G-Quadruplexes are four-stranded nucleic acid structures typically stabilized by GGGG tetrads. These structures are intrinsically fluorescent, which expands the known scope of nucleic acid function and raises the possibility that they could eventually be used as signaling components in label-free sensors constructed from DNA or RNA. In this study, we systematically investigated the effects of mutations in tetrads, loops, and overhanging nucleotides on the fluorescence intensity and maximum emission wavelength of >500 sequence variants of a reference DNA G-quadruplex. Some of these mutations modestly increased the fluorescence intensity of this G-quadruplex, while others shifted its maximum emission wavelength. Mutations that increased the fluorescence intensity were distinct from those that increased the maximum emission wavelength, suggesting a trade-off between these two biochemical properties. The fluorescence intensity and maximum emission wavelength were also correlated with multimeric state: the most fluorescent G-quadruplexes were monomers, while those with the highest maximum emission wavelengths typically formed dimeric structures. Oligonucleotides containing multiple G-quadruplexes were in some cases more fluorescent than those containing a single G-quadruplex, although this depended on the length and sequence of the spacer linking the G-quadruplexes. These experiments provide new insights into the properties of fluorescent G-quadruplexes and should aid in the development of improved label-free nucleic acid sensors.

Artificial evolution experiments typically use libraries of ∼1015 sequences and require multiple rounds of selection to identify rare variants with a desired activity. Based on the simple structures of some aptamers and nucleic acid enzymes, we hypothesized that functional motifs could be isolated from significantly smaller libraries in a single round of selection followed by high-throughput sequencing. To test this idea, we investigated the catalytic potential of DNA architectures in which twelve or fifteen randomized positions were embedded in a scaffold present in all library members. After incubating in either the presence or absence of lead (which promotes the nonenzymatic cleavage of RNA), library members that cleaved themselves at an RNA linkage were purified by PAGE and characterized by high-throughput sequencing. These selections yielded deoxyribozymes with activities 8- to 30-fold lower than those previously isolated under similar conditions from libraries containing 1014 different sequences, indicating that the disadvantage of using a less diverse pool can be surprisingly small. It was also possible to elucidate the sequence requirements and secondary structures of deoxyribozymes without performing additional experiments. Due to its relative simplicity, we anticipate that this approach will accelerate the discovery of new catalytic DNA and RNA motifs.

G-quadruplexes are noncanonical nucleic acid structures formed by stacked guanine tetrads. They are capable of a range of functions and thought to play widespread biological roles. This diversity raises an important question: what determines the biochemical specificity of G-quadruplex structures? The answer is particularly important from the perspective of biological regulation because genomes can contain hundreds of thousands of G-quadruplexes with a range of functions. Here we analyze the specificity of each sequence in a 496-member library of variants of a reference G-quadruplex with respect to five functions. Our analysis shows that the sequence requirements of G-quadruplexes with these functions are different from one another, with some mutations altering biochemical specificity by orders of magnitude. Mutations in tetrads have larger effects than mutations in loops, and changes in specificity are correlated with changes in multimeric state. To complement our biochemical data we determined the solution structure of a monomeric G-quadruplex from the library. The stacked and accessible tetrads rationalize why monomers tend to promote a model peroxidase reaction and generate fluorescence. Our experiments support a model in which the sequence requirements of G-quadruplexes with different functions are overlapping but distinct. This has implications for biological regulation, bioinformatics, and drug design.