Nucleotide replacement at two sites can be directed by modified single-stranded oligonucleotides in vitro and in vivo

Abstract

Studies involving the alteration of DNA sequences by modified single-stranded oligonucleotides in vitro and in vivo have revealed potential applications for functional genomics. Repair of a replacement, deletion, or insertion mutation has already been achieved with molecules having lengths between 25 and 74 bases. But, other vector parameters still remain to be explored. Here, the position of the single base in the vector directing the alteration was examined and the optimal site was found to be at or near the center of the vector. If that position is staggered 3′ or 5′, the frequencies of gene repair in vitro decreases. The potential of a single vector to direct two nucleotide changes at a specific site in a target sequence was also examined. Both targeted bases are corrected together at the same frequency if the sites are separated by three bases, but conversion linkage decreases precipitously when the distance is expanded to 15 and 27 nucleotides, respectively. These results suggest that single oligonucleotides can be used to direct nucleotide exchange at two independent sites, a reaction characteristic that may be useful for many genomics applications.

Introduction

A technology that could alter the genome at precise locations within the coding region of a gene would be a powerful tool for functional genomics. Current techniques for gene modification or disruption, while effective, usually require a significant amount of preparatory work or are limited by DNA sequence restrictions. The former problem is most often encountered in the generation of “knock-out” models where the aim is to disable the expression of a specific gene. And, the preparation of the targeting molecules is dependent on the availability of cloning sites and size limitations. The latter problem is best exemplified by the genomic alteration vector—the triplex-forming oligonucleotide (TFO). While these small synthetic vectors have advanced the field and are clearly a simplification of the overall process and their activity is restricted by binding dynamics to certain nucleotide arrays in the genome, reducing their overall utility [1]. The TFO strategy did, however, demonstrate that nucleic acid modification need not rely on complex vectors, but rather could occur in a facile manner and generate a wide array of genetic mutations, such as knock-outs, knock-downs, and altered codons. Taken together then, the ideal vector would be a synthetic molecule that is simple to design and produce, but that can alter specific single nucleotides with high fidelity without promoting secondary events.

One experimental strategy developed for such a site-specific effect employs a chimeric RNA/DNA oligonucleotide. The design of this molecule was prompted by the discovery that RNA–DNA hybrids are highly active in recombination assays where the length of homology between pairing partners is reduced [2], [3]. The chimera is a self-complementary oligonucleotide that folds into a double-hairpin configuration that can act as a template for gene repair of a point mutation (see [4] for review). It contains 2′-O-methyl RNA, a modification that makes the chimera resistant to RNAseH activity. The chimera also features hairpin-capped ends, thus avoiding destabilization or destruction by helicases or exonucleases, and making its half-life in cells or serum substantially longer than uncapped duplexes. The molecule is designed such that it aligns in perfect register with specific DNA sequences in the target gene, except for a single base, creating a mismatch at a unique site. The structural distortion created by the mismatched base pair is envisioned to attract the endogenous DNA repair systems, resulting in a change in sequence in the gene. While undergoing the initial DNA pairing phase of the process, RNA stabilizes reaction intermediates and, due to the increase in stability, the chimera–gene complex enables gene conversion at an elevated frequency. The functionality of this chimeric molecule has been shown in mammalian [5], [6], [7], [8], [9], plant [10], [11], [12], and yeast [4], [13] systems, as well as in animal models [14], [15], [16], [17] and cell-free extracts [18]. The extra stability conferred by the RNA portion of the chimera on the reaction intermediate undergoing the mutagenesis or repair is likely to have lead to the elevated level of base exchange. Several other techniques have also been used to modify nucleotides or reverse mutation. One of them, short fragment homologous recombination [19] relies on amplification of PCR products and the integration of this population into a homologous site in the genome. If this technique can be refined, it might become quite useful for gene replacement. Presently, however, the integrative vector is undefined and the mechanism of integration unknown.

In an attempt to simplify the design of the synthetic oligonucleotide vector, as well as to elucidate the biologically active component of the chimeric oligonucleotide without a loss of specificity, systematic dissection of the chimera structure was undertaken [20]. After testing each section of the intact chimera separately, single-stranded molecules containing RNA were found to be devoid of significant gene conversion activity, while the unmodified DNA single-stranded molecule displayed only 20% of the activity of the chimera. The lower rates of repair in both cases could be attributable to a rapid (perhaps partial) degradation of the single-stranded molecules. To overcome this problem, we have designed a new generation of vectors using modified linkages between the terminal residues, which now resist nuclease digestion [20].

These changes resulted in a substantial increase in the efficiency with which nucleotide modification took place. The current (“best”) repair vector is a 74-mer with the three terminal linkages comprised of phosphorothioate bonds. It is designed to be perfectly homologous to the target gene sequence, with the exception of a single, centrally placed base, which upon binding to the region of homology in the target forms a single D-loop structure containing a mismatched, base pair. The resulting distortion attracts the repair machinery, which, in turn, can catalyze base exchange at the target site [21]. Most of the information elucidating this mechanism has come primarily from studies carried out in the true yeast, Saccharomyces cerevisiae [4], [13], [22], [23]. This genetically tractable organism provides, in the long run, an excellent model on which to build and test functional genomics strategies for higher eukaryotes. But, in the short-term, the development of this oligonucleotide approach has intrinsic merit for modifying the genomes of lower eukaryotes. In these studies, oligonucleotides of lengths varying from 25 to 74 nucleotides were found to be active in the repair of both replacement (point) and frameshift mutations [23]. Similar results were obtained using unmodified single-stranded DNA molecules ([22] and references therein).

Since structural alterations have led to improved vector activity, it seemed prudent to examine the importance of mismatch position and the capacity to engineer multiple base exchanges at adjacent or juxtaposed nucleotides. Multiple base alterations in the same gene can assure that genes targeted for knock-out are irreversibly disabled. In this study, we define the optimal position for base mispairing between the vector and the target, and demonstrate, for the first time, that multiple mutations can be created in the same gene, directed by a single vector.