Conformational preferences of inosine and its methyl derivatives: Comparison of the AMBER derived force field parameters and reparameterization of the glycosidic torsion parameters

doi:10.21203/rs.3.rs-2486374/v1

Inosine is one of the most abundant post-transcriptionally modified ribonucleosides which is known to play a major role in several important biological processes and is of great therapeutic importance. The growing importance of this modified ribonucleoside in therapeutics suggests the requirement of further theoretical studies involving inosine and its derivatives and ensuring the accuracy of their force field parameters is crucial for such theoretical studies to be reliable. The present study reports the validation of the AMBER derived force field parameter sets for inosine as well as examination of the transferability of the available revised sets of glycosidic and gamma torsion parameters corresponding to the respective canonical nucleosides based on detailed comparison of different conformational features from replica exchange molecular dynamics. We also report newly developed sets of partial atomic charges and glycosidic torsion parameters (𝛘_KOL0) for inosine and its methyl derivatives. These parameters, in combination with the AMBER FF99 parameters (Cheatham, T. E., III; Cieplak, P.; Kollman, P. A. J. Biomol. Struct. Dyn. 1999, 16, 845 − 862; Aduri, R.; Psciuk, B. T.; Saro, P.; Taniga, H.; Schlegel, H. B.; SantaLucia, J. J. Chem. Theory Comput. 2007, 3, 1464 − 1475), and the recommended bsc0 correction for the gamma torsion (Pérez, A.; Marchán, I.; Svozil, D.; Sponer, J.; Cheatham, T. E.; Laughton, C. A.; Orozco, M. Biophys. J. 2007, 3817 − 3829.), reproduced the conformational properties of inosine and its 1-methyl derivative in agreement with experimental (NMR) data. In this study, we have also predicted the conformational preferences for the other two methyl derivatives of inosine, i.e., 2’-O-methylinosine and 1,2’-O-dimethylinosine using the revised sets of glycosidic torsion parameters.

Post-transcriptional modifications

inosine

methyl derivatives

conformational preferences

force field parameters

Post transcriptional modifications are ubiquitously found in RNA. The MODOMICS database currently reports more than 140 such naturally occurring post-transcriptionally modified nucleosides [1]. Some post-transcriptional modifications have been found to be conserved across various organisms [2]. Some of these post-transcriptional modifications play an important role in the folding and stability of RNA molecules and can also have a significant impact on RNA-protein interactions [2–4].

The inosine (I) modification was the first noncanonical nucleotide which was identified within the anticodon loop of tRNA at the first (wobble) position (position 34) [5]. It is one of the most abundant post-transcriptional modifications in RNA. Hydrolytic deamination of adenosine (A) at C6 position catalyzed by the family of enzymes called ADARs (adenosine deaminases that act on RNA) results in the formation of inosine [6, 7]. This process is known as A to I editing of RNA. The family of enzymes known as ADATs (Adenosine Deaminases that act on tRNA) catalyze the A to I editing at the wobble position of tRNA [6, 7]. This modification has also been identified in mRNA (as an epigenetic modification), primarily in the non-coding regions, e.g. 5’-UTRs, 3’-UTRs and intronic regions mostly within the Alu repeat sequences [7–12]. Inosine has also been found to occur at position 37 in the tRNA^Ala anticodon loop [13]. Interestingly, deamination of adenosine to inosine has also been reported to occur in DNA/RNA hybrid structures [14]. A to I editing has been reported to have significant impact on the structure and stability of RNA [15–17], RNA interference [18–21], splicing [9, 22–24], protein recoding [7, 25, 26], innate immune response [27, 28] and viral defense [29–31] etc. Alteration in the regulation of the A to I editing process leads to diseases, e.g., autoimmune disorders [32–34], neurological disorders [35–37], different types of cancer [38–41], etc. The potential therapeutic activities, e.g., antiviral and antitumor activities of inosine and synthetic inosine analogues have been reported in several studies [42–45].

The structure of inosine resembles the structure of the canonical nucleoside guanosine except for the exocyclic amino group (the amino group attached to the C2 atom) [6, 46] (Fig. 1). Hence this modification is recognized as guanosine by ribosome and other cellular machinery and this can lead to protein recoding, i.e., the recoding of genetic information [13, 47]. An NMR study reports a large predominance of SYN population and preference for the C2′-endo (SOUTH) sugar pucker conformation of inosine in solution [48]. Several theoretical and experimental studies have been conducted to investigate the structural and thermodynamic consequences due to the presence of inosine in RNA duplexes [6, 49–53]. Notably, the nearest neighbor thermodynamic parameters for I·U and I·C base pairs have been developed by Znosko and colleagues [50, 51].

There are three naturally occurring derivatives of inosine (I; INO) reported so far, which are 1-methylinosine (m¹I; 1MI), 2’-O-methylinosine (Im; MRI) and 1,2’-O-dimethylinosine (m¹Im; MMI) [1]. The three letter codes for the modified residues in the present study are according to Aduri et al. [54]. 1-methylinosine and 1,2’-O-dimethylinosine have been reported to occur in tRNA [2, 55–60] and 2’-O-methylinosine (Im) has been identified in rRNA [61]. NMR data suggests that 1-methylinosine prefers the SOUTH conformation of sugar pucker and g⁺ conformation of 𝛄 torsion [62].

Force field parameters derived from the Aduri et al. [54] parameters for inosine and the methyl derivatives of inosine under this study are available within AMBER. In the present study, we have validated/investigated/examined the transferability of the different glycosidic torsion parameters and 𝛄 torsion parameters for the canonical nucleoside adenosine to inosine in combination with the parameters for inosine provided by Aduri et al. [54]. In a recent study, Sakuraba et al. [53] applied the glycosidic torsion parameters (𝛘_OL3) for adenosine developed by Zgarbová et al. [63] in combination with the AMBER99 parameters for the standard nucleic acids [64] and the bsc0 𝛄 parameters [65] to inosine in RNA duplexes and derived the nearest neighbor parameters for I·C base pairs in good agreement with experimental data. We have now compared the effect of the application of different combinations of available parameter sets with the objective of determining the most effective strategy to replicate the experimentally observed conformational distribution of inosine as well as its methyl derivatives under this study.

Computational details

GAUSSIAN09 [66] software suite was used to perform all the quantum mechanical (QM) calculations. All molecular mechanics and molecular dynamics calculations were carried out using the AMBER16 [67] molecular modeling suite.

Ab initio potential energy surface (PES) scan

For the preparation of the initial geometries of the modified ribonucleoside inosine (INO) and three of its methyl derivatives 1-methylinosine (1MI), 2’-O-methylinosine (MRI) and 1,2’-O-dimethylinosine (MMI), the mean values of the bonds, angles and dihedral angles for the ribose sugars reported by Gelbin et al. [68] were incorporated using the molecular structure editor MOLDEN [69]. Planar geometries were considered for the base moieties. The geometries of the modified ribonucleosides were kept either in the C3′-endo/g⁺ conformation or in the C2′-endo/g⁺ conformation by fixing the corresponding torsional angles to definite values. The 𝛄 torsion angle (O5′-C5′-C4′-C3′) was fixed at 54° corresponding to the g⁺ conformation as observed in the standard A-form RNA [70]. The 𝜹 (C5′-C4′-C3′-O3′) and O4′-C1′-C2′-C3′ dihedral angles were fixed at 81° and − 24°, respectively to compel the nucleoside geometries to stay in the C3’-endo sugar puckering. To force the geometries to the C2′-endo sugar pucker conformation, the values of the 𝜹 (C5′-C4′-C3′-O3′) and O4′-C1′-C2′-C3′ dihedral angles were set to 140° and 32° respectively. Four initial geometries (i.e., SC1, SC2, SC3 and SC4) with definite values of the H5T-O5′-C5′-C4′ and C1′-C2′-O2′-HO2′ torsional angles were prepared for each of the modified nucleosides, to either facilitate or obstruct the base-sugar hydrogen bonding interactions by keeping the nucleosides either in C3′-endo or in C2′-endo sugar pucker conformation (Table S1). The schemes SC1-SC4 were chosen following the values of the torsional angles corresponding to the four schemes chosen in Yildirim et al. [71]. The SC1 and SC2 conformational schemes were kept in the C2′-endo conformation while SC3 and SC4 were kept in the C3′-endo conformation. To obstruct any hydrogen bonding interaction between H3T or O2′ and base, so that these interactions do not affect the glycosidic torsion energy profile, the C4′-C3′-O3′-H3T torsion was fixed at -148° for all the initial geometries.

For each of initial geometries for each of the modified residues, a gas phase PES scan was executed around the glycosidic torsion angle (O4′-C1′-N9-C8) with an increase in its value by 5° resulting in 72 conformations for each nucleoside geometry. During the PES scan, the structures were optimized using the HF/6-31G* level of theory. During the geometry optimization step, the dihedral angles mentioned in Table S1, were kept frozen to obtain a smooth QM energy profile. The QM energies (E_QM) corresponding to each of the 72 conformations (for each scheme) were calculated using the MP2/6-31G* level of theory (Fig. 2). The quantum mechanically optimized structures (along with the atom names) of inosine and the three of its methyl derivatives are shown in Figure S1.

RESP fitting

For the reoptimization of the partial atomic charges for the modified residues, we followed a multiconformational approach using geometry optimized minimum energy conformations for all four schemes. The new sets of partial atomic charges for inosine and its methyl derivatives were developed corresponding to the quantum mechanically optimized structures using the RESP (Restrained Electrostatic Potential fitting) fitting method [72, 73] using the R.E.D. version III.52 perl program [74]. The partial atomic charges for the atoms of the ribonucleosides are provided in the supporting information (Table S2).

Molecular mechanical energy minimization

The molecular mechanical (MM) energies (E_MM) were calculated for the 288 quantum mechanically (QM) optimized conformers (corresponding to the four schemes) for inosine and its methyl derivatives (Fig. 3). During the MM energy minimizations, the dihedral angles (as mentioned in Table S1) were restrained to the values corresponding to the QM optimized geometries by application of a force constant of 1500 Kcal/mol-rad². The starting structure for the MM energy minimization step were the structures identical with the QM optimized geometry obtained from the PES scan. The 5′-phosphate group was replaced with a hydrogen (5′-OH) and a hydrogen atom (3′-OH) was added to the 3′ end of the original topology provided by Aduri et al. [54] to create the topologies for the modified nucleosides used in this study. The parameters corresponding to the 5′-OH and 3′-OH groups were obtained from the FF99 force field parameter set [64]. During the MM energy minimization, all the glycosidic torsion parameters corresponding to AMBER derived parameter sets [54, 64] were set to 0. The MM energy minimization consisted of 500 steps of steepest descent minimization followed by 99500 steps of conjugate gradient minimization. A long-range cut-off of 12 Å was applied to the non-bonded interactions during the energy minimization in vacuum.

Fitting glycosidic torsion (𝛘) parameters

The potential energy due to the glycosidic torsion (𝛘) angle is represented by the differences between the QM energies (E_QM) and the MM energies (E_MM) (as shown in Eq. 1).

E_𝛘 = E_QM - E_MM (1)

For each modification, the 4 x 72 = 288 E_𝛘 values corresponding to the 288 conformations obtained from Eq. (1) were fitted to the Fourier (cosine) series shown in Eq. (2) (V_n1, V_{n2 ,} 𝞍₁ and 𝞍₂ are the fitting parameters).

Where 𝞍₁ and 𝞍₂ are the dihedral angles (O4’-C1’-N9-C8) and (C2’-C1’-N9-C8) respectively. V_n1 and V_n2 represent the potential energy barriers around the (O4’-C1’-N9-C8) and (C2’-C1’-N9-C8) dihedral angles. A separate fitting was carried out to calculate V_n1 and V_n2 for each of the residues.

Molecular mechanical (MM) energies were re-calculated corresponding to the 288 quantum mechanically (QM) optimized geometries of inosine and each of the methyl derivatives of inosine in this study with the new sets of glycosidic torsion parameters sets FF99_𝛘_KOL0 (Figs. 3, S2-4). The revised sets of glycosidic torsion parameters for inosine and its methyl derivatives are tabulated in Table 1.

Table 1

Revised sets of glycosidic torsional parameters for inosine (INO) and its methyl derivatives (1MI, MRI and MMI).
	Modified nucleoside/s	Torsion/s	n	V_n
𝛘_KOL0	INO	O4’-C1’-N9-C8 (OS-CT-N*-Q1)	1 2 3 4	0.0464629 1.21305 -0.0132474 0.458791
	INO	C2’-C1’-N9-C8 (CT-CT-N*-Q1)	1 2 3 4	1.1051 0.0343393 -0.00648998 0.560197
	1MI	O4’-C1’-N9-C8 (OS-CT-N*-Q1)	1 2 3 4	0.246844 1.38382 0.130711 0.651255
		C2’-C1’-N9-C8 (CT-CT-N*-Q1)	1 2 3 4	1.30067 0.197192 0.15392 0.749508
	MRI	O4’-C1’-N9-C8 (OS-CT-N*-Q1)	1 2 3 4	-0.131643 1.01949 -0.108929 0.281191
	MRI	C2’-C1’-N9-C8 (CT-CT-N*-Q1)	1 2 3 4	0.981303 -0.272234 -0.154972 0.506944
	MMI	O4’-C1’-N9-C8 (OS-CT-N*-Q1)	1 2 3 4	-0.0897368 0.985703 -0.132695 0.305859
	MMI	C2’-C1’-N9-C8 (CT-CT-N*-Q1)	1 2 3 4	1.0236 -0.26396 -0.195876 0.515077

Replica exchange molecular dynamics simulations

The REMD [75] simulations were carried out using the multisander approach in AMBER16 [67] molecular modeling suite. The PDB format files corresponding to the quantum mechanically optimized geometries and the parameter and topology files for the modified residues provided by Aduri et al. [54] are available in the AMBER18 package [76] and these files were modified to obtain the initial structures of the modified nucleosides under this study. The starting structures of the modified nucleosides for REMD simulations were in the NORTH/g⁺/ANTI conformation. The tLEAP module (available in the AMBER18 package) was used for the addition of hydrogens, neutralization and solvation of the systems. Each of the modified residues was solvated using TIP3P [77] water molecules within a truncated octahedral box having the closest distance of 9 Å between the edge of the box and any solute atom. For inosine, simulations were carried out using the Aduri et al. [54] parameters for inosine (mentioned as ff_A) and different combinations of the Aduri et al. [54] parameters with the revised sets of parameters for the glycosidic torsion (𝛘) and gamma (𝛄) torsion angles, i.e., ff_A [54], ff_A_𝛄_bsc0 [54, 65], ff_A_𝛘_OL3 [54, 63], ff_A_𝛘_OL3_𝛄_bsc0[54, 65, 63], ff_A_𝛘_YIL [54, 71], ff_A_𝛘_YIL_𝛄_bsc0 [54, 65, 71], ff_A_𝛘_SHAW [54, 78], ff_A_𝛘_SHAW_𝛄_bsc0 [54, 65, 78], ff_A_𝛘_SHAW_𝛄_SHAW [54, 78], ff_A_𝛘_ROC [54, 79], ff_A_𝛘_ROC_𝛄_bsc0 [54, 65, 79], ff_A_𝛘_ROC_𝛄_ROC [54, 79], ff_A_RNA_ROC [54, 79] and the newly developed force field parameters FF99_𝛘_KOL0_𝛄_bsc0. The newly derived parameter sets FF99_𝛘_KOL0_𝛄_bsc0 (developed for each of the modified residues separately) were used for the simulations of the methyl derivatives of inosine. The detailed description of the force field parameter sets used in this study is provided in Table 2.

Table 2

Relevant details of the force fields used in this study
Force fields	Applied revised parameters for torsion (s)	Definition of the applied torsional terms
ff_A	None	The AMBER parameters derived from Aduri et al. [54] parameters for inosine.
ff_A_𝛄_bsc0	𝛄_bsc0	The AMBER parameters derived from Aduri et al. [54] parameters for inosine in combination with the revised 𝛄 torsional parameters developed by Perez et al. [65]
ff_A_𝛘_OL3	𝛘_OL3	The AMBER parameters derived from Aduri et al. [54] parameters for inosine in combination with the revised glycosidic torsion parameters for adenosine developed by Zgarbova et al. [63]
ff_A_𝛘_OL3_𝛄_bsc0	𝛘_OL3 + 𝛄_bsc0	The AMBER parameters derived from Aduri et al. [54] parameters for inosine, 1-methylinosine, 2'-O-methylinosine, and 1,2'-O-dimethylinosine each in combination with the revised glycosidic torsion parameters for adenosine developed by Zgarbova et al. [63] and the revised 𝛄 torsional parameters developed by Perez et al. [65]
ff_A_𝛘_YIL	𝛘_YIL	The AMBER parameters derived from Aduri et al. [54] parameters for inosine in combination with the revised glycosidic torsion parameters for adenosine developed by Yildirim et al. [71]
ff_A_𝛘_YIL_𝛄_bsc0	𝛘_YIL + 𝛄_bsc0	The AMBER parameters derived from Aduri et al. [54] parameters for inosine in combination with the revised glycosidic torsion parameters for adenosine developed by Yildirim et al. [71] and the revised 𝛄 torsional parameters developed by Perez et al.[65]
ff_A_𝛘_SHAW	𝛘_SHAW	The AMBER parameters derived from Aduri et al. [54] parameters for inosine in combination with the revised glycosidic torsion parameters for adenosine developed by Tan et al. [78]
ff_A_𝛘_SHAW_𝛄_bsc0	𝛘_SHAW + 𝛄_bsc0	The AMBER parameters derived from Aduri et al. [54] parameters for inosine in combination with the revised glycosidic torsion parameters for adenosine developed by Tan et al. [78] and the revised 𝛄 torsional parameters developed by Perez et al. [65]
ff_A_𝛘_SHAW_𝛄_SHAW	𝛘_SHAW + 𝛄_SHAW	The AMBER parameters derived from Aduri et al. [54] parameters for inosine in combination with the revised glycosidic torsion parameters for adenosine and the revised 𝛄 torsional parameters developed by Tan et al. [78].
ff_A_𝛘_ROC	𝛘_ROC	The AMBER parameters derived from Aduri et al. [54] parameters for inosine in combination with the revised glycosidic torsion parameters for adenosine developed by Aytenfisu et al. [79]
ff_A_𝛘_ROC_𝛄_bsc0	𝛘_ROC+ 𝛄_bsc0	The AMBER parameters derived from Aduri et al. [54] parameters for inosine in combination with the revised glycosidic torsion parameters for adenosine developed by Aytenfisu et al. [79] along with the revised 𝛄 torsional parameters developed by Perez et al. [65]
ff_A_𝛘_ROC_𝛄_ROC	𝛘_ROC+ 𝛄_ROC	The AMBER parameters derived from Aduri et al. [54] parameters for inosine in combination with the revised glycosidic torsion parameters for adenosine and the revised 𝛄 torsional parameters developed by Aytenfisu et al. [79].
ff_A_RNA_ROC	𝛘_ROC+ 𝛄_ROC	The AMBER parameters derived from Aduri et al. [54] parameters for inosine in combination with the revised bond, angle and dihedral parameters corresponding to the C4’ atom, the revised glycosidic torsion parameters for adenosine and the revised 𝛄 torsional parameters developed by Aytenfisu et al. [79]. The bond, angle and dihedral parameters for the C8 atom (except for the glycosidic torsion parameters) in the RNA_ROC force field are obtained from the AMBER99 parameter sets).
FF99_𝛘_KOL0_𝛄_bsc0	𝛘_KOL0	The default AMBER99 parameters (Chetham et al. 1999) [REF] for the canonical nucleosides in combination with the the revised glycosidic torsion parameters and the reoptimized partial atomic charges developed (by method 2) in this study for inosine and its methyl derivatives along with the revised 𝛄 torsional parameters developed by Perez et al.[65] The bond angle and torsion parameters necessary for the N1-methyl group was added from the the AMBER parameters derived from Aduri et al. [54] parameters.
CHARMM36m	N.A.	CHARMM36m parameters for inosine developed by Xu et al, (2016) [83].

Table 3

Effect of the application of different gamma (𝛄) torsion parameters on the C3′-C4′ bond distance and C5′-C4′-C3′ angle of inosine
Bond/angle	ff_A_𝛘_SHAW	ff_A_𝛘_SHAW_𝛄_bsc0	ff_A_𝛘_SHAW_𝛄_SHAW	ff_A_𝛘_ROC	ff_A_𝛘_ROC_𝛄_bsc0	ff_A_𝛘_ROC_𝛄_ROC	ff_A_RNA_ROC	CHARMM36m
C3′-C4′ bond distance (avg)	1.5452 ± 0.00038	1.5453 ± 0.00015	1.5474 ± 0.00017	1.5437 ± 0.00015	1.5441 ± 0.00008	1.5444^(a)	1.5416± 0.0001	1.5393^(b)
C5′-C4′-C3′ angle (avg)	113.42 ± 0.008	114.1 ± 0.03	113.68 ± 0.064	113.48 ± 0.059	114.28± 0.032	113.74^(a)	113.77 ± 0.023	115.11^(b)
^aThe values corresponding to a set of 16 ns REMD simulation. ^bThe values corresponding to the concatenated trajectory of 200 ns MD simulation. The other values reported here are obtained from three independent sets (set 1, set 2 and set 3) of 16 ns REMD simulations (at 300 K).

Each of the solvated systems was energy minimized in two steps. The first set of energy minimization consisted of 500 steps of steepest descent followed by 500 steps of conjugate gradient optimization with the nucleosides restrained with a positional restraining force of 500 kcal/(mol Å²). The next set of energy minimization consisting of 1000 steps of steepest descent followed by 1500 steps of conjugate gradient optimization was carried out in the absence of any positional restraining force. Then the energy-minimized systems were equilibrated in two steps. In the first step, the systems were heated from 0 to 300 K in 20 ps with a 2 fs time step using constant volume dynamics with the nucleosides held fixed with a positional restraining force of 10 kcal/(mol-Å2). In the next step of equilibration, the systems were equilibrated without any restraint at 300 K for 200 ps with a 2 fs time step using constant pressure dynamics (reference pressure of 1 atm with a pressure relaxation time of 2 ps). The Berendsen barostat [80] was used for pressure control.

The final coordinates obtained after the completion of the equilibration steps were taken as the input coordinates for the REMD simulations. Before the REMD production run, the REMD equilibration step was carried out for 1 ns with a 2 fs time step with constant volume dynamics in which each system was equilibrated at 16 target temperatures that spanned over a range of 300 − 400 K (i.e., at T = 300.0, 305.8, 311.7, 317.8, 323.9, 330.2, 336.6, 343.1, 349.7, 356.5, 363.4, 370.5, 377.6, 384.9, 392.4, and 400.0 K). These equilibrated systems were used for the REMD production runs which consisted of 2000 cycles in constant volume and 4000 steps of molecular dynamics with a 2 fs time step before the attempted exchange between the 16 replicas with the neighboring temperatures. The exchange acceptance ratio between each pair of replicas was between 0.21 and 0.28. The frames were saved at 2 ps intervals. The REMD production runs generated simulation of 16 ns for each of the replicas, giving 256 ns of total simulation time. Three sets of REMD simulations were performed for each force field system combination yielding a total simulation of 768 ns in aggregate for inosine for each of the force field parameter sets (except for ff_A_𝛘_ROC_𝛄_ROC, for which one set of 16 ns REMD simulation was carried out),. For the methyl derivatives of inosine under study, one set of 16 ns REMD simulation (256 ns of total simulation) was carried out using the ff_A_𝛘_YIL_𝛄_bsc0, ff_A_𝛘_OL3_𝛄_bsc0 and FF99_𝛘KOL0_𝛄_bsc0 parameters respectively. Langevin dynamics (with random velocity scaling with 1 ps^− 1 collision frequency) was used for the propagation of the trajectories. To constrain the bonds that involved hydrogen atoms, the SHAKE [81] algorithm was used. Particle mesh Ewald (PME) method [82] was used for handling the electrostatic interactions and to include nonbonded interactions, a long-range cutoff of 8 Å was considered.

Additionally, we have also carried out two independent sets of normal MD simulations (100ns each) of inosine with the CHARMM36m [83] parameters for inosine at 298 K. The method for these simulations is provided in the supporting information.

Analyses of conformational ensembles

We have analyzed the simulated ensembles in terms of the population of sugar puckering conformations, percentage of SYN or ANTI conformations of the glycosidic (𝛘) torsional angle, the conformations corresponding to the 𝛄 torsional angle, and hydrogen bonding characteristics.

The convention as given in Saenger [84] was followed for the atom names and the dihedral angle nomenclature. For calculating the magnitude of the pseudorotation angle we followed Altona and Sundaralingam [85]. For direct comparison of the simulated conformational distributions and the equilibrium distributions of the pseudorotation angle (P) observed in experimental data (NMR), the pseudorotation angular space was divided into two regions of sugar puckering, i.e. NORTH (270° ≤ P < 90°) and SOUTH (90° ≤ P < 270°) which included the ideal C3′-endo and C2′-endo regions, respectively[86, 87]. For both residues the 𝛘 torsional angle was defined by the atoms O4′−C1′−N9 − C4. The angular range of 170°−300° corresponding to the 𝛘 torsional angle was considered as ANTI while that of 30°-90° was considered as the SYN conformation and the values which were beyond these two ranges were referred as OTHERS [68, 88, 89].

For the calculation of the γ torsional angle (measured with respect to O5′−C5′−C4′−C3′), the angular ranges of 60 ± 30°, 300 ± 30°, and (180 ± 30°) were considered as g⁺, g⁻, and TRANS conformations respectively and values outside these ranges were referred as OTHERS.

The conformational ensembles were also analyzed based on the C3′-O3′ bond distance, and values of the C2′-C3′-O3′ and C4′-C3′-O3′ sugar valence angles which were reported to depend on the sugar pucker conformation [68]. The values of C3′-C4′ bond length and C5′-C4′-C3′ angle (which were also found to be dependent upon the conformation of sugar puckering [68]) were also calculated to see how different sets of gamma torsional parameters influence the sugar pucker conformation of the nucleosides. Hydrogen bond formation was calculated using the cpptraj tool from Ambertools18 [76] and was considered if the distance between the donor and the acceptor atoms was ≤ 3 Å and the donor-hydrogen-acceptor angle was ≥ 135°. The averages and the standard deviations of the fraction (in %) of sugar pucker, the fraction (in %) of base orientation states, the fraction (in %) of 𝛄 conformational states, the values of the bond distances and angles as mentioned above were calculated from the three sets of 16ns REMD simulations (for the equilibrium ensembles at 300 K). The improper torsion angle 𝜿′ (C4 − N9 − C1′−C8–180°) was calculated to examine the pyramidalization of N9 (glycosidic nitrogen) and 𝜿′ = 0° corresponds to the idealized planar configuration of N9 (glycosidic nitrogen). The cpptraj tool from Ambertools18 [76] was used for the calculation of the different torsion angles, bond lengths and angles mentioned above.

Distribution of the pseudorotation angle (P)

For inosine, the Aduri et al. parameter sets [54] (ff_A) underestimated the population of NORTH sugar pucker (Fig. 4). ff_A_𝛘_OL3, ff_A_𝛘_OL3_𝛄_bsc0, ff_A_𝛘_YIL, ff_A_𝛘_YIL_𝛄_bsc0, ff_A_𝛘_SHAW, ff_A_𝛘_SHAW_𝛄_bsc0, ff_A_𝛘_ROC and ff_A_𝛘_ROC_𝛄_bsc0 predicted a larger population of NORTH sugar pucker than what was observed with the ff_A force field (Table S3, Fig. 4). The application of the revised γ torsion parameters (parmbsc0) [65] led to an improvement in the predicted NORTH population for each of the ff_A variant force fields and the values were closer to the experimentally observed population distribution. But the application of the revised γ torsion parameters (𝛄_SHAW) developed by Tan et al. [78] to the ff_A_𝛘_SHAW parameters and application of the revised γ torsion parameters (𝛄_ROC) developed by Aytenfisu et al. [79] to the ff_A_𝛘_ROC force field (for both ff_A_𝛘_ROC_𝛄_ROC and ff_A_RNA_ROC force fields) resulted in a decrease in the equilibrium population of the NORTH conformers and the equilibrium shifted almost entirely towards the SOUTH conformation. The ff_A_𝛘_ROC_𝛄_bsc0 parameters were able to reproduce the experimentally observed population of the NORTH and SOUTH conformers of inosine [48]. ff_A_𝛄_bsc0, ff_A_𝛘_SHAW, ff_A_𝛘_SHAW_𝛄_bsc0, ff_A_𝛘_ROC and the CHARMM36m force field parameters also predicted the population of the NORTH conformers close to the experimentally observed population and were able to reproduce the preference of the inosine residue towards the SOUTH conformation of sugar pucker. The FF99_𝛘_KOL0_𝛄_bsc0 force field parameters which is the combination of the AMBER99 parameters with the revised glycosidic torsion parameters (𝛘_KOL0) developed in the present study along with the recommended bsc0 correction (parmbsc0), predicted a population of the NORTH conformers which was much higher than that predicted by the ff_A parameters but also higher than the experimentally observed population.

For the three methyl derivatives of inosine under this study, i.e. 1-methylinosine, 2’-O-methylinosine and 1,2’-O-dimethylinosine, we applied the ff_A_𝛘_YIL_𝛄_bsc0, ff_A_𝛘_OL3_𝛄_bsc0, and the newly developed FF99_𝛘_KOL0_𝛄_bsc0 parameters sets respectively (Table S4). For 1-methylinosine, the ff_A_𝛘_YIL_𝛄_bsc0 parameters predicted the preference of the NORTH conformation while with the ff_A_𝛘_OL3_𝛄_bsc0 and the FF99_𝛘_KOL0_𝛄_bsc0 parameters a preference towards the SOUTH conformation was observed in agreement with the NMR data [62]. For the 2’-O-methylinosine residues, the ff_A_𝛘_YIL_𝛄_bsc0 and ff_A_𝛘_OL3_𝛄_bsc0 parameter sets predicted a preference for the SOUTH conformation while for 1,2’-O-dimethylinosine, these two force fields predicted a preference for the NORTH conformation. For each of these modifications, the newly developed FF99_𝛘_KOL0_𝛄_bsc0 parameter sets predicted a slight preference of the SOUTH conformation over the NORTH conformation.

Distribution of the glycosidic torsion angle (χ )

The ff_A force field failed to significantly populate the SYN region [48] for inosine even with the application of the bsc0 correction (parmbsc0) and predicted a preference towards the ANTI conformation (Table S3, Fig. 5). With the ff_A_𝛘_OL3, ff_A_𝛘_OL3_𝛄_bsc0, ff_A_𝛘_YIL, ff_A_𝛘_YIL_𝛄_bsc0, ff_A_𝛘_SHAW, ff_A_𝛘_SHAW_𝛄_bsc0 and ff_A_𝛘_SHAW_𝛄_SHAW force fields, the predicted population of the SYN conformers were greater than the ANTI conformers (Fig. 5). ff_A_𝛘_ROC and ff_A_𝛘_ROC_𝛄_bsc0 force fields generated smaller population of SYN conformers than the population of the ANTI conformers. The predicted population of SYN conformers by these force field parameters were greater than what were predicted by the ff_A and the ff_A_𝛄_bsc0 force fields. On the other hand, with ff_A_RNA_ROC, the ANTI base orientation was observed to be preferred and, the distribution of the glycosidic torsion was similar to what were observed with the ff_A and the ff_A_𝛄_bsc0 force fields. For the different combinations of glycosidic torsion and gamma torsion parameters with the ff_A parameters, the OTHERS region of base orientation was observed to be significantly populated. With the CHARMM36m [83] parameters, the SYN base orientation was preferred and the population of conformers with the OTHERS conformation were significantly reduced compared to the ff_A force field variants. The newly developed force field parameters FF99_𝛘_KOL0_𝛄_bsc0 generated the distribution of base orientation states similar to that generated by the CHARMM36m parameters.

For each of the three methyl derivatives of inosine in the present study, ff_A_𝛘_OL3_𝛄_bsc0, and ff_A_𝛘_YIL_𝛄_bsc0 parameter sets as well as the newly developed FF99_𝛘_KOL0_𝛄_bsc0 parameters, predicted the preference for SYN base orientation. With the application of the FF99_𝛘_KOL0_𝛄_bsc0 parameters, the population of the OTHERS conformers were much lower than those predicted by the other force fields.

The 2D correlation maps (Figures S5-8) revealed that for inosine, the ff_A and ff_A_𝛄_bsc0 force fields generated conformers which preferentially adopted the SOUTH/ANTI conformation. The ff_A_𝛘_OL3, ff_A_𝛘_OL3_𝛄_bsc0, ff_A_𝛘_YIL, ff_A_𝛘_YIL_𝛄_bsc0, ff_A_𝛘_SHAW, ff_A_𝛘_SHAW_𝛄_bsc0 force fields generated a greater population of SOUTH/SYN conformers of inosine than what were observed with the ff_A and ff_A_𝛄_bsc0 force fields. The ff_A_𝛘_ROC and ff_A_𝛘_ROC_𝛄_bsc0 force fields predominantly sampled the SOUTH/ANTI conformers. The ff_A_𝛘_SHAW_𝛄_SHAW generated a larger population of SOUTH/SYN conformers than that of the other conformers, but predicted an excess of SOUTH conformation. The ff_A_RNA_ROC force field showed a preference for the SOUTH/ANTI conformation. With the FF99_𝛘_KOL0_𝛄_bsc0 parameters, the SOUTH/SYN region was observed to be the most populated followed by the NORTH/SYN region and similar distribution was observed for the CHARMM36m parameters. For the methyl derivatives of inosine also, the FF99_𝛘_KOL0_𝛄_bsc0 parameters predicted a preference for the SOUTH/SYN conformation.

Distribution of the gamma torsion angle (γ)

The ff_A, ff_A_𝛘_OL3, ff_A_𝛘_YIL, ff_A_𝛘_SHAW and ff_A_𝛘_ROC force fields generated smaller populations of the g⁺ conformers for both the nucleosides, but the values corresponding to the ff_A_𝛘_YIL, were observed to be closer to that observed in NMR for inosine (Table S3, Fig. 6). As was observed in our earlier studies [90, 91], the application of the 𝛄_bsc0 parameters [65] resulted in larger (~ 90%) population of the g⁺ conformation compared to the experimentally observed values for both modified residues. The ff_A_𝛘_ROC_𝛄_ROC parameter sets (i.e., without the application of the revised parameters for the C4’ atom included in the RNA_ROC parameters) also failed to predict the preference for g⁺ conformation. The ff_A_𝛘_SHAW_𝛄_SHAW and the ff_A_RNA_ROC force fields predicted the population of the g⁺ conformers close to the experimentally observed population for inosine. With each of the ff_A_𝛘_OL3_𝛄_bsc0, ff_A_𝛘_YIL_𝛄_bsc0 and FF99_𝛘_KOL0_𝛄_bsc0 parameter sets, in the methyl derivatives of inosine studied in this work the g⁺ conformation was observed to be preferred (Table S4, Figure 7).

The 2D correlation maps (Figures S9-12) revealed that the ff_A, ff_A_𝛘_OL3, ff_A_𝛘_YIL, and ff_A_𝛘_SHAW force fields sampled nearly equal populations of the SOUTH/g⁺ and SOUTH/TRANS conformers of inosine and the population of the SOUTH/g- conformers were lesser than that of these two types of conformers. For the ff_A_𝛘_ROC force field, the population of the SOUTH/TRANS conformers were greater than that of the other conformers for each of the nucleosides. Inclusion of the 𝛄_bsc0 parameters with each of these force fields shifted the distribution towards the g⁺ conformation and also increased the population of the NORTH/g⁺ conformers for both nucleosides. With the ff_A_𝛘_SHAW_𝛄_SHAW and the ff_A_RNA_ROC force fields, a large population of SOUTH/g⁺ conformers was observed for each of the residues. The FF99_𝛘_KOL0_𝛄_bsc0 parameters predicted nearly equal populations of the SOUTH/g⁺ and NORTH/g + conformers for inosine (Figure S12). For the CHARMM36m parameters, the SOUTH/g⁺ region was observed to be the most populated followed by the NORTH/g⁺ region (Figure S13). With the FF99_𝛘_KOL0_𝛄_bsc0 parameters, each of the methyl derivatives of inosine, in the current study, preferentially adopted the SOUTH/g⁺ conformation, but the population of the NORTH/g⁺ conformers were also high (Figure S14). The time series plots of pseudorotation angle (P), glycosidic torsion angle (χ) and gamma torsion angle (𝛄) for inosine corresponding to the CHARMM36m parameters are provided in Figure S15.

Hydrogen bonding characteristics

The SYN conformation of inosine has been reported to be stabilized by the formation of intramolecular O5′-H5T—N3 hydrogen bond while the ANTI conformers can form the O2′-HO2′---N3 hydrogen bond [92]. The ff_A, ff_A_𝛄_bsc0 force fields generated much smaller number of conformers with the O5′-H5T—N3 hydrogen bonding interactions than what was predicted by the other force fields (Table S5). While with the ff_A_𝛘_OL3, ff_A_𝛘_YIL, ff_A_𝛘_SHAW and ff_A_𝛘_C force fields, the number of O5′-H5T—N3 hydrogen bonding conformers increased a little, the ff_A_𝛘_ROC force field generated a lower number of O5′-H5T—N3 hydrogen bonding conformers. Each of the ff_A_𝛘_OL3_𝛄_bsc0, ff_A_𝛘_YIL_𝛄_bsc0, ff_A_𝛘_SHAW_𝛄_bsc0, and ff_A_𝛘_SHAW_𝛄_SHAW force fields generated significantly greater number of conformers with O5′-H5T—N3 hydrogen bond than what was observed with the ff_A parameter sets. The application of the 𝛄_bsc0 resulted in an increase in the number of conformers with O5′-H5T—N3 hydrogen bond and did not result in any significant change in the number of O2′-HO2′---N3 hydrogen bonding conformers. With the ff_A_𝛘_ROC, ff_A_𝛘_ROC_𝛄_bsc0, and ff_A_𝛘_ROC_𝛄_ROC force fields, the population of the O2′-HO2′---N3 hydrogen bonding conformers were significantly higher than that with the other force fields. Interestingly, ff_A_𝛘_SHAW_𝛄_SHAW force field predicted a larger number of O5′-H5T—N3 hydrogen bonding conformers than all the other force fields. On the other hand, ff_A_𝛘_ROC_𝛄_ROC force field predicted small numbers of both O5’-H5T—N3 hydrogen bonding conformers and O2′-HO2′---N3 hydrogen bonding conformers. The FF99_𝛘_KOL0_𝛄_bsc0 and the CHARMM36m parameters also predicted a much greater populations of conformers with the O5′-H5T—N3 hydrogen bond than those of the conformers with the O2′-HO2′---N3 hydrogen bond. In general, it was observed that, for the force fields corresponding to which the SYN base orientation was preferred than the ANTI base orientation, the population of the conformers with O5′-H5T—N3 hydrogen bond was also observed to be higher (Table S5).

For 1-methylinosine also, a greater populations of conformers with the O5′-H5T—N3 hydrogen bond than those of the conformers with the O2′-HO2′---N3 hydrogen bond was observed for each of the ff_A_𝛘_OL3_𝛄_bsc0, ff_A_𝛘_YIL_𝛄_bsc0, and the newly developed FF99_𝛘_KOL0_𝛄_bsc0 parameter sets, suggesting the stabilization of the SYN conformation (Table S6). But with FF99_𝛘_KOL0_𝛄_bsc0, the population of the O2′-HO2′---N3 hydrogen bonding conformers were lower than what was observed for the other parameter sets. For 2'-O-methylinosine and 1,2’-O-dimethylinosine, ff_A_𝛘_YIL_𝛄_bsc0 and FF99_𝛘_KOL0_𝛄_bsc0 predicted a greater frequency of the O5′-H5T—N3 hydrogen bond than the ff_A_𝛘_OL3_𝛄_bsc0 parameter sets.

Distribution of the C3′-O3 ′ bond distance, C2′-C3′-O3′ and C4′-C3′-O3′ sugar valence angles for inosine

The values of the C3′-O3′ bond distance, C2′-C3′-O3′ and C4′-C3′-O3′ sugar valence angles have been reported to be correlated with the distribution of the pseudorotation angle [68] and hence these values might influence the distribution of the sugar pucker. According to Gelbin et al. [68] the mean values of the C3′-O3′ bond distance for the C2′-endo conformation of ribonucleosides was 1.427 Å and that corresponding to the C3′ endo conformation of ribonucleosides was 1.417 Å. The mean values of the C2′-C3′-O3′ sugar valence angle was reported to be 109.5° for the C2′-endo conformation of ribonucleosides and 113.7° for the C3′-endo conformation of ribonucleosides [68]. The mean values of the C4′-C3′-O3′ sugar valence angle was reported to be 109.4° for the C2′-endo conformation of ribonucleosides and 113.0° for the C3′-endo conformation of ribonucleosides [68].

Hence In the present study, we investigated the correlation of the C3′-O3′ bond distance and C2′-C3′-O3′ and C4′-C3′-O3′ sugar valence angles corresponding to the different combinations of glycosidic torsion and gamma torsion parameters to understand the role of the different sets of glycosidic torsion and gamma torsion parameters in modulating the sugar puckering state (Figures S16-24). We observed that, in general, the mean value of C3′-O3′ bond distance for inosine to be similar for all the force fields and the value was ~ 1.45 Å for the ff_A variant force fields. For the FF99_𝛘_KOL0_𝛄_bsc0 and the CHARMM36m parameters, the values were slightly lesser and ~ 1.42 Å and 1.44 Å respectively (Table S7). The 2D correlation maps (Figures S16-18) revealed a significant correlation of the C3′-O3′ bond distance with the pseudorotation angle. The SOUTH conformers explored a greater range of values of the C3′-O3′ bond distance than that of the NORTH conformers. (Figures S16-18).

The C2′-C3′-O3′ sugar valence angle for inosine (Table S7) was found to be on an average ~ 110.6° for all the force fields except ff_A_𝛘_ROC_𝛄_ROC, ff_A_𝛘_ROC_𝛄_ROC, and ff_A_𝛘_SHAW_𝛄_SHAW, which predicted the values of this angle to be ~ 109°, ~ 110° and 109.6° which may indicate a tendency of the 𝛄_ROC and the 𝛄_SHAW parameters to shift the pseudorotation equilibrium entirely towards the SOUTH conformation. However, the FF99_𝛘_KOL0_𝛄_bsc0 and the CHARMM36m parameters predicted this angle to be higher than the other parameter sets, i.e., ~ 111.4 ° and ~ 114° respectively. In general, the 2D correlation maps (Figures S19-21) suggested that the values of the C2′-C3′-O3′ angle for the NORTH conformers were greater than those of the SOUTH conformers.

The average C4′-C3′-O3′ sugar valence angle for inosine (Table S7) was found to be ~ 109.4°. But with the ff_A_𝛘_ROC_𝛄_ROC, ff_A_RNA_ROC, ff_A_𝛘_SHAW_𝛄_SHAWthe values were observed to be ~ 107.5°, ~ 110.4°, ~ 108.2° respectively. The FF99_𝛘_KOL0_𝛄_bsc0 and the CHARMM36m parameters predicted this angle to be higher than the other parameter sets, i.e., ~ 111.6 ° and ~ 112° respectively. In general, the 2D correlation maps (Figures S22-24) suggested that the values of the C4′-C3′-O3′ angle for the NORTH conformers were greater than those of the SOUTH conformers.

Distribution of the C3 ′ -C4 ′ bond distance and C5′-C4′-C3′ angle for inosine

The values of the C3′-C4′ bond distance and C5′-C4′-C3′ angle have been reported to be correlated also with the distribution of the pseudorotation angle [68] and changes in the parameters for the 𝛄 torsion angle can result in the changes in the values of the C3′-C4′ bond distance, C5′-C4′-C3′ angle which might lead to the changes in the distribution of the pseudorotation angle. According to Gelbin et al. [68], the mean values of the C3′-C4′ bond distance for the C2′-endo conformation of ribonucleosides was 1.527 Å and that corresponding to the C3′-endo conformation of ribonucleosides was 1.521 Å. The mean values of the C5′-C4′-C3′ angle was reported to be 115.2° for the C2′-endo conformation of ribonucleosides and 116° for the C3′-endo conformation of ribonucleosides [68].

In the present study we checked the changes in the values of the C3′-C4′ bond distance, C5′-C4′-C3′ angle for the ff_A_𝛘_SHAW, ff_A_𝛘_SHAW_𝛄_bsc0, ff_A_𝛘_SHAW_𝛄_SHAW, ff_A_𝛘_ROC, ff_A_𝛘_ROC_𝛄_bsc0, ff_A_𝛘_ROC_𝛄_ROC and ff_A_RNA_ROC force fields to further investigate the effect of the application of different sets of 𝛄 parameters on the sugar puckering states of ribonucleosides (Table 4, Figs. 8–9 and S25-28). The observed changes in the mean values of the C3′-C4′ bond distance and the C5′-C4′-C3′ angle for the nucleosides under this study upon application of different 𝛄 torsion parameters revealed the effect of these parameters on the pseudorotation equilibrium.

The mean values of the C3′-C4′ bond distance corresponding to the ff_A_𝛘_SHAW and ff_A_𝛘_SHAW_𝛄_bsc0 force fields were observed to be similar and were ~ 1.545 Å (Table 4). But upon inclusion of the 𝛄_SHAW parameters, i.e., with the ff_A_𝛘_SHAW_𝛄_SHAW force field, the values very slightly increased and were observed to be ~ 1.547 Å. Both ff_A_𝛘_ROC, ff_A_𝛘_ROC_𝛄_bsc0 and ff_A_𝛘_ROC_𝛄_ROC force fields predicted the C3′-C4′ bond distance of ~ 1.544 Å, while ff_A_RNA_ROC force field reduced the value to ~ 1.542 Å. The difference between the values corresponding to the ff_A_𝛘_ROC_𝛄_ROC and ff_A_RNA_ROC force fields might have arised due to the application of the revised parameters for the C4’ atom (corresponding to the RNA_ROC force field parameters) in the later.

With the ff_A_𝛘_SHAW force field, the C5′-C4′-C3′ angle was found to be ~ 113.4°. On inclusion of the 𝛄_bsc0 parameters (i.e. the ff_A_𝛘_SHAW_𝛄_bsc0 force field) the value was observed to be ~ 114.1° and on inclusion of the 𝛄_SHAW parameters (i.e. the ff_A_𝛘_SHAW_𝛄_SHAW force field) the value was found to be ~ 113.7°. The ff_A_𝛘_ROC, ff_A_𝛘_ROC_𝛄_bsc0, ff_A_𝛘_ROC_𝛄_ROC and ff_A_RNA_ROC force fields respectively predicted the values ~ 113.5°, ~ 114.3°, ~ 113.7° and ~ 113.7° for the C5′-C4′-C3′ angle. The CHARMM36m parameters predicted the values of the C3′-C4′ bond distance and the C5′-C4′-C3′ angle to be ~ 1.539 Å and 115.1° respectively.

These observations suggested that for the inosine residue although with the application of the 𝛄_bsc0 parameters [65] the C3′-C4′ bond distance did not change to a significant extent, the value of the C5′-C4′-C3′ angle significantly increased with a concomitant shift in the pseudorotation equilibrium slightly towards the NORTH sugar pucker. With the application of the 𝛄_ROC parameters[79], although the value of the C5′-C4′-C3′ angle did not change significantly, a decrease in the value of the C3′-C4′ bond distance was observed with a concomitant shift in the pseudorotation equilibrium towards the SOUTH sugar pucker.

Pyramidalization of the glycosidic nitrogen (N9) of inosine and its methyl derivatives

The deviation of the idealized planar structure of nucleosides has been suggested by earlier studies and this phenomenon can result in a complex effect on the accuracy of the currently available force field parameters [93, 94]. Sychrovsky et al, (2009), based on their extensive statistical analyses of ultrahigh resolution crystal structures, along with QM and MD studies, reported the existence of this novel structural property, i.e. pyramidalization of glycosidic nitrogen of purine and pyrimidine nucleobases within DNA and RNA oligomers [93]. The pyramidalization of the glycosidic nitrogen has been reported to depend on the local geometry of the nucleoside residue, i.e. the base orientation and the sugar pucker conformation and inversion of the conformation of the glycosidic nitrogen was observed upon the change of base orientation from SYN to ANTI [93]. In another study, they reported the effect of base pairing and site-specific isolation on the N9-pyramidalization of dG (deoxyguanosine) in DNA G-quadruplex [94].

As mentioned earlier, the magnitude of the improper torsion angle 𝜿′ (C4 − N9 − C1′−C8–180°) corresponds to the extent of N9-pyramidalization and a value of 𝜿′ equal to 0° indicates the idealized planar configuration of N9 (glycosidic nitrogen) [94]. In the present study, we compared the values of 𝜿′ torsion for inosine predicted by the newly derived FF99_𝛘_KOL0_𝛄_bsc0 and the CHARMM36m force field parameter and analyzed the dependence of the N9-pyramidalization on the orientation of the glycosidic torsion angle (𝛘) for inosine and its methyl derivatives (with the FF99_𝛘_KOL0_𝛄_bsc0 parameters) (Tables S8-9, Figures S29-30).

For inosine, the extent of N9-pyramidalization was observed to be slightly greater with the FF99_𝛘_KOL0_𝛄_bsc0 parameters than what was observed with the CHARMM36m parameters. The extents of N9-pyramidalization were found to be similar for inosine and its derivatives in this study. With the CHARMM36m parameter sets, nearly equal populations of SYN conformers with -ve 𝜿′ values and those with + ve 𝜿′ values were preferentially sampled. The population of the ANTI conformers with -ve 𝜿′ values and those with + ve 𝜿′ values were also nearly equal. With the FF99_𝛘_KOL0_𝛄_bsc0 parameters, SYN conformers with -ve 𝜿′ values were observed to be preferred for inosine and its methyl derivatives.

In this work, we have studied the conformational preferences of inosine (which is a derivative of the canonical nucleoside adenosine) and three of its methyl derivatives from the conformational ensembles obtained from REMD simulations (normal/regular MD simulations for the CHARMM36m parameters). Our preliminary/primary objective was to validate the force field parameters derived from the Aduri et al. parameter sets [54] for the inosine modification. Besides that, we also aimed at validating the different combinations of the Aduri et al. parameter sets for inosine with the available revised sets of glycosidic torsion parameters and gamma torsion parameters for adenosine to check whether these parameters are transferable to inosine for theoretically replicating the experimentally observed conformational distribution or if reoptimization of parameters for the individual nucleosides would be required. In an earlier study on a set of uridine derivatives, we reported that Aduri et al. parameters [54] were transferable only for particular cases [95].

Our analyses revealed that the Aduri et al. parameters [54] failed to reproduce the experimental (NMR) observations. Among the other force field parameter combinations, the ff_A_𝛘_OL3_𝛄_bsc0 and ff_A_𝛘_SHAW_𝛄_bsc0 parameters predicted the conformational preferences of inosine, close to the experimental values. The CHARMM36m parameters generated the conformational properties of inosine in excellent agreement with the experimental data [48]. The newly developed glycosidic torsion parameters in combination with the amber99 (FF99) parameters [64] and the revised 𝛄 torsion parameters developed by Perez et al. [65], i.e., the FF99_𝛘_KOL0_𝛄_bsc0 force field including the reoptimized sets of partial atomic charges also predicted the conformational preferences in general agreement with the experimental values. Hence, we developed the FF99_𝛘_KOL0_𝛄_bsc0 parameter sets for each of the methyl derivatives under this study following the generalized approach. For the methyl derivatives of inosine, the FF99_𝛘_KOL0_𝛄_bsc0 parameters predicted the predominance of the SOUTH/SYN/g⁺ conformers. The predicted populations of SOUTH and g⁺ conformers for 1-methylinosine were in agreement with the NMR data. The O5′-H5T—N3 hydrogen bonding interaction was observed to contribute in the stabilization of the preferred SYN conformation of inosine and its methyl derivatives.

The results obtained in this study also suggested the ability of the 𝛄_bsc0 parameters to shift the pseudorotation equilibrium slightly towards the NORTH sugar pucker from an excess of SOUTH sugar pucker for inosine. We report here our observations on the effect of the application of the recommended bsc0 correction (i.e. inclusion of the 𝛄_bsc0 parameters) on the value of the C5′-C4′-C3′ angle (which has been reported to be correlated with the sugar pucker conformation [68]). On the other hand, the tendency of the 𝛄_SHAW [78] and 𝛄_ROC [79] parameters to shift the pseudorotation equilibrium almost entirely towards the SOUTH sugar pucker could also be observed and both these parameters predicted lower values of the C5′-C4′-C3′ angle than what was observed with the 𝛄_bsc0 parameters [65]. The inclusion of the 𝛄_ROC parameters [79] also led to a decrement in the value of the C4′-C3′-O3′ angle which might be correlated with the predominance of the SOUTH conformation as according to Gelbin et al. [68], the value of this angle is lower corresponding to the SOUTH sugar pucker conformation. Our study highlights the strengths and weaknesses of the recent revisions of the AMBER force field parameters for the canonical nucleosides in sampling the conformations of the modified ribonucleosides similar to the experimentally observed conformations.

We also proposed a new set of force field parameters applying the bond, angles and dihedral parameters from the amber99 parameters (FF99) [64] for the canonical nucleosides and only the revised bond, angle and dihedral parameters dependent on the modification (methylation in this work) from Aduri et al. parameters [54] to increase the transferability of the parameters. We present a revised set of glycosidic torsion parameters (𝛘_KOL0) developed individually for inosine and its methyl derivatives along with a reoptimized set of partial atomic charges. In combination with the amber99 parameters and the revised set of 𝛄 torsion parameters (parmbsc0) [65], i.e., FF99_𝛘_KOL0_𝛄_bsc0, this was able to generate the conformational properties of inosine and its 1-methyl derivative in good agreement with the NMR data. The conformational characteristics of the other two methyl derivatives of inosine were also predicted using the FF99_𝛘_KOL0_𝛄_bsc0 parameters. Our results and the revised set of force field parameters might be helpful in the theoretical study of complex RNA structures containing these modifications.

Funding

Department of Science and Technology – Science and Engineering Research Board funding (EMR/2016/007753).

Department of Science and Technology – Innovation in Science Pursuit for Inspired Research Senior Research Fellowship (DST/INSPIRE Fellowship/2018/IF180895).

Conflicts of interest

There are no conflicts to declare.

Consent to participate

Not applicable.

Consent for publication

Not applicable.

Availability of data and material

All data and software are available. Input files required to reproduce the raw data and the output data are available in the supporting information. Additional data which support the findings in this study are available upon request from the corresponding author. The software and webservers used in this study are as follows AMBER16: http://ambermd.org/ . AMBER18: http://ambermd.org/. CHARMM (free version): https://www.charmm.org/ PyRED: https://upjv.q4md-forcefieldtools.org/REDServer-Development/ . Pymol: https://pymol.org/2/ . Avogadro: https://avogadro.cc/ . R 3.5.0: https://cran.r-project.org/ .

Code availability

Not applicable

Author Contributions

N.D., and A.L. jointly conceived and designed the study. N.D. generated the data and analyzed the results. All authors reviewed the manuscript and approved the final version.

ACKNOWLEDGMENTS

The authors acknowledge the support provided by the SERB funding (EMR/2016/007753). N.D. acknowledges support from the DST INSPIRE Senior Research Fellowship (DST/ INSPIRE Fellowship/2018/IF180895). Some of the simulations and calculations were performed at the Poznan Supercomputing and Networking Center.

Boccaletto P, Machnicka MA, Purta E, et al (2018) MODOMICS: a database of RNA modification pathways. 2017 update. Nucleic Acids Research 46:D303–D307
Björk GR (1995) Genetic dissection of synthesis and function of modified nucleosides in bacterial transfer RNA. Prog Nucleic Acid Res Mol Biol 50:263–338
Nachtergaele S, He C (2017) The emerging biology of RNA post-transcriptional modifications. RNA Biology 14:156–163
Lewis CJT, Pan T, Kalsotra A (2017) RNA modifications and structures cooperate to guide RNA–protein interactions. Nature Reviews Molecular Cell Biology 18:202–210
Holley RW, Everett GA, Madison JT, Zamir A (1965) Nucleotide Sequences in the Yeast Alanine Transfer Ribonucleic Acid. Journal of Biological Chemistry 240:2122–2128
Martin FH, Castro MM, Aboul-ela F, Tinoco I Jr (1985) Base pairing involving deoxyinosine: implications for probe design. Nucleic Acids Res 13:8927–8938
Nishikura K (2010) Functions and regulation of RNA editing by ADAR deaminases. Annu Rev Biochem 79:321–349
Paul MS, Bass BL (1998) Inosine exists in mRNA at tissue-specific levels and is most abundant in brain mRNA. EMBO J 17:1120–1127
Athanasiadis A, Rich A, Maas S (2004) Widespread A-to-I RNA editing of Alu-containing mRNAs in the human transcriptome. PLoS Biol 2:e391
Blow M, Futreal PA, Wooster R, Stratton MR (2004) A survey of RNA editing in human brain. Genome Res 14:2379–2387
Kim DDY, Kim TTY, Walsh T, et al (2004) Widespread RNA editing of embedded alu elements in the human transcriptome. Genome Res 14:1719–1725
Levanon EY, Eisenberg E, Yelin R, et al (2004) Systematic identification of abundant A-to-I editing sites in the human transcriptome. Nat Biotechnol 22:1001–1005
Bian Z, Ni Y, Xu J-R, Liu H (2019) A-to-I mRNA editing in fungi: occurrence, function, and evolution. Cell Mol Life Sci 76:329–340
Zheng Y, Lorenzo C, Beal PA (2017) DNA editing in DNA/RNA hybrids by adenosine deaminases that act on RNA. Nucleic Acids Res 45:3369–3377
Siegfried NA, Metzger SL, Bevilacqua PC (2007) Folding cooperativity in RNA and DNA is dependent on position in the helix. Biochemistry 46:172–181
Krepl M, Otyepka M, Banáš P, Šponer J (2013) Effect of guanine to inosine substitution on stability of canonical DNA and RNA duplexes: molecular dynamics thermodynamics integration study. J Phys Chem B 117:1872–1879
Brümmer A, Yang Y, Chan TW, Xiao X (2017) Structure-mediated modulation of mRNA abundance by A-to-I editing. Nat Commun 8:1255
Scadden ADJ, Smith CWJ (2001) RNAi is antagonized by A→I hyper‐editing. EMBO reports 2:1107–1111
Knight SW, Bass BL (2002) The role of RNA editing by ADARs in RNAi. Mol Cell 10:809–817
Nishikura K (2006) Editor meets silencer: crosstalk between RNA editing and RNA interference. Nat Rev Mol Cell Biol 7:919–931
Kawahara Y, Zinshteyn B, Sethupathy P, et al (2007) Redirection of silencing targets by adenosine-to-inosine editing of miRNAs. Science 315:1137–1140
Valente L, Nishikura K (2005) ADAR gene family and A-to-I RNA editing: diverse roles in posttranscriptional gene regulation. Prog Nucleic Acid Res Mol Biol 79:299–338
Solomon O, Oren S, Safran M, et al (2013) Global regulation of alternative splicing by adenosine deaminase acting on RNA (ADAR). RNA 19:591–604
Parada GE, Munita R, Cerda CA, Gysling K (2014) A comprehensive survey of non-canonical splice sites in the human transcriptome. Nucleic Acids Res 42:10564–10578
Pullirsch D, Jantsch MF (2010) Proteome diversification by adenosine to inosine RNA-editing. RNA Biology 7:205–212
Licht K, Hartl M, Amman F, et al (2019) Inosine induces context-dependent recoding and translational stalling. Nucleic Acids Res 47:3–14
Daniel C, Silberberg G, Behm M, Öhman M (2014) Alu elements shape the primate transcriptome by cis-regulation of RNA editing. Genome Biol 15:R28
Mannion NM, Greenwood SM, Young R, et al (2014) The RNA-editing enzyme ADAR1 controls innate immune responses to RNA. Cell Rep 9:1482–1494
Cattaneo R, Schmid A, Eschle D, et al (1988) Biased hypermutation and other genetic changes in defective measles viruses in human brain infections. Cell 55:255–265
Taylor DR, Puig M, Darnell MER, et al (2005) New antiviral pathway that mediates hepatitis C virus replicon interferon sensitivity through ADAR1. J Virol 79:6291–6298
Samuel CE (2011) Adenosine deaminases acting on RNA (ADARs) are both antiviral and proviral. Virology 411:180–193
Laxminarayana D, O’Rourke KS, Maas S, Olorenshaw I (2007) Altered editing in RNA editing adenosine deaminase ADAR2 gene transcripts of systemic lupus erythematosus T lymphocytes. Immunology 121:359–369
Shallev L, Kopel E, Feiglin A, et al (2018) Decreased A-to-I RNA editing as a source of keratinocytes’ dsRNA in psoriasis. RNA 24:828–840
Zhang H, Fu Q, Shi X, et al (2020) Human A-to-I RNA editing SNP loci are enriched in GWAS signals for autoimmune diseases and under balancing selection. Genome Biol 21:288
Akbarian S, Smith MA, Jones EG (1995) Editing for an AMPA receptor subunit RNA in prefrontal cortex and striatum in Alzheimer’s disease, Huntington's disease and schizophrenia. Brain Research 699:297–304
Kawahara Y, Kwak S, Sun H, et al (2003) Human spinal motoneurons express low relative abundance of GluR2 mRNA: an implication for excitotoxicity in ALS. J Neurochem 85:680–689
Khermesh K, D’Erchia AM, Barak M, et al (2016) Reduced levels of protein recoding by A-to-I RNA editing in Alzheimer’s disease. RNA 22:290–302
Ishiuchi S, Tsuzuki K, Yoshida Y, et al (2002) Blockage of Ca2 -permeable AMPA receptors suppresses migration and induces apoptosis in human glioblastoma cells. Nature Medicine 8:971–978
Cenci C, Barzotti R, Galeano F, et al (2008) Down-regulation of RNA editing in pediatric astrocytomas: ADAR2 editing activity inhibits cell migration and proliferation. J Biol Chem 283:7251–7260
Galeano F, Tomaselli S, Locatelli F, Gallo A (2012) A-to-I RNA editing: the “ADAR” side of human cancer. Semin Cell Dev Biol 23:244–250
Xu X, Wang Y, Liang H (2018) The role of A-to-I RNA editing in cancer development. Curr Opin Genet Dev 48:51–56
Hogg RS, Rhone SA, Yip B, et al (1998) Antiviral effect of double and triple drug combinations amongst HIV-infected adults: lessons from the implementation of viral load-driven antiretroviral therapy. AIDS 12:279–284
De Clercq E, Li G (2016) Approved Antiviral Drugs over the Past 50 Years. Clin Microbiol Rev 29:695–747
Hashimoto M, Taguchi K, Ishiguro T, et al (2018) Pharmacokinetic properties of a novel inosine analog, 4′-cyano-2′-deoxyinosine, after oral administration in rats. PLOS ONE 13:e0198636
Sliva J, Pantzartzi CN, Votava M (2019) Inosine Pranobex: A Key Player in the Game Against a Wide Range of Viral Infections and Non-Infectious Diseases. Adv Ther 36:1878–1905
Corfield PW, Hunter WN, Brown T, et al (1987) Inosine.adenine base pairs in a B-DNA duplex. Nucleic Acids Res 15:7935–7949
Basilio C, Wahba AJ, Lengyel P, et al (1962) Synthetic polynucleotides and the amino acid code. V. Proc Natl Acad Sci U S A 48:613–616
Rossi C, Picchi M, Tiezzi E, et al (1990) Conformational and dynamic investigation in solution of inosine and its molecular complex, inosiplex, by proton and carbon NMR spectroscopy. Magn Reson Chem 28:348–354
Watkins NE Jr, SantaLucia J Jr (2005) Nearest-neighbor thermodynamics of deoxyinosine pairs in DNA duplexes. Nucleic Acids Res 33:6258–6267
Wright DJ, Rice JL, Yanker DM, Znosko BM (2007) Nearest Neighbor Parameters for Inosine·Uridine Pairs in RNA Duplexes. Biochemistry 46:4625–4634
Wright DJ, Force CR, Znosko BM (2018) Stability of RNA duplexes containing inosine·cytosine pairs. Nucleic Acids Res 46:12099–12108
Skinner A, Yang C-H, Hincks K, et al (2020) Experimental and theoretical rationalization for the base pairing abilities of inosine, guanosine, adenosine, and their corresponding 8-oxo-7,8-dihydropurine, and 8-bromopurine analogues within A-form duplexes of RNA. Biopolymers 111:e23410
Sakuraba S, Iwakiri J, Hamada M, et al (2020) Free-Energy Calculation of Ribonucleic Inosines and Its Application to Nearest-Neighbor Parameters. J Chem Theory Comput 16:5923–5935
Aduri R, Psciuk BT, Saro P, et al (2007) AMBER Force Field Parameters for the Naturally Occurring Modified Nucleosides in RNA. J Chem Theory Comput 3:1464–1475
Yamaizumi Z, Ihara M, Kuchino Y, et al (1982) Archaebacterial tRNA contains 1-methylinosine at residue 57 in T psi C-loop. Nucleic Acids Symp Ser 209–213
McCloskey JA, Crain PF, Edmonds CG, et al (1987) Structure determination of a new fluorescent tricyclic nucleoside from archaebacterial tRNA. Nucleic Acids Research 15:683–693
Grosjean H, Constantinesco F, Foiret D, Benachenhou N (1995) A novel enzymatic pathway leading to 1-methylinosine modification inHaloferax volcaniitRNA. Nucleic Acids Research 23:4312–4319
Björk GR, Jacobsson K, Nilsson K, et al (2001) A primordial tRNA modification required for the evolution of life? EMBO J 20:231–239
Jin X, Lv Z, Gao J, et al (2019) AtTrm5a catalyses 1-methylguanosine and 1-methylinosine formation on tRNAs and is important for vegetative and reproductive growth in Arabidopsis thaliana. Nucleic Acids Res 47:883–898
Dutta N, Deb I, Sarzynska J, Lahiri A (2022) Inosine and its methyl derivatives: Occurrence, biogenesis, and function in RNA. Prog Biophys Mol Biol 169-170:21–52
Gray MW (1976) O2’-Methylinosine, a constituent of the ribosomal RNA of Crithidia fasciculata. Nucleic Acids Research 3:977–988
Wishart DS, Knox C, Guo AC, et al (2009) HMDB: a knowledgebase for the human metabolome. Nucleic Acids Research 37:D603–D610
Zgarbová M, Otyepka M, Sponer J, et al (2011) Refinement of the Cornell et al. Nucleic Acids Force Field Based on Reference Quantum Chemical Calculations of Glycosidic Torsion Profiles. J Chem Theory Comput 7:2886–2902
Cheatham TE 3rd, Cieplak P, Kollman PA (1999) A modified version of the Cornell et al. force field with improved sugar pucker phases and helical repeat. J Biomol Struct Dyn 16:845–862
Pérez A, Marchán I, Svozil D, et al (2007) Refinement of the AMBER Force Field for Nucleic Acids: Improving the Description of α/γ Conformers. Biophysical Journal 92:3817–3829
Antony J, Rendell AP, Yang R, et al (2011) Modelling the Runtime of the Gaussian Computational Chemistry Application and Assessing the Impacts of Microarchitectural Variations. Procedia Computer Science 4:281–291
Case DA, Betz RM, Cerutti DS, et al (2016) Amber 2016; University of California: San Francisco, 2016.
Gelbin A, Schneider B, Clowney L, et al (1996) Geometric parameters in nucleic acids: Sugar and phosphate constituents. J Am Chem Soc 118:519–529
Schaftenaar G, Noordik JH (2000) Journal of Computer-Aided Molecular Design. 14:123–134
Richardson JS, Schneider B, Murray LW, et al (2008) RNA backbone: consensus all-angle conformers and modular string nomenclature (an RNA Ontology Consortium contribution). RNA 14:465–481
Yildirim I, Stern HA, Kennedy SD, et al (2010) Reparameterization of RNA χ Torsion Parameters for the AMBER Force Field and Comparison to NMR Spectra for Cytidine and Uridine. Journal of Chemical Theory and Computation 6:1520–1531
Bayly CI, Cieplak P, Cornell W, Kollman PA (1993) A well-behaved electrostatic potential based method using charge restraints for deriving atomic charges: the RESP model. The Journal of Physical Chemistry 97:10269–10280
Cieplak P, Cornell WD, Bayly C, Kollman PA (1995) Application of the multimolecule and multiconformational RESP methodology to biopolymers: Charge derivation for DNA, RNA, and proteins. Journal of Computational Chemistry 16:1357–1377
Vanquelef E, Simon S, Marquant G, et al (2011) R.E.D. Server: a web service for deriving RESP and ESP charges and building force field libraries for new molecules and molecular fragments. Nucleic Acids Research 39:W511–W517
Sugita Y, Okamoto Y (1999) Replica-exchange molecular dynamics method for protein folding. Chemical Physics Letters 314:141–151
Case DA, Ben-Shalom IY, Brozell SR, et al (2018) AMBER 2018; University of California: San Francisco, 2018.
Jorgensen WL, Chandrasekhar J, Madura JD, et al (1983) Comparison of simple potential functions for simulating liquid water. The Journal of Chemical Physics 79:926–935
Tan D, Piana S, Dirks RM, Shaw DE (2018) RNA force field with accuracy comparable to state-of-the-art protein force fields. Proc Natl Acad Sci U S A 115:E1346–E1355
Aytenfisu AH, Spasic A, Grossfield A, et al (2017) Revised RNA Dihedral Parameters for the Amber Force Field Improve RNA Molecular Dynamics. J Chem Theory Comput 13:900–915
Berendsen HJC, Postma JPM, van Gunsteren WF, et al (1984) Molecular dynamics with coupling to an external bath. J Chem Phys 81:3684–3690
Ryckaert J-P, Ciccotti G, Berendsen HJC (1977) Numerical integration of the cartesian equations of motion of a system with constraints: molecular dynamics of n-alkanes. Journal of Computational Physics 23:327–341
Darden T, York D, Pedersen L (1993) Particle mesh Ewald: An N⋅log(N) method for Ewald sums in large systems. J Chem Phys 98:10089–10092
Xu Y, Vanommeslaeghe K, Aleksandrov A, et al (2016) Additive CHARMM force field for naturally occurring modified ribonucleotides. J Comput Chem 37:896–912
Saenger W (1984) Principles of Nucleic Acid Structure. Springer Advanced Texts in Chemistry
Altona C, Sundaralingam M (1972) Conformational analysis of the sugar ring in nucleosides and nucleotides. A new description using the concept of pseudorotation. J Am Chem Soc 94:8205–8212
(1983) IUPAC-IUB Joint Commission on Biochemical Nomenclature (JCBN). Abbreviations and symbols for the description of conformations of polynucleotide chains. Recommendations 1982. Eur J Biochem 131:9–15
Saenger W (1984) Structures and Conformational Properties of Bases, Furanose Sugars, and Phosphate Groups. Principles of Nucleic Acid Structure 51–104
Foloppe N, Hartmann B, Nilsson L, MacKerell AD Jr (2002) Intrinsic conformational energetics associated with the glycosyl torsion in DNA: a quantum mechanical study. Biophys J 82:1554–1569
Foloppe N, Nilsson L (2005) Toward a Full Characterization of Nucleic Acid Components in Aqueous Solution: Simulations of Nucleosides. The Journal of Physical Chemistry B 109:9119–9131
Dutta N, Sarzynska J, Lahiri A (2020) Molecular Dynamics Simulation of the Conformational Preferences of Pseudouridine Derivatives: Improving the Distribution in the Glycosidic Torsion Space. J Chem Inf Model 60:4995–5002
Dutta N, Deb I, Sarzynska J, Lahiri A (2021) Data-informed reparameterization of modified RNA and the effect of explicit water models: Application to pseudouridine and derivatives
Alvarez-Ros MC, Alcolea Palafox M (2013) Molecular structure of the nucleoside analogue inosine using DFT methods: Conformational analysis, crystal simulations and possible behaviour. Journal of Molecular Structure 1047:358–371
Sychrovsky V, Foldynova-Trantirkova S, Spackova N, et al (2009) Revisiting the planarity of nucleic acid bases: Pyramidilization at glycosidic nitrogen in purine bases is modulated by orientation of glycosidic torsion. Nucleic Acids Res 37:7321–7331
Sychrovský V, Sochorová Vokáčová Z, Trantírek L (2012) Guanine bases in DNA G-quadruplex adopt nonplanar geometries owing to solvation and base pairing. J Phys Chem A 116:4144–4151
Deb I, Sarzynska J, Nilsson L, Lahiri A (2014) Conformational preferences of modified uridines: comparison of AMBER derived force fields. J Chem Inf Model 54:1129–1142

Table 1. Revised sets of glycosidic torsional parameters for inosine (INO) and its methyl derivatives (1MI, MRI and MMI).

Modified

nucleoside/s

Torsion/s

n

V_n

𝛘_KOL0

INO

O4’-C1’-N9-C8

(OS-CT-N*-Q1)

1

2

3

4

0.0464629

1.21305

-0.0132474

0.458791

C2’-C1’-N9-C8

(CT-CT-N*-Q1)

1

2

3

4

1.1051

0.0343393

-0.00648998

0.560197

1MI

O4’-C1’-N9-C8

(OS-CT-N*-Q1)

1

2

3

4

0.246844

1.38382

0.130711

0.651255

C2’-C1’-N9-C8

(CT-CT-N*-Q1)

1

2

3

4

1.30067

0.197192

0.15392

0.749508

MRI

O4’-C1’-N9-C8

(OS-CT-N*-Q1)

1

2

3

4

-0.131643

1.01949

-0.108929

0.281191

C2’-C1’-N9-C8

(CT-CT-N*-Q1)

1

2

3

4

0.981303

-0.272234

-0.154972

0.506944

MMI

O4’-C1’-N9-C8

(OS-CT-N*-Q1)

1

2

3

4

-0.0897368

0.985703

-0.132695

0.305859

C2’-C1’-N9-C8

(CT-CT-N*-Q1)

1

2

3

4

1.0236

-0.26396

-0.195876

0.515077

Table 2. Relevant details of the force fields used in this study

Force fields	Applied revised parameters for torsion (s)	Definition of the applied torsional terms
ff_A	None	The AMBER parameters derived from Aduri et al. [54] parameters for inosine.
ff_A_𝛄_bsc0	𝛄_bsc0	The AMBER parameters derived from Aduri et al. [54] parameters for inosine in combination with the revised 𝛄 torsional parameters developed by Perez et al. [65]
ff_A_𝛘_OL3	𝛘_OL3	The AMBER parameters derived from Aduri et al. [54] parameters for inosine in combination with the revised glycosidic torsion parameters for adenosine developed by Zgarbova et al. [63]
ff_A_𝛘_OL3_𝛄_bsc0	𝛘_OL3+ 𝛄_bsc0	The AMBER parameters derived from Aduri et al. [54] parameters for inosine, 1-methylinosine, 2'-O-methylinosine, and 1,2'-O-dimethylinosine each in combination with the revised glycosidic torsion parameters for adenosine developed by Zgarbova et al. [63] and the revised 𝛄 torsional parameters developed by Perez et al. [65]
ff_A_𝛘_YIL	𝛘_YIL	The AMBER parameters derived from Aduri et al. [54] parameters for inosine in combination with the revised glycosidic torsion parameters for adenosine developed by Yildirim et al. [71]
ff_A_𝛘_YIL_𝛄_bsc0	𝛘_YIL + 𝛄_bsc0	The AMBER parameters derived from Aduri et al. [54] parameters for inosine in combination with the revised glycosidic torsion parameters for adenosine developed by Yildirim et al. [71] and the revised 𝛄 torsional parameters developed by Perez et al.[65]
ff_A_𝛘_SHAW	𝛘_SHAW	The AMBER parameters derived from Aduri et al. [54] parameters for inosine in combination with the revised glycosidic torsion parameters for adenosine developed by Tan et al. [78]
ff_A_𝛘_SHAW_𝛄_bsc0	𝛘_SHAW + 𝛄_bsc0	The AMBER parameters derived from Aduri et al. [54] parameters for inosine in combination with the revised glycosidic torsion parameters for adenosine developed by Tan et al. [78] and the revised 𝛄 torsional parameters developed by Perez et al. [65]
ff_A_𝛘_SHAW_𝛄_SHAW	𝛘_SHAW + 𝛄_SHAW	The AMBER parameters derived from Aduri et al. [54] parameters for inosine in combination with the revised glycosidic torsion parameters for adenosine and the revised 𝛄 torsional parameters developed by Tan et al. [78].
ff_A_𝛘_ROC	𝛘_ROC	The AMBER parameters derived from Aduri et al. [54] parameters for inosine in combination with the revised glycosidic torsion parameters for adenosine developed by Aytenfisu et al. [79]
ff_A_𝛘_ROC_𝛄_bsc0	𝛘_ROC+ 𝛄_bsc0	The AMBER parameters derived from Aduri et al. [54] parameters for inosine in combination with the revised glycosidic torsion parameters for adenosine developed by Aytenfisu et al. [79] along with the revised 𝛄 torsional parameters developed by Perez et al. [65]
ff_A_𝛘_ROC_𝛄_ROC	𝛘_ROC+ 𝛄_ROC	The AMBER parameters derived from Aduri et al. [54] parameters for inosine in combination with the revised glycosidic torsion parameters for adenosine and the revised 𝛄 torsional parameters developed by Aytenfisu et al. [79].
ff_A_RNA_ROC	𝛘_ROC+ 𝛄_ROC	The AMBER parameters derived from Aduri et al. [54] parameters for inosine in combination with the revised bond, angle and dihedral parameters corresponding to the C4’ atom , the revised glycosidic torsion parameters for adenosine and the revised 𝛄 torsional parameters developed by Aytenfisu et al. [79]. The bond, angle and dihedral parameters for the C8 atom (except for the glycosidic torsion parameters) in the RNA_ROC force field are obtained from the AMBER99 parameter sets).
FF99_𝛘_KOL0_𝛄_bsc0	𝛘_KOL0	The default AMBER99 parameters (Chetham et al. 1999) [REF] for the canonical nucleosides in combination with the the revised glycosidic torsion parameters and the reoptimized partial atomic charges developed (by method 2) in this study for inosine and its methyl derivatives along with the revised 𝛄 torsional parameters developed by Perez et al.[65] The bond angle and torsion parameters necessary for the N1-methyl group was added from the the AMBER parameters derived from Aduri et al. [54] parameters.
CHARMM36m	N.A.	CHARMM36m parameters for inosine developed by Xu et al, (2016) [83].

Table 3. Effect of the application of different gamma (𝛄) torsion parameters on the C3′-C4′ bond distance and C5′-C4′-C3′ angle of inosine

Bond/angle	ff_A_𝛘_SHAW	ff_A_𝛘_SHAW_𝛄_bsc0	ff_A_𝛘_SHAW_𝛄_SHAW	ff_A_𝛘_ROC	ff_A_𝛘_ROC_𝛄_bsc0	ff_A_𝛘_ROC_𝛄_ROC	ff_A_RNA_ROC	CHARMM36m
C3′-C4′ bond distance (avg)	1.5452 ± 0.00038	1.5453 ± 0.00015	1.5474 ± 0.00017	1.5437±0.00015	1.5441±0.00008	1.5444^(a)	1.5416± 0.0001	1.5393^(b)
C5′-C4′-C3′ angle (avg)	113.42±0.008	114.1 ±0.03	113.68 ±0.064	113.48± 0.059	114.28± 0.032	113.74^(a)	113.77±0.023	115.11^(b)
^aThe values corresponding to a set of 16 ns REMD simulation. ^bThe values corresponding to the concatenated trajectory of 200 ns MD simulation. The other values reported here are obtained from three independent sets (set 1, set 2 and set 3) of 16 ns REMD simulations (at 300 K).

No competing interests reported.

Conformational preferences of inosine and its methyl derivatives: Comparison of the AMBER derived force field parameters and reparameterization of the glycosidic torsion parameters

Status:

Version 1

Abstract

Figures

Introduction

Method

Computational details

Ab initio potential energy surface (PES) scan

RESP fitting

Molecular mechanical energy minimization

Fitting glycosidic torsion (𝛘) parameters

E_𝛘 = E_QM - E_MM (1)

Replica exchange molecular dynamics simulations

Analyses of conformational ensembles

Results And Discussion

Distribution of the pseudorotation angle (P)

Distribution of the glycosidic torsion angle (χ )

Distribution of the gamma torsion angle (γ)

Hydrogen bonding characteristics

Distribution of the C3′-O3 ′ bond distance, C2′-C3′-O3′ and C4′-C3′-O3′ sugar valence angles for inosine

Distribution of the C3 ′ -C4 ′ bond distance and C5′-C4′-C3′ angle for inosine

Pyramidalization of the glycosidic nitrogen (N9) of inosine and its methyl derivatives

Conclusion

Declarations

References

Tables

Additional Declarations

Supplementary Files

Status:

Version 1

Conformational preferences of inosine and its methyl derivatives: Comparison of the AMBER derived force field parameters and reparameterization of the glycosidic torsion parameters

Status:

Version 1

Abstract

Figures

Introduction

Method

Computational details

Ab initio potential energy surface (PES) scan

RESP fitting

Molecular mechanical energy minimization

Fitting glycosidic torsion (𝛘) parameters

E𝛘 = EQM - EMM (1)

Replica exchange molecular dynamics simulations

Analyses of conformational ensembles

Results And Discussion

Distribution of the pseudorotation angle (P)

Distribution of the glycosidic torsion angle (χ )

Distribution of the gamma torsion angle (γ)

Hydrogen bonding characteristics

Distribution of the C3′-O3 ′ bond distance, C2′-C3′-O3′ and C4′-C3′-O3′ sugar valence angles for inosine

Distribution of the C3 ′ -C4 ′ bond distance and C5′-C4′-C3′ angle for inosine

Pyramidalization of the glycosidic nitrogen (N9) of inosine and its methyl derivatives

Conclusion

Declarations

References

Tables

Additional Declarations

Supplementary Files

Status:

Version 1

E_𝛘 = E_QM - E_MM (1)