Skip to main content
NCBI home page
ACS AuthorChoice logoLink to ACS AuthorChoice
. 2022 Sep 13;126(37):7114–7125. doi: 10.1021/acs.jpcb.2c04649

Computational Modeling of RNA Aptamers: Structure Prediction of the Apo State

Shuting Yan †, Muslum Ilgu †,‡,§, Marit Nilsen-Hamilton †,‡,§, Monica H Lamm †,*
PMCID: PMC9512008  PMID: 36097649

Abstract

graphic file with name jp2c04649_0012.jpg

RNA aptamers are single-stranded oligonucleotides that bind to specific molecular targets with high affinity and specificity. To design aptamers for new applications, it is critical to understand the ligand binding mechanism in terms of the structure and dynamics of the ligand-bound and apo states. The problem is that most of the NMR or X-ray crystal structures available for RNA aptamers are for ligand-bound states. Available apo state structures, mostly characterized by crystallization under nonphysiological conditions or probed by low resolution techniques, might fail to represent the diverse structural variations of the apo state in solution. Here, we develop an approach to obtain a representative ensemble of apo structures that are based on in silico RNA 3D structure prediction and in vitro experiments that characterize base stacking. Using the neomycin-B aptamer as a case study, an ensemble of structures for the aptamer in the apo (unbound) state are validated and then used to investigate the ligand-binding mechanism for the aptamer in complex with neomycin-B.

Introduction

Nucleic acid aptamers are single-stranded oligonucleotides that bind to specific molecular targets with high affinity. The molecular targets vary from small molecules to peptides and proteins.1 Due to their ligand recognition specificity, aptamers are considered as possible substitutes for antibodies and have potential for use in medical applications as therapeutics or diagnostic agents.2 Aptamer-based sensors have been developed for disease biomarkers and environmental and food analysis applications.3−5

Aptamer structures can be resolved at atomic scale by NMR spectroscopy and X-ray crystallography. The problem is that most structures of RNA aptamers deposited in the Protein Data Bank (https://www.rcsb.org/) are for aptamers in the ligand-bound state. Among the aptamer structures available, fewer than 25% are for apo aptamers. Most available apo state structures were determined by crystallographic means6−16 under high salt concentrations, other nonphysiological reagents, low pH, or antibodies as crystallization chaperones. As constrained by the crystal lattice, the observed structural changes are unlikely to represent solution structures. The single crystal structure might not represent the range of structures that apo state RNAs can adopt in solution. It has been pointed out that, in the absence of ligand, the RNA exists as an ensemble of conformations ranging from a minimal tertiary structure to the bound-like state.13 Solution structures can also be probed using methods such as selective 2′-hydroxyl acylation analyzed by primer extension (SHAPE), small-angle x-ray scattering (SAXS), and nucleotide analog interference mapping. However, these are relatively low-resolution techniques that identify changes in flexibility of base positions, global shape, and the influence of substitutions on RNA binding capability and structure. In studies where the apo state is to be investigated by a molecular dynamics (MD) simulation, the initial structure is derived by simply deleting the ligand from a ligand-bound complex. However, it is difficult to assess whether the conformations from MD simulations performed in this way are representative of the apo state without experimental support. Hence, a method to obtain valid apo aptamer structures is of great importance to advance our understanding of the binding mechanisms for aptamers.

The prediction of the 3D structure of an RNA from its primary sequence is a desired route to achieve this goal. Computational modeling plays a major role in providing structural insights based on sequence.17−19 The secondary structure can be predicted by comparative sequence analysis and free energy minimization.20 The tertiary structure can be predicted via three main approaches including ab initio modeling based on physics, comparative modeling based on the homology to the template structure, and knowledge-based modeling based on statistical potential or machine learning.21,22 Many tools have been developed for RNA 3D structure prediction, including RNAComposer,23 iFoldRNA,24 3dRNA,25 FARFAR2,26 SimRNA,27 and MC-Fold|MC-Sym pipeline.28 The MC-Fold|MC-Sym pipeline was selected for structure prediction in this study due to its success in predicting a noncanonical base pair and its use of an exhaustive conformational search. The approach described in this study is not limited to the use of MC-Fold|MC-Sym. In principle, any tool that predicts 3D structures for single-stranded RNA can be used to generate starting structures. The key aspects of RNA 3D structure prediction accuracy are non-Watson–Crick interactions and acceptable clash. The potential energy and clash score are criteria commonly used to select the best candidates from the structure prediction.29−31 It has been shown that the use of MD simulation improves prediction accuracy,32 and MD is often applied to refine the predicted structures with the addition of solvent and counterions.33 Hence, experimental information involving non-Watson–Crick interactions can be utilized to improve prediction accuracy. For example, coaxial stacking observed from experiments was considered a restraint of model selection together with the potential energy profile.31 In this study, base stacking data obtained from 2-aminopurine (2AP) fluorescence experiments for the pentaloop were used to select the ensemble of conformations that are derived from the larger set of 3D structures predicted for the apo state of the neomycin-B aptamer.

Multiple MD simulations started from different initial conditions (conformations or velocities) would greatly improve the sampling compared with a single trajectory, especially from a thermodynamics standpoint. The importance of multiple independent simulations and its application to obtain better statistics on simulation results is gaining recognition.34−37 When multiple independent simulations show reproducible results and convergence of a structural and dynamic process, complete sampling of the thermally accessible conformational ensemble of a biomolecule is considered to be achieved.

Previous studies have shown that presenting an ensemble of molecular conformations, which adhere to certain experimental constraints, provides a means to characterize the inherent flexibility of biomolecules that cannot be achieved by simply examining the conformation for a single structure.38,39 Especially for RNA aptamers in the absence of ligand, it has been shown that, although the stem structures are generally stable, the single-stranded bulges and loop regions, including ligand binding sites, exist as an ensemble of conformations lacking defined structures.36 Thus, the determination of an ensemble of conformations for the apo state, rather than a single structure, is a promising approach to characterize the structure of an apo aptamer. Moreover, an ensemble of conformations with as few structures as feasible would be the desired starting point for an investigation that proceeds with multiple independent MD simulations.

In this study, we conducted a case study to test our approach of apo aptamer structure prediction. The RNA aptamer investigated here is 23 nucleotides long and selectively binds neomycin-B (PDB ID: 1NEM(40)). This aptamer was selected for validation of our approach using the available bound structure determined by NMR as the reference. We performed a clustering procedure on aptamer structures that was predicted from the MC-Fold|MC-Sym pipeline28 for the RNA aptamer sequence. The structures in the pool were clustered on the basis of root mean square deviation (RMSD) cutoffs. Using this procedure, we approximated a representative ensemble of conformations for the apo state of the aptamer by assigning a relative population weight to each cluster on the basis of experimentally determined base stacking of the apo state. The experimental data was obtained using 2AP fluorescence detection with 2AP base substitutions at specific positions in the aptamer. For the aptamer studied here (NEO1A), stacking of the critical bases A14 and A16 identified by Ilgu et al.41 was applied for model selection. To validate this approach for predicting an ensemble of aptamer structures, we compared the predicted ensembles with structures that were obtained by deleting the ligand from the available NMR structure and then running MD simulation. Our results indicate that the ensemble constructed from the MC-Fold|MC-Sym pipeline broadly sampled the conformation space and delineated the fluctuations and correlations of the residues in the aptamer.

In summary, we present an approach to generate an ensemble of conformations for the apo state of an RNA aptamer. Looking forward, this approach for structure prediction of the apo state has the potential to provide insight about binding mechanisms for aptamers.

Methods

The nomenclature used in this paper to describe the aptamer cases is defined in Table 1.

Table 1. Definitions for the Names Used for Aptamers Described in This Work.

name definition reference
1NEM model 5 NMR-derived structure for the aptamer in complex with neomycin-B; fifth model from the PDB entry for 1NEM Jiang et al.40
apo-1NEM structures derived from molecular dynamics simulations for the apo aptamer that began with an apo structure that was derived by deleting the coordinates for neomycin-B from the 1NEM model 5 structure this work
bound-NEO1A aptamer in complex with neomycin-B, studied with 2D-NMR and 2AP fluorescence measurements Ilgu et al.;41 this work
apo-NEO1A aptamer in the absence of ligand, studied with 2D-NMR and 2AP fluorescence measurements Ilgu et al.;41 this work
bound-RNA structures derived for the NEO1A sequence from molecular docking for the representative apo-RNA structures in complex with neomycin-B this work
apo-RNA structures derived for the NEO1A sequence from molecular dynamics simulations for the apo aptamer that began with an ensemble of apo structures predicted in silico this work
Open in a new tab

Obtaining an Ensemble of Conformations from Predicted RNA 3D Models

In this section, we present the procedure developed to determine the ensemble of conformations from the predicted RNA 3D models based on experimental measurements of base stacking in the loop of an aptamer, which has the structure of a stem loop when in complex with its ligand.40 On the basis of the data reported in ref (41), the stacking fractions of A14 and A16 in the apo state measured from steady state fluorescence studies were used here as input to select conformations from the set of predicted models. Here, we present our method for clustering the predicted models by coordinate distance and the assignment of relative populations to clusters of conformations on the basis of the average stacking fraction of A14 and A16 calculated in the cluster. The workflow to construct a representative conformation ensemble of the apo state from RNA 3D structure prediction consists of four major steps as outlined in Scheme 1. In the first step, the 3D models are clustered on the basis of structural similarity. Clusters with small weights are considered insignificant and are discarded. In the second step, structures are clustered into small groups (about 5 structures each). Within each small group, those structures with low energy scores (as determined by MC-Sym) are retained. In the third step, all the remaining structures are clustered and 400 sets of weights are assigned to the clusters to avoid overfitting. In the last step, structures discarded from the previous two steps can be selected and added back to the ensemble for minimal square error of the experimental stacking data. Each step is described in further detail in the following paragraphs.

Scheme 1. Workflow Used to Determine the Conformation Ensemble from an RNA 3D Prediction.

Scheme 1

Open in a new tab

Step 1: Select Representative Clusters from Prediction

Predicted 3D models of the apo aptamer were generated using the MC-Fold|MC-Sym pipeline.28 First, the primary sequence of the aptamer was entered as input to MC-Fold, and secondary structures were returned as output. Second, the secondary structure with the lowest free energy was then submitted to MC-Sym for the 3D structure prediction, from which 1000 putative 3D models were obtained. These models were postprocessed in VMD42 (visual molecular dynamics) for structural alignment, which eliminates the effect of rotation and translation to align the structures prior to clustering.

The coordinates of the predicted models were read by the package Bio3d.43 This was followed by a hierarchical clustering procedure conducted in R.44 A hierarchical clustering of the 1000 structures with 2232 variables (corresponding to the 3D coordinates of 744 atoms) was conducted with the hclust function using the ward.D2 method. The Euclidean distance was used in the clustering procedure. Clusters were obtained by cutting the dendrogram tree at a ratio of 1/2.5 of the maximum height. Here, the 1/2.5 ratio was selected as the cutoff to avoid the size of the clusters being too big or too small. A nucleotide base was considered stacked with another base if the center of mass (COM) distance between the two bases was smaller than 0.5 nm and the center-normal angle was smaller than 50°.45 The average stacking fractions for the bases A14 and A16 were calculated for each cluster, and a 2-by-n matrix, denoted as X (n is the number of clusters), was constructed using the average stacking fraction data. The weights of the clusters were calculated on the basis of the linear model

graphic file with name jp2c04649_m001.jpg

where X is a 2-by-n matrix of base stacking fractions, β is an n-by-1 vector of coefficients, ε is a 2-by-1 vector of errors, and y is a 2-by-1 vector of responses. The stacking fractions for bases A14 and A16 from the 2AP fluorescence experiment26 were used as the response vector, y. The estimate of the unknown β is b, obtained from the following equation

graphic file with name jp2c04649_m002.jpg

where X is not a square matrix.

All clusters with weights smaller than 0.01 were removed until all the clusters had positive weights greater than 0.01.

Step 2: Select Structures with the Lowest Energy from the Clusters

The remaining structures were brought to a second round of clustering for structure selection. The number of clusters was set to be k = number of clusters/5. Here, 5 was selected as a relatively small number to be the size of the clusters for screening low energy structures. The clusters with weights smaller than 0.01 were removed. For the clusters where both bases A14 and A16 stacked in all the structures, only the structures with the minimum score or Pscore from MC-Sym were retained. The procedure continued iteratively until the number of structures was smaller than 20.

Step 3: Avoid Overfitting

The structures retained from last step were clustered the same as for the first round: the number of clusters was determined by cutree at 1/2.5 of the maximum height. N copies of the stacking fraction data were generated from the means and standard deviations of 2AP experimental data. When one assumed the flipping of the bases A14 and A16 was independent and set N equal to 20, there were 400 combinations of the stacking fractions of A14 and A16. Thus, 400 sets of weights were calculated for the clusters. The clusters with mean weight values smaller than the standard deviation were removed due to not consistently having a positive weight.38

Step 4: Add Back Structures

In the last step, structures selected from those removed in the previous step were added back as follows. First, one structure was added back followed by clustering and the weight calculation. The number of clusters was determined as cutting the tree at 1/a of the maximum height. If the sum of weights was smaller than 1, one more structure would be added back whose weight was assigned to be 1 minus the sum of the weights in the previous step. The square error of the weighted stacking fraction to experimental stacking fraction was calculated. The combination of structures together with clustering parameter a, which generated the minimum square error, was chosen as the final ensemble.

Molecular Docking of Neomycin-B to the Predicted Aptamer Models

Molecular docking was performed via AutoDock Vina46 as the first step on the representative structures from apo MD simulations. The COM of the RNA aptamer was used to define the center of the grid box. We defined grid dimensions as a cube of length 50 Å, which is long enough to span the RNA molecule in all three dimensions. Aptamer models were selected as the center of the largest cluster from the second half of the 100 ns MD simulations of the apo state. Ligand coordinates were from model 5 of the NMR structure of NEO1A. The exhaustiveness of the searching algorithm was set to be 100. Twenty docked poses were generated for each model. The second step of docking uses the Amber GB/SA scoring function via Dock6, which involves the minimization/MD/minimization protocol.47 We performed 500 steps of minimization via a conjugate gradient approach and 10 000 steps of MD simulation followed by another 500 steps of minimization. All ligand and aptamer atoms were flexible in the sampling. The final structures were evaluated with the Amber score and ranked.

Molecular Dynamics Simulations of Apo Models in the Ensemble

Structures selected as part of the conformation ensemble using 2AP fluorescence information from the apo-NEO1A aptamer were the initial structures for the apo-RNA simulations. To prepare for the simulations, the models were parametrized with Amber99sb force field using GROMACS.48 Eleven Mg2+ ions were added to VMD42 plugin Cionize for charge neutralization. The systems were then solvated with TIP3P water molecules,49 energy-minimized via the steepest descent, and equilibrated while holding the aptamer and ions fixed. The equilibration step was run for 100 ps using the NVT (298 K) and NPT (298 K and 1 bar) ensembles. Simulations were carried out under constant temperature and pressure (298 K and 1 bar). The aptamer and ionic solvent were independently coupled to external heat baths with a relaxation time of 0.1 ps. The particle Mesh Ewald method50 was used to treat long-range electrostatics. All bonds were constrained using the LINCS algorithm.51 The integration step size was 2 fs, and the final production run for each independent simulation was 100 ns, consistent with other published studies. For example, Banáš et al.52 tested different force fields on the stability of RNA tetraloop hairpins with simulations of 100, 300, 800, and 1000 ns and found that simulations of a 100 ns duration are sufficient for MD characterization of the existing RNA structures.

Ten simulations were conducted with the same protocol for apo-1NEM starting from different initial structures. The initial structures were selected every 10 ps in the 100 ps NPT equilibration under a higher temperature. The purpose for equilibration under a higher temperature was to create variations in the aptamer structure for better sampling.

Parameters for Loop Conformation Characterization

Stacking of loop bases was analyzed with the stacking score defined by Condon et al.45 First, the COM of each base was calculated using non-hydrogen atoms. For adenine, the base plane was defined with COM, C8, and N6. For guanine, the base plane was defined with COM, C8, and O6. For cytosine, the base plane was defined with COM, O2, and N6. For uracil, the base plane was defined with COM, O2, and O6. The stacking scores were then defined by the following three criteria: (1) the distance between the COMs, (2) the center-normal angle (ω), and (3) the normal–normal angle (Ξ) provide a measure of stacking. The center-normal angle ω is the angle between the vector connecting the COMs of two bases and the normal vector of one base plane, measuring the overlap between bases. The normal–normal angle Ξ is the angle between normal vectors of two base planes. The stacking score as a continuous score from −2 to 2 was defined as follows: If the COM distance (d0) ≤ 3.5 Å, the stacking score was 1. If 3.5 Å < d0 < 5.0 Å, the score was decreased as r–3 from 1 to 0. If d0 > 5.0 Å, the bases were considered to be unstacked and the score was 0 in the absence of considering the angles ω and Ξ. If ω ≤ 25°, the score was incremented +1. If 25°< ω ≤ 50°, the score was linearly decreased from 1 to 0. If ω > 50°, bases were considered to be unstacked and the score was 0. The angle Ξ describes whether the base pair is a parallel stack or T-shape. If 45°< Ξ < 135°, the bases are perpendicularly stacked and the stack score was multiplied by −1. Otherwise, the score was multiplied by +1, which indicated parallel stacking. Stacking scores of all possible base pairs formed by A14 or A16 with bases G9-U21 were calculated. When one estimated the stacking fraction of A14 or A16 in the MD simulations, the maximum absolute values of the stacking scores of all possible base pairs were taken as the stacking score for the respective base. This maximum score counted as the strongest stacking interaction, including both parallel and T-shaped stacking, that A14 or A16 could form with another base. If this value was greater than 1, the base A14 or A16 was considered to be in-stack. The average fraction of time being in-stack over MD trajectory was then calculated for A14 and A16.

Chemicals and RNAs

Neomycin-B was obtained as its sulfate salt from Sigma-Aldrich. All RNA oligonucleotides were from Integrated DNA Technologies Inc. (IDT, Coralville, IA) and were maintained at −20 °C in deionized distilled water (ddH2O) until use. Sigma-Aldrich was the source of sodium cacodylate, and cacodylic acid was obtained from Amresco. All other chemicals were from Fisher Scientific. The buffer contained 200 mM NH4CI, 80 mM KCl, 80 mM Na-cacodylate, and 5 mM MgCl2 at pH 7.4. The sequence of the 2AP-labeled RNA aptamer 2AP15NEO1A is given by GGACUGGGCGAXAAGUUUAUCC where X is 2-aminopurine.

Isothermal Titration Calorimetry (ITC)

ITC experiments were performed with a Microcal VP-ITC microcalorimeter (Northampton, MA). Data were analyzed using nonlinear least-squares curve fitting in Origin7.0 (OriginLab Corp.).

NEO1A in the reaction cell and neomycin-B in the syringe were prepared in the same buffer for each experiment as previously described.41 All solutions were degassed at room temperature, and following thermal equilibrium at 25 °C and an initial 60 s delay, 30 serial injections of neomycin-B were added at an interval of 300 s into the stirred sample cell (1.4 mL) containing the NEO1A variant at a stirring rate of 310 rpm at 25 °C. The heat associated with each titration peak was integrated and plotted against the respective molar ratio. Control experiments were performed to correct for the heats of dilution from the titrants by making identical injections of the titrant solution into a cell containing only the respective buffer, and these values were subtracted from the titration of the titrant solution into the reaction cell. Data were analyzed using the standard one-binding site model fitting (nonlinear least-squares curve) in Origin7.0 (OriginLab Corp., Northampton, MA).

Measurement of Steady State 2AP Fluorescence

The 2AP in 2AP15NEO1A provides a measurable fluorescence intensity, which reflects the extent of solvation of the base, and was used to identify the effect of neomycin-B binding on the pentaloop structure of NEO1A. The binding of neomycin-B was monitored by a Cary Eclipse spectrofluorometer (Varian, Palo Alto, CA). The fluorescence of 2AP15NEO1A was measured with quartz cells with a 1 cm path length with excitation (λex) at 307, emission (λem) at 370 nm, and an emission slit of 5 nm.41

To establish the impact of the ligand addition on the loop structure, 1.0 μM 2AP15NEO1A was incubated in Buffer F in the presence and absence of neomycin-B (5 or 10 μM), a condition of saturating ligand concentration. The values for 2AP15NEO1A without ligand were subtracted from the values for 2AP15NEO1A with neomycin to obtain the fluorescence value associated with ligand binding. The effects of aminoglycoside binding to the aptamer were observed as decreased 2AP fluorescence, which indicates a movement of the 2AP toward a more hydrophobic environment such as that present when it is in the stacked position.

Results and Discussion

We begin by presenting experimental data for the behavior of the base G15 of NEO1A in its apo state and upon binding neomycin-B. We then present computational data to establish the validity for the ensemble of conformations that were obtained for the apo state of the neomycin-B binding aptamer through the structure prediction and MD simulation computational workflow.

Base G15 Moves into a Stacked Conformation upon Binding Neomycin-B

The systems apo-2AP15NEO1A and bound-2AP15NEO1A were studied with 2AP fluorescence to determine the positioning of G15 in the apo and bound states. Figure 1A shows that binding of neomycin-B by 2AP15NEO1A follows the same isotherm as the binding of neomycin-B NEO1A, which demonstrates that the aptamer binding parameters are unaffected by the 2AP modification to G15. Figure 1B shows that the fluorescence of 2AP-modified G15 decreases from the apo state to the bound state, indicating that G15 moves into a stack upon binding.

Figure 1.

Figure 1

Open in a new tab

(A) Comparison of heat isotherms for neomycin-B (Neo-B) binding to NEO1A and 2AP15NEO1A. (B) Fluorescence data for 2AP15NEO1A with and without Neo-B in buffers F and N.

Validation Approach and Comparison Data

A validation procedure was established by comparing the computational data generated in this work to the experimental data (this work and ref (41)) for apo-NEO1A. There is not a 3D structure of the apo-NEO1A to which a direct comparison can be made. However, a 3D structure of NEO1A in complex with neomycin-B has been determined by NMR, PDB ID: 1NEM.40 It is reasonable to suppose that the apo-NEO1A might be modeled by deleting the coordinates for the neomycin-B ligand from the complex and subsequently relaxing to the apo-NEO1A structure with MD. Using the alternative procedure described in the Methods, apo-1NEM was generated for comparison.

Ensemble of Apo Aptamer Conformations Produced from the Structure Prediction Workflow

Using the procedures described in the Methods, 11 structures for apo-RNA were selected from the 1000 structures generated from the primary aptamer sequence using the MC-Fold and MC-Sym pipeline. These were selected on the basis of the earlier finding from fluorescence studies41 that bases A14 and A16 are in-stack in the apo state with stacking fractions of 0.93 and 0.91, respectively. We clustered the 1000 predicted structures on the basis of conformational similarity and then applied the base stacking constraints. The constraints for base stacking fractions of A14 and A16 were applied independently, since in the fluorescence studies,41 the 2AP replacements for A14 and for A16 were conducted in separate experiments. When the computational workflow is executed, it is assumed that the base stacking network is unchanged when 2AP replaces an adenine in the aptamer.53 The resulting 11 structures for apo-RNA shown in Figure 2 were used as the initial structures for independent MD simulations. Each simulation included 100 ns of production time. The production trajectories from the independent MD simulations were combined to create an 1100 ns pseudotrajectory for apo-RNA. The apo-RNA pseudotrajectory was analyzed, and structural properties were compared to findings from previous NMR40,41 and 2AP fluorescence studies,41 as discussed below, to validate its suitability as a representation of the apo state conformational ensemble.

Figure 2.

Figure 2

Open in a new tab

Structures of the 11 conformations for apo-RNA determined from the structure prediction workflow. The bases in the stem (green), binding pocket (cyan), and pentaloop regions (magenta) are shown, along with the C12-G18 pair (orange). The pentaloop bases G13, A14, G15, A16, and A17 are labeled for the structure at the top left of the figure.

Stem and Binding Pocket Regions Are Structured in the Apo State

The previous 2D NMR study of apo-NEO1A41 established that the stem and binding pocket for the neomycin-B aptamer are well-structured in the apo state. To determine if the same is true for apo-RNA and apo-1NEM, the root mean square (RMS) fluctuations were calculated for all bases in the aptamer, obtained from multiple independent MD simulations. The RMS results are shown in Figure 3. Overall, the fluctuations are found to be lower for the binding pocket and stem regions, indicating that these regions are more structured than the pentaloop for both apo-RNA and apo-1NEM. The RMS fluctuation for base G15 is lower for apo-1NEM than for apo-RNA.

Figure 3.

Figure 3

Open in a new tab

Root mean square (RMS) fluctuation of residues in (A) apo-RNA and (B) apo-1NEM. The average RMS fluctuation for each case was calculated from the associated multiple independent MD simulations. Error bars indicate the 95% confidence interval.

The Apo State Conformation Has Regions of Overlap with the Conformation Observed in the 1NEM Model 5 Structure Determined for the Aptamer in Complex with Neomycin-B

Using cluster analysis, on the basis of the RMSD, for the apo-RNA and apo-1NEM pseudotrajectories, the dominant clusters for the loop and binding pocket regions of the aptamer were determined. For each dominant cluster, its center was determined and designated as the dominant apo state structure. Then, the dominant apo-RNA and apo-1NEM structures were aligned to the 1NEM model 5 structure, as shown in Figure 4. The binding pocket region for the dominant apo state structures have the same conformation as 1NEM model 5.

Figure 4.

Figure 4

Open in a new tab

Cluster center of the dominant cluster for the loop and binding pocket regions of the apo state determined by the clustering structures obtained from the pseudotrajectory from MD simulation for (A) apo-RNA and (B) apo-1NEM. Clustering is performed with respect to the RMSD. The dominant cluster structure for the apo state (pink) is aligned to the 1NEM model 5 (green). The bases A14, G15, and A16 are annotated for each structure.

The loop region for the dominant apo-RNA structure differs from 1NEM model 5 in the following way. In the apo-RNA structure, base G15 is flipped out, while in 1NEM model 5, base A16 is flipped out. The backbone of the apo-RNA loop is shifted relative to the backbone of the 1NEM model 5 loop, such that the G15 base in apo-RNA resides in a more hydrophilic environment, similar to the positioning in which the A16 base in the 1NEM model 5 is found. The differences observed in the positions of the bases G15 and A16 between the dominant apo-NEO1A structure and the 1NEM model 5 structure are consistent with the previous 2D NMR study41 that showed large chemical shifts for G15 and A16 between the apo and bound states. The loop region for the dominant apo-1NEM structure has the same conformation as 1NEM model 5.

In the Apo State, Base G15 Is Found in a Hydrophilic Environment and Base A16 Is Found in a Hydrophobic Environment

Previous fluorescence studies41 found that A16 exists in a more hydrophobic environment in the apo state compared with the bound state. The fluorescence study in this work found that G15 exists in a more hydrophilic environment in the apo state. To determine the environment of the bases G15 and A16 in apo-RNA and apo-1NEM, the solvent accessible surface area (SASA) was calculated for MD pseudotrajectories. For comparison, the value of SASA for bases G15 and A16 in 1NEM model 5 was also calculated. The distribution of SASA for G15 and A16 in apo-RNA and apo-1NEM is shown in Figure 5, along with the SASA value in 1NEM model 5. The distribution of SASA for G15 in apo-RNA is shifted toward higher values than that for apo-1NEM. This indicates that G15 for apo-RNA is in a more hydrophilic environment, in agreement with the fluorescence results for apo-2AP15NEO1A. The distribution of SASA for A16 in apo-RNA is shifted toward lower values than that for apo-1NEM, indicating a greater likelihood of finding A16 in a hydrophobic environment in apo-RNA. This agrees with the previously reported findings for apo-2AP16NEO1A in which 2AP substitutes for A16 in NEO1A.

Figure 5.

Figure 5

Open in a new tab

Distribution of solvent accessible surface area (SASA) for bases A14, G15, and A16 in the apo state structures from the pseudotrajectories from MD simulation for (A) apo-RNA and (B) apo-1NEM. The SASA for the bases A14, G15, and A16 in the 1NEM model 5 structure is denoted by a dashed blue line.

Base A16 Does Not Stack as Often as Expected

In the apo state, when a particular base appears in a hydrophobic environment, one may presume that base is stacking with other bases in the aptamer.41 For apo-NEO1A, it was determined that A16 has a stacking fraction of 0.91.41 In the MD pseudotrajectories, the base stacking fractions for A16 are found to be 0.5 (apo-RNA) and 0.1 (apo-1NEM).

Bound Complex

Docking was performed on 11 selected models as well as 11 dominant cluster center structures from MD simulations starting from these selected models. The binding score from AutoDock Vina shows that the complexes of the MD dominant structure and neomycin-B are more stable than the starting MC-Sym prediction for most models in Figure 6. Thus, complexes of the MD dominant structure and neomycin-B were refined via Amber score docking in Dock6 for the final bound state results.

Figure 6.

Figure 6

Open in a new tab

Scores obtained after forming aptamer–ligand complexes by docking. Aptamer structures considered are the initial MC-Sym predicted structures and the dominant cluster center structures obtained after performing MD simulations on the initial structures. Twenty docking poses were generated for each aptamer–ligand complex.

The loop bases (A14, G15, and A16) in bound-RNA complexes remained in similar environments as in apo-RNA MD simulations, as shown in Figure 7. For comparison, the Amber scores from docking are also shown for the NMR complexes (PDB ID: 1NEM) in Figure 8. The scores of the lowest free energy complexes from each of the 11 models are all within the range of the 1NEM complexes. The docked structures for model 398 have lower scores than the 1NEM complexes.

Figure 7.

Figure 7

Open in a new tab

Distribution of solvent accessible surface area (SASA) for bases A14, G15, and A16 in the bound state structures for (A) bound-RNA and (B) bound-1NEM. The SASA for bound-RNA is obtained from 20 aptamer–ligand complexes for each of the 11 predicted aptamer models. The SASA for bound-1NEM is obtained from 100 ns MD simulations that start from model 5 of 1NEM. The SASA for the bases A14, G15, and A16 in the 1NEM model 5 structure is denoted by a dashed blue line.

Figure 8.

Figure 8

Open in a new tab

Amber score of complexes from docking.

To further study the conformations of the aptamer in the bound state, base stacking and hydrogen bonding in the loop were investigated. The fraction of structures with base A14 found to be in-stack is shown in Figure 9. Base A14 is in-stack in all the ligand-bound 1NEM complexes (bound-1NEM). In most complexes formed from docking each representative apo-RNA structure to the ligand (bound-RNA), A14 is frequently in-stack. The average number of hydrogen bonds between the loop bases (A14, G15, and A16) and the ligand is shown in Table 2. In the bound-1NEM complexes, base A16 forms 5 hydrogen bonds, on average, with the ligand across the 9 models. In the bound-RNA complex from the prediction, neomycin-B forms one hydrogen bond, on average, with G15 in model 691 and one hydrogen bond, on average, with A16 in model 789. The bound-RNA models predict that neomycin-B is not as tightly surrounded by the flipped base as in the bound-1NEM structures. This suggests that either A16 or G15, but not both, are flipped out-of-stack when interacting with the ligand while A14 remains in-stack. Occurrences of this type would correspond to a GNRA motif. The GNRA motif is present in the bound-RNA complexes, and examples are shown in Figure 10. When A16 is flipped out as the aptamer interacts with the ligand, the other four bases in the pentaloop form a GAGA motif. When G15 is flipped out, the other four bases form a GAAA motif. Thus, the pentaloop is predicted to adjust its conformation for ligand binding with more than one option for conformational adjustment.

Figure 9.

Figure 9

Open in a new tab

Stacking fraction of base A14 in the ligand-bound state. Nine structures are included in bound-1NEM. For the remaining 11 models of bound-RNA, 20 structures are included in each model.

Table 2. Average Number of Hydrogen Bonds between Loop Bases and Neomycin-B.

  average number of hydrogen bonds with neomycin-B
structure A14 G15 A16
1NEM 0 0 5
m168 0 0 0
m266 0 0 0
m283 0 0 0
m398 0 0 0
m43 0 0 0
m448 0 0 0
m596 0 0 0
m626 0 0 0
m691 0 1 0
m789 0 0 1
m821 0 0 0
Open in a new tab

Figure 10.

Figure 10

Open in a new tab

Screenshots from bound-RNA complexes that show the pentaloop adopts a GNRA motif conformation. (A) GAAA motif with G15 flipped out in model 266 bound-RNA. (B) GAGA motif with A16 flipped out in model 789 bound-RNA complex. The pentaloop bases are annotated, and for clarity, the ligand is not shown.

Conclusions

The Ensemble of the Conformations Approach Provides a Reasonable Estimate of the Apo-RNA State

On the basis of the considerations above, the apo-RNA ensemble is a reasonable approximation for the apo state of the neomycin-B aptamer (NEO1A). The apo-RNA structure maintains a structured stem and binding pocket throughout the MD pseudotrajectory. The base G15 in apo-RNA was found to be in a hydrophilic environment, which agrees with fluorescence measurements performed on apo-2AP15NEO1A. The base A16 in apo-RNA was found to exist in a hydrophobic environment yet was not in-stack as often as expected. In the loops for apo-RNA and apo-NEO1A, the GNRA motif is composed of G13-A14-A16-A17 with G15 flipped out. The dominant structure of the apo-RNA loop has similar characteristics to the 1NEM model 5 loop, yet the two loop structures are not in complete agreement. In both structures, the pentaloop forms a GNRA motif with four bases in-stack and the fifth base being flipped out-of-stack. The results suggest that the pentaloop is adjustable as there is more than one option for it to arrange into a GNRA motif. In the 1NEM model 5 loop and bound-NEO1A loop, the GNRA motif is composed of G13-A14-G15-A17 with A16 flipped out, while the bound-RNA loop predicts a GNRA motif G13-A14-A16-A17.

The workflow introduced has been validated for one test case. While the results are promising, further investigations with other aptamer test cases are needed to determine if the workflow is a reliable procedure to generate the 3D structure of the apo state of an aptamer. We emphasize that the workflow presented here can be used with any available 3D structure prediction tool, such as those described in refs (23−28).

Concerning the case of apo-1NEM, it is noteworthy that the structures sampled in the apo-1NEM pseudotrajectory remained fairly close to the 1NEM model 5 structure, despite the fact that the apo structure was equilibrated at high temperature to generate independent starting configurations and the simulations were conducted for similar amounts of time (∼1000 ns). It is important to emphasize that this alternative procedure for modeling the apo-RNA state is only possible for cases where the three-dimensional structure for the aptamer in complex with the ligand has been determined experimentally.

Acknowledgments

This work was supported by the Department of Chemical and Biological Engineering at Iowa State University.

The authors declare no competing financial interest.

References

  1. Krüger A.; de Jesus Santos A. P.; de Sá V.; Ulrich H.; Wrenger C. Aptamer Applications in Emerging Viral Diseases. Pharmaceuticals 2021, 14 (7), 622. 10.3390/ph14070622. [DOI] [PMC free article] [PubMed] [Google Scholar]
  2. Germer K.; Leonard M.; Zhang X. RNA Aptamers and Their Therapeutic and Diagnostic Applications. Int. J. Biochem. Mol. Biol. 2013, 4 (1), 27–40. [PMC free article] [PubMed] [Google Scholar]
  3. Kozitsina A. N.; Svalova T. S.; Malysheva N. N.; Okhokhonin A. V.; Vidrevich M. B.; Brainina K. Z. Sensors Based on Bio and Biomimetic Receptors in Medical Diagnostic, Environment, and Food Analysis. Biosensors 2018, 8 (2), 35. 10.3390/bios8020035. [DOI] [PMC free article] [PubMed] [Google Scholar]
  4. Ozturk M.; Nilsen-Hamilton M.; Ilgu M. Aptamer Applications in Neuroscience. Pharmaceuticals (Basel) 2021, 14 (12), 1260. 10.3390/ph14121260. [DOI] [PMC free article] [PubMed] [Google Scholar]
  5. Banerjee S.; Nilsen-Hamilton M.. Aptamers for Infectious Disease Diagnosis; IntechOpen, 2019; 10.5772/intechopen.86945. [DOI] [Google Scholar]
  6. Huang L.; Wang J.; Lilley D. M. J. The Structure of the Guanidine-II Riboswitch. Cell Chemical Biology 2017, 24 (6), 695–702.e2. 10.1016/j.chembiol.2017.05.014. [DOI] [PMC free article] [PubMed] [Google Scholar]
  7. Jenkins J. L.; Krucinska J.; McCarty R. M.; Bandarian V.; Wedekind J. E. Comparison of a PreQ1 Riboswitch Aptamer in Metabolite-Bound and Free States with Implications for Gene Regulation. J. Biol. Chem. 2011, 286 (28), 24626–24637. 10.1074/jbc.M111.230375. [DOI] [PMC free article] [PubMed] [Google Scholar]
  8. Vicens Q.; Mondragón E.; Batey R. T. Molecular Sensing by the Aptamer Domain of the FMN Riboswitch: A General Model for Ligand Binding by Conformational Selection. Nucleic Acids Res. 2011, 39 (19), 8586–8598. 10.1093/nar/gkr565. [DOI] [PMC free article] [PubMed] [Google Scholar]
  9. Stagno J. R.; Liu Y.; Bhandari Y. R.; Conrad C. E.; Panja S.; Swain M.; Fan L.; Nelson G.; Li C.; Wendel D. R.; White T. A.; Coe J. D.; Wiedorn M. O.; Knoska J.; Oberthuer D.; Tuckey R. A.; Yu P.; Dyba M.; Tarasov S. G.; Weierstall U.; Grant T. D.; Schwieters C. D.; Zhang J.; Ferré-D’Amaré A. R.; Fromme P.; Draper D. E.; Liang M.; Hunter M. S.; Boutet S.; Tan K.; Zuo X.; Ji X.; Barty A.; Zatsepin N. A.; Chapman H. N.; Spence J. C. H.; Woodson S. A.; Wang Y.-X. Structures of Riboswitch RNA Reaction States by Mix-and-Inject XFEL Serial Crystallography. Nature 2017, 541 (7636), 242–246. 10.1038/nature20599. [DOI] [PMC free article] [PubMed] [Google Scholar]
  10. Ramakrishnan S.; Stagno J. R.; Conrad C. E.; Ding J.; Yu P.; Bhandari Y. R.; Lee Y.-T.; Pauly G.; Yefanov O.; Wiedorn M. O.; Knoska J.; Oberthür D.; White T. A.; Barty A.; Mariani V.; Li C.; Brehm W.; Heinz W. F.; Magidson V.; Lockett S.; Hunter M. S.; Boutet S.; Zatsepin N. A.; Zuo X.; Grant T. D.; Pandey S.; Schmidt M.; Spence J. C. H.; Chapman H. N.; Wang Y.-X. Synchronous RNA Conformational Changes Trigger Ordered Phase Transitions in Crystals. Nat. Commun. 2021, 12 (1), 1762. 10.1038/s41467-021-21838-5. [DOI] [PMC free article] [PubMed] [Google Scholar]
  11. Huang L.; Serganov A.; Patel D. J. Structural Insights into Ligand Recognition by a Sensing Domain of the Cooperative Glycine Riboswitch. Mol. Cell 2010, 40 (5), 774–786. 10.1016/j.molcel.2010.11.026. [DOI] [PMC free article] [PubMed] [Google Scholar]
  12. Wilt H. M.; Yu P.; Tan K.; Wang Y.-X.; Stagno J. R. Tying the Knot in the Tetrahydrofolate (THF) Riboswitch: A Molecular Basis for Gene Regulation. J. Struct. Biol. 2021, 213 (1), 107703. 10.1016/j.jsb.2021.107703. [DOI] [PMC free article] [PubMed] [Google Scholar]
  13. Stoddard C. D.; Montange R. K.; Hennelly S. P.; Rambo R. P.; Sanbonmatsu K. Y.; Batey R. T. Free State Conformational Sampling of the SAM-I Riboswitch Aptamer Domain. Structure 2010, 18 (7), 787–797. 10.1016/j.str.2010.04.006. [DOI] [PMC free article] [PubMed] [Google Scholar]
  14. Shelke S. A.; Shao Y.; Laski A.; Koirala D.; Weissman B. P.; Fuller J. R.; Tan X.; Constantin T. P.; Waggoner A. S.; Bruchez M. P.; Armitage B. A.; Piccirilli J. A. Structural Basis for Activation of Fluorogenic Dyes by an RNA Aptamer Lacking a G-Quadruplex Motif. Nat. Commun. 2018, 9 (1), 4542. 10.1038/s41467-018-06942-3. [DOI] [PMC free article] [PubMed] [Google Scholar]
  15. Sjekloća L.; Ferré-D’Amaré A. R. Binding between G Quadruplexes at the Homodimer Interface of the Corn RNA Aptamer Strongly Activates Thioflavin T Fluorescence. Cell Chemical Biology 2019, 26 (8), 1159–1168.e4. 10.1016/j.chembiol.2019.04.012. [DOI] [PMC free article] [PubMed] [Google Scholar]
  16. Huang H.; Suslov N. B.; Li N.-S.; Shelke S. A.; Evans M. E.; Koldobskaya Y.; Rice P. A.; Piccirilli J. A. A G-Quadruplex-Containing RNA Activates Fluorescence in a GFP-like Fluorophore. Nat. Chem. Biol. 2014, 10 (8), 686–691. 10.1038/nchembio.1561. [DOI] [PMC free article] [PubMed] [Google Scholar]
  17. Cruz J. A.; Blanchet M.-F.; Boniecki M.; Bujnicki J. M.; Chen S.-J.; Cao S.; Das R.; Ding F.; Dokholyan N. V.; Flores S. C.; Huang L.; Lavender C. A.; Lisi V.; Major F.; Mikolajczak K.; Patel D. J.; Philips A.; Puton T.; Santalucia J.; Sijenyi F.; Hermann T.; Rother K.; Rother M.; Serganov A.; Skorupski M.; Soltysinski T.; Sripakdeevong P.; Tuszynska I.; Weeks K. M.; Waldsich C.; Wildauer M.; Leontis N. B.; Westhof E. RNA-Puzzles: A CASP-like Evaluation of RNA Three-Dimensional Structure Prediction. RNA 2012, 18 (4), 610–625. 10.1261/rna.031054.111. [DOI] [PMC free article] [PubMed] [Google Scholar]
  18. Miao Z.; Adamiak R. W.; Blanchet M.-F.; Boniecki M.; Bujnicki J. M.; Chen S.-J.; Cheng C.; Chojnowski G.; Chou F.-C.; Cordero P.; Cruz J. A.; Ferré-D’Amaré A. R.; Das R.; Ding F.; Dokholyan N. V.; Dunin-Horkawicz S.; Kladwang W.; Krokhotin A.; Lach G.; Magnus M.; Major F.; Mann T. H.; Masquida B.; Matelska D.; Meyer M.; Peselis A.; Popenda M.; Purzycka K. J.; Serganov A.; Stasiewicz J.; Szachniuk M.; Tandon A.; Tian S.; Wang J.; Xiao Y.; Xu X.; Zhang J.; Zhao P.; Zok T.; Westhof E. RNA-Puzzles Round II: Assessment of RNA Structure Prediction Programs Applied to Three Large RNA Structures. RNA 2015, 21 (6), 1066–1084. 10.1261/rna.049502.114. [DOI] [PMC free article] [PubMed] [Google Scholar]
  19. Miao Z.; Adamiak R. W.; Antczak M.; Batey R. T.; Becka A. J.; Biesiada M.; Boniecki M. J.; Bujnicki J. M.; Chen S.-J.; Cheng C. Y.; Chou F.-C.; Ferré-D’Amaré A. R.; Das R.; Dawson W. K.; Ding F.; Dokholyan N. V.; Dunin-Horkawicz S.; Geniesse C.; Kappel K.; Kladwang W.; Krokhotin A.; Łach G. E.; Major F.; Mann T. H.; Magnus M.; Pachulska-Wieczorek K.; Patel D. J.; Piccirilli J. A.; Popenda M.; Purzycka K. J.; Ren A.; Rice G. M.; Santalucia J.; Sarzynska J.; Szachniuk M.; Tandon A.; Trausch J. J.; Tian S.; Wang J.; Weeks K. M.; Williams B.; Xiao Y.; Xu X.; Zhang D.; Zok T.; Westhof E. RNA-Puzzles Round III: 3D RNA Structure Prediction of Five Riboswitches and One Ribozyme. RNA 2017, 23 (5), 655–672. 10.1261/rna.060368.116. [DOI] [PMC free article] [PubMed] [Google Scholar]
  20. Seetin M. G.; Mathews D. H.. RNA Structure Prediction: An Overview of Methods In Bacterial Regulatory RNA; Methods in Molecular Biology; Humana Press, 2012; Vol. 905, pp 99–122; 10.1007/978-1-61779-949-5_8. [DOI] [PubMed] [Google Scholar]
  21. Dufour D.; Marti-Renom M. A. Software for Predicting the 3D Structure of RNA Molecules. Wiley Interdisciplinary Reviews: Computational Molecular Science 2015, 5 (1), 56–61. 10.1002/wcms.1198. [DOI] [Google Scholar]
  22. Li J.; Chen S.-J. RNA 3D Structure Prediction Using Coarse-Grained Models. Front. Mol. Biosci. 2021, 8, 720937. 10.3389/fmolb.2021.720937. [DOI] [PMC free article] [PubMed] [Google Scholar]
  23. Popenda M.; Szachniuk M.; Antczak M.; Purzycka K. J.; Lukasiak P.; Bartol N.; Blazewicz J.; Adamiak R. W. Automated 3D Structure Composition for Large RNAs. Nucleic Acids Res. 2012, 40 (14), e112 10.1093/nar/gks339. [DOI] [PMC free article] [PubMed] [Google Scholar]
  24. Ding F.; Sharma S.; Chalasani P.; Demidov V. V.; Broude N. E.; Dokholyan N. V. Ab Initio RNA Folding by Discrete Molecular Dynamics: From Structure Prediction to Folding Mechanisms. RNA 2008, 14 (6), 1164–1173. 10.1261/rna.894608. [DOI] [PMC free article] [PubMed] [Google Scholar]
  25. Wang J.; Wang J.; Huang Y.; Xiao Y. 3dRNA v2.0: An Updated Web Server for RNA 3D Structure Prediction. International Journal of Molecular Sciences 2019, 20 (17), 4116. 10.3390/ijms20174116. [DOI] [PMC free article] [PubMed] [Google Scholar]
  26. Watkins A. M.; Rangan R.; Das R. FARFAR2: Improved De Novo Rosetta Prediction of Complex Global RNA Folds. Structure 2020, 28 (8), 963–976.e6. 10.1016/j.str.2020.05.011. [DOI] [PMC free article] [PubMed] [Google Scholar]
  27. Boniecki M. J.; Lach G.; Dawson W. K.; Tomala K.; Lukasz P.; Soltysinski T.; Rother K. M.; Bujnicki J. M. SimRNA: A Coarse-Grained Method for RNA Folding Simulations and 3D Structure Prediction. Nucleic Acids Res. 2016, 44 (7), e63–e63. 10.1093/nar/gkv1479. [DOI] [PMC free article] [PubMed] [Google Scholar]
  28. Parisien M.; Major F. The MC-Fold and MC-Sym Pipeline Infers RNA Structure from Sequence Data. Nature 2008, 452 (7183), 51–55. 10.1038/nature06684. [DOI] [PubMed] [Google Scholar]
  29. Athyala P. K.; Kanwar J. R.; Alameen M.; Kanwar R. K.; Krishnakumar S.; Watson J.; Vetrivel U.; Narayanan J. Probing the Biophysical Interaction between Neocarzinostatin Toxin and EpCAM RNA Aptamer. Biochem. Biophys. Res. Commun. 2016, 469 (2), 257–262. 10.1016/j.bbrc.2015.11.109. [DOI] [PubMed] [Google Scholar]
  30. Ahirwar R.; Nahar S.; Aggarwal S.; Ramachandran S.; Maiti S.; Nahar P. In Silico Selection of an Aptamer to Estrogen Receptor Alpha Using Computational Docking Employing Estrogen Response Elements as Aptamer-Alike Molecules. Sci. Rep 2016, 6, 21285. 10.1038/srep21285. [DOI] [PMC free article] [PubMed] [Google Scholar]
  31. Huang W.; Kim J.; Jha S.; Aboul-ela F. The Impact of a Ligand Binding on Strand Migration in the SAM-I Riboswitch. PLoS Comput. Biol. 2013, 9 (5), e1003069 10.1371/journal.pcbi.1003069. [DOI] [PMC free article] [PubMed] [Google Scholar]
  32. Jeddi I.; Saiz L. Three-Dimensional Modeling of Single Stranded DNA Hairpins for Aptamer-Based Biosensors. Sci. Rep. 2017, 7 (1), 1178. 10.1038/s41598-017-01348-5. [DOI] [PMC free article] [PubMed] [Google Scholar]
  33. Golub E.; Albada H. B.; Liao W.-C.; Biniuri Y.; Willner I. Nucleoapzymes: Hemin/G-Quadruplex DNAzyme-Aptamer Binding Site Conjugates with Superior Enzyme-like Catalytic Functions. J. Am. Chem. Soc. 2016, 138 (1), 164–172. 10.1021/jacs.5b09457. [DOI] [PubMed] [Google Scholar]
  34. Suresh G.; Srinivasan H.; Nanda S.; Priyakumar U. D. Ligand-Induced Stabilization of a Duplex-like Architecture Is Crucial for the Switching Mechanism of the SAM-III Riboswitch. Biochemistry 2016, 55 (24), 3349–3360. 10.1021/acs.biochem.5b00973. [DOI] [PubMed] [Google Scholar]
  35. Yildirim I.; Kierzek E.; Kierzek R.; Schatz G. C. Interplay of LNA and 2′-O-Methyl RNA in the Structure and Thermodynamics of RNA Hybrid Systems: A Molecular Dynamics Study Using the Revised AMBER Force Field and Comparison with Experimental Results. J. Phys. Chem. B 2014, 118 (49), 14177–14187. 10.1021/jp506703g. [DOI] [PubMed] [Google Scholar]
  36. Warfield B. M.; Anderson P. C. Molecular Simulations and Markov State Modeling Reveal the Structural Diversity and Dynamics of a Theophylline-Binding RNA Aptamer in Its Unbound State. PLoS One 2017, 12 (4), e0176229 10.1371/journal.pone.0176229. [DOI] [PMC free article] [PubMed] [Google Scholar]
  37. Negi I.; Mahmi A. S.; Seelam Prabhakar P.; Sharma P. Molecular Dynamics Simulations of the Aptamer Domain of Guanidinium Ion Binding Riboswitch YkkC-III: Structural Insights into the Discrimination of Cognate and Alternate Ligands. J. Chem. Inf. Model. 2021, 61 (10), 5243–5255. 10.1021/acs.jcim.1c01022. [DOI] [PubMed] [Google Scholar]
  38. Vammi V.; Lin T.-L.; Song G. Enhancing the Quality of Protein Conformation Ensembles with Relative Populations. Journal of biomolecular NMR 2014, 58 (3), 209–225. 10.1007/s10858-014-9818-2. [DOI] [PubMed] [Google Scholar]
  39. Zhu G.; Liu W.; Bao C.; Tong D.; Ji H.; Shen Z.; Yang D.; Lu L. Investigating Energy-based Pool Structure Selection in the Structure Ensemble Modeling with Experimental Distance Constraints: The Example from a Multidomain Protein Pub1. Proteins: Struct., Funct., Bioinf. 2018, 86, 501. 10.1002/prot.25468. [DOI] [PubMed] [Google Scholar]
  40. Jiang L.; Majumdar A.; Hu W.; Jaishree T. J.; Xu W.; Patel D. J. Saccharide-RNA Recognition in a Complex Formed between Neomycin B and an RNA Aptamer. Structure 1999, 7 (7), 817–827. 10.1016/S0969-2126(99)80105-1. [DOI] [PubMed] [Google Scholar]
  41. Ilgu M.; Fulton D. B.; Yennamalli R. M.; Lamm M. H.; Sen T. Z.; Nilsen-Hamilton M. An Adaptable Pentaloop Defines a Robust Neomycin-B RNA Aptamer with Conditional Ligand-Bound Structures. RNA 2014, 20 (6), 815–824. 10.1261/rna.041145.113. [DOI] [PMC free article] [PubMed] [Google Scholar]
  42. Humphrey W.; Dalke A.; Schulten K. VMD: Visual Molecular Dynamics. J. Mol. Graphics 1996, 14 (1), 33–38. 10.1016/0263-7855(96)00018-5. [DOI] [PubMed] [Google Scholar]
  43. Grant B. J.; Rodrigues A. P. C.; ElSawy K. M.; McCammon J. A.; Caves L. S. D. Bio3d: An R Package for the Comparative Analysis of Protein Structures. Bioinformatics 2006, 22 (21), 2695–2696. 10.1093/bioinformatics/btl461. [DOI] [PubMed] [Google Scholar]
  44. R: The R Project for Statistical Computing; https://www.r-project.org/ (accessed 2018-07-07).
  45. Condon D. E.; Kennedy S. D.; Mort B. C.; Kierzek R.; Yildirim I.; Turner D. H. Stacking in RNA: NMR of Four Tetramers Benchmark Molecular Dynamics. J. Chem. Theory Comput. 2015, 11 (6), 2729–2742. 10.1021/ct501025q. [DOI] [PMC free article] [PubMed] [Google Scholar]
  46. Trott O.; Olson A. J. AutoDock Vina: Improving the Speed and Accuracy of Docking with a New Scoring Function, Efficient Optimization and Multithreading. J. Comput. Chem. 2010, 31 (2), 455–461. 10.1002/jcc.21334. [DOI] [PMC free article] [PubMed] [Google Scholar]
  47. Lang P. T.; Brozell S. R.; Mukherjee S.; Pettersen E. F.; Meng E. C.; Thomas V.; Rizzo R. C.; Case D. A.; James T. L.; Kuntz I. D. DOCK 6: Combining Techniques to Model RNA-Small Molecule Complexes. RNA 2009, 15 (6), 1219–1230. 10.1261/rna.1563609. [DOI] [PMC free article] [PubMed] [Google Scholar]
  48. Abraham M. J.; Murtola T.; Schulz R.; Páll S.; Smith J. C.; Hess B.; Lindahl E. GROMACS: High Performance Molecular Simulations through Multi-Level Parallelism from Laptops to Supercomputers. SoftwareX 2015, 1–2, 19–25. 10.1016/j.softx.2015.06.001. [DOI] [Google Scholar]
  49. Jorgensen W. L.; Chandrasekhar J.; Madura J. D. Comparison of Simple Potential Functions for Simulating Liquid Water. J. Chem. Phys. 1983, 79 (2), 926–935. 10.1063/1.445869. [DOI] [Google Scholar]
  50. Essmann U.; Perera L.; Berkowitz M. L. A Smooth Particle Mesh Ewald Method. J. Chem. Phys. 1995, 103 (19), 8577–8593. 10.1063/1.470117. [DOI] [Google Scholar]
  51. Hess B.; Bekker H.; Berendsen H. J. C.; Fraaije J. G. E. M. LINCS: A Linear Constraint Solver for Molecular Simulations. J. Comput. Chem. 1997, 18, 1463–1472. . [DOI] [Google Scholar]
  52. Banáš P.; Hollas D.; Zgarbová M.; Jurečka P.; Orozco M.; Cheatham T. E.; Šponer J.; Otyepka M. Performance of Molecular Mechanics Force Fields for RNA Simulations: Stability of UUCG and GNRA Hairpins. J. Chem. Theory Comput. 2010, 6 (12), 3836–3849. 10.1021/ct100481h. [DOI] [PMC free article] [PubMed] [Google Scholar]
  53. HALL K. B.; WILLIAMS D. J. Dynamics of the IRE RNA Hairpin Loop Probed by 2-Aminopurine Fluorescence and Stochastic Dynamics Simulations. RNA 2004, 10 (1), 34–47. 10.1261/rna.5133404. [DOI] [PMC free article] [PubMed] [Google Scholar]

Articles from The Journal of Physical Chemistry. B are provided here courtesy of American Chemical Society

RESOURCES