The interactions of nanoparticles (NPs) with single stranded nucleic acids (NAs) have important implications in gene delivery, and nanotechnological and biomedical applications. Here, the complexation of cationic ligand functionalized gold nanoparticles with single stranded deoxyribose nucleic acid (DNA) and ribonucleic acid (RNA) are examined using all atom molecular dynamics simulations. The results indicated that complexation depends mostly on charge of nanoparticle, and, to lesser extent, sequence and type of nucleic acid. For cationic nanoparticles, electrostatic interactions between charged ligands and the nucleic acid backbone dominate binding regardless of nanoparticle charge. Highly charged nanoparticles bind more tightly and cause compaction of the single-stranded NAs through disruption of intrastrand π–π stacking and hydrogen bonding. However, poly-purine strands (polyA-DNA, polyA-RNA) show less change in structure than poly-pyrimidine strands (polyT-DNA, polyU-RNA). Overall, the results show that control over ssNA structure may be achieved with cationic NPs with a charge of more than 30, but the extent of the structural changes depends on sequence.

In gene therapy, foreign nucleic acids (NAs) are introduced into cells to affect gene expression. This field holds great promise to treat a number of genetic disorders and diseases.1 However, effective delivery vehicles are required for transporting genetic material into cells. Positively charged dendrimers or nanoparticles (NPs) can be used to bind negatively charged NAs through electrostatic interactions and act as delivery vectors.2,3 Ligand functionalized inorganic NPs are particularly promising in this area of research3,4 because of potential stimuli responsiveness and tunable surface chemistry. Control over shape, size, and charge of NP-nucleic acid complex is crucial for its delivery efficiency and transport into cells. For example, studies of the cellular uptake of citrate stabilized gold nanoparticles have shown strong size effects, with maximum uptake occurring with 50 nm particles,5 whereas increasing nanoparticle ligand hydrophobicity has been shown to increase cytotoxicity of nanoparticles.6 

The interactions between NPs and nucleic acids can be engineered to occur in a predictable manner. Recent studies show that positively charged nanoparticles7 and deoxyribose nucleic acid (DNA) micelles8 were able to interact with and across the cellular membrane. Additionally, the NP ligand chemical identity and charge can control biodistribution, with charged nanoparticles accumulating in different organ regions as compared to uncharged NPs.9 However, the NP design rules for effective gene delivery are yet to be formulated.

Computational techniques can provide atomic level details of the mechanism of interactions between nanoparticles with DNA and predict the required properties for NPs. All previous computational investigations focused on interactions between NP and double stranded DNA (dsDNA). For example, atomistic molecular dynamic (MD) simulations have shown that cationic dendrimers cause different conformational changes to dsDNA depending on the dendrimer generation and overall charge.10 Smaller dendrimers bind and conform to the structure of dsDNA, resulting in local deformation of the helix. In contrast, binding of larger, higher generation dendrimers causes bending of the dsDNA. We have recently reported the effect of cationic ligand functionalized nanoparticle binding on structure of dsDNA and double stranded ribonucleic acid (RNA) (Refs. 11 and 12). High concentrations of nanoparticles with low charge have been shown to damage dsDNA by inducing DNA strand separation.12 For low concentrations of nanoparticles, the bending of dsDNA around the nanoparticle depending on the NP charge; i.e., nanoparticles with low charge bind to dsDNA without affecting its conformation, but nanoparticles with higher charge (50% charge neutralization) cause dsDNA bending. Response of dsDNA to NP binding can be further modulated through changes in the solution salt concentration.11 

Though structural conformations of single stranded DNA and RNA (ssDNA and ssRNA) are essential for recognition by other nucleic acids and proteins and their biological functions, the nature of noncovalent complexation of a NP with a ssDNA or ssRNA remain much less well characterized and understood. Many applications rely on ssDNA's ability to hybridize with complementary strands through base pairing, which is used to assemble complex shapes and structures such as DNA origami13 or tiles,14 or to induce a nanoscale motion such as “nanorobots”15 or DNA walkers.16 Thus, the NP binding to nucleic acids may induce changes that can be used to further modify/control the processes of hybridization and interrupt RNA's and DNA's intramolecular interactions. Here, we investigate the binding of ligand functionalized cationic gold nanoparticles to single stranded DNA and RNA using all-atom molecular dynamics simulations. We perform simulations with nanoparticles of three different total charges: +6, +30, and +60, which are varied by number of charged (protonated) ligands. To observe the differences which may be imparted by nucleic acid sequence, either poly-adenine (polyA-DNA and polyA-RNA), poly-thymine (polyT-DNA), or poly-uracil (polyU-RNA) is investigated. Our study shows that the mechanism of NP binding to ssDNA and ssRNA depends on NP charge.

All-atom MD simulations were performed using the amber 14 software package17 with the GPU-accelerated PMEMD code.18–20 Single stranded NAs were simulated using the ff99bsc0 (Ref. 21) force field. The force field parameters and the preparation of ligand functionalized nanoparticles with metal core diameter of approximately 1.5 nm were described previously.12 The nanoparticle core consisted of 135 gold atoms functionalized with 60 S-(CH2)11-R organic ligands, where R is an end-group. The end-group of the ligands were either an ammonium cation (R = NH4+) or a methyl group (R = CH3), allowing for the variation of nanoparticle charge by varying the number of charged nanoparticle ligands. Single stranded poly-adenine (polyA-DNA or polyA-RNA), single stranded poly-thymine (polyT-DNA), and single stranded poly-uracil (polyU-RNA) were each 60 bases long. All nucleic acid sequences were equilibrated (folded) in implicit solvent for at least 60 ns prior to introduction of NPs for simulation of nanoparticle binding. Systems containing NPs and DNA were first simulated in implicit solvent for 20 ns then subjected to an explicit solvent refinement for 10 ns. Implicit solvent was chosen for simulations of folding an initial binding because the reduced viscosity in implicit solvent allows for greater sampling of conformational space and faster conformational convergence.22 

Implicit solvent simulations were performed using igb = 1 option in amber, which uses the pairwise descreening method of Hawkins et al.23 This implicit solvent technique has been found to be a good model for nucleic acids in solution.24 In explicit solvent, systems were solvated with TIP3P water and sodium chloride (NaCl) ions with force field parameters from Joung and Cheatham25 at a concentration of 0.1 M NaCl. Simulation analysis was performed using CPPTRAJ (Ref. 26) and in-house scripts, while images were created using visual molecular dynamics.27 

Implicit solvent simulation systems were prepared using a four-step protocol for minimization and heating.12 The system was first subjected to a 10 000 step steepest descent minimization, followed by constrained 20 kcal/mol Å2 heating to 300 K and two consecutive constrained MD equilibrations with 1 and 0.1 kcal/mol Å2 constraints correspondingly. Production simulations were run with no constraints with implicit salt concentration of 0.5 M and 1 fs time step using a Berendsen28 thermostat with no cut-off for nonbonded interactions. Simulations in implicit solvent were carried out for at least 20 ns. The implicit solvent simulations were also used for analysis of initial nanoparticle binding to the nucleic acid chain.

Final stable conformations from implicit solvent simulation were used to perform explicit solvent refinement simulations. The explicitly solvated system was first minimized using a steepest descent method for 10 000 cycles. Subsequently, the system was heated to 300 K over 100 picoseconds using a Langevin thermostat. The heating step was performed using an isothermal-isobaric (NPT) ensemble to allow the solvent to adjust to the appropriate density. Production simulations were performed using the pmemd.cuda in the amber software suite. For production simulations, we used the isothermal-isobaric ensemble with a Berendsen thermostat28 with a 9 Å cut-off for nonbonded interactions. Long-range electrostatic interactions were calculated using a particle mesh Ewald summation. Simulations were run for a minimum of 10 ns in explicit solvent for structure refinement, of which the last 5 ns was used for analysis.

The nanoparticle ligands are hydrophobic alkyl chains capped with either an ammonium cation or a methyl group. Thus, binding of such NPs to NAs can be achieved through contacts with the positively charged end-group (electrostatic and hydrogen bonding) or through hydrophobic contacts with the alkyl chain. Figure 1 shows a comparison of intermolecular contacts of hydrophobic and electrostatic part of the ligand for +6 and +60 NPs with polyT-DNA and polyU-RNA [Figs. 1(d) and 1(e)] and the corresponding simulations snapshots [Figs. 1(f) and 1(g)]. Here, NA residues are considered to be “in contact” with the NP when an interatomic distance of less than 3.5 Å between NAs atoms and either the ammonium cation end-group (charged contact) or of the hydrophobic carbon chain (hydrophobic contact) is achieved. Initially, all NP atoms are farther than 3.5 Å away from NAs and the number of NP/NA atom contacts is zero. The number of NP-NA contacts increases as the simulation progresses, and levels off by about 10 ns simulations time for all systems. Figures 1(d) and 1(e) show that highly charged NPs bind to nucleic acids and reach a stable conformation more rapidly than NPs with low charge. Our simulations indicate that the interaction type between the nanoparticle and NAs depends on the NP charge. The analysis shows that each charged group is able to interact with more than one base; for example, the number of charged contacts between the +6 NP and polyT-DNA and poly-U RNA bases is around 11–14, meaning that each charged ligand can interact with more than one base (Fig. S1).31 For the +6 NP, the interactions of NAs with the hydrophobic part of the NP ligand is more dominant than with the charged end group [Fig. 1(d)]. This trend is reversed for the +60 NP where nucleic acids make a greater number of contacts with the charged ligand end groups than with the hydrophobic part of the ligands. Due to constraints in the ligand movement around the NP, not all of the ligands are able to bind to the nucleic acid chain [Figs. 1(b), 1(c), S3(d), and S3(h)]. This results in a lower number of nucleic acid contacts per charged ligand for the +60 NP relative to the +6 NP. The observed trends are the same independent of sequence and type of nucleic acid (Fig. S2). Increasing nanoparticle charge also increases the number of hydrophobic contacts with the nucleic acid as charged end groups are able to bind with the DNA through both electrostatic and hydrophobic interactions (Fig. S1).

Fig. 1.

Simulation snapshots and NP-NA contacts. Snapshots for NPs with (a) polyT-DNA in solution, with (b) +6 NP and with (c) +60 NP. (d) and (e) NP-NA contacts data for the (d) +6NP and the (e) +60 NP. Number of NA contacts with charged ligand end groups are in red, and the number of contacts with hydrophobic ligand chain are in black for polyT-DNA (solid lines) and polyU-RNA (dashed lines). Snapshots for (f) NPs with polyU-RNA in solution, with (g) +6 NP and with (h) +60 NP.

Fig. 1.

Simulation snapshots and NP-NA contacts. Snapshots for NPs with (a) polyT-DNA in solution, with (b) +6 NP and with (c) +60 NP. (d) and (e) NP-NA contacts data for the (d) +6NP and the (e) +60 NP. Number of NA contacts with charged ligand end groups are in red, and the number of contacts with hydrophobic ligand chain are in black for polyT-DNA (solid lines) and polyU-RNA (dashed lines). Snapshots for (f) NPs with polyU-RNA in solution, with (g) +6 NP and with (h) +60 NP.

Close modal

We observed that change in nucleic acid conformation upon binding with a NP depends on NP charge. One way to detect structural changes is by considering the surface area of the nucleic acid molecules which interacts with the solvent in a complex, or the solvent accessible surface area (SASA). Nanoparticle binding to nucleic acids is expected to decrease the SASA of the NA molecule because the NP and NP ligands blocks part of the surface area of the molecule available to solvent. Shown in Fig. 2 is the change in SASA for polyT-DNA (a), polyA-DNA (b), polyU-RNA (c), and polyA-RNA (d) upon binding with various NPs. The SASA is divided into contributions from the nucleic acid backbone (red) and from the nucleic acid bases (gray) in this plot.

Fig. 2.

Solvent accessible surface area for (a) polyT-DNA, (b) polyU-RNA, (c) polyA-DNA, and (d) polyA-RNA as a function of interactions with various nanoparticles. The contribution to SASA from the backbone and bases are colored as red and gray, respectively.

Fig. 2.

Solvent accessible surface area for (a) polyT-DNA, (b) polyU-RNA, (c) polyA-DNA, and (d) polyA-RNA as a function of interactions with various nanoparticles. The contribution to SASA from the backbone and bases are colored as red and gray, respectively.

Close modal

In general, SASA of hydrophobic molecules is lower than SASA of hydrophilic molecules in aqueous solutions. For example, for the nanoparticles considered here, the SASA for nanoparticle with uncharged alkyl ligands in 0.1 M NaCl is 7.3 × 103 ± 130 Å2 while the SASA for the most highly charged nanoparticle is 9.5 × 103 ± 140 Å2, representing an increase of nearly 30% due to charge or protonation. As the number of charged groups on the nanoparticles increases, the surface area of the nanoparticle increases due to electrostatic repulsion between the end-groups and increase hydrophilicity of the nanoparticle.

The structure of ssNAs in aqueous solution represents partially folded conformation which occurs through minimization of the hydrophobic surface area of the nucleic acid bases and through the formation of intramolecular interactions such as hydrogen bonding and π stacking. Thus, the charged NA backbone is accessible to solvent. In our simulations, we observe that in the absence of NP, the SASA for NA molecules is dominated by the polar, highly charged backbone (Fig. 2). A change in SASA of nucleic acids upon NP binding indicates NA structural changes.

All nucleic acid sequences under study show similar trends in SASA upon interaction with NPs. In the absence of NPs, the backbone SASA is higher than that of the bases due to partially folded conformation of nucleic acids [Figs. 1(a) and 1(f)]. In general, the binding of charged NPs caused a decrease in SASA of the NA backbone, with higher nanoparticle charge causing a greater reduction in backbone SASA. This indicates that interactions of charged ligand with the NA backbone change hydration properties and conformation of the backbone [Figs. 1(a)–1(c) and 1(f)–1(h)]. For ssNAs alone or NAs bound to nanoparticles with low charge (+6), the SASA of the NA backbone remains greater than the SASA of the NA bases and also corresponds to a smaller change in the conformational properties of NAs. However, our results indicate that the ability of a nanoparticle to change the conformation of nucleic acids is related to the nucleic acid sequence. As can be seen in Fig. 2, the pyrimidine samples (polyU-RNA and polyT-DNA) show greater response to nanoparticle binding, and a greater tendency for bases to become exposed. For RNA, a NP charge greater than +30 (50% charge neutralization) causes the SASA of the NA bases to be greater than that of the backbone. Similar behavior occurs for polyT-DNA with +60 NPs (100% charge neutralization). However, for polyA-DNA, this trend is not observed and the SASA of the backbone remains higher than that of the bases. An increase in SASA of the NA bases indicates that the NP can significantly affect the intramolecular interactions within single stranded nucleic acids.

Single stranded DNA does not form a well-defined structure, as compared to dsDNA, even though bases participate in base pairing and base stacking interactions. Due to the formation of these interactions, ssDNA molecule in solution folds on itself to form various structural conformations. For example, poly-guanine and poly-cytosine in DNA can form G-quadruplexes or i-motifs,29 while complementary sequences may form hairpin loops. ssRNA molecules are well-known to fold into complex three-dimensional structures that are particularly important for function.30 

To more clearly define the effect of nanoparticles on nucleic acid structure, we analyzed the intramolecular interactions of the nucleic acid molecules. Table I shows data for the number of unique base to base hydrogen bonds and π stacking interactions as well as the occupancy for these interactions during the last 5 ns in explicit solvent. Occupancy is defined as the average percentage of time an interaction was present for the last 5 ns of simulation time. For example, for polyU-RNA in solution, 134 unique base-base hydrogen bonds were observed, and on average, a hydrogen bond was present for 12.5% of the simulation time.

Table I.

Number of unique hydrogen bonds and π-stacking interactions and their averages for nucleic acids in solution and NA-NP systems. The average hydrogen bonding occupancy is the average percentage of time the interactions were present in the last 5 ns of simulation.

# Unique H-bondsAvg H-bond occupancy (%)# Unique Pi bondsAvg Pi stack occupancy (%)
polyT-DNA 54 17.5 ± 17.2 50 62.5 ± 35.9 
6NH3 + polyT-DNA 49 14.0 ± 15.6 44 59.3 ± 37.2 
30NH3 + polyT-DNA 44 17.0 ± 18.7 44 55.9 ± 39.5 
60NH3 + polyT-DNA 27 11.2 ± 14.0 30 62.13 ± 35.0 
polyU-RNA 134 12.1 ± 13.8 43 28.1 ± 28.6 
6NH3 + polyU-RNA 131 12.0 ± 16.3 31 27.1 ± 27.4 
30NH3 + polyU-RNA 99 12.4 ± 15.6 29 33.1 ± 29.1 
60NH3 + polyU-RNA 105 11.0 ± 14.8 27 31.7 ± 27.7 
# Unique H-bondsAvg H-bond occupancy (%)# Unique Pi bondsAvg Pi stack occupancy (%)
polyT-DNA 54 17.5 ± 17.2 50 62.5 ± 35.9 
6NH3 + polyT-DNA 49 14.0 ± 15.6 44 59.3 ± 37.2 
30NH3 + polyT-DNA 44 17.0 ± 18.7 44 55.9 ± 39.5 
60NH3 + polyT-DNA 27 11.2 ± 14.0 30 62.13 ± 35.0 
polyU-RNA 134 12.1 ± 13.8 43 28.1 ± 28.6 
6NH3 + polyU-RNA 131 12.0 ± 16.3 31 27.1 ± 27.4 
30NH3 + polyU-RNA 99 12.4 ± 15.6 29 33.1 ± 29.1 
60NH3 + polyU-RNA 105 11.0 ± 14.8 27 31.7 ± 27.7 

Binding of highly charged nanoparticles to polyU-RNA or polyT-DNA leads to a decrease in intrastrand hydrogen bonds and π-stacking interactions (Table I). Moreover, the increase in SASA of bases (Fig. 2) is directly correlated with a significant decrease in the number of unique hydrogen bonds and π-stacking interactions between bases. For example, binding of +6 NP does not change the number of unique hydrogen bonds within polyU-RNA as compared to NA in solution; however, interactions with highly charged NPs (+30 or +60 NP) lead to about 20%–25% decrease in the number of unique hydrogen bonds. For polyT-DNA, binding of +60 NP causes ∼40% decrease in the number of unique H-bond and a similar decrease in π-stacking interactions as compared to other NPs. Similar trends were observed for the polyA-DNA and polyA-RNA systems (Table S1). For all systems, the number of unique intramolecular interactions within the NA decreases with nanoparticle binding; however, the occupancy does not change significantly. This suggests that nanoparticle binding disrupts NA intramolecular interactions within its immediate proximity, i.e., only a section of the NA structure closest to the NP binding is disrupted.

The local disruption of the NA structure due to NP binding can be further characterized by the distribution of NA intramolecular interactions. In Fig. 3, scatter plots per NA residue show the location of intramolecular interactions for polyT-DNA and polyU-RNA, where blue circles and red triangles represent π-stacking interactions and hydrogen bonding, respectively. The occupancy of the interaction is indicated by the intensity of the color. Here, for clarity, we show only the intramolecular interactions which had occupancy greater than 20%. For hydrogen bonding, the x-axis represents the H-bond donor, and the y-axis represents the H-bond acceptor. Therefore, red triangles at (20,30), for example, represent a hydrogen bond with the 20th base being the hydrogen bond donor and the 30th base being the hydrogen bond acceptor. For π-stacking interactions, since no donor or acceptor is specified, points at (20,30) and (30,20) are equivalent. Thus, only π interactions on the upper portion of the graph are shown. Plots for polyA-DNA and polyA-RNA are shown in Fig. S2. Line plots adjacent to the scatter plots is the distance between the NA base's and the NPs center of mass, where the dashed straight line shows the NP radius of gyration (Rg). When this NA-NP distance approaches Rg, this means that these NAs bases are within close proximity and directly interacts with NP (shaded squares of the graph, Fig. 3).

Fig. 3.

Intrastrand hydrogen bonding and π stacking interactions for (a)–(d) polyT-DNA and (e)–(h) polyU-RNA for (a) and (e) NA in solutions and NA with NP of charge (b) and (f) +6, (c) and (g) +30, (d) and (h) +60 shown as scatter plots. Line plots show the distance between NA bases and NP center of mass, where the dashed line indicates the radius of gyration of the nanoparticles. Nucleic acid bases which are close to the nanoparticle ligands are indicated by shaded regions. The arrows in (b) show the direction of increasing distance from the NP core. Scatter plots show location of π-stacking interactions (blue circles) and hydrogen bonding (red triangles) on strand. A darker color indicates a higher occupancy. The corresponding snapshots are shown beside the plots.

Fig. 3.

Intrastrand hydrogen bonding and π stacking interactions for (a)–(d) polyT-DNA and (e)–(h) polyU-RNA for (a) and (e) NA in solutions and NA with NP of charge (b) and (f) +6, (c) and (g) +30, (d) and (h) +60 shown as scatter plots. Line plots show the distance between NA bases and NP center of mass, where the dashed line indicates the radius of gyration of the nanoparticles. Nucleic acid bases which are close to the nanoparticle ligands are indicated by shaded regions. The arrows in (b) show the direction of increasing distance from the NP core. Scatter plots show location of π-stacking interactions (blue circles) and hydrogen bonding (red triangles) on strand. A darker color indicates a higher occupancy. The corresponding snapshots are shown beside the plots.

Close modal

For DNA in solution, π-π stacking interactions occur primarily for bases which are next to one another (π stacking is diagonal line across plot) [Figs. 3(a) and 3(e)]. PolyT-DNA shows a region at the end of the molecule, between bases 50 and 60, where many hydrogen bonds between nonadjacent bases occur. PolyU-RNA in solution, in contrast, forms a loop in the middle of the molecule, as indicated by the presence of hydrogen bonds between nonadjacent base pairs.

Nanoparticle binding causes changes in NA intramolecular interactions. For polyT-DNA, even though NPs of varying charge bind to the DNA chain in different locations, the plots show that disruption to the DNA intramolecular structure occurs primarily within the region where nanoparticles are bound. For example, the +6 NP binds to the DNA at bases 9–12, 22–27, 33–4, and 58. These regions show a reduced number of hydrogen bonds, as can be seen by comparing Figs. 3(a) and 3(b). The hydrogen bond-rich region at the end of the molecule (bases 50–60) remains mostly unaffected. In contrast, the +30 NP is bound close to bases 37–60, and the hydrogen bond region at the end of the polyT-DNA strand is disrupted. Binding of the most highly charged NP (+60) completely disrupts ssDNA's initial structure, as all DNA bases, 1–60, shows almost a complete loss of intrastrand hydrogen bonding. Pi stacking interactions for adjacent DNA bases, however, are well maintained. Thus, the decrease in unique π–π stacking interactions observed in Table I occur to lower occupancy π–π stacking interactions which occur through nonadjacent base pairs. For polyU-RNA, the occupancy of π-stacking interactions is much lower than the occupancy for these interactions in polyT-DNA. However, a greater number of hydrogen bonding between bases occurs. Similar to DNA, the intrastrand interactions are affected locally by NP binding.

In this paper, we examined the possibility of noncovalent complexation of single stranded nucleic acids with charged ligand functionalized gold nanoparticles using all-atom molecular dynamics simulations. Our results indicated that complexation depends mostly on the charge of nanoparticle, and, to a lesser extent, sequence and type of nucleic acid. Complexation of NAs with nanoparticles of low charge primarily occurs through hydrophobic rather than electrostatic contacts, while highly charged nanoparticles bind to NAs through electrostatic interactions. Complexation with nanoparticles with charges above +30 causes disruption of nucleic acid structural conformation, loss of hydrogen bonding, and π–π stacking interactions, which occurs simultaneously with an increase in solvent accessible surface area of the nucleic acid bases.

We show that complexation is also sequence dependent, as poly-purines (polyA-DNA and polyA-RNA) show less structural response to NP binding and smaller NA conformational change than the poly-pyrimidines (polyT-DNA and polyU-RNA), with the structure of polyA-DNA changing the least. Sequences containing only polyU-RNA or polyT-DNA show a greater structural response to nanoparticle binding, which is attributed to changes in the NAs solvent accessible surface area and loss of stabilizing intermolecular interactions.

The stability of NP-NA complexes is important for their successful use in gene delivery and nanotechnology applications. Overall, our results show that control over NP binding strength and ssNA structure can be achieved through the use of NPs with a charge of more than +30. Our results indicated that NP charge is crucial for NA structural stability and NA-NP contact area, which may affect the ability of the nucleic acids to hybridize with complementary strands. The three dimensional structure of NA dictates interactions with other molecules; thereby, the changes induced by nanoparticle binding may change behavior and function of NAs.

The authors thank A. L. Kwansa for his administration of the GPU systems used to perform these simulations. This study was supported by the National Science Foundation (CBET-1403871, CMMI-1150682, and DMR-1121107). J.A.N. was supported by the National Science Foundation Graduate Research Fellowship under Grant No. DGE-0946818.

1.
T.
Wirth
,
N.
Parker
, and
S.
Ylä-Herttuala
,
Gene
525
,
162
(
2013
).
2.
C.
Dufès
,
I. F.
Uchegbu
, and
A. G.
Schätzlein
,
Adv. Drug Delivery Rev.
57
,
2177
(
2005
).
3.
P.
Ghosh
,
G.
Han
,
M.
De
,
C. K.
Kim
, and
V. M.
Rotello
,
Adv. Drug Delivery Rev.
60
,
1307
(
2008
).
4.
V.
Sokolova
and
M.
Epple
,
Angew. Chem. Int. Ed.
47
,
1382
(
2008
).
5.
B. D.
Chithrani
,
A. A.
Ghazani
, and
W. C. W.
Chan
,
Nano Lett.
6
,
662
(
2006
).
6.
S. T.
Kim
,
K.
Saha
,
C.
Kim
, and
V. M.
Rotello
,
Acc. Chem. Res.
46
,
681
(
2013
).
7.
T.
Mizuhara
,
D. F.
Moyano
, and
V. M.
Rotello
,
Nano Today
11
,
31
(
2016
).
8.
H.
Liu
,
Z.
Zhu
,
H.
Kang
,
Y.
Wu
,
K.
Sefan
, and
W.
Tan
,
Chemistry
16
,
3791
(
2010
).
9.
S. G.
Elci
 et al,
ACS Nano
10
,
5536
(
2016
).
10.
B.
Nandy
,
P. K.
Maiti
, and
A.
Bunker
,
J. Chem. Theory Comput.
9
,
722
(
2013
).
11.
J. A.
Nash
,
A.
Singh
,
N. K.
Li
, and
Y. G.
Yingling
,
ACS Nano
9
,
12374
(
2015
).
12.
J. G.
Railsback
,
A.
Singh
,
R. C.
Pearce
,
T. E.
McKnight
,
R.
Collazo
,
Z.
Sitar
,
Y. G.
Yingling
, and
A. V.
Melechko
,
Adv. Mater.
24
,
4261
(
2012
).
13.
P. W. K.
Rothemund
,
Nature
440
,
297
(
2006
).
14.
S. H.
Park
,
C.
Pistol
,
S. J.
Ahn
,
J. H.
Reif
,
A. R.
Lebeck
,
C.
Dwyer
, and
T. H.
LaBean
,
Angew. Chem.
118
,
749
(
2006
).
15.
S. M.
Douglas
,
I.
Bachelet
, and
G. M.
Church
,
Science
335
,
831
(
2012
).
16.
J.-S.
Shin
and
N. A.
Pierce
,
J. Am. Chem. Soc.
126
,
10834
(
2004
).
17.
R.
Salomon-Ferrer
,
D. A.
Case
, and
R. C.
Walker
,
Wiley Interdiscip. Rev. Comput. Mol. Sci.
3
,
198
(
2013
).
18.
R.
Salomon-Ferrer
,
A. W.
Götz
,
D.
Poole
,
S.
Le Grand
, and
R. C.
Walker
,
J. Chem. Theory Comput.
9
,
3878
(
2013
).
19.
A. W.
Götz
,
M. J.
Williamson
,
D.
Xu
,
D.
Poole
,
S.
Le Grand
, and
R. C.
Walker
,
J. Chem. Theory Comput.
8
,
1542
(
2012
).
20.
S.
Le Grand
,
A. W.
Götz
, and
R. C.
Walker
,
Comput. Phys. Commun.
184
,
374
(
2013
).
21.
A.
Pérez
,
I.
Marchán
,
D.
Svozil
,
J.
Sponer
,
T. E.
Cheatham
,
C. A.
Laughton
, and
M.
Orozco
,
Biophys. J.
92
,
3817
(
2007
).
22.
R.
Anandakrishnan
,
A.
Drozdetski
,
R. C.
Walker
, and
A. V.
Onufriev
,
Biophys. J.
108
,
1153
(
2015
).
23.
G. D.
Hawkins
,
C. J.
Cramer
, and
D. G.
Truhlar
,
J. Phys. Chem.
100
,
19824
(
1996
).
24.
Z.
Gong
,
Y.
Xiao
, and
Y.
Xiao
,
J. Biomol. Struct. Dyn.
28
,
431
(
2010
).
25.
I. S.
Joung
and
T. E.
Cheatham
,
J. Phys. Chem. B
112
,
9020
(
2008
).
26.
D. R.
Roe
and
T. E.
Cheatham
,
J. Chem. Theory Comput.
9
,
3084
(
2013
).
27.
W.
Humphrey
,
A.
Dalke
, and
K.
Schulten
,
J. Mol. Graphics
14
,
33
(
1996
).
28.
H. J. C.
Berendsen
,
J. P. M.
Postma
,
W. F.
van Gunsteren
,
A.
DiNola
, and
J. R.
Haak
,
J. Chem. Phys.
81
,
3684
(
1984
).
29.
R.
Pederson
 et al, “
Properties of DNA
,” in
Handbook of Nanomaterials Properties
, edited by
B.
Bhushan
,
D.
Luo
,
S. R.
Schricker
,
W.
Sigmund
, and
S.
Zauscher
(
Springer
,
Berlin
,
2014
), pp.
1125
1157
.
30.
S. E.
Butcher
and
A. M.
Pyle
,
Acc. Chem. Res.
44
,
1302
(
2011
).
31.
See supplementary material at http://dx.doi.org/10.1116/1.4966653 for additional data on NP-nucleic acid binding and intrasrand interactions of polyA-DNA and polyA-RNA.

Supplementary Material