Strong DNA deformation required for extremely slow DNA threading intercalation by a binuclear ruthenium complex

DNA intercalation by threading is expected to yield high affinity and slow dissociation, properties desirable for DNA-targeted therapeutics. To measure these properties, we utilize single molecule DNA stretching to quantify both the binding affinity and the force-dependent threading intercalation kinetics of the binuclear ruthenium complex Δ,Δ-[μ‐bidppz‐(phen)4Ru2]4+ (Δ,Δ-P). We measure the DNA elongation at a range of constant stretching forces using optical tweezers, allowing direct characterization of the intercalation kinetics as well as the amount intercalated at equilibrium. Higher forces exponentially facilitate the intercalative binding, leading to a profound decrease in the binding site size that results in one ligand intercalated at almost every DNA base stack. The zero force Δ,Δ-P intercalation Kd is 44 nM, 25-fold stronger than the analogous mono-nuclear ligand (Δ-P). The force-dependent kinetics analysis reveals a mechanism that requires DNA elongation of 0.33 nm for association, relaxation to an equilibrium elongation of 0.19 nm, and an additional elongation of 0.14 nm from the equilibrium state for dissociation. In cells, a molecule with binding properties similar to Δ,Δ-P may rapidly bind DNA destabilized by enzymes during replication or transcription, but upon enzyme dissociation it is predicted to remain intercalated for several hours, thereby interfering with essential biological processes.


INTRODUCTION
When synthesizing DNA-targeted drugs, high DNA binding affinity and slow DNA binding kinetics are considered essential aspects that enhance their therapeutic capability (1)(2)(3)(4)(5). Higher affinities maximize the treatment effect with minimal dose exposure, lower association rates enable selective binding molecules to scan DNA for a specific targeted sequence, and lower dissociation rates ensure disruption of DNA transcription and duplication (1,3,4,(6)(7)(8)(9). These desired properties are found in ligands that intercalate DNA by threading (10)(11)(12)(13), an interaction that first requires nonintercalating moieties of a ligand to pass between DNA base pairs before the intercalative binding occurs (6,7,14,15). The threading step is required for the ligand to bring the intercalating moiety in close proximity to the base pairs in order to reach an equilibrium state, resulting in a very slow binding process (6,14,16,17). This is in contrast to conventional intercalation, which occurs by simple insertion between the DNA base pairs.
First synthesized more than a decade ago (18), the binuclear ruthenium complex , -[-bidppz-(phen) 4 Ru 2 ] 4+ ( , -P) (Figure 1) exhibits extremely slow DNA intercalation kinetics, as determined by luminescence and circular dichroism (CD) bulk experiments, due to the threading intercalation binding mechanism (6,12,19). Early bulk studies overestimated the dissociation half-life of , -P from calf thymus DNA due to catalytic interactions by the surfactant molecules sodium dodecyl sulfate (SDS) (11,20). A revised method using competitive DNA binding reported a dissociation half-life of 38 h at 37 • C, two orders of magnitude slower than the SDS measurements (11). Furthermore, , -P has been shown to have a preference for ATrich DNA sequences (7,(21)(22)(23). The equilibrium dissociation constant for the DNA-, -P complex and the molecular mechanism governing the threading intercalation have not been quantitatively determined, due to the limitations  of traditional techniques for investigating ligands with extremely slow kinetics and high binding affinities.
Binding by intercalation increases the equilibrium separation between the base pairs, providing a direct measurement of DNA-, -P complex formation (4,10,(24)(25)(26)(27), e.g. by using single molecule DNA stretching. However, such measurements are typically performed with pulling rates much faster than the time required for slow DNA binding ligands to equilibrate (10,24). Instead, the elongation can be mechanically probed for a single DNA molecule at constant applied force while the intercalation interaction approaches equilibrium (4,10,24). Optical tweezers are able to resolve single DNA molecule intercalation elongation of a few nanometers while maintaining an applied constant force with 1 pN resolution (10,24,26,(28)(29)(30). Paramanathan et al. demonstrated the ability of such an approach to characterize the extremely slow process of DNA-, -P threading intercalation at a , -P concentration of 2 nM (10), showing that higher forces exponentially facilitate this process. However, this study did not explore the equilibrium binding or the concentration-and force-dependent individual forward and reverse threading processes. We later developed an optical tweezers-based approach that allows complete characterization of very slow intercalation processes, and applied it to characterize the activity of the historically first anticancer antibiotic Actinomycin D (ActD) (24). Based on our findings we suggested a novel mechanism for the anticancer activity of this drug involving its ability to bind transcription-destabilized DNA exponentially stronger and faster than stable chromosomal DNA, and to remain intercalated on a very long timescale (24). The current study utilizes the same approach toward complete characterization of DNA threading intercalation by , -P.

MATERIALS AND METHODS
In all experiments, dual-beam optical tweezers (lasers wavelength 830 nm) were used to trap a streptavidin-coated ∼5.6 m polystyrene bead with a typical optical trap stiffness of 80 pN/m, while a second identical bead was held by a micropipette with a ∼1 m tip. A single bacteriophage -DNA, labeled on opposite strands with biotin, was attached between the two beads and a piezoelectric positioner controlled the micropipette displacements (±10 nm) to maintain a fixed applied force (±1 pN) on the DNA molecule (26). After the attachment between the two beads, the DNA-only stretching curve ( Figure 2, black curve) is first measured at a pulling rate ∼200 nm/s. Then, the DNA is stretched rapidly (∼2 s) to reach the targeted constant force and the elongation from the DNA-only equilibrium extension due the threading intercalation by , -P is recorded until the DNA-, -P complex reaches its equilibrium extension. Typically the force feedback measurements were obtained every 200 ms, while the force feedback system can react to force changes as fast as 50 ms by mechanically repositioning the micropipette to recover the targeted force. The experiments were conducted in a flow cell chamber volume of ∼100 l while continuously flowing solution of the desired , -P concentration at a flow rate of ∼2 l/s, holding the force constant. Constant force measurements of at least three DNA molecules were acquired at each force and , -P concentration. All experiments were carried out at 21 • C and buffer conditions of 10 mM Tris, 100 mM NaCl and pH 8. The , -P complex was synthesized as described elsewhere (14).

Equilibrium extension measurements at fixed force yield DNA-, -P titration curves and saturated complex elasticity
Equilibrium DNA extension measurements were performed for forces ranging from 7 pN to 60 pN at ligand concentrations from 5 nM to 150 nM, as shown in Figure 2A along with the previously reported measurements at 2 nM (10). Throughout the measurements, the force was maintained constant (±1 pN), which is illustrated in Figure 2A for a 40 pN measurement at 5 nM , -P concentration. At each constant force the extension of the DNA-, -P complex increased exponentially with time and converged to an equilibrium value, L eq (F,C), (per bp) ( Figure 2B). The time evolution of the DNA-, -P complex extension at all forces is well fit by a single exponential with a net relaxation rate, k total (F, C) (4,24): supporting a single molecular mechanism for threading intercalation, at least under our force-facilitated conditions. This confirms previously reported results from optical tweezers experiments for -DNA, and it is in agreement with bulk CD measurements for AT-DNA (10,19). However, previous luminescence measurements and CD measurements on mixed sequence DNA were fitted with multiple exponential equations, indicating multiple step behavior for , -P binding. It is possible that such bulk observations, which were made under very low applied forces, reflect more subtle non-threading processes (19). Such nonthreading processes may include pre-intercalation binding to dsDNA, ligand rearrangements due to DNA-DNA contacts, as well as post-intercalation DNA-ligand interactions that do not contribute to DNA elongation. In contrast, the rate that is found in single DNA molecule experiments is measured exclusively from the mechanical length change due to intercalation on the timescale of tens of minutes.
For each constant force we obtained extension measurements at increasing concentration until we reached the saturated extension, L sat (F). The equilibrium extensions per bp of the DNA-, -P complex, L eq (F,C), are presented in Figure 2A for several force values between 10 and 60 pN. For each force the fractional equilibrium binding of ligand (F, C), which ranges from zero to one, was determined by comparing L eq (F,C) to the saturated extension at the same force, L sat (F), (4,24): where L DNA (F) is the DNA extension in the absence of ligand binding. Afterward, we use (F, C) as a function of force at each fixed concentration to obtain the equilibrium force-extension curves L eq (F, C) (and its reciprocal function F(L eq ,C)). It follows from Equation (2) that the equilibrium force-extension L eq (F, C) for each ligand concentration C can be calculated as a linear combination of L DNA (F) and L sat (F) weighted with the fraction of each component, governed by (F,C), according to the relationship: Fits to the experimental force-extension curves obtained by using Equation (3) (3) and fit to the worm-like chain (WLC) model of polymer elasticity yields an effective persistence length of ∼2 nm. This is significantly lower than the persistence length of the saturated DNA-monomer -P complex of ∼14 nm (25), which intercalates DNA without threading.
Our observation that the complete L eq (F,C) curves calculated in this manner fit the equilibrium extensions obtained as a result of relaxation at each concentration supports the assumption that the intercalated DNA is indeed a mixture of the ligand-free and ligand-saturated DNA to a good approximation. It is important to note that, due to slow, force-dependent ligand binding, the equilibrium forceextension curve cannot be obtained directly, as this would require pulling on the timescale of many hours. To illustrate this point, a force-extension curve obtained after stretching the DNA to 60 pN at 50 nM, waiting at constant force until equilibrium is reached and then releasing at a rate of 200 nm/s shows very different elasticity due to the nonequilibrium nature of the stretching curve (cross symbols, Figure 2A). In principle, a very slow release (over many hours) of the DNA-, -P complex from its equilibrium extension at 60 pN would reflect the equilibrium extension fit obtained from Equation (3).

, -P-DNA binding affinity and the site size are strongly affected by force
The measured equilibrium length of the DNA-, -P complex L eq (F,C) is presented in Figure 3A as a function of concentration. The McGhee-von Hippel binding isotherm (26,27,(31)(32)(33) is employed to fit these measurements by comparing the values obtained from Equation (2) to the fit values obtained from Equation (4), shown as dashed lines in Figure 3A.
There are two fitting parameters in Equation (4), the equilibrium dissociation constant K d and the binding site size n, which are varied independently to fit the experimental L eq (F,C) curves at each force. The binding curves could not be fit well to a single value of n for all curves. The fitted K d (F) values decrease with force from 21±4 nM at 10 pN to 2.8±0.1 nM at 60 pN (see Figure 3B). In addition, the fitted value of the binding site size of , -P, n(F), decreases from 3.7±0.2 at 10 pN to 1.7±0.1 at 60 pN ( Figure 3C). Analyzing the fitted K d (F) values in terms of the expected exponential force dependence for ligand binding processes that elongate DNA (4,24,25,34) gives a zero-force equilibrium dissociation constant K d (0) of 44±2 nM. We also obtain a value of x eq = 0.19±0.01 nm, which represents the increase in DNA equilibrium extension due to one ligand binding event.
The measured force-dependent DNA-, -P complex elongation per bp observed at saturation, L sat (F) = L sat (F) − L DNA (F), along with the fitted value of the binding site size n(F), provides an independent estimate of x eq at each force Substituting the fitted values of n(F) and L sat (F) into Equation (6), we obtain the force-independent value of DNA elongation due to the single , -P intercalation event, x eq = 0.23±0.01 nm, averaged over all measured forces. The result is in reasonable agreement with the value of the same parameter x eq = 0.19±0.01 nm obtained by fitting K d (F) to Equation (5). The decreasing binding site size shown in Figure 3C implies that higher forces promote additional ligand intercalation at saturation, such that at 60 pN almost every DNA stack can be intercalated. In contrast, at 10 pN only every 3.7 stacks are intercalated at saturation.

Kinetics analysis unveils pronounced base pair displacements required for binding
The kinetics of , -P binding to DNA can be followed as an increase in the DNA-ligand complex length over time, which fits well to a single exponential (Equation (1) and Figure 2B) with a net relaxation rate k total = k on + k off, where k on and k off are the bimolecular forward and the unimolecular reverse threading rates, respectively. As the DNA-, -P complex relaxation kinetics is determined at a fixed force, all rates pertain to that particular force. Because we also determined K d (F), we can use this information to obtain the individual k on (F) and k off (F) rates at each constant force (24): Furthermore, k on (F) and k off (F) at each ligand concentration can be individually fitted to an exponential dependence on force, which provides a measurement of the zero-force rate, k on/off (0), and the change in DNA length between the unbound and the forward transition state, x on , and the intercalated and the reverse transition state, x off , respectively, k on/off (F) = k on/off (0)e x on/off F/kT (9) Figure 4A presents the force-dependent kinetics analysis at 5 nM concentration. Interestingly, both k on (F) and k off (F) are strongly facilitated by force, due to the significant complex elongations required for both ligand association (x on = 0.33±0.01 nm) and dissociation (x off = 0.14±0.01 nm). Similar dynamic displacements are obtained from analyzing the kinetics at all other concentrations below saturated binding (data not shown). The difference between the measured complex elongations to the transition state in the forward and reverse directions provides another independent estimate of the equilibrium complex elongation upon a single , -P intercalation event ( Figure 5), x eq = x onx off = 0.19±0.01 nm, which agrees with the two other measurements of this parameter discussed above. The forward rates k on (F) for all measured forces show a linear dependence on ligand concentration, whereas the backward rates k off (F) are independent of concentration, as expected for a bimolecular process. For example, the forward rate k on at 30 pN ranges from ∼10 −4 s −1 ( ∼2 h) at 2 nM to ∼10 −2 s −1 ( ∼2 min) at 150 nM, as illustrated in Figure 4B. The slope from the linear concentration dependence at each measured force gives the association rate k a (F), as shown in Figure 4C for all measured forces. By fitting the association rates to the exponential dependence on force (Equation (9)), we obtain a zero-force association rate of k a (0)∼10 4 M −1 s −1 . This bimolecular association rate is ∼5 orders of magnitude slower than an association rate limited by diffusion.

Equilibrium intercalation properties of , -P
DNA threading by , -P is dramatically facilitated by force, as indicated by the exponential decrease in the DNA-ligand equilibrium dissociation constant with applied stretching force ( Figure 3B). This , -P -intercalated DNA state is stabilized by force because each intercalation event results in a DNA elongation of x eq = 0.19 nm. Also, the DNA stretching force appears to relieve significant steric constraints against , -P threading, leading to saturated , -P-DNA intercalation every 1.7 bp at 60 pN force (Figure 3C), in contrast to the observed saturated intercalation of only every ∼3.7 bp at low forces (10 pN).
In Table 1, we compare the parameters describing equilibrium , -P binding with the analogous parameters for the corresponding mononuclear complex, Ru(phen) 2 dppz 2+ ( -P), also measured by DNA stretching with optical tweezers (25). Only the equilibrium intercalation by these two related ligands can be compared, as the rapid intercalation of the -P molecules on our DNA stretching timescale ∼1 min does not allow studies of its intercalation kinetics (25). At low forces (F = 10 pN), the intercalative binding site size of , -P (3.7 bp) is somewhat larger than that of -P (2.9 bp), as could be expected from the larger size of the former complex. However, increasing force has a much more dramatic effect on the binding site size of , -P, which decreases by 2 bp to 1.7 bp at 60 pN, relative to the binding site size of -P, which decreases by only 0.4 bp ( Table 1). Surprisingly, the equilibrium DNA elongation upon monomer -P intercalation x eq = 0.38 nm is almost twice as large as the elongation upon , -P intercalation, x eq = 0.2 nm. Also, the measured zero-force equilibrium dissociation constant of the , -P molecule K d = 44 nM is 25-fold smaller than for -P, suggesting additional interactions that drive , -P intercalation, despite the similar dppz stacking moiety for these two ligands. Finally, the flexibility of the equilibrium ligand-saturated DNA complex for both monomerand dimer-ligands is much higher than for the ligand-free double-stranded B-from DNA, as described by the corresponding persistence length of these complexes of 14 nm, 2 nm and 50 nm, respectively.
These results may be interpreted in terms of the recent crystal structure of , -P intercalated into a doublestranded DNA 6-mer d(CGTACG) (16). According to this study, the , -P molecule traverses the DNA double helix and displaces an AT base pair from the DNA stack to position the two bulky Ru centers on the opposite groove sides of the double helix. These A and T bases that are displaced into the minor groove make extensive stacking contacts with the phen aromatic moieties of the Ru center located in the minor groove. The overall , -P-DNA binding mode was termed 'insertion threading intercalation', as opposed to non-displacive intercalation, which occurs for the mono-nuclear ruthenium compound -P (35,36). The difference between these two types of intercalation provides a potential explanation for the difference in DNA elongation after these binding events, as observed in our DNA stretching experiments. Indeed, the classical intercalation typical of the -P ligand leads to duplex elongation by ∼0.38 nm, which is similar to the length of a normal base pair stack ∼0.34 nm, as for most of the other classic intercalators (25,27,37). In contrast, the inserted aromatic dppz moiety of , -P takes the place of the extruded base, thereby leading to a much smaller DNA elongation. The more severe duplex deformation caused by insertion threading intercalation is also consistent with the much smaller persistence length of the saturated , -P-DNA complex of 2 nm observed here. While this is much smaller than the B-form DNA persistence length of 50 nm, it still involves 4-6 intercalated DNA base pairs, in agreement with the relatively straight conformation of the final equilibrium bound state, in which the DNA duplex can be described as a distorted form of B-DNA and appears to be underwound (16).
such that G thread ∼ 11.5 k B T = 6.8 kcal/mol is the free energy of DNA deformation involved in the rate limiting step of , -P threading. This free energy is consistent with the complete melting of 1-2 base pairs in the middle of the DNA duplex. While the stability of the single terminal duplex bp is only 2-4 k B T, the creation of the helix/coil boundaries associated with local duplex disruption can easily introduce an additional 8-10 k B T (42). The free energy cost of threading deformation is much smaller than its enthalpy estimated from the temperature dependence of threading, discussed above. This result is not surprising, as the large enthalpy cost of threading deformation can be partially compensated by the associated entropy gain.
In vivo it is likely that the very slow threading of , -P into stable polymeric dsDNA would prevent this ligand from intercalating most chromosomal DNA if it reaches the nucleus. However, DNA undergoing high rates of transcription and replication, such as may be found preferentially in cancer cells, may already be destabilized by RNA polymerase or other motor proteins, facilitating rapid initial binding to these locations, followed by very slow dissociation. Therefore, the slow dissociation kinetics combined with its high DNA binding affinity highlights the potential therapeutic use of ligands with binding properties similar to , -P (43,44).