Altered biochemical specificity of G-quadruplexes with mutated tetrads

Abstract

A fundamental motif in canonical nucleic acid structure is the base pair. Mutations that disrupt base pairs are typically destabilizing, but stability can often be restored by a second mutation that replaces the original base pair with an isosteric variant. Such concerted changes are a way to identify helical regions in secondary structures and to identify new functional motifs in sequenced genomes. In principle, such analysis can be extended to non-canonical nucleic acid structures, but this approach has not been utilized because the sequence requirements of such structures are not well understood. Here we investigate the sequence requirements of a G-quadruplex that can both bind GTP and promote peroxidase reactions. Characterization of all 256 variants of the central tetrad in this structure indicates that certain mutations can compensate for canonical G-G-G-G tetrads in the context of both GTP-binding and peroxidase activity. Furthermore, the sequence requirements of these two motifs are significantly different, indicating that tetrad sequence plays a role in determining the biochemical specificity of G-quadruplex activity. Our results provide insight into the sequence requirements of G-quadruplexes, and should facilitate the analysis of such motifs in sequenced genomes.

Figure 1.

Comparative sequence analysis of duplex and G-quadruplex structures. (A) Chemical structures of G-C and A-T base pairs. (B) Hypothetical sequence alignment of an evolutionary conserved hairpin. Covariations in the alignment are shown in orange. (C) Chemical structure of a G-G-G-G tetrad. (D) Hypothetical sequence alignment of an evolutionary conserved G-quadruplex. Covariations in the alignment based on those identified in this work are shown in orange.

Figure 2.

Mutagenesis of a tetrad in a DNA G-quadruplex with GTP-binding and peroxidase activity. (A) Topological isoforms of G-quadruplexes formed from different combinations of parallel and antiparallel strands. (B) Primary sequence and possible structure of the reference construct used in these experiments. Mutated positions in the central tetrad are numbered. Note that our experiments do not indicate which strands are next to one another in this structure. (C) Peroxidase activity of the reference construct using the substrate ABTS. Solid orange curve = the most active previously described construct (ATTGGGAGGGATTGGGTGGG); solid blue curve = the reference construct; dotted orange curve = a 17 nucleotide random sequence pool. (D) Circular dichroism spectrum of the reference construct. DA = differential absorption. (E) Retention of the reference construct in a gel filtration column compared to single and double-stranded markers. Random sequence = a 17 nucleotide random sequence pool; single stranded = a randomly generated, 17 nucleotide oligonucleotide with the sequence GACTGCCTCGTCACGAT; double stranded = a mix of the single-stranded oligonucleotide GACTGCCTCGTCACGAT and its reverse complement. The experiment in panel (C) was performed at 10 μM DNA concentration in a buffer containing 200 mM KCl, 1 mM MgCl₂, 20 mM HEPES pH 8, 0.05% Triton X-100, 0.5 μM hemin, 1% DMSO, 5 mM ABTS, and 600 μM H₂O₂. Experiments in panels (D) and (E) were performed at 10 μM DNA concentration in a buffer containing 200 mM KCl, 1 mM MgCl₂, 20 mM HEPES pH 7.1, and 10 nM unlabeled GTP. In panel (E), reported values represent the average of three experiments and error bars indicate one standard deviation.

Figure 3.

Sequence requirements of G-quadruplexes with GTP-binding and peroxidase activity. (A) Measuring the GTP-binding activity of G-quadruplex variants by gel filtration. (B) Heat map showing the GTP-binding activity of all possible variants of the central tetrad in the reference sequence. (C) Measuring the peroxidase activity of G-quadruplex variants using the substrate ABTS. (D) Heat map showing the peroxidase activity of all possible variants of the central tetrad in the reference sequence. All GTP-binding assays were performed at 10 μM DNA concentration in a buffer containing 200 mM KCl, 1 mM MgCl₂, 20 mM HEPES pH 7.1, and 10 nM ³²P-γ-GTP. All peroxidase assays were performed at 10 μM DNA concentration in a buffer containing 200 mM KCl, 1 mM MgCl₂, 20 mM HEPES pH 8, 0.05% Triton X-100, 0.5 μM hemin, 1% DMSO, 5 mM ABTS, and 600 μM H₂O₂. Reported values in panels (B) and (D) represent the GTP-binding or peroxidase activity obtained from a single experiment. All measurements >20% of the value of the reference construct in this screen were repeated two more times, and those with an average value > 20% of the reference construct are listed in Tables 1 and 2.

Figure 4.

Correlated mutational effects in G-quadruplexes with GTP-binding and peroxidase activity. (A) GTP-binding activity of all possible single mutant variants of the reference construct. (B–D) GTP-binding activity of all possible double, triple, and quadruple mutant variants of the reference construct. (E) Peroxidase activity of all possible single mutant variants of the reference construct. (F–H) Peroxidase activity of all possible double, triple and quadruple mutant variants of the reference construct. In panels (B–D) and (F–H), the solid blue line indicates the expected relationship for independent mutational effects. All experiments in panels (A–D) were performed at 10 μM DNA concentration in a buffer containing 200 mM KCl, 1 mM MgCl₂, 20 mM HEPES pH 7.1, and 10 nM ³²P-γ-GTP. All experiments in panels (E–H) were performed at 10 μM DNA concentration in a buffer containing 200 mM KCl, 1 mM MgCl₂, 20 mM HEPES pH 8, 0.05% Triton X-100, 0.5 μM hemin, 1% DMSO, 5 mM ABTS, and 600 μM H₂O₂. In panels (A) and (E), reported values represent the average of three experiments and error bars indicate one standard deviation. In panels (B–D) and (F-H), reported values represent the GTP-binding or peroxidase activity obtained from a single experiment. In all panels, GTP-binding and peroxidase activity is expressed relative to that of the reference construct.

Figure 5.

Compensatory mutations in G-quadruplex structures. (A) Overview of the strategy used to identify compensatory mutations. Variants containing such mutations were chosen based on both activity (which was required to be >20% of that of the reference construct) and extent of rescue (which was required to be at least 5-fold higher than expected based on multiplying single mutation effects). (B) Expected and observed GTP-binding activity for the G-G-G-G to G-A-T-T compensatory change. In each case, expected GTP-binding activity was calculated by multiplying the effects of the mutations that make up the G-G-G-G to G-A-T-T mutant, which are indicated below each pair of measurements. Observed activity is the experimentally measured GTP-binding activity of the indicated mutant. (C) Expected and observed peroxidase activity for the G-G-G-G to G-G-C-C and G-G-G-G to G-G-T-C compensatory changes. In each case, expected peroxidase activity was calculated by multiplying the effects of the individual mutations that make up the G-G-G-G to G-G-C-C or G-G-G-G to G-G-T-C mutant, which are indicated below each pair of measurements. Observed activity is the experimentally measured peroxidase activity of the indicated mutant. Experiments in panel (B) were performed at 10 μM DNA concentration in a buffer containing 200 mM KCl, 1 mM MgCl₂, 20 mM HEPES pH 7.1, and 10 nM ³²P-γ-GTP. Experiments in panel (C) were performed at 10 μM DNA concentration in a buffer containing 200 mM KCl, 1 mM MgCl₂, 20 mM HEPES pH 8, 0.05% Triton X-100, 0.5 μM hemin, 1% DMSO, 5 mM ABTS, and 600 μM H₂O₂. In all panels, GTP-binding and peroxidase activity is expressed relative to that of the reference construct. Reported values represent the average of three experiments and error bars indicate one standard deviation. For expected GTP-binding and peroxidase activity, standard deviations were calculated using standard methods of propagation of error.

Figure 6.

The circular dichroism spectra of single mutant variants of the reference construct are similar to those of parallel strand G-quadruplexes. (A) GTP-binding activity of all single mutant variants of the reference construct. Measurements were performed as described in the legend to Figure 3. (B) Peroxidase activity of all single mutant variants of the reference construct. Measurements were performed as described in the legend to Figure 3. (C) Circular dichroism spectra of all single mutant variants of the reference construct. In each graph, the blue curve represents the circular dichroism spectrum of the reference construct, and the orange curve represents the circular dichroism spectrum of the indicated mutant. DA = differential absorption. All experiments in panel (C) were performed at 10 μM DNA concentration in a buffer containing 200 mM KCl, 1 mM MgCl₂, 20 mM HEPES pH 7.1 and 10 nM unlabeled GTP.

Figure 7.

Sequence models for G-quadruplexes that bind GTP and promote peroxidase reactions. (A) Comparing the effects of mutations in the 5’, central and 3’ tetrad on the ability of the reference construct to bind GTP. The color scheme used to indicate the activity of each mutant matches that in Figure 3. (B) Sequence model for G-quadruplexes that bind GTP. 5’ tetrad = positions 1, 5, 10 and 14; central tetrad = positions 2, 6, 11 and 15; 3’ tetrad = positions 3, 7, 12 and 16; spacer nucleotides = 4, 8, 9, 13 and 17. (C) Expected and observed activities of 10 randomly chosen sequences that satisfy the requirements of our model. The blue line indicates the relationship if expected and observed activities were identical, and was calculated by assuming that spacer sequence does not influence activity. GTP-binding activity is expressed relative to that of the reference construct. Experiments were performed at 10 μM DNA concentration in a buffer containing 200 mM KCl, 1 mM MgCl₂, 20 mM HEPES pH 7.1 and 10 nM ³²P-γ-GTP. (D) Comparing the effects of mutations in the 5’, central and 3’ tetrad on the ability of the reference construct to promote peroxidase reactions. The color scheme used to indicate the activity of each mutant matches that in Figure 3. (E) Sequence model for G-quadruplexes that promote peroxidase reactions. 5’ tetrad = positions 1, 5, 10 and 14; central tetrad = positions 2, 6, 11 and 15; 3’ tetrad = positions 3, 7, 12 and 16; spacer nucleotides = 4, 8, 9, 13 and 17. (F) Expected and observed activities of 30 randomly chosen sequences that satisfy the requirements of our model. The blue line indicates the relationship if expected and observed activities were identical, and was calculated by assuming that spacer sequence does not influence activity. Peroxidase activity is expressed relative to that of the reference construct. Experiments were performed at 10 μM DNA concentration in a buffer containing 200 mM KCl, 1 mM MgCl₂, 20 mM HEPES pH 8, 0.05% Triton X-100, 0.5 μM hemin, 1% DMSO, 5 mM ABTS and 600 μM H₂O₂. In panels (B) and (E), the IUPAC nucleotide code was used to indicate the nucleotides that are permitted to occur at variable positions in each sequence model. W = A or T; H = A, C or T; D = A, G or T; B = C, G or T; N = A, C, G or T.

Figure 8.

Identification of evolutionary conserved G-quadruplexes in the human genome that bind GTP and promote peroxidase reactions. (A) Abundance and evolutionary conservation of G-quadruplexes that bind GTP and promote peroxidase reactions in various vertebrate clades. (B) Sequence alignment of an evolutionary conserved G-quadruplex that contains mutations in a tetrad consistent with the sequence requirements of GTP-binding activity. This example occurs in an intergenic region on chromosome 4. (C) Sequence alignment of an evolutionary conserved G-quadruplex that contains mutations in a tetrad consistent with the sequence requirements of peroxidase activity. This example occurs in an intron shared by two genes on chromosome 10.