Gauging the nanotoxicity of h2D-C2N towards Single Stranded DNA

Subscriber access provided by UNIV OF DURHAM

Functional Nanostructured Materials (including low-D carbon)

Gauging the nanotoxicity of h2D-C2N towards Single Stranded DNA: An in – silico Molecular Simulation Approach Titas Kumar Mukhopadhyay, Kalishankar Bhattacharyya, and Ayan Datta ACS Appl. Mater. Interfaces, Just Accepted Manuscript • DOI: 10.1021/acsami.8b00494 • Publication Date (Web): 03 Apr 2018 Downloaded from http://pubs.acs.org on April 3, 2018

Just Accepted “Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted online prior to technical editing, formatting for publication and author proofing. The American Chemical Society provides “Just Accepted” as a service to the research community to expedite the dissemination of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts appear in full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been fully peer reviewed, but should not be considered the official version of record. They are citable by the Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered to authors. Therefore, the “Just Accepted” Web site may not include all articles that will be published in the journal. After a manuscript is technically edited and formatted, it will be removed from the “Just Accepted” Web site and published as an ASAP article. Note that technical editing may introduce minor changes to the manuscript text and/or graphics which could affect content, and all legal disclaimers and ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors or consequences arising from the use of information contained in these “Just Accepted” manuscripts.

is published by the American Chemical Society. 1155 Sixteenth Street N.W., Washington, DC 20036 Published by American Chemical Society. Copyright © American Chemical Society. However, no copyright claim is made to original U.S. Government works, or works produced by employees of any Commonwealth realm Crown government in the course of their duties.

Page 1 of 41 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

ACS Applied Materials & Interfaces

Gauging the nanotoxicity of h2D-C2N towards Single Stranded DNA: An in – silico Molecular Simulation Approach Titas Kumar Mukhopadhyay, Kalishankar Bhattacharyya, Ayan Datta* Department of Spectroscopy, Indian Association for the Cultivation of Science, 2A and 2B Raja S.C.Mullick Road, Jadavpur, Kolkata – 700032, West Bengal, India. Abstract Recent toxicological assessments of graphene, graphene oxides and some other 2D materials have shown them to be substantially toxic at the nanoscale where they inhibit and eventually disrupt biological processes. These shortfalls of graphene and analogs have resulted in a quest for novel biocompatible two dimensional materials with minimum cytotoxicity. In this article, we demonstrate C2N (h2D-C2N), a newly synthesized two dimensional porous graphene analog to be non – nanotoxic towards genetic materials from an “in – silico” point of view through sequence dependent binding of different polynucleotide single stranded DNA onto it. The calculated binding energy of nucleobases and the free energy of binding of polynucleotides follow the common trait: cytosine > guanine > adenine > thymine and are well within the limits of physisorption. Ab – initio simulations completely exclude the possibility of any chemical reaction, demonstrating purely non covalent binding of nucleobases with C2N through a crucial interplay between hydrogen bonding and π – stacking interactions with the surface. Further, we show that the extent of distortion inflicted upon ssDNA by C2N is negligible. Analysis of the density of states (DOS) of the nucleobase – C2N hybrids confirm minimum electronic perturbation of the bases after adsorption. Most importantly, we demonstrate the potency of C2N in nucleic acid transportation via reversible binding of ssDNA. The plausible use of C2N as a

1 ACS Paragon Plus Environment

ACS Applied Materials & Interfaces 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

template for DNA repair is illustrated through an example of C2N assisted complementary ssDNA winding. Keywords: Porous 2D systems, Genetic materials, Non – covalent interactions, nanotoxicity, Molecular Dynamics. Introduction: Notwithstanding that hybrid of biomolecules and inorganic nanostructures scarcely cohabitate in nature, the flourishing biomedical applications of such composite materials have encouraged researchers for the last two decades to devote a great deal of endeavor towards such technologies.1 Particularly, two dimensional inorganic nanomaterials have embarked on a new journey through nanomedicine ever since the discovery of graphene in 2004, owing to their remarkable physical, mechanical and chemical properties as well as unique electronic structures.2,3 In 2008, the demonstration of Graphene oxide as an efficient nanocarrier for drug delivery by Dai et al. excited material scientists and biologists alike leading to worldwide interdisciplinary collaborations which resulted in the exploration of widespread application of graphene as well as graphene oxides in the realms of biosensing, drug delivery, gene therapy and transportation, photodynamic therapy and tissue engineering among many others.4-11 Posing one step ahead, graphene have been functionalized to design a variety of composite materials which have successfully been used as templates for drug delivery and as probes for diagnostic imaging such as magnetic resonance imaging (MRI), computed tomography (CT), fluorescent imaging and even in radio – imaging.12-16 Nonetheless, in recent years a large variety of nanomaterials have been found to be toxic as they interrupt different biological processes through interactions at the nanoscale.17,18 Hence, it has become imperative to perform extensive in - vitro and in - vivo toxicity assessments with


Page 2 of 41

Page 3 of 41 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


graphene and graphene based materials to ensure safety prior to the practical applications in medical science. Recent studies have revealed that the sharp corners and irregular edges of graphene can penetrate cell membranes and interrupt the intra – cellular functions through strong interactions.19-22 Compacted graphene sheets and less densely packed graphene oxides have also been shown to be toxic towards mammalian fibroblasts.23 Moreover, pure 2D monolayer graphene is significantly hematotoxic, though the hemolytic properties reduce with increasing number of layers.24 In addition, graphene substantially reduces protein – protein interaction and may cleave the secondary and tertiary structures.25 Regarding environmental impacts, Walker and coworkers found that owing to its inherent hydrophobicity, once graphene finds its way into ground and/or surface water, it tends to travel further floating on the surface, thereby reducing the effective contact area between water and air, which is a potent threat to aquatic lives.26 These inadequacies of graphene and analogous materials have resulted in the quest for other 2D materials exhibiting lower nano – toxicity.27-29 One of the newest additions to the family of 2D crystals is C2N, which was first synthesized by Mahmood et al in 2015 through wet chemical reactions in a bottom – up approach.30 The structure of 2D C2N (h2D-C2N) consists of periodic holes surrounded by electronegative nitrogens. It behaves as a semiconductor, showing a direct bandgap of 1.96 eV and a field effect transistor fabricated based on monolayer C2N exhibits a high on/off ratio of ~107. C2N is predicted to have promising future in opto - electronics, semiconductor industry, sensing technology and photovoltaics among many others.31-36 As a novel 2D material, C2N may have biological and medical applications as well, albeit the question of bio – compatibility still persists. To the best of our knowledge there are no previous reports that unequivocally evaluate the nano - toxicity of C2N. To this regard, in this article, we evaluated the bio – compatibility of



C2N towards single stranded DNA through a multidimensional ‘in silico’ approach involving electronic structure calculations and both classical and ab – initio molecular dynamics simulations. Since, Nucleic acids (DNA and RNA) are the most important genetic materials for life we have chosen adsorption of single stranded DNA (ssDNA) on C2N as a platform to judge the nano – toxicity of the material. The adsorption mechanism is scrutinized through large scale all – atom classical molecular dynamics simulations which show interesting structural features of ssDNA bound to the surface which are characteristic for C2N only and not observed for graphene or other 2D materials. Periodic density functional theory calculations are carried out to investigate the structural and electronic perturbation inflicted on the surface upon adsorption of bases. Possibility of chemical reactions between the surface and nucleobases are also encountered employing ab – intio molecular dynamics simulations. We demonstrate reversible ssDNA binding onto C2N, which may be experimentally achieved via appropriate surface passivation. As a final point, we show how biological processes interrupted by graphene can occur in presence of C2N, taking a simple example of duplex formation from two separated complementary ssDNA molecules over C2N. The present article provides sufficient evidences to prove 2D - C2N as a material with substantially low nano – toxicity together with its prospect for reversible ssDNA binding and DNA repair. Computational details: In this work, C2N is modeled as a monolayer of dimension 5.4 nm × 6.5 nm containing 650 carbon and 325 nitrogen atoms (Figure 1(a)). Four ssDNA structures are prepared, each consisting of 12 nucleotides and corresponding to poly A, G, C and T respectively. C2N is placed along the XY plane and for both perpendicular and parallel orientations the ssDNA is kept at


Page 4 of 41

Page 5 of 41 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


least 1.5 nm away from the C2N surface. For each ssDNA, two different initial configurations are generated by placing the DNA at different positions along the XY plane but keeping the vertical distance with C2N fixed. The resulting structures are solvated in a water box of dimension 5.4 nm × 6.5 nm × 7 nm (Figure 1(b)) and the negative charges due to phosphate backbone were neutralized by adding required number of potassium ions. Periodic boundary conditions are applied in all three directions to model an infinite C2N sheet. We adopt a multistep minimization and equilibration protocol as the stability of the ssDNA – C2N composite is the focus of our interest. Details of the protocol are mentioned in the supporting information file. Finally, 300 ns NVT production simulations are carried out with structures obtained at the end of the equilibration simulations using two different initial configurations for each system. We get very similar results for both the systems ensuring that the delineated mechanism is indeed reliable and our results are not suffered by metastable states. For the simulations of ssDNA in the absence of C2N, the same protocol is followed for minimization as well as equilibration and production simulations are performed for 150 ns. In all simulations, isothermal conditions are maintained using Langevin dynamics with a damping coefficient of 5 ps-1.37 Langevin piston method is employed to maintain a constant pressure of 1 atm.38 As the C2N sheet is constrained along the XY plane, therefore, constant pressure coupling is applied along the z direction only. For this purpose, 100 fs piston period, 50 fs damping time constant and 300 K piston temperatures were considered. Particle mesh Ewald (PME) method with 1 Å grid is employed to calculate the periodic electrostatics and a 2 fs time step is used to integrate equations of motions using Velocity Verlet algorithm.39 We use SHAKE algorithm to hold rigid covalent bonds with hydrogen atoms.40 Non-bonded interactions are calculated with a cutoff distance of 12 Å and atomic coordinates are stored after every 10 ps for



the trajectory analysis. Packmol utility is used for the preparation of initial configurations, NAMD 2.12 for classical molecular dynamics simulations, VMD for visualization and our in house Tcl scripts as well as VMD plug-ins for the analysis of data.41-43 We used TIP3P model for water and CHARMM General Force Field (CGenFF) parameters (version 3.0.1) for ssDNA.44,45 The carbon and nitrogen atoms of C2N are assigned partial charges of 0.24e and – 0.48e respectively, which were calculated by Zhou and co-workers using HF/6-31G* level of optimization along with RESP parametrization.46 The Lennard Jones parameters associated with CGenFF atom types CG2R61 (aromatic carbon) and NG311 (sp2 hybridized pyrazine nitrogen) are assigned to carbon and nitrogen atoms. Reversible free energy changes accompanying adsorption are calculated in terms of potential of mean forces (PMF), using the adaptive biasing force (ABF) algorithm implemented in NAMD.47 Details of the PMF calculations are mentioned in the supporting information file. For analysis, a contact between ssDNA and C2N is considered if any of the ssDNA atoms is within a cutoff distance of 0.6 nm to the C2N surface and is 𝑁

𝑟 +0.6𝑛𝑚

𝑁𝑠𝑠𝐷𝑁𝐴 𝐶 𝑁 calculated as 𝑁𝑐 (𝑡) = ∑𝑖=12 ∑𝑗=1 ∫𝑟 𝑖 𝑖

𝛿[𝑟(𝑡) − 𝑟𝑗 (𝑡)] 𝑑𝑟. Contact surface area (CSA)

is defined as half of the difference between the solvent accessible surface area (SASA) of the ssDNA – C2N composite system and the sum of the SASAS’s of the C2N surface and ssDNA calculated separately. To define a hydrogen bond, we have used the following geometric criteria: D – A distance ≤ 0.35 nm and D – H – A angle ≥ 1200. Two nucleobase residues 𝑖 𝑎𝑛𝑑 𝑗 are considered to be π - stacked if the COM distances between them 𝑑𝑖𝑗 ≤ 0.5 nm (𝑖 = 𝑗 ± 1 for sequential and 𝑖 = 𝑗 ± 2, … … , 𝑗 + 11 for non – sequential stacking). All periodic Density Functional Theory (DFT) calculations are performed by VASP package.48 Core – valance interactions are accounted by projector augmented wave (PAW) and the Perdew-Burke-Ernzerhof (PBE) exchange-correlation functional is used within the 6 ACS Paragon Plus Environment

Page 6 of 41

Page 7 of 41 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


generalized gradient approximation (GGA).49 Ionic cores are represented by ultrasoft pseudopotentials and van der Waals corrections are incorporated by Grimme’s DFT-D2 method.50 We prepare a 2×2×1 supercell from the optimized unit cell of C2N monolayer. Along the z direction, 15 Å of vacuum is used to avoid the interaction with periodically propagated images. Brillouin zone integrations are described by the 5×5×1 Monkhorst-Pack K-point for structural relaxation and 13×13×1 K – point is used for density of states calculations. For structural optimization, cut off energy of 400 eV is used and the energy as well as forces required for convergence on each atom is set to 10-4 ev and 10-3 eV/Å respectively. Adsorption energies of nucleobases over C2N monolayer were calculated using the formula ΔEad = EC2N-base – (EC2N + Ebase), where ΔEad is the adsorption energy of the molecules on C2N and EC2N-base, Ebase and EC2N represent the total energies of the optimized nucleobase – C2N composite, isolated nucleobases and C2N layer respectively. Ab – initio molecular dynamics (AIMD) simulations are carried out using the QUICKSTEP program implanted in CP2K package.51 This algorithm uses a Gaussian planewave (GPW) approach where Kohn-Sham orbitals are generated in an atom-centered Gaussian basis. Auxiliary plane wave basis sets are used to describe electronic charge density.52 Under GPW approach, the norm-conserving Goedecker, Teter, and Hutter (GTH) pseudopotentials have been used to describe the core electrons as well as nuclei, double –ζ valance polarized basis set (DZVP-MOLOPT-SR-DTH) are used to treat electrons in the valence shell.53 We employed Grimme’s DFT-D2 method to incorporate non-covalent interactions.54 All simulations are carried out at 300K in the NVT ensemble where isothermal condition is maintained using NoséHoover thermostat.55,56 Velocity-Verlet algorithm is used to integrate the equations of motion taking 0.5fs as the time step and all the trajectories were propagated for at least 15 ps. 7 ACS Paragon Plus Environment


Results and Discussion: Adsorption of ssDNA onto C2N: To decipher the toxicity induced by C2N towards ssDNA, it is highly necessary to understand the underlying mechanism of ssDNA binding. To this regard, we performed simulations (300 ns each) with 12 nucleotide long ssDNA’s corresponding to poly – A, T, G and C respectively in a parallel orientation with respect to the basal plane of C2N situated beneath. The mechanism of ssDNA binding on C2N can be categorized into three hierarchical steps namely anchoring (0 – 40 ns), adsorption (40 – 120 ns) and reorganization (120 ns and beyond). Figure 1 (c – f) represents the snapshots of typical intermediate geometries of polyadenine ssDNA binding with C2N. Similar snapshots for other three polynucleotide ssDNA are given in the supporting information file (Figure S1(a-e)). Stability of the final adsorption geometries can be judged on the basis of root mean square displacement (rmsd) of the ssDNA (Figure S1), which saturates within 180 200 ns of timescales for all the simulations. To analyze the dynamics of adsorption, we first plot the time evolution of the center of mass (COM) distances between C2N and different ssDNA in Figure 2(a). Initially ssDNA is placed at least 1.5 nm above C2N surface and they are completely non – interacting (Figure 1(c)). However, as the simulations start, COM distance first increases and undergoes some initial fluctuation between 0 - 10 ns representing solvent assisted free diffusion of the ssDNA. Thereafter, distance gradually reduces until ~ 100 ns and attains constancy around 1 nm, which certainly indicates stepwise adsorption of ssDNA on C2N. To compare the sequence dependent adsorption propensity and pattern we calculated the number of contacts (N) as well as the contact surface area (CSA) between the C2N surface and the adsorbing ssDNA (Figures 2(b) and 2(c) respectively). Initially, both N as well as CSA is zero for all the simulations when the ssDNA is distant to C2N. After few nanoseconds, ssDNA comes 8 ACS Paragon Plus Environment

Page 8 of 41

Page 9 of 41 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


close to the surface and make the first contact through π - π stacking of a terminal nucleobase (Figure 1(d)), which is understood through the comparison of the rmsd of each of the twelve residues for all four ssDNA molecules. Figure 2(d) shows that for each ssDNA one of the terminal residues has very low rmsd compared to others for the last 200 ns of the simulations. Therefore, this residue is conformationally locked most of the time and in fact, the base behaves as an ‘anchor’ to the surface which is maintained throughout the simulation after adsorption. As observed from Figure 2(b) and 2(c), during the above events, both N and CSA increase sharply and concurrently although in a stepwise fashion, as different segments of the ssDNA gets sequentially absorbed onto the surface and more interactions between C2N and ssDNA are built up. In all of our simulations, we see N and CSA to increase and fluctuate up to ~ 200 ns compared to ~ 50 ns and ~ 150 ns for the adsorption of ssDNA over graphene oxide and graphene respectively as oberved by Xu et al in a recent study.57 Thus the adsorption of ssDNA over C2N is kinetically sluggish, presumably due to the porous structure of the surface which results in lower adsorption energy.



Figure 1: (a) Dimensions of the model C2N sheet (C and N atoms are represented by cyan and blue spheres respectively) and (b) shape of the periodic box used in the Classical MD simulations and (c-f) snapshots representing hierarchical stages of adsorption of polyadenine ssDNA onto 2D - C2N at 300 K.

Figure 2: Time evolution of (a) COM distances, (b) number of contacts and (c) contact surface areas between C2N surface and the four ssDNA employed in this study, (d) average RMSD for each of the twelve residues for all the adsorption simulations involving polynucleotide ssDNA parallel to C2N. It is known that nucleobases strongly interact with graphene and carbon nanotubes predominantly through π – π stacking interactions.58 Similar to graphene, the structure of C2N consists of 2D array of aromatic rings, albeit intervened by pores. In addition, each of these pores


Page 10 of 41

Page 11 of 41 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


are surrounded by six electronegative nitrogen atoms and therefore, the sugar phosphate backbone and the bases of ssDNA can interact with C2N through hydrogen bonding interactions as well. To shed light into the mechanistic details, we calculated the number of hydrogen bonds formed and interaction energies between ssDNA and C2N. As shown by Figure 3(a), within first few nanoseconds, hydrogen bonds start to build up between ssDNA and C2N as they make contacts. Interestingly, for all the simulations, H – bonds undergo a rapid increase between 40 – 120 ns to reach the maxima and then subsequently decrease between 120 – 150 ns to attain an equilibrated value which is maintained throughout the rest of the timescale investigated. This observation is also supported by Figure 3(b) which shows an initial rapid increase in the ssDNA – C2N interaction energy between 40 and 120 ns followed by a small decrease to reach saturation. In harmony with N and CSA, the ssDNA – C2N interaction energies follow the order poly T > poly - C > poly - A ≈ poly – G. Partitioning the interaction energies into electrostatic and vdW component (Figure S3(a-d)) reveal that both of these energies change concurrently and in an approximately similar fashion, the electrostatic component being less contributing than the vdW part. The force field used in the present study recognizes π - π stacking in terms of vdW interactions and therefore, the decrease in energy between 40 – 120 ns is accompanied by a synergistic progression of π stacking as well as H – bonding interactions between ssDNA and C2N. Also, Figure 2(b-d) show that, after 40 ns the COM distances remains approximately constant, but N as well as CSA first increases and then fluctuates between 40 – 120 ns (Figures 2(b, c)). Therefore, after anchoring, the ssDNA rapidly comes closer to C2N through simultaneous hydrogen bonding and π - π stacking interactions (up to ~ 40 ns). This process continues till ssDNA searches for a favorable adsorption geometry where interaction with C2N is maximum (~ 40 – 150 ns) and gets bound to the surface.



anchoring adsorption


reorganization

Page 12 of 41

reorganization

Figure 3: Time evolution of the (a) number of hydrogen bonds and (b) interaction energy between the four polynucleotide ssDNA molecules and C2N considered in this study. ssDNA adsorption on a polar surface is a complex thermodynamic phenomenon and DNA – surface interactions are not solely responsible for the outcome of the event.59 For instance, the above analyses do not give insights regarding the behavior of ssDNA after adsorption, especially the loss of ssDNA – C2N hydrogen bonding and interaction energy during this period of time (> 120 ns). Therefore, we scrutinized the effect of adsorption on the structure and intra residue interactions of ssDNA. To this regard, we calculated the number of intra – residue stacking interactions and the radius of gyration (Rg) of the ssDNA. In Figures 4(a) and 4(b), both pi stacking and Rg shows two strikingly different trends. For poly –A and poly – G, Rg smoothly decreases from the very beginning and continues to decrease till 300 ns to produce a folded and coiled structure, while the self-stacking first remains constant up to ~ 40 ns, but increases between 40 - 120 ns, before again approaching constancy. On the other hand, for poly – C and poly – T, Rg first decreases and then fluctuates before reaching a stable value while stacking between the bases decrease until 150 ns and then again increases to a small extent 12 ACS Paragon Plus Environment

Page 13 of 41 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


between 150 – 300 ns. In addition, they are adsorbed in a linearly extended conformation than poly A and G, which is easily understood from the much higher Rg values at the end of 300 ns (Figure 4(b)) and the respective snapshots (Figures 1(e) and 1(f)). These observations suggest a crucial interplay between intra residue ssDNA π stacking and DNA - C2N interactions. The structure of adenine (A) and guanine (G) consists of two aromatic rings compared to one ring in cytosine (C) and thymine (T). Therefore, A and G more efficiently self – stack than C and T. As a consequence, during adsorption, poly A and poly G maintains while poly T and poly C loses most of the intra DNA π – π stacking interactions. However, after adsorption individual nucleobase residues reorganize through an increase in stacking to attain a configuration in phase space where intra residue stacking as well as interaction with C2N is optimum. The kinetics of reorganization is faster for poly A and G (up to 120 ns) in comparison with poly T and C (up to 300 ns) which is again ascribed to the formers advanced stacking efficiency. Nevertheless, during self – stacking, DNA – surface π stacking interactions are gradually lost as some of the bases desorb from the surface which is also followed by the loss of some H - bonds. That is why we observe H - bonding as well as interaction energies between ssDNA – C2N to deplete after adsorption, albeit to a small extent. In addition, constancy of neighboring base stacking along with gradual reduction in Rg for poly A and G ssDNA between 120 and 300 ns is further suggestive of non – sequential nucleobase stacking, which in turn implies weaker bonding between these ssDNA and C2N. It should be noted that, although the capability of self stacking for poly A and poly G are greater than both poly C and poly T due to two aromatic rings in the former, but these tendencies cannot overcome the tendency of stacking with surface. Interestingly, poly C shows higher number of H - bonds until the end of our simulations as cytosine prefers to self – stack rather than stacking with C2N. This self – stacking is achieved



Page 14 of 41

through an orientation where most of the bases are approximately perpendicular to the surface but hydrogen bonded to the nitrogen rich pores. This observation is explored in greater details in later sections. anchoring adsorption


reorganization

reorganization

Figure 4: Variation of (a) Intra – residue π – π stacking interactions between neighboring nucleobases and (b) radius of gyration (Rg) for the four polynucleotide ssDNA during 300 ns of the simulations. We also performed simulations with poly G and poly C ssDNA initially perpendicular to C2N. However, both of them get adsorbed in a similar geometry to that observed in an initially parallel orientation, going through identical hierarchical steps and showing similar dynamical properties (Figures S4 and S5). Therefore, the initial arrangement of ssDNA has negligible influence on binding. Effects of heating: Conformations of ssDNA strands on two dimensional C2N can be history dependent and even metastable states may arise where the simulations may remain trapped even after 300 ns. To rule 14 ACS Paragon Plus Environment

Page 15 of 41 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


out such possibilities and additionally to explore the effects of heating on the binding of ssDNA with C2N, we have taken the final structures of the production simulations for all four ssDNA over C2N at the end of 300 ns and heated from 300 K to 350 K over 10 ns. Then, they are simulated for 100 ns at 350 K temperature and 1 atm pressure in an isothermal isobaric ensemble. Figure 5 (a – d) displays the snapshots showing the structures of the ssDNA - C2N hybrids for all four polynucleotides both before and after heating and Figure S12 represents the time evolution of different dynamical quantities for these simulations. Increase in temperature leads to various events in the surface – ssDNA composite system. It is known that at higher ranges of temperatures the intra – ssDNA interactions are disrupted to some extent. For all the ssDNA – C2N composites, when heated, we observe the COM distances to remain nearly unaltered suggesting no abrupt change in the hybrid architecture or desorption during heating (Figure S12(a)). However, we observe number of contacts between the surface and ssDNA to undergo a small increase (Figure S12(b)) which clearly supports that the thermodynamically preferred interaction is the ssDNA – C2N interaction and not the intra – ssDNA nucleobase stacking, as mentioned previously. This observation is in agreement with a recent study by Walsh et al who observed a higher degree of dsDNA binding to graphene, leading to structural disruption in different conformations of an adenosine binding DNA aptamer on graphene after carrying out simulated annealing.60 However, for C2N, the extent of increase in stacking with surface upon heating is indeed lesser compared to graphene. Regarding the structure of ssDNA, we observe the radius of gyration (Rg) for all of them to increase, albeit to very small extents (Figure S12(c)). Poly A is an exception among these for which increase in Rg upon heating is higher, leading to the loss of compactness (Figure 5(a)). As shown in Figure S12(d), number of intra – DNA π stacking interactions slightly decreases for both poly – A and poly – G which does



not cause adsorbed poly – G to expand (Figure 5(b)). Therefore, the increase in Rg for poly – A is certainly due to loss of nonsequential stacking interactions and not because of the disruption of native sequential stacking. This may be ascribed to the higher stacking energy between poly – G nucleobases compared to poly – A (ΔEGG > ΔEAA), as reported in previous studies.61 On the other hand, poly C and poly T are already adsorbed in an extended conformation. Therefore, with increase in temperature, they further interact with the surface and become somewhat more elongated (Figures 5(c) and 5(d)), thereby marginally increasing the radius of gyration. During heating, each of the polynucleotide further interacts with the surface, thereby increasing the number of H – bonds (Figure S12(e)). It should be noted that, although on heating both π – π stacking and hydrogen bonds between ssDNA and surface increases, but the distribution of these interactions are dependent on the polynucleotide sequences. For example, both poly – C and poly – T are strongly adsorbed on C2N, but the former has more propensity towards hydrogen bonding while the latter rely on π stacking interactions for binding. Nevertheless, at higher ranges of temperature, H – bonded conformations are entropically more preferred compared to other conformers bound through π – π stacking. As expected, poly – C shows highest number of hydrogen bonds among all four polynucleotides, followed by poly – A and poly – G respectively which is in agreement with the simulations performed at 300 K. Indeed, the increase in contacts and H – bonds are accompanied with a small decrease in the interaction energy, as shown in Figure S12(f). However, the interaction energy trend remains the same poly – T > poly – C > poly – A ~ poly – G. Overall, heating does not change the binding trends and leads to no structural disruption, only the nucleotide residues interact somewhat more with the surface.


Page 16 of 41

Page 17 of 41 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


Figure 5: Snapshots representing the structures (both top and side views) of the ssDNA - C2N hybrids for (a) poly – A, (b) poly – G, (c) poly – C and (d) poly – T respectively before and after heating at 350 K for 110 ns in total. For poly – A, expansion of the compact structure is observed upon heating because of the loss of non – sequential stacking, keeping the sequential stacking nearly intact, while for others the structure remains nearly unchanged. Thermodynamics of sequence dependent ssDNA binding: To disentangle the multifaceted thermodynamics of polynucleotide ssDNA binding on C2N, we adopted a step up approach by increasing the size of the adsorbed species. First we calculated the binding energies of different nucleobases on C2N using periodic DFT methods. The binding energy sequence becomes C > G > A > T (Table 1) which is different from those reported for graphene and hexagonal boron nitride in previous studies.62 Interestingly, among five different adsorption geometries considered (Figure S6(a – e)), the perpendicular orientation of all nucleobases over C2N is found to be more favored over the parallel orientations resulting from 17 ACS Paragon Plus Environment


stacking of bases over benzene or pyrazene rings of the surface (Figure 6 (a – d)). In these perpendicular conformations, there are one or more hydrogen bonds between the bases and the electronegative surface nitrogen atoms surrounding the pore, as indicated from the low N – H∙ ∙ ∙N distances in Figure S7(a – d), but they lack in stabilization due to stacking. However, magnitudes of the binding energies are similar to those found for graphene and therefore we speculate that individual nucleobases have equivalent affinity towards graphene and C2N, although the modes of interactions with these materials are fundamentally different.62,63 Further discussion regarding the choices of adsorbed nuceobase geometries through extensive sampling, decomposition of the binding energies and comparison with classical MD gas phase free energies can be found in the supporting information file (Figures S6, S7, S11 and Tables S1 – S3). To incorporate the signature of temperature, solvent molecules and sugar phosphate backbone, we calculated the reversible free energies for adsorption in terms of potential of mean forces (PMF) for dinucleotides and polynucleotides corresponding to A, G, C and T respectively on C2N 𝑏 ̅̅̅̅𝑛𝑢−2 (Figure 7 (a) and 7(b)). The binding PMF per nucleotide for dinucleotides (∆𝐺 ) follow the

order C > G > A > T which is exactly the same as observed for the binding energies of free bases calculated employing DFT (Table 1). However, for polynucleotides, binding free energy requirements have the sequence: poly T (- 43 kcal/mol) > poly C (- 36 kcal/mol) > poly G (- 33 kcal/mol) > poly A (- 28 kcal/mol) which is different from that observed for dinucleotides. As shown in Table 1, the magnitudes of average binding free energies per nucleotide calculated for 𝑏 𝑏 𝑏 ̅̅̅̅𝑛𝑢−12 ̅̅̅̅𝑛𝑢−12 ̅̅̅̅𝑛𝑢−2 the 12 base long sequences (∆𝐺 ) are much less than those (|∆𝐺 | < |∆𝐺 |) for the

dinucleotides in the cases of poly A, poly G and poly C, in spite of carrying strongly adsorbing bases. Contrary to the above observation, poly T shows increase in free energy per nucleotide in moving from dinucleotides to polynucleotides. From DFT calculations, we observed thymine to


Page 18 of 41

Page 19 of 41 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


show least binding energy among all the nucleobases and hence, the enhanced binding free energy of poly T certainly has a large contribution from backbone – surface interactions. We, assign these observations as direct consequences of backbone – surface interactions as well as prevalence of base stacking over stacking with C2N. Increasing the polynucleotide chain increases the interaction of the sugar phosphate backbone with surface and tends to increase the free energy of adsorption. On the other hand, longer polynucleotides are able to perform enhanced self - stacking and after adsorption they undergo reorganization to regain those stacks, thereby reducing the interaction with surface. The relative strengths of these two opposing events guide the overall spontaneity of sequence dependent binding of ssDNA on C2N. For poly T, backbone surface interactions outweigh inefficient self - stacking while for all other polynucleotides, self – stacking appears to be the decisive factor during adsorption. Also, a large entropic penalty for adsorption reduces the thermodynamic binding affinity with increasing polynucleotide length. Table 1: Thermodynamic quantities related to the binding of ssDNA on C2N surface. ∆𝑬𝒃 represent the binding energies of the most stable optimized structures of single nucleobases adsorbed over C2N obtained from DFT while ̅̅̅̅ ∆𝑮𝒃𝒏𝒖−𝟐 and ̅̅̅̅ ∆𝑮𝒃𝒏𝒖−𝟏𝟐 are the average binding energies per nucleotide for different dinucleotides and polynucleotides respectively obtained through PMF calculations. Energies carry the unit of kcal/mol. Nucleobase

∆𝑬𝒃

̅̅̅̅ ∆𝑮𝒃𝒏𝒖−𝟐

̅̅̅̅ ∆𝑮𝒃𝒏𝒖−𝟏𝟐

A

-20.6

-3.9

-2.3

G

-23.0

-5.1

-2.8

C

-27.7

-7.5

-3.0

T

-13.3

-2.2

-3.6



(a)

(b)

(c)

(d)

Figure 6: Side and top views of the most stable optimized structures of nucleobases on C2N: (a) adenine, (b) guanine, (c) cytosine and (d) thymine, as obtained from periodic DFT calculations.


Page 20 of 41

Page 21 of 41 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


Figure 7: Potential of mean forces (PMF) corresponding to the adsorption of different (a) dinucleotides and (b) polynucleotides onto C2N surface at 300K. Free energies for the non – interacting desorbed state is considered as zero which corresponds to δ = 3 nm and 1 nm for (a) and (b) respectively. Reversibility of ssDNA binding: To show the reversibility of ssDNA binding to C2N, we adopted a force field control protocol. We ran several simulations with modified force field parameters, each for 50 ns, with the final structure of poly – T adsorbed over C2N collected at 300 ns (Figure 8(a)). In one simulation, partial charges of C2N are withdrawn and the adsorbed structure is observed to remain nearly unaltered (Figure 8(b)). On the other hand, we ran three different simulations with modified Lennard Jones parameters of C2N so that the van der Waals interactions with ssDNA are scaled to 75%, 50% and 25% of the original magnitude, keeping charges over C2N intact. In each of the simulations we observe poly T to slowly desorb from the surface, which is understood through gradual increase in the COM distance (Figure 9(a)) with concomitant reduction in contacts with C2N (Figure 9(b)) and loss of interaction energy (Figure 9(c)). With progressively decreasing van der Waals interaction, the desorption kinetics becomes faster. For instance, upon decreasing the vdW interactions to 25% and 50%, ssDNA is observed to be entirely dislodged from the surface (COM distance > 1.3 nm) within only 20 ps and 40 ps respectively (Figures 9(c) and 9(d)). However, upon scaling the vdW interactions to 75%, we observed the surface to weakly carry the DNA with a few contacting atoms, rather than its complete release (Figures 8(e) and 9(b)). Therefore, hydrogen bonding and electrostatics help the DNA to get adsorbed and indeed behave as a favorable addition to vdW interactions, but certainly cannot hold the hybrid architecture when vdW interactions are reduced just by 25%. Note that, poly – T has highest binding affinity 21 ACS Paragon Plus Environment


Page 22 of 41

to the surface and therefore, all other polynucleotides having looser binding are expected to show higher degrees of such reversibility. Inspired from such results we envision that reversible binding of ssDNA onto C2N can be experimentally achieved through appropriate surface passivation techniques and screening of vdW interactions, having far reaching consequences particularly in the domain of controlled nucleic acid transference.64,65

0.25

Ecol = 0

(b)

Evdw

(c)

(a) 0.75

0.5

Evdw

(e)

Evdw

(d)

Figure 8: Snapshots representing structural changes through the alteration of force – field parameters for poly T: (a) initial structure, (b) structure when charges over C2N are removed, (c-e) structures at the end of 50 ns when DNA – C2N van der Waals interactions are reduced to 75%, 50% and 25% of the original magnitude respectively.


Page 23 of 41 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


Figure 9: Time evolution of (a) COM distances, (b) number of contacts and (c) interaction energies between poly T ssDNA and C2N during different force field modification simulations. Timescales in the negative regime (-20 ns to 0ns) indicates the last 20 ns of the 300 ns production simulation and are given for comparison only. On the Nano – toxicity of C2N: To establish the non – cytotoxicity of C2N at nanoscale, we have to ensure that (i) biomolecules get physically adsorbed to the material without perturbing their individual electronic and structural properties, (ii) bio – molecular fragments do not chemically react with the material and (iii) biological processes can occur even in presence of surface. In the next section, we check whether C2N satisfies these necessary conditions for non – nanotoxicity taking ssDNA and its components as substrates. The magnitudes of binding energies of nucleobases calculated from DFT and adsorption free energies for polynucleotides calculated from MD simulations are well below the range of typical chemisorptions. However, classical MD force fields are incapable of tracking chemical reactions while DFT results provide a static picture taking no notice of thermal as well as solvent effects. To this regard, we perform ab - initio molecular dynamics simulations (AIMD) of different nucleobases adsorbed over C2N monolayer. For each base, two production NVT simulations are performed at least for 15 ps at 300 K corresponding to the most stable adsorbed perpendicular and parallel geometries obtained from DFT and both of the simulations give very similar results. Figures 10(a) and 10(b) display snapshots representing the initial and final structures of typical purine and pyrimidine bases adenine and cytosine respectively, initially adsorbed on 2D - C2N in a perpendicular fashion and the observations for guanine and thymine



are very similar to them. The equilibrated final structures of both the bases are neither completely parallel not perpendicular to the surface. To track the conformational evolution we plot the angle between z axis and the molecular axis of nucleobases as defined in Figure 10(c). For the first few picoseconds nucleobases undergo fluctuation (θ ≈ 10 – 20) and then rapidly flip into a tilted conformation nearly perpendicular to the surface, the molecular rotation being confirmed from the increase in the angle (θ) from θ ~ 20 to 90 in Figures 10(d) and 10(e). The spontaneous flipping suggests that small energy barriers between the parallel and perpendicular conformations of the nucleobases are easily overwhelmed due to thermal effects. Whatever initial orientation we start with, the equilibrated structure is an intermediate between the above two extremes which is obtained through a crucial interplay of π stacking to the surface pyrazene rings while simultaneously performing hydrogen bonding with the nitrogen atoms of the pore (Figure 10(f)). We also observe similar tilted gas phase conformations of nucleobases on C2N in our PMF calculations (Discussed in supporting information file). These observations lead to the conclusion that in presence of solvent molecules, neighboring residues and extended backbone, individual nucleobases in a C2N - polynucleotide hybrid would easily be able to shuffle between different local energy minima. The ultimate distribution of nucleobase orientations are guided by the above mentioned effects instead of their individual binding energies which directly support the reduction in binding energy per nucleobase with increasing length of the polynucleotide, as observed in our free energy calculations. In none of the AIMD simulations we observe the planarity of the nucleobases or C2N to disrupt which provides necessary confidence regarding the unreactive nature of these DNA fragments towards C2N. Also, calculation of density of states (DOS) of pure C2N and C2N – nucleobase hybrids suggest negligible perturbation of the electronic states of C2N upon adsorption (Figure S8(a-d)). Thus we can firmly conclude that,


Page 24 of 41

Page 25 of 41 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


backbone mediated adsorption of constituent bases of nucleic acids on C2N surface proceed only through non – covalent interactions without hampering the electronic and chemical features of individual systems. Indeed, Klein and coworkers have demonstrated some unusual conformations of ssDNA in presence of single walled carbon nanotubes, which have been attributed to backbone mediated helical wrapping.58

Figure 10: (a, b) Snapshots representing initial structures and the geometries after flipping, (c) definitions of the angles θ for the analyses of nucleobase flipping, (d, e) Time evolutions of θ and 25 ACS Paragon Plus Environment


(f) average NH (base)∙∙∙N (C2N) distances representing hydrogen bonding (d

Gauging the nanotoxicity of h2D-C2N towards Single Stranded DNA

Recommend Documents