Distance-dependent duplex DNA destabilization proximal to G-quadruplex/i-motif sequences

G-quadruplexes and i-motifs are complementary examples of non-canonical nucleic acid substructure conformations. G-quadruplex thermodynamic stability has been extensively studied for a variety of base sequences, but the degree of duplex destabilization that adjacent quadruplex structure formation can cause has yet to be fully addressed. Stable in vivo formation of these alternative nucleic acid structures is likely to be highly dependent on whether sufficient spacing exists between neighbouring duplex- and quadruplex-/i-motif-forming regions to accommodate quadruplexes or i-motifs without disrupting duplex stability. Prediction of putative G-quadruplex-forming regions is likely to be assisted by further understanding of what distance (number of base pairs) is required for duplexes to remain stable as quadruplexes or i-motifs form. Using oligonucleotide constructs derived from precedented G-quadruplexes and i-motif-forming bcl-2 P1 promoter region, initial biophysical stability studies indicate that the formation of G-quadruplex and i-motif conformations do destabilize proximal duplex regions. The undermining effect that quadruplex formation can have on duplex stability is mitigated with increased distance from the duplex region: a spacing of five base pairs or more is sufficient to maintain duplex stability proximal to predicted quadruplex/i-motif-forming regions.

2 Temperature-dependent absorption profiles. Absorption was monitored at 260 nm and 295 nm over a temperature gradient to assess duplex, G-quadruplex, and i-motif thermal stabilities (Figures S1A, S2A, S3A, S4A) (1,2). In all cases, duplex melting could be identified by an increase in absorbance at 260 nm (and slightly at 295 nm) upon thermal denaturation of the duplex. Complete duplex melting was sometime observed to induce a change in baseline, because the molecular environment adjacent to the G-quadruplex is altered. The Gquadruplex was found to be highly stable, and the onset of its dissociation was accompanied by a strong hypochromic shift at 295 nm, while absorption at 260 nm increased slightly as precedented (2). In turn, i-motif melting was characterised by a hyperchromic shift and occurred at high temperature depending on the acidity of the buffer (3). At pH 4.0, its melting temperature was found to be 73.5 °C (σ = 1.0, n = 9), which is in excellent agreement with previous reports (Tm = 69.4 °C at pH 4.4) (4). In all cases, the resulting melting curves were fully reversible, i.e. heating and cooling curves superimposed well and hysteresis was negligible if present at all. This indicates that at the rate of temperature change used (0.25 °C/min), folding and unfolding reaches equilibrium.

Concentration-dependent UV melting experiments. Total strand concentration in UV
melting profiles was varied between 1 µM and 10 µM. Increasing DNA concentration led to an increase in Tm of the duplex, while the thermal stability of the G-quadruplex and the imotif remained unaltered (Figures S1D; S2D,F; S3D,F; S4D,F). This concentration dependence of duplex stability proves that its formation is a bimolecular process. In turn, both the G-quadruplex and the i-motif only form from one strand, which is consistent with the experimental design depicted in Figure 1 in the main text. It is important to note that experiments at a total strand concentration of 6 µM were repeated in order to be consistent.
Minor deviations within the accuracy of the method were observed in some cases, but the replicates originating from different stock solutions measured on different instruments were generally found to be in excellent agreement (5). 3 Van't Hoff plots. Van't Hoff analysis requires duplex formation and dissociation to be theoretically described by a two-state model, which assumes that that the two strand are either maximally paired or entirely dissociated (6). Other than by performing van't Hoff analysis, thermodynamic parameters can as well be determined from a plot of 1/Tm versus ln(total strand concentration/4) (Equation 1) (6): As expected, these plots were linear in all cases, allowing for determination of ΔH° and ΔS° from the slope and the intercept (Figures S1E, S2E, S3E, S4E). The resulting thermodynamic parameters were always within 10 % difference, suggesting that the two-state model is valid.
Temperature-dependent CD spectra. CD spectra were recorded at different temperatures in 80 mM KCl, 10 mM Britton-Robinson buffer and in the presence of a consistent total strand concentration of 6 µM. In order to properly characterise both the duplex system and the quadruplex system (where present, N = 2 -see Equation 2), spectral measurements obtained for a blank sample containing only dissolved salt and buffer were subtracted from the raw ellipticity data. The resulting spectra revealed the existence of an isoelliptic point in the relevant temperature range (Figures S1E, S2C, S3C, S4C). Analyses of the CD spectra suggest that the contributions of the single-strands and the double-stranded DNA to the overall spectrum are linearly dependent over the entire region of wavelengths investigated, further supporting the two-state hypothesis (7).
Subtraction of blank measurements containing both buffer and the duplex-forming strand from raw ellipticity spectra were used to elucidate the contribution of the four-stranded structure to the overall spectrum at different temperatures ( Figures S2B, S3B, S4B, N = 1, see below). In the case of the G-quadruplex, a strong positive band around 295 nm and at 260 nm appeared at 70 °C and lower was observed that is consistent with the melting profiles and with the previously reported high thermodynamic stability of the Tetrahymena G-quadruplex (8), 4 which is slightly decreased through the duplex overhang (9). The signature observed in the CD spectra was found to be in good agreement with the spectrum reported for (T 2 G 4 ) 4 alone (Tetrahymena G-quadruplex with two additional T bases 5' of the G-quadruplex) and may be attributed to a 3+1 antiparallel topology (10).In the case of the i-motif, a strong positive peak was observed around 280 nm and a negative peak around 260 nm was observed at 70 °C and lower, in excellent agreement with the literature precedent (4). CD in all figures are expressed as the mean nucleoside residue ellipticity [θ] and have been calculated using: where θ denotes the ellipticity in mdeg (the instrument output), N is the total number of strands contributing to the observed signal, n i is the number of nucleoside residues, c i is the strand concentration, and l is the pathlength of the cuvette (11).
pH-dependent UV melting profiles and CD spectra. At pH 4.0, the thermal stability of the G-quadruplex was found to be significantly decreased ( Figure S5). Interestingly, this phenomenon was not observed at pH 4.4. G-quadruplex topology and stability have previously been reported to be insensitive to pH (12). As the sequences used involved Gquadruplexes in isolation, the observed quadruplex destabilisation is most likely due to protonation of adenine N1 (pKa(N1H + ) = 3.71) and/or cytosine N3 (pKa(N3H + ) = 4.31 ± 0.07) within the sequences adjacent to the G-quadruplex (13).

Supplementary Tables
Supplementary Table S1. UV melting data. Experimental error is given in ± 1σ. Propagation of experimental error: