown in Figure 5 panel A. A burst phase is apparent at the start of the reaction which is typical of “ping-pong” kinetics involving a covalently-linked enzyme-substrate intermediate. This phase was usually flowed by a linear phase from which the Vinital was estimated as the gradient of a best fit straight line; at high inhibitor concentrations, curvature of the line could complicate estimation of this gradient.
As with any enzyme reaction, it was first necessary to determine the KM of the enzyme for the specific substrate under investigation. Figure 5 panel B shows the dependence of Vmax on [poly(dA)·poly(rU)] from which KM was determined (KM= 9.99±1.39 µM for poly(dA)·poly(rU)). Subsequent kinetic runs were performed with [S] = 4KM to ensure maximal velocity (39). Formally, the inhibitor strength should be measured as the Ki. However, this requires repetitions of the assay at ranges of [I] and [S] which may become expensive in both time and the quantity of enzyme consumed; for most purposes it is adequate to measure the IC50. Data obtained for a range of concentrations of the most potent inhibitor, compound 1d, and the curve-fit for the determination of the IC50 (6.53±0.03 µM) are shown in Figure 5 panel C.
The RNase H assay can be used in a survey mode to screen series of compounds at a fixed concentration in order to select the most potent inhibitors for detailed investigation. Figure 4, Panel B shows data for a range of analogues of biarylpyrimidine 1a inhibiting digestion of the poly(dA)·poly(U) hybrid. Significantly, the rank order of potency of inhibition thus detected followed the magnitude of the ΔTm shifts detected in the in the melting of mixtures assay (Figure 4, panel A) even though this melting assay was performed at non-saturating concentrations of ligand using a different buffer system.
Materials for the competition dialysis assay in a variety of different formats containing arrays of 13, 19 and 48 nucleic acid structures have been presented in several publications (21, 22, 25, 34, 35)
This work supported by award number R01GM077422 from the National Institute of General Medical Sciences. The content is solely the responsibility of the authors and does not necessarily represent the official vies of the NIGMS or NIH.
1The assays described herein were performed using a CARY400Bio spectrophotometer equipped with a multi cell (6+6) block and Peltier heating control. A matched par of Hellma masked quartz cuvettes (114-QS) was used. The test solution cuvette was placed in the front position and buffer in the back (or nothing where relative absorbance data were required); sample volumes were 1.2 ml. The temperature was monitored using a temperature probe placed in a third cuvette, containing buffer, placed in the position adjacent to the sample cuvette in the block.
2Attempts at dilution of the enzyme stock into reaction buffer prior to use resulted in loss of activity. The enzyme preparation is sufficiently fluid at room temperature to be pipetted into the reaction cuvette. Thorough mixing by gentle inversion 12 times initiates the reaction.
3Tris buffer was used as made up from the supplied solid; the pH was not checked or adjusted.
4Precision digital pipettes were used for all test compound and nucleic acid measurements.
5Before use, each of the four duplexes should be melted individually for quality assurance. The absorbance change observed for each duplex is about the same.
6It is critical that the cuvettes are cleaned thoroughly: during the KM study, cuvettes were soaked for 3–6 h in 2% Hellmanex solution between reactions.
7From the total absorbance change at “infinity” (about 6.5 h for poly(dA)·poly(U)), and assuming complete release of rU nucleotides from the hybrid, the absorbance data could be scaled to nmoles of rU released (1 Abs unit 287.08 nmoles rU) and hence rate data converted to pmol(rU) s−1.