Dual Binding of an Antibody and a Small Molecule Increases the Stability of TERRA G-Quadruplex

In investigating the binding interactions between the human telomeric RNA (TERRA) G-quadruplex (GQ) and its ligands, it was found that the small molecule carboxypyridostatin (cPDS) and the GQ-selective antibody BG4 simultaneously bind the TERRA GQ. We previously showed that the overall binding affinity of BG4 for RNA GQs is not significantly affected in the presence of cPDS. However, single-molecule mechanical unfolding experiments revealed a population (48 %) with substantially increased mechanical and thermodynamic stability. Force-jump kinetic investigations suggested competitive binding of cPDS and BG4 to the TERRA GQ. Following this, the two bound ligands slowly rearrange, thereby leading to the minor population with increased stability. Given the relevance of G-quadruplexes in the regulation of biological processes, we anticipate that the unprecedented conformational rearrangement observed in the TERRA-GQ–ligand complex may inspire new strategies for the selective stabilization of G-quadruplexes in cells.


Materials
Small molecule ligand carboxypyridostatin (cPDS) was synthesized as previously described. [1] The antibody (BG4) was prepared as previously described. [2] Aqueous stock solution (10 mM) of cPDS was stored at -80 °C, 1 mM working aliquots were stored at -20 °C, and BG4 was stored at 4 °C. DNA oligonucleotides were purchased either from Integrated DNA Technologies (IDT) (http://www.idtdna.com) or Gene Link (http://www.genelink.com) and further purified by denaturing polyacrylamide gel and stored at -20 °C. Cloned plasmid DNA template for in vitro RNA transcription was sequenced at the University of Maine DNA sequencing facility. TERRA synthetic oligonucleotide for ELISA binding assay was purchased from Sigma Aldrich.

TERRA Construct
The DNA-RNA hybrid construct was prepared by inserting the encoding TERRA quadruplex-forming sequence (5'-TTAGGG TTAGGG TTAGGG TTAGGG TTA -3') and cloning into a pBR322 plasmid (NEB, Ipswich, MA) between EcoRI and HindIII sites. The template for in vitro RNA synthesis (MEGAscript, Invitrogen) was generated from PCR amplification of the cloned plasmid (between the 3836 th and 1473 rd nucleotides) using primers containing the T7 promoter. Handle A (527 bp) was obtained from PCR amplification of the unmodified pBR322 plasmid (between the 3836 th and 1 st nucleotides) with an EcoRI digestion site on the forward primer, followed by EcoRI (NEB, Ipswich, MA) digestion and end-labeling with DNA Polymerase I, Large Fragment (Klenow, NEB) in the presence of digoxigenin (Dig)-11-dUTP (Roche, Indianapolis, IN). Handle B (1473 bp) was prepared from PCR amplification of the unmodified pBR322 plasmid (between the 31 st and 1473 th nucleotides) with a 5'-biotin labeled reverse primer. The control construct was prepared similarly by replacing the TERRA sequence with the wild type sequence between the EcoRI and HindIII sites of pBR322.
The DNA-RNA hybrid constructs were prepared by annealing the RNA prepared by using MEGAscript® T7 high yield transcription kit (Ambion) and the two DNA handles via heating at 95 °C for 5 min and then gradually cooling down to room temperature for 4 h.

Single-Molecule Experiments
Laser tweezers experiments were performed by attaching the DNA-RNA hybrid constructs (~0.25 ng/μL) to the anti-Dig antibody coated polystyrene beads (diameter 2.10 μm, Spherotech, Lake Forest, IL) via the Dig/anti-Dig linkage for 30 min at room temperature. These beads and the streptavidin coated beads (1.87 µm, Spherotech, Lake Forest, IL) were injected into a microfluidic chamber and dispersed into a 1 mL 10 mM Tris buffer with 100 mM KCl, 1 mM EDTA and 1.5 μL RNase inhibitor (RNAsecure, Invitrogen) at pH 7.4. The buffer contains 5 μM cPDS and/or 50 nM BG4. The two different beads were separately trapped by two laser foci in a house-made laser tweezers instrument. [3] The DNA-RNA construct was tethered between the two trapped beads by moving the trapped beads together. Each tethered construct was repeatedly stretched and relaxed by force-ramping between 0 and 60 pN at a loading rate of 5.5 pN/s to record the force-extension (F-X) curves at 1000 Hz using Labview ® program.
Histograms of ∆L and rupture force were prepared by measuring 470 total molecules with and without ligands. Variation between individual molecules is not significant. Single molecules were confirmed by a 65 pN plateau in rupture force observed for each F-X curve which indicates the denaturation of single dsDNA molecules under tension, [4] or single breakage of the tethered molecule to 0 pN. Accurate distance measurement of the instrument was confirmed by determining the contour length per DNA nucleotide. First, F-X curves of a DNA construct that contains a DNA hairpin, 5'-GC(T) 10 GCTTTTGC(A) 10 GC-3', with a tetrathymine loop (underscored), sandwiched between two dsDNA handles as described in literature, [5] were obtained. These F-X curves were then fitted into a sequential model [5] that uses two Worm Like Chain (WLC) functions [4] to represent the ssDNA hairpin and dsDNA handles.
, in which x is the end-to-end distance at a particular force (F), L is the contour length, k B is the Boltzmann constant, T is absolute temperature, P is the persistence length, and S is the stretch modulus. A value of 0.44 ± 0.02 nm per DNA nucleotide was obtained, which is identical to that measured by Block and coworkers [6] and within the same range as determined by others. [7][8][9][10] For the TERRA GQ construct, the values of P (10.6 ± 1.8 nm) and S (191 ± 61 pN) for the hybrid DNA/RNA handles were obtained from sequential fitting of the F-X curves. [5] These values are consistent with those reported in literature. [11]

Population Deconvolution with Nanometer Resolution (PoDNano)
This data analysis method is used to identify the predominant ∆L populations in a ∆L histogram. [12] First, the change-in-extension (∆x) for each F-X curve due to the unfolding of a secondary structure, sandwiched between two nucleic acid handles, was obtained at a particular force F from each set of unfolding and refolding traces. Average standard error (σ) values from the regions immediately after and before the rupture transitions were obtained for each ∆L.
Using this value as the width of a Gaussian kernel, a kernel density distribution was constructed for the ∆L of all unfolding transitions. The most probable populations in the constructed ∆L histogram were identified by performing 3,000 random resampling bootstrapping analyses.
From each resampling, two highest peaks were identified by the Igor ® program (WaveMatrics, Portland, OR). These selected peaks were used to construct the PoDNano histograms.

Force-Jump Experiments
The detailed force-jump procedures have been described elsewhere. [13] Briefly, the TERRA construct tethered between the two beads as stated above was stretched until the structure was unfolded, then relaxed to 0 pN in 10 ms, followed by incubation for varied time durations (0-180 s) to monitor the refolding of the structure. Rupture events in the subsequent unfolding curves that start at 10 pN by another force jump indicate the refolding of secondary structures during the incubation.
The TERRA GQ formation probability with or without ligand for each incubation time was calculated as the ratio of the number of unfolding events observed in subsequent curves to the total subsequent pulling events (folded + unfolded).

Change in Free Energy of Unfolding (∆ ∆ ∆ ∆G unfold )
Change in free energy for the mechanical unfolding of the TERRA structures (∆G unfold ) were calculated by the Jarzynski equation for non-equilibrium systems, [14][15]    where N is the number of observations and W is the nonequilibrium work done to unfold the structure(s), which is equivalent to the hysteresis area between unfolding and refolding F-X curves.

Fitting Equations for Ligand Binding Assays
To determine the folding rate constant, k fold , from the kinetic measurements in the forcejump experiments with or without a ligand, the experimental data of the probability of unfolded and folded species were fitted into the two-state kinetic model shown below, [16] Here, P represents the folding probability, t is the incubation time, and k fold and k unfold are folding and unfolding rate constants, respectively. These equations were used to fit the refolding probability of GQ at different incubation times.