Orientational Binding of DNA Guided by the C - ACS Publications

Mar 13, 2017 - (A) Representative snapshot of dsDNA binding on the C2N monolayer showing ..... (1) Srivastava, S.; Verma, A.; Frankamp, B. L.; Rotello...
0 downloads 0 Views 5MB Size
Download PDF
Orientational Binding of DNA Guided by the C2N Template

Zonglin Gu,† Lin Zhao,† Shengtang Liu,† Guangxin Duan,† Jose Manuel Perez-Aguilar,‡ Judong Luo,*,§ Weifeng Li,*,† and Ruhong Zhou*,†,‡,∥ †

Institute of Quantitative Biology and Medicine, SRMP and RAD-X, Collaborative Innovation Center of Radiation Medicine of Jiangsu Higher Education Institutions, Soochow University, Suzhou 215123, China ‡ Computational Biological Center, IBM Thomas J. Watson Research Center, Yorktown Heights, New York 10598, United States § Department of Oncology, The Affiliated Hospital of Nanjing Medical University, Changzhou No.2 People’s Hospital, Changzhou, 213003, China ∥ Department of Chemistry, Columbia University, New York, New York 10027, United States S Supporting Information *

ABSTRACT: A detailed understanding of the interactions between biomolecules and nanomaterial surfaces is critical for the development of biomedical applications of these nanomaterials. Here, we characterized the binding patterns and dynamics of a double stranded DNA (dsDNA) segment on the recently synthesized nitrogenized graphene (C2N) with both theoretical (including classical and quantum calculations) and experimental approaches. Our results show that the dsDNA repeatedly exhibits a strong preference in its binding mode on the C2N substrate, displaying an upright orientation that is independent of its initial configurations. Interestingly, once bound to the C2N monolayer, the transverse mobility of the dsDNA is highly restricted. Further energetic and structural analyses reveal that the strength and position of the binding is guided by the favorable π−π stacking between the dsDNA terminal base pairs and the benzene rings on the C2N surface, accompanied by a simultaneous strong nanoscale dewetting that provides additional driving forces. The periodic atomic charge distributions on C2N (from its unique porous structure) also cause the formation of local highly dense first solvation shell water clusters, which act as further steric hindrance for the dsDNA migration. Furthermore, free energy profiling calculated by the umbrella sampling technique quantitatively supports these observations. When compared to graphene, C2N is found to show a milder attraction to dsDNA, which is confirmed by experiments. This orientational binding of DNA on the C2N substrate might shed light on the design of template-guided nanostructures where their functions can be tuned by specialized biomolecular coating. KEYWORDS: nitrogenized graphene, C2N, orientational binding, biocompatibility, molecular dynamics, electrophoresis, porous nanostructure

T

with poly(ethylene glycol) (PEG) or serum proteins, graphene and graphene oxide display enhanced biocompatibility, which has prompted their use in drug delivery and imaging applications.31−37 Moreover, graphene surfaces containing nanopores have been proposed as potential candidates for DNA sequencing, where the engineered morphology of the graphene surface would regulate the migration and translocation of DNA.38−41 Recently, the successful synthesis of nitrogenized porous graphene monolayer, C2N, has attracted great interest.42−49 First, it exhibits an extremely high thermal stability, which can resist temperatures close to 700 °C. It also has a tunable wide-

he combination of biomolecules, such as DNA and proteins, with nanoscale materials has opened opportunities for the design of advanced composite nanomaterials with interesting optical, electronic, and mechanical properties.1−10 In biomedical applications, these composite nanomaterials can serve as therapeutic platforms for highly targeted and controlled drug delivery technologies. In this context, the specific interactions between biomolecules and nanomaterial surfaces are determining factors that guide the morphologies of the interfacial nanostructures with synergistic effects on their functions. In recent decades, two-dimensional (2D) nanomaterials have attracted great attention for biomedical applications11−13 due to their outstanding physicochemical properties.14−24 For instance, graphene and graphene derivatives (e.g., graphene oxide) can be used as highly efficient antimicrobial materials because their active surfaces exhibit high cytotoxicity.25−30 In contrast, when coated © 2017 American Chemical Society

Received: January 11, 2017 Accepted: March 13, 2017 Published: March 13, 2017 3198

DOI: 10.1021/acsnano.7b00236 ACS Nano 2017, 11, 3198−3206

Article

www.acsnano.org

Article

ACS Nano

Figure 1. (A) Top view of the crystal structure of the C2N monolayer; the C and N atoms are depicted as cyan and blue spheres, respectively. Initial configurations for simulation with dsDNA embedded in a water box adopting a perpendicular (B) and a parallel (C) orientation relative to the C2N surface. Potassium (K+) counterions are shown as yellow spheres.

Figure 2. (A) Representative snapshot of dsDNA binding on the C2N monolayer showing the upright dsDNA orientation. (B) Top view of the local contacting structure of the terminal base pairs (colored red) and the C2N surface. (C) Distribution contour map of two contacting bases on the C2N structure. (D) Time evolution of the number of contacts between dsDNA and C2N (black line) and the contacting angle defined between the dsDNA axial direction and C2N monolayer (red line). (E) The vdW (black) and Coulombic (red) energy interaction terms between dsDNA and C2N. (F) Number of intra-DNA hydrogen bonds as a function of time.

segment on the C2N monolayer is greatly restricted in the transverse directions. Energetic analyses and free energy profiling highlight the significance of the C2N porous structure for the biased dsDNA binding, where periodic energy wells and barriers exist for the binding and diffusion of the DNA base pairs. Although C2N demonstrates high capability to attract and interact with DNA, the internal dsDNA structure is well preserved upon surface binding. We believe the insights revealed from the C2N-template guided orientational DNA binding might find applications in the design of advanced nano−bio interfaces that require specific biocoating for their functions.

ranging band gap, implying its possible applications in electronic and optoelectronic devices.44,46 Moreover, C2N encapsulated with cobalt oxide exhibits high catalytic activity for hydrogen generation.47 The structure of C2N monolayer is shown in Figure 1 (also see Figure S1 in Supporting Information). Structurally, the C2N monolayer can be treated as sp2-hybridized carbon (C) rings, distributed in a honeycomb lattice, connected through nitrogen (N) bridges; six N atoms constitute a surface hole (Figure 1A). Due to the larger electronegativity of the nitrogen atom in the C−N bonds, there are intrinsic electron transfers in the C2N monolayer. This electron redistribution is expected not only to increase the solubility of C2N, which is an important feature desired for its biological usage, but also to engender distinctive properties that are absent in a pristine graphene surface. Based on these perspectives, we have explored the interactions of the C2N monolayer with a double stranded DNA (dsDNA) segment to illustrate its interactions with biomolecules. From molecular dynamics simulations, an orientational dsDNA binding on C2N is observed: the dsDNA structure binds to C2N through its terminal base pair and adopts a perpendicular orientation relative to the C2N surface. More interestingly, the translocation of the dsDNA

RESULTS AND DISCUSSION Stable Upright dsDNA Orientation with Preferential Binding Site. In our simulations, two initial configurations of dsDNA relative to C2N were considered: (i) with the dsDNA principal axis perpendicular to the C2N surface (indicated in Figure 1B and abbreviated as DNA⊥C2N); (ii) with the dsDNA principal axis parallel to the C2N surface (shown in Figure 1C and abbreviated as DNA∥C2N). Our current C2N model parameters, including atomic partial charges, are 3199

DOI: 10.1021/acsnano.7b00236 ACS Nano 2017, 11, 3198−3206

Article

ACS Nano

Figure 3. (A) Binding pathway process of dsDNA onto the C2N showing the characteristic orientational flipping. (B) Number of contacts between dsDNA and C2N as a function of time. (C) Time evolution of the contacting angle defined between the dsDNA axial direction and C2N monolayer; data was collected from three independent trajectories; (D) Distribution contour map of contacting bases. Results from panels A and C correspond to the trajectory colored in black in panel B.

a contact between DNA and C2N was considered if the distance between any DNA atom is within a distance of 6.0 Å of any C2N atom. The time evolution of the number of contacts is shown in Figure 2D, where two distinctive representative stages for the binding event are identified (although the detailed time scales may differ from different trajectories, the overall trend is consistent in all of them): (i) Early binding stage (14.5−21 ns), part of the dsDNA atoms have formed direct contacts with C2N (contacts ∼27). The binding is not strong enough to restrain the dsDNA motion so that the dsDNA can still diffuse along the C2N surface; (ii) Stable binding stage (22 ns and afterward), the terminal base pair has formed stable π−π stacking interactions with C2N, with the number of contacts reaching a maximum value (contacts ∼72). The position of the dsDNA is restricted at close proximity of the identified C2N binding site. The relative orientation of the dsDNA upon binding to C2N was quantitatively described by the angle formed between the C2N plane and the vector connecting the two dsDNA extremes. As illustrated in Figure 2D, during the 1000 ns simulation, the dsDNA always maintains the characteristic perpendicular orientational binding on the C2N monolayer. Interestingly, the formation of π−π stacking interactions is accompanied by a local nanoscale “dewetting” at the dsDNA− C2N interface. This dewetting process occurs at a very short time scales (∼21−22 ns), which is consistent with our previous findings55,56 (more discussions below). Similar trend is also observed when the interaction energy is decomposed into its van der Waals (vdW) and Coulombic (Coul) contributions (Figure 2E). At ∼21 ns, the magnitude of the vdW interactions displays an earlier and much faster decay than the Coulombic interactions, suggesting that the vdW contributions are dominant for the dsDNA binding process. The concerns about the biological safety and compatibility of nanomaterials are primary factors for their biological and medical applications since many nanomaterials are, in general, considered to be potentially toxic. For instance, graphene demonstrates strong capabilities to interact with biomolecules (such as protein, DNA, and lipids), albeit causing severe structural distortions.19,25,57,58 Along these lines, it is worth

obtained from quantum calculations and validated (see more details in MD Simulation and Supporting Information Figure S2). Periodic boundary conditions were used in all the simulations, with sufficiently large water boxes to avoid artifacts (Figure S3). From the analysis of the DNA⊥C2N simulations, we observed that the dsDNA segment diffuses freely in the solvent before approaching the C2N surface. Notably, in the three independent simulations, similar binding patterns of dsDNA binding onto the C2N surface were observed. As indicated in Figure 2A, the binding mode between dsDNA and the C2N monolayer shows an orientation of the axis of dsDNA in a perpendicular direction to the C2N monolayer. Details about the local contacting interface are illustrated in Figure 2B, where the two nucleotide bases located at the terminal position establish partial π−π stacking interactions with the C2N structure. These types of interactions (π−π stacking) are commonly found in composite materials formed by biomolecules and carbon-based nanomaterials, for example, graphene and various carbon nanotubes (CNTs).19,50 It is interesting that the migration of dsDNA in the transverse direction is highly restricted after binding, with the dsDNA maintaining an upright position (animation is supplied as Supporting Information). The distribution of the two contacting bases (AT base pairs in this case) on the C2N surface, as represented by their center of mass, is shown in Figure 2C. From this representation, the binding sites are identified to be highly localized nearby the “C−N bridges” (i.e., C−N bonds; see Figure 2C). Also, from one of the simulations, where the contacting bases were GC, the distribution of the contacting bases displays a similar profile as the one presented in Figure 2C, with again high occupancy above the “C−N bridges” (results of parallel trajectories can be found in Figures S4 and S5 in the Supporting Information). This is in sharp contrast with cases where biomolecular entities are adsorbed on either graphene or CNTs, where different studies suggest that the adsorbents can slide on the sp2-carbon surface rather uniformly.25,51−54 The dsDNA binding process is characterized by monitoring the intimate contacting atoms between dsDNA and C2N. Here, 3200

DOI: 10.1021/acsnano.7b00236 ACS Nano 2017, 11, 3198−3206

Article

ACS Nano

Figure 4. (A) Water density along the normal direction of the C2N plane, where the C2N is set at Z = 0. (B) Top view of the DNA−C2N local contacting area; only water molecules that are within 0.5 nm to DNA atoms are shown. (C) Potentials of mean force (PMF) of a nucleotide adenine moving along the C2N surface in two representative pathways as indicated by the red arrows. (D) Density map of the densest water layer (corresponding to peak 1 in panel A) on the C2N surface.

noticing that, in all our trajectories, the DNA duplex has wellmaintained its internal structure after binding to C2N. As can be seen in Figure 2F, the number of hydrogen bonds between the DNA strands remains almost constant throughout the simulations implying that its internal structure is not affected by the interaction with the C2N surface, which in turn suggests an adequate biocompatibility between C2N and dsDNA. Observation of dsDNA Orientations Flipping from Parallel to Upright. That the identified binding configurations of dsDNA onto the C2N surface may be strongly related to the initial configurations (and thus that the aforementioned preference in binding orientation may be an artifact of the limited sampling) was further investigated by including additional control simulations that started from a different initial configuration. As mentioned in the MD Simulation section, here we carried out additional simulations from a totally orthogonal configuration, the DNAC2N configuration (parallel to C2N surface), which was intended to corroborate and support the dsDNA binding specifics identified in the above DNA⊥C2N simulations. Overall, in the simulations starting from an initial parallel orientation (DNA∥C2N), an adsorption with a comparable binding mode to the above simulations with an initial perpendicular orientation (DNA⊥C2N) was observed. Figure 3A depicts representative configurations that illustrate the binding pathway process of one such trajectory of the DNA∥C2N system. The dsDNA, initially positioned parallel to the C2N monolayer, moved on the surface retaining its relative orientation in the beginning. At ∼18 ns, some atoms of the dsDNA began to directly interlock (i.e., anchoring) with the C2N surface (Figure 3A) with a number of contacts of ∼20. A few nanoseconds after the first interactions were established, a relatively fast “flipping” of the dsDNA structure was observed and completed at ∼42 ns, resulting in a perpendicular binding pattern similar to that

observed above for the DNA⊥C2N system. The number of contacts reached the maximum value (contacts ∼66) and lasted for the rest of simulation as shown in Figure 3B. The time evolution of the dsDNA angles (axial direction) with respect to the C2N for all the three trajectories is presented in Figure 3C, from which it is clear that the dsDNA flipping can be completed in a few nanoseconds (see animations in the Supporting Information), confirming that the upright dsDNA orientation is strongly favored. During the binding process, the contacting base pair can diffuse on the C2N monolayer with a higher distribution probability localized at the “C−N bridge” regions (see the migration path in Figure 3D), which is consistent with the our previous findings (Figure 2D). Results from the parallel trajectories also revealed similarities in the dsDNA binding kinetics, see Figures S6 and S7 in the Supporting Information. Interfacial Water Play a Significant Role in DNA Binding and Migration. The behavior of interfacial water and, particularly, the role of the C2N first solvation shell water on the dsDNA binding and migration are further examined in this subsection. As expected, the nonuniform partial charge distribution on the C2N surface (0.24e for C and −0.48e for N) result in an overall more hydrophilic surface of C2N compared to graphene. Figure 4A shows the water density along the normal direction of the C2N surface, where three detectable water density peaks (water layers) can be identified near the C2N surface. Especially interesting is the water layer that peaks at around −3.5 Å from the surface (labeled as peak 1), which displays a water density about 3-fold of that of bulk water. Despite the potential high local density (more below), these interfacial water molecules can be squeezed once the dsDNA terminal base pairs approach (as illustrated in Figure 4B) resulting in a so-called nanoscale dewetting (or drying), which provides a strong driving force for the dsDNA adsorption and 3201

DOI: 10.1021/acsnano.7b00236 ACS Nano 2017, 11, 3198−3206

Article

ACS Nano

Figure 5. Control simulations of dsDNA on graphene reveal the deleterious dsDNA denaturation on (A) one base pair and (B) two base pairs at the interface. (C) Number of contacts between dsDNA and graphene (Gra) as a function of time; the results of dsDNA with C2N are also shown for clear comparison. (D) local binding pattern of the dsDNA binding on the C2N. (E) Solutions of graphene and C2N. (F) Estimations of residual DNA ratios in the supernatants as indicated by the agarose gel electrophoresis of the supernatants. (G) cDNA concentration after treatment of C2N and graphene. Graphene can attract ∼52.3% percent of DNA, while C2N only attracts 6.2%. The control runs are indicated as “Con” in (F) and (G).

characteristic binding.55,56 After drying, the favorable π−π stacking between the dsDNA terminal base pairs and the benzene rings on the C2N surface results in a tight and stable binding. To further quantify the role of these interfacial (or first solvation shell) water molecules on the dsDNA migration along the C2N surface, we performed the potential of mean force (PMF) analyses by pulling an adenine nucleobase along two highly symmetric surface pathways as indicated in Figure 4C. The PMF is calculated using the umbrella sampling method (the pulling force is applied on the center of mass of the adenine nucleobase). The stable binding site, which corresponds to the free energy global minimum, is located above the “C−N bridges”, which further confirms the findings discussed above. Migration over the benzene ring encounters a barrier of 9.8 kJ/mol, which would likely be inaccessible at room temperature. In addition, there is also an energy barrier for adenine binding just on the top of the center of the porous region (i.e., those holes surrounded by the N atoms). The restrained dsDNA binding on the C2N surface is further confirmed by calculating the friction coefficient of dsDNA on C2N (taking graphene as a comparison). Our results clearly reveal a much higher friction of dsDNA with C2N than that with graphene (for detailed analyses, please refer to Figure S8 and accompanying descriptions in the Supporting Information). The structural periodicity of C2N has a crystal constant of 8.30 Å (hole-to-hole distance), which is compatible with the sizes of nucleotides and amino acids. The distribution of the first solvation shell water molecules is thus regulated by the surface porous structure where periodic energy wells and barriers exist. As illustrated in Figure 4B, there are water molecules occupying these porous structural cavities, corresponding to the intimate peak (peak 2 in Figure 4A). Detailed analyses reveal that the waters in this first solvation shell have their dipoles uniformly pointing to the C2N surface (Figure S9 in the Supporting Information). Consequently, the water oxygens (negatively charged) are exposed to the dsDNA adsorbate. As the dsDNA backbone is also negative charged,

the perpendicular orientational binding on C2N is the most favorable pattern, which minimizes the electrostatic repulsion energy. Moreover, the most dense water layer (peak 1 in Figure 4A) is also highly localized on the C2N monolayer, with most of the water molecules residing above the carbon benzene rings (Figure 4D). These local highly dense water clusters effectively obstruct the movement of the bound dsDNA, since the dsDNA base pairs have to squeeze out these water molecules during a transverse migration. To support these, we have conducted indepth analysis of the C2N and graphene surfaces by calculating a fundamental physical parameter, the friction. Our results clearly show that the friction of the dsDNA on C2N is rather larger, on the order of 102, than that of graphene (for detailed analyses, please refer to Figure S8 in the Supporting Information). This strongly supports the high energy barriers found for the dsDNA migration on C2N. Interestingly, the K+ ions do not play an important role in the dsDNA binding, partly because K+ ions are mostly located in the grooves of the dsDNA duplex (but not at its terminal base pairs) and partly because even if some K+ ions do bind onto the C2N surface, they are favorably positioned at the holes surrounded by the N atoms (through Coulombic attractions) and thus they do not block the partial π−π stacking between the dsDNA terminal base pairs and the C2N benzene rings (see Figure 2B). Control Simulations and Agarose Gel Electrophoresis Confirm a Stronger DNA Adsorption in Graphene than in C2N. To better disclose the structural significance of in dsDNA binding onto the C2N nanomaterial, we have considered control simulations of dsDNA binding on the structural analogue, graphene. Figure 5A,B illustrates two representative structures of the already bound dsDNA on graphene. In Figure 5A, the end base pair at the interface is denatured, that is, the interbase H-bonds are broken and the released bases form a direct binding with the graphene atoms; this is an indication of the strong attraction between dsDNA and graphene. In Figure 5B, an additional base pair is also denatured. During this process, the number of contacts between dsDNA and graphene also increases in a step by 3202

DOI: 10.1021/acsnano.7b00236 ACS Nano 2017, 11, 3198−3206

Article

ACS Nano

biocompatible 2D-nanomaterial of C2N. Meanwhile, the orientational binding of DNA on C2N substrate might also provide insights on the design of template-guided nanostructures where their functions can be tuned by specialized biomolecular coating at bio−nano interface.

step manner as shown in Figure 5C. The deleterious binding of graphene to dsDNA was also observed in the trajectory with an initial parallel dsDNA orientation (Figure S10 in Supporting Information) that has been reported in a previous study59 in which both multilayered graphene and carbon nanotube (CNT) arrays demonstrate an obvious dsDNA denaturation capacity through strong hydrophobic and π−π stacking interactions. The binding strength of dsDNA to graphene and C2N is quantitatively compared by calculating the binding affinities of nucleotides on each of the nanomaterials. Our PMF results (Figure S11 in Supporting Information) clearly indicate that the nucleotide loading on graphene is systematically stronger than that on C2N. Moreover, it is found that the strong graphene−dsDNA binding is not correlated with the restriction of the dsDNA transverse movement. By monitoring the dsDNA trajectories along the 1000 ns simulation, we found that the dsDNA covers a significantly broader area on the graphene surface, indicating that dsDNA can diffuse in a rapid and highly dynamic manner (Figure S12 in the Supporting Information). This is in sharp contrast to the aforementioned highly restricted binding mode of dsDNA on C2N. Additionally, from agarose gel electrophoresis experiments, we also found that there is a larger amount of residual DNA sample in the supernatant after exposure to C2N than to graphene (Figure 5E,F). This supports the predicted stronger absorption capacity of graphene for dsDNA, relative to the C2N surface. In detail, under the same experimental conditions, graphene can attract ∼52.3% of DNA, which is significantly larger than C2N (∼6.2%; Figure 5G). The experimental results have confirmed the findings from the MD simulations regarding the different absorption capacity of these two nanomaterials.

SIMULATION AND EXPERIMENTAL METHODS MD Simulation. The C2N model used in our simulations has a dimension of 83.2 × 86.4 Å2 and is constituted by 1440 C and 720 N atoms. The dsDNA model (sequence ATCGATCGATCGATCG) adopted a B-form and was constructed by the Nucleic Acid Builder (NAB) as implemented in the Amber Tools package.60 The dsDNA was initially placed above the C2N monolayer with a minimum distance of 15 Å. The different starting locations of dsDNA with respect to C2N sheet were generated by translating the dsDNA segment in the transverse directions (x and y) by 1 nm, resulting in three independent trajectories. The complex was then solvated in an 83.2 × 86.4 × 96.0 Å3 water box, with potassium ions added to neutralize the negative charge of dsDNA (ions were added by randomly replacing solvent molecules with monatomic ions with the genion tool in GROMACS). Periodic boundary conditions were used in all the simulations, and necessary checks were made to ensure that the DNA strands maintain a sufficient minimum distance to avoid direct interactions between images (Figure S3). The total number of water molecules in our simulations are 27769 for DNA⊥C2N system and 27820 for DNA//C2N system, respectively. All MD simulations were conducted with the GROMACS package.61 The AMBER99sb force field62 was employed (with “NB” and “CA” atom types utilized for the nitrogen and carbon atoms in C2N, respectively, to be specific). The atomic charges were further optimized using Gaussian 09 at a HF/6-31G* level and parametrized with the RESP method, yielding a value of 0.24e for C and −0.48e for N, respectively (the details of the calculations can be found in the Supporting Information). The TIP3P water model63 was utilized in all the simulated systems. As the coordinates of C2N atoms were frozen during the simulation, constant pressure coupling (1 atm) was applied at the Z-direction of the simulation box during the pre-equilibrium stage by the Berendsen barostat with a coupling coefficient of τP = 1 ps, followed by a production simulation phase conducted in the NVT ensemble (300 K) using velocity-rescale thermostat64 with a coupling coefficient of τT = 0.1 ps. The electrostatic interactions were treated using the particle mesh Ewald (PME) method65,66 (using fourth-order interpolation; the maximum Fourier spacing for the FFT gird is 0.12 nm; the electrostatic energy tolerance is 1 × 10−5), with the van der Waals (vdW) interactions calculated with a cutoff distance of 10 Å. All the bonds involving hydrogen atoms were maintained at their constant equilibrium lengths with the LINCS algorithm.67 Each system was first energy-minimized with the steepest descent algorithm, followed by 20 ns of pre-equilibration with position restraints applied on the dsDNA atoms (after 20 ns, the box length in the Z-direction and total energy converge), random initial velocities were assigned to dsDNA and solvent atoms. For each configuration, 1000 ns of data-production simulation was conducted. A time step of 2.0 fs was used, and data were collected every 1 ps. The validity of the classical C2N model used in our current study was further tested by considering a benchmark case, that is, the adsorption process of a water molecule probe on the C2N surface. The energy profiles corresponding to the adsorption process were generated with both MD calculations and quantum mechanics calculations. The energetic and structural characteristics were well reproduced (results can be found in the Supporting Information, Figure S2), confirming that the classical C2N model used here is adequate. Control simulations of the dsDNA binding with graphene (abbreviated as Gra) were also conducted to compare the different DNA kinetics on two nanomaterials. The carbon atoms of graphene were assigned as “CA” atom types in AMBER99sb force field.62 The graphene model used in our simulations has a dimension of 83.5 × 89.3 Å2 and is constituted by 2856 carbon atoms. We used similar

CONCLUSION In summary, we have investigated the structural and dynamical properties of a dsDNA binding onto a recently synthesized graphene derivative C2N monolayer, using both atomistic molecular dynamics simulations and agarose gel electrophoresis assays. Different orientations (perpendicular and parallel) of the dsDNA segment relative to this 2D-nanomaterial were employed. Our simulations of the orthogonal systems reveal that dsDNA can stably bind to the C2N monolayer through its terminal base pairs while preserving its internal structure. Remarkably, the binding mode presents a dsDNA molecule preferentially adopting an upright perpendicular orientation regardless of the dsDNA initial configuration. Once adsorbed, the transverse migration of the dsDNA is highly restricted on the C2N surface, with its upright orientation maintained. Through detailed analyses, two distinct structural properties that guide the orientational DNA binding have been highlighted: (1) The porous surface structure biases the dsDNA binding toward a preferential binding site near the “C−N bridges”, where the favorable π−π stacking can be formed between the dsDNA terminal base pairs and the C2N benzene rings. This close interaction is accompanied by a simultaneous nanoscale dewetting (drying), which provides additional driving force for the dsDNA adsorption and characteristic binding. (2) The periodic distribution of local highly dense solvation water clusters on the C2N further restricts the adsorbed dsDNA transverse migration. Both simulations and agarose gel electrophoresis assays confirm a stronger and more disruptive DNA adsorption onto graphene than C2N, indicating a more 3203

DOI: 10.1021/acsnano.7b00236 ACS Nano 2017, 11, 3198−3206

Article

ACS Nano *E-mail: [email protected] (R. H. Z.).

simulation protocols as those for the dsDNA−C2N simulations. The dsDNA binding on graphene, through a parallel orientation, has been extensively studied in a previous work,59 demonstrating that a fast denaturation of dsDNA occurs upon binding. Thus, here we have only considered the case for the orthogonal orientation, dsDNA⊥Gra. The total number of water molecules in our simulations are 27778, and the box dimensions are 83.5 × 89.3 × 96.3 Å3, which is comparable to the size of the C2N simulations. Two independent trajectories of 1000 ns were generated. Potential of Mean Force (PMF) Profiling. The PMF for nucleotide diffusion on C2N was calculated with an umbrella sampling method.68−70 The transverse distance (d) to the starting point along the representative path was restrained at a certain value (d0) with a harmonic force

ORCID

Weifeng Li: 0000-0002-0244-2908 Ruhong Zhou: 0000-0001-8624-5591 Notes

The authors declare no competing financial interest.

ACKNOWLEDGMENTS We thank the group of Prof. Jong-Beom Baek group at UNIST, South Korea, for providing us with the C2N samples for experiments. We thank Zaixing Yang, Seung-gu Kang, Tien Huynh, and Xuanyu Meng for helpful discussions. This work was partially supported by the National Natural Science Foundation of China (Grants 11304214, 11574224, and 21320102003). A Project Funded by the Priority Academic Program Development of Jiangsu Higher Education Institutions (PAPD), and Jiangsu Provincial Key Laboratory of Radiation Medicine and Protection. RZ acknowledges the support from IBM Blue Gene Science Program Program (Grants W1258591, W1464125, and W1464164).

F = k × (d − d0) where k is the force constant (2000 kJ mol−1 nm−2). The sampling window has a resolution of 0.05 nm. At each d0, the system was first equilibrated for 2 ns followed by a 10 ns productive simulation. PMF profiles were obtained with the g_wham tool that implements the Weighted Histogram Analysis Method. Experimental Section. Cell Culture. A549 cells were obtained from American Type Culture Collection (ATCC) (Rockville, MD). Cells were maintained in Dulbecco’s modified Eagle’s medium (DMEM, Gibco), supplemented with 10% fetal bovine serum (FBS), L-glutamine (5 mmol/L), penicillin (100 U/mL), and streptomycin (100 U/mL) (Invitrogen, Carlsbad, CA) at 37 °C in a humidified 5% CO2 atmosphere. Agarose Gel Electrophoresis. To quantitatively compare the DNA adsorption to C2N and graphene, we conducted agarose gel electrophoresis. The DNA sample was prepared from the A549 cells. Technically, RNA was first extracted from A549 cells using TRIzol reagent (Invitrogen, USA) and isolated. After that, 500 ng of RNA was used to produce DNA by the reverse transcriptase method (Invitrogen, USA). C2N (synthesized by Prof. Jong-Beom Baek group, UNIST, South Korea) or graphene (purchased from Chengdu Organic Chemical Company, Chinese Academy of Science) were added to the DNA solution with the concentration ratio of 1:1 for DNA loading. Pure DNA solution without any treatment was used as control. After 24 h treatment at room temperature, all the samples were centrifuged at 13500 rpm for 30 min. Finally, the supernatant was collected and analyzed by electrophoresis through 0.8% agarose gels containing 0.01% Gelred (Beyotime Institute of Biotechnology, Beijing, China). The gels were photographed under ultraviolet light. The density of all bands was analyzed with FluorChem M fluorescent chemiluminescence imaging analysis system (α technologies, USA) to give the residual DNA concentration in the supernatant. Error was estimated as mean squared deviations of three independent samples.

REFERENCES (1) Srivastava, S.; Verma, A.; Frankamp, B. L.; Rotello, V. M. Controlled Assembly of Protein-nanoparticle Composites through Protein Surface Recognition. Adv. Mater. 2005, 17, 617. (2) Calzolai, L.; Franchini, F.; Gilliland, D.; Rossi, F. ProteinNanoparticle Interaction: Identification of the Ubiquitin-Gold Nanoparticle Interaction Site. Nano Lett. 2010, 10, 3101−3105. (3) Lynch, I.; Salvati, A.; Dawson, K. A. Protein-Nanoparticle Interactions What Does the Cell See? Nat. Nanotechnol. 2009, 4, 546− 547. (4) Lynch, I.; Dawson, K. A. Protein-nanoparticle interactions. Nano Today 2008, 3, 40−47. (5) Tang, L.; Chang, H.; Liu, Y.; Li, J. Duplex DNA/Graphene Oxide Biointerface: From Fundamental Understanding to Specific Enzymatic Effects. Adv. Funct. Mater. 2012, 22, 3083−3088. (6) Zhang, Q.; Qiao, Y.; Hao, F.; Zhang, L.; Wu, S.; Li, Y.; Li, J.; Song, X.-M. Fabrication of a Biocompatible and Conductive Platform Based on a Single-Stranded DNA/Graphene Nanocomposite for Direct Electrochemistry and Electrocatalysis. Chem. - Eur. J. 2010, 16, 8133−8139. (7) Chong, Y.; Ge, C.; Yang, Z.; Garate, J. A.; Gu, Z.; Weber, J. K.; Liu, J.; Zhou, R. Reduced Cytotoxicity of Graphene Nanosheets Mediated by Blood-Protein Coating. ACS Nano 2015, 9, 5713−5724. (8) Ge, C.; Du, J.; Zhao, L.; Wang, L.; Liu, Y.; Li, D.; Yang, Y.; Zhou, R.; Zhao, Y.; Chai, Z.; Chen, C. Binding of Blood Proteins to Carbon Nanotubes Reduces Cytotoxicity. Proc. Natl. Acad. Sci. U. S. A. 2011, 108, 16968−16973. (9) Schwartz, M. P.; Alvarez, S. D.; Sailor, M. J. Porous SiO2 Interferometric Biosensor for Quantitative Determination of Protein Interactions: Binding of Protein A to Immunoglobulins Derived from Different Species. Anal. Chem. 2007, 79, 327−334. (10) He, P. L.; Hu, N. F. Electrocatalytic Properties of Heme Proteins in Layer-by-layer Films Assembled with SiO2 Nanoparticles. Electroanalysis 2004, 16, 1122−1131. (11) Feng, L.; Liu, Z. Graphene in Biomedicine: Opportunities and Challenges. Nanomedicine 2011, 6, 317−324. (12) Sanchez, V. C.; Jachak, A.; Hurt, R. H.; Kane, A. B. Biological Interactions of Graphene-Family Nanomaterials: An Interdisciplinary Review. Chem. Res. Toxicol. 2012, 25, 15−34. (13) Geim, A. K. Graphene: Status and Prospects. Science 2009, 324, 1530−1534. (14) Tielrooij, K. J.; Piatkowski, L.; Massicotte, M.; Woessner, A.; Ma, Q.; Lee, Y.; Myhro, K. S.; Lau, C. N.; Jarillo-Herrero, P.; van Hulst, N. F.; Koppens, F. H. L. Generation of Photovoltage in

ASSOCIATED CONTENT

S Supporting Information *

The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/acsnano.7b00236. Movie 1 showing one representative MD trajectory starting from the DNA⊥C2N initial configuration (AVI) Movie 2 showing one representative MD trajectory starting from the DNA∥C2N initial configuration (AVI) Further details and discussions on the quantum calculation setups, model validation, additional trajectories, friction coefficients, water orientation at interface, control runs of DNA on graphene, PMF calculations, and DNA diffusion on graphene (PDF)

AUTHOR INFORMATION Corresponding Authors

*E-mail: wfl[email protected] (W. F. L.). *E-mail: [email protected] (J. D. L.). 3204

DOI: 10.1021/acsnano.7b00236 ACS Nano 2017, 11, 3198−3206

Article

ACS Nano

thermal Therapy, and Magnetic Resonance Imaging. Nano Res. 2012, 5, 199−212. (34) Pan, Y.; Bao, H.; Sahoo, N. G.; Wu, T.; Li, L. Water-Soluble Poly(N-isopropylacrylamide)-Graphene Sheets Synthesized via Click Chemistry for Drug Delivery. Adv. Funct. Mater. 2011, 21, 2754−2763. (35) Sun, X.; Liu, Z.; Welsher, K.; Robinson, J. T.; Goodwin, A.; Zaric, S.; Dai, H. Nano-Graphene Oxide for Cellular Imaging and Drug Delivery. Nano Res. 2008, 1, 203−212. (36) Yang, K.; Zhang, S.; Zhang, G.; Sun, X.; Lee, S.-T.; Liu, Z. Graphene in Mice: Ultrahigh in vivo Tumor Uptake and Efficient Photothermal Therapy. Nano Lett. 2010, 10, 3318−3323. (37) Yue, H.; Wei, W.; Gu, Z.; Ni, D.; Luo, N.; Yang, Z.; Zhao, L.; Garate, J. A.; Zhou, R.; Su, Z.; Ma, G. Exploration of Graphene Oxide as an Intelligent Platform for Cancer Vaccines. Nanoscale 2015, 7, 19949−19957. (38) Sathe, C.; Zou, X.; Leburton, J.-P.; Schulten, K. Computational Investigation of DNA Detection Using Graphene Nanopores. ACS Nano 2011, 5, 8842−8851. (39) Qiu, H.; Sarathy, A.; Leburton, J.-P.; Schulten, K. Intrinsic Stepwise Translocation of Stretched ssDNA in Graphene Nanopores. Nano Lett. 2015, 15, 8322−8330. (40) Wells, D. B.; Belkin, M.; Comer, J.; Aksimentiev, A. Assessing Graphene Nanopores for Sequencing DNA. Nano Lett. 2012, 12, 4117−4123. (41) Shankla, M.; Aksimentiev, A. Conformational Transitions and Stop-and-go Nanopore Transport of Single-stranded DNA on Charged Graphene. Nat. Commun. 2014, 5, 5171. (42) Mahmood, J.; Lee, E. K.; Jung, M.; Shin, D.; Jeon, I.-Y.; Jung, S.M.; Choi, H.-J.; Seo, J.-M.; Bae, S.-Y.; Sohn, S.-D.; Park, N.; Oh, J. H.; Shin, H.-J.; Baek, J.-B. Nitrogenated Holey Two-dimensional Structures. Nat. Commun. 2015, 6, 6486. (43) Sakaushi, K.; Antonietti, M. Carbon- and Nitrogen-Based Organic Frameworks. Acc. Chem. Res. 2015, 48, 1591−1600. (44) Zhang, R.; Li, B.; Yang, J. Effects of Stacking Order, Layer Number and External Electric Field on Electronic Structures of Fewlayer C2N-H2D. Nanoscale 2015, 7, 14062−14070. (45) Zhu, L.; Xue, Q.; Li, X.; Wu, T.; Jin, Y.; Xing, W. C2N: An Excellent Two-dimensional Monolayer Membrane for He sSeparation. J. Mater. Chem. A 2015, 3, 21351−21356. (46) Sahin, H. Structural and Phononic Characteristics of Nitrogenated Holey Graphene. Phys. Rev. B: Condens. Matter Mater. Phys. 2015, 92, 085421. (47) Mahmood, J.; Jung, S.-M.; Kim, S.-J.; Park, J.; Yoo, J.-W.; Baek, J.-B. Cobalt Oxide Encapsulated in C2N-h2D Network Polymer as a Catalyst for Hydrogen Evolution. Chem. Mater. 2015, 27, 4860−4864. (48) Chaban, V. V.; Prezhdo, O. V. Nitrogen-Nitrogen Bonds Undermine Stability of N-Doped Graphene. J. Am. Chem. Soc. 2015, 137, 11688−11694. (49) Guan, S.; Cheng, Y.; Liu, C.; Han, J.; Lu, Y.; Yang, S. A.; Yao, Y. Effects of Strain on Electronic and Optic Properties of Holey Twodimensional C2N Crystals. Appl. Phys. Lett. 2015, 107, 231904. (50) Yang, Z.; Wang, Z.; Tian, X.; Xiu, P.; Zhou, R. Amino Acid Analogues Bind to Carbon Nanotube via pi-pi Interactions: Comparison of Molecular Mechanical and Quantum Mechanical Calculations. J. Chem. Phys. 2012, 136, 025103. (51) Yan, L. Y.; Li, W.; Fan, X. F.; Wei, L.; Chen, Y.; Kuo, J.-L.; Li, L.J.; Kwak, S. K.; Mu, Y.; Chan-Park, M. B. Enrichment of (8,4) SingleWalled Carbon Nanotubes Through Coextraction with Heparin. Small 2010, 6, 110−118. (52) Yan, L. Y.; Li, W.; Mesgari, S.; Leong, S. S. J.; Chen, Y.; Loo, L. S.; Mu, Y.; Chan-Park, M. B. Use of a Chondroitin Sulfate Isomer as an Effective and Removable Dispersant of Single-Walled Carbon Nanotubes. Small 2011, 7, 2758−2768. (53) Johnson, R. R.; Kohlmeyer, A.; Johnson, A. T. C.; Klein, M. L. Free Energy Landscape of a DNA-Carbon Nanotube Hybrid Using Replica Exchange Molecular Dynamics. Nano Lett. 2009, 9, 537−541. (54) Roxbury, D.; Jagota, A.; Mittal, J. Sequence-Specific SelfStitching Motif of Short Single-Stranded DNA on a Single-Walled Carbon Nanotube. J. Am. Chem. Soc. 2011, 133, 13545−13550.

Graphene on a Femtosecond Timescale through Efficient Carrier Heating. Nat. Nanotechnol. 2015, 10, 437−443. (15) Lee, W. C.; Kim, K.; Park, J.; Koo, J.; Jeong, H. Y.; Lee, H.; Weitz, D. A.; Zettl, A.; Takeuchi, S. Graphene-templated Directional Growth of an Inorganic Nanowire. Nat. Nanotechnol. 2015, 10, 423− 428. (16) Scrace, T.; Tsai, Y.; Barman, B.; Schweidenback, L.; Petrou, A.; Kioseoglou, G.; Ozfidan, I.; Korkusinski, M.; Hawrylak, P. Magnetoluminescence and Valley Polarized State of a Two-dimensional Electron Gas in WS2Monolayers. Nat. Nanotechnol. 2015, 10, 603. (17) Surwade, S. P.; Smirnov, S. N.; Vlassiouk, I. V.; Unocic, R. R.; Veith, G. M.; Dai, S.; Mahurin, S. M. Water Desalination using Nanoporous Single-layer Graphene. Nat. Nanotechnol. 2015, 10, 459− 464. (18) Liu, T.; Wang, C.; Gu, X.; Gong, H.; Cheng, L.; Shi, X.; Feng, L.; Sun, B.; Liu, Z. Drug Delivery with PEGylated MoS2 Nano-sheets for Combined Photothermal and Chemotherapy of Cancer. Adv. Mater. 2014, 26, 3433−3440. (19) Gu, Z. L.; Yang, Z. X.; Wang, L. L.; Zhou, H.; Jimenez-Cruz, C. A.; Zhou, R. H. The Role of Basic Residues in the Adsorption of Blood Proteins onto the Graphene Surface. Sci. Rep. 2015, 5, 10873. (20) Yang, G.; Gong, H.; Liu, T.; Sun, X.; Cheng, L.; Liu, Z. Twodimensional Magnetic WS2@Fe3O4 Nanocomposite with Mesoporous Silica Coating for Drug Delivery and Imaging-guided Therapy of Cancer. Biomaterials 2015, 60, 62−71. (21) Xi, Q.; Zhou, D.-M.; Kan, Y.-Y.; Ge, J.; Wu, Z.-K.; Yu, R.-Q.; Jiang, J.-H. Highly Sensitive and Selective Strategy for MicroRNA Detection Based on WS2 Nanosheet Mediated Fluorescence Quenching and Duplex-Specific Nuclease Signal Amplification. Anal. Chem. 2014, 86, 1361−1365. (22) Liu, M.; Yin, X.; Zhang, X. Double-Layer Graphene Optical Modulator. Nano Lett. 2012, 12, 1482−1485. (23) Bolotin, K. I.; Sikes, K. J.; Jiang, Z.; Klima, M.; Fudenberg, G.; Hone, J.; Kim, P.; Stormer, H. L. Ultrahigh Electron Mobility in Suspended Graphene. Solid State Commun. 2008, 146, 351−355. (24) Mao, H. Y.; Laurent, S.; Chen, W.; Akhavan, O.; Imani, M.; Ashkarran, A. A.; Mahmoudi, M. Graphene: Promises, Facts, Opportunities, and Challenges in Nanomedicine. Chem. Rev. 2013, 113, 3407−3424. (25) Tu, Y.; Lv, M.; Xiu, P.; Huynh, T.; Zhang, M.; Castelli, M.; Liu, Z.; Huang, Q.; Fan, C.; Fang, H.; Zhou, R. Destructive Extraction of Phospholipids from Escherichia coli Membranes by Graphene Nanosheets. Nat. Nanotechnol. 2013, 8, 594−601. (26) Chen, J.; Peng, H.; Wang, X.; Shao, F.; Yuan, Z.; Han, H. Graphene Oxide Exhibits Broad-spectrum Antimicrobial Activity against Bacterial Phytopathogens and Fungal Conidia by Intertwining and Membrane Perturbation. Nanoscale 2014, 6, 1879−1889. (27) Mangadlao, J. D.; Santos, C. M.; Felipe, M. J. L.; de Leon, A. C. C.; Rodrigues, D. F.; Advincula, R. C. On the Antibacterial Mechanism of Graphene Oxide (GO) Langmuir-Blodgett films. Chem. Commun. 2015, 51, 2886−2889. (28) Bianco, A. Graphene: Safe or Toxic? The Two Faces of the Medal. Angew. Chem., Int. Ed. 2013, 52, 4986−4997. (29) Perreault, F.; de Faria, A. F.; Nejati, S.; Elimelech, M. Antimicrobial Properties of Graphene Oxide Nanosheets: Why Size Matters. ACS Nano 2015, 9, 7226−7236. (30) Hu, W.; Peng, C.; Luo, W.; Lv, M.; Li, X.; Li, D.; Huang, Q.; Fan, C. Graphene-Based Antibacterial Paper. ACS Nano 2010, 4, 4317−4323. (31) Darden, T.; York, D.; Pedersen, L. Particle Mesh Ewald - An N.Log(N) Method for Ewald Sums in Large Systems. J. Chem. Phys. 1993, 98, 10089−10092. (32) Liu, J.; Cui, L.; Losic, D. Graphene and Graphene Oxide as New Nanocarriers for Drug Delivery Applications. Acta Biomater. 2013, 9, 9243−9257. (33) Ma, X.; Tao, H.; Yang, K.; Feng, L.; Cheng, L.; Shi, X.; Li, Y.; Guo, L.; Liu, Z. A Functionalized Graphene oxide-iron Oxide Nanocomposite for Magnetically Targeted Drug Delivery, Photo3205

DOI: 10.1021/acsnano.7b00236 ACS Nano 2017, 11, 3198−3206

Article

ACS Nano (55) Zhou, R.; Huang, X.; Margulis, C. J.; Berne, B. J. Hydrophobic Collapse in Multidomain Protein Folding. Science 2004, 305, 1605−9. (56) Liu, P.; Huang, X.; Zhou, R.; Berne, B. J. Observation of a Dewetting Transition in the Collapse of the Melittin Tetramer. Nature 2005, 437, 159−62. (57) Gu, Z.; Yang, Z.; Chong, Y.; Ge, C.; Weber, J. K.; Bell, D. R.; Zhou, R. Surface curvature relation to protein adsorption for carbonbased nanomaterials. Sci. Rep. 2015, 5, 10886. (58) Yang, Z.; Ge, C.; Liu, J.; Chong, Y.; Gu, Z.; Jimenez-Cruz, C. A.; Chai, Z.; Zhou, R. Destruction of Amyloid Fibrils by Graphene through Penetration and Extraction of Peptides. Nanoscale 2015, 7, 18725−18737. (59) Zhao, X. Self-Assembly of DNA Segments on Graphene and Carbon Nanotube Arrays in Aqueous Solution: A Molecular Simulation Study. J. Phys. Chem. C 2011, 115, 6181−6189. (60) Case, D. A.; Betz, R.; Cerutti, D. S.; Cheatham, T.E., III; Darden, T. A.; Duke, R. E.; Giese, T. J.; Gohlke, H.; Goetz, A. W.; N, H.; Izadi, S.; Janowski, P.; Kaus, J.; Kovalenko, A.; Lee, T. S.; LeGrand, S.; Li, P.; Lin, C.; Luchko, T.; Luo, R.; Madej, B.; Mermelstein, D.; Merz, K. M.; Monard, G.; Nguyen, H.; Nguyen, H. T.; Omelyan, I.; Onufriev, A.; Roe, D. R.; Roitberg, A.; Sagui, C.; Simmerling, C. L.; Botello-Smith, W. M.; Swails, J.; Walker, R. C.; Wang, J.; Wolf, R. M.; Wu, X.; Xiao, L.; Kollman, P. A. Amber 2016; University of California, 2016. (61) Hess, B.; Kutzner, C.; van der Spoel, D.; Lindahl, E. GROMACS 4: Algorithms for Highly Efficient, Load-balanced, and Scalable Molecular Simulation. J. Chem. Theory Comput. 2008, 4, 435−447. (62) Lindorff-Larsen, K.; Piana, S.; Palmo, K.; Maragakis, P.; Klepeis, J. L.; Dror, R. O.; shaw, D. E. Improved Side-chain Torsion Potentials for the Amber ff99SB Protein Force Field. Proteins: Struct., Funct., Genet. 2010, 78, 1950−1958. (63) Jorgensen, W. L.; Chandrasekhar, J.; Madura, J. D.; Impey, R. W.; Klein, M. L. Comparison of Simple Potential Functions for Simulating Liquid Water. J. Chem. Phys. 1983, 79, 926−935. (64) Bussi, G.; Donadio, D.; Parrinello, M. Canonical Sampling through Velocity Rescaling. J. Chem. Phys. 2007, 126, 014101. (65) Darden, T.; York, D.; Pedersen, L. Particle Mesh Ewald - An N.Log(N) Method for Ewald Sums in Large Systems. J. Chem. Phys. 1993, 98, 10089−10092. (66) Essmann, U.; Perera, L.; Berkowitz, M. L.; Darden, T.; Lee, H.; Pedersen, L. G. A Smooth Particle Mesh Ewald Method. J. Chem. Phys. 1995, 103, 8577−8593. (67) Hess, B.; Bekker, H.; Berendsen, H. J. C.; Fraaije, J. LINCS: A linear constraint solver for molecular simulations. J. Comput. Chem. 1997, 18, 1463−1472. (68) Torrie, G. M.; Valleau, J. P. Non-Physical Sampling Distributions In Monte-Carlo Free-Energy Estimation - Umbrella Sampling. J. Comput. Phys. 1977, 23, 187−199. (69) Kumar, S.; Rosenberg, J. M.; Bouzida, D.; Swendsen, R. H.; Kollman, P. A. Multidimensional Free-Energy Calculations Using The Weighted Histogram Analysis Method. J. Comput. Chem. 1995, 16, 1339−1350. (70) Roux, B. The Calculation of The Potential of Mean Force Using Computer-Simulations. Comput. Phys. Commun. 1995, 91, 275−282.

3206

DOI: 10.1021/acsnano.7b00236 ACS Nano 2017, 11, 3198−3206