Kinetic Characterization of an Extremely Slow DNA Binding Equilibrium

Fredrik Westerlund, Pa1r Nordell, Bengt Norde´n, and Per Lincoln* ... After adding poly(dAdT)2 as a sequestering agent to B or P bound to ct-DNA, the...
0 downloads 0 Views 138KB Size
Download PDF
9132

J. Phys. Chem. B 2007, 111, 9132-9137

Kinetic Characterization of an Extremely Slow DNA Binding Equilibrium Fredrik Westerlund, Pa1 r Nordell, Bengt Norde´ n, and Per Lincoln* Department of Chemical and Biological Engineering, Chalmers UniVersity of Technology, SE-41296 Gothenburg, Sweden ReceiVed: March 16, 2007; In Final Form: May 18, 2007

We here exploit the recently reported thermodynamic preference for poly(dAdT)2 over mixed-sequence calf thymus (ct) DNA of two binuclear ruthenium complexes, ∆∆-[µ-bidppz(bipy)4Ru2]4+ (B) and ∆∆-[µ-bidppz(phen)4Ru2]4+ (P), that bind to DNA by threading intercalation, to determine their intrinsic dissociation rates. After adding poly(dAdT)2 as a sequestering agent to B or P bound to ct-DNA, the observed rate of change in luminescence upon binding to the polynucleotide reflects the rate of dissociation from the mixed sequence. The activation parameters for the threading and dissociation rate constants allow us for the first time to characterize the thermodynamics of the exceedingly slow threading intercalation equilibrium of B and P with ct-DNA. The equilibrium is found to be endothermic by 33 and 76 kJ/mol, respectively, and the largest part of the enthalpy difference between the complexes originates from the forward threading step. At physiological temperature (37 °C) B and P have dissociation half-lives of 18 and 38 h, respectively. This is to our knowledge the slowest dissociating noncovalently bound DNA-drug reported. SDS sequestration is the traditional method for determination of rate constants for cationic drugs dissociating from DNA. However, the rates may be severely overestimated for slowly dissociating molecules due to unwanted catalysis by the SDS monomers and micelles. Having determined the intrinsic dissociation rates with poly(dAdT)2 as sequestering agent, we find that the catalytic effect of SDS on the dissociation rate may be up to a factor of 60, and that the catalysis is entropy driven. A simple kinetic model for the SDS concentration dependence of the apparent dissociation rate suggests an intermediate that involves both micelles and DNA-threaded complex.

Introduction The mechanisms by which small molecules interact with DNA is a subject of great interest, both for understanding processes occurring in nature and for design of molecules that can bind to DNA in a specific way, for example in diagnostic and therapeutic applications. The most common way to achieve DNA binding specificity is through thermodynamic control, but kinetic factors are also crucial and slow dissociation from DNA is considered to be a most important property of cancertherapeutic DNA-binding drugs.1 We have recently shown that kinetic control of threading intercalation to DNA can result in high specificity for long stretches of AT base pairs.2 Binuclear ruthenium complexes of general formula [µ-bidppz(L)4Ru2]4+ (bidppz ) 11,11′-bi(dipyrido[3,2-a:2′3′-c]phenazinyl), the complex with L ) 1,10-phenanthroline will here be denoted P and that with L ) 2,2′-bipyridine will be denoted B; see Figure 1) intercalate into DNA by threading a coordinated ruthenium ion through the DNA base stack, ending up with one subunit in each groove of DNA and the bridging bidppz ligand sandwiched between the DNA bases.3-6 When intercalated, in the excited state the bridging ligand is protected from hydrogen bonding to water, which gives rise to a very large increase in luminescence (the “light-switch effect”).7-9 Intercalation and dissociation kinetics can thereby be followed by the change in emission intensity with time. The threading is extremely slow for mixed-sequence calf thymus DNA (ct-DNA) at room temperature in 10 mM NaCl buffer,4 but is conveniently followed at 50 °C and an ionic strength of 150 mM NaCl where t1/2, the time to reach half * Telephone: +46-31-772 30 55. Fax: +46-31-772 38 58. E-mail: [email protected].

Figure 1. Structure of the complexes. Top: [µ-bidppz(phen)4Ru2]4+ (P). Bottom: [µ-bidppz(bipy)4Ru2]4+ (B).

completion, is found to range between 6 min for ΛΛ-P and 132 min for ΛΛ-B (Λ denoting the propeller-like ruthenium coordination to be left handed, ∆ right handed).2 The kinetic selectivity for long tracts of AT base pairs is exceptional; the most discriminatory complex, ΛΛ-B, binds 2500 times faster to poly(dAdT)2 than to mixed-sequence ct-DNA at 50 °C, and the difference is even larger at physiological temperature (“kinetic recognition”).2

10.1021/jp072126p CCC: $37.00 © 2007 American Chemical Society Published on Web 07/03/2007

DNA Threading Kinetics Although the association kinetics of these complexes is relatively straightforward to study, the thermodynamic parameters are difficult to assess from binding isotherms due to the extremely slow equilibration. Thus, the thermodynamic parameters have to be evaluated from kinetic data, necessitating the dissociation rates as well as the association rates to be determined with sufficient accuracy. A well-established method to study dissociation of hydrophobic, cationic molecules from DNA is detergent sequestering, i.e., using negatively charged surfactant micelles (typically sodium dodecyl sulfate, SDS) as a scavenger for the dissociated molecules.10-16 The micelles are used to shift the DNA binding equilibrium by lowering the concentration of free cationic molecules in the bulk. The ratelimiting step is the dissociation from DNA if the binding of the dissociated cations to the micelles is diffusion-controlled and thereby much faster than the actual dissociation from DNA. The negatively charged micelles are furthermore assumed not to interact with the DNA, or the DNA-bound ligands, as electrostatic repulsion should keep them at distance from the polyanion DNA. However, we have recently shown that the rate of dissociation of threading intercalators from DNA, determined using SDS sequestering, may be severely overestimated due to catalysis by the SDS monomers and micelles.17 In this study we use poly(dAdT)2 as an alternative scavenger and study, for the first time, the dissociation of the threading intercalators ∆∆-P and ∆∆-B (hereafter referred to only as P and B) from ct-DNA. We find the intrinsic dissociation to be extremely slow (dissociation halflife of 38 h for P at physiological temperature) and that dissociation using SDS micelles as scavenger is up to 60 times faster than sequestration with poly(dAdT)2. The rate constants and activation energies allow us to estimate thermodynamic parameters for the intercalation into and the uncatalyzed dissociation from mixed-sequence ct-DNA of P and B. Furthermore, using a simple kinetic model to characterize the influence of SDS concentration on the dissociation rate, we show that an intermediate that involves both micelles and DNAthreaded complex has to be included to account for the saturation at high SDS concentrations. Materials and Methods Chemicals. All experiments were performed in 150 mM NaCl, 1 mM sodium cacodylate buffer, pH 7. The ruthenium complexes (P and B) were synthesized as described elsewhere,4 calf thymus DNA (ct-DNA) was purchased from Sigma-Aldrich, and poly(dAdT)2 was purchased from Amersham Biosciences. Sodium dodecyl sulfate (SDS) was purchased from SigmaAldrich and stock solutions were made in buffer. Sample Preparation. Stock solution of ct-DNA was prepared by dissolving lyophilized ct-DNA in buffer to a concentration of approximately 10 mM bases. The solution was stirred overnight and filtered three times through a 0.7 µm polycarbonate filter. Stock solution of poly(dAdT)2 was made at a concentration of approximately 3 mM in buffer. Stock solutions of P and B were made by dissolving the complexes in buffer. For the experiments the complex concentration of P and B was kept at 3.75 µM and the DNA concentration was kept at 60 µM bases. The concentrations of all duplex nucleic acid samples were determined by measuring the absorbance on a Cary 4B spectrophotometer, using 260 ) 6600 cm-1 M-1 (ct-DNA) and 262 ) 6600 cm-1 M-1 (poly(dAdT)2), respectively. The concentrations of P and B were determined similarly by using extinction coefficients of 75 800 and 65 000 at 410 nm, respectively. Before use, the samples for the dissociation experiments were equilibrated by heating to 50 °C overnight.

J. Phys. Chem. B, Vol. 111, No. 30, 2007 9133 Fluorescence Measurements. The association and dissociation kinetics of P and B were studied as changes in emission with time using a Varian Eclipse spectrofluorimeter, equipped with a Peltier cooled heating block for control of sample temperature. The excitation wavelength was 410 nm, and the emission was recorded at 620 nm. The steady-state spectra were recorded on a Varian Eclipse spectrofluorimeter with excitation wavelength of 410 nm. Mathematic Model for the SDS Catalyzed Dissociation. The rates of dissociation of DNA-bound ligand (DLtot), measured as the decrease in luminescence, for P and B bound to ct-DNA and poly(dAdT)2 after addition of SDS, can be satisfactorily fitted as first-order decays (see Supporting Information):

d[DL]tot ) -kobs[DL]tot dt

(1)

To account for the dependence of the observed rate constant on the SDS concentration, we assume that kobs is a sum of the rate constants from three parallel processes: (i) the intrinsic dissociation rate constant k-1, (ii) the pseudo-first-order rate constant kmono for a SDS monomer catalyzed dissociation, and (iii) a Michaelis-Menten type of expression for a micellecatalyzed dissociation, which gives rise to the nonlinearity observed at higher SDS concentrations:

kobs ) k-1 + kmono[SDS]mono +

k3[SDS]micelle [SDS]micelle + kM

(2)

We here distinguish between SDS monomers, the concentration of which is assumed to be equal to the critical micelle concentration (cmc), and SDS micelles (the concentration of which is assumed to be constant during the dissociation process, due to the large excess of SDS compared to DNA and metal complex): [SDS]micelle ) [SDS]tot - cmc. Figure 6 shows that the model fits the data well, and Table 2 gives the values for the corresponding microscopic rate constants, obtained with cmc set to 0.9 mM. Since the total SDS concentration at the first data point for each compound is very close to the cmc, the rate constant at this point gives kobs ) k-1 + kmono(cmc), leaving k3 and kM to be determined by the four data points at higher SDS concentrations. Since all k-1 + kmono(cmc) values are small, the observed pseudo-first-order rate constant will, at moderate SDS concentrations, be approximately (k3/kM)[SDS]micelle, and at high total SDS concentrations the rate will approach the maximum rate given by kobs ) k3. The form of the empirical rate law (iii) suggests a mechanism in which the reaction proceeds through an intermediary DNAligand-micelle complex, DLM (comprising an unspecified number of SDS monomers): k2

8 DLM f D + LM DL + SDSmicelle 79 k -2

(3)

The total concentration of DNA bound ligand is given by

[DL]tot ) [DL] + [DLM]

(4)

The change of [DLM] with time is given by

d[DLM] ) k2([DL]tot - [DLM])[SDS]micelle dt (k3 + k-2)[DLM] (5)

9134 J. Phys. Chem. B, Vol. 111, No. 30, 2007

Westerlund et al.

Figure 2. P (top panel) and B (bottom panel) bound to ct-DNA (blue) and poly(dAdT)2 (green) in its final binding mode, respectively. Dashed black line is final emission for an equilibrated sample when a stoichiometric amount of poly(dAdT)2 has been added to a sample with complex bound to ct-DNA in its final binding mode. Excitation wavelength 410 nm. Each panel has been normalized separately. Measurements performed at 25 °C K in 150 mM sodium chloride, 1 mM sodium cacodylate, pH 7.

Assuming steady-state conditions for [DLM], i.e., setting eq 5 equal to zero, and dividing by k2 gives

[DLM] )

[DL]tot[SDS]micelle [SDS]micelle + kM

(6)

where

kM )

k3 + k-2 k2

(7)

can be interpreted as a variant of a Michaelis-Menten constant. Alternatively, without setting eq 5 equal to zero, but instead assuming a fast preequilibrium step

DL + SDSmicelle T DLM eq 6 is again obtained, but kM is now simply the preequilibrium (dissociation) constant:

kM ) k-2/k2 Determination of Activation Parameters. The activation enthalpy (∆Hq) and entropy (∆Sq) were calculated as

∆Hq ) Ea - RT

(8)

∆Sq ) R ln(Ah/kTe)

(9)

and

where Ea and A are determined from the Arrhenius plot of ln(kobs) (see eq 1) versus T-1 in Figure 3. Results When ∆∆-[µ-bidppz(phen)4Ru2]4+ (P) and ∆∆-[µ-bidppz(bipy)4Ru2]4+ (B) are first intercalated into ct-DNA and thereafter incubated with an equivalent amount of poly(dAdT)2, the emission spectra become practically superimposable with the spectra for binding to poly(dAdT)2 only (Figure 2), indicating a large thermodynamic preference for the latter polynucleotide.2 The large difference in emission quantum yield (5 times) for P when bound to poly(dAdT)2 and ct-DNA (Figure 2, top panel) makes it possible to study the binding distribution

between the two nucleic acids by monitoring the change in emission with time. For B the emission quantum yield is similar when bound to both types of DNA, but since the redistribution is associated with a 50 nm wavelength shift to the red, the dissociation from ct-DNA is easily monitored by measuring the change in emission at 620 nm (Figure 2, bottom panel). Since, for both complexes, the binding to poly(dAdT)2 is much faster than the dissociation from ct-DNA, the change in emission with time after adding poly(dAdT)2 to a sample preequilibrated with ruthenium complex and ct-DNA is effectively only reflecting dissociation from the latter. Furthermore, the rate of dissociation from ct-DNA does not increase when the concentration of added poly(dAdT)2 is increased above the stoichiometric amount (data not shown), confirming the strong thermodynamic preference of both P and B for poly(dAdT)2. The highly favored binding to poly(dAdT)2 leads to that, contrary to earlier studies,17 no large excess of scavenger DNA is needed. Figure 3 shows association kinetics for P and B threading into ct-DNA as well as dissociation kinetics from ct-DNA, using either poly(dAdT)2 or SDS (14.4 mM) as scavenger for the dissociated complexes, at three different temperatures. This is the first measurement of the extraordinarily slow intrinsic dissociation of B, and even more so of P, from random-sequence ct-DNA. The dissociation of P has a half-life of 360 min at 50 °C. Extrapolating to physiological temperature (37 °C) using the calculated activation parameters (see below) gives a dissociation half-life of 2300 min, equal to 38 h. B dissociates somewhat faster with a half-life of 110 min at 50 °C, extrapolated to 870 min (18 h) at 37 °C. This should be compared with common intercalators that usually dissociate from DNA with a half-life of less than 1 min.18-20 To the right in Figure 3 Arrhenius plots are shown from which activation parameters are determined (Table 1). The rate constants were obtained by fitting the first 90% of the association or dissociation processes to a monoexponential expression (see Supporting Information). The activation enthalpies for association are 164 and 133 kJ/mol for P and B, respectively, whereas the activation enthalpies for dissociation using poly(dAdT)2 as scavenger are smaller and more similar, 88 and 100 kJ/mol, respectively. The rate constants and the activation parameters allow determination of the thermodynamic parameters of the threading intercalation equilibrium (Table 1). The process is endothermic with a standard enthalpy of reaction more than twice as high for P, 76 kJ/mol, as for B, 33 kJ/mol. Furthermore, for both P and B the activation enthalpy for dissociation is slightly higher with 14.4 mM SDS as scavenger than with poly(dAdT)2 as scavenger. The development of a method to study the intrinsic dissociation rate makes it possible to investigate the catalytic effect of SDS17 on the dissociation of P and B from ct-DNA further. Figure 4 (top section) shows dissociation of P and B from ctDNA sequestered by poly(dAdT)2 and increasing concentrations of SDS, respectively. The dissociation rate is considerably higher when SDS is used as scavenger for both complexes. For dissociation of P and B using 14.4 mM SDS as scavenger, the rate constants are 58 and 30 times higher, respectively, compared to using poly(dAdT)2 as scavenger. There is also a pronounced effect on the rate of dissociation when the SDS concentration is decreased. The rate of dissociation decreases 6 times when the SDS concentration is decreased from 14.4 to 0.9 mM for P. The corresponding effect for B is a 4 times decrease in rate. Since the SDS concentration is above the critical micelle concentration (cmc ≈ 0.9 mM at 150 mM NaCl) for all concentrations studied, the effect seen must be attributed to the

DNA Threading Kinetics

J. Phys. Chem. B, Vol. 111, No. 30, 2007 9135

Figure 3. Left: Association to and dissociation from ct-DNA for P (top panel) and B (bottom panel) studied at 320 K (red), 325 K (black), and 330 K (green). Dissociation is studied using either 14.4 mM SDS (traces to the left) or a stoichiometric amount of poly(dAdT)2 (traces to the right) as sequestering agent. Y-axis values correspond to amount of complex bound to ct-DNA. For Y ) 1 all complex is bound to ct-DNA, and for Y ) 0 no complex is bound to ct-DNA, meaning that for the dissociation experiments all complex has bound to the sequestering agent. Right: The natural logarithm of the rate constant for association (open symbols) and dissociation plotted versus the inverse temperature (in kelvin) for P (squares) and B (circles). Dissociation is studied using either 14.4 mM SDS (half-filled symbols) or a stoichiometric amount of poly(dAdT)2 (filled symbols) as sequestering agent. Measurements performed in 150 mM sodium chloride, 1 mM sodium cacodylate, pH 7.

TABLE 1: Activation Enthalpies and Entropies for Threading into and Dissociation from ct-DNA of P and B at 55 °C (328 K) ∆Hq/kJ/mol association SDSa poly(dAdT)2a

P

B

P

B

164 106 88

133 108 100

87 31 1

61 34 17

∆H°/kJ/mol equilibrium

T∆Sq/kJ/mol

∆G°/kJ/mol

P

B

P

B

76

33

-10

-11

a

Dissociation is studied using either 14.4 mM SDS or a stoichiometric amount of poly(dAdT)2 as sequestering agent. Measurements performed in 150 mM sodium chloride, 1 mM sodium cacodylate, pH 7.

SDS micelles. Noticeably, the rate dependence on SDS concentration is not linear and seems to saturate at higher concentrations. To investigate if the catalyzing effect of SDS depends on DNA sequence, SDS sequestered dissociation of P and B from poly(dAdT)2 was studied (Figure 4, bottom section). Just as for ct-DNA, the effect of SDS differs when comparing P and B: P is considerably more sensitive to SDS concentration than B, and the difference is even larger than for ct-DNA. Whereas the dissociation is twice as fast for P compared to B at the highest SDS concentration studied (14.4 mM), B dissociates 10 times faster than P at the lowest SDS concentration (0.9 mM). Furthermore, B shows a saturation at higher concentration, similar to dissociation from ct-DNA, but no such saturation is seen for P dissociating from poly(dAdT)2 (Figure 6). A simple model that could account for the saturation of the rate at higher SDS concentrations was obtained by including an intermediary DNA-ligand-micelle complex in the dissociation process, as described in the Materials and Methods section. Table 2 shows the rate constants determined by fitting the model to the data as shown in Figure 6. Interestingly, kM, the (pseudo)equilibrium constant for the dissociation of the assumed DNAligand-micelle intermediate, is very similar for the three systems that show a nonlinear SDS dependence, but significantly different for P dissociating from poly(dAdT)2. Discussion Poly(dAdT)2 as Scavenger. We find poly(dAdT)2 to be well suited as a scavenger in dissociation studies of binuclear

Figure 4. Top section: Dissociation of P (top panel) and B (bottom panel) from their final binding to ct-DNA using poly(dAdT)2 (thick line) or SDS (thin lines) as sequestering agent. Bottom section: Dissociation of P (top panel) and B (bottom panel) from their final binding to poly(dAdT)2 using SDS as sequestering agent. SDS concentrations are from left to right: 14.4, 7.2, 3.6, 1.8, and 0.9 mM. Y-axis values correspond to the amount of complex bound to ct-DNA (top section) or poly(dAdT)2 (bottom section). For Y ) 1 all complex is bound to DNA, and for Y ) 0 all complex is bound to the sequestering agent. Measurements performed at 325 K in 150 mM sodium chloride, 1 mM sodium cacodylate, pH 7.

complexes B and P from mixed-sequence DNA. Only a stoichiometric amount, relative to ct-DNA, was needed due to the high affinity for these complexes to the synthetic polymer. Since the threading intercalation to poly(dAdT)2 is comparably fast (t1/2 < 3 min already at 25 °C),2 the observed rate of binding to the polymer will very closely reflect the rate of dissociation. However, this method requires a significant difference between poly(dAdT)2 and the other nucleic acid in the luminescence properties for the threaded binuclear complex. Thus, we cannot exclude that a small fraction of the binuclear complexes are bound more firmly to some sites of the mixed-sequence ct-DNA, which happen to have similar quantum yield and spectral shape

9136 J. Phys. Chem. B, Vol. 111, No. 30, 2007

Figure 5. Kinetic scheme for the threading intercalation of P and B. A is a complex electrostatically bound to DNA, C is a nonthreaded complex bound in one of the grooves of DNA, and D is a complex bound to DNA by threading intercalation. The equilibrium AC is much faster than the equilibrium AD. The rate for the threading is thus given by the equation to the right.

Figure 6. Rate constants for dissociation of P (open symbols) and B (filled symbols) from ct-DNA (squares) and poly(dAdT)2 (circles). Fit of eq 2 as solid curves, using rate constants given in Table 2.

as in poly(dAdT)2. Furthermore, the absence of distinct spectral differences2 makes the method less suitable for the study of the dissociation of the ΛΛ-enantiomers of B and P. Activation Parameters. Using sequestration with poly(dAdT)2 as scavenger enables us to determine the true intrinsic rates of dissociation of the threading intercalated binuclear ruthenium complexes P and B from mixed-sequence ct-DNA. We find that both complexes dissociate extremely slowly and that the dissociation rates determined earlier using SDS sequestration3 are grossly overestimated (up to 60 times). The dissociation rate for P from ct-DNA presented here (estimated dissociation half-life of 38 h at 37 °C) is to our knowledge the slowest dissociating noncovalently bound molecule reported. Although still extremely slow, B dissociates approximately twice as fast as P from ct-DNA at 37 °C. Since dissociation has to involve opening of one or more base pairs in the double helix, the faster dissociation for B could be expected due to the smaller steric bulkiness of the bipyridine complex. Surprisingly, the activation enthalpy barrier is higher for dissociation of B than for P (Table 1). For threading intercalation of the ΛΛenantiomers into poly(dAdT)2 a similar observation of a higher activation barrier for the smaller bipyridine complex has been reported.2 This difference in activation enthalpy was ascribed to a possible transient stabilization of an opened base pair by stacking with an auxiliary phenanthroline ligand. On the contrary, we here find that the activation enthalpy barrier for threading intercalation of B into ct-DNA is lower than for P. We have earlier shown that P and B differ in their DNA binding geometry in the initially bound state.21 Whereas P has an angle of the bridging ligand relative to the helix axis of approximately 45°, B binds with its bridging ligand much more parallel to the DNA bases. Figure 5 shows a plausible kinetic scheme for the binding of these complexes to DNA that connects the groovebound (C) and the intercalated (D) states with a common intermediate state (A). The groove binding preequilibrium (Aa) is, since C is formed immediately upon mixing, orders of magnitudes faster than the threading intercalation equilibrium (A a D). Thus, the rate of threading is dependent on the equilibrium constant (K ) kAC/kCA) for the groove-bound state.

Westerlund et al. This means that even if the actual forward threading process A f D has a similar activation enthalpy for both B and P, the A a C preequilibrium can have a quite different ∆H°, attributing the difference in the overall activation enthalpy to the difference of the initially bound states. SDS-Catalyzed Dissociation. The influence of SDS on the dissociation rate is significant, and the rate increases markedly with increasing SDS concentration (Figure 4). An anionic amphiphilic molecule like SDS can be expected to associate to the positively charged DNA-bound ruthenium complexes. With increasing SDS concentration the amount of SDS aggregates will increase, both as micelles in bulk and associated to the metal complex on DNA, possibly as “hemimicelles”. Support for the involvement of SDS aggregates associated with the DNA-drug complex is provided by the gradual saturation of the dissociation rates observed at high SDS concentrations in all cases, except for P dissociating from poly(dAdT)2. P bound to poly(dAdT)2 has in earlier studies been found to eventually redistribute after the initial intercalation step to favor complexes bound closer together.5 If the complex ions are bound closer together, a multibase pair opening catalyzed by bound SDS micelles may release more bound complex than with more uniformly distributed complex ions, as with ct-DNA, preventing the saturation seen for the other combinations of DNA and complex. We model the SDS-sequestered dissociation of P and B from ct-DNA and poly(dAdT)2 in Figure 6 by assuming that the observed rate for the first-order dissociation of total DNA-bound ligand is a sum of the intrinsic rate (rate constant k-1), the pseudo-first-order rate (kmono) for the SDS monomer catalyzed dissociation and a Michaelis-Menten type of expression for the micelle-catalyzed dissociation processes, respectively (see Materials and Methods). Figure 6 shows the best fit of the model and Table 2 gives the values for the corresponding microscopic rate constants. The most interesting observation is that the (pseudo)equilibrium constant kM, for dissociation of SDS micelles from the DNA-ligand-micelle intermediate, is similar for all systems, except for P dissociating from poly(dAdT)2. The similar kM values, resulting from the similar curvatures of the kobsvs [SDS] plots, suggest that it might primarily be the charge of the metal complex, and not the detailed structure, that is important for the weak binding of the SDS micelles to the DNA-bound ligand. Interestingly, the activation enthalpy barrier is not lowered for the SDS-catalyzed sequestration compared to the intrinsic dissociation (Table 1); it is in fact found to be slightly higher, which shows that the catalysis is entropy driven. This conclusion is consistent with the notion of an interaction, driven by release of counterions and water, between preorganized micelles and exposed hydrophobic parts of the intercalated complex in the rate-determining step. The entropy-enthalpy compensation, evident from comparison of the standard reaction enthalpies and free energies for the threading equilibrium of P and B, supports this notion of the importance of shielding hydrophobic surfaces from water in the threading intercalation. The conclusion that hydrophobic interactions are of primary importance for the large catalytic effect of SDS is in agreement with the theoretical predictions made in a recent analysis of the SDS effect on the dissociation of ∆∆-[µ-c4(cpdppz)2(phen)4Ru2]4+ from poly(dAdT)2.22 The catalytic effect varies with type of DNA and type of complex, and it is meaningful to discuss the mechanism of catalysis in terms of structure of a transition state landscape where hydrophobic interactions are important. The fundament

DNA Threading Kinetics

J. Phys. Chem. B, Vol. 111, No. 30, 2007 9137

TABLE 2: Rate Constants for SDS Sequestered Dissociation of B and P from Calf Thymus DNA (CT) and Poly(dAdT)2 (AT) k-1/min-1 B CT B AT P CT P AT a

0.0064 n.d.a 0.0019 n.d.a

ks/min-1 M-1

kM/M

k3/min-1

k3/kM/min-1 M-1

17 18 2.1 2.2

0.0097 0.012 0.0090 0.086b

0.29 0.14 0.18 1.64b

30 12 20 19

Not determined; set to zero in eq 2. b Lower limits; only their ratio k3/kM is accurately determined from the data.

of DNA stability is the hydrophobic interactions that provide the base stacking, and it is reasonable to believe that opening of the base stack, required to let a threading intercalator pass through, will lead to unfavorable exposures of the DNA bases to water. In such a landscape any extra hydrophobic functionalities can be anticipated to have catalytic effects by forming transient hydrophobically attractive structures, lowering the activation free energy: we may speak of “hydrophobic catalysis”. Conclusions From this kinetic study of the thermodynamic parameters of dissociation of tetravalent cationic ruthenium complexes from DNA, the following may be concluded: (a) The dissociation of the complexes from ct-DNA, studied for the first time in the absence of catalyzing SDS by using poly(dAdT)2 as scavenger, shows that P and B dissociate with dissociation half-lives of 38 and 18 h at physiological temperature, respectively. (b) The difference in activation enthalpy is much larger for association than for dissociation, which can be ascribed to the different binding geometries in the initial groove-bound state for the two complexes. (c) SDS catalyzes the dissociation up to 60 times and the catalytic effect has been characterized with respect to dependence on structural and thermodynamic properties, and the activation enthalpy barrier for dissociation is not lowered when using SDS compared to when using poly(dAdT)2, implying that the catalysis is entropy driven. Acknowledgment. We are grateful to the Swedish Science Council (VR) for financial support of this project. Supporting Information Available: Fitting of monoexponential decays to experimental dissociation data at 328 K. This

material is available free of charge via the Internet at http:// pubs.acs.org. References and Notes (1) Mu¨ller, W.; Crothers, D. M. J. Mol. Biol. 1968, 35, 251. (2) Nordell, P.; Westerlund, F.; Wilhelmsson, L. M.; Norde´n, B.; Lincoln, P. Angew. Chem., Int. Ed. 2007, 46, 2203. (3) Wilhelmsson, L. M.; Westerlund, F.; Lincoln, P.; Norde´n, B. J. Am. Chem. Soc. 2002, 124, 12092. (4) Wilhelmsson, L. M.; Esbjo¨rner, E. K.; Westerlund, F.; Norde´n, B.; Lincoln, P. J. Phys. Chem. B 2003, 107, 11784. (5) Westerlund, F.; Wilhelmsson, L. M.; Norde´n, B.; Lincoln, P. J. Phys. Chem. B 2005, 109, 21140. (6) Nordell, P.; Lincoln, P. J. Am. Chem. Soc. 2005, 127, 9670. (7) Friedman, A. E.; Chambron, J. C.; Sauvage, J. P.; Turro, N. J.; Barton, J. K. J. Am. Chem. Soc. 1990, 112, 4960. (8) Hiort, C.; Lincoln, P.; Norde´n, B. J. Am. Chem. Soc. 1993, 115, 3448. (9) Westerlund, F.; Eng, M. P.; Winters, M. U.; Lincoln, P. J. Phys. Chem. B 2007, 111, 310. (10) Gabbay, E. J.; Grier, D.; Fingerle, R. E.; Reimer, R.; Levy, R.; Pearce, S. W.; Wilson, W. D. Biochemistry 1976, 15, 2062. (11) Wilson, D. W.; Grier, D.; Reimer, R.; Bauman, J. D.; Preston, J. F.; Gabbay, E. J. J. Med. Chem. 1976, 19, 381. (12) Wilson, W. D.; Krishnamoorthy, C. R.; Wang, Y. H.; Smith, J. C. Biopolymers 1985, 24, 1941. (13) Fox, K. R.; Brassett, C.; Waring, M. J. Biochim. Biophys. Acta 1985, 840, 383. (14) Wakelin, L. P. G.; Atwell, G. J.; Rewcastle, G. W.; Denny, W. A. J. Med. Chem. 1987, 30, 855. (15) White, R. J.; Phillips, D. R. Biochem. Pharmacol. 1988, 38, 331. (16) Phillips, D. R.; Greif, P. C.; Boston, R. C. Mol. Pharmacol. 1988, 33, 225. (17) Westerlund, F.; Wilhelmsson, L. M.; Norde´n, B.; Lincoln, P. J. Am. Chem. Soc. 2003, 125, 3773. (18) Foerster, W.; Stutter, E. Int. J. Biol. Macromol. 1984, 6, 114. (19) Chaires, J. B.; Dattagupta, N.; Crothers, D. M. Biochemistry 1985, 24, 260. (20) Macgregor, R. B., Jr.; Clegg, R. M.; Jovin, T. M. Biochemistry 1985, 24, 5503. (21) Lincoln, P.; Norde´n, B. Chem. Commun. 1996, 2145. (22) Marcus, R. A. J. Phys. Chem. B 2005, 109, 21419.