Binding of Cationic Conjugated Polymers to DNA: Atomistic

Feb 25, 2011 - Simulations of Adducts Involving the Dickerson's Dodecamer. Julien Preat,*. ,†. David Zanuy,. ‡. Eric A. Perp`ete,. † and Carlos ...
0 downloads 0 Views 4MB Size
Download PDF
ARTICLE pubs.acs.org/Biomac

Binding of Cationic Conjugated Polymers to DNA: Atomistic Simulations of Adducts Involving the Dickerson’s Dodecamer Julien Preat,*,† David Zanuy,‡ Eric A. Perpete,† and Carlos Aleman*,‡,§ †

Unite de Chimie Physique Theorique et Structurale, Facultes Universitaires Notre-Dame de la Paix, rue de Bruxelles, 61, B-5000 Namur, Belgium ‡ Departament d’Enginyeria Química, E. T. S. d’Enginyeria Industrial de Barcelona, Universitat Politecnica de Catalunya, Diagonal 647, 08028 Barcelona, Spain § Center for Research in Nano-Engineering, Universitat Politecnica de Catalunya, Campus Sud, Edifici C0 , C/Pasqual i Vila s/n, Barcelona E-08028, Spain

bS Supporting Information ABSTRACT: We here describe the investigation at the atomistic level of the structure, stability, and dynamics of several complexes resulting from the interaction of oxidized poly(3,4-ethylenedioxythiophene) with the well-known Dickerson’s dodecamer sequence. Four specific arrangements have been selected as referential structures for molecular dynamics simulations, and the resulting independent trajectories tend to converge in two distinguishable models with the strongest interactions. The first one presents a coiled DNA strand enveloping the oligomer chain, whereas in the second model, the conducting polymer chain and the disorganized DNA strand are facing sideby-side. Analysis of the intermolecular interactions indicates that the electrostatic interactions involving the negatively charged DNA phosphates and the positively charged units of the oligomer are much more frequent in the first model. In addition, aside from these electrostatic interactions, specific O 3 3 3 H and S 3 3 3 H hydrogen bonds, π-π stacking, and NH 3 3 3 π interactions have been detected. Among all of these four specific interactions, we show that the π-π stacking is the most abundant and shows the best stability, whereas O 3 3 3 H hydrogen bonds are also frequent with long lifetimes. At the end, we have to underline that these specific interactions are predominant for the thymine and the guanine, which is in perfect agreement with previous experimental observations.

’ INTRODUCTION The control of the interactions between π-conjugated polymers and bioentities, for example living cells,1-4 proteins,5-7 and DNA,8-20 is not only an exciting but also an essential research area for the development of advanced biotechnological applications. Within this context, π-conjugated polymers have proven utility in a range of biosensor applications based on DNA detection.19,20 Motivated by these applications, several groups have focused their investigations on the interactions of cationic (oxidized) π-conjugated polymers and oligomers with DNA. Specifically, both electrostatic and hydrophobic interactions are implicated in the formation of the complexes,14-16 singlestranded DNA being identified as a more effective quencher than the double-stranded DNA of the π-conjugated polymers because of the greater flexibility and increased hydrophobicity of the former.17,18 We have recently found that π-conjugated polymers bearing polar functional groups are able to act as hydrogen bonding donors, acceptors, or both.10,21-25 For example, the side groups of some substituted poly(thiophene) derivatives are able to form specific interactions with well-defined nucleotide sequences of r 2011 American Chemical Society

plasmid DNA.10,21,23 This selectivity reflects that π-conjugated polymers 3 3 3 DNA adducts are stabilized not only by electrostatic nonspecific interactions but also by additional interactions that depend on the spatial disposition and orientation of the chemical groups. Among specific interactions, hydrogen bonds have been found to be more important than other weak interactions, like π-π stacking and hydrophobic.24 In a very recent experimental study involving a π-conjugated polymer in both oxidized (positively charged) and reduced (neutral) states, we proposed a mechanism for the formation of the adducts with DNA25 that consists of an initial stabilization of the complexes through nonspecific electrostatic interactions, followed by small structural rearrangements that allow establishing specific hydrogen bonds between the polar groups of the π-conjugated polymer and selected DNA bases. This mechanism requires a structural alteration of the B-DNA double helix, which undergoes a drastic transformation, as observed by circular dichroism and UV-vis Received: January 5, 2011 Revised: February 11, 2011 Published: February 25, 2011 1298

dx.doi.org/10.1021/bm200022n | Biomacromolecules 2011, 12, 1298–1304

Biomacromolecules

ARTICLE

Scheme 1

spectroscopy. Therefore, the very high degree of exposition detected for DNA bases was attributed to the effect of the πconjugated polymer, which promotes the DNA unfolding into two separate strands.21,25 Poly(3,4-ethylenedioxythiophene), hereafter abbreviated PEDOT (Scheme 1), is a π-conjugated polymer with particular abilities to interact with plasmid DNA.10,23,25 PEDOT has attracted considerable interest because of a combination of properties: low oxidation potential, good optical transparency, high conductivity (up to 500 S/cm), exceptional thermal and chemical stabilities, fast doping-undoping processes, and excellent biocompatibility.26-33 Electrophoretic and spectroscopic studies on mixtures of plasmid DNA and both oxidized and reduced PEDOT have shown the formation of stable adducts, the formation of interactions with specific nucleotide sequences being evidenced through the protection imparted by this material against restriction enzymes.10,23,25 Furthermore, sophisticated first-principle calculations using the MP2 quantum mechanical method indicated that the binding strength between the 3,4ethylenedioxythiophene (EDOT) monomeric unit and DNA bases grows in the following order: adenine (A) < cytosine (C) < guanine (G) ≈ thymine (T).23 In this work, we provide microscopic details of the interactions between oxidized PEDOT and DNA, discussing both the structure of the PEDOT 3 3 3 DNA complexes and the impact of the specific interactions on their stability. For this purpose, atomistic molecular dynamics (MD) simulations have been carried out considering two interacting molecules immersed in a water solvent box. Specifically, independent trajectories have been computed for the four different systems, which differ in the relative disposition of the PEDOT molecule and the DNA strand.

’ MOLECULAR MODELS PEDOT has been represented using an oligomer of 20 repeating units in the oxidized state, a net positive charge being located at every two repeating units. The number of positive charges supported by each monomeric unit of PEDOT produced under the experimental conditions used to study the interactions with plasmid DNA was found to be þ0.549.30 It should be noted that this oxidation degree, which was determined by standard ion chromatography, is accurately reproduced by the model used in this work for PEDOT, that is, [(EDOT0.539þ)n(ClO4-)0.539n]solid ≈ [(EDOT2)þCl-]n, where ClO4- refers to the counterions used in the experiments. Recent studies based on PEDOT samples with oxidation degrees ranging from 0.14 to 1.05 positive charges per repeating unit, which were obtained by modifying (reducing and oxidizing, respectively) the polymer after electrodeposition, indicated that the interaction with DNA is affected by the doping

Figure 1. Stick representation of the two molecular chains in the four models considered for the PEDOT 3 3 3 DNA complex at the beginning of the MD simulations.

level.25 However, in this work, we considered only the chemical characteristics of PEDOT, as frequently prepared (i.e., the conditions used in ref 30). DNA has been simulated considering a single strand with sequence 50 -CGCGAATTCGCG-30 . It should be noted that the latter sequence corresponds to Dickerson’s dodecamer,33 a well-known sequence with three primary characteristic tracts: CG, AA, and TT. The formation of stable PEDOT 3 3 3 DNA adducts has been investigated considering four molecular models that differ in the relative orientation of the two chains. These models, which have been used as starting points in MD simulations, are represented in Figure 1 and can be described as follows: Model I. The DNA strand adopts the same conformation as that in the B-DNA double helix, whereas the π-conjugated backbone of the PEDOT chain faces the phosphate groups of the DNA. Model II. The DNA strand adopts the same conformation as that in B-DNA double helix, whereas the π-conjugated backbone of the PEDOT chain faces the bases of the DNA. Model III. The DNA strand presents a coiled conformation, which was generated to wrap the PEDOT chain. Model IV. The DNA strand shows a randomly generated coiled conformation, which is facing the PEDOT chain. Each of these four models was placed in the center of orthorhombic simulation box with dimensions 47.46  47.46  126.55 Å3, full of previously equilibrated water molecules. The box dimensions were chosen to avoid biased results that would result in the violation of periodic boundary conditions. The solvent consists of 9992 water molecules (dilute aqueous solution of DNA), previously equilibrated at constant pressure (1 atm) and constant temperature (298 K), that is, NPT conditions. Fifteen positively charged sodium atoms and fourteen negative charged chloride atoms were added to the simulation box to reach electric neutrality.

’ COMPUTATIONAL DETAILS All MD trajectories were generated using the scalable computer program NAMD.34 The energy was calculated using the AMBER all1299

dx.doi.org/10.1021/bm200022n |Biomacromolecules 2011, 12, 1298–1304

Biomacromolecules atom force-field.35 The parameters required for DNA were taken from AMBER libraries,35,36 whereas water molecules were represented using the TIP3 model.37 Equilibrium parameters for EDOT units were derived from quantum mechanical calculations at the UB3LYP/6-31þG(d,p) level.38 All other force-field parameters, with the obvious exception of the electrostatic charges, were adapted from the AMBER libraries. Atomic electrostatic parameters for oxidized EDOT units (i.e., EDOTþ0.5) were derived from the ESP charges of (EDOT)42þ calculated at the UHF/6-31þG(d,p) level. In all simulations, the numerical integration step is 2 fs, and atom pair distance cutoffs are applied at 14.0 Å to compute nonbonding van der Waals interactions. To avoid discontinuities in the Lennard-Jones potential, a switch function is applied to allow a continuum decay of the energy when the atom pair distances are >12.0 Å. For electrostatic interactions, we compute the nontruncated electrostatic potential throughout Ewald summations.39 The real space term was determined by the van der Waals cut off (14 Å), whereas the reciprocal term was estimated by interpolation of the effective charge into a charge mesh with a grid thickness of five points per volume unit, that is, particle-mesh Ewald (PME) method.39 Periodic boundary conditions were applied using the nearest image convention, and the nonbonded pair list was updated every 1000 steps (1 ps). Bond lengths and bond angles of EDOT units were frozen at their equilibrium values, which were extracted from quantum mechanical calculations.38 The molecular parameters of DNA were not restrained. To facilitate intermolecular rearrangements, the inter-ring dihedral angles of the PEDOT molecule were allowed to rotate. Models I and II were manually constructed using graphic software. Model III was obtained by generating a random coiled DNA, which wraps the all-anti PEDOT molecule, followed by a gas-phase optimization using the conjugated gradient for 5.0  103 steps to relax structural tensions. The DNA arrangement used to construct model IV was derived from a 50 ns NPT MD simulation in the gas phase; that is, counterions (Naþ) to reach the electric neutrality were the only elements introduced in the simulation box. Once constructed, model IV was minimized similarly to model III. The four models were equilibrated prior to any production trajectory calculation. Consecutive rounds of MD short runs were performed to equilibrate the density of the system as well as its temperature and pressure. Therefore, once the complex was placed at the center of the simulation box, the solvent structure was optimized by keeping the complex frozen (belly conditions) and allowing the water molecules to move freely using an NPT simulation (1 atm and 298 K). Next, the entire system was brought to the simulation conditions. First, the energy was minimized by performing 1.0  105 steps of steepest descent. After this, two consecutive MD short runs were performed: 60 ps of NVT MD (thermal relaxation at 298 K) and 50 ps of NPT MD (isobaric relaxation at 1 atm and 298 K). Both temperature and pressure were controlled by the weak coupling method, the Berendsen thermobarostat,40 using a time constant for heat bath coupling and a pressure relaxation time of 2 ps. All NPT production simulations (after thermal and isobaric relaxation) were 10 ns long, and coordinates were stored every 1 ps for subsequent analysis. Interactions were included on the basis of the following geometric criteria: (a) For electrostatic interactions between the EDOT units and the phosphate groups, the distance between the interacting atoms is shorter than 6.5 Å. (b) For hydrogen bonds, the O 3 3 3 H or S 3 3 3 H distance is shorter than 3.0 Å. (c) For π-π stacking, the distance between the centers of mass of the stacked rings is shorter than 4.0 Å. (d) For N-H 3 3 3 π interactions, the distance between the hydrogen atom and the center of mass of the ring is shorter than 3.0 Å.

’ RESULTS AND DISCUSSION Figure 2 displays the root-mean-square deviations (rmsd’s) with respect to the ideal models (Figure 1). The high rmsd values

ARTICLE

Figure 2. Evolution of the root-mean-square deviation (rmsd) for the four PEDOT 3 3 3 DNA models relative to the corresponding initial structures. (See Figure 1.)

Figure 3. Stick representation of the two molecular chains in the four models considered for the PEDOT 3 3 3 DNA complex at the end of the MD simulations.

evidence a drastic reorganization of the four complexes. Complexes I, II, and IV are the least stable; the average rmsd’s calculated with respect to the initial arrangements are 34.57 ( 2.87, 35.29 ( 2.42, and 41.73 ( 0.58 Å, respectively. Figure 3 displays the molecular organization after a 10 ns MD and shows that I and II undergo significant changes in both the CP and the DNA. In both complexes, the PEDOT significantly bends, whereas the starting regular conformation of the DNA transforms into a nonregular distorted arrangement, even though the strand retains a quite elongated shape. These two structures present significant differences with respect to that obtained for IV at the end of the simulation. Therefore, in IV, the DNA presents a coiled conformation with a lower degree of elongation than in I and II; additionally, the bending of the PEDOT chain is higher. Analysis of the bending of the PEDOT chain, which has been determined by considering the angle defined by the geometric center of the whole molecule and the geometric centers of the first and last repeating units, indicates a 1300

dx.doi.org/10.1021/bm200022n |Biomacromolecules 2011, 12, 1298–1304

Biomacromolecules

Figure 4. Evolution of the distance between the geometric centers of the two molecules (R) for the four PEDOT 3 3 3 DNA models.

deformation with respect to the ideally planar conformation of 32.52 ( 1.60, 32.64 ( 3.25, and 38.43 ( 2.13° for I, II, and IV, respectively. The rmsd of III (14.45 ( 1.90 Å) is significantly smaller than those obtained for the other three complexes. This should be attributed not only to the fact that the DNA strand preserves its initial disposition, wrapping the PEDOT chain (Figure 3), but also to the low degree of bending of the latter (25.75 ( 2.02°). Accordingly, the highest concordance between the initial and final structures was obtained for complex III. The most striking feature of the structural changes displayed in Figure 2 is that they are produced during the equilibration runs. Indeed, the rmsd of the snapshot recorded at the end of the equilibration process (i.e., just before the production runs) calculated with respect to the ideal model (Figure 1) is 34.57, 35.29, 14.45, and 41.73 Å for I, II, III, and IV, respectively. Several protocols including variations in the simulation time, introduction of restrictions to fix structural parameters during the relaxation and thermalization, slower heating processes, and so on were enforced to minimize the apparition of oversized structural reorganizations. In all cases of simulations, the simulations converged to structures similar to those above-described. These features allow us to conclude that the interaction between the CP and the DNA strand does not follow an ideal organization such as those displayed in Figure 1 but a structure in which the DNA is completely coiled. Furthermore, the formation of the complexes is very rapid and, once achieved, the adducts only undergo small rearrangements, as is reflected by the relatively small fluctuations of the RMSDs during the production trajectories (Figure 2). These results are fully consistent with the hypotheses proposed in previous works using indirect experimental observations.21,23,25 Figure 4 illustrates the temporal evolution of the distance between the geometric centers of the PEDOT and DNA chains (R) for the four models under study. As can be seen, R remains relatively stable during whole trajectories with the average values of 5.81 ( 0.60, 3.92 ( 0.47, 3.17 ( 0.42, and 8.06 ( 0.39 Å for I, II, III, and IV, respectively. These values correlate not only with the bending deformation of the PEDOT chain but also with the distortion undergone by the DNA strand; that is, the average rmsd of the DNA strand calculated with respect to the initial model is 6.98 ( 1.35, 4.20 ( 1.33, 0.84 ( 0.56, and 10.66 ( 1.22 for I, II, III, and IV, respectively. The average total energy (ET) of the four simulated systems, which is provided in the Supporting Information (Table S1), indicates that the most favored complexes are II and IV, whereas

ARTICLE

I and III are less stable. The standard deviations are in all cases lower than 0.1%, indicating that the four complexes are well equilibrated; therefore, the fluctuations of ET remain notably small during the whole simulations. Analysis of the different contributions to ET indicates that the four complexes are dominated by the attractive electrostatic term (Eel), whereas the van der Waals contribution (EvdW) is repulsive in all cases. Finally, a residual contribution corresponds to the repulsive bonding energy (Eb), which corresponds to the difference between ET and the sum of Eel and EvdW. Thus, in all cases Eb is lower than EvdW. A more detailed analysis of the energy components is detailed in the Supporting Information (Table S2). The strength of the interaction between the DNA and PEDOT molecules is reflected by Eint and favors to I and, especially, III. In contrast, PEDOT 3 3 3 DNA intermolecular interaction energy is ∼40% higher for II and IV, even though these are the most stable complexes (Table S1 in the Supporting Information). Inspection of the nonbonding contributions indicates that the interaction between the two macromolecules is dominated by the electrostatic term in all cases (Table S2 in the Supporting Information). The PEDOT 3 3 3 DNA interaction is remarkably favored by electrostatics in I and III, as the weight of this contribution is ∼94% of Eint. This weight decreases to ∼74% in II and IV. Amazingly, the energy of the DNA (EDNA) is significantly more stable (∼60%) for II and, especially, for IV than for I and III, whereas the energy of PEDOT (EPEDOT) is similar in all cases. The contribution associated with the DNA hydration (EDNA,W) is least favorable for IV because the biomolecule is considerably screened by bending of the PEDOT chain (Figure 3). Because the ether groups of the dioxane rings are the only polar of the PEDOT chain, the influence of the intermolecular arrangement on the hydration of the CP (EPEDOT,W contribution) is very small. Summing of the contributions associated with the interaction energy PEDOT 3 3 3 DNA, the energy of DNA, and the energy of PEDOT (Eint þ EDNA þ EPEDOT in Table S2 in the Supporting Information) turn upside down the relative stability ranking. That is, complexes I and III become significantly more stable than II and IV, indicating that the higher stability of II and IV with respect to I and III (Table S1 in the Supporting Information) was due to the solvent effects. Moreover, the addition of DNA 3 3 3 water and PEDOT 3 3 3 water energy contributions to the previous sums (Eint þ EDNA þ EPEDOT þ EDNA,W þ EPEDOT, W, in Table S2 in the Supporting Information) clearly evidence that the preference toward II and IV (see the Supporting Information, Table S1) is an artifact and an exclusive consequence of the energy of water. As a result, we retain I and III as the more stable complexes once the solute 3 3 3 solvent contributions are added. In addition to the obvious electrostatic interactions between the charged phosphate groups of DNA and the charged repeating units of PEDOT, we looked for additional specific interactions in our four models. Accordingly, we traced the presence of H 3 3 3 O and S 3 3 3 H hydrogen bonds, π-π stacking, and N-H 3 3 3 π interactions between EDOT units and the individual nucleotide bases in all conformations recorded during the trajectories. Results, which are provided in the Supporting Information (Table S3), indicate that II and IV present the four classes of specific interactions. However, no N-H 3 3 3 π interaction was found in I nor in III. It is worth noting that the O 3 3 3 H hydrogen bonds in the four complexes involve G, whereas T is the base with least participation in such type of interactions. Other interesting 1301

dx.doi.org/10.1021/bm200022n |Biomacromolecules 2011, 12, 1298–1304

Biomacromolecules

ARTICLE

Figure 5. Accumulated lifetime (in %) of (a) π-π stacking, (b) N-H 3 3 3 π, (c) S 3 3 3 H, and (d) O 3 3 3 H PEDOT 3 3 3 DNA interactions in complexes I, II, III, and IV. Within each category, each interaction with a given nucleotide base is represented by an individual bar.

features are that T and G are involved in the π-π interactions and that N-H 3 3 3 π interactions are restricted to T and G. Additionally, we look for recognition at the primary (CG, AA, and TT) and secondary (GA, AT, and TC) tracts. It is worth noting that II is able to recognize the CG, AT, and TT tracts, even though in the former tract specific interactions are predominantly formed with guanine. III and IV show preferences toward the AT and CG tracts, respectively. Recognition of tracts with a higher number of bases was not identified by the models under study. These results combined with those obtained for individual bases suggest that the molecular recognition exerted by the PEDOT chain involves individual bases rather than tracts. The accumulated lifetimes of the specific interactions, which have been expressed in percentage, are schematically represented in Figure 5. The most frequent interaction is, independently of both the model and the nucleotide bases, the π-π stacking. This interaction is particularly relevant in complexes II and IV. In the former complex, the thiophene rings interact with six different bases (2 T, 1 G, and 3 C), whereas in I, III, and IV the number of π-π stacking interactions reduces to 3, 1, and 4, respectively. The distinctive feature of IV is the stability of the interactions, which remains formed for 85% (T) or even the whole simulation time (A, T, and G). On the other hand, Figure 5 shows that both the N-H 3 3 3 π and the S 3 3 3 H are not frequent nor particularly stable interactions. Although there are six bases (2 T and 4 G)

able to participate in the N-H 3 3 3 π interactions of II, the largest accumulated lifetime is only 15% with average lifetimes of ∼2%. Complexes I, III, and IV do not form any N-H 3 3 3 π interaction during the whole MD simulation. Regarding the S 3 3 3 H hydrogen bonds, I shows both the largest number of interacting bases (1 A and 1 C) and the largest accumulated lifetime (16.8% for C). The results obtained for the O 3 3 3 H hydrogen bond are particularly engaging. Specifically, complex III, which shows the strongest PEDOT 3 3 3 DNA interaction (Table S2 in the Supporting Information) forms O 3 3 3 H hydrogen bonds with six nucleotide bases, three of them being G (i.e., the fourth G of the DNA strand participates in the S 3 3 3 H hydrogen bond). Moreover, the accumulated lifetime of one of such O 3 3 3 H interactions with G is 97%, whereas the other two are 10 and 17%. Finally, the number of electrostatic interactions (i.e., the number of EDOT units that interact with at least one of the 11 phosphate groups contained in the DNA strand) was found to be 3, 5, 7, and 3 for I, II, III, and IV, respectively. This is feature evidence that the strength of the PEDOT 3 3 3 DNA interaction in III, which is reflected by the lowest Eint, is essentially due to the large number of electrostatic interactions. This conclusion is fully consistent with the analysis of the different terms that contribute to Eint (Table S2 in the Supporting Information). Once such interactions have been established, the results displayed in Figure 5 indicate that additionally the PEDOT chain forms specific 1302

dx.doi.org/10.1021/bm200022n |Biomacromolecules 2011, 12, 1298–1304

Biomacromolecules interactions with the DNA strand. These interactions preferentially involve the G bases when the DNA strand wraps the polymer chain. The average intermolecular distances for the specific interactions detected in complexes I-IV between the CP and the DNA strand are listed in the Supporting Information (Table S4). The most relevant features can be summarized as follows. In all cases, specific interactions remain stable during the whole simulation lifetime, as is reflected by the low standard deviations. Thus, the temporal evolution of the different interactions indicates that, once formed, the intermolecular interactions retain a length around the equilibrium values with very small oscillations until they break up (data not shown). The strongest specific interactions, which are identified by the smallest intermolecular distances, are present in III. In this complex, the interactions between the two molecules involve not only O 3 3 3 H and S 3 3 3 H hydrogen bonds with very short distances but also very strong π-π stacking interactions with T and G. Finally, the most striking observation refers to the high specific affinity of PEDOT toward selected nucleotide bases. Independently of the model, the formation of specific interactions with C and A is significantly less frequent than with T and, especially, G. In general terms, the two former bases do not involve N-H 3 3 3 π interactions or S 3 3 3 H hydrogen bonds, and the π-π stacking interactions are identified in only half of the models. In contrast, all four types of specific interactions are detected for T and G, π-π stacking and the O 3 3 3 H hydrogen bond being the most abundant. Finally, it should be noted that prediction and understanding at the molecular level of the influence of the DNA in the optical and electric properties of PEDOT are highly desirable. However, calculation of these properties on PEDOT 3 3 3 DNA adducts requires sophisticated quantum mechanical methods combined with large basis sets. Unfortunately, at present time, application of such high -level quantum mechanical calculations is restricted to systems with less than 100 atoms because they require a huge amount of computational resources.

’ CONCLUSIONS The structure, stability, and dynamics of complexes formed by PEDOT in the oxidized state with a single-stranded DNA featuring Dickerson’s dodecamer sequence have been investigated using atomistic MD simulations. Four models have been used as starting structures for independent trajectories, which eventually converged in two different models. In one of them, the DNA strand adopts a coiled conformation wrapping the PEDOT chain (III), whereas in the other, the PEDOT chain and the disordered DNA strand, which can be present shapes with different degrees of elongation (i.e., very elongated as in I and II or closer to a coiled shaped like in IV), are faced side-by-side. Although complexes II and IV are the most stable in terms of total energy, the strongest PEDOT 3 3 3 DNA interactions correspond to I and III. Indeed, the latter two complexes become the more stable after the internal energy of water is removed. A detailed analysis indicates that the interactions between the negatively charged phosphate groups and the positively charged repeating units of PEDOT are more frequent in III, and this should be attributed to the particular disposition of the DNA strand covering the CP. In addition to the electrostatic interactions, the formation of specific O 3 3 3 H and S 3 3 3 H hydrogen bonds, π-π stacking, and N-H 3 3 3 π interactions have been detected. In general terms, π-π stacking interactions have been

ARTICLE

found to be the most frequent and stable interactions, whereas O 3 3 3 H hydrogen bonds are also very abundant and show relatively large accumulated lifetimes, especially for III. In terms of occurrence, all of these specific interactions are more frequent with T and G than with A and C. These preferences are particularly remarkable for III, in which the four G bases form hydrogen bonds with the PEDOT chain.

’ ASSOCIATED CONTENT

bS

Supporting Information. Total energies, contributions to the total energy (van der Waals and electrostatics), energetic analysis describing intra- and intermolecular interactions, and description of the specific interactions and average distances for the specific interactions. This material is available free of charge via the Internet at http://pubs.acs.org.

’ AUTHOR INFORMATION Corresponding Author

*E-mail: [email protected] (J.P.); [email protected] (C.A.).

’ ACKNOWLEDGMENT This work has been supported by MICINN and FEDER funds (project number MAT2009-09138) and by the DIUE of the Generalitat de Catalunya (contract numbers 2009SGR925 and XRQTC). Computer resources were generously provided by the “Centre de Supercomputacio de Catalunya” (CESCA) and the “Barcelona Supercomputing Center” (BSC). J.P. and E.A.P. thank the Belgian National Fund for Scientific Research (FNRS) for their respective positions. Support for the research of C.A. was received through the prize “ICREA Academia” for excellence in research funded by the Generalitat de Catalunya. ’ REFERENCES (1) Liu, B.; Bazan, G. C. Chem. Mater. 2004, 16, 4467–4476. (2) Green, R. A.; Lovell, N. H.; Poole-Warren, L. A. Biomaterials 2009, 30, 3637–3644. (3) Lee, J. W.; Serna, F.; Nickels, J.; Schmidt, C. E. Biomacromolecules 2006, 7, 1692–1695. (4) del Valle, L. J.; Estrany, F.; Armelin, E.; Oliver, R.; Aleman, C. Macromol. Biosci. 2008, 8, 1144–1151. (5) Kros, A.; van Howell, S. W. F. M.; Sommerdijk, N. A. S. M.; Nolte, R. J. M. Adv. Mater. 2001, 13, 1555–1557. (6) Azioune, A.; Chehimi, M. M.; Miksa, B.; Basinska, T.; Slomkowski, S. Langmuir 2002, 18, 1150–1156. (7) Sanghvi, A. B.; Miller, K. P. H.; Belcher, A. M.; Schmidt, C. E. Nat. Mater. 2005, 4, 496–502. (8) Peng, H.; Zhang, L.; Spires, J.; Soeller, C.; Travas-Sejdic, J. Polymer 2007, 48, 3413–3419. (9) Yamamoto, T.; Shimizu, T.; Kurokawa, E. React. Funct. Polym. 2000, 43, 79–84. (10) Ocampo, C.; Armelin, E.; Estrany, F.; del Valle, L. J.; Oliver, R.; Sepulcre, F.; Aleman, C. Macromol. Mater. Eng. 2007, 292, 85–94. (11) Fan, Y.; Chen, X. T.; Trigg, A. D.; Tung, C. H.; Kong, J. M.; Gao, Z. Q. J. Am. Chem. Soc. 2007, 129, 5437–5443. (12) Dawn, A.; Nandi, A. K. Macromolecules 2005, 38, 10067–10073. (13) Sun, C. J.; Gaylord, B. S.; Hong, J. W.; Liu, B.; Bazan, G. C. Nat. Protoc. 2007, 2, 2148–2151. (14) Xu, Q. H.; Gaylord, B. S.; Wang, S.; Bazan, G. C.; Moses, D.; Heeger, A. J. Proc. Natl. Acad. Sci. U.S.A. 2004, 101, 11634–11639. 1303

dx.doi.org/10.1021/bm200022n |Biomacromolecules 2011, 12, 1298–1304

Biomacromolecules

ARTICLE

(15) Xu, Q. H.; Wang, S.; Korystov, D.; Mikhailovsky, A.; Bazan, G. C.; Moses, D.; Heeger, A. J. Proc. Natl. Acad. Sci. U.S.A. 2005, 102, 530–535. (16) Gaylord, B. S.; Heeger, A. J.; Bazan, G. C. J. Am. Chem. Soc. 2003, 125, 896–900. (17) Wang, S.; Liu, B.; Gaylord, B. S.; Bazan, G. C. Adv. Funct. Mater. 2003, 13, 463–467. (18) Xia, F.; Xiaolei, Z.; Yang, R.; Xiao, Y.; Kang, D.; Vallee-Belise, A.; Gong, X.; Heeger, A. J.; Plaxco, K. V. J. Am. Chem. Soc. 2010, 132, 1252–1254. (19) Liu, B.; Bazan, G. C. Proc. Natl. Acad. Sci. U.S.A. 2005, 102, 589–593. (20) Kim, Y.; Whitten, J. E.; Swager, T. M. J. Am. Chem. Soc. 2005, 127, 12122–12130. (21) Teixeira-Dias, B.; del Valle, L. J.; Estrany, F.; Armelin, E.; Oliver, R.; Aleman, C. Eur. Polym. J. 2008, 44, 3700–3707. (22) Pfeiffer, P.; Armelin, E.; Estrany, F.; del Valle, L. J.; Cho, L. Y.; Aleman, C. J. Polym. Res. 2008, 15, 225–234. (23) Aleman, C.; Teixeira-Dias, B.; Zanuy, D.; Estrany, F.; Armelin, E.; del Valle, L. J. Polymer 2009, 50, 1965–1974. (24) Zanuy, D.; Aleman, C. J. Phys. Chem. B 2008, 112, 3222–3230. (25) Teixeira-Dias, B.; Zanuy, D.; del Valle, L. J.; Estrany, F.; Armelin, E.; Aleman, C. Macromol. Chem. Phys. 2010, 211, 1117–1126. (26) Jonas, F.; Shrader, L. Synth. Met. 1991, 41, 831–836. (27) Heywang, G.; Jonas, F. Adv. Mater. 1992, 4, 116–118. (28) Dietrich, M.; Heinze, J.; Heywang, G.; Jonas, F. J. Electroanal. Chem. 1994, 369, 87–92. (29) Chiu, W. W.; Sejdic, J. T.; Cooney, R. P.; Bowmaker, G. A. Synth. Met. 2005, 155, 80–88. (30) Ocampo, C.; Oliver, R.; Armelin, E.; Aleman, C.; Estrany, F. J. Polym. Res. 2006, 13, 193–200. (31) Tamburri, E.; Orlanducci, S.; Toschi, F.; Terranova, M. L.; Passeri, D. Synth. Met. 2009, 159, 406–414. (32) del Valle, L. J.; Aradilla, D.; Oliver, R.; Sepulcre, F.; Gamez, A.; Armelin, E.; Aleman, C.; Estrany, F. Eur. Polym. J. 2007, 43, 2342–2349. (33) Drew, H. R.; Wing, R. M.; Takano, T.; Broka, C.; Tanaka, S.; Itakura, K.; Dickerson, R. E. Proc. Natl. Acad. Sci. U.S.A. 1981, 78, 2179–2183. (34) Kale, L.; Skeel, R.; Bhandarkar, M.; Brunner, R.; Gursoy, A.; Krawtz, N.; Phillips, J.; Shinozaki, A.; Varadarajan, K.; Schulten, K. J. Comput. Phys. 1999, 151, 283–312. (35) Cornell, W. D.; Cieplak, P.; Bayly, C. I.; Gould, I. R.; Merz, K. M.; Ferguson, D. M.; Spellmeyer, D. C.; Fox, T.; Cadwell, J. W.; Kollman, P. A. J. Am. Chem. Soc. 1995, 117, 5179–5197. (36) Wang, J.; Cieplak, P.; Kollman, P. A. J. Comput. Chem. 2000, 21, 1049–1074. (37) Jorgensen, W. L.; Chandrasekhar, J.; Madura, J. D.; Impey, R. W.; Klein, M. L. J. Chem. Phys. 1983, 79, 926–935. (38) Casanovas, J.; Aleman, C. J. Phys. Chem. C 2007, 111, 4823–4830. (39) Darden, T.; York, D.; Pedersen, L. J. Chem. Phys. 1993, 98, 10089–10092. (40) Berendsen, H. J. C.; Postma, J. P. M.; van Gunsteren, W. F.; Di Nola, A.; Haak, J. R. J. Chem. Phys. 1984, 81, 3684–3690.

1304

dx.doi.org/10.1021/bm200022n |Biomacromolecules 2011, 12, 1298–1304